The Hedgehog pathway, critical to vertebrate development, is organized in primary cilia. Activation of signaling causes the Hedgehog receptor Ptch1 to exit cilia, allowing a second receptor, Smo, to accumulate in cilia and activate the downstream steps of the pathway. Mechanisms regulating the dynamics of these receptors are unknown, but the ubiquitination of Smo regulates its interaction with the intraflagellar transport system to control ciliary levels. A focused screen of ubiquitin-related genes identified nine required for maintaining low ciliary Smo at the basal state. These included cytoplasmic E3s (Arih2, Mgrn1, and Maea), a ciliary localized E3 (Wwp1), a ciliary localized E2 (Ube2l3), a deubiquitinase (Bap1), and three adaptors (Kctd5, Skp1a, and Skp2). The ciliary E3, Wwp1, binds Ptch1 and localizes to cilia at the basal state. Activation of signaling removes both Ptch1 and Wwp1 from cilia, thus providing an elegant mechanism for Ptch1 to regulate ciliary Smo levels.
Cilia are microtubule-based, hair-like organelles protruding from the surface of most eukaryotic cells. These organelles are evolutionarily conserved from unicellular eukaryotes to mammals and provide diverse motility and sensory functions to cells (Satir and Christensen, 2007). Ciliary dysfunction causes a variety of degenerative and developmental diseases collectively known as the ciliopathies (Reiter and Leroux, 2017). Cilia are complex structures made up of 1,000 or more unique proteins (Pazour et al., 2005). These organelles rely on proteins synthesized in the cell and transported into the cilium by intraflagellar transport (IFT; Bhogaraju et al., 2013; Kozminski et al., 1993; Rosenbaum and Witman, 2002; Scholey, 2003). The IFT system consists of trains built from IFT-A, IFT-B, and BBSome subcomplexes carried along cilia by kinesin-2 (Cole et al., 1998; Snow et al., 2004) and dynein-2 motors (Pazour et al., 1998). The ∼22 proteins of IFT-A and IFT-B are thought to be adaptors that connect the motors to the diverse cargos (Bhogaraju et al., 2013; Mukhopadhyay et al., 2010; Palenik et al., 2007). The eight subunits of the BBSome function primarily for the removal of ciliary proteins such as Smoothened (Smo) and Gpr161 (Lechtreck et al., 2009; Liew et al., 2014; Nozaki et al., 2019; Nozaki et al., 2018), although they may have roles in delivery as well (Berbari et al., 2008). Mutations in the BBSome-encoding genes do not block ciliary assembly but cause functional defects in sensory perception and signaling. In contrast, mutations in most IFT-A and IFT-B genes cause ciliary assembly defects. However, there are exceptions, such as Ift25 and Ift27. These two subunits of IFT-B are not required for ciliary assembly, but mutