RABL2 positively controls localization of GPCRs in mammalian primary cilia

Primary cilia are microtubule-based and membrane-enclosed protrusions present at the surface of most mammalian cells. They nucleate from basal bodies, evolved from an older centriole of two perpendicularly oriented centrioles during interphase (Kobayashi and Dynlacht, 2011; Sánchez and Dynlacht, 2016). The older (mother) centriole is characterized by the presence of distal and subdistal appendages (DAP and SAP) at its distal region, which are absent in the younger (daughter) centriole. Dysfunction of primary cilia is implicated in numerous genetic diseases with pleiotropic symptoms, called ciliopathies. Cancer development is also thought to be correlated with primary cilia because loss of the organelles is often observed in several tumors. The primary cilium functions as a cellular sensor by receiving extracellular stimuli and sending them into the cell body. To serve as a sensory organelle, the primary cilium contains many signaling components, such as enzymes, channels and membrane receptors, including G-protein-coupled receptors (GPCRs) (Hilgendorf et al., 2016; McIntyre et al., 2016; Mukhopadhyay et al., 2017; Nachury, 2018; Schou et al., 2015).

GPCRs constitute the largest membrane protein family in the human body, comprising more than 800 members. At present, ∼30 GPCRs, including rhodopsin, serotonin receptor 6 (HTR6), somatostatin receptor 3 (SSTR3), melanin concentrating hormone receptor 1 (MCHR1), neuropeptide Y receptor type 2 (NPY2R), orphan receptor GPR161 and melanocortin 4 receptor (MC4R) have been reported to be specifically concentrated in the ciliary membrane compartment (Hilgendorf et al., 2016; Siljee et al., 2018). Although the ciliary membrane is continuous with extra-ciliary membrane, entry and exit of the ciliary membrane proteins are strictly regulated by ‘ciliary gates’ at the base of the organelle (Garcia-Gonzalo and Reiter, 2017). Two characteristic structures, the transition zone (TZ) and transition fiber (TF), are thought to serve as the gates. The TF is identical to the DAP, and connects the distal end of the mother centriole to the plasma membrane. The TZ is distal to the TF, and contains Y-shaped links that connect the ciliary microtubules termed axoneme to the ciliary membrane. Ciliary GPCRs are delivered by specific transporting complexes to cross the gates. In addition, ciliary-targeting sequences (CTS) on ciliary GPCRs are presumably a prerequisite for their sorting (Hilgendorf et al., 2016; Schou et al., 2015).

Intraflagellar transport (IFT) machinery, which is essential for the formation and function of cilia, comprises protein complexes that bi-directionally move on the ciliary axoneme to transport cargo proteins (Ishikawa and Marshall, 2011). Two protein complexes, called IFT-A and IFT-B, mediate retrograde and anterograde transport, respectively, via association with motor proteins. As an additional role, IFT-A promotes entry of various GPCRs into primary cilia cooperating with its binding partner Tubby-like protein 3 (TULP3) (Mukhopadhyay et al., 2010). The BBSome, a multi-protein complex composed of Bardet–Biedl syndrome proteins, has been initially reported to be responsible for entry of GPCRs into primary cilia, and has recently been shown to play further roles in exit of ciliary GPCRs upon their activation (Jin et al., 2010; Nachury, 2018; Ye et al., 2018). Moreover, a series of small GTPases including ARF4, RAB8 and RAB11 regulate rhodopsin sorting in photoreceptor cells (Deretic et al., 2005; Mazelova et al., 2009; Wang and Deretic, 2015; Wang et al., 2017a, 2012).

