RBD11, a bioengineered Rab11-binding module for visualizing and analyzing endogenous Rab11

Rab GTPases belong to the Ras superfamily of small GTPases and play important roles in membrane trafficking in eukaryotic cells (reviewed in Hutagalung and Novick, 2011; Zhen and Stenmark, 2015; Pfeffer, 2017; Homma et al., 2021). Like other Ras-like GTPases, Rabs cycle between a GTP-bound active state and a GDP-bound inactive state, and the active Rabs promote various membrane trafficking steps, including vesicle budding, tethering, docking and fusion, through interaction with their specific effectors. Rab cycling is spatiotemporally controlled by two regulatory enzymes, a guanine nucleotide exchange factor (GEF) and a GTPase-activating protein (GAP) (reviewed in Ishida et al., 2016; Lamber et al., 2019). Rabs constitute the largest subfamily of the Ras superfamily, and ∼60 different Rabs have been identified in mammals, as opposed to only 11 Rabs in the unicellular budding yeast (Diekmann et al., 2011; Klöpper et al., 2012). The expansion of Rab isoforms in multicellular eukaryotes, especially in higher eukaryotes, is generally thought to be related to the complexity of their tissues, which consist of highly specialized, differentiated cells that contain unique membrane trafficking pathways. The function of each Rab in mammals has been gradually elucidated, but the precise function and localization of most mammalian Rabs remains largely unknown.

Several methods of investigating the function and localization of specific Rabs, such as overexpression of constitutively active or negative (CA/CN) Rab mutants have been developed (reviewed in Fukuda, 2010). One such method uses fluorescently tagged Rab-binding (or effector) domains (RBDs) to visualize ‘endogenous’ Rabs and inhibit their functions. However, a drawback of this method is that many of the RBDs bind to several distinct Rabs (Fukuda et al., 2008; Gillingham et al., 2014), and, with a few exceptions, their Rab-binding specificity has never been thoroughly investigated (Fukuda et al., 2008, 2011; Nottingham et al., 2011; Espinosa et al., 2009; Ohishi et al., 2019). Even the representative effector proteins [e.g. rabenosyn-5, Rab-interacting lysosomal protein (RILP) and Rab11-FIPs] of the well-characterized, evolutionarily conserved Rabs (Rab5, Rab7 and Rab11 proteins, respectively) bind to several distinct Rabs (Eathiraj et al., 2005; Fukuda et al., 2008; Kelly et al., 2009; Matsui et al., 2012; Schafer et al., 2016). Thus, careful evaluation of their effects on membrane trafficking is necessary when their RBDs are used as dominant-negative constructs, because they can trap several distinct Rabs. Thus, an artificial RBD that can recognize a ‘single Rab isoform’ must be generated by bioengineering techniques to overcome this problem.

In this study, we bioengineered and developed an RBD that is specific for active Rab11 (herein referring to the Rab11A and Rab11B isoforms unless otherwise indicated) from the C-terminal domain of Rab11-FIP2 and Rab11-FIP4 (Cullis et al., 2002; Hales et al., 2001; Lindsay and McCaffrey, 2002; Wallace et al., 2002) and named it RBD11. We then demonstrated that RBD11 allowed visualization of endogenous Rab11 without altering its distribution or single-lumen formation in three-dimensional (3D) cysts formed by Madin–Darby canine kidney (MDCK) cells. We also developed a tandem RBD11, named 2×RBD11, as a dominant-negative construct and showed that expression of 2×RBD11 in MDCK 3D cysts induced a multi-lumen phenotype, the same as occurs in Rab11-deficient cysts (Bryant et al., 2010; Mrozowska and Fukuda, 2016a; Homma et al., 2019). In addition, we developed a tetracycline (Tet)-inducible 2×RBD11 and an artificially oligomerized domain (FM)-tagged RBD11 (Rivera et al., 2000) and demonstrated their usefulness in temporally and reversibly analyzing Rab11-dependent membrane trafficking.

