RAB19 directs cortical remodeling and membrane growth for primary ciliogenesis

SUMMARY

Primary cilia are sensory organelles that utilize the compartmentalization of membrane and cytoplasm to communicate signaling events, yet how formation of a cilium is coordinated with reorganization of the cortical membrane and cytoskeleton is unclear. Using polarized epithelia, we find that cortical actin clearing and apical membrane partitioning occur where the centrosome resides at the cell surface prior to ciliation. RAB19, a previously uncharacterized RAB, associates with the RAB GAP TBC1D4 and the HOPS tethering complex to coordinate cortical clearing and ciliary membrane growth which is essential for ciliogenesis. This RAB19-directed pathway is not exclusive to polarized epithelia, as RAB19 loss in non-polarized cell types blocks ciliogenesis with a docked ciliary vesicle. Remarkably, inhibiting actomyosin contractility can substitute for the function of the RAB19-complex and restore ciliogenesis in knockout cells. Together, this work provides a mechanistic understanding behind a cytoskeletal clearing and membrane partitioning step required for ciliogenesis.

eTOC

Primary cilia are sensory organelles, yet building these extracellular structures requires reorganization of the plasma membrane and cortical cytoskeleton. Jewett et al. describe a RAB19-driven trafficking pathway that coordinates cortical clearing with ciliary membrane growth. This pathway is required for primary ciliogenesis in both polarized and non-polarized cell types.

Graphical Abstract

INTRODUCTION

The primary cilium, present in most vertebrate cell types, is an essential sensor and regulator of signaling, cell cycle progression, and extracellular cues (Goetz and Anderson, 2010, Seeger-Nukpezah and Golemis, 2012, Ke and Yang, 2014). Defects in primary ciliogenesis result in a number of genetic, multisystemic diseases termed ciliopathies (Hildebrandt et al., 2011). Primary cilia nucleate from the mother centriole and project into the extracellular space. Cilia formation requires extension of axoneme microtubules, membrane remodeling to ensheath the axoneme, and creation of a boundary, called the transition zone, which restricts access to and from the cilium creating a unique cellular compartment (Schmidt et al., 2012, Garcia-Gonzalo and Reiter, 2012, Tanos et al., 2013). Two primary ciliogenesis pathways have been described (Sorokin, 1962, Sorokin, 1968). The first is a well characterized intracellular pathway common in fibroblasts and non-polarized cells. In this pathway, regulatory factors and membranes are recruited to a nuclear-proximal centrosome and the ciliary axoneme begins to grow in the cell interior before migrating toward the cell surface to fuse with the plasma membrane. The second is an extracellular pathway, common in polarized epithelia, which involves direct anchoring of the centrosome to the apical plasma membrane, followed by recruitment of regulatory factors and extension of an axoneme. Despite differences in the location of ciliary initiation, both pathways result in a cilium that extends into the extracellular space to send and receive signals.

Although the cilium is typically thought of as a microtubule-based structure, recent work implicates a role for the actin cytoskeleton in cilium formation and maintenance. Polymerized and branched F-actin networks appear to inhibit ciliogenesis (Kim et al., 2010, Drummond et al., 2018), whereas balanced actin contractility is critical for centrosome migration to the cell surface during ciliogenesis (Lemullois et al., 1988, Dawe et al., 2009, Pitaval et al., 2017, Pitaval et al., 2010). In addition, previous research has shown a void in apical membrane proteins and the actin cytoskeleton where the cilium emerges (Meder et al., 2005, Vieira et al., 2006, Francis et al., 2011, Reales et al., 2015). Because the cell membrane and underlying actin cytoskeleton provide both structural support for cell shape and function as a barrier to cellular entry and exit, traversing the cell cortex may be an additional requirement to accomplish ciliogenesis by both pathways. Yet, how axoneme extension and membrane specialization is coordinated with cortical membrane and actin remodeling is largely unknown.

The RAB family of small GTPases are master regulators of protein and lipid recruitment to cell locations at precise times and function in tandem with microtubule and actin networks. RAB specificity is determined by Guanine nucleotide Exchange Factors (GEFs) and GTPase Activating Proteins (GAPs) which regulate the nucleotide state (Lamber et al., 2019). RAB trafficking pathways are critical for building a functional cilium with a specialized membrane domain (Nachury et al., 2007, Yoshimura et al., 2007, Babbey et al., 2010, Blacque et al., 2018). In particular, a RAB11-RAB8 cascade directs formation of a ciliary vesicle at the mother centriole and recruitment of transition zone proteins to promote coordinated membrane and axoneme growth (Knödler et al., 2010, Westlake et al., 2011, Lu et al., 2015). Because cortical rearrangements during ciliogenesis would allow the cilium to bypass the cell surface and create a specialized membrane domain, we speculated that RABs may serve additional roles in coordinating cortex remodeling with cilia formation.

Primary ciliogenesis is typically studied in the context of non-polarized cell types utilizing the intracellular pathway, and much less is known about the timing and coordination of extracellular ciliogenesis with epithelial polarization. We performed a comprehensive time course analysis of ciliogenesis together with the polarity program in renal epithelial cells. Prior to ciliation, centrosome migration towards the apical cell surface results in cortical clearing of both the actin cytoskeleton and apical membrane proteins. After cortical clearing, ciliation proceeds in two phases: first through recruitment of ciliary membrane and exposure to the extracellular space; followed by accumulation of modified tubulin at the axoneme, which coincides with complete polarization of the monolayer. In searching for RABs that might regulate this cortical clearing prior to ciliogenesis, we discovered that a poorly characterized RAB, RAB19, interacts with the RAB GAP TBC1D4 and the HOPS late endosomal tethering complex to facilitate cortical clearing and ciliary membrane growth. Importantly, this RAB19 mechanism is not limited to polarized epithelia but also critical for ciliogenesis in RPE1 cells utilizing the intracellular pathway and in vivo, suggesting that RAB19-dependent cortical remodeling is a conserved element of ciliogenesis.

RESULTS

Dynamics of extracellular ciliogenesis in polarized epithelia

In epithelial cells, ciliation is tightly coupled with polarity establishment (Bacallao et al., 1989, Reinsch and Karsenti, 1994); however, a characterization of the relative timing of these processes is lacking. We thus performed a time course analysis of ciliation and polarization in Madin-Darby Canine Kidney II (MDCK) cells. Cells were plated on collagen-coated transwell filters, fixed, and analyzed by immunofluorescence every 24 hours for 8 days. MDCK cells were plated sparsely to reach about 20% confluency by day 1. At this time, cells are not yet polarized, and volume projections demonstrate that the centrosome, labeled by a pair of γ-tubulin puncta, resides near the nucleus (Figure 1A). By day 2, the centrosome has moved towards the apical actin cortex and most cells have created an apical clearing in both cortical actin (Figures 1A and 1B) and the composition of the plasma membrane labeled with an antibody to GP135 (also called podocalyxin) (Figures S1A, S1B). At this point, cells have not yet ciliated; however, the unciliated centrosome resides in this clearing (Figures 1A–C). We noticed that beginning on day 5, the γ-tubulin labeled centrosomes separate in most cells (Figures 1A, S1C). One of the centrosomes remains at the apical cortex, whereas the other is found 1–4 μm into the cell. This is not centrosome duplication and progression into G2, as most γ-tubulin foci contain only a single centriole (Figures S1D and S1E). By days 7–8, cells are a fully polarized, confluent monolayer as defined by transepithelial resistance measurements (TER) (Figure 1C). The clearing of cortical actin and apical membrane proteins is retained throughout polarization and ciliation (Figures 1A–D, S1A, S1B, S1F).

Figure 1. Dynamics of extracellular ciliogenesis in polarized epithelia.

Time course analysis of MDCK cells grown on transwell filters and fixed every 24 hours for 8 days. (A) Cells stained for Hoechst, γ-tubulin, phalloidin, and ARL13B. Brackets denote region where cortical actin clearing occurs. (B) Line scan quantitation across apical cell surface measuring relative actin, γ-tubulin, and ARL13B intensities. n_B = 3/n_T = 10. (C) Percent of cells with ARL13B-positive or ARL13B- and Actub-positive primary cilium (left Y-axis) and TER (right Y-axis). n_B = 3/n_T = 3 (ciliation) and n_B = 3/n_T = 8 (TER). (D) Cells stained for phalloidin, ARL13B, and Actub. (E) Representative images depicting time-lapse of centrosome (RFP-PACT) migration towards cortical actin (SirActin). Scale bars, 1 μm. Image width, 5.5 μm. (F) Quantitation from time-lapse of mean centrosome position relative to cell cortex and actin intensity over time. Graph shows mean ± one half SD for n = 11 cells. *P < 0.01 for all time points when compared to t = 0. (G) Quantitation from time-lapse showing cortical actin intensities above and flanking the centrosome. Graph shows mean ± one half SD for n_B = 11. *P < 0.01 for all time points when compared to t = 0. (H) SIM images showing γ-tubulin, ARL13B, and CP110. (I) Quantitation of mother centriole uncapping (left y-axis) and ARL13B accumulation at the centrosome (right y-axis). n_B = 2/n_T = 5. (J) Model schematic of ciliation and polarity in MDCK cells. Data represent mean ± SEM unless otherwise indicated.

To establish the temporal dynamics of centrosome migration relative to cortical actin clearing, centrosomes were visualized relative to the actin cortex in live MDCK cells. Actin intensity levels at the apical cell surface directly above the centrosome decrease by 50% as the centrosome approaches the cell cortex (Figures 1E, 1F, Movie S1). The centrosome then dwells about 400 nm away from the cortex while the actin intensity at the clearing continues to decrease and the width of the clearing increases to about 2 μm. (Figures 1E, 1F, S1G). 30% of these centrosomes exhibit dynamic oscillations relative to the cell cortex and the cortical actin clearance positively correlates with centrosome proximity to the cortex, such that when the centrosome is less than 400 nm from the cortex, actin is cleared, but when the centrosome is further away from the cell cortex, actin fills in (Figure S1H). Interestingly, at the same time actin intensities decrease directly above the centrosome, actin intensities flanking this cleared region increase without a corresponding increase in cortical actin thickness (Figures 1G, S1I). Thus, centrosome migration to the apical cell surface is tightly coupled with decreasing local cortical actin and increasing cortical actin density adjacent to this clearing as MDCK cells polarize prior to ciliogenesis.

