Abstract
Efficient transport of proteins into the primary cilium is a crucial step for many signaling pathways. Dysfunction of this process can lead to the disruption of signaling cascades or cilium assembly, resulting in developmental disorders and cancer. Previous studies on ciliary trafficking were mostly focused on the membrane-embedded receptors. In contrast, how soluble proteins are delivered into the cilium is poorly understood. In our work, we identify the exocyst complex as a key player in the ciliary trafficking of soluble Gli transcription factors. Considering that the exocyst mediates intracellular vesicle transport, we demonstrate that soluble proteins, including Gli2/3 and Lkb1, can use the endosome recycling machinery for their delivery to the primary cilium. Finally, we identify GTPases: Rab14, Rab18, Rab23, and Arf4 involved in vesicle-mediated Gli protein ciliary trafficking. Our data pave the way for a better understanding of ciliary transport and uncover novel transport mechanisms inside the cell.
Introduction
Hedgehog (Hh) signaling is essential for embryonic patterning and organ morphogenesis1. Malfunctions of this pathway can lead to developmental disorders and cancer. The expression of Hh target genes is controlled by Gli transcription factors: Gli1 which acts as an activator, and Gli2/Gli3, which display both activator and repressor functions2.
Processing of Gli transcription factors to activator and repressor forms requires their efficient transport to the primary cilium, which integrates proteins necessary to a variety of Gli modifications3–8. Cilia are indispensable for the transduction of the Hh signal and the translocation of Gli activators into the nucleus9. In humans, defects in the ciliary function and the trafficking of ciliary proteins often result in developmental defects associated with the dysfunction of the Hh/Gli cascade.
Gli proteins are large and slowly diffusing proteins, so it is puzzling how they accumulate at the cilium within a mere few minutes upon signal reception10. This accumulation is a result of a three-step process: (1) targeted transport to the cilium base, (2) gated entry through a diffusion barrier, and (3) active trafficking along the cilium. The mechanisms of Gli transition across the diffusion barrier and the model of transport from the base to the tip are relatively well-described11–13. However, it is still unclear how Gli proteins are delivered so quickly and precisely from the cytoplasm to the cilium base.
Most previous studies on protein delivery to the ciliary base were focused on membrane proteins. Three different transport routes have been described for their delivery from the Golgi complex to the primary cilium14. Some ciliary proteins first reach the plasma membrane and then move to the ciliary membrane by lateral transport15. Others reach the base of the cilium using regulated vesicular transport, either directly or through the recycling trafficking pathway16.
The process of protein trafficking to the primary cilium is supported by many players involved in endocytosis and the vesicle transport machinery17,18. Prominent among them are small GTPases, which act as molecular switches that allow for the guidance of their associated vesicles19–21. In addition to GTPases, the protein ciliary trafficking depends on several multiprotein complexes, such as the BBsome22,23 and the exocyst24,25. The exocyst is a conserved protein complex that mediates the tethering of secretory vesicles to the plasma membrane26. It interacts with the ciliary transport machinery to transport transmembrane proteins necessary for ciliogenesis and signaling16,27,28.
In our quest to identify the molecular machinery that delivers Gli proteins to the cilium base, we performed a proteomic analysis of Gli3 interactors. Interestingly, among Gli3-binding proteins, we detected several exocyst subunits26. Loss-of-function assays show the dependence of Gli2 and Gli3 ciliary localization on the exocyst. Consistent with the role of this complex in vesicle trafficking, we show that Gli2 uses intracellular vesicles as trafficking vehicles. In addition, several small GTPases, including Rab14, Rab18, Rab23, and Arf4, regulate the ciliary transport of Gli2. Finally, we show that this vesicle-based transport machinery is used for the ciliary delivery of Lkb1, another soluble protein that concentrates at cilia.
Results
The exocyst complex interacts with Gli3
To identify proteins that help guide Gli proteins to the primary cilium, we immunoprecipitated proteins that interact with Gli3 in cells treated with the Smoothened (Smo) agonist SAG29. Cells were separated into “nuclear” and “cytoplasmic” fractions and then immunoprecipitated with anti-Gli3 antibodies. The eluates were separated using SDS-PAGE, and prominent bands were submitted for MS-based protein identification (Fig. 1A).
We identified 473 high confidence Gli3 interactors by rejecting frequent IP/MS contaminants based on the CRAPome database30. In this dataset, we found well-known Gli interaction partners, such as SuFu, Kif7, and Xpo731–34. The dataset was enriched for proteins involved in intraciliary and vesicle transport, chromatin remodeling, and DNA repair (Fig. 1B) and contained components of multi-subunit ciliary transport complexes, including the BBSome and the exocyst (Fig. 1C, Supplementary Table S1).