To develop a specific RBD that binds to active Rab11 alone, we turned our attention to the well-known Rab11 effectors, the Rab11-FIP proteins (reviewed in Horgan and McCaffrey, 2009). Five different Rab11-FIPs (Rab11-FIP1–Rab11-FIP5) have been identified in humans and mice (Fig. 1A) and their binding properties to several Rabs, such as Rab11 and Rab25, have been well characterized (Prekeris et al., 2000, 2001; Hales et al., 2001; Lindsay et al., 2002; Wallace et al., 2002; Lall et al., 2013, 2015). However, their Rab-binding specificity for all mammalian Rabs had never been thoroughly investigated. To comprehensively identify their Rab-binding specificity, we cloned cDNAs encoding the C-terminal domain (i.e. Rab-binding domain; RBD) of mouse Rab11-FIP1–Rab11-FIP5 (location indicated by the brackets in Fig. 1A) and performed yeast-two hybrid assays using 62 different constitutively negative (CN) Rab mutants [e.g. Rab11A(N124I)] or constitutively active (CA) Rab mutants [e.g. Rab11A(Q70L)] as bait (Fig. 1C). Consistent with the results of previous studies, all of the Rab11-FIPs bound to the CA form of Rab11A and Rab11B (red boxes in Fig. 1C), but each of them exhibited slightly different Rab-binding specificity. In addition, several previously unknown interactions of Rab11-FIPs with Rabs (i.e. Rab20 and Rab42) were also observed; for example, Rab11-FIP2 bound to the CN form of Rab11A and Rab11B (Rab11A/B) and Rab20 (white boxes in Fig. 1C), and to the CA form of Rab11A/B, Rab14 and Rab25, whereas Rab11-FIP4 bound to the CA form of Rab11A/B and Rab42. Based on their domain organizations and the sequence similarity of their RBDs (Fig. 1A,B), Rab11-FIPs are classified into two groups, class I (Rab11-FIP1, Rab11-FIP2 and Rab11-FIP5) and class II (Rab11-FIP3 and Rab11-FIP4). The class I Rab11-FIPs are characterized by binding to the CA form of Rab14 (yellow boxes in Fig. 1C) (Fukuda et al., 2008; Lall et al., 2015), whereas the class II Rab11-FIPs are characterized by binding to the CA form of Rab42 (blue boxes in Fig. 1C). We especially noted that all of the Rab11-FIPs except Rab11-FIP4-C bound to the CN form of Rab11A and Rab11B (orange boxes in Fig. 1C) and to the GDP-bound form of Rab11 (Junutula et al., 2004). Rab11-FIP2 also bound to Rab11A/B(S25N), another CN form of Rab11A/B (Hales et al., 2001; Lindsay and McCaffrey, 2002), whereas Rab11-FIP4 did not (Fig. S1A,B). In view of its narrow Rab-binding specificity and strict GTP-dependency, we decided to use Rab11-FIP4-C as the main backbone to develop an active Rab11-specific binding module.

Since the class II Rab11-FIPs bound to Rab42 in addition to Rab11 (Fig. 1C), we next investigated whether Rab42 binds to the Rab11-binding site of Rab11-FIP4-C (or Rab11-FIP3-C). To do so, we deleted one-third of the C-terminal portion of Rab11-FIP4-C (or Rab11-FIP3-C) (Fig. 2A; see also Fig. 3A), which is known to be essential for Rab11 binding (Eathiraj et al., 2006; Jagoe et al., 2006; Shiba et al., 2006), and tested the ability of Rab11 and Rab42 to bind to Rab11-FIP3-ΔC and Rab11-FIP4-ΔC by yeast two-hybrid assays (Fig. 2B). The results showed that Rab11-FIP3-ΔC and Rab11-FIP4-ΔC interacted with neither Rab11A/B nor Rab42, suggesting that Rab11 and Rab42 bind to the same region of Rab11-FIP3 and Rab11-FIP4. To artificially alter the Rab-binding specificity of Rab11-FIP4-C, we then prepared two chimeric proteins between Rab11-FIP2-C and Rab11-FIP4-C (black bars and gray bars, respectively, in Fig. 2A). Although Rab11-FIP4/2-C2 completely lacked Rab11- and Rab42-binding ability, Rab11-FIP4/2-C1 bound to the CA form of Rab11A/B, but not to the CA form of Rab42 (Fig. 2C). Fortunately, Rab11-FIP4/2-C1 bound strongly to active Rab11A/B and weakly to Rab25 (sometimes called Rab11C), but it did not bind to inactive Rabs (Fig. 2D).

To further manipulate the Rab25-binding ability of Rab11-FIP4/2-C1, we performed site-directed mutagenesis, especially focusing on amino acids that are conserved only in Rab11-FIP4, because Rab11-FIP4-C alone did not bind to Rab25 in our two-hybrid assays (green boxes in Fig. 1C). Sequence comparisons of the RBDs of Rab11-FIPs enabled us to identify Met-616 (corresponding to Ile-481 of Rab11-FIP2) and Asp-625 (corresponding to Glu-490 of Rab11-FIP2) as unique residues in Rab11-FIP4 (green background in Fig. 3A). When the glutamate residue of Rab11-FIP4/2-C1 (red arrowhead on the right in Fig. 3A) was mutated to an alanine residue, the resulting Rab11-FIP4/2-C1-EA mutant hardly interacted with Rab25(CA) (lane 4 in Fig. 3B). When the isoleucine residue of Rab11-FIP4/2-C1, which corresponds to Ile-480 of Rab11-FIP2, which is essential for Rab11 interaction (Jagoe et al., 2006), was further mutated to methionine (red arrowhead on the left in Fig. 3A), the resulting Rab11-FIP4/2-C1-IM/EA mutant completely lost its Rab11/25(CA)-binding ability (lane 6 in Fig. 3B). We therefore decided to use Rab11-FIP4/2-C1-EA as RBD11 (for ‘RBD specific for active Rab11’) and Rab11-FIP4/2-C1-IM/EA (referred to as RBD11-mut hereafter) as an ideal negative control for RBD11.