In non-polarized cells, ciliation is preceded by remodeling of the mother centriole, including removal of the capping protein CP110 (Spektor et al., 2007). Structured illumination microscopy (SIM) reveals that mother centriole uncapping occurs on days 3–4 as indicated by loss of CP110 from one of the centrioles (Figures 1H, 1I). Furthermore, this remodeling of the mother centriole correlates with accumulation of the ciliary membrane marker ARL13B (Figures 1H, 1I). Remarkably, MDCK cells appear to undergo ciliation in two phases. First, ARL13B is recruited to the centrosome. By day 4, most cells have ARL13B at the apical actin clearing where the centrosome resides, but the ciliary axonemal marker Acetylated tubulin (Actub) is not yet detectable (Figures 1A–D, S1A). By day 7–8 however, Actub has accumulated at the proximal end of almost all primary cilia (Figures 1C and 1D).

To determine when the cilium becomes extracellular during ciliogenesis, we created a stable cell line expressing pHluorin-tagged (pH-sensitive) Smoothened (pH-SMO) and performed an IN/OUT assay (Kukic et al., 2016). Cells were fixed and stained with an anti-GFP antibody prior to permeabilization to only label cilia that have emerged from the cell and exposed pH-SMO. By day 4, most cilia are extracellular as determined by colocalization of pH-SMO and anti-GFP signals (Figures S1K, S1J). This suggests that ARL13B observed on day 4 is not simply accumulation of ARL13B at the mother centriole, but most likely a nascent, extracellular cilium.

In summary, our time course analyses of polarization and ciliation in MDCK cells suggest a model (Figure 1J) whereby early in the polarity process, centrosome migration to the apical cell surface results in a localized clearing of cortical actin and apical membrane proteins, defined hereafter as “cortical clearing” (CC). Concurrently, the density of cortical actin flanking the clearing increases. Then, mother centriole remodeling occurs and ARL13B accumulates at the mother centriole. These nascent ciliary membranes are exposed to the extracellular space. Eventually, the daughter centriole moves away from the cell surface, and ARL13B ciliary structures continue to assemble. Finally, Actub labels the ciliary axoneme 7–8 days after plating cells.

RAB19 localizes to the inner periphery of the cortical clearing

Our data suggest that topological changes in membrane at the cell surface are coordinated with the cytoskeleton to drive cortical remodeling prior to ciliogenesis. Because the RAB family of small GTPases are master regulators of protein and lipid recruitment to specific cellular locations, we visualized the localization of several candidate RABs and found that RAB19 exhibited striking localization changes during MDCK polarization. Prior to polarity establishment, RAB19 localizes to vesicles enriched at the apical cell surface (Figures 2A, S2A), however as cortical actin begins to clear at the future ciliation site, RAB19 localizes just inside the periphery of the clearing (Figure S2B). Then, as cells accumulate ARL13B at the centrosome on day 4, RAB19 forms a cluster of vesicles immediately surrounding ARL13B inside the CC (Figure 2B). This periciliary localization is maintained throughout polarity establishment and maintenance (Figures 2C, 2D, S2C, S2D). Given that apical transport of vesicles is often dependent on microtubule networks, we next asked whether RAB19 periciliary localization requires microtubules. Upon microtubule depolymerization, RAB19 is lost from the inner periphery of the CC and the clearing decreases in size (Figures 2E, compare to 2D). In summary, RAB19 clusters at the inner periphery of the CC just prior to accumulation of ARL13B at the centrosome, and maintenance of this actin clearing requires microtubule-dependent RAB19 periciliary localization (Figure 2F).

Figure 2. RAB19 localizes to the inner periphery of the cortical clearing.

Time course analysis of MDCK cells stably expressing GFP-RAB19 in the RAB19 KO background grown on transwell filters. (A-D) XY and XZ images of cells fixed and stained for Hoechst, phalloidin, and ARL13B. Corresponding line scans across apical cell surface measuring relative intensities on right. Arrows point to cilium where actin clearing occurs and RAB19 localizes. (E) Cells treated with 3nM nocodazole for 3 hours prior to fixation on day 8. Arrow points to small actin clearing where RAB19 localization is lost. Corresponding line scan across the apical cell surface measuring relative intensities on right. Note loss of RAB19 at clearing and decreased width of actin clearing. (F) Model schematic summarizing Rab19 localization. n_B = 3/n_T = 10. Data represent mean ± SEM.

RAB19 functions in cortical clearing and ciliary membrane growth

Given RAB19 localization patterns at the apical cell surface, we asked if RAB19 was required for CC. We used CRISPR/Cas9 to create a RAB19 knockout (KO) MDCK cell line and analyzed polarity progression in RAB19 KO cells as we did in wild-type (WT) cells. RAB19 KO cells show no defects in apical centrosome localization (Figures 3A, S3A); however, CC is inhibited in RAB19 KO cells (Figures 3A, 3B, S3B, S3C). Live imaging of RAB19 KO cells further reveals defects in cortical actin clearing specifically in the region directly above the centrosome. Conversely, actin networks flanking the centrosome are unchanged compared to WT cells (Figures 3C, 3D, S3D, Movie S2). TER measurements indicate that RAB19 KO cells form confluent monolayers (Figure S3E). Furthermore, no defects in the cell cycle, apical basal cell polarity, tight junctions, Golgi stack morphology, centriole distance, or medial cytoskeletal networks are observed in RAB19 KO cells (Figures S3F–L). We next examined cilia formation in cells lacking RAB19. About 50% of RAB19 KO cells accumulate ARL13B at the centrosome, but these structures are not well retained at later time points, and most cells do not show Actub staining (Figures 3E, 3F). Importantly, the CC and ciliation phenotypes are rescued by stable expression of GFP-RAB19 (Figures S3M, S3N, 2A–D), and overexpression of a RAB19 dominant negative mutant phenocopies loss of cilia in RAB19 KO cells (Figures S3N, S3O). Thus, RAB19 promotes CC for efficient ciliogenesis in polarized epithelia.

Figure 3. RAB19 functions in cortical clearing and ciliary membrane growth.

Time course analysis of MDCK RAB19 KO cells grown on transwell filters and fixed every 24 hours for 8 days. (A) Cells stained for Hoechst, γ-tubulin, phalloidin, and ARL13B. (B) Integrated actin intensities through 2 μm region centered on γ-tubulin in WT and RAB19 KO cells. n_B = 2/n_T = 5. ns and *P < 0.05 compare WT and KO at each time point. (C) Representative images depicting time-lapse of centrosome migration towards cortical actin in RAB19 KO cells. Scale bars, 1 μm. Image width, 5.5 μm. (D) Quantitation from time-lapse showing cortical actin intensity over time above and flanking centrosome in WT and Rab19 KO cells. Graph shows mean ± one half SD for n_B = 13. *P < 0.01 compare WT and KO at each time point. (E) RAB19 KO cells stained for phalloidin, ARL13B, and Actub. (F) Percent of WT and RAB19 KO cells with ARL13B-positive or ARL13B- and Actub-positive primary cilium. n_B = 3/n_T = 3. (G) Distribution of four classes of phenotypes in WT and RAB19 KO cells. n_B = 2/n_T = 3. (H) RAB19 KO cells stably expressing either Flag-RAB19 or constitutively active mutant (RAB19^Q76L) stained for Flag, ARL13B, and Actub. (I) Quantitation of ciliary length. n_B = 3/n_T = 12. ***P < 0.0005. (J) Model schematic of RAB19 function in CC and ciliary membrane growth. Data represent mean ± SEM unless otherwise indicated.

Because a subset of RAB19 KO cells demonstrate CC, we asked whether these cells also accumulate ARL13B at the centrosome. Like WT cells, RAB19 KO cells display no clearing and no ARL13B on day 1, but by day 2 some cells show a clearing without ARL13B (Figure 3G), consistent with the clearing occurring before ciliation. At later time points, most WT cells exhibit both a clearing and ARL13B at the centrosome. Conversely, in RAB19 KO cells, a new population of cells emerge on day 3 that have ARL13B accumulation without a clearing. This population persists through later time points, along with a mixture of the other three phenotypes (Figure 3G). These data suggest that RAB19 has an additional function in ciliary membrane delivery that may be coordinated with CC. To further investigate RAB19’s role in ciliary membrane delivery, we asked whether overactivated RAB19 affects cilium length. We expressed either Flag-RAB19 or Flag-RAB19^Q76L, the RAB19 constitutively active mutant, in RAB19 KO cells and measured ciliary length. Cells expressing Flag-RAB19 form cilia of a similar length to WT cells. However, cells expressing Flag-RAB19^Q76L form unusually long cilia labeled with both ARL13B and Actub (Figures 3H, 3I). Altogether, these data suggest that RAB19 functions in both CC and ciliary membrane growth.

One possibility for lack of coordination between CC and ciliary membrane growth is defective mother centriole remodeling for primary ciliogenesis. To test this, we performed SIM and found that CP110 uncapping occurs in RAB19 KO cells but is delayed by 1 day compared to WT cells (Figures S3P, S3Q). Moreover, the RAB19 KO cells that accumulate ARL13B at the mother centriole demonstrate increased cytoplasmic ARL13B staining and smaller centriolar ARL13B puncta (Figures S3P, compare to 1E). We next tested whether the portion of RAB19 KO cells that undergo CC and centrosomal ARL13B accumulation form an extracellular cilium. In WT cells, 85–90% of ARL13B-positive structures at the centrosome are extracellular on days 3–4 (Figures 1I, S1J) compared to RAB19 KO cells, in which about half of the ARL13B-positive structures are extracellular on day 4 and this correlates with the percentage of cells that have both a clearing and ARL13B (Figures 3F–G, S3R–S). In summary, RAB19 coordination of CC with ciliary membrane growth is required to build an extracellular cilium (Figure 3J).

