The Rab GTPase-binding protein EHBP1L1 and its interactors CD2AP/CIN85 negatively regulate the length of primary cilia via actin remodeling

Primary cilia are organelles consisting of axonemal microtubules and plasma membranes, and they protrude from the cell surface to the extracellular region and function in signal sensing and transduction. The integrity of cilia, including the length and structure, is associated with signaling functions; however, factors involved in regulating the integrity of cilia have not been fully elucidated. Here, we showed that the Rab GTPase-binding protein EHBP1L1 and its newly identified interactors CD2AP and CIN85, known as adaptor proteins of actin regulators, are involved in ciliary length control. Immunofluorescence microscopy showed that EHBP1L1 and CD2AP/CIN85 are localized to the ciliary sheath. EHBP1L1 depletion caused mislocalization of CD2AP/CIN85, suggesting that CD2AP/CIN85 localization to the ciliary sheath is dependent on EHBP1L1. Additionally, we determined that EHBP1L1- and CD2AP/CIN85-depleted cells had elongated cilia. The aberrantly elongated cilia phenotype and the ciliary localization defect of CD2AP/CIN85 in EHBP1L1-depleted cells were rescued by the expression of WT EHBP1L1, although this was not observed in the CD2AP/CIN85-binding–deficient mutant, indicating that the EHBP1L1–CD2AP/CIN85 interaction is crucial for controlling ciliary length. Furthermore, EHBP1L1- and CD2AP/CIN85-depleted cells exhibited actin nucleation and branching defects around the ciliary base. Taken together, our data demonstrate that the EHBP1L1–CD2AP/CIN85 axis negatively regulates ciliary length via actin network remodeling around the basal body.

Primary cilia are antenna-like structures consisting of central axonemal microtubules and membranes. They are exposed to the extracellular environment and function as signaling centers (1,2). Mutations in ciliary protein-encoding genes, which regulate ciliary length, traffic, and signal transductions, cause a number of disorders termed ciliopathies (3).
Previous studies have described two pathways of ciliogenesis: extracellular and intracellular (2). The extracellular pathway starts with docking of the basal body to the plasma membrane, which occurs mostly in epithelial polarized cells. After docking, the ciliary membrane elongates, accompanied by an axonemal microtubule extension. The intracellular pathway, which is predominant in fibroblasts, neural progenitor cells, and retinal pigment epithelium (RPE) cells, is initiated by the accumulation of preciliary vesicles (PCVs) on the distal appendages of the mother centriole, which is known as the basal body. PCVs fuse together to form a ciliary vesicle (CV). The CV then becomes a nascent ciliary membrane with two distinct domains, ciliary sheath and shaft, surrounding the growing axoneme. Finally, the antenna-like ciliary structure is formed by the fusion of the nascent ciliary membrane to the plasma membrane, followed by membrane and axoneme elongation (4,5).
Rab GTPases function in various membrane trafficking pathways. GTP-bound Rab proteins recruit specific binding proteins to the membrane and regulate trafficking events such as vesicle fusion, fission, tethering, motility, and membrane remodeling (6).
Rab8 is a member of the Rab family, which is conserved among eukaryotes, and it functions in polarized traffic. In mammalian epithelial cells, Rab8 localizes to post-Golgi vesicles, tubules, or the endocytic recycling compartment (ERC) and regulates exocytic trafficking to the plasma membrane (7,8). Rab8 also localizes and plays a role in the cilia, which are also regarded as polarized membrane domains (9,10). However, the precise role that Rab8 plays in cilia remains unclear.
We investigated the involvement of EHBP1L1 in primary cilia integrity, in addition to its role in the ERC and exocytic trafficking in epithelial cells. In this study, we demonstrated that EHBP1L1 and its newly identified binding proteins CD2AP and CIN85 localize to the ciliary sheath membrane, which is a subdomain of the CV, and negatively regulate ciliary length. Our study also showed that depletion of EHBP1L1 and CD2P/CIN85 causes actin nucleation and branching defects around the basal body, suggesting that actin network formation is crucial for maintaining an appropriate length of primary cilia.