To confirm that RBD11 directly binds to Rab11 in a GTP-dependent manner, we performed direct binding assays using purified components [T7-tagged RBD11 and GST–Rab11A(CA/CN)] (Fig. 3C). As anticipated, T7–RBD11 strongly bound to GST–Rab11A(CA), and hardly bound to GST–Rab11A(CN) at all (lanes 11 and 12 in Fig. 3C). Although T7–Rab11-FIP2-C also preferentially bound to GST–Rab11(CA) based on the results of the precipitation assays, it always bound to GST–Rab11(CN) more strongly than RBD11 did (lanes 9 and 12 in Fig. 3C), consistent with the results of the yeast two-hybrid assays described above. Since in addition to binding to Rab11, Rab11-FIP4 has been reported to bind to Arf6, another type of Ras-like small GTPase (Fielding et al., 2005; Shiba et al., 2006), we also confirmed by yeast two-hybrid assays that RBD11 does not trap active Arf6 (Fig. S1C).

Next, we investigated whether RBD11 also selectively binds to Rab11 in cultured mammalian cells by using a recently established Rab KO collection (Homma et al., 2019), because we thought that the results obtained above in yeast cells may not be able to be simply applied to the Rab-binding specificity of RBD11 in mammalian cells. When EGFP-tagged Rab11-FIP2-C, RBD11, and RBD11-mut were each stably expressed in wild-type (control), Rab11A/B-KO (Rab11-KO), Rab14-KO, Rab25-KO, and Rab42-KO MDCK cells, both Rab11-FIP2-C and RBD11 showed a punctate distribution in the control cells, whereas RBD11-mut exhibited a cytosolic distribution (far left column in Fig. 3D). It should be noted that Rab11-KO cells were the only Rab-KO cells in which RBD11 had a completely cytosolic distribution (middle row in Fig. 3D). By contrast, a punctate distribution of Rab11-FIP2-C was still observed even in Rab11-KO cells, although its punctate signals were clearly decreased (top row in Fig. 3D), suggesting that Rab11-FIP2-C traps Rabs other than Rab11A/B in cultured mammalian cells. We then immunostained for endogenous Rab11 with a specific antibody and confirmed that RBD11, and not RBD11-mut, colocalized with endogenous Rab11 (Fig. 3E). Consistent with the recycling endosomal localization of Rab11, RBD11 did not colocalize with any other organelle markers, including GM130 (also known as GOLGA2; a Golgi marker), EEA1 (early endosome antigen 1; an early endosome marker), LBPA (lysobisphosphatidic acid; a late endosome marker), and LAMP2 (lysosomal-associated membrane protein 2; a lysosome marker) (Fig. S2A). Intriguingly, RBD11 did not colocalize with transferrin receptor (TfR), another recycling endosome marker, indicating that RBD11 can visualize Rab11-positive recycling endosomes, but not TfR-positive recycling endosomes (bottom panels in Fig. S2A). Such distinct intracellular localization of Rab11 and TfR has also been reported in other cell types such as PC12 cells (Kobayashi and Fukuda, 2013). Moreover, Rab11 is known to localize apical recycling endosomes (Bryant et al., 2010), whereas transferrin (Tf) recycling predominantly occurs at the basolateral membrane of polarized MDCK cells (Fuller and Simons, 1986), suggesting that Tf recycling mainly occurs independently of Rab11 in MDCK cells. Indeed, it has previously been reported that Tf recycling occurs in MDCK cells even in the presence of a dominant negative Rab11A(S25N) (Gallo et al., 2014).

To investigate the effect of RBD11 on Rab11-dependent membrane trafficking events other than Tf recycling, we focused on single-lumen formation by 3D cysts of MDCK cells, because a multi-lumen phenotype was clearly observed in Rab11-KO cysts and Rab11-knockdown (KD) cysts (far left image in Fig. 4B) (Bryant et al., 2010; Homma et al., 2019). We prepared 3D MDCK cysts stably expressing EGFP-tagged RBD11, RBD11-mut or Rab11-FIP2-C and visualized their luminal domain (i.e. apical domain) by staining with anti-ezrin antibody. Since both Rab11-FIP2-C and RBD11 recognized endogenous Rab11 (Fig. 3D,E), we initially expected that their expression should induce a typical multi-lumen phenotype by trapping endogenous Rab11 (i.e. by serving as a ‘Rab11 trapper’). However, what we found was, although many cysts stably expressing EGFP–Rab11-FIP2-C contained multiple small lumens (upper right image in Fig. 4B), contrary to our expectations, stable expression of EGFP–RBD11 (or EGFP–RBD11-mut) failed to impair the formation of a single lumen (lower panels in Fig. 4B).