Because exocyst, a multi-subunit protein complex involved in vesicle transport and docking35, had previously been implicated in the trafficking of proteins to primary cilia, we decided to focus on its components as potential mediators of the Gli proteins delivery to the cilium base. The exocyst has mostly been studied in the context of its binding to intracellular vesicles and the plasma membrane, but the subunits that were specifically enriched in the Gli3 interactome are positioned away from the putative lipid-binding surface of the complex, consistent with Gli3 being a soluble, rather than a lipid-embedded protein (Fig. 1D).
In agreement with the proteomic data, Gli3, as well as Gli2, co-immunoprecipitate with Sec5 (Fig. 2A). Moreover, Sec5 and Gli2 tightly colocalize in cells, as shown using the proximity ligation assay (Fig. 2B). Similarly, overexpressed Sec3, Sec5, and Sec8 interact with the constitutively active Gli2 mutants Gli2(P1-6A) (Fig. 2C, D)36. We decided to use Gli2(P1-6A) in most experiments because it localizes to cilia in the absence of upstream activation, allowing us to study its trafficking independently of the transport of membrane proteins regulating endogenous Gli proteins, such as Smo and Ptch37,38.
To identify the Gli2 domain responsible for interaction with the exocyst, we performed co-immunoprecipitation of Sec3/5/8 with the N-terminal domain of Gli2 and a construct lacking the N-terminus. The exocyst subunits interact with the N-terminus of Gli2 (HA-Gli2-N) but interact only weakly with Gli2(P1-6A)-ΔN (Fig. 2E, F).
Trafficking of Gli2 to cilia depends on the exocyst
Because the exocyst is required for the trafficking of some ciliary proteins, we hypothesized that the loss-of-function of the exocyst could impair Gli ciliary localization. To test this assertion, we knocked down individual exocyst subunits in cells expressing Gli2(P1-6A). Both shRNA- (Fig. 3A) and siRNA-mediated knockdown (Fig. 3B) of exocyst subunits resulted in a significant reduction of Gli2(P1-6A) ciliary localization (Fig. 3C and D).
Similarly, mislocalization of Sec5 using the mitochondrial trap39 impairs the ciliary trafficking of Gli2(P1-6A). We fused Sec5 with the mitochondrial protein Tom20 and mScarlet and co-expressed the resulting Tom20-mScarlet-Sec5 construct with Gli2(P1-6A) (Fig. 4A). We observed a reduced Gli2 ciliary level in cells overexpressing the Tom20-mScarlet-Sec5 mitochondrial trap, compared to those overexpressing two negative control constructs – Tom20-mScarlet and mScarlet-Sec5 (Fig. 4B).
Finally, the exocyst inhibitor endosidin2 reduces Gli2(P1-6A) ciliary localization in the stable cell line after just two hours of treatment (Fig. 4C).
Because the exocyst binds to Gli2 mostly via its N-terminal domain (Fig 2E, F), we suspected that removing the N-terminus would impair Gli2 ciliary accumulation. Accordingly, we observed a strong reduction of the Gli2(P1-6A)-ΔN mutant localization in the primary cilium compared to the full-length protein (Fig. 4D).
Having demonstrated that the exocyst is required for the trafficking of Gli to cilia, we wondered if the localization of the exocyst is affected by Hh pathway activation. Indeed, the treatment with SAG increases the amount of Sec3 and Sec5 at the ciliary base suggesting that the exocyst is co-transported with Gli proteins upon pathway activation (Fig. 4E).
Gli2 associates with intracellular vesicles
While the best-known role of the exocyst complex is the transport of vesicle-embedded membrane proteins, our results suggest that soluble cytoplasmic Gli proteins may also use the exocyst as a vehicle for intracellular trafficking. We, therefore, wondered if Gli proteins use vesicles for their transport into the cilium. To verify this hypothesis, we used super-resolution AiryScan microscopy to image cells co-expressing HA-Gli2(P1-6A) and EGFP-Sec5, and surprisingly, we observed Gli2 around Sec5-positive vesicle-like structures. It suggests that Gli2 could accumulate on the surface of vesicles, where it could interact directly with the exocyst (Fig. 5A).
We also looked at Gli2 localization by immunogold electron microscopy. In HEK293T cells overexpressing EGFP-Gli2(P1-6A), we observed EGFP-positive clusters adjacent to membrane vesicle-like structures (Fig. 5B).
To check if Gli-positive structures represent intracellular vesicles, we isolated vesicles using cell fractionation. HA-Gli2(P1-6A), endogenous Gli3, and Sec5 co-fractionated with the endosome marker EEA1 in the endosomal fraction. ERK was used as the cytoplasmic control marker. The total abundance of proteins in fractions we showed by silver staining (Fig. 5C).
The most likely explanation for our results is that Gli proteins are transported on the surface of vesicles towards the ciliary base. The two potential sources of these vesicles are the Golgi apparatus via the exocytic pathway40,41 and the plasma membrane by endocytosis42–44. Firstly, we inhibited endocytosis using two inhibitors: dynasore45 and pitstop246 in cells expressing constitutively active Gli2. Surprisingly, after 2h of dynasore treatment, we observed an almost complete inhibition of Gli2 ciliary accumulation. This effect was independent of Smo because treatment with two Smo inhibitors cyclopamine and vismodegib did not affect the Gli2(P1-6A) ciliary level (Fig. 5E and Fig. S2).