Results
EHBP1L1 localizes in the ciliary sheath during early primary ciliogenesis and negatively regulates ciliary length To determine whether EHBP1L1 is involved in primary cilia formation, two cell lines, hTERT-RPE1 and NIH3T3, whose ciliogenesis is initiated via the intracellular pathway, were immunostained with an EHBP1L1 antibody. Conventional confocal microscopy images showed that EHBP1L1 was localized around the primary cilia, as indicated by acetylated tubulin (Fig. 1A). Under serum starvation conditions, 91.7% of RPE-1 cells possessed primary cilia while only 24.0% of cilia in ciliated cells were EHBP1L1-positive (Fig. 1B). The majority of EHB1P1L1-positive cilia were colabeled with myosin-Va (Myo-Va), which temporarily associates with the nascent ciliary membrane during early ciliogenesis and disappears from the developed cilia in RPE-1 cells (5) (Fig. 1B). Moreover, the lengths of EHBP1L1-positive cilia, as indicated by acetylated tubulin, were significantly shorter than those of EHBP1L1-negative cilia (2.77 ± 0.60 μm and 3.34 ± 0.63 μm, respectively) (Fig. 1C). Additional analysis was conducted using super-resolution structured illumination microscopy (SR-SIM). SR-SIM analysis revealed that EHBP1L1 localized to the compartment surrounding acetylated tubulin-positive and Arl13b-positive structures that represent the axoneme and the inside of the ciliary membrane called the ciliary shaft membrane during early ciliogenesis, respectively (5) (Fig. 1, D and E). EHBP1L1 distribution overlapped well with Myo-Va, which was previously shown to localize to the ciliary sheath, which is the outer region of the growing ciliary membrane (5) (Fig. 1, D and E). These data suggest that EHBP1L1 localizes to the ciliary sheath and functions during early ciliogenesis (Fig. 1E).
Next, we determined whether EHBP1L1 plays a role in primary ciliogenesis. To do this, we depleted EHBP1L1 protein using two independent siRNA oligonucleotides ( Fig. 2A). Compared with the length of primary cilia in control cells (3.18 ± 0.50 μm), EHBP1L1-depleted cells had longer cilia (siRNA #1:3.76 ± 0.59 μm; siRNA #2:3.63 ± 0.60 μm) (Fig. 2, B and C). These data suggest that EHBP1L1 is involved in the shortening of primary cilia. To identify the region in EHBP1L1 that is involved in the regulation of primary ciliary length, a series of Flag-tagged deletion mutants (Figs. 2D and S1) were expressed in EHBP1L1-depleted cells, and phenotypic rescue of ciliary length was assessed. Of the four proteins targeted to the ciliary region (EHBP1L1 full-length, ΔPR, ΔCH, and ΔC2) (Fig. S1), only ΔPR-positive cilia were longer than the endogenous EHBP1L1-positive cilia in control cells (2.66 ± 0.45 μm), and their lengths were similar to those of cilia in EHBP1L1-depleted cells (3.49 ± 0.51 μm and 3.59 ± 0.85 μm, respectively) (Fig. 2E). These data indicate that the ΔPR mutant failed to rescue the ciliary phenotype (length) and suggest that the PR region of EHBP1L1 is necessary for the control of ciliary length.
We also demonstrated that the EHBP1L1 mutant lacking the bMERB domain (ΔbMERB) necessary for GTP-Rab8 binding failed to localize to CVs (Fig. S1). Thus, these data clearly indicate the importance of Rab8 in ciliary targeting of EHBP1L1-CD2AP/CIN85.