RAB-like (RABL) family proteins are considered as atypical RAB small GTPases because they lack prenylation sites at their C-terminal region as well as some consensus residues in the G domain that are universally conserved in GTP-binding proteins (Blacque et al., 2018; Colicelli, 2004). Interestingly, some RABL members, such as RABL2 (herein, referring to the near-identical human paralogs RABL2A and RABL2B), RABL4 (also known as IFT27) and RABL5 (also known as IFT22) are known to be involved in the IFT machinery (Blacque et al., 2018). RABL2 interacts with IFT74 and IFT81, components of IFT-B, and promotes initiation of anterograde IFT (Kanie et al., 2017; Nishijima et al., 2017). Mutations in rodent Rabl2 or human RABL2B cause male infertility with defects in structure and motility of sperm and obesity, reminiscent of symptoms observed in patients suffering from ciliopathies (Hosseini et al., 2017; Lo et al., 2012; Wang et al., 2017b; Yi Lo et al., 2016). RABL2 localizes to the mother centriole through a centrosomal protein, CEP19, which has been shown to cause morbid obesity when mutated (Kanie et al., 2017; Nishijima et al., 2017; Shalata et al., 2013). Besides, recruitment of RABL2 and CEP19 depends on the centrosomal proteins CEP350 and FOP (also known as FGFR1OP) (Kanie et al., 2017; Mojarad et al., 2017; Nishijima et al., 2017). However, the role of RABL2 at the ciliary base and its detailed localization mechanism are still unclear.

In this study, we found that RABL2 is recruited to the mother centriole in a manner that is dependent on DAP proteins such as CEP164 and CEP83. This result could explain why RABL2 localizes only to the mother centriole. Silencing of RABL2 and its co-factors, CEP19 and IFT-B, induces mis-localization of the ciliary GPCRs GPR161 and HTR6. By contrast, ectopic expression of active RABL2 causes accumulation of these GPCRs in primary cilia. Furthermore, we demonstrated that RABL2 physically associates with ciliary GPCRs. Present findings suggest that RABL2 positively controls localization of ciliary GPCRs cooperating with CEP19 and IFT-B at the ciliary base, probably around the DAP, in mammalian cells.

Previous papers have shown that RABL2 and its binding protein CEP19 localize to the centriole through the centrosomal proteins FOP and CEP350 (Kanie et al., 2017; Mojarad et al., 2017; Nishijima et al., 2017). However, FOP and CEP350 localize to both mother and daughter centrioles, while RABL2 and CEP19 proteins are observed only at the mother centriole (Fig. S1A) (Kanie et al., 2017; Mojarad et al., 2017; Nishijima et al., 2017). Therefore, we questioned the possibility of whether RABL2 and CEP19 depend on unidentified protein(s) for their specific localization at the mother centriole. Since the location of endogenous RABL2 and CEP19 grossly overlapped with the DAP protein CEP164 as visualized by immunofluorescence analysis in human retinal pigment epithelial (RPE1) cells (Fig. S1A), we employed super-resolution-structured illumination microscopy (SR-SIM) to further assess the colocalization pattern of the two proteins. SR-SIM analysis indicated that the SAP protein ninein localized to the mother centriole more proximally than CEP164, as determined by analyzing a side view of the mother centriole (Fig. S1B). In contrast, RABL2 and CEP19 are very close to CEP164 in the same view (Fig. 1A,B), suggesting that RABL2 and CEP19 can connect with DAP proteins for their mother centriole-specific localization. We then set out to perform knockdown experiments and asked whether their foci on the centriole were impacted by introduction of siRNA targeting themselves and CEP164 in RPE1 cells (Fig. 1C,D; Fig. S1C–E). Whereas ablation of RABL2 did not influence CEP19, RABL2 disappeared upon CEP19 loss, indicating that RABL2 relies on CEP19 for its recruitment to the mother centriole as reported previously (Kanie et al., 2017; Nishijima et al., 2017) (Fig. 1C,D). By contrast, although loss of RABL2 or CEP19 was ineffective against CEP164, knockdown of CEP164 clearly disrupted RABL2 and CEP19 foci, suggesting that RABL2 and CEP19 assemble on the mother centriole through CEP164 (Fig. 1C,D). We further conducted silencing of other DAP components, such as CEP89, FBF1 and CEP83 (Fig. S1F) (Tanos et al., 2013). Knockdown of CEP89 or FBF1, which assemble on the DAP independently of CEP164, did not influence RABL2, CEP19 and CEP164 (Fig. 1F; Fig. S1G) (Tanos et al., 2013). In contrast, silencing of another DAP protein, CEP83, which is known to be required for location of CEP164 (Tanos et al., 2013), hampered RABL2 and CEP19 as well as CEP164 localization (Fig. 1E,F; Fig. S1G). We did not observe detectable alteration in RABL2, CEP19 or CEP164 foci upon knockdown of TALPID3, which forms a ring around the distal end of centrioles (data not shown) (Kobayashi et al., 2014). Taken together, these results suggest that RABL2 and CEP19 hierarchically localize to the mother centriole via the DAP proteins CEP164 and CEP83 (Fig. S1G).