If the dynasore effects are a consequence of the reduced rate of new vesicle formation, we would expect these effects to be fully reversible once the proper formation of vesicles is restored. We used a pulse-chase assay with 2h vismodegib + dynasore treatment, and then we washed out dynasore from the media and collected cells at several time points. We observed a clear recovery of Gli2 ciliary transport within 1h from the dynasore washout (Fig. 5D).
We also used another inhibitor of endocytosis – pitstop2. Because of its lethal effect on NIH3T3 fibroblasts in less than 30min, we treated cells with pitstop2 for 15 min, followed by a 30min incubation without the drug. Similar to the dynasore effects, we observed a decrease of Gli2 ciliary level in pitstop2-treated cells (Fig. 5F).
To determine if the vesicle transport from the cis-Golgi was also important for Gli2 ciliary trafficking, we treated stable HA-Gli2(P1-6A) cells with brefeldin A, a Golgi-disrupting drug47. We did not observe changes in Gli2 ciliary localization after 2h treatment (Fig. 5G).
The stimulation of target gene transcription by Gli2 is enhanced by its localization at the cilium9,48. We expected that dynasore would inhibit Hh target gene transcription in cells stably expressing the Gli2(P1-6A). Indeed, the expression of the Hh target gene Gli1 was decreased after dynasore treatment, although the expression of HA-Gli2(P1-6A) was unchanged (Fig. 5H).
Rab and Arf proteins mediate Gli2 transport
The trafficking of vesicles in cells is guided by the reversible association of small GTPases, especially from the Rab and Arf families18,49. Because their association with vesicles and associated proteins is transient, we hypothesized that under the stringent conditions of our initial co-IP/MS, the Gli-associating GTPases may have been washed away from the bait protein. Thus, we performed another co-IP/MS, with less stringent detergents, using HA-Gli2(P1-6A) as bait in cells that either had normal cilia or were devoid of cilia by means of overexpression of a dominant-negative mutant Kif3a motor50. We expected the GTPases promoting Gli ciliary trafficking to be associated with Gli2 in ciliated, but not in unciliated cells (Fig. 6A).
We identified 200 high-confidence interactors (<10% FDR in the CRAPome database) including the same well-known regulators of Gli, such as SuFu, Kif7, Xpo7, and Spop34,32,51,52, as well as component proteins of the cilium and basal body (Fig. 6B, Supplementary Table S2). Among proteins associated with Gli2(P1-6A) in ciliated cells were Rab14, Rab5c, Rab11b, Rab18, and Arf4 (Fig. 6C). In addition, we tested two other Rab-family GTPases: the well-known Hh regulator Rab2353–55 and Rab8, which cooperates with the exocyst56,57.
Initially, we established by co-IP that Rab14, Rab18, Rab23, and Arf4 proteins interact with Gli2(P1-6A) (Fig. 7A). In contrast, two Rab GTPases that had been implicated in ciliary trafficking of membrane proteins: Rab8 and Rab11a, do not strongly bind to Gli2(P1-6A) (Fig. S3A).
Subsequently, we performed loss-of-function experiments using shRNA and CRISPR/Cas9 mutagenesis. The knockdown of Rab14, Rab18, and Arf4 caused the reduction of the Gli2(P1-6A) ciliary level (Fig. 7B-D). Likewise, the CRISPR/Cas9-mediated Rab14, Rab18, and Rab23 knockout also significantly decreased the Gli2(P1-6A) ciliary accumulation (Fig. 7E). Moreover, we engineered cell lines expressing dominant-negative Rab23S51N and Arf4T31N mutants from doxycycline-inducible promoters. Consistent with shRNA- and CRISPR/Cas9-based experiments, we observed a significant decrease of Gli2(P1-6A) ciliary accumulation in cells expressing Arf4 and Rab23 mutants(Fig.7F).
The trafficking of Lkb1, but not Ubxn10, depends on endocytosis and the exocyst
To establish if the mechanism of transport to cilia by endocytic vesicles is unique to Gli proteins or more common, we imaged several HA or GFP tagged soluble ciliary candidate proteins: HA-Dvl258, Kap3a-EGFP59, HA-Lkb160, HA-Mek161, HA-Nbr162, HA-Raptor63, Tbx3-GFP64, and Ubxn10-GFP65. Only two proteins clearly localized at cilia in NIH/3T3: Ubxn10-GFP, and HA-Lkb1 (Fig. 8A and Fig. S4).
To examine if the ciliary serine-threonine kinase Lkb1 uses an analogous transport mechanism, we treated stable expressing HA-Lkb1 cells with dynasore and observed decreased Lkb1 ciliary level (Fig. 8D). Similar to Gli2, ciliary accumulation of HA-Lkb1 also dropped after the shRNA knockdown of Sec3/5/8 (Fig. 8B). Accordingly, we detected HA-Lkb1 in the endosomal fraction (Fig. 7G). Finally, we observed using co-IP that Lkb1 binds to the exocyst subunits (Fig. 8F).