Several other RAB proteins have been implicated in primary cilia formation, so we wondered how RAB19 integrates with these established ciliary RABs. RAB11/RAB8-directed trafficking events promote the formation of preciliary structures during early steps in ciliogenesis, so we examined the possibility of a functional interplay between RAB19 and RAB11. RAB19 loss does not affect centrosomal RAB11 (Figure S4A), and cells expressing the dominant negative RAB11 mutant exhibit normal RAB19 localization (Figure S4B). Interestingly, expression of the RAB11 dominant negative mutant results in CC defects (Figure S4C); however, we cannot rule out the possibility that this is a result of general polarization problems. Thus, RAB19 functions independently of the early RAB11/RAB8 pathway to promote CC and ciliary membrane growth in polarized epithelia.

RAB19 regulates intracellular ciliogenesis

We next wondered if RAB19 function was broadly applicable to ciliogenesis in non-polarized cell types. Human Retinal Pigmented Epithelial (RPE1) cells are a common intracellular cilia model. We generated an RPE1 RAB19 KO cell line using CRISPR/Cas9. Similar to polarized epithelia, RAB19 KO in RPE1 cells causes cilia loss but no defects in mother centriole uncapping or remodeling (Figures 4A–D, S4D). This cilia phenotype can be rescued by exogenous GFP-RAB19 (Figures 4B, S4E). Moreover, transmission electron microscopy of RAB19 KO cells reveals that most cilia are blocked at the ciliary vesicle stage and those that do make it past this stage form a stubby cilium (Figure 4E). These stubby cilia are always intracellular when analyzed with the IN/OUT assay (Kukic et al., 2016) (Figure S4F), but it is worth noting that most RAB19 KO cells do not accumulate ciliary pH-SMO to the same degree as WT cells. We next asked whether the intraflagellar transport (IFT) machinery or transition zone to control trafficking to and from the cilium were disrupted in RAB19 KO cells. Both the localization of a transition zone protein (RPGRIP1L) and two IFT proteins (IFT88 and 140) are unaffected in RAB19 KO cells (Figures 4D, 4F, S4G). Thus, RAB19 promotes CC and/or ciliary membrane growth in both the polarized extracellular and intracellular ciliogenesis pathways (Figure 4J).

Figure 4. RAB19 regulates intracellular ciliogenesis.

(A-F) RPE1 cells grown on glass coverslips and serum starved for 48 hours prior to fixation. (A) RPE1 WT and RAB19 KO cells stained for Hoechst, ARL13B, and Actub. (B) Percent of cells with a primary cilium in WT, RAB19 KO, and RAB19 KO stably expressing GFP-RAB19 cells. n_B = 2/n_T = 9, 8, 7, respectively. ***P < 0.005. (C) SIM of RPE1 WT and RAB19 KO cells stained for centrin, Actub, and CP110. Arrows point to uncapped mother centriole. Uncapping occurs in 100% of cells; n = 26 for WT, n = 21 for RAB19 KO. (D) SIM of RPE1 WT and RAB19 KO cells stained for GT335, CEP164, and RPGRIP1L. Arrows point to transition zone. (E) Serial TEM of RPE1 WT and RAB19 KO cells. Percentages indicate cells found at each stage. n_B = 2/n_T = 3 WT, n_B = 2/n_T = 9 RAB19 KO. (F) SIM on RPE1 WT and RAB19 KO cells stained for GT335 and IFT88. Arrows point to transition zone. (G) Tiled confocal microscopy images of control and rab19 probed fluorescent HCR in situ on WT embryos at 3dpf. (H) Protein schematic of zebrafish rab19 showing RAB domain in yellow and di-prenylation motif at C-terminus in magenta. The rab19^sa24592 mutant introduces a premature stop before prenylation motif. (I) Bright-field images of wt and homozygous rab19^sa24592 mutant zebrafish at 3dpf. (J) Model schematic depicting RAB19 function during intracellular ciliogenesis. Data represent mean ± SEM.

In zebrafish, rab19 is expressed in ciliated organs such as the otic vesicle, lateral line, kidneys, and spinal cord at 3 days post fertilization (dpf) (Figure 4G). To test whether rab19 is required for ciliogenesis in vivo, we analyzed rab19 mutant zebrafish (rab19^sa24592) that generate a premature stop in the rab19 gene. The resulting rab19 mutant lacks a prenylation motif and cannot insert into membranes (Figure 4H). Consistent with zebrafish phenotypes associated with ciliopathy genes (Song et al., 2016), rab19 homozygous mutant fish demonstrate dorsal body curvature, small heads, and heart edema at 3 dpf (Figure 4I). These mutant fish cannot survive past 4 dpf. While this in vivo data is preliminary, it suggests that RAB19’s function during ciliogenesis is broadly conserved in vertebrates.

TBC1D4 interacts with RAB19 and is required for primary ciliogenesis

We next asked what RAB19 interacts with to coordinate CC and ciliary membrane growth. Co-immunoprecipitation/mass spectrometry on MDCK cells expressing Flag-RAB19 identified the RAB GAP, TBC1D4 (also called AS160), and every subunit of the HOPS late endosomal tethering complex (Figure 5A). TBC1D4 has a closely related family member called TBC1D1 (Figure 5B) and the Drosophila ortholog for both interacts with RAB19 (Gillingham et al., 2014). They are typically studied for their role in insulin-dependent glucose transporter trafficking and show redundancy or cooperativity in a cell-type specific manner (Sakamoto and Holman, 2008, Fontanesi and Bertolini, 2013, Hatakeyama et al., 2019). Through glutathione bead pulldown assays, we validated the binding of GST-RAB19 to TBC1D1 and TBC1D4 in MDCK cells (Figures 5C, 5D) and found that this interaction is specific to RAB19 and not RAB11A (Figure 5E).

Figure 5. TBC1D4 interacts with RAB19 and is required for primary ciliogenesis.

(A) Spectral counts of candidate proteins identified through co-immunoprecipitation/mass spectrometry on Flag-RAB19 in its GTP-locked, GDP-locked, and constitutively active (RAB19^Q76L) forms. The full list of proteins is in Table S1. (B) Domain schematic of TBC1D4 and the closely related family member TBC1D1. Both proteins contain two phospho-tyrosine binding domains (PTB), a calmodulin binding domain (CBD), two coiled-coil domains (CC1 and CC2), a RAB-GAP domain, and several known phosphorylation sites (P in red circle). (C) Pulldown with recombinant GST-RAB19 and MDCK cell lysate followed by Western Blot probing for TBC1D1 or TBC1D4. (D) Quantitation of TBC1D1 and TBC1D4 band intensities for GTP- and GDP-locked RAB19. N_B = 3. (E) Pulldown with recombinant GST-RAB11a and MDCK cell lysate followed by Western Blot probing for TBC1D1 or TBC1D4. (F) Percent of cells with a primary cilium for WT, TBC1D1 KO, TBC1D4 KO, GFP-TBC1D4 rescue, and GFP-TBC1D4^R972K rescue. n_B = 3/n_T = 6, 7, 5, 3, 3, respectively. (G) WT and TBC1D4 KO cells stained for GP135, phalloidin, and Actub. Arrows denote CC. (H) Integrated actin intensities through 2 μm region around γ-tubulin in WT and TBC1D4 KO cells. n_B = 2/n_T = 5. ns and * compare WT and KO for each condition. (I) RPE1 RAB19 KO cells stably expressing GFP-RAB19 serum starved for 48 hours and stained for Hoechst, γ-tubulin, and TBC1D4. (J) RPE1 WT and RAB19 KO cells serum starved for 48 hours and stained for Hoechst, TBC1D4, and PCM1. (K) Model schematic of RAB19 and TBC1D4 function during ciliogenesis. ***P < 0.0005. Uncropped blots shown in Figures S7A–D. Data represent mean ± SEM.

If RAB19 and TBC1D1 or TBC1D4 function in the same ciliogenesis pathway, we hypothesized that loss of function of either protein should phenocopy RAB19 KO cells. CRISPR/Cas9-generated KO of TBC1D1 has no effect on ciliogenesis, whereas KO of TBC1D4 results in loss of cilia (Figures 5F, S5A–D). Given the cilia phenotype in TBC1D4 KO cells, we tested whether CC could occur. Remarkably, on day 4, TBC1D4 KO cells have no CC defects compared to WT cells, but such defects are observed by day 8 (Figures 5G, 5H). This suggests that TBC1D4 is not required for early CC steps, but perhaps interacts with RAB19 to regulate ciliary membrane delivery or CC maintenance at later time points.

Because TBC1D4 is a RAB GAP, we tested if its GAP activity is required for ciliogenesis. Whereas stably expressing WT GFP-TBC1D4 rescues loss of cilia, a GAP-dead mutant does not rescue ciliation (Figures 5F, S5E, S5F). TBC1D4 is a known GAP for the ciliary RAB, RAB8 (Miinea C et al., 2005), but its GAP activity towards RAB19 has never been tested. To this end, we performed an in vitro GAP assay and found that TBC1D4 shows mild GAP activity towards RAB19 and this GAP activity is slightly decreased with the GAP-dead mutant of TBC1D4 (Figure S5G). To further analyze this GAP activity in cells, we compared the localization of RAB8 and RAB19 at the centrosome in WT and TBC1D4 KO cells. Both RAB8 and RAB19 localization to the centrosome is slightly but significantly increased in TBC1D4 KO cells (Figures S5H–J). In addition to RAB8, TBC1D4 is a GAP for RAB10, which associates with cilia and negatively regulates ciliogenesis (Miinea C et al., 2005, Babbey et al., 2010, Dhekne et al., 2018). We thus asked if TBC1D4 KO cells inhibit cilia formation through failure to inactivate RAB10. When the dominant negative mutant of RAB10 (RAB10^T23N) is expressed in WT and TBC1D4 KO cells, no defects in cilia formation are observed in WT cells overexpressing RAB10^T23N, and TBC1D4 KO cells overexpressing RAB10^T23N remain unable to ciliate. Therefore, overactive RAB10 does not contribute to the cilia defects observed in TBC1D4 KO cells. Collectively, these data suggest that RAB19 and TBC1D4 function in the same pathway to regulate ciliogenesis partly through the GAP activity of TBC1D4.