CD2AP/CIN85 proteins interact with EHBP1L1
To gain further insights into the function of EHBP1L1 in regulating ciliary length, we attempted to identify the protein that binds to its PR region. Glutathione S-transferase (GST) pull-down assay using the PR region of EHBP1L1 generated several coprecipitated proteins that were absent in the assay conducted using the C2 domain (Fig. 3A). Mass spectrometry analysis revealed that Bin1, CD2AP, CAPZA1, CAPZA2, and CAPZB were coprecipitated with the PR region of EHBP1L1 (Fig. 3B). We previously reported that the EHBP1L1-Bin1dynamin axis regulates exocytic trafficking in the endocytic recycling compartment (11). However, the relationship between EHBP1L1, an adaptor for the actin network regulator CD2AP, and the barbed-end actin capping protein CAPZ has not been explored. Thus, we focused on CD2AP and CAPZ proteins.
Given that CAPZ proteins reportedly bind CD2AP and its paralog CIN85 via capping protein interaction motifs (15,16) and CD2AP/CIN85 have three SRC homology 3 (SH3) domains (SH3-A, SH3-B, and SH3-C) (Fig. 3C) that are known to bind to PR sequences (17,18), the CD2AP-CAPZ complex was co-pulled down with the EHBP1L1 PR region via the SH3 domain in CD2AP. We examined the direct interaction between CD2AP and the EHBP1L1 PR region using the GST pull-down assay. The results clearly showed direct binding of the full-length and N-terminal regions of CD2AP to GST-EHBP1L1-PR (Fig. 3, C and D). Further experiments demonstrated that the second SH3 domain (SH3-B) interacted directly with GST-EHBP1L1-PR (Fig. 3D). We also confirmed direct binding between the PR region of EHBP1L1 and CIN85 (Fig. 3, C and E).
Interaction between EHBP1L1 and CD2AP/CIN85 is crucial for EHBP1L1's regulation of ciliary length To determine the minimal EHBP1L1 PR region necessary for binding to CD2AP/CIN85, the PR region of EHBP1L1 was divided into four fragments, each containing several SH3binding proline-X-X-proline (PXXP) motifs (Fig. 4A). GST pull-down assays of these fragments bound to CD2AP-SH3-B revealed that the second PR region fragment (PR2) was required for binding to CD2AP-SH3-B (Fig. 4B). We further EHBP1L1 controls ciliary length via CD2AP/CIN85 examined the interaction between full-length CD2AP and CIN85 and either the PR region of EHBP1L1 (residues 442-595) or the mutant EHBP1L1-PR in which all four proline residues (positions 497 and 499-501) were replaced with alanine (EHBP1L1-PR P4A). Consistent with the data shown in EHBP1L1-P4A mutant (Fig. 4, E and F) in control and EHBP1L1 siRNA-treated cells. Protein expression was confirmed using immunoblotting (Fig. S1). Both WT and P4A mutant were localized around the cilia (Fig. S1). In control siRNA-treated cells, neither WT nor P4A mutant expression affected ciliary length (Fig. 4, E and F). In EHBP1L1 siRNAtreated cells, elongated cilia (also shown in Fig. 2) was suppressed by the expression of the WT protein but not the P4A mutant (Fig. 4, E and F). These data suggest that the interaction between EHBP1L1 and the CD2AP/CIN85 proteins is crucial for EHBP1L1's ability to maintain a certain length for primary cilia.
CD2AP/CIN85 proteins regulate ciliary length by remodeling the actin network To further assess the relationship between CD2AP/CIN85 and EHBP1L1, cells were costained with antibodies against CD2AP, CIN85, Arl13b, and EHBP1L1 and observed using SR-SIM. CD2AP was localized to the same region as CIN85, which was distinguishable from the Arl13b-positive ciliary shaft, whereas CIN85 overlapped with EHBP1L1 (Fig. 5A). These data indicate that CD2AP/CIN85 localized to the ciliary sheath together with EHBP1L1 (Fig. 5B). The number of CD2AP-and CIN85-positive cilia was significantly lower in EHBP1L1depleted cells. Ciliary localization defects in CD2AP and CIN85 in EHBP1L1-depleted cells were rescued by the expression of WT EHBP1L1 but not by the expression of the EHBP1L1-P4A mutant (Fig. 5, C and D). These results suggest that recruitment of CD2AP/CIN85 to the ciliary sheath is dependent on EHBP1L1. Next, we investigated the role of CD2AP/CIN85 in the determination of ciliary length using siRNA (Fig. 5E). Compared with control cells, CD2AP/CIN85depleted cells had longer cilia, and this trait was suppressed by exogenous expression of either CD2AP or CIN85 (Fig. 5F). These data indicate that the EHBP1L1-CD2AP/CIN85 axis controls ciliary length. Since CD2AP family proteins and the CAPZ complex regulate actin network formation by enhancing actin branching (19) and actin nucleation at the centrosome/basal body in an Arp2/Arp3-dependent manner (20), we examined the distribution of Arp2 to monitor the actin-branched network around the basal body in EHBP1L1-and CD2AP/CIN85depleted cells. We observed significantly reduced Arp2 fluorescence intensity around the basal body (Fig. 6, A and D), although the quantity of Arp2 protein did not change (Fig. 6C).
In addition, we also observed that fluorescent-dye-conjugated phalloidin-labeled actin filaments around the basal body in EHBP1L1-and CD2AP/CIN85-depleted cells had reduced signal intensity (Fig. 6, B and E). Taken together, these results suggest that EHBP1L1-CD2AP/CIN85-CAPZ-mediated actinbranched network formation around the basal body controls ciliary length.

