Establishment and analysis of conditional Rab1- and Rab5-knockout cells using the auxin-inducible degron system

The Rab family of small GTPases plays important roles in membrane trafficking in all eukaryotes, and ∼60 different Rab isoforms have been identified in mammals. Each Rab isoform functions as a molecular switch by cycling between an active GTP-bound form and an inactive GDP-bound form. The active form localizes on the surface of a specific intracellular membrane where it recruits its specific binding partner, called a Rab effector, which promotes the budding, transport, tethering, docking and/or fusion of vesicles (reviewed in Stenmark, 2009; Hutagalung and Novick, 2011; Pfeffer, 2017; Homma et al., 2021). In the past few decades, dominant-active (or -negative) experiments using a constitutively active (or negative) Rab mutant and small interfering RNA (siRNA)-mediated knockdown experiments have often been performed to analyze the function of individual Rabs in cellular functions (reviewed in Fukuda, 2010). However, recent advances in genome editing technologies, specifically in CRISPR/Cas9 technology (reviewed in Hsu et al., 2014), have enabled us to generate knockout (KO) cells even at the cultured mammalian cell level and to assess the function of a target gene product under ‘complete null’ conditions. We recently established comprehensive Rab family KO Madin–Darby canine kidney (MDCK) cell lines by knocking out multiple Rab paralogs (for example, Rab3A–Rab3D) simultaneously; however, no Rab1-KO or Rab5-KO cells were obtained (Homma et al., 2019; Oguchi et al., 2020).

Two Rab1 paralogs (Rab1A and Rab1B) and three Rab5 paralogs (Rab5A, Rab5B and Rab5C), referred to hereafter collectively as Rab1 and Rab5, respectively, unless otherwise specified, are present in all vertebrates (Diekmann et al., 2011), and they appear to function redundantly in cell survival and/or growth (Homma et al., 2019), thereby precluding the establishment of corresponding KO cells. Although Rab1 and Rab5 are known to regulate endoplasmic reticulum (ER)-to-Golgi trafficking and early endosome trafficking (or formation), respectively (reviewed in Martinez and Goud, 1998; Langemeyer et al., 2018), whether these trafficking events are completely impaired in the absence of Rab1 or Rab5 has never been demonstrated experimentally. Thus, the development of corresponding conditional KO (CKO) cells could be expected to resolve this issue.

In order to assess the impact of complete Rab1 or Rab5 protein depletion on membrane trafficking, we for the first time established Rab1-CKO and Rab5-CKO cell lines, in which the endogenous Rab1 and Rab5 genes, respectively, were completely disrupted, by using the auxin-inducible protein degradation system. In brief, auxin-inducible degron (AID) technology enables rapid degradation of AID-tagged proteins of interest in an auxin-dependent manner (Nishimura et al., 2009): AID-tagged proteins (for example, AID–Rab5A in Rab5-CKO cells) are recognized and ubiquitylated by an F-box transport inhibitor response 1 (TIR1)-containing SCF ubiquitin ligase complex in an auxin-dependent manner, and then the ubiquitylated target proteins are rapidly degraded by proteasomes. We evaluated the Rab1-CKO and Rab5-CKO cells in relation to cell survival, cell growth, and organelle distribution and size at various time points after auxin exposure. Although the Rab1-CKO and Rab5-CKO cells had started to die by 18 h and 48 h, respectively, after auxin exposure, we were able to assess the effects of acute Rab1 depletion (6–12 h after auxin exposure) and Rab5 depletion (12–36 h after auxin exposure) on Golgi morphology and the number of early endosomes, respectively. The results showed that Rab1 and Rab5 are essential for the maintenance of Golgi morphology and for the formation of early endosomes, respectively, consistent with the findings in previous reports (Gorvel et al., 1991; Wilson et al., 1994; Haas et al., 2007; Zeigerer et al., 2012; Aizawa and Fukuda, 2015). We also analyzed the effect of Rab1 or Rab5 depletion on the morphology and distribution of various organelles and discovered a previously unknown function of Rab1 in regulating the spatial distribution of early endosomes. We also observed an unaltered number of lysosomes in Rab5-depleted cells. Based on our findings, we discuss the usefulness of Rab1-CKO and Rab5-CKO cells as valuable tools for analyzing the functions of Rab1 and Rab5, respectively, in various cellular events under virtually complete null conditions.

We previously demonstrated that Rab1 and Rab5 are essential for the survival and the growth, respectively, of MDCK cells. Since it was impossible to obtain KO cell lines for these genes using conventional CRISPR/Cas9-mediated genome editing technology (Homma et al., 2019), we sought to generate CKO cell lines using the auxin-inducible protein degradation system (Nishimura et al., 2009) as a means of investigating the functions of Rab1 and Rab5. To apply this system, AID-tagged Rab1B or Rab5A was stably and exogenously expressed in MDCK cells prior to completely knocking out the endogenous Rab1 (Rab1A and Rab1B) or Rab5 (Rab5A, Rab5B and Rab5C) genes, and then tetracycline-inducible TIR1 was stably introduced (see Figs S1 and S2A for details), thereby allowing doxycycline (Dox)- and auxin-dependent rapid depletion of exogenous AID–Rab1B and AID–Rab5A proteins (Fig. 1A,B). For simplicity, we herein refer to the cells obtained as Rab1-CKO cells and Rab5-CKO cells. Such temporally restricted expression of TIR1 was necessary because this protein exhibits high basal ubiquitin ligase activity that would allow degradation of AID–Rab1B (or AID–Rab5A) even in the absence of auxin, as reported previously (Yesbolatova et al., 2019).