We next investigated the consequences of RABL2 depletion. There are two nearly identical RABL2 paralogs, RABL2A and RABL2B, on the human genome (Wong et al., 1999). We transfected siRNAs targeting both RABL2A and RABL2B and verified that levels of RABL2 mRNA were substantially reduced by quantitative PCR using primers that are common for two paralogs in RPE1 cells (Fig. S1C). Protein levels of RABL2 were also considerably decreased by the siRNA treatments (Fig. S1D). Since a recent study showed that RABL2 is required for primary cilia formation by analyzing RPE1 cells whose RABL2A and RABL2B were doubly knocked out (Kanie et al., 2017), we first examined cilia formation in RABL2-depleted RPE1 cells. In line with the previous report, ablation of RABL2 led to a significant decrease in number of primary cilia in serum-starved RPE1 cells (Fig. S2A).

As RABL2 proteins belong to the RAB family of small GTPases, generally involved in trafficking of various intracellular molecules, we surmised that RABL2 might play a role in trafficking of ciliary proteins as well as ciliogenesis. We then tested whether localization of ciliary GPCRs was affected by RABL2 depletion. To this end, we first evaluated GPR161, which is a rhodopsin-type GPCR negatively regulating hedgehog signaling and known to be concentrated in primary cilia of RPE1 cells (Hirano et al., 2017; Mukhopadhyay et al., 2013). Although GPR161 was predictably observed in primary cilia in control cells, ablation of RABL2 led to substantially decreased localization of GPR161 to the organelle (Fig. 2A–C), suggesting that RABL2 is required for targeting GPR161 to the primary cilia. We next asked whether overexpression of RABL2 could influence the localization of GPR161. RABL2 retains most of the G-domains and is able to associate with guanine nucleotides (Kanie et al., 2017). We thus ectopically expressed wild-type (WT), constitutively active (QL) or negative (SN) RABL2B and examined localization of GPR161 in RPE1 cells. Expression of RABL2B/QL drastically increased ciliary GPR161 and RABL2B/WT induced a lower degree of elevation, in comparison with mock- and RABL2B/SN-expressing cells (Fig. 2D,E). These results suggest that activated RABL2 promotes localization of GPR161 in primary cilia. To evaluate whether RABL2 controls localization of other ciliary GPCRs, we next investigated the serotonin receptor HTR6, which is predominantly expressed in brain and localizes to neuronal primary cilia (Brailov et al., 2000; Hamon et al., 1999). To this end, we generated IMCD3 kidney cells stably expressing HTR6. While most cells showed specific enrichment of HTR6 in their primary cilia, silencing of RABL2 significantly lowered ciliary expression of HTR6 (Fig. 2F–H; Fig. S2B). Conversely, ciliary HTR6 was remarkably augmented by ectopic expression of RABL2B/QL (Fig. 2I,J). We confirmed that enforced RABL2B/SN expression disturbs ciliation in RPE1 and HTR6-IMCD3 cells, consistent with previous reports (Fig. S2C,D) (Kanie et al., 2017; Nishijima et al., 2017). Collectively, these results suggest that RABL2 positively controls localization of the ciliary GPCRs GPR161 and HTR6.

As loss of CEP19 results in the disappearance of RABL2 from the mother centriole (Fig. 1C,D), we next tested whether CEP19 depletion influences on GPR161 localization in primary cilia. Silencing of CEP19 significantly suppressed localization of ciliary GPR161 as well as ciliogenesis (Fig. 3A–C; Figs S1E, S2A), suggesting that CEP19-dependent RABL2 location to the mother centriole is necessary for GPR161 to localize to the primary cilia.