Discussion
The lengths of primary cilia are often aberrant in cells carrying mutated ciliopathy-associated genes (21-31), suggesting that tight regulation of ciliary length is crucial for ciliary functions. The maintenance of ciliary length is dependent on the integrity of the ciliary membrane and axonemal microtubules (32,33). In addition, several lines of evidence have indicated that actin plays an important role in controlling ciliary length. Cells treated with drugs or siRNAs targeting actin-regulating genes that inhibit actin polymerization showed an increased ciliated cell number as well as elongated cilia (34)(35)(36).
Although the precise role of actin in ciliogenesis is unclear, several mechanisms have been proposed. For instance, the actin network around the centrosome/basal body may inhibit the vesicle trafficking necessary to form CVs or elongate the  . EHBP1L1 binds CD2AP family proteins. A, cell lysate was incubated with immobilized GST-C2 or PR (442-595 a.a.) region of EHBP1L1. Isolated proteins were analyzed using SDS-PAGE and silver staining. Arrows indicate specific proteins that coprecipitated with GST-PR but not GST-C2. B, summary of proteins identified using mass spectrometry. C, schematic drawing of domain structures in the CD2AP protein and its deletion forms (full-length, N and C) and CIN85. D, GST-pull-down (PD) assay using purified recombinant proteins. Hisx6-tagged full-length CD2AP, its deletion forms (Full, N, and C), MBP-tagged SH3 domains of CD2AP (SH3-A, -B, and -C), and GST-tagged PR region of EHBP1L1. GST protein was used as a control. The gels were stained with Coomassie brilliant blue (CBB). E, GST PD assay using GST or GST-PR and Hisx6-tagged CIN85. CC, coiled-coil domain; CPI, capping protein-interaction motif; GST, Glutathione S-transferase. PR, proline-rich region; SH3, Src-homology 3 domain.   Figure 5. EHBP1L1-CD2AP/CIN85 axis controls ciliary length. A, comparison of ciliary localization of CD2AP and CIN85 to Arl13b (ciliary shaft) and EHBP1L1 (ciliary sheath). The stained cells were imaged using super-resolution structured illumination microscopy (SR-SIM). The line scan profiles at positions marked with white lines are also shown. Scale bar represents 1 μm. B, graphical summary of the positional relationship between EHBP1L1 and other ciliary proteins. C, EHBP1L1-depleted cells and EHBP1L1-depleted cells expressing WT or P4A mutant EHBP1L1 were stained with antibodies against CD2AP, EHBP1L1 controls ciliary length via CD2AP/CIN85 ciliary membrane (37). Given the association between actin filaments and the basal body (20), it is likely that the actin network determines the position of the basal body close to the plasma membrane, a step that is important for ciliogenesis (4). CD2AP and the paralog CIN85 are adaptor proteins that interact with the membrane or signaling proteins via the SH3 domain (18) and with barbed-end actin capping proteins (CAPZs) via the capping protein interaction motif (16). Importantly, previous studies demonstrated that the C-termini of CD2AP family proteins have the ability to uncap CAPZ from actin filaments and subsequently polymerize actin (38,39). Moreover, CD2AP associates with and stabilizes actin filaments and bundles (40,41). In this study, we demonstrated that CD2AP/CIN85 is recruited by EHBP1L1 to the ciliary sheath, a subdomain of the CV, via the SH3 domain ( Figs. 1 and 5). Our study also showed that depletion of EHBP1L1 and CD2AP/ CIN85 causes the actin filaments around the basal body to become unstable, as indicated by the absence of the Arp2/3 complex (Fig. 6). Thus, our findings strongly suggest that CD2AP/CIN85 functions as an adaptor that links CV to actin filaments and an actin regulator that stabilizes and develops the actin network, thereby negatively regulating primary ciliogenesis. Notably, the CH domain in EHBP1L1 can bind directly to actin (42,43). Despite the nonessential role of the CH domain CIN85 (green), and Arl13b (red). Enlarged pictures of cilia are shown on the right. Scale bar represents (main images) 5 μm; (insets) 2 μm. D, percentages of ciliary localization of CD2AP (upper) and CIN85 (lower) in control, EHBP1L1-depleted, and EHBP1L1-depleted cells expressing WT or P4A mutant were indicated on the graph. n = 3 experiments. E, the depletion efficiencies and exogenous expression of CD2AP and CIN85 were examined by immunoblotting using control or specific siRNA (#1 and #2)-transfected cells. GAPDH was used as the loading control. F, ciliary length indicated by ac-tubulin was measured and is indicated on the graph (n = 89-128 cells). Error bars represent the S.D. p values were calculated using ordinary one-way ANOVA followed by Tukey's test. Statistical significance was set at *p < 0.05, **p < 0.01, and ***p < 0.001. Dotted circles indicate a 1 μm radius around the PCNT-labeled basal body. Magnified images of the circled areas are shown in the insets. Scale bar represent (main images) 2 μm; (insets) 0.5 μm. D and E, fluorescence intensities of Arp2 and Alexa-568 phalloidin within a 1 μm radius around the basal body were measured and plotted against siRNA-treated cells (control, CD2AP/CIN85, or EHBP1L1 #1 and #2). n = 50 for each experiment. Error bars represent S.D. p values were calculated using Student's t test. Statistical significance was set at ***p < 0.001. C, cell lysates treated with siRNAs (control, CD2AP/CIN85, or EHBP1L1 #1 and #2) were immunoblotted with EHBP1L1, CD2AP, CIN85, and Arp2 antibodies. GAPDH was used as the loading control. PCNT, pericentrin; RPE, retinal pigment epithelium.