To determine why RBD11 failed to affect single-lumen formation, we focused on another biochemical property of Rab11-FIP2-C, that is, its dimerization activity (Lindsay and McCaffrey, 2002; Junutula et al., 2004; Jagoe et al., 2006; Wei et al., 2006), because a dimerized protein would trap its ligand more efficiently by increasing its local concentration. Consistent with the results of previous studies, the results of yeast two-hybrid assays showed that Rab11-FIP2-C formed homodimers, but that RBD11 did not exhibit homodimerization activity (compare lanes 3 and 4 in Fig. 4C). We therefore attempted to confer a dominant-negative function on RBD11 by arranging RBD11 in tandem (named 2×RBD11) (Fig. 4A). When EGFP-tagged 2×RBD11 was stably expressed in MDCK cells, large Rab11-positive puncta, which also colocalized with EGFP–2×RBD11 (insets in the left column of Fig. 4D), were observed in the perinuclear region. Moreover, these puncta did not colocalize with other organelles, including the Golgi, early endosomes, late endosomes, lysosomes and TfR-positive recycling endosomes, where Rab11 was not present (Fig. S2B). By contrast, no such large puncta were observed in EGFP–2×RBD11-mut-expressing cells (right column of Fig. 4D). In contrast to the original EGFP–RBD11 and EGFP–2×RBD11-mut, EGFP–2×RBD11 was found to significantly inhibit single-lumen formation in 3D cysts (Fig. 4E).

Finally, we attempted to create two additional RBD11-based tools capable of temporally inhibiting the function of endogenous Rab11. First, we established MDCK cell lines in which expression of 2×RBD11 (or 2×RBD11-mut as a control) was specifically induced by Tet (Fig. 5A, upper two constructs) and confirmed its doxycycline (Dox)-inducible expression (Fig. 5B). We then used the Tet-inducible system to investigate the possible involvement of Rab11 in single-lumen formation by 3D cysts and to determine the stage at which Rab11 functions during lumenogenesis. As shown in Fig. 5C,D, Tet-induced expression of 2×RBD11 in the first half of cyst growth (from day 1 to day 3; Fig. 5C,Di–iv) resulted in a significant increase in the number of cysts containing multiple lumens, whereas its expression in the latter half did not (from day 4 to day 7; Fig. 5C,Dv). By contrast, Tet-induced expression of 2×RBD11-mut had no effect on lumenogenesis under any conditions (Fig. 5C), indicating that the multi-lumen phenotype induced by 2×RBD11 in the first half of cyst growth is attributable to inhibition of endogenous Rab11.

Because 2×RBD11, not RBD11 alone, inhibited single-lumen formation by 3D cysts (Fig. 4E), we assumed that artificial regulation of RBD11 dimerization (i.e. a rapid transition between monomer and dimer) in living cells would allow us to temporally and reversibly inhibit the function of endogenous Rab11. To test our assumption, we focused on a drug-regulated homodimerization domain [i.e. the FM domain; a mutant FKBP12 (also known as FKBP1A) that contains a Phe-to-Met substitution] (Rivera et al., 2000) and prepared FM-tagged RBD11 and RBD11-mut (Fig. 5A, lower two constructs). In the absence of D/D solubilizer, FM–RBD11 forms a dimer, which presumably inhibits the function of endogenous Rab11, the same as 2×RBD11 does. By contrast, in the presence of D/D solubilizer, FM–RBD11 is a monomer, which is unlikely to have any effect on the function of Rab11. We therefore treated MDCK cells stably expressing FM–RBD11 (or FM–RBD11-mut) with D/D solubilizer for the times indicated in Fig. 5E and examined its effect on lumenogenesis of 3D cysts. Consistent with the results obtained with the Tet-inducible 2×RBD11 described above, treatment of FM–RBD11-expressing cells with D/D solubilizer only during the first 2 days of cyst growth completely reversed the inhibitory effect of FM–RBD11, and the cysts formed a normal single lumen (Fig. 5Ei,Fi). By contrast, when cells were exposed to D/D solubilizer after two days of cyst growth, a significantly increased number of cysts containing multiple lumens was observed even in the presence of D/D solubilizer (Fig. 5Eii–iv,Fiv). Again, the Rab11-binding-deficient FM–RBD11-mut had no effect on lumenogenesis under any conditions (Fig. 5E). All of these findings are highly consistent with the results of previous studies showing that apical membrane proteins are transcytosed to the newly formed apical domain through Rab11-positive endosomes in the early stage of cyst formation (Schlüter et al., 2009) and that Rab11 mediates the formation of a single apical membrane initiation site (Bryant et al., 2010; Mrozowska and Fukuda, 2016a). Since these Rab11-dependent events occur 16–36 h after the start of cyst growth (Mrozowska and Fukuda, 2016b), it is reasonable to expect that inhibition of endogenous Rab11 with 2×RBD11 or FM–RBD11 during the initial 48-h period of cyst growth is the most effective means of inducing a multi-lumen phenotype.