Another soluble ciliary protein that we studied was Ubxn10. Dynasore treatment did not negatively affect the ciliary trafficking of Ubxn10-GFP (Fig. 8E). Unlike for Gli2(P1-6A), we observed no effect of Sec5 knockdown on Ubxn10 ciliary localization (Fig. 8C). Consistent with these results, we detected Ubxn10 predominantly in the cytosolic cell fraction (Fig. 8H).
Discussion
The cilium is an essential organelle that relays environmental signals to the nucleus. Nevertheless, the mechanism of the signaling protein delivery to cilia is still poorly understood, especially for soluble proteins. To gain a better understanding of cytoplasmic proteins’ transport to cilia we studied Gli transcription factors, large soluble proteins that accumulate at the tip of the cilium before their conversion into transcriptional activators 5,9,11.
Using proteomic screening, we found that Gli proteins interact with the exocyst, a complex implicated in ciliary delivery of membrane receptors16,35. We found that loss-of-function of the exocyst by RNAi, mitochondrial trap, or drug treatment decreases ciliary localization of the constitutively active mutant Gli2(P1-6A) independently of their effect on transmembrane Hh signaling proteins Ptch and Smo.
On a molecular level, we show that the N-terminal region of Gli proteins binds to the subcomplex I of the exocyst26,66. This agrees with our data and published reports suggesting agree that the N-terminal domain is necessary for the Gli proteins ciliary accumulation5,9,11. The N-terminus is, however, not sufficient for Gli ciliary transport, with other domains, particularly the central domain of Gli2/35,9 likely participating in other stages of ciliary translocation.
Our results suggest that soluble cytoplasmic proteins, like Gli2/3, can use the exocyst as a vehicle for intracellular trafficking. The exocyst was shown to collaborate with the BLOC-1 complex and IFT20 in the transport of membrane proteins polycystin-2 and fibrocystin to cilia16. However, IFT20 does not interact with HA-Gli2(P1-6A) (Fig. S3A). This suggests that the exocyst may mediate Gli protein ciliary trafficking independently of IFT20, which implies that the pathways directing membrane and soluble cilium components are somewhat divergent. Importantly, the exocyst can be transported to the cilium despite IFT20 loss-of-function16.
Consistent with the requirement of the exocyst in the transport of Gli2 to cilia, it appears that Gli2 is associated, at least transiently, with intracellular vesicles. Interestingly, the subunits of the exocyst that most strongly interact with Gli2 are positioned away from the putative lipid-facing surface of the complex26,66, indicating that the exocyst may form a tether between vesicle lipids and soluble proteins. Structural ciliary proteins had been previously found to be attached to the outer surfaces of intracellular vesicles carrying ciliary membrane proteins in Chlamydomonas67. We now provide functional data that corroborate and extend these findings. Protein delivery by vesicles to the cilium is persistent and essential for maintaining proper cilium function and structure68,69. Thus, the strategy of using vesicles as universal carriers of proteins, both soluble and membrane-embedded, to cilia, solves the logistical problem of homing many protein classes onto the tiny cilium base.
The trafficking of vesicles in cells is coordinated by the small GTPases from the Rab and Arf families. Intriguingly, we found that Rab14, Rab18, Rab23, and Arf4, interact with Gli2 and are essential for its accumulation in the ciliary compartment. The Rab14 GTPase localizes at early endosomes and plays a role in protein exchange between the endosomes and the Golgi compartment70–73, and exocytic vesicle targeting74. On the other hand, Rab18 is usually associated with the endoplasmic reticulum and lipid droplets75,76. Intriguingly, we identify COPI and TRAPP complex components in Gli2(P1-6A) and Gli3 interactomes, and these complexes have been implicated in lipid droplet recruitment of Rab1877. This suggests that Gli may recruit Rab18 via TRAPPII and COPI to promote ciliary trafficking. Interestingly, all three of the above GTPases: Rab18, Rab14, and Arf4, were recently identified as proximity interactors of the cilium base-localized kinase Ttbk278, strengthening the case for their involvement in the targeting of Gli-laden vesicles to the cilium.
Finally, Rab23 had previously been implicated in Hh signaling and ciliary transport of receptors79. Rab23 is described as a negative regulator of the Hh pathway but several different mechanisms have been proposed, from affecting Smo to directly regulating Gli proteins54,80,53. Here, we propose Rab23 as one of the key players in the trafficking of Gli transcription factors into the primary cilium. This is consistent with the recently discovered role of Rab23 in the transport of another soluble protein, Kif17, to primary cilia and with the ciliary and early endosome enrichment of Rab2381,82.