To confirm that endogenous Rab1 and Rab5 proteins were actually depleted, we first performed immunoblot analyses using specific antibodies. As anticipated, no endogenous Rab1 and Rab5 bands were observed in Rab1-CKO and Rab5-CKO cells, respectively (Fig. 1B, arrowheads), and exogenous FLAG-tagged AID–Rab1B and AID–Rab5A proteins were detected instead (Fig. 1B, arrows). These AID–Rab bands had completely disappeared 24 h after simultaneous exposure to 1 μg/ml Dox and 500 µM auxin (simply referred to as auxin exposure below, unless otherwise specified) (Fig. 1B, third and fourth lanes). Since no AID–Rab protein band was detected by immunoblotting, even after prolonged exposure of the blots, AID–Rab proteins should be completely degraded by 24 h after auxin exposure under our experimental conditions. Moreover, FLAG–AID–Rab1B and FLAG–AID–Rab5A colocalized well with two different Golgi markers (GM130, also known as GOLGA2, and GalNT2) and with early endosome markers (EEA1 and VPS35), respectively (Fig. S2B), suggesting that N-terminal AID-tagging had no effect on localization of the Rab proteins. We then evaluated the degradation efficiency of FLAG–AID–Rab1B and FLAG–AID–Rab5A after auxin exposure at the times indicated in Fig. 1C. The results of the immunoblot analysis showed that almost all of the FLAG–AID–Rab1B protein in the Rab1-CKO cells was degraded within 6 h of auxin exposure, whereas exposure to auxin for at least 12 h was required for degradation of FLAG–AID–Rab5A protein in the Rab5-CKO cells (Fig. 1C). Similar results were obtained in immunofluorescence analyses of Rab1-CKO and Rab5-CKO cells (Fig. 1D), although a few Rab1-CKO and Rab5-CKO cells were still positive for anti-FLAG tag antibody even after auxin exposure for 6 h and 12 h, respectively.

Because Rab1 and Rab5 are essential for cell survival and/or growth (Homma et al., 2019), after being exposed to auxin the Rab1-CKO and Rab5-CKO cells we established would eventually undergo cell death and/or cell growth arrest. Thus, in order to functionally analyze Rab1 or Rab5 without any side effects from cell death, it was important to identify the periods between the disappearance of AID–Rab1B and AID–Rab5A proteins and the onset of cell death or growth arrest of the CKO cells. To do so, we first investigated the time-dependent changes in the morphology and growth rate of Rab1-CKO and Rab5-CKO cells after auxin exposure (Fig. 2A,B). To visualize cell morphology, CKO cells were fixed at the times indicated in Fig. 2A, and their plasma membrane was stained with Alexa Fluor-conjugated wheat germ agglutinin (WGA). Some of the Rab1-CKO cells shrank and had become small and rounded by 18 h after auxin exposure (Fig. 2A, yellow arrowheads), and the number of cells subsequently decreased dramatically (Fig. 2A, far right panel in the top row). Consistent with the results of the morphological analysis, we confirmed that the Rab1-CKO cells appeared to grow normally until 12 h after auxin exposure but significantly decreased in number thereafter (Fig. 2B, left). These findings led us to hypothesize that apoptosis is actively induced in Rab1-CKO cells 12 h after auxin exposure, and we detected activation of caspase-3 and/or caspase-7, which are known to be activated in the executive phase of apoptosis (Boatright and Salvesen, 2003), in almost all of the dying Rab1-CKO cells 36 h after auxin exposure (Fig. 2C; Fig. S3A). Conversely, some of the Rab5-CKO cells had begun to elongate 36 h after auxin exposure (Fig. 2A, red asterisks), and dendrite-like structures were observed after 48 h of auxin exposure (Fig. 2A, red arrows). Consistent with the time of onset of morphological changes in the Rab5-CKO cells, the number of cells gradually decreased thereafter (Fig. 2B, right panel).

Based on all of the above findings taken together, we concluded that the optimal time periods for functional assays of Rab1-CKO and Rab5-CKO cells are 6–12 h and 12–36 h, respectively, after auxin exposure (Fig. 2D, green bars). During these periods, the AID–Rab1B and AID–Rab5A proteins were almost completely depleted (i.e. virtually complete null conditions), and the CKO cells were able to grow without any changes in their morphology. Based on these criteria, we decided to perform the subsequent functional assays of Rab1-CKO and Rab5-CKO cells principally at 6 h and 24 h, respectively, after auxin exposure.

In our attempt to characterize Rab1-KO or Rab5-KO phenotypes by using the Rab1-CKO or Rab5-CKO cell lines we had established, we first focused on widely recognized Rab1 and Rab5 functions. Rab1 is known to mainly localize at the Golgi and to be required for maintenance of Golgi morphology (Wilson et al., 1994; Alvarez et al., 2003; Haas et al., 2007; Aizawa and Fukuda, 2015), and it also regulates ER-to-Golgi vesicular trafficking (Schwaninger et al., 1992; Martinez et al., 2016; Westrate et al., 2020). To evaluate the effect of acute depletion of Rab1 protein on Golgi morphology, we immunocytochemically stained Rab1-CKO cells and control wild-type (WT) cells for GM130 (cis-Golgi marker) and GalNT2 (trans-Golgi marker). In the WT cells, the cis- and trans-Golgi marker signals were observed in close proximity to each other (indicating the presence of Golgi stacks), and they were clustered in the perinuclear region, where they formed a compact ribbon structure (Fig. 3A, left panels). In the Rab1-CKO cells, both Golgi markers were also observed in the perinuclear region under resting conditions (i.e. in the absence of auxin), but some signals were clearly dispersed in the cytoplasm (Fig. 3A, middle panels). This weak Golgi dispersion phenotype is likely to be attributable to the lack of Rab1A protein in the Rab1-CKO cells (Fig. 1B), because re-expression of EGFP–Rab1A in Rab1A-KO cells rescued this phenotype (data not shown). However, it is noteworthy that both cis- and trans-Golgi markers were still observed in close proximity to each other even in the dispersed Golgi, strongly suggesting that Golgi stacks themselves had been retained in the Rab1-CKO cells. In contrast, the auxin-exposed Rab1-CKO cells showed more prominent dispersion of the Golgi, and clear separation between the cis- and trans-Golgi markers was observed (Fig. 3A, right panels), indicating the loss of cisternal stacks of the Golgi. These results indicate that Rab1 plays an important role in maintaining not only the perinuclear Golgi ribbon structure in mammalian cells but also the cis- and trans-Golgi cisternal stacks.