In the present study, we performed mutational and chimeric analyses of Rab11-FIP2 and Rab11-FIP4, and succeeded in developing an artificial protein module named RBD11 that specifically binds to the active Rab11 isoforms, that is active Rab11A and Rab11B, both in vitro and in cultured cells. We think that the RBD11 we developed has several advantages over the RBD of Rab11-FIP2. The first advantage is its exclusive Rab11-binding specificity and specific recognition of the GTP-bound form of Rab11 (Fig. 3). The second advantage is that a strict negative control for RBD11, named RBD11-mut, a point mutant of RBD11 that completely lacks Rab11-binding ability, can be used to determine the subcellular localization (Fig. 3D,E) and inhibit the function of endogenous Rab11 (Fig. 4E). The third advantage is that, since, in contrast to the original Rab11-FIP2-C construct (Fig. 4B), stable expression of RBD11 itself in MDCK cells had no effect on the function of Rab11, it can be used as a tool to visualize endogenous, active Rab11 in vivo without altering its localization or inhibiting its function. It should be noted, however, that tandem RBD11 (2×RBD11) and artificially dimerized RBD11 (FM–RBD11) significantly inhibited single-lumen formation by 3D MDCK cysts, thereby leading to a multi-lumen phenotype (Figs 4E and 5), which is characteristic of Rab11-deficient cysts (Fig. 4B). Thus, Tet-inducible 2×RBD11 and artificially dimerized RBD11 are unique tools that can be used to temporally and reversibly analyze the function of Rab11 at the endogenous protein level.

Several methods for analyzing Rab11-mediated membrane trafficking have become available thus far. The most widely used method is overexpression of a CA or CN form of Rab11. However, a drawback of this method is that CA/CN Rabs have sometimes affected membrane trafficking even though the corresponding Rabs have not been endogenously expressed. By contrast, KD of Rab11 with a specific siRNA and KO of Rab11 by genome-editing technologies are powerful methods for analyzing the function of endogenous Rab11. However, because two Rab11 isoforms, Rab11A and Rab11B, are present in mammals and function redundantly, at least in epithelial morphogenesis by MDCK cells (Homma et al., 2019), simultaneous KD or KO of Rab11A/B is necessary, and KD/KO efficiency is a limiting factor. Moreover, off-target effects of siRNA or guide RNA should also be considered, and appropriate rescue experiments are generally required. In that sense, the dimeric form of FM–RBD11 and monomeric form of FM–RBD11 (or RBD11 and RBD11-mut) developed in this study are ideal positive and negative controls, respectively, for analyzing the localization and function of endogenous Rab11. A Förster resonance energy transfer (FRET)-based Rab11 sensor, named AS-Rab11, has previously been reported as a means of visualizing Rab11 activation and inactivation in vivo (Campa et al., 2018). Although AS-Rab11 enables spatiotemporal visualization of Rab11 activation and inactivation, it is incapable of inhibiting the function of endogenous Rab11. Conversely, RBD11 is unable to visualize Rab11 activation and inactivation of Rab11 (i.e. it only visualizes active Rab11), but FM–RBD11 is capable of temporally inhibiting the function of endogenous Rab11. Similarly, optogenetically oligomerized Rab11 has been reported to inhibit the function of Rab11 (Nguyen et al., 2016). Although this tool is superior to FM–RBD11 in terms of spatial regulation, it does not directly inhibit the function of endogenous Rab11 and instead inhibits the trafficking of recycling endosomes, where ectopically expressed oligomerized Rab11 is present. Based on all of the above findings taken together, we think that the RBD11 tools developed in this study will serve as powerful tools for initial assessments of the function of endogenous Rab11 in many cell types, because they are easily expressed by means of plasmid transfection or retrovirus infection without the need for any special equipment. However, use of different Rab11 tools, including RBD11, in combination will certainly be necessary to fully understand the spatiotemporal regulation of Rab11-mediated membrane trafficking.

In conclusion, we have bioengineered a Rab11-specific binding module, named RBD11, and further developed FM–RBD11, which is capable of visualizing endogenous Rab11 in the monomer state and inhibiting the function of endogenous Rab11 in the dimer state. Because Rab11 is highly conserved in vertebrates, our RBD11 tools could be applied to various mammalian cell lines, and even to animal models. It might be possible to apply this strategy to Rabs by using their specific effector domains with FM-tag, which will contribute to the elucidation of the molecular mechanisms of Rab-mediated membrane trafficking in the future.