In addition to Rab family GTPases, we found Gli2 to associate with Arf4, which functions in sorting ciliary cargo at the Golgi and is a crucial regulator of ciliary receptor trafficking83,84. Arf4 binds the ciliary targeting signal of rhodopsin and controls the assembly of the Rab11a-Rabin8-Rab8 module for the proper delivery of cargo to the ciliary base85. Although Rab8 and Rab11a were found to cooperate with both the exocyst and Arf485 in the targeting of ciliary cargos, we found that the expression of dominant-negative Rab8 and Rab11a did not negatively affect Gli2 ciliary accumulation, with Rab8 DN actually promoting higher Gli2 accumulation in cilia (Fig. S3B). Moreover, we did not find Rab8 or Rab11a among interactors of Gli2 and Gli3. Instead, among Gli2 interactors was a Rab11a ortholog Rab11b, which had also been implicated in ciliogenesis and found to associate with Rab886,87. Disentangling the roles of the two Rab11 orthologs as well as Rab8/Rabin8 in the trafficking of soluble ciliary components will be an interesting subject for future studies.
Many of the implicated Rab/Arf proteins had been known to associate both with Golgi-derived exocytic vesicles and with plasma membrane-derived endosomes. To decipher the relative importance of these two potential vesicle sources, we used pharmacological inhibitors to show that Gli2 is likely delivered to cilia via endocytic vesicle trafficking rather than the canonical secretory pathway.
In addition to Gli2, other soluble ciliary proteins can adopt a similar transport mechanism. Specifically, we show that Lkb1 levels at primary cilia drop upon exocyst loss-of-function and inhibition of endocytosis. Like Gli2, Lkb1 associates with intracellular vesicles and interacts with the exocyst. In contrast, another soluble ciliary component Ubxn10 localizes at the cilium normally in cells depleted of Sec5 or treated with dynasore. This suggests that while the vesicle-mediated transport is important for the ciliary localization of some cytoplasmic proteins, others use different routes of ciliary trafficking.
In summary, we describe a novel mechanism for the transport of soluble cytoplasmic proteins to primary cilia, which relies on the association of these proteins with dynamically cycling endocytic vesicles. While we identify several key players in the ciliary trafficking of these vesicles, further work will dissect the precise sequence of events that are involved in this process. In particular, it will be interesting to discover potential similarities and differences between the canonical ciliary targeting pathways for membrane proteins, such as polycystin 2, fibrocystin, Smo, and rhodopsin with those described here for soluble ciliary proteins. Our work brings us closer to gaining a broad understanding of ciliary trafficking and the coordinated transport of proteins among membrane compartments.
Materials and methods
Constructs and molecular cloning
Gli2/3 constructs were cloned based on the Gli2(P1-6A) mutant previously described36 tagged with the N-terminal 3xHA. Initially, Gli2 fragments were amplified by PCR and then cloned into the pENTR2B (Life Technologies) vector by Gibson assembly88 using the NEBuilder® HiFi DNA Assembly Master Mix (NEB). Subsequently, the constructs were shuttled into pEF/FRT/V5-DEST (Life Technologies) using the Gateway method (Gateway LR Clonase II mix; Life Technologies). Plasmids with Sec3/5/8, Rab8/11/14/18, and Arf4 on the pEGFP vector were ordered from the Addgene site (Tab. 1). Rab23 wild type and mutant cDNA sequences were obtained by DNA synthesis (DNA Strings; Thermo) and cloned by Gibson assembly into the LT3GEPIR plasmid ordered from addgene (Tab. 1). Tom20 sequence was amplified from mouse cDNA and then fused with mScarlet cloned from pmScaret (addgene, Tab. 1) and Sec5 by Gibson assembly in the pEGFP-C3 vector with the EGFP sequence removed by restriction digestion. Other soluble proteins sequences of Dvl2, Nbr1, Mek1, Lkb1, Raptor were amplified from mouse cDNA and cloned into the pENTR2B with 3xHA tag vector by Gibson assembly. Ubnx10 was cloned from pHAGE-NGFP-UBXD3 - gift from M. Raman65. Tbx3 was cloned from the construct with Tbx3-Myc - a gift from A. Moon64. pEGFP-Kap3a was a gift from P. Avasthi and pEGFP-Rab11a was a gift from M. Miaczynska.
Cell culture
HEK293T (ATCC) and NIH/3T3 Flp-In (Thermo) cells were maintained in media composed of DMEM (high glucose; Biowest), sodium pyruvate (Thermo), stable glutamine (Biowest), non-essential amino acids (Thermo), 10% fetal bovine serum (EurX), and penicillin/streptomycin solution (Thermo). HA-Gli2(P1-6A) and HA-LKB1 NIH3T3 stable cell lines were generated using the Flp-In system according to the manufacturer’s protocols (Thermo Fisher). Stable cell lines were reselected with hygromycin on every other passage to preserve selection pressure.
To stimulate ciliogenesis the cells were cultured in the same medium but containing 0.5% FBS for 24h before fixing. For activation of the Hh pathway, we used SAG (Smoothened agonist) treatment 200nM for 24h. Transient transfections of cells we performed using the JetPrime reagent (Polyplus) according to the manufacturer’s protocol.