To assess the functional involvement of Rab1 in ER-to-Golgi trafficking, we used a signal sequence (ss)-fused EGFP with an artificial oligomerization domain (FM4) and a transmembrane domain (TMD), named ssEGFP–FM4–TMD (Homma et al., 2019), as a model transmembrane cargo and stably expressed it in both WT and Rab1-CKO cells. In the steady state, ssEGFP–FM4–TMD is retained in the ER, because it forms large aggregates via its FM4 domain, but when treated with D/D solubilizer to disassociate the FM4-domain-mediated oligomers, ssEGFP–FM4–TMD starts to be transported from the ER to the Golgi (∼30 min), and ultimately to the plasma membrane (∼60 min) (Homma et al., 2019). In Rab1-CKO cells not exposed to auxin, ssEGFP–FM4–TMD had reached the Golgi (as determined by colocalization with GM130 and GalNT2) 30 min after D/D solubilizer treatment, the same as in the WT cells (Fig. 3B; Fig. S3B, lower left and middle panels), whereas almost all of the ssEGFP–FM4–TMD signals in the auxin-exposed Rab1-CKO cells remained within the reticular ER structure even in the presence of the D/D solubilizer (Fig. 3B; Fig. S3B, lower right panels). Intriguingly, a few ssEGFP–FM4–TMD signals were also observed in the cis-Golgi compartment (Fig. 3B, lower right panel), but none or hardly any were detected in the trans-Golgi compartment (Fig. S3B, lower right panel). Thus, acute depletion of Rab1 protein severely inhibited the ER-to-Golgi trafficking of secretory cargos and disrupted cis-to-trans trafficking within the Golgi.

To determine whether transport of ssEGFP–FM4–TMD to the plasma membrane was completely abrogated under Rab1-depleted conditions, we specifically visualized cell-surface-localized ssEGFP–FM4–TMD using anti-GFP antibody without permeabilizing the cells, conditions under which the antibody is unable to recognize intracellular ssEGFP–FM4–TMD (Homma et al., 2019). In the Rab1-CKO cells not exposed to auxin, no signals were observed on the cell surface, whereas 60 min after D/D solubilizer treatment strong ssEGFP–FM4–TMD signals were detected over the entire cell surface, the same as was observed for the WT cells (Fig. 3C, left and middle panels). In contrast, no ssEGFP–FM4–TMD signals were detected on the cell surface of the auxin-exposed Rab1-CKO cells after D/D solubilizer treatment (Fig. 3C, right panels), confirming that Rab1 is essential for the transport of secretory cargos from the ER to the plasma membrane through the Golgi.

Since the Golgi is functionally linked to other organelles, including the ER and endosomes via membrane trafficking, and to microtubules, we next investigated the impact of acute Rab1 depletion on the morphology and distribution of other organelles and the cytoskeleton using immunocytochemistry (Fig. 4A). Although neither the morphology nor the distribution of the ER, mitochondria, peroxisomes, microtubules or actin filaments appeared to be affected by Rab1 protein depletion (Fig. 4A, right panels), we were surprised to find that early endosomes had greatly accumulated in the perinuclear region of the Rab1-CKO cells after auxin exposure (Fig. 4A, yellow arrowheads in EEA1 panels; Fig. S3C for z-stacked images). We also observed perinuclear clustering of recycling endosomes (Fig. 4A, yellow arrows in TfR and Rab11 panels) and weak accumulation of late endosomes and lysosomes in the perinuclear region of the auxin-exposed Rab1-CKO cells (Fig. 4A, LBPA and LAMP1 panels). Taken together, these results suggest that Rab1 is involved in the subcellular distribution of early endosomes, recycling endosomes, late endosomes and lysosomes in addition to maintenance of the compacted Golgi in the perinuclear region.

Since Rab1 depletion affected endosomal and lysosomal localization, we next evaluated its effect on the uptake, recycling and/or degradation of representative endocytic cargos. First, we assessed the transferrin (Tf) recycling efficiency in Rab1-CKO cells by performing a pulse-chase analysis using Alexa Fluor 594-conjugated Tf (594–Tf). Under control conditions, internalized 594–Tf signals were localized around the perinuclear region and then gradually decreased as they returned to the plasma membrane (Fig. 4B, top row). In contrast, under Rab1-depleted conditions, the internalized 594–Tf signals accumulated to a greater extent in the perinuclear region (Fig. 4B, bottom row), especially at 5 min and 20 min of the chase, corresponding to the early endosome accumulation phenotype of Rab1-CKO cells. The results of a quantitative analysis confirmed significantly delayed Tf recycling efficiency in Rab1-depleted cells in comparison with that of the control cells (Fig. 4C). We also assessed the degradation efficiency of epidermal growth factor (EGF) receptor (EGFR) in Rab1-CKO cells. Although a significant amount of immature EGFR (detected as a lower molecular weight EGFR band) was observed in the Rab1-depleted cells (asterisk in Fig. 4D), presumably due to the blocking of ER-to-Golgi trafficking in the absence of Rab1 (Fig. 3B; Fig. S3B), there was no significant difference between the control and Rab1-depleted cells in the degradation efficiency of active EGFR. Taken together, these results suggest that Rab1-mediated proper early (or recycling) endosome distribution is required for efficient recycling of endocytic cargos (such as Tf) but is dispensable for degradation of endocytic cargos (such as EGFR).