We next focused on IFT-B, an essential protein complex for anterograde transport in primary cilia (Ishikawa and Marshall, 2011). Despite the fact that endogenous RABL2 is detected only at the mother centriole/basal body (Fig. S1A), ectopically expressed GFP-fused RABL2B/QL entered into ciliary shaft and substantially colocalized with IFT88, a component of the IFT-B complex (Fig. S3A). We therefore assumed that RABL2 controls trafficking of ciliary GPCR through IFT-B. Ectopic expression of RABL2/QL enhanced IFT88 signal in primary cilia and conversely, silencing of RABL2 decreased IFT88 in the organelles, suggesting that RABL2 positively regulates anterograde IFT (Fig. S3B–D). We subsequently asked whether IFT88 depletion causes defects in the localization of GPR161. Silencing of IFT88 significantly impeded GPR161 targeting to primary cilia in addition to assembly of the organelles (Fig. 3A–C; Figs S2A, S3E). These data suggest that RABL2 controls trafficking of ciliary GPCR through anterograde IFT mediated by the IFT-B complex.

TULP3 is known to be required for delivery of GPR161 into primary cilia (Mukhopadhyay et al., 2013). Silencing of TULP3 actually impaired localization of ciliary GPR161 as predicted (Fig. S3F–H). We found that combined knockdown of RABL2 and TULP3 significantly perturbed ciliary targeting of GPR161 compared to singly silencing the genes (Fig. 3D,E). We also observed a similar additive effect upon dual silencing of IFT88 and TULP3 (Fig. 3F,G). Furthermore, TULP3 depletion did not alter the distribution of IFT88 (Fig. S3I). Collectively, these results suggest that RABL2 regulates localization of ciliary GPR161 through IFT-B, but TULP3 exerts its effect independently of RABL2–IFT-B (see Discussion).

Mounting studies have shown that there is a physical interaction between ciliary GPCRs and their carrier proteins (Badgandi et al., 2017; Deretic et al., 2005; Dong and Wu, 2013; Jin et al., 2010; Mukhopadhyay et al., 2017). This led us to verify whether RABL2 actually associates with ciliary GPCRs in cells. We transfected plasmids expressing Flag–RABL2B and GPR161 or HTR6 in HEK293T cells, performed anti-Flag immunoprecipitation, and demonstrated that these GPCRs were co-precipitated with Flag–RABL2B (Fig. 4A,B). Reciprocal immunoprecipitations using anti-GPR161 or anti-HTR6 antibody confirmed their association with RABL2B (Fig. 4A,B). These data suggest that RABL2 physically interacts with the ciliary GPCRs GPR161 and HTR6. We further observed that RABL2B co-precipitates more HTR6 than another serotonin receptor HTR7 (Fig. 4C, Fig. S4A), which does not localize to primary cilia (Berbari et al., 2008), suggesting that RABL2 preferentially associates with ciliary GPCRs. We next investigated the guanine nucleotide dependency of RABL2 for its binding to GPCRs. We detected unambiguous interactions between RABL2B/WT and the active mutant (RABL2B/QL) with CEP19 (Fig. S4B), and between RABL2B/QL and IFT88 (Fig. 4A,B) consistent with previous studies (Kanie et al., 2017; Nishijima et al., 2017), indicating that RABL2B mutants bind to the expected guanine nucleotides. Nevertheless, we observed that GPR161 and HTR6 were co-precipitated with RABL2B/QL and RABL2B/SN with marginally lower binding efficacies than RABL2B/WT (Fig. 4A,B). These results suggest that RABL2 interacts with ciliary GPCRs irrespective of its guanine nucleotide-bound form.