We next explored TBC1D4 localization. We generated a polyclonal antibody to both the N- and C-fragments of TBC1D4 (Figure S5A). Both of our antibodies and commercial antibodies do not recognize dog (MDCK) TBC1D4 by immunofluorescence; however, our antibody does recognize mouse and human TBC1D4. Therefore, we used a mouse polarized epithelial line (mIMCD3 cells) and human RPE1 cells to analyze TBC1D4 localization. TBC1D4 is found both at the centrosome and at centriolar satellites in both cell types as indicated by colocalization with PCM1 (Figures 5I, 5J, S5L). Similar localization is observed in cells stably expressing GFP-TBC1D4, and the GAP activity of TBC1D4 has no effect on this localization (Figure S5M). Moreover, TBC1D4 localization is unchanged in RAB19 KO cells (Figure 5J), suggesting that RAB19 is not required for TBC1D4 localization. Finally, a co-occurrence analysis on RAB19 and TBC1D4 localization shows that TBC1D4 is mostly at the periphery of pericentrosomal GFP-RAB19 vesicles (Figures 5I and S5N). Taken together, our data suggest that TBC1D4 is a centrosome- and satellite-localized protein that interacts with RAB19 to regulate ciliary membrane growth or CC maintenance during primary ciliogenesis (Figure 5K).

The HOPS complex is a RAB19 effector required for primary ciliogenesis

In addition to TBC1D4, RAB19 proteomics identified every subunit of the HOPS tethering complex (Figure 5A). The HOPS complex comprises six individual protein subunits (Figure 6A) that co-assemble to mediate vesicle-membrane tethering during late stages of the endosomal pathway (Balderhaar and Ungermann, 2013). We validated this interaction with pulldown assays and found that the HOPS subunits, VPS11 and VPS41, preferentially interact with RAB19 in its GTP-bound, active state and fail to interact with RAB11A (Figures 6B–D). Recent studies suggest that the VPS41 subunit of the HOPS complex can moonlight alone in other cellular functions (Asensio et al., 2013, Pols et al., 2013); however, in our system, VPS41 is detected only in the glycerol gradient fractions predicted to have molecular weights similar to the HOPS complex in both WT and RAB19 KO MDCK cells (Figure S6A). This suggests that in MDCK cells, VPS41 functions together with the HOPS complex, and RAB19 is not required for HOPS complex formation.

Figure 6. The HOPS complex is a RAB19 effector required for primary ciliogenesis.

(A) Cartoon of HOPS complex. Four subunits (VPS33, VPS16, VPS18, VPS11) are shared with the closely related CORVET complex whereas the two blue subunits (VPS41 and VPS39) are specific to HOPS. (B) Pulldown with recombinant GST-RAB19 and MDCK cell lysate followed by Western Blot probing for VPS11 or VPS41. (C) Quantitation of VPS11 and VPS41 band intensities for GTP- and GDP-locked RAB19. n_B = 3. (D) Pulldown with recombinant GST-RAB11A and MDCK cell lysate followed by Western Blot probing for VPS11 or VPS41. (E) WT and VPS41 KO cells stained for GP135, phalloidin, and Actub. Arrows denote CC. (F) Integrated actin intensities through 2 μm region around γ-tubulin in WT and VPS41 KO cells. n_B = 2/n_T = 5. * compares WT and KO for each condition. (G) Percent of cells with a primary cilium for WT, VPS41 KO #1, VPS41 KO #2, WT cells stably expressing GFP-VPS41, and VPS41 KO cells stably expressing GFP-VPS41. n_B = 2/n_T = 6, 6, 4, 3, 3, respectively. (H) SIM on RPE1 WT and RAB19 KO cells serum starved for 48 hours and stained for centrin, VPS11, and pericentrin (PCNT). (I) Model schematic of RAB19 and HOPS complex function during ciliogenesis. ***P < 0.0005, *P < 0.05. Uncropped blots shown in Figure S7E–G. Data represent mean ± SEM.

To determine if HOPS regulates CC and ciliogenesis, we used CRISPR/Cas9 to create a VPS41 KO MDCK cell line, a subunit required for HOPS complex assembly. Consistent with RAB19 KO cells, VPS41 KO cells display CC defects (Figures 6E, 6F). Moreover, VPS41 KO cells show the most severe cilia loss in both polarized cells grown on filters and semi-polarized cells grown on coverslips (Figures 6G, S6B–D). We attempted to rescue this cilia phenotype by stably expressing GFP-VPS41 in VPS41 KO cells, but these cells show severe polarization and cell-adherence defects (Figures 6G, S6E). Additionally, stable expression of GFP-VPS41 in WT cells are also unable to robustly form cilia (Figures 6G, S6E). However, the cilia and CC phenotypes were recapitulated with a second VPS41 KO allele (Figures 6G, S6F) providing evidence that these are not off-target phenotypes, and instead VPS41 functions as part of the HOPS complex where stoichiometries are critical.

We next investigated the relationship between RAB19 and HOPS-labeled vesicles. VPS11-positive vesicles cluster around the centrosome and this localization is dependent on RAB19 (Figure 6H). To determine whether RAB19 and HOPS localize to the same vesicles, we performed a co-occurrence analysis on GFP-Rab19 and VPS11 vesicles and find that approximately half of RAB19-positive vesicles overlap with VPS11-labeled vesicles (Figures S6G, S6H). The HOPS complex was originally identified as a tethering complex regulating late endosome, lysosome, and autophagosome pathways (Balderhaar and Ungermann, 2013). Consistent with HOPS regulation of late endosomal trafficking (Rink et al., 2005, Chirivino et al., 2011), we observe enlarged early endosomes when HOPS complex formation is disrupted (Figure S6I). We thus wondered whether RAB19 could be interacting with HOPS as part of the lysosomal degradation pathway. The vast majority of RAB19 vesicles do not overlap with CD63-positive lysosomes (Figures S6J, S6K). In summary, the HOPS complex is a RAB19 effector functioning in CC and ciliogenesis through a non-lysosomal pathway.

Our data identify RAB19, TBC1D4, and the HOPS complex to localize and function at the pericentrosomal region to regulate CC and ciliary membrane growth. To test whether RAB19 mediates assembly of TBC1D4 and HOPS into a complex, we performed pulldowns with recombinant GST-TBC1D4 and WT or RAB19 KO MDCK cell lysates, probing for the HOPS subunits VPS41 and VPS11. VPS41 and VPS11 co-precipitate with the C-terminus of TBC1D4 and this binding is significantly decreased in RAB19 KO cells (Figures S6L, S6M). Moreover, RAB19, TBC1D4, VPS41, and VPS11 are all found in the same glycerol fraction (Figure S6A), and we do not observe changes in TBC1D4 or VPS41 protein levels in RAB19 KO cells (Figure S6N). Overall, this indicates that RAB19 recruits the HOPS complex to the centrosome and mediates interaction with TBC1D4 to coordinate CC and ciliary membrane growth during primary ciliogenesis.

RAB19 antagonizes actin contractility to facilitate ciliogenesis

Recent work has linked actin networks to ciliogenesis without defining a clear function or regulatory mechanism (reviewed in (Mirvis et al., 2018)). In polarized epithelia, contractile actin networks are highly enriched at the apical cell surface (Vasquez et al., 2014). To test whether RAB19 harnesses these Myosin II-based contractile actin networks to regulate CC during ciliogenesis, we treated both WT and RAB19 KO MDCK cells with the ROCK inhibitor Y27632 to disrupt actomyosin contractility. WT cells treated with the ROCK inhibitor show faster ciliation with CC and rapid Actub accumulation at the cilium by day 4 (Figures 7A, compare to S1A, 7E). Remarkably, RAB19 KO-treated cells recover cilia formation, although these cilia are shorter than their WT counterparts (Figures 7B, 7F–H). Moreover, CC is rescued in ROCK inhibited RAB19 KO cells (Figures 7B, compare to S3B, 7E). Given that RAB19 interacts with TBC1D4 and the HOPS complex, we asked whether inhibiting actin contractility also rescued the cilia and CC defects in these KO backgrounds. ROCK inhibitor treatment has no effect on ciliation or CC in TBC1D4 KO cells (Figures 7C, compare to 5G, 7E–G). Conversely, VPS41 KO cells treated with the ROCK inhibitor form a CC but remain unable to ciliate (Figures 7D, compare to 6F, 7E–G). Therefore, these data indicate that contractile actomyosin networks in polarized epithelia inhibit ciliogenesis and are remodeled by RAB19 and the HOPS complex either directly or indirectly (Figure 7O), whereas TBC1D4 is likely responsible for ciliary membrane growth and not CC.

Figure 7. RAB19 antagonizes actin contractility to facilitate ciliogenesis.