Rab5 mainly localizes at early endosomes (Chavrier et al., 1990), and it is known to regulate the homotypic fusion of endocytic vesicles and early endosomes (Gorvel et al., 1991) and to orchestrate endosome maturation steps (Rink et al., 2005; reviewed in Stenmark, 2009). Moreover, knockdown of all three Rab5 isoforms in mouse liver tissue has been reported to result in loss of early endosomes, late endosomes and lysosomes (Zeigerer et al., 2012). To determine whether Rab5 is actually essential for the formation of these organelles even in cultured mammalian cells, we observed the changes in the number of early endosomes, late endosomes and lysosomes in Rab5-CKO cells every 12 h after auxin exposure by immunofluorescence labeling of two different early endosome markers (EEA1 and VPS35) (Jimenez-Orgaz et al., 2018), lysobisphosphatidic acid (LBPA), and LAMP1, respectively (Fig. 5A; Fig. S4A), and quantified their numbers per cell area (Fig. 5B) or per cell (Fig. S4B). Both EEA1-postive and VPS35-positive punctate signals had almost completely disappeared 12 h after auxin exposure, which clearly coincided with the disappearance of FLAG–AID–Rab5A protein (Fig. 1C,D). Late endosomes also disappeared, but more slowly than the early endosomes disappeared (Fig. 5B; Fig. S4B); some LBPA-positive punctate signals remained even 12 h after auxin exposure, but they had almost completely disappeared after auxin exposure for an additional 12 h. Unexpectedly, however, the number of LAMP1-positive punctate signals per cell area (or per cell) was unaltered (or had slightly increased) even 36 h after auxin exposure (Fig. 5B; Fig. S4B). The decreased numbers of EEA1-positive and VPS35-positive puncta are unlikely to be attributable to reduced protein expression levels (e.g. degradation of EEA1 and VPS35), because the levels of EEA1, VPS35 and LAMP1 protein expression were unaltered in Rab5-CKO cells irrespective of the presence of auxin (Fig. 5C, middle three panels). However, the protein expression level of low-density lipoprotein receptor (LDLR) dramatically increased in the auxin-exposed Rab5-CKO cells (Fig. 5C, second panel from the bottom), a finding that is consistent with a previous report on the results of Rab5 knockdown in mouse liver tissue (Zeigerer et al., 2012). As has been suggested by the previous report, some compensatory mechanism may have been upregulated to overcome the reduction in LDL uptake within 24 h after Rab5 protein depletion.

Since Rab5 is essential for maintaining early and late endosomal structures, we also investigated the impact of acute Rab5 depletion on the morphology and distribution of other organelles and the cytoskeleton by immunocytochemistry (Fig. S4C). At 24 h after auxin exposure, the Golgi ribbon structure in Rab5-CKO cells had become more compact than in WT cells and Rab5-CKO cells not exposed to auxin. In contrast, however, the localization and distribution of other organelles, including recycling endosomes, ER and peroxisomes, and of the cytoskeleton (both microtubules and actin filaments) in Rab5-CKO cells were unaffected by auxin. Intriguingly, no noticeable changes were observed in the localization or distribution of recycling endosomes, which are closely related to the endocytic pathway, even after exposing Rab5-CKO cells to auxin, suggesting that Rab5 does not directly regulate the localization or distribution of recycling endosomes.

Since Rab5 has previously been shown to be involved in both fluid-phase uptake and cargo uptake by receptor-mediated endocytosis in experiments performed by dominant-negative approaches using a constitutively negative Rab5 mutant (Bucci et al., 1992; Stenmark et al., 1994), we investigated the effect of acute Rab5 protein depletion on endocytic uptake in auxin-exposed Rab5-CKO cells. We first focused on internalization of EGF as a model of the receptor-mediated endocytic pathway. Many Alexa Fluor 488–EGF (488–EGF)-positive punctate signals were detected in Rab5-CKO cells not exposed to auxin, indicating that the internalized 488–EGF had reached the early endosomes and/or late endosomes, the same as in WT cells. In contrast, only a small number of punctate 488–EGF signals was observed in the auxin-exposed Rab5-CKO cells, and the small dot-like signals in the background had instead increased (Fig. 6A). Quantification of the number of 488–EGF punctate signals greater than 0.2 µm in diameter confirmed a significant decrease in 488–EGF puncta in the auxin-exposed Rab5-CKO cells (Fig. 6B).

EGF is recognized by EGFR, and ligand-bound EGFR is internalized by endocytosis and ultimately degraded in lysosomes (reviewed in Bakker et al., 2017). Because of the dramatic reduction in EGF punctate signals in the auxin-exposed Rab5-CKO cells, we assumed that EGFR degradation would not occur even after EGF exposure. As anticipated, the results of EGFR degradation assays in the auxin-exposed Rab5-CKO cells showed that the EGFR protein level had slightly decreased 1 h after addition of EGF, but it subsequently remained unaltered until 4 h after auxin exposure (Fig. 6C, right five lanes). In contrast, however, the EGFR protein level rapidly decreased in a time-dependent manner after the addition of EGF under control conditions (i.e. in WT cells and Rab5-CKO cells not exposed to auxin; Fig. 6C, left five lanes and middle five lanes, respectively). The impaired EGFR degradation observed in the auxin-exposed Rab5-CKO cells was unlikely to have been caused by lysosomal dysfunction, because highly acidic LysoTracker-positive lysosomal compartments and normal acid protease activity (as monitored using Magic Red) were still observed in Rab5-CKO cells even after 24 h of auxin exposure (Fig. 6D). Thus, the impaired EGFR degradation in Rab5-deficient cells was likely to have been directly attributable to impaired EGFR trafficking to lysosomes.