In this study, we showed that RABL2–CEP19 assembles on the centriole in a manner that is dependent on CEP164 and CEP83. Since CEP164 and CEP83 constitute core DAP proteins on the mother centriole, this assembly pathway can illustrate how RABL2 and CEP19 localize only to the mother centriole. CEP83 is placed at the root of the DAP and is indispensable for localization of four other core DAP proteins, whereas CEP164 localizes at the tip of a branch via CEP83–SCLT1 (Tanos et al., 2013) (Fig. S1G). Based on this knowledge and the further support of our findings, RABL2–CEP19 is likely to link to CEP164 (Fig. S1G). However, we did not succeed in detecting co-immunoprecipitation between CEP164 and CEP19 or RABL2 (data not shown), raising the possibilities that unidentified protein(s) might mediate the interaction between CEP164 and RABL2–CEP19 or that they associate with an undetectably low affinity. Previous studies demonstrated that depletion of FOP or CEP350, which localize at both mother and daughter centrioles, causes disappearance of CEP19 and RABL2 (Kanie et al., 2017; Mojarad et al., 2017; Nishijima et al., 2017). However, although FOP and CEP350 localize to the inner body of the centriole near the SAP, CEP19 and RABL2 have been shown to locate closer to the DAP than FOP and CEP350 (Kanie et al., 2017). Our SR-SIM analysis exhibited near-complete colocalization of RABL2 and CEP19 with CEP164 (Fig. 1A,B). In addition, CEP164 is known to locate along the length of the DAP (Yang et al., 2018). Hence, it is conceivable that CEP19 and RABL2 are able to associate with CEP164 around the ‘neck’ region of the DAP. In conclusion, we propose that RABL2–CEP19 connects to and depends on both FOP–CEP350 and CEP164 for its localization at around the DAP of the mother centriole (Fig. S1G).

The present findings suggest a model in which RABL2 promotes trafficking of GPCRs, such as GPR161 and HTR6, to primary cilia by mediating anterograde IFT. It has been reported that IFT components gather at the TF, and the machinery presumably assembles there prior to entering the ciliary shaft (Yang et al., 2015). Therefore, we hypothesize that RABL2 contributes to load GPCRs on IFT machinery near to the TF and, subsequently, the RABL2–IFT–GPCR complex moves to the ciliary tip. Colocalization of RABL2B/QL and IFT88 throughout primary cilia supports this idea (Fig. S3A). TULP3, together with IFT-A, plays a role in entry of multiple ciliary GPCRs including GPR161 into the primary cilia (Mukhopadhyay et al., 2010). We showed here that combinatorial knockdown of RABL2 or IFT88 and TULP3 additively induce mis-localization of GPR161 as compared to singularly silencing the genes (Fig. 3D–G). These data suggest that TULP3 and IFT-A, and RABL2 and IFT-B act at distinct steps for delivery of ciliary GPCR in that TULP3 and IFT-A mediate entry of GPR161 from cytosol to the base of primary cilia and thereafter RABL2 and IFT-B convey the received GPCR to the ciliary shaft. In this model, linkage of two machineries probably occurs at the DAP/TF. Localization of the GPCR dopamine D1 receptor (D1R, also known as DRD1) at primary cilia has been reported to be interrupted by loss of IFT-B components or TULP3 (Badgandi et al., 2017; Leaf and Von Zastrow, 2015), like GPR161. Therefore, D1R and GPR161 may travel through common TULP3- and IFT-B-dependent pathways in that RABL2 could mediate two machineries. Future work will be needed to assess whether D1R requires RABL2 for its localization in primary cilia.

Here, we detected co-precipitation of ciliary GPCRs with RABL2, indicating that the carrier RABL2 binds to cargo GPCRs in cells. GTP-locked RABL2/QL specifically interacted and colocalized with IFT88 in the ciliary shaft and, conversely, RABL2/SN is absent from centrioles probably because this mutant is not able to bind to CEP19 (Fig. 4A,B; Figs S3A, S4B; data not shown) (Kanie et al., 2017). Based on these data, we presumed that RABL2/QL preferentially associates with GPCRs. However, RABL2 unpredictably bound to GPCRs regardless of its guanine nucleotide form. These data may imply that guanine nucleotide form of RABL2 determines its subcellular localization via interactions with IFT-B and CEP19 but do not impinge on its binding to ciliary GPCRs.