The following antibodies were obtained commercially: anti-Rab11 rabbit polyclonal antibody (Invitrogen, Carlsbad, CA; #71-5300), which recognizes both Rab11A and Rab11B, anti-GM130 mouse monoclonal antibody (BD Biosciences, San Jose, CA; #610823), anti-EEA1 mouse monoclonal antibody (BD Biosciences; #610456), anti-LBPA mouse monoclonal antibody (Echelon Biosciences, Salt Lake City, UT; #Z-PLBPA), anti-LAMP2 mouse monoclonal antibody (Thermo Fisher Scientific, Waltham, MA; #MA-28269), anti-TfR mouse monoclonal antibody (Invitrogen; #13-6800), anti-β-actin mouse monoclonal antibody (Applied Biological Materials; #G043), anti-ezrin mouse monoclonal antibody (Abcam, Cambridge, UK; #ab4069), horseradish peroxidase (HRP)-conjugated anti-GFP polyclonal antibody (MBL, Nagoya, Japan; #598-7), HRP-conjugated anti-mouse-IgG goat polyclonal antibody (SouthernBiotech, Birmingham, AL; #1031-05), Alexa Fluor 555⁺-conjugated anti-mouse-IgG goat polyclonal antibody (Thermo Fisher Scientific; #A32727), and Alexa Fluor 555⁺-conjugated anti-rabbit-IgG goat polyclonal antibody (Thermo Fisher Scientific; #A32732). Other reagents used in this study were also obtained commercially: doxycycline (FUJIFILM Wako Pure Chemical Corporation, Osaka, Japan) and D/D solubilizer (Takara Bio, Shiga, Japan).

cDNAs encoding the C-terminal 102 amino acids (a.a.) of mouse Rab11-FIP1/RCP, the C-terminal 124 a.a. of mouse Rab11-FIP2, the C-terminal 100 a.a.of Rab11-FIP3, the C-terminal 99 a.a. of mouse Rab11-FIP4, and the C-terminal 184 a.a.of mouse Rab11-FIP5/Rip11 were amplified from the Marathon-Ready adult mouse brain and testis cDNAs (Clontech/Takara Bio) by performing PCR using the standard molecular biology techniques. After verifying their sequences, they were subcloned into the pGAD-C1 vector (James et al., 1996) to perform yeast two-hybrid assays. Deletion mutants of Rab11-FIP3 and Rab11-FIP4 (ΔC), chimeric mutants between Rab11-FIP2 and Rab11-FIP4 (FIP4/2-C1 and FIP4/2-C2), FIP4/2-C1 point mutants [EA (called RBD11) and IM/EA (called RBD11 mut); see Fig. 3A for details] were also prepared by the standard molecular biology techniques, including the PCR sewing technique. The Rab11-FIP2-C, RBD11 and RBD-mut cDNA fragments were subcloned into the pEGFP-C1 vector (Clontech/Takara Bio), pEGFP-C1-FM vector (Hirano et al., 2016), pMRX-IRES-puro-EGFP vector (a kind gift from Dr Shoji Yamaoka, Tokyo Medical and Dental University, Tokyo, Japan; Saitoh et al., 2003), pRetroX-TetOne-Puro vector (Clontech/Takara Bio) and/or pEF-T7 tag vector (Fukuda et al., 1999).