Finally, we assessed the effect of acute Rab5 protein depletion on the internalization of Tf (another cargo of receptor-mediated endocytosis) and BSA (a cargo of fluid-phase endocytosis), whose internalization is known to be regulated by Rab5 (Stenmark et al., 1994; Li et al., 1994; Bucci et al., 1995). Consistent with the EGF uptake results described above, punctate signals (>0.2 µm diameter) of both Tf and BSA were also significantly decreased in auxin-exposed Rab5-CKO cells (Fig. 6E,F). These results taken together indicate that Rab5 is an essential component for the proper maturation of endosomes in both the receptor-mediated endocytic pathway and the fluid-phase endocytic pathway.

We have previously demonstrated that both Rab1 and Rab5 are required for cell survival and/or growth, and thus Rab1-KO and Rab5-KO cells cannot be obtained by conventional CRISPR/Cas9-mediated KO technology (Homma et al., 2019). In the present study, we generated Rab1-CKO and Rab5-CKO MDCK cells, in which the endogenous Rab1 genes (Rab1A and Rab1B) and Rab5 genes (Rab5A, Rab5B and Rab5C) were completely disrupted and AID-tagged Rab1B and Rab5A, respectively, were stably expressed (Fig. 1). After exposing these cells to auxin, the AID-tagged Rabs were degraded within 12 h, and we were able to assess the function of Rab1 and Rab5 under virtually null Rab1 and Rab5 protein conditions, although the Rab1-CKO and Rab5-CKO cells eventually died (>18 h and >48 h, respectively, after auxin exposure; Fig. 2). Using these CKO cells enabled us to confirm functions of Rab1 and Rab5 that have been demonstrated previously in knockdown and/or dominant-negative experiments, and they also enabled the discovery of previously unknown phenotypes of Rab1- and Rab5-depleted cells (discussed below).

After acute depletion of Rab1 protein, the Golgi was severely fragmented and dispersed into the cytoplasm, and the cisternal stack structures (cis-to-trans) were completely disrupted (Fig. 3A). Disruption of the Golgi ribbon structure (i.e. Golgi fragmentation) by expressing dominant-negative mutants of Rab1 or by siRNA-mediated Rab1 knockdown has previously been reported (Alvarez et al., 2003; Monetta et al., 2007). Since the interaction between Rab1B and GBF1 is involved in COP1 recruitment, disruption of this interaction is likely to be the cause of the Golgi fragmentation phenotype in Rab1-depleted cells. Given that Rab1 is also known to interact with several Golgi-resident tethering factors as Rab1 effectors, including p115 (also known as USO1), GM130 and golgin84 (also known as GOLGA5) (Allan et al., 2000; Moyer et al., 2001; Weide et al., 2001; Diao et al., 2003), the disruption of the Golgi stacks in Rab1-depleted cells may have been caused by impairment of the interactions between Rab1 and these coiled-coil-domain-containing tethering factors on the Golgi membrane. However, since there have been no reports on their function in the Golgi cisternal stack structures, other unknown Rab1 effectors may be involved in the maintenance of the Golgi stack structures. Alternatively, known Rab1 effectors, such as p115, GM130 and golgin84, might function redundantly in the stack structures. Further research will be necessary to identify the molecular mechanism by which Rab1 regulates the Golgi stack structures, and our Rab1-CKO cells should be useful for investigating this mechanism in the future.

In addition to the functions of Rab1 in the Golgi stack and ribbon structures, Rab1 is also important for the ER-to-Golgi trafficking of secretory proteins (Fig. 3B,C). In the absence of Rab1 protein, the model secretory cargo ssEGFP–FM4–TMD did not reach the trans Golgi or, consequently, the cell surface. Intriguingly, however, although most ssEGFP–FM4–TMD signals were retained in the ER, some ssEGFP–FM4–TMD signals were also detected on dispersed cis-Golgi compartments, suggesting the presence of a Rab1-indepenent ER-to-Golgi trafficking mechanism. Alternatively, the ssEGFP–FM4–TMD signals on the cis-Golgi may merely be attributable to incomplete degradation of AID–Rab1B protein after auxin exposure.

Moreover, we discovered a novel phenotype under Rab1-depleted conditions: perinuclear accumulation of early endosomes. Since nocodazole, a microtubule-depolymerizing reagent, completely reversed this phenotype (data not shown), the perinuclear accumulation of early endosomes occurs in a microtubule-dependent manner. Whether this phenotype is caused by a direct or indirect effect of inhibited ER-to-Golgi transport upon Rab1 depletion is currently unknown. However, since it has previously been reported that Rab1 also localizes at early endosomes and regulates the motility of early endocytic vesicles (Mukhopadhyay et al., 2011; Sandin et al., 2012), we hypothesize that Rab1 directly regulates endosomal distribution. There are two possible hypotheses to explain the perinuclear clustering of early endosomes in the absence of Rab1: (1) Rab1 is involved in the promotion of anterograde transport of early endosomes along microtubules, and (2) Rab1 inhibits the microtubule-dependent retrograde transport of early endosomes. We favor the former hypothesis, because Rab1A has been shown to interact with kinesin-1 via SKIP (also known as PLEKHM2) and to regulate anterograde melanosome transport on microtubules in melanocytes, although in living cells, the majority of Rab1A is present in the Golgi, and only a small amount of Rab1A is found on melanosomes (melanosome-localized Rab1A is not evident in fixed cells) (Ishida et al., 2012, 2015). Thus, future investigation of Rab1 localization on early endosomes in living cells will be necessary, and the molecular mechanism by which Rab1 regulates the transport of early endosomes is the next important issue that must be addressed in a future study. We also observed mild perinuclear accumulation of recycling endosomes, late endosomes and lysosomes, but this phenotype is most likely due to an indirect effect of perinuclear early endosome accumulation, because early endosomes are directly or indirectly connected to these organelles. Perinuclear endosome accumulation is likely to affect the anterograde transport of endocytic cargos to the plasma membrane, and we observed delayed Tf recycling in Rab1-depleted cells (Fig. 4C).