(A-D) MDCK WT (A), RAB19 KO (B), TBC1D4 KO (C), and VPS41 KO (D) cells stained for GP135, phalloidin, and Actub. Cells were treated with 10 μM Y27632 for 48 hours prior to fixation. Arrows point to CC. (E) Integrated actin intensities through 2 μm region around γ-tubulin in WT and KO cells. n_B = 2/n_T = 5. * indicates difference in TBC1D4 KO compared to WT and other KOs. All other comparisons are not significant. (F) MDCK WT and KO cells treated with 10μM Y27632 for 48 hours prior to fixation and stained for ARL13B, Actub, and phalloidin. (G) Percent of cells with a primary cilium in untreated and Y27632-treated MDCK WT and KO cells. n_B = 2/n_T = 5, 5, 6, 6, 2, 2, 2, 2, respectively. (H) Cilia length quantitation with ARL13B and Actub markers in MDCK WT and RAB19 KO cells on day 8. n_B = 2/n_T = 30, 29, 24, respectively. (I,J,N) RPE1 WT and RAB19 KO cells treated with DMSO control (I), 10μM Y27632 (J), and 0.5μM Cytochalasin D (N) for 48 hours in serum starvation media. Cells stained for Hoechst, γ-tubulin, and Actub. (K) Percent of cells with a primary cilium in RPE1 WT and RAB19 KO cells for DMSO control, Y-27632, and Cytochalasin D-treated conditions. n_B = 2/n_T = 5, 4, 3, 3, 2, 2, respectively. (L) Quantitation of extracellular cilia on dorsal and ventral side in RPE1 WT and RAB19 KO cells treated with 10μM Y27632 from IN/OUT assay. n_B = 2/nT = 2. (M) IN/OUT assay in RPE1 WT and RAB19 KO cells treated with 10μM Y27632. (O-P) Model schematics depicting the role of actin contractility in CC and ciliogenesis in MDCK (O) and RPE1 (P) cells. ***P < 0.0005, **P < 0.005, *P < 0.05. Data represent mean ± SEM.

To test if ciliogenesis in non-polarized cell types was also rescued by inhibition of actomyosin contractility upon RAB19 loss, we treated WT and RAB19 KO RPE1 cells with Y27632. Similar to polarized cells, inhibition of actomyosin contractility also rescues cilia formation in RAB19 KO RPE1 cells utilizing the intracellular pathway (Figures 7I–K). Most of these cilia are extracellular in both WT and RAB19 KO cells (Figures 7L, 7M), although KO cells show decreased ciliary SMO levels compared to WT. Interestingly, whereas WT cells have equal frequencies of forming a cilium on the dorsal or ventral surface, RAB19 KO cells treated with the ROCK inhibitor always form extracellular cilia on the ventral surface of the cell (Figure 7L). In RPE1 cells, the ventral side contains contractile stress fibers, so we wondered if ciliary rescue was specific to actin contractility or more broadly applicable to general disruption of actin networks. To test this, we depolymerized actin with Cytochalasin D treatment. Consistent with the literature (Kim et al., 2010), WT cells form unusually long cilia, whereas minimal rescue of ciliation is observed in RAB19 KO cells (Figures 7K, 7N). Thus, RAB19 regulates a subpopulation of actomyosin-based movements during ciliogenesis in both polarized and non-polarized cell types (Figures 7O, 7P).

DISCUSSION

The primary cilium is a partially extracellular organelle of specialized membrane composition resulting in a region around the cilium devoid of apical membrane proteins and cortical actin (Francis et al., 2011). However, the timing, formation, and mechanism of this cortical actin and apical membrane clearing remained uncertain. We performed a time-course characterization of primary ciliogenesis in polarized epithelia and found that cortical cytoskeletal and plasma membrane changes at the apical cell surface are coordinated with centrosome docking and occur several days prior to complete ciliogenesis. This CC is driven by a RAB19 trafficking pathway which runs in parallel to mother centriole remodeling and ciliary vesicle formation. Proteomic analysis uncovered RAB19 interactions with the RAB GAP TBC1D4 and HOPS vesicle tethering complex, and our functional studies reveal a mechanism whereby RAB19 forms a complex with TBC1D4 and HOPS to coordinate CC with ciliary membrane growth. Importantly, this RAB19 mechanism is also required for ciliogenesis in non-polarized cell types, and zebrafish rab19 mutants show phenotypes consistent with those of known ciliopathy genes.

Crosstalk between RAB19-directed membrane remodeling and cytoskeletal networks

RAB proteins are well-established regulators of protein and lipid recruitment with defined roles in coordinating cortical membrane and cytoskeleton topology (Dhekne et al., 2014, Jewett et al., 2017, Fremont et al., 2017). We discovered that RAB19, a previously uncharacterized RAB, also fits into this category. RAB19 shows dynamic localization at the CC in a peri ciliary ring and is functionally required for CC. The primary cilium is a microtubule-based organelle, yet the actin cytoskeleton is closely interweaved with cilia formation and maintenance (Mirvis et al., 2018). Our results suggest an additional role for actin contractility in CC during ciliogenesis that functions antagonistically with RAB19. Given that Myosin II inhibition rescues CC and ciliogenesis defects in RAB19 KO cells, and disruption of RAB19 periciliary localization through brief microtubule depolymerization results in decreased size of the actin clearing, we propose that RAB19 actively restricts contractile actin networks from the clearing and perhaps mediates crosstalk between microtubules and cortical actin networks. Moreover, RAB19’s function in CC is specific to actin networks directly above the centrosome, as actin networks surrounding this cleared region are unaffected in RAB19 KO cells.

Mechanistic insights into RAB19-mediated formation of TBC1D4 and HOPS complex

We identified TBC1D4 specifically as a RAB19 interacting partner that functions in cilia formation and localizes to the centrosome and satellites. Our data suggest that TBC1D4 is a weak GAP for RAB19, but TBC1D4 KO cells do not phenocopy overexpression of a RAB19 constitutively active mutant, whose longer cilia phenotype will be interesting to tease apart in the future. TBC1D4 is also a GAP for RAB8 both in vitro and in cells (Miinea C et al., 2005). Given that RAB11-RAB8-directed ciliary vesicle formation can proceed in the absence of RAB19 and CC, it is possible that TBC1D4 may link these pathways through functioning as a GAP for RAB8 and perhaps also RAB19 at the centrosome. This allows release of RAB19 cargo and effectors at the centrosome and conversion of the RAB8 ciliary vesicle into a cilium. However, our data does not rule out the possibility that TBC1D4 interacts with RAB19 in another context and it is likely that TBC1D4 has multiple functions during ciliogenesis.

The HOPS complex is a membrane tethering complex studied for its role in the late endosomal stages of endo-lysosomal trafficking (Balderhaar and Ungermann, 2013). Our work supports an additional function for the HOPS complex in a non-lysosomal trafficking pathway required for CC and ciliogenesis. We propose that the HOPS complex is a RAB19 effector acting as a membrane tether, bringing two membranes into close enough proximity to fuse. It remains to be seen which two organelles, although potential candidates include RAB19 vesicles with the cortical membrane or the ciliary vesicle with another membrane-bound organelle. In support of our data, VPS41 and other HOPS subunits were identified as positive regulators of ciliary signaling in a genome wide CRISPR screen (Breslow et al., 2018). Moreover, while this manuscript was in preparation, work implicating VPS39, the other HOPS-specific subunit, in ciliogenesis was published (Iaconis et al., 2020).

In summary, our work identifies a CC and membrane delivery step required for primary ciliogenesis governed by RAB19 interaction with TBC1D4 and the HOPS complex. Although we identified this RAB19-clearing pathway in polarized epithelia, our data suggest a similar mechanism to occur in cells with nuclear-proximal centrosomes. In the future, it will be interesting to further tease apart the relationship between RAB19 and actin networks, especially in clearing a path for an intracellular cilium to make contact with the plasma membrane.

STAR Methods

RESOURCE AVAILABILITY

Lead Contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Rytis Prekeris (rytis.prekeris@cuanschutz.edu).

Materials Availability

Antibodies and plasmids generated in this study are available upon request.

Data and Code Availability

The published article includes all datasets generated during this study. This study did not generate new code.

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Cell lines

MDCK II (ATCC), hTERT RPE-1 (ATCC), and mIMCD3 (ATCC) cells were cultured at 37°C with 5% CO₂ in DMEM with 10% FBS and penicillin/streptomycin. Cell lines were authenticated through STR testing. Each line was checked against the ATCC and DSMZ databases for fingerprinting. Cells were routinely screened for mycoplasma.

Zebrafish lines and husbandry

All animal work was approved by the Institutional Animal Care and Use Committee at the University of Colorado School of Medicine. All stocks were maintained in a heterozygous state. Embryos were raised at 28.5°C in embryo medium (5 mM NaCl, 0.17 mM KCl, 0.33 mM CaCl, 0.33 mM MgSO4, pH to 7.4 with sodium bicarbonate) and staged according to hours/days post-fertilization and morphological criteria (Kimmel et al., 1995). Sex cannot be determined at the embryonic and larval stages. RAB19^sa24592 mutants were obtained from the Zebrafish International Resource Center (ZIRC) and outcrossed to the AB WT strain for two generations before using for experiments. RAB19^sa24592 homozygous mutants were validated through PCR and restriction enzyme digest genotyping and siblings from the same clutch were used as WT controls.

METHOD DETAILS

Immunoprecipitation and proteomics analysis of RAB19

Flag-RAB19 WT and constitutively active mutant (RAB19^Q76L) were cloned into the pLVX-Puro lentiviral vector and transfected into MDCK cells along with viral packaging plasmids D8.9 and pVSVG. Single clones expressing Flag-RAB19 WT and Flag-RAB19^Q76L were selected for experiments. Immunoprecipitation and proteomics analysis of RAB19 binding partners was performed as previously described (Willenborg et al., 2011). Mass spectrometry was performed by the University of Colorado Proteomics Core Facility. RAB19 binding partners were analyzed under three different conditions: WT RAB19 locked in its GTP-bound form using GTPγS, WT RAB19 in its GDP-bound form, and constitutively active RAB19 (RAB19^Q76L) in which the GTP hydrolysis activity is inhibited. IgG was used as a control. Proteomics generated a list of 2066 proteins, which we narrowed down based on the following criteria: 1) absent from IgG sample or at least 6-fold more enriched in RAB19 samples; 2) at least 2-fold more enriched in RAB19-GTP or constitutively active RAB19 than RAB19-GDP sample. We then eliminated proteins that were high on the CRAPome (a contaminant repository for affinity purification-mass spectrometry data) (Mellacheruvu et al., 2013), mitochondrial, or involved in transcription. The complete list of proteomics hits can be found in Table S1.