All inhibitors were suspended in DMSO and used with indicated times. The following concentrations of inhibitors were used: dynasore (40μM, Sigma), endosidin2 (200μM, Sigma), pitstop2 (30μM, Sigma), brefeldin A (5μg/ml, Sigma).
Large scale co-IP/MS on Gli3
NIH/3T3 cells were cultured to confluence on 50 15cm dishes and starved overnight to promote ciliogenesis. They were treated with 100nM SAG for 4h. The cells were fractionated into nuclear and cytoplasmic fractions as previously described101. Each fraction was immunoprecipitated overnight with 150μL Dynabeads-Protein G (Invitrogen) covalently cross-linked with goat-anti-Gli3 (AF3690; R&D Systems; 30μg antibody per fraction). The beads were washed with the following buffers: harsh RIPA lysis buffer (50mM Tris pH 7.4, 150mM NaCl, 2% Nonidet P-40, 500mM LiCl, 1mM DTT, 0.25% sodium deoxycholate, 0.1% SDS, protease and phosphatase inhibitors), RIPA lysis buffer supplemented with 0.8M urea, and mild 0.1% NP-40 lysis buffer (50mM Tris pH 7.4, 150mM NaCl, 0.1% Nonidet P-40, 1mM DTT, 1% glycerol, phosphatase inhibitors). The samples were eluted from beads using preheated 2x Laemmli sample buffer without DTT at 85°C for 5 min. The samples were then reduced and alkylated using DTT and iodoacetamide and loaded onto a 6% SDS-PAGE gel. The gel was stained using the GelCode Blue reagent (Pierce) and prominent bands were excised using a sterile scalpel and submitted for further processing to MS Bioworks (Ann Arbor, MI). The bands were destained and subjected to in-gel digest using trypsin. Each gel digest was analyzed by nano LC/MS/MS with a Waters NanoAcquity HPLC system interfaced to a ThermoFisher LTQ Orbitrap Velos. Peptides were loaded on a trapping column and eluted over a 75μm analytical column at 350nL/min; both columns were packed with Jupiter Proteo resin (Phenomenex). The mass spectrometer was operated in data-dependent mode, with MS performed in the Orbitrap at 60,000 FWHM resolution and MS/MS performed in the LTQ. The fifteen most abundant ions were selected for MS/MS. Data were searched using a local copy of Mascot with the following parameters: Enzyme: Trypsin, Database: IPI Mouse v3.75 (forward and reverse appended with common contaminants), Fixed modification: Carbamidomethyl (C), Variable modifications: Oxidation (M), Acetyl (N-term, K), Pyro-Glu (N-term Q), Deamidation (N,Q), Phospho (S,T,Y), GlyGly (K), Mass values: Monoisotopic, Peptide Mass Tolerance: 10 ppm, Fragment Mass Tolerance: 0.5 Da, Max Missed Cleavages: 2. Mascot DAT files were parsed into the Scaffold algorithm for validation, filtering, and to create a nonredundant list per sample. Data were filtered using a minimum protein value of 90%, a minimum peptide value of 50% (Prophet scores), and requiring at least two unique peptides per protein.
To determine high-confidence Gli3 interactors, we rejected all proteins found in more than 10% of negative control affinity purification/MS experiments in the CRAPome database102 (FDR < 10%). Enrichment of proteins representing specific Gene Ontology terms was performed using PANTHER with GO-Slim Cellular Component and GO-Slim Biological Process terms 103.
Large scale co-IP/MS on HA-Gli2 (P1-6A) in ciliated and non-ciliated cells
NIH/3T3 cells stably expressing HA-Gli2 (P1-6A) were transduced either with the control vector or with a retroviral vector encoding the dominant-negative variant of Kif3a (headless – amino acids 441-701 of the mouse Kif3a; dnKif3a) and selected with puromycin to eliminate untransduced cells. Each cell line was expanded from a single clone and ciliogenesis or lack thereof was verified by immunofluorescence.
Both cell lines were starved for 36h and lysed in a gentle lysis buffer (50mM Tris pH 7.4, 150mM NaCl, 0.1% Nonidet P-40, 5% glycerol, protease and phosphatase inhibitors) and scraped at 4°C. The lysate was clarified for 30min at 15,000xg and the supernatant was immunoprecipitated for 2h at 4°C with Dynabeads-protein G covalently coupled to the rat anti-HA antibody (Roche). The beads were washed 3×5 min. with the lysis buffer and 1×5min with the lysis buffer with the addition of 350mM NaCl (total NaCl concentration 500mM). Protein was eluted from beads using 2x Laemmli sample buffer at 37°C for 30min with vigorous mixing (500rpm).