Using Rab5-depleted cells enabled us to confirm that Rab5 is essential for the formation of early endosomes and late endosomes in cultured mammalian cells (Fig. 5); however, acute depletion of Rab5 protein did not affect the number of lysosomes. Several models of lysosome biogenesis have been proposed (reviewed in Trivedi et al., 2020), and our findings support the model in which lysosomes are not directly derived from endosomes that form and mature after endocytosis. In contrast to our findings, Rab5 knockdown in mouse liver tissue has previously been shown to cause loss of lysosomes as well as loss of early and late endosomes (Zeigerer et al., 2012). Such an apparent discrepancy may be explained by the difference between the assay times: lysosome numbers were analyzed 36 h after auxin stimulation in this study as opposed to 4–5 days after siRNA administration in the previous study. Thus, we cannot rule out the possibility that more time (>36 h) is needed for the number of lysosomes to decrease under Rab5-depleted conditions, although the Rab5-depleted cells had started to die by 48 h after auxin exposure (Fig. 2B), which precluded an analysis in our experimental paradigm. However, since internalized endocytic cargos generally reach lysosomes within several hours after internalization, we consider this possibility to be unlikely. Another possible explanation is that lysosome biogenesis differs among cell types, for example between mouse hepatocytes and MDCK cells. Further work will be necessary to understand the mechanism responsible for maintaining lysosomal numbers in Rab5-depleted MDCK cells.

We also observed a dramatic reduction in the punctate signals of internalized cargos from receptor-mediated endocytosis (EGF and Tf) and fluid-phase endocytosis (BSA) in Rab5-depleted cells (Fig. 6), suggesting that Rab5 is essential for internalization of endocytic cargos and/or endosome maturation. However, there have been several contradictory opinions as to whether Rab5 promotes internalization of endocytic cargos (Bucci et al., 1992; Stenmark et al., 1994; McLauchlan et al., 1998; Barbieri et al., 2000; Dinneen and Ceresa, 2004; Chen et al., 2009), and a more detailed analysis of our Rab5-CKO cells in the future may resolve this issue.

It is now widely believed that several endocytic pathways other than the clathrin-dependent pathway are present in single cells (reviewed in Sigismund et al., 2021), and some of them presumably function in a Rab5-independent manner. It has recently been reported that endo-lysosomal maturation for internalized cell-penetrating peptides, homeoproteins and polyamines depends mainly on Rab14 rather than on Rab5 (Trofimenko et al., 2021). Thus, the Rab5-CKO cells established in our study should serve as an ideal tool for analyzing such Rab5-independent endocytic pathways.

We have established Rab1-CKO and Rab5-CKO MDCK cells to provide tools for conducting a functional analysis of Rab1 and Rab5 under virtually complete KO conditions between 6–12 h and 12–36 h, respectively, after auxin exposure. Although MDCK cells are capable of forming an epithelial cell sheet or a three-dimensional cyst, we were unable to investigate the epithelial morphogenesis of these CKO cells, because it takes more than 48 h to form epithelial cell structures (Mrozowska and Fukuda, 2016). Thus, one drawback of using our tools is the limited time period for conducting analyses because of the resulting eventual cell death. However, these CKO cells should be useful for analyzing various physiological phenomena that occur within 12 h (in the case of Rab1) or 36 h (in the case of Rab5) after auxin exposure. Such phenomena include autophagosome formation and mTORC1 regulation, in which Rab1 and Rab5 are thought to be involved (reviewed in Ao et al., 2014; Jin et al., 2021). Rab5-CKO cells should also be useful for analyzing early endocytic events in various endocytic pathways. Future investigation of these cellular phenomena in Rab1-CKO and Rab5-CKO cells should shed light on the novel functions of Rab1 and Rab5, respectively, and the mechanisms underlying them.

Rabbit polyclonal antibodies against Rab1A, Rab1B, Rab5A and Rab5B/C were raised against respective glutathione S-transferase (GST)-tagged Rabs as an antigen and purified from antiserum as described previously (Aizawa and Fukuda, 2015; Mrozowska and Fukuda, 2016). The following antibodies were obtained commercially: anti-FLAG M2 mouse monoclonal antibody (#F1804), anti-GalNT2 rabbit polyclonal antibody (#HPA011222) and anti-β-tubulin mouse monoclonal antibody (#T4026) from Sigma-Aldrich (St Louis, MO, USA); anti-LBPA mouse monoclonal antibody (#Z-PLBPA) from Echelon Biosciences (Salt Lake City, UT, USA); anti-GFP rabbit polyclonal antibody (#598) from MBL (Tokyo, Japan); anti-Myc (9E10) mouse monoclonal antibody (#sc-40) from Santa Cruz Biotechnology (Dallas, TX, USA); anti-GM130 mouse monoclonal antibody (#610823) and anti-EEA1 mouse monoclonal antibody (#610456) from BD Biosciences (San Jose, CA, USA); anti-PEX14 rabbit polyclonal antibody (#10594-1-AP) from ProteinTech (Chicago, IL, USA); anti-NogoA rabbit polyclonal antibody (#AHP1799) from Bio-Rad (Hercules, CA, USA); anti-LDLR rabbit monoclonal antibody (#NB110-57162) from Novus Biologicals (Centennial, CO, USA); anti-β-actin mouse monoclonal antibody (#G043) from Applied Biological Materials (Richmond, BC, Canada); anti-VPS35 goat polyclonal antibody (#ab10099), anti-LAMP1 rabbit polyclonal antibody (#ab24170), and anti-mitochondria (MTC02) mouse monoclonal antibody (#ab3298) from Abcam (Cambridge, UK); anti-TfR mouse monoclonal antibody (#13-6800), anti-Rab11 rabbit polyclonal antibody (#71-5300), anti-LAMP2 mouse monoclonal antibody (#MA5-28269) and Alexa Fluor-conjugated secondary antibodies (#A21052, #A32723, #A32727, #A32731, and #A32732) from Thermo Fisher Scientific (Waltham, MA, USA). Other reagents used in this study were also obtained commercially: Alexa Fluor 568 Phalloidin, Alexa Fluor 594-conjugated wheat germ agglutinin (WGA; 1:1800 dilution for 1 h), Alexa Fluor 488 EGF complex (488–EGF), Alexa Fluor 594-conjugated transferrin (594–Tf), Alexa Fluor 488-conjugated bovine serum albumin (488–BSA), LysoTracker Red DND-99 (10 nM) and CellEvent^TM caspase-3/7 green detection reagent from Thermo Fisher Scientific; 3-indoleacetic acid (auxin) from Sigma-Aldrich; doxycycline from Fujifilm Wako Pure Chemical (Osaka, Japan); D/D solubilizer from Takara Bio (Shiga, Japan); and Cathepsin B Assay from Immunochemistry Technologies (Bloomington, MN, USA; 1:1000 dilution).