cDNAs encoding the following CN mutants of mouse or human Rabs were also produced by the standard molecular biology techniques: Rab1A(N124I), Rab1B(N121I), Rab2A(N119I), Rab2B(N119I), Rab3A(N135I), Rab3B(N135I), Rab3C(N143I), Rab3D(N135I), Rab4A(N126I), Rab4B(N121I), Rab5A(N133I), Rab5B(N147I), Rab5C(N134I), Rab6A(N126I), Rab6B(N126I), Rab6C(N126I), Rab41/6D(N144I), Rab7(N125I), Rab7B/42(N124I), Rab8A(N121I), Rab8B(N121I), Rab9A(N124I), Rab9B(N124I), Rab10(N122I), Rab11A(N124I), Rab11B(N124I), Rab12(N154I), Rab13(N121I), Rab14(N124I), Rab15(N121I), Rab17(N132I), Rab18(N122I), Rab19(N130I), Rab20(N113I), Rab21(N130I), Rab22A(N118I), Rab22B(N118I), Rab23(N121I), Rab24(T120I), Rab25(N125I), Rab26(N181I), Rab27A(N133I), Rab27B(N133I), Rab28(N129I), Rab29(N125I), Rab30(N122I), Rab32(N141I), Rab33A(N151I), Rab33B(N148I), Rab34(S166I), Rab35(N120I), Rab36(T171I), Rab37(N143I), Rab38(N127I), Rab39A(H127I), Rab39B(H123I), Rab40A(N126I), Rab40AL(N126I), Rab40B(N126I), Rab40C(N126I), Rab43/41(N129I), and Rab42/43(H127I). The nomenclature of the Rabs in this study is in accordance with the National Center for Biotechnology Information (NCBI) database, and the names of several Rabs in the report by Itoh et al. (2006) are different (indicated by slash in Fig. 1C). The Rab CN mutants lacking a 3′ region that encodes cysteine residue(s) for geranylgeranylation were subcloned into the pGBD-C1 vector [named pGBD-C1-Rabs(CN)ΔCys; James et al., 1996]. Mouse Arf6-Q67L (CA form) and Arf6-T27N (CN form) cDNA fragments (Kobayashi and Fukuda, 2012) were also subcloned into the pGBD-C1 vector. pGBD-C1-Rabs(CA)ΔCys vectors were prepared as described previously (Fukuda et al., 2008). pGBD-C1-Rab11A(S25N)ΔCys, -Rab11B(S25N)ΔCys, -Rab14(S25N)ΔCys, -Rab20(T19N)ΔCys, -Rab25(T26N)ΔCys, and -Rab42(T23N)ΔCys were also prepared as described previously (Tamura et al., 2009). Mouse Rab11A(Q70L) (the CA form; Itoh et al., 2006) and Rab11A(N124I) (the CN form) cDNA fragments were subcloned into the pGEX-4T-3 vector (GE Healthcare, Buckinghamshire, UK). The sequences of the oligonucleotides used for plasmid constructions in this study are available from the corresponding authors on request. RBD11 tools and Rab panel tools for yeast two-hybrid assays (pGBD-C1-Rabs) are available from RIKEN BioResource Center in Japan (https://dnaconda.riken.jp/search/depositor/dep005893.html; Cat# [RDB18788-RBD18925]).

The yeast strain (PJ69-4A), medium, culture conditions, and transformation protocol used were as described previously (James et al., 1996). The yeast two-hybrid assays were performed using pGBD-C1-Rabs(CA/CN)ΔCys and pGAD-C1-Rab11-FIPs-C or pGAD-C1-RBD11 (WT and mutants) as described previously (Fukuda et al., 2008, 2011). Yeast cells on the selection medium [SC-AHLW: synthetic complete (SC) medium lacking adenine, histidine, leucine and tryptophan] were incubated at 30°C for ∼1 week.

COS-7 cells and MDCK cells (parental and Rab-KO MDCK-II cells; RIKEN BioResource Center, Cat# RCB5112, RCB5115, RCB5125, RCB5139, and RCB5148) (see Homma et al., 2019; Rab11-KO#27) were cultured at 37°C in Dulbecco's modified Eagle's medium (DMEM) (FUJIFILM Wako Pure Chemical Corporation) supplemented with 10% fetal bovine serum, 100 units/ml penicillin G, and 100 µg/ml streptomycin in a 5% CO₂ incubator. One day after plating COS-7 cells in a 6-cm dish (3×10⁵ cells), plasmids were transfected into the cells by using Lipofectamine 2000 (Thermo Fisher Scientific) according to the manufacturer's instructions.

For retrovirus production, Plat-E cells (a kind gift from Dr Toshio Kitamura, The University of Tokyo, Tokyo, Japan; Morita et al., 2000) were plated on a 35mm-dish (4×10⁵ cells/dish) and incubated for 24 h. The cells were transiently transfected with pMRX and pLP/VSVG plasmids (Thermo Fisher Scientific) by using Lipofectamine 2000. After 24 h, the medium was replaced with fresh medium, and the cells were cultured for an additional 24 h. The medium was then collected and centrifuged at 17,900 g for 3 min to remove debris. The virus-containing medium was added to the MDCK cell culture with 8 µg/ml polybrene. Uninfected cells were removed by treatment with 1 µg/ml puromycin.

MDCK cells were suspended in the culture medium containing 12 mM HEPES, pH 7.2 and 2 mg/ml collagen I on ice. The mixture was then dispensed into a 24-well plate and maintained at 37°C for 1 h. After adding 2 ml of culture medium to each well, the cells were cultured for 7 or 8 days. Then, 2 µg/ml of Dox (Fig. 5C,D) or 250 nM D/D solubilizer (Fig. 5E,F) was added to the culture medium for the times indicated in each figure.