Eluted proteins were submitted for mass spectrometric protein identification to MS Bioworks (Ann Arbor, MI). The entire amount of sample was separated ~1.5cm on a 10% Bis-Tris Novex mini-gel (Invitrogen) using the MES buffer system. The gels were stained with coomassie and excised into ten equally sized segments. Gel segments were processed using a robot (ProGest, DigiLab) with the following protocol:⍰Washed with 25mM ammonium bicarbonate followed by acetonitrile. Reduced with 10mM dithiothreitol at 60°C followed by alkylation with 50mM iodoacetamide at RT. Digested with trypsin (Promega) at 37°C for 4h. Quenched with formic acid and the supernatant was analyzed directly without further processing.
The gel digests were analyzed by nano LC/MS/MS with a Waters M-class HPLC system interfaced to a ThermoFisher Fusion Lumos. Peptides were loaded on a trapping column and eluted over a 75μm analytical column at 350nL/min; both columns were packed with Luna C18 resin (Phenomenex). A 30min gradient was employed (5h LC/MS/MS per sample). The mass spectrometer was operated in data-dependent mode, with MS and MS/MS performed in the Orbitrap at 60,000 FWHM resolution and 15,000 FWHM resolution, respectively. APD was turned on. The instrument was run with a 3s cycle for MS and MS/MS. Data were searched using a local copy of Mascot with the following parameters: Enzyme: Trypsin, Database: Swissprot Mouse (concatenated forward and reverse plus common contaminants), Fixed modification: Carbamidomethyl (C), Variable modifications: Oxidation (M), Acetyl (Protein N-term), Deamidation (NQ), Mass values: Monoisotopic, Peptide Mass Tolerance: 10 ppm, Fragment Mass Tolerance: 0.02 Da, Max Missed Cleavages: 2. Mascot DAT files were parsed into the Scaffold software for validation, filtering, and to create a nonredundant list per sample. Data were filtered at 1% protein and peptide level FDR and requiring at least two unique peptides per protein.
Viral transduction
For lentivirus production, we transfected HEK293T cells with pRSV-rev, pMDLg/pRRE, pMD2.G lentiviral packaging vectors (addgene, Tab. 1) and the construct encoding our protein or shRNA or sgRNA of interest, and then after 2 days, we collected the virus-containing medium and added it to target cells. We used puromycin to select transduced cells.
siRNA mediated knockdown
For siRNA-mediated knockdown of Sec5, we used the Sec5 ON-TARGET plus siRNA set of four siRNAs with non-targeting controls (Horizon Dharmacon). For siRNA transfection, we used Lipofectamine RNAiMAX (Thermofisher). Each siRNA was introduced at 40 pmol/well on a 24-well plate for 48h.
shRNA mediated knockdown
shRNAs were cloned into pLKO.1-TRC cloning vector (Tab. 1). Targeting sequences were designed using the BlockIT software from the Thermo-Fisher website.
CRISPR-Cas9-mediated mutagenesis
CRISPR-Cas9-mediated mutagenesis was performed on NIH/3T3 Flp-In cells stably expressing HA-Gli2(P1-6A) and Cas9 (Tab. 1). sgRNA sequences were designed using the Broad Institute sgRNA designer tool 104 and cloned into the pLentiGuide-puro vector (addgene, Tab. 1). We transduced the target cells with lentiviruses carrying the sgRNA of interest and either fixed 72h later or subjected to antibiotic selection.
Immunostaining and microscopy
Cells were cultured on glass coverslips. After low-serum starvation to promote ciliogenesis, we fixed cells in 4% [w/v] paraformaldehyde in PBS for 15min at room temperature (RT) and then washed 3 x 10min in phosphate buffer saline (PBS). Subsequently, cells were blocked and permeabilized in 5% [w/v] donkey serum in 0.2% [w/v] Triton X-100 in PBS. We incubated cells with the primary antibodies diluted in blocking buffer overnight at 4°C. Next, we washed the coverslips 3 x 10min with 0.05% [w/v] Triton X-100 in PBS, followed by incubation with secondary antibodies in the blocking buffer for 1 hour at RT. Cells were washed as above and mounted onto slides using a fluorescent mounting medium with DAPI (ProLong Diamond, Thermo). We acquired images on an inverted Olympus IX-73 fluorescent microscope equipped with a 63x uPLANAPO oil objective and the Photometrics Evolve 512 Delta camera. For superresolution microscopy, we used the Zeiss LSM800 confocal microscope with the Airyscan detector and Plan Apochromat 63x/1.4 Oil DIC objective.
For the quantitative analysis of fluorescence intensities, images were acquired with the same settings of exposure time, gain, offset, and illumination. Fluorescent intensities were measured in a semi-supervised manner by a custom ImageJ script. To calculate the Gli ciliary accumulation, we calculated the log10 values of the ratios of intensities of the fluorescent signal at the tip of the primary cilium and the surrounding background in each cell.
Co-immunoprecipitation
We performed co-immunoprecipitation using Pierce Anti-HA Magnetic Beads (Life Technologies) or using Dynabeads-protein G (Thermo) magnetic beads with primary antibodies (anti-GFP Genetex No#GTX113717; anti-Sec5 Proteintech No#12751-1-AP) cross-linked using dimethyl pimelimidate (Life Technologies).