pMK106 plasmids (expression clone for an auxin-based degron system; Nishimura et al., 2009) were obtained from the RIKEN BioResource Research Center (#RDB08469). AID and Oryza sativa TIR1 (OsTIR1) cDNAs were amplified by using specific oligonucleotides with appropriate restriction enzyme sites and subcloned into the pMRX-IRES-bsr-FLAG vector (Saitoh et al., 2003; Homma et al., 2019) and pRetroX-TetOne-Puro vector (Clontech/Takara Bio, #634307), respectively, by standard molecular biology techniques (see Fig. 1A for details). For efficient subcloning, internal BglII and NotI sites of OsTIR1 were mutated by site-directed mutagenesis without altering their amino acid sequences. The mouse Rab1B (Itoh et al., 2006) and Rab5A cDNAs (Kuroda et al., 2002) were subcloned into the pMRX-IRES-bsr-FLAG-AID vector described above (named pMRX-IRES-puro-FLAG-AID-Rab1B and -Rab5A, respectively). The single guide RNA/Cas9-encoding vectors for Rab1A, Rab1B, Rab5A, Rab5B and Rab5C were prepared as described previously (Homma et al., 2019). The pMRX plasmids constructed in this study and pMRX-bsr-ssEGFP-FM4-TMD (Homma et al., 2019) have been deposited in the RIKEN BioResource Research Center (https://dnaconda.riken.jp/search/depositor/dep005893.html; #RDB19429–19431 and #RDB17365). pLp-VSVG was obtained from Thermo Fisher Scientific.

WT and mutant (Rab1B-KO and Rab5A/Rab5B-KO) Madin–Darby canine kidney (MDCK) strain II cells were prepared as described previously (Homma et al., 2019) (RIKEN BioResource Research Center, #RCB5148, #RCB5100 and #RCB5104, respectively). To generate Rab1-CKO cells, Rab1B-KO cells (Rab1A: WT/−2 bp; and Rab1B: +1 bp/−86 bp) were first established (step I) as described previously (Homma et al., 2019). This cell line was infected with retroviruses expressing FLAG–AID–Rab1B without drug selection (step II) and then transfected with sgRNA/Cas9-encoding plasmids targeting Rab1A (step III). The cells were selected in 2 μg/ml puromycin, subcloned by limiting dilution, and then sequenced and immunoblotted (Kinoshita et al., 2021) to identify clones that expressed FLAG–AID–Rab1B and had lost endogenous Rab1A expression. Finally, Tet-inducible OsTIR1 was retrovirally introduced into Rab1A/Rab1B-KO cells expressing FLAG–AID–Rab1B without drug selection, and the cells were subcloned by limiting dilution to isolate highly Dox-responsive clones (step IV) (see also the scheme for generating Rab1-CKO cells in Fig. S1A). Similarly, in order to generate Rab5-CKO cells, we first established Rab5A/Rab5B-KO cells (Rab5A: +1 bp; and Rab5B: +1 bp/−5 bp) (step I). The Rab5A/Rab5B-KO cell line was then infected with retroviruses expressing FLAG–AID–Rab5A without drug selection (step II) and transfected with sgRNA/Cas9-encoding plasmids targeting Rab5C (step III). The cells were selected in 2 μg/ml puromycin, subcloned by limiting dilution, and then sequenced and immunoblotted (Kinoshita et al., 2021) to identify clones that expressed FLAG–AID–Rab5A and had lost endogenous Rab5C expression. Finally, Tet-inducible OsTIR1 was retrovirally introduced into Rab5A/Rab5B/Rab5C-KO cells expressing FLAG–AID–Rab5A without drug selection, and the cells were subcloned by limiting dilution to isolate highly Dox-responsive clones (step IV) (see also the scheme for generating Rab5-CKO cells in Fig. S1B). Rab1-CKO and Rab5-CKO cells established in this study are also available from the RIKEN BioResource Research Center (#RCB5389 and #RCB5390, respectively). The WT and mutant MDCK cells were cultured at 37˚C under 5% CO₂ in DMEM (Fujifilm Wako Pure Chemical) supplemented with 10% fetal bovine serum (FBS), 100 U/ml penicillin G (Meiji Seika Pharma, Tokyo, Japan) and 100 μg/ml streptomycin (Meiji Seika Pharma). For retrovirus production, Plat-E cells, a kind gift from Dr Toshio Kitamura (The University of Tokyo, Tokyo, Japan), were cultured on a poly-L-lysine-coated culture plate under the same culture conditions as the MDCK cells.