CEP19- or RABL2-knockout RPE1 cells have substantial ciliation defects, although cilia are formed to a small degree and with frailty (Kanie et al., 2017). Another paper demonstrated that CEP19-knockout MEF cells form primary cilia identically to wild-type MEFs (Shalata et al., 2013). These studies may imply that CEP19 and RABL2 are required but not essential for ciliary assembly and their necessities vary by cell type. Loss-of-function mutation in CEP19 causes morbid obesity in both human and mouse (Shalata et al., 2013). In addition, knocking out of the Rabl2 gene induces increased weight in mice (Yi Lo et al., 2016). Several studies have hypothesized that obesity is due to mis-localization of ciliary GPCRs in the central neurons. Mis-targeting of NPY2R to neuronal primary cilia appears to cause compromised energy balances, resulting in obesity (Loktev and Jackson, 2013). It has been recently shown that obesity is provoked by impaired localization of the ciliary GPCR MC4R, the most common factor in obesity pathogenesis (Siljee et al., 2018). From these and our studies, it would be plausible that dysfunction of RABL2 or CEP19 causes mis-localization of ciliary GPCRs at the central nervous system, thereby leading to obesity, even without anomaly in primary cilia assembly. In addition, given overexpression of active RABL2 (RABL2/QL) enhances accumulation of ciliary GPCRs, RABL2 might be a valid therapeutic target against obesity caused by mis-targeting of GPCRs to primary cilia.

HEK293T cells (from Brian D. Dynlacht, Dept. of Pathology, NYU, USA) were grown in DMEM (Nacalai tesque) supplemented with 10% calf serum (Thermo Fisher Scientific) and 100 units/ml penicillin and 100 μg/ml streptomycin (P/S) (Nacalai tesque). hTert-RPE1 cells (from Brian D. Dynlacht) and Lenti-X 293T (from Masatoshi Hagiwara, Dept. of Anatomy and Developmental Biology, Kyoto University, Japan) cells were grown in DMEM supplemented with 10% fetal bovine serum (FBS) (Biosera) and P/S. IMCD3 cells (from Koji Ikegami, Dept. of Cellular and Molecular Anatomy, Hamamatsu University School of Medicine, Japan) and IMCD3 cells stably expressing HTR6 (HTR6-IMCD3) were grown in DMEM/F12 (Nacalai tesque) supplemented with 10% FBS and P/S. All cells are treated with MC210 (DS pharma) to remove mycoplasma. In the case of IMCD3 cells, serum starvation was performed in DMEM/F12 without FBS.

Lentivirus supernatant was produced by co-transfection of pLVX-HTR6-IRES-Puro, Δ8.9, pcRev and VSVG plasmids into Lenti-X 293T cells using PEI Max. The virus supernatant was harvested 72 h post-transfection and concentrated using a Lenti-X Concentrator (Clonetech). IMCD3 cells were incubated with virus in the presence of 5 µg/ml polybrene (Nacalai tesque) in serum starvation medium for 24 h. The cells were cultured in medium with 3 µg/ml puromycin for 24 h and 2 µg/ml puromycin for 6 more days. The expression of HTR6 in surviving cells was evaluated through immunofluorescence.

Antibodies used in this study include mouse anti-glutamylated tubulin (GT335) (1:1000, Adipogen, AG-20B-0020), rabbit anti-RABL2B (1:100, 11588-1-AP; cross reacts with RABL2A), rabbit anti-IFT88 [1:500 for immunofluoresence (IF), 1:1000 for western blotting (WB), 13967-1-AP], rabbit anti-GPR161 (1:500 for IF, 1:1000 for WB, 13398-1-AP), rabbit anti-CEP164 (1:200, 22227-1-AP), rabbit anti-TULP3 (1:1000, 13637-1-AP) (all from Proteintech), mouse anti-centrin (1:1000, 04-1624), mouse anti-ninein (1:200, 79-160-7) (all from Millipore), rabbit anti-CEP19/C3orf34 (1:1000, Abcam, ab74989), goat anti-CEP164 (1:200, sc-240226), mouse anti-β-actin (1:1000, sc-47778), rabbit anti-GFP (1:1000, sc-9996) (all from Santa Cruz Biotechnology), rabbit anti-FLAG (1:1000, F7425), mouse anti-FLAG (M2) (1:1000, F1804), rabbit IgG (1:1000, I5006) (all from Sigma Aldrich) and rabbit anti-Myc (1:1000, MBL, 562). Rabbit anti-HTR6 antibody was produced by immunizing synthetic peptide containing CFVTDSVEPEIRQHPLGS (amino acids 421–437 of mouse HTR6) into rabbits (1:2000, MBL). Antibodies were purified by affinity chromatography using the peptide-conjugated resin from antiserum.