EHBP1L1 controls ciliary length via CD2AP/CIN85
in the control of ciliary length (Fig. 2, D and E), we hypothesize that the direct interaction between EHBP1L1 and actin may support the role of CD2AP/CIN85 in actin network formation. Previous studies have demonstrated that Rab8 is a positive regulator of primary cilia formation that is involved in PCV assembly during early ciliogenesis and vesicle trafficking to the ciliary base via various binding proteins that interact with the membrane, centrosome/basal body, and cytoskeleton; however, whether these proteins bind directly to Rab8 has not been determined (9,10,(44)(45)(46)(47)(48). In summary, overexpression of the WT or GTP-restricted form of Rab8 promotes ciliogenesis and leads to extensive cilia. In contrast, depletion of Rab8 and various Rab8-binding proteins completely inhibits ciliary formation or reduces ciliary length. Contrary to the cited studies, we revealed that EHBP1L1-CD2AP/CIN85 has a negative effect on ciliary length, as indicated by the elongated cilia in EHBP1L1-and CD2AP/CIN85-depleted cells. These findings suggest that the role of Rab8 in ciliogenesis and ciliary integrity is unexpectedly complex.
In this study, we discovered a novel mechanism for primary ciliogenesis involving Rab8. Our findings suggest that Rab8 plays an indispensable role in ciliary membrane trafficking and actin network formation via EHBP1L1-CD2AP/CIN85, thus supporting the concept that tight regulation of the actin network is crucial for controlling ciliary length.
CD2AP deletion constructs were inserted into the pMAL-pre-His vector to produce Hisx6-MBP-tagged proteins.