Plat-E cells (∼80% confluent in a 3.5-cm dish) were transfected with 2 μg of each pMRX plasmid and 1 μg of pLp-VSVG in 2 ml culture medium using Lipofectamine 2000 (Thermo Fisher Scientific). One day after transfection, the medium was replaced with fresh medium, and the transfected cells were incubated for an additional day. The medium was then collected and filtered through a Millex-GV 0.45 μm syringe filter (Merck, Darmstadt, Germany). For virus infection, the retrovirus-containing medium was added to the culture dish of MDCK cells, and the cells were cultured for 24 h in the presence of 8 μg/ml polybrene.

For immunostaining, cells were fixed with 4% paraformaldehyde (PFA) for 10 min, permeabilized with 50 μg/ml digitonin in phosphate-buffered saline (PBS) for 5 min or in 0.3% Triton X-100 in PBS for 1 min (for GalNT2 and mitochondria staining), and then blocked with 3% bovine serum albumin (BSA) in PBS for 15 min. The permeabilized cells were incubated with primary antibodies in PBS containing 3% BSA (3% BSA/PBS) for 1 h, washed with PBS three times, and then incubated with appropriate Alexa Fluor-conjugated secondary antibodies and DAPI in 3% BSA/PBS for 1 h. All procedures were performed at room temperature. Primary antibodies to detect the following proteins and organelles were used at the dilutions indicated: FLAG (1:3500 dilution), Rab1B (1:300 dilution), Rab5A (1:300 dilution), GM130 (1:300 dilution), GalNT2 (1:300 dilution), GFP (1:2000 dilution), EEA1 (1:200 dilution), VPS35 (1:300 dilution), LBPA (1:500 dilution), TfR (1:500 dilution), Rab11 (1:500 dilution), LAMP1 (1:500 dilution), LAMP2 (1:1000 dilution), NogoA (1:300 dilution), mitochondria (MTC02; 1:500 dilution), PEX14 (1:300 dilution) and β-tubulin (1:500 dilution). Fluorescence images were obtained with a confocal laser-scanning microscope (Fluoview 1000; Olympus, Tokyo, Japan) through a Plan-Apochromat 60×/1.35 oil immersion objective lens or U-Plan-S-Apochromat 100×/1.45 oil immersion objective lens and Fluoview software (Olympus). Unless otherwise specified, fluorescence images are single-section images obtained with the confocal laser-scanning microscope.

Cells were lysed with an SDS sample buffer (62.5 mM Tris-HCl, pH 6.8, 2% 2-mercaptoethanol, 2% SDS, 10% glycerol and 0.02% Bromophenol Blue), and the lysates were boiled for 15 min. Denatured samples were analyzed by 8% or 15% SDS–polyacrylamide gel electrophoresis and transferred to polyvinylidene difluoride (PVDF) membranes (Merck Millipore, Burlington, MA, USA). The blots were blocked for 1 h at room temperature with 1% skimmed milk in PBS containing 0.1% Tween-20 (PBS-T), and were incubated for 1 h with appropriate primary antibodies diluted in the blocking buffer. Primary antibody dilutions were as follows: FLAG, 1:5000 dilution; Myc, 1:5000 dilution; β-actin, 1:10,000 dilution; Rab5B/C, 1:500 dilution; EEA1, 1:500 dilution; and others, 1:1000 dilution. After washing the blots with PBS-T three times, the blots were incubated for 1 h at room temperature with appropriate secondary HRP-conjugated antibodies diluted in the blocking buffer, and then washed with PBS-T three times again. Finally, chemiluminescence signals were detected by using an ECL substrate (Bio-Rad) and a chemiluminescence imager (ChemiDoc Touch; Bio-Rad).

For the EGF and BSA uptake assays, WT and Rab5-CKO MDCK cells were pre-incubated at 37°C in serum-free medium for 12 h and then incubated in serum-free medium containing 2 µg/ml 488–EGF for 20 min or 50 µg/ml 488–BSA for 1 h at 37°C. For the Tf uptake assays, cells were incubated for 20 min at 37°C in complete medium containing 25 µg/ml 594–Tf. The cells treated with fluorescently labeled endocytic cargos were washed with cold PBS twice and fixed with 4% PFA. Fluorescence images of endocytic cargos were analyzed using the confocal laser-scanning microscope as described above.

After Dox and auxin (or DMSO) exposure for 6 h, Rab1-CKO MDCK cells were washed with PBS three times and cultured for 20 min at 37°C in serum-free medium. The cells were then incubated on ice for 1 h with cold serum-free medium containing 25 µg/ml 594–Tf. After washing the cells with cold 0.2% BSA in PBS three times, surface-bound Tf uptake was initiated by incubation at 37°C in complete medium, and the cells were fixed with 4% PFA for the times indicated in Fig. 4B,C. All of these procedures were carried out in the presence of Dox and auxin (or DMSO as a control). The internalized 594–Tf signals were detected with a confocal laser-scanning microscope, and average signal intensity per cell was calculated.

For the EGFR degradation assays, WT, Rab1-CKO and Rab5-CKO MDCK cells were pre-incubated in serum-free medium in 24-well plate for 18–24 h with or without the addition of 1 µg/ml doxycycline and 500 µM auxin for 6 h (Rab1-CKO) or 24 h (Rab5-CKO). After pre-incubation at 37°C for 30 min in the medium containing 100 µg/ml cycloheximide, the cells were treated with 200 ng/ml EGF for the times indicated. The entire level of EGFR protein was checked by immunoblotting with the anti-EGFR antibody.

Quantified data were statistically analyzed using an two-tailed, unpaired Student's t-test, and we used a P-value of 0.05 or less as the criterion for a significant difference.

We thank 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/doi/10.1242/jcs.259184.