To generate Flag–RABL2B, a human RABL2B fragment encoding residue 1–228 was amplified by PCR using forward primer 5′-AAGAATTCATGGCAGAAGACAAAACCAAACC-3′ and reverse primer 5′-AAGTCGACTCAGCTGTGGGGAGAGGCCG-3′, and sub-cloned into pCMV5-Flag (Kobayashi et al., 2011). RABL2B/Q80L and S35N constructs were made by PCR-based mutagenesis using forward primer 5′-TGGAGCGGTTCCAGAGCATG-3′ and reverse primer 5′-GGCCTGCCGTGTCCCAAAAG-3′, and forward primer 5′-AACAAACTCATGGAGAGATTTC-3′ and reverse primer 5′-TTTGCCCACTGCGCTGTCTC-3′, respectively. To generate GFP–RABL2B proteins, human RABL2B fragments encoding residues 1–228, 1–228/Q80L or 1–228/S35N were excised from pCMV5-Flag-hRABL2B and subcloned into pEGFP-C2 (Clontech). To generate GPR161, a human GPR161 fragment encoding residues 1–529 was amplified by PCR using forward primer 5′-AAGAATTCGCCACCATGAGCCTCAACTCCTCCCT-3′ and reverse primer 5′-AAGTCGACTCATCTCTGCTCGGCAGCTA-3′, and sub-cloned into pCMV5. To generate HTR6, a mouse HTR6 fragment encoding residues 1–440 was amplified from pEGFP-N3-HTR6 by PCR using forward primer 5′-AACTCGAGGCCACCATGGTTCCAGAGCC-3′ and reverse primer 5′-TTGGATCCTCAGTTCATGGGGGAACCAAGTG-3′, and sub-cloned into pLVX-IRES-Puro. To generate HTR6–Myc, a mouse HTR6 fragment encoding residues 1–440 was amplified from pEGFP-N3-HTR6 by PCR using forward primer 5′-AAGAATTCGCCACCATGGTTCCAGAGCCCGGCCC-3′ and reverse primer 5′-AAGTCGACGTTCATGGGGGAACCAAGTG-3′, and sub-cloned into pMyc-C-CMV5. To generate HTR7–Myc, a mouse HTR7 fragment encoding residues 1–448 was excised from pmCherry-HTR7 and sub-cloned into pMyc-C-CMV5. pMyc-C-CMV5 was derived from pFlag-C-CMV5 by replacing the Flag sequence with a Myc sequence (Hori et al., 2008). Plasmid expressing CEP19–GFP was obtained from Sehyun Kim and Brian D. Dynlacht.

Plasmid transfection into RPE1 cells was performed using FuGENE HD (Promega) or Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. Plasmid transfection into HEK293T or Lenti-X 293T cells was performed using PEI Max (Polysciences) according to the manufacturer's instructions.

siRNA oligonucleotides used in this study were siRABL2#1 (Dharmacon D-008404-01), siRABL2#2 (Dharmacon D-008404-03), siRABL2#3 (Dharmacon D-008404-04), simRABL2#1 (5′-GACACGACCAAGAGCCCAUtt-3′, Sigma), simRABL2#2 (5′-GAAACUAGACAGGCGUGAUtt-3′, Sigma) (the ‘m’ refers to mouse), siCEP19#1 (Ambion s39800), siCEP19#2 (Ambion s39801), siCEP164#1 (Ambion s22615), siCEP164#2 (5′-GACCUUGAAACCAGAGCUAtt-3′, Sigma), siCEP83#1 (Ambion s27524), siCEP83#2 (Ambion s27526), siCEP89#1 (Ambion s39619), siCEP89#2 (Ambion s39620), siFBF1#1 (Ambion s39897), siFBF1#2 (Ambion s39898), siIFT88#1 (Ambion s15623), siIFT88#2 (Ambion s224870), siTULP3#1 (5′-GAAACAAACGUACUUGGAUtt-3′, Sigma) and siTULP3#2 (5′-GCAGCUAGAAAGCGGAAAAtt-3′, Sigma). A mixture of #1 and #2 siRNA was used for silencing of CEP83, Cep89 or FBF1. The siRNA for luciferase (siLuc) was described previously (Kobayashi et al., 2017). For RNAi, 4×10⁴ RPE1 cells or 1×10⁵ HTR6-IMCD3 cells were seeded in 24-well plates and cultured for 24 h. After transfection of 15 pmol siRNA using Lipofectamine RNAiMAX (Invitrogen), cells were cultured in normal medium for 24 h and subsequently incubated in serum starvation medium for 48 h.