Protein expression and purification
Bacterial expression plasmids were transformed into Rosetta 2 (DE3) pLys cells (Novagen) and incubated at 18 C for 16 h in the presence of 0.25 mM IPTG. The cell lysates were incubated with Ni-NTA agarose (Qiagen) for 2 h and bound proteins were eluted with 200 mM imidazole. The proteins were dialyzed against PBS, snap-frozen with liquid N 2 , and stored at −80 C until use.

Glutathione S-transferase pull-down and mass spectrometry
GST pull-down of purified proteins and pull-down using the cell lysate were performed as previously described (11). Protein samples were separated on NuPAGE Novex Bis-Tris gels (Thermo Fisher Scientific) and then silver stained as previously described (49). The bands were excised from the gel, and the proteins reduced, alkylated, and digested with trypsin in Tris-buffered saline for 16 h at 37 C. The proteins were analyzed using Q-Exactive mass spectroscopy (Thermo Fisher Scientific) at the Osaka University Center for Medical Research and Education. Searches were conducted against the EHBP1L1 controls ciliary length via CD2AP/CIN85 Mascot server (v2.3; Matrix Science) and International Protein Index (mouse, Ver. 3.77 or 3.87; EMBL-EBI) databases.

Transfection and immunofluorescence
To analyze cilia, 0.5 × 10 5 cells were cultured on coverslips in 24-well plates. To express EGFP-Myo-Va, 0.5 μg of pEGFP-C3-Myo-Va plasmid was transfected using Viafect reagent (Promega) in medium containing 10% FBS. After 48 h, the medium was replaced with serum-free medium, and the cells were incubated for a further 24 h. For knockdown experiments, 5 pmol of siRNA was transfected using Lipofectamine RNAi-Max reagent (Thermo Fisher Scientific) in medium containing 10% FBS. After 48 h, siRNA was transfected again in serum-free medium, and cells were incubated for a further 72 h. Cells were then fixed with 3% paraformaldehyde in PBS for 10 min and washed twice with 50 mM NH 4 Cl in PBS and twice with PBS. For Arp2 and pericentrin staining, cells were fixed with methanol and acetone (1:1) for 20 min at −20 C and then rinsed with PBS. The cells were then incubated with primary and secondary antibodies in the presence of 0.1% saponin. To visualize F-actin, paraformaldehyde-fixed and saponin-permeabilized cells were incubated with Alexa-568 phalloidin (Thermo Fisher Scientific) for 60 min and then rinsed with PBS. The coverslips were mounted with ProLong Diamond Antifade Mountant containing DAPI (Thermo Fisher Scientific) and observed using Olympus IX71 and Nikon A1R HD25 confocal microscopes. The lengths of cilia indicated by acetylated tubulin or Arl13b staining were measured using ImageJ and analyzed using Microsoft Excel and GraphPad Prism statistical software. Images were processed using Adobe Photoshop.

Super-resolution structured illumination microscopy
Samples were imaged using an ELYRA S.1 microscope with a 100×/1.46 oil-immersion objective (Carl Zeiss). Images and fluorescence intensity profiles were analyzed using ZEN 2011 software (Carl Zeiss) and image J.

Data availability
All the data analyzed in this study are included in the main article and supporting information.
Supporting information-This article contains supporting information.