Cells were fixed with 10% trichloroacetic acid (TCA; for MDCK cysts) or 4% paraformaldehyde (PFA; for other cell cultures), permeabilized with 0.1% Triton X-100 in phosphate-buffered saline (PBS) for 3 min (for MDCK cysts) or 50 µg/ml digitonin in PBS for 5 min (for other cell cultures), and incubated with a blocking solution (1% bovine serum albumin in PBS) at room temperature for 1 h (for MDCK cysts) or 20 min (for other cell cultures). The cells were then incubated for 1 h at room temperature with primary antibodies [anti-Rab11 (1:300 dilution), anti-GM130 (1:500 dilution), anti-EEA1 (1:500 dilution), anti-LBPA (1:500 dilution), anti-TfR (1:500 dilution), anti-LAMP2 (1:500 dilution), anti-GFP (1:2000 dilution), and anti-ezrin antibody (1:300 dilution)], then for 1 h at room temperature with Alexa Fluor 555⁺-conjugated anti-rabbit IgG together with DAPI. Only the fixation step was performed on the samples to be examined for EGFP fluorescence alone (Fig. 3D). All samples were examined through a confocal fluorescence microscope (Fluoview 1000; Olympus, Tokyo, Japan) equipped with a Plan-Apochromat 100×/1.45 NA oil-immersion objective lens.

Protein extracts were obtained from cells that had been lysed with a lysis buffer [50 mM HEPES-KOH, pH 7.2, 150 mM NaCl, 1% Triton X-100, 1 mM EDTA, and protease inhibitor cocktail (Roche, Basel, Switzerland)] and boiled for 5 min with an SDS sample buffer. Proteins were separated by 16% SDS-PAGE and transferred to a polyvinylidene difluoride membrane (Merck Millipore, Burlington, MA) by electroblotting. The blots were blocked for 30 min with 1% skimmed milk in PBS containing 0.1% Tween-20, and after incubation for 1 h with primary antibodies (1:5000 dilution), they were incubated for 1 h with appropriate HRP-conjugated secondary antibodies. The entire procedure was performed at room temperature. Chemiluminescence signals were visualized by means of the Immobilon Western Chemiluminescent HRP substrate (EMD Millipore, Burlington, MA) and detected with a chemiluminescence imager (ChemiDoc Touch; Bio-Rad, Hercules, CA).

GST–Rab11A(CA), GST–Rab11A(CN) and control GST alone were expressed in E. coli JM109 and purified with gluthathione–Sepharose beads (GE Healthcare) through a standard protocol. For GTP/GDP loading, 10 µg of GST–Rab11A(CA) or GST–Rab11A(CN) was incubated for 20 min at 4°C with 100 µl of 50 mM HEPES-KOH, pH 7.2, 150 mM NaCl, 2.5 mM EDTA, and 0.1% Triton X-100, and then with 1 µl each of 1 M MgCl₂ (final 10 mM) and 50 mM GTPγS (final 0.5 mM) or 100 mM GDP (final 1 mM). COS-7 cells (6-cm dish) transiently expressing T7-tagged Rab11-FIP2-C or RBD11 were lysed for 1 h at 4°C with 400 µl of 50 mM HEPES-KOH, pH 7.2, 250 mM NaCl, 1 mM MgCl₂, 1% Triton X-100, and 1×protease inhibitor cocktail (Roche). After centrifugation at 20,000 g for 10 min, the supernatant was recovered and incubated for 1 h at 4°C with anti-T7 tag-antibody-conjugated agarose beads (wet volume 30 µl). The beads coupled with T7-tagged proteins were washed three times with 400 µl of 50 mM HEPES-KOH, pH 7.2, 150 mM NaCl, 1 mM MgCl₂ and 0.1% Triton X-100 (washing buffer). The beads coupled with purified T7-tagged proteins were incubated for 1 h at 4°C with 100 µl of the solution containing GTPγS-loaded GST–Rab11A(CA) or GDP-loaded GST–Rab11A(CN) described above. After washing the beads with 400 µl of the washing buffer three times, proteins bound to the beads were analyzed by performing 15% SDS-PAGE and staining with Coomassie Brilliant Blue R-250.

The amino acid sequences of the C-terminal region of Rab11-FIPs (102 a.a. of Rab11-FIP1, 124 a.a. of Rab11-FP2, 100 a.a. of Rab11-FIP3, 99 a.a. of Rab11-FIP4 and 101 a.a. of Rab11-FIP5) were aligned by using the ClustalW software program (version 2.1; available at http://clustalw.ddbj.nig.ac.jp/top-e.html) set at the default parameters and their phylogenetic tree was drawn by the neighbor-joining method.

One-way ANOVA and Tukey's test or the two-tailed Student's unpaired t-test were used to perform the statistical analysis, and P<0.05 was used as the criterion for statistical significance.

We thank Drs Toshio Kitamura and Shoji Yamaoka for kindly donating materials, Kazuyasu Shoji for technical assistance, and members of the Fukuda laboratory for valuable discussions.

The peer review history is available online at https://journals.biologists.com/jcs/article-lookup/134/7/jcs257311/