For Flag–RABL2B and GPCR interaction studies, cells were lysed with lysis buffer (50 mM Hepes pH 7.5, 150 mM NaCl, 5 mM MgCl2, 0.5% NP-40, 1 mM DTT, 0.5 mM PMSF, and 2 µg/ml leupeptin) at 4°C for 30 min. For immunoprecipitation, 1 mg of the resulting supernatant after centrifugation (17,000 g for 10 min) was incubated with anti-Flag agarose beads (Sigma) at 4°C for 2 h. The resin was washed with lysis buffer, and the bound polypeptides were analyzed by SDS-PAGE and immunoblotting. 10 µg of lysate was loaded in the input (IN) lane.

For Flag–RABL2B and GFP–CEP19 interaction studies, cells were lysed with lysis buffer at 4°C for 10 min. 1 mg of the resulting supernatant after centrifugation was incubated with anti-Flag agarose beads at 4°C for 30 min, and subsequently washed with lysis buffer, and the bound polypeptides were analyzed by SDS-PAGE and immunoblotting.

Cells were fixed with cold methanol for 5 min or 4% PFA in phosphate-buffered saline (PBS) for 10 min, and permeabilized with 0.2% Triton X-100 in PBS for 10 min. For SR-SIM analysis, cells were washed once with PBS and fixed with 10% (v/v) trichloriacetic acid (TCA) for 15 min at 4°C, followed by permeabilization with 0.1% Triton X-100 for 5 min. Slides were blocked with 5% BSA in PBS prior to incubation with primary antibodies. Primary and secondary antibodies were diluted with 5% BSA. For RABL2B staining, primary antibody was diluted with Can Get Signal Immunostain A (TOYOBO). Secondary antibodies used were Alexa Fluor 350-, Alexa Fluor 488- or Alexa Fluor 594-conjugated donkey anti-mouse, anti-rabbit or anti-goat IgG (Invitrogen). Cells were stained with Hoechst 33342 (Nacalai tesque) to visualize DNA. Mounted slides with PermaFluor Mounting Medium (Thermo Scientific) were observed and photographed using AxioObserver (Zeiss) with a 63× lens. Image analysis was performed using Photoshop (Adobe). Quantification analysis of fluorescence intensity was performed using ImageJ (Nozaki et al., 2017).

The SR-SIM imaging was performed using an ELYRA S.1 microscope (Zeiss) equipped with an Andor iXon 885 EMCCD camera, a 100×/1.40 NA oil-immersion objective and four laser beams (405, 488, 561 and 642 nm) (Takahashi et al., 2018). Serial z-stack sectioning was carried out at 101 nm intervals. Z-stacks were recorded with three phase-changes and five grating rotations for each section. The microscope was calibrated with 100 nm fluorescent beads to calculate both lateral and axial limits of image resolution. Images were reconstituted with Zen software (Zeiss).

Quantitative PCR was performed as described previously (Kobayashi et al., 2017). Primers are listed in Table S1.

Differences were considered significant when P<0.05 (**P<0.01; *P<0.05).

We thank Koji Ikegami (Hamamatsu Medical University) for pmCherry-HTR7 and IMCD3 cells, Brian D. Dynlacht and Sehyun Kim (New York University) for pLVX-IRES-Puro, pEGFP-N3-hCEP19 and hTert RPE1 and HEK293T cells, Takanari Inoue (Johns Hopkins University) for pEGFP-N3-HTR6, Masatoshi Hagiwara (Kyoto University) for Lenti-X 293T cells, Δ8.9, pcRev, and VSVG plasmids, and Toshiaki Katada and Kenji Kontani (University of Tokyo) for pCMV5-Flag and pFlag-C-CMV5. We thank the SR-SIM imaging facility in the Center for Medical Research and Education, Graduate School of Medicine, Osaka University. We thank S. Kim for valuable comments.