Protein expression and purification

PCR primer sequences are listed in the Key Resources Table. Mouse RAB19 cDNA was a gift from Dr. Marino Zerial (Lütcke et al., 1995). Human GST-RAB11a and rat GST-RAB14 were gifts from Dr. Jagath Junutula (Genentech). Human TBC1D1 and TBC1D4 cDNA were obtained from Harvard PlasmID Repository. N-terminal, middle region, and C-terminal truncations of TBC1D1 and TBC1D4 were generated using the YASPIN secondary structure prediction tool (IBIVU) to avoid disrupting structured regions. The following amino acid regions were used: TBC1D1-N (1–380), TBC1D1-Middle (381–783), TBC1D1-C (785–1168), TBC1D4-N (1–505), TBC1D4-Middle (506–898), TBC1D4-C (899–1298). Full-length and truncation constructs of GST-tagged proteins were generated through restriction enzyme cloning into expression vector pGEX-KG and purified using the BL21 Codon Plus E. coli strain. E. coli cultures were grown at 16°C and induced with 125 μM IPTG overnight. Cells were lysed using a French press and then incubated with glutathione agarose beads. Beads were washed with wash buffer (PBS containing 0.1% Tween, 100 mM additional NaCl). To reduce contamination from heat-shock proteins, beads were incubated with 5mM ATP at 37°C for 15 minutes prior to elution with 25 mM glutathione or cleavage with thrombin. For purification of RAB proteins, 5 mM MgCl2 was added to all buffers, EDTA was excluded, and proteins were kept in reducing conditions. Protein concentration was measured with a Bradford assay and validated by SDS-PAGE and Coomassie staining. We were unable to successfully purify the middle region fragment and full-length protein for both TBC1D1 and TBC1D4, thus only the N- and C-terminal fragments were used for experiments.

Generation and purification of TBC1D1 and TBC1D4 polyclonal antibodies

Purified GST-TBC1D1 and TBC1D4 N- and C-terminal truncations were sent to Pocono Rabbit Farm and Laboratory, Inc. Rabbit anti-sera was affinity purified with Affigel beads and eluted with 0.1 M Glycine, pH 2.5. Both N- and C-terminal antibodies were validated via Western Blot using knockout cell lines.

GST pulldown assays

GST pulldown assay were performed as previously described (Willenborg et al., 2011). For pulldowns with GST-RAB19, nucleotide loading was performed as described in proteomics (See Immunoprecipitation and proteomics analysis of RAB19). After elution with 1% SDS, samples were analyzed by SDS-PAGE and stained with Coomassie blue or immunoblotted with fluorescent secondaries. Blots were imaged on a LI-COR Odyssey Scanner and analyzed in FIJI.

GAP assay

Purified recombinant RAB11, RAB14, RAB19, TBC1D1-C, TBC1D4-C, TBC1D1-C^R854K, and TBC1D4-C^R972K were dialyzed into 20 mM HEPES and used according the CytoPhos Phosphate Assay Kit (Cytoskeleton, Inc. BK054). Briefly, RAB proteins were GTP loaded and 3 μg of each protein was added to the reaction mixture along with 0.3 mM GTP. Absorbance was measured at 650 nm 10 minutes after addition of GTP using a BioTek Cytation 3 plate reader. Reactions were performed in technical triplicate and the average value was used for analysis. The experiment was performed at least 5 times for each condition.

Glycerol gradient fractionation assay

Twelve glycerol fractions with concentrations ranging from 23–45% increasing in 2% increments were prepared in resuspension buffer (20mM HEPES pH 7.4, 100 mM KCl, 1 mM MgCl2, 1 mM DTT, 1% Triton X-100). Glycerol fractions were prechilled to 4°C and then layered in 100 μL volumes starting with the heaviest fraction in prechilled tubes. Cell lysates were GTP/GDP loaded and then layered on top of the glycerol fractions. Samples were spun at 40,000 × g for 18 hours and fractions were disassembled in 100 μL volumes, mixed with 4x SDS loading dye and analyzed by Western Blot. Molecular weight standard and purified thyroglobulin fractions were prepared similarly and then analyzed by Coomassie.

Immunofluorescence

For polarized MDCK ciliogenesis experiments, 50,000 cells were plated on collagen type I-coated Transwell filters (Corning 3460) as described in (Ott and Lippincott-Schwartz, 2012) and grown for 8 days to polarize and reach full ciliation capacity. For actin drug treatments, cells were treated for 48 hours prior to fixation. In some experiments noted in the text, cells were plated on collagen-coated glass coverslips. For RPE-1 ciliogenesis experiments, cells were plated on collagen type I-coated coverslips and grown to about 85% confluency. To induce ciliogenesis, cells were washed once with 1x PBS and then grown in serum depleted media (DMEM with 0.5% FBS) for 48 hours. For actin drug treatments, cells were treated for 48 hours in serum depleted media prior to fixation. Cells were fixed with 4% paraformaldehyde for 20 minutes at room temperature or with 100% ice cold Methanol for 10 minutes at −20°C. Cells were blocked for 1–2 hours in block buffer (PBS, 0.1% Triton X-100, 10% normal donkey serum). Primary antibodies were diluted in block buffer and incubated overnight at room temperature. Cells were washed with PBSTx before adding secondary antibodies for 1–2 hours at room temperature. Cells were washed again before mounting in VectaShield and sealing with nail polish or mounting in Prolong Gold. Coverslips for all experiments were #1.5. Primary and secondary antibodies and drugs are listed in the Key Resources Table.

Confocal microscopy and image analysis

Confocal images were acquired using a Nikon Eclipse Ti 2 inverted A1 Confocal with a 100x TIRF objective (NA 1.45). For XZ projections, images were captured at Nyquist sampling and reconstructed in FIJI. Images were processed in FIJI (Schindelin et al., 2012) and Adobe Photoshop. Figures were made in Adobe Illustrator. A minimum of three biological replicates were performed for each experiment.

Structured illumination microscopy (SIM)

Super resolution imaging was performed on the Nikon 3D SIM system (Ti 2 Eclipse) with a 100x TIRF objective (NA 1.45). Images were captured with a complementary metal-oxide semiconductor camera (Orca-Flash 4.0; Hamamatsu). Raw images were reconstructed using the Nikon Elements image stack reconstruction algorithm.

Time-lapse imaging

Time-lapse imaging was performed using an inverted microscope (Ti Eclipse) with a 100× Plan-Apochromat (NA 1.43) objective lens (Nikon) and a spinning disk confocal module (CSU-X1; Yokogawa). Images were captured with a charge-coupled device (CCD) camera (iXon X3; Andor Technology). Time-lapse movies were acquired with Slidebook (3i - Intelligent Imaging Innovations) with a z-step size of 0.3 μm at 15 minutes interval. Temperature (37°C), CO2 (5%) and humidity were controlled using the LiveCell microscope stage incubator (Pathology Devices, Inc.).

Transmission electron microscopy

Cells were serum starved for 24 hours and fixed with 2% glutaraldehyde in 0.1 M sodium cacodylate buffer for 1 hour at room temperature. After three washes in 0.1 M cacodylate buffer, samples were post-fixed in 1% osmium tetroxide in 0.1 M cacodylate buffer for 30 minutes. Following subsequent washes with ddH₂O, cells were pre-stained with 0.5% uranyl acetate for 1 hour. Then cells were dehydrated in graded ethanol series and embedded in Polybed 812 resin (PolyScience). 70 nm sections were cut using a Leica EM UC7 microtome and stained with uranyl acetate and lead citrate. Electron micrographs were acquired using a transmission electron microscope (FEI Tecnai T-12).

Trans-epithelial resistance (TER) measurements

MDCK cells were grown on collagen-coated transwell filters (see Cell Culture section) and resistance measurements were taken every 24 hours for 8 days using a Millicell ERS-2 Voltohmmeter (Millipore). Three measurements per well, one for each space between the plastic prongs of the filter holder, were averaged and subtracted from the average of the blank well containing a collagen-coated filter without any cells.

Generating MDCK and RPE-1 CRISPR knockout lines

MDCK cells stably expressing Tet-inducible Cas9 (Dharmacon Edit-R inducible lentiviral Cas9 nuclease) were grown in a 12-well dish to about 75% confluency before treatment with doxycycline at a final concentration of 1 μg/mL for 24 hours to induce Cas9 expression. Cells were then transfected with crRNA:tracrRNA mix as described for DharmaFECT Duo co-transfection protocol (Horizon Discovery Cat# T-2010-xx). Transfected cells were incubated for 24 hours before trypsinizing and plating for individual clones. Individual clones were screened through genotyping PCR and sanger sequencing. To generate RPE-1 RAB19 knockout line, the same protocol was used except instead of an inducible Cas9 line, the crRNA:tracrRNA mix was co-transfected with Edit-R mKate2-Cas9 nuclease mRNA (Horizon Discovery #CAS11859). FACS was performed 24 hours after transfection and mKate2 positive cells were plated for individual clones which were then genotyped. All CRISPR gRNAs and genotyping primers are listed in the Key Resources Table. Primers were designed with the NCBI Primer-BLAST tool (Ye et al., 2012). Sequences of CRISPR alleles are listed in Table S2.