For the production of whole-cell lysates, cells were lysed in 4°C in lysis buffer (50 mM Tris at pH 7.4, 1% NP-40 [v/v], 150 mM NaCl, 0.25% sodium deoxycholate [v/v], protease inhibitor cocktail [1× EDTA-free protease inhibitors, Sigma], 10mM NaF). 1/10 part of the clarified lysate was saved as an input fraction, and the rest was subjected to immunoprecipitation.
After adding beads, binding of the protein of interest was performed overnight with gentle rotation at 4°C. The next day, beads were washed 4 x 10min in 4°C in the same lysis buffer to remove unbound proteins, and complexes were eluted off the beads using 2x SDS sample buffer at 37C for 30min. We analyzed the composition of eluent using the SDS-PAGE and Western Blot method.
SDS-PAGE and Western Blot
Proteins were denaturated for 30min at 65°C and resolved by SDS-PAGE. Afterward, we performed electrotransfer onto a nitrocellulose membrane. Immunocomplexes were detected using an enhanced chemiluminescence detection system (Clarity or Clarity Max, Bio-rad) on Amersham Imager 680 and 800 as 16-bit grayscale TIFF files. The molecular weight of proteins was estimated with pre-stained protein markers (Bio-rad).
Proximity Ligation Assay
We performed the proximity ligation assay105 using the Duolink PLA Kit (Merck) according to the manufacturer’s protocol. Anti-Sec5 and anti-Gli2 primary antibodies (Tab. 3) were used to detect sites of interaction between the proteins in NIH/3T3 Flp-In cells.
Endosome Isolation
The Trident Endosome Isolation Kit (Genetex) was used to fractionate cell lysates according to the manufacturer’s protocol.
Electron Microscopy
HEK293 cells expressing EGFP-Gli2(P-16A) were fixed on the dish with 4% PFA in 0.2M phosphate buffer and 0.25% sucrose. The samples were sent to Biocenter Oulu Electron Microscopy Core Facility and there processed for EM and immunogold labeled with anti-GFP. Imaging was performed on Sigma HD VP FE-SEM equipped with ET-SE and In-lens SE detectors, VPSE G3 detector for low vacuum mode, and 5Q-BSD detector.
Data analysis
The statistical data analysis was performed using Microsoft Excel and R/RStudio. For the processing of the fluorescence images, we used the FiJi/ImageJ suite. Statistical significance was calculated using Student’s t-test for experiments involving two experimental groups, or ANOVA and Tukey posthoc test for multiple comparisons.
Acknowledgments
We would like to thank Marta Miączyńska, Małgorzata Maksymowicz, Jarosław Cendrowski, and the members of the CeNT Bio PI discussion group, and the Laboratory of Molecular and Cellular Signalling for insightful discussion and helpful suggestions. We thank M. Raman, A. Moon, and P. Avasthi for sharing their reagents with us and Addgene contributors for making their plasmids available (see Table 1). This work was supported by the following grants from the National Science Centre (NCN): SONATA BIS 2014/14/E/NZ3/00033 and PRELUDIUM 2018/29/N/NZ3/01523.
References
- 1.↵
- 2.↵
- 3.↵
- 4.
- 5.↵
- 6.
- 7.
- 8.↵
- 9.↵
- 10.↵
- 11.↵
- 12.
- 13.↵
- 14.↵
- 15.↵
- 16.↵
- 17.↵
- 18.↵
- 19.↵
- 20.
- 21.↵
- 22.↵
- 23.↵
- 24.↵
- 25.↵
- 26.↵
- 27.↵
- 28.↵
- 29.↵
- 30.↵
- 31.↵
- 32.↵
- 33.
- 34.↵
- 35.↵
- 36.↵
- 37.↵
- 38.↵
- 39.↵
- 40.↵
- 41.↵
- 42.↵
- 43.
- 44.↵
- 45.↵
- 46.↵
- 47.↵
- 48.↵
- 49.↵
- 50.↵
- 51.↵
- 52.↵
- 53.↵
- 54.↵
- 55.↵
- 56.↵
- 57.↵
- 58.↵
- 59.↵
- 60.↵
- 61.↵
- 62.↵
- 63.↵
- 64.↵
- 65.↵
- 66.↵
- 67.↵
- 68.↵
- 69.↵
- 70.↵
- 71.
- 72.
- 73.↵
- 74.↵
- 75.↵
- 76.↵
- 77.↵
- 78.↵
- 79.↵
- 80.↵
- 81.↵
- 82.↵
- 83.↵
- 84.↵
- 85.↵
- 86.↵
- 87.↵
- 88.↵
- 89.
- 90.
- 91.
- 92.
- 93.
- 94.
- 95.
- 96.
- 97.
- 98.
- 99.
- 100.
- 101.↵
- 102.↵
- 103.↵
- 104.↵
- 105.↵
- 106.
- 107.