Flow cytometry and fluorescence activated cell sorting (FACS)

All cell sorting experiments were performed at the University of Colorado Cancer Center Flow Cytometry Shared Resource Facility using a MoFlo XDP 100. For expression of fluorescently labeled proteins, WT and mutant cell lines were gated equivalently. To measure DNA content per cell, cells were prepared as described in (Krishan, 1975) and analyzed using a Beckman Coulter FC500 flow cytometer. Doublets were excluded from the analysis using the peak versus integral gating method. ModFit LT software (Verity Software House, Topsham, ME) was used for cell cycle analysis.

IN/OUT cilia assay

MDCK WT and RAB19 KO cell stably expressing pHluorin-Smoothened (pLVX-pHluorin-Smoothened vector was a gift from Dr. Derek Toomre) were plated on collagen-coated filters, fixed after 4 days and stained as described in (Kukic et al., 2016). Briefly, culture media was removed, cells were gently washed in PBS, and then fixed in 4% PFA in PBS for 10 minutes. Cells were then blocked in 5% normal donkey serum in PBS for 30 minutes followed by incubation with anti-GFP primary antibody solution for 1 hour. Cells were fixed again for 10 minutes, permeabilized with 0.1% Triton-X, and incubated in another primary antibody solution containing anti-Actub for 1 hour. After gentle washing, cells were incubated with secondary antibodies and Hoechst, followed by washing and mounting on slides. For RPE1 WT and RAB19 KO cells, cells were plated on collagen coated glass coverslips, serum starved for 48 hours, fixed, and stained in the same way as above.

Genotyping zebrafish

Fish tissue was placed in lysis buffer (10 mM Tris pH 8.3, 50 mM KCl, 1.5 mM MgCl₂, 0.3% Tween-20, 0.3% NP-40 in water) with 2% Proteinase K (Invitrogen Cat # 25530049). Lysis reactions were placed at 55°C for 4 hours, then 95°C for 20 minutes to inactivate Proteinase K. Tubes were mixed gently, then spun down briefly and stored at 4°C. A PCR/Restriction Enzyme-based assay was used to genotype RAB19^sa24592 mutant fish lines. A 200 bp region of genomic DNA surrounding the mutation site was amplified by PCR. The PCR product was then digested with Hpy188IM for 1 hour at 37°C, and the resulting product was run on a 2% agarose gel. The restriction enzyme Hpy188III cuts WT genomic DNA into 2 pieces but cannot cut the mutant sequence, thus the banding pattern was indicative of WT, heterozygous, or homozygous fish. PCR primer sequences are listed in the Key Resources Table. Primers were designed with the NCBI Primer-BLAST tool (Ye et al., 2012).

Zebrafish HCR in situ

Expression of RAB19 mRNA in zebrafish embryos was detected using fluorescent hybridization chain reaction (HCR) in situ hybridization (Choi et al., 2014). Custom probes against RAB19 (LOC555645) mRNA were designed by Molecular Instruments (www.molecularinstruments.com), and care was taken to ensure that probes overlapped regions shared across previously annotated RAB19 isoforms. Whole mount in situ hybridization was performed following published protocols for zebrafish (Molecular instruments: MI-Protocol-HCRv3-Zebrafish). Briefly, WT zebrafish embryos were reared at 28°C to 3 dpf in 0.003% 1-phenyl-2-thiourea (PTU) diluted in embryo medium to inhibit pigment cell development. Embryos were fixed in 4% paraformaldehyde overnight at 4°C and dehydrated in 100% methanol at −20°C. Prior to hybridization, embryos were briefly digested in a 10 μg/mL Proteinase K solution for 10 minutes at room temperature. Custom RAB19 probes were diluted in hybridization buffer to a final concentration of 4 picomoles/mL, and hybridization was performed at 37°C overnight. HCR amplification was performed overnight at room temperature using an Alexafluor647 label and diluting hairpin amplification oligos to a final concentration of 60 picomoles/mL. Embryos were mounted in 80% glycerol and imaged using a Leica TCS SP8 confocal microscope. All image stitching and processing was performed in LASX software.

QUANTIFICATION AND STATISTICAL ANALYSIS

Time-lapse imaging analysis

Image analysis: (I) centrosome relative position to cortical actin, (II) cortical actin intensity at and flanking the centrosome, (III) width of cortical actin clearance and (IV) cortical actin thickness. Fluorescence quantification of centrosome position relative to cortical actin was performed by a semi-automated strategy that utilizes the FIJI macro scripting language and plugins (Schindelin et al., 2012). All movies were cropped to generate image stacks that contain one cell per field of view.

(I) To track centrosomes, the cartesian coordinate (XYZ) of the brightest pixel within each centrosome was extracted via the FIJI Trackmate plugin (Tinevez et al., 2017). To remove spurious noise, all image stacks were processed with a median filter. Centrosome cartesian coordinates were confirmed by visual inspection for all time frames. Frames with inaccurate centrosome coordinates were excluded from analysis. Based on the centrosome XY coordinates, a 3D image stack (5.1 μm × 5.1 μm × 5.1 μm) that centers around the centrosome was cropped for each time frame. The relative z-distance between the centrosome and cortical actin was quantified based on the brightest cortical actin plane. To generate the average plots depicted in the figures, all movies were aligned by the frame when the centrosome is 2 planes (0.6 μm) below the brightest cortical actin plane.

(II) To measure actin fluorescence intensity at the centrosome, a 5 × 5-pixel box (0.5 μm × 0.5 μm) was centered over the centrosome and the total actin fluorescence intensity within the box was measured. The actin fluorescence intensity flanking the centrosome was sampled using a 2-pixel wide region that is radially positioned 2.5 μm away from the centrosome. A correction factor was applied to account for size differences of the sampling boxes used to measure the actin fluorescence intensity at and flanking the centrosome. For both measurements, the actin fluorescence intensities were sampled at the brightest apical cortical actin plane. Background actin fluorescence intensity within the cell was sampled with a 5 × 5-pixel box positioned at 2 z-planes (0.6 μm) below the brightest apical actin plane and subtracted. The same actin fluorescence intensity measurement strategy was applied to all time frames. Actin fluorescence intensity was normalized to the first time point.

(III and IV) To measure the width of cortical actin clearance, a fluorescence-based criterion was used. Using the Fiji Plot Profile plugin, we measured the drop in cortical actin fluorescence intensity at the brightest cortical actin plane (intensity value ≤ 25% from the peak fluorescence intensity qualifies for length measurement). To avoid spurious noise, a 11-pixel (1.2 μm) wide line that spans across the centrosome was used. To quantify for the thickness of cortical actin, a similar strategy was performed. However, all image stacks were first pre-processed to project the XZ dimension (FIJI Z-functions plugin; Reslice). Cortical actin thickness was measured at two regions that are positioned 2.2 μm away from the center of the centrosome using a 11-pixel (1.2 μm) wide line and averaged (intensity value ≥ 75% from the peak fluorescence intensity qualifies for length measurement). Peak actin fluorescence intensity was normalized to 1 for each time frame.

Statistical Analysis

All graphs show mean ± SEM unless otherwise indicated. Biological replicates from separate culture passages are denoted as n_B and technical replicates from separate fields of view or cells are denoted as n_T in the figure legends. Datasets were assessed for normal distribution using the Shapiro-Wilk normality test. A t-test was performed on all normally distributed datasets and Mann-Whitney U test was performed for datasets not normally distributed.

Supplementary Material

2

Movie S1. Centrosome migration and cortical actin clearing in MDCK WT cells, Related to Figure 1. In WT MDCK cells, the centriole migrates towards the cell cortex and cortical actin clearance occurs. Movie illustrates the XY (top) and XZ (side) projections. Image has been smoothed for visualization of process. XY projection dimensions, 5.5 × 5.5 μm. XZ projection dimensions, 5.5 × 5.1 μm. Frame interval, 36 minutes.

3

Movie S2. Centrosome migration and cortical actin clearing in MDCK RAB19 KO cells, Related to Figure 3. In RAB19 KO cells, the centriole migrates towards cell cortex and cortical actin clearance does not occur. Movie illustrates the XY (top) and XZ (side) projections. Image has been smoothed for visualization of process. XY projection dimensions, 5.5 × 5.5 μm. XZ projection dimensions, 5.5 × 5.7 μm. Frame interval, 60 minutes.

4

Table S1. List of proteins from mass spectrometry on RAB19 interacting partners, Related to Figure 5.

KEY RESOURCES TABLE.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Centrin clone 20H5	Sigma	Cat#04-1624
ARL13B clone N295B/66	Antibodies Inc	Cat#73-287
γ-Tubulin polyclonal (DQ-19)	Sigma	Cat#T3195
γ-Tubulin monoclonal	Sigma	Cat#T5326
ODF2	Sigma	Cat#HPA001874
Acetylated-tubulin polyclonal	Cell Signaling	Cat#5335
Acetylated-tubulin monoclonal	Sigma	Cat#T7451
Podocalyxin/GP135	DSHB	Cat#3F2/D8
Pericentrin (PCNT)	Abcam	Cat#ab4448
Flag	Sigma	Cat#F3165
VPS41 (E-10)	Santa Cruz Biotech	Cat#sc-377271
VPS11 (C-12)	Santa Cruz Biotech	Cat#sc-515094
PCM1	Cell Signaling	Cat#5213
RAB11	Life Technologies	Cat#715300
GT335	Adipogen	Cat#AG-20B-0020-C100
CEP164	Santa Cruz Biotech	Cat#sc-240226
RPGRIPL1	Protein Tech	Cat#55160-1-AP
IFT140	Protein Tech	Cat#17460-1-AP
IFT88	Protein Tech	Cat#13967-1-AP
CD63	Gift from Andrew Peden	N/A
CP110	Proteintech	Cat#12780-1-AP
GFP	Life Technologies	Cat#A-11120
FIP5	Prekeris Lab	N/A
Cingulin (CGN)	Prekeris Lab	N/A
β-Catenin	Transduction labs	Cat#C19220
TBC1D1	Gift from Makoto Kanzaki	N/A
TBC1D4-Nterm	This paper	N/A
TBC1D4-Cterm	This paper	N/A
Alexa 488 Anti-Rabbit secondary	Jackson ImmunoResearch	Cat#711-545-152
Alexa 594 Anti-Rabbit secondary	Jackson ImmunoResearch	Cat#711-585-152
Alexa 488 Anti-Mouse secondary	Jackson ImmunoResearch	Cat#715-545-150
Alexa 594 Anti-Mouse secondary	Jackson ImmunoResearch	Cat#715-585-150
Alexa 488 Anti-Mouse IgG2a Secondary	Invitrogen	Cat#A-21131
Alexa 488 Anti-Mouse IgG1 Secondary	Invitrogen	Cat#A-21121
Alexa 568 Anti-Mouse IgG1 Secondary	Invitrogen	Cat#A-21124
Alexa 568 Anti-Mouse IgG2b Secondary	Invitrogen	Cat#A-21144
Alexa 647 Anti-Rabbit Secondary	Invitrogen	Cat#A-21245
Alexa-568 Phalloidin	Invitrogen	Cat#A12380
Chemicals, Peptides, and Recombinant Proteins
Y27632	Enzo Life Sciences	Cat#ALX-270-333-M001
Cytochalasin D	ThermoFisher	Cat#PHZ1063
Nocodazole	Millipore Sigma	Cat#M1404
SirActin	Cytoskeleton	Cat#CY-SC001
Critical Commercial Assays
CytoPhos Endpoint Phosphate Assay	Cytoskeleton	Cat#BK054
Experimental Models: Cell Lines
MDCK II	ATCC	CRL-2936
hTERT RPE-1	ATCC	CRL-4000
mIMCD-3	ATCC	CRL-2123
Experimental Models: Organisms/Strains
Zebrafish rab19sa24592 allele	Zebrafish International Resource Center	ZFIN ID: ZDB-ALT-140106-279
Oligonucleotides
ZFish RAB19 Genotyping Forward: TGTTCGAGGACGCCTGTACG	This paper	N/A
ZFish RAB19 Genotyping Reverse: TCCTTCGATTGGGTGTGAGTC	This paper	N/A
RPE-1 RAB19 KO Genotyping Forward: CATTACCTGTGGGAACCCCT	This paper	N/A
RPE-1 RAB19 KO Genotyping Reverse: TCCACTGTCATTAGAAACATGCAGA	This paper	N/A
MDCK RAB19 KO Genotyping Forward: TGGGAACTCCTGATGCGAAC	This paper	N/A
MDCK RAB19 KO Genotyping Reverse: CTAAGCCAAACGCTTTCCGAG	This paper	N/A
MDCK TBC1D1 KO Genotyping Forward: GAGTTCGATGACACGTTCGC	This paper	N/A
MDCK TBC1D1 KO Genotyping Reverse: TGAAGAGCATAGTTCGATTGTCCT	This paper	N/A
MDCK TBC1D4 KO Genotyping Forward: ATATGTAAACACAGACCCCACCC	This paper	N/A
MDCK TBC1D4 KO Genotyping Reverse: AGAGGAGAGCGAACACCATGA	This paper	N/A
MDCK VPS41 KO Genotyping Forward: GCTGGTGAAGTTGTAGCAGT	This paper	N/A
MDCK VPS41 KO Genotyping Reverse: CTGGAAAAAGCCACCACCAAG	This paper	N/A
pLVX-GFP-RAB19 Forward: CCGGCGAATTCCAGTTCTCCAGCTCATCCA	This paper	N/A
pLVX-GFP-RAB19 Reverse: AGGGCCCTCAACAAGTACAGCGGGTG	This paper	N/A
pLVX-GFP-RAB19T31N SDM Forward: GCTGAACCACACAGTTCTTCCCCACGTTGGAGTCCC	This paper	N/A
pLVX-GFP-RAB19T31N SDM Reverse: GGGACTCCAACGTGGGGAAGAACTGTGTGGTTCAGC	This paper	N/A
pLVX-GFP-RAB19Q76L SDM Forward: GACACGGCAGGCCTGGAGCGCTTCC	This paper	N/A
pLVX-GFP-RAB19Q76L SDM Reverse: GGAAGCGCTCCAGGCCTGCCGTGTC	This paper	N/A
pLVX-Flag-RAB19 Forward: TATAGAATTCTCAGTTCTCCAGCTCATCC	This paper	N/A
pLVX-Flag-RAB19 Reverse: ATCTAGATTAACAAGTACAGCGGGT	This paper	N/A
pLVX-GFP-TBC1D4 Forward: AGGGCCCGAGCCGCCCAGCT	This paper	N/A
pLVX-GFP-TBC1D4 Reverse: GCTCTAGATTATGGCTTATTTCCTATCTT	This paper	N/A
pLVX-GFP-TBC1D4R972K SDM Forward: GAGTAGGAAACGTCTTTCCTAAATCCACGAGAATCG	This paper	N/A
pLVX-GFP-TBC1D4R972K SDM Reverse: CGATTCTCGTGGATTTAGGAAAGACGTTTCCTACTC	This paper	N/A
pGEX-KG-RAB19 Forward: CCGGCGAATTCTACAGTTCTCCAGCTCAGCCA	This paper	N/A
pGEX-KG-RAB19 Reverse: TACAAGCTTTCAACAAGTACAGCGGGT	This paper	N/A
pGEX-KG-TBC1D1 Full Forward: CCGCCATGGGTGAACCAATAACATTCACAGCAAGG	This paper	N/A
pGEX-KG-TBC1D1 Full Reverse: ATCCCATGGTTAGTCGCCCGTGGGCTCG	This paper	N/A
pGEX-KG-TBC1D1-N 1-380 Forward: CCGGGAATTCTAGAACCAATAACATTCACAGCAAGG	This paper	N/A
pGEX-KG-TBC1D1-N 1-380 Reverse: ATCTCTCGAGTTAAGCTGTCTGCTGCACTG	This paper	N/A
pGEX-KG-TBC1D1-Middle 381-783 Forward: TATCTAGACAAGGCGCCAGCCCAGCT	This paper	N/A
pGEX-KG-TBC1D1-Middle 381-783 Reverse: TATACTCGAGTTATCTTCCTGGAGTGCTAAGCATCTTTTCCCA	This paper	N/A
pGEX-KG-TBC1D1-C 785-1168 Forward: CCGGGAATTCTATCAAAAATTAAGTTTGACATGGAAAAAATGCA	This paper	N/A
pGEX-KG-TBC1D1-C 785-1168 Reverse: TATGAGCTCTTATCTTCCTGGAGTGCTAAGCATCTTTTCC	This paper	N/A
pGEX-KG-TBC1D1R854K SDM Forward: ATGGGTGTGTAGGAAAGGTTTTCCCAAGGTCAATAAGAATCG	This paper	N/A
pLVX-GFP-TBC1D1R854K SDM Reverse: CGATTCTTATTGACCTTGGGAAAACCTTTCCTACACACCCAT	This paper	N/A
pGEX-KG-TBC1D4 Full Forward: TATCTAGACGAGCCGCCCAGCTGC	This paper	N/A
pGEX-KG-TBC1D4 Full Reverse: TATGAGCTCTTAGTCGCCCGTGGGCTCG	This paper	N/A
pGEX-KG-TBC1D4-N 1-505 Forward: CCGGGAATTCTAGAGCCGCCCAGCTGCAT	This paper	N/A
pGEX-KG-TBC1D4-N 1-505 Reverse: ATCTCTCGAGTTAGACTGGCTTCATTTTCTGAACTCT	This paper	N/A
pGEX-KG-TBC1D4-Middle 506-898 Forward: TATCTAGACAGTGACCAGGAAGAAAATG	This paper	N/A
pGEX-KG-TBC1D4-Middle 506-898 Reverse: TATACTCGAGTTATAACAACTTCTTATCCCAAG	This paper	N/A
pGEX-KG-TBC1D4-C 899-1298 Forward: CCGGGAATTCTAAACTGCAGAGCTAAAATCAG	This paper	N/A
pGEX-KG-TBC1D4-C 899-1298 Reverse: TATGAGCTCTTAGTCGCCCGTGGGCTCG	This paper	N/A
pLVX-GFP-VPS41 Forward: AGGGCCCGCGGAAGCAGAGGAGCAG	This paper	N/A
pLVX-GFP-VPS41 Reverse: GCTCTAGACTATTTTTTCATCTCCAAAATTGCACTTCC	This paper	N/A
MDCK RAB19 KO gRNA: GTGAGGAACAGACTCGAAGG	This paper	N/A
MDCK TBC1D1 KO gRNA: TTTAAGAAGGAATTGCACGA	This paper	N/A
MDCK TBC1D4 KO gRNA: CTGTTCTTCTCAGAATCCGA	This paper	N/A
MDCK VPS41 KO gRNA: AACCCAAACTGAAGTATGAA	This paper	N/A
RPE1 RAB19 KO gRNA: TCGCCCTTGCTCACGCTGCCACCAGCCATC	This paper	N/A
Recombinant DNA
pLVX-puro	Clontech	Cat#632164
pGEX-KG	Prekeris Lab	N/A
Software and Algorithms
FIJI	Schneider et al., 2012	https://imagej.nih.gov/ij/
Primer-BLAST	Ye et al., 2012	https://.ncbi.nlm.nih.gov/tools/primer-blast/
Adobe Illustrator	Adobe Inc	N/A
Other

Highlights.

• Cortical clearing of actin and plasma membrane is required for primary ciliogenesis.

• Centrosome migration to the cell cortex drives cortical clearing in polarized cells.

• A RAB19 trafficking complex coordinates cortical clearing with cilia growth.

• Inhibition of actin contractility can substitute functionally for the RAB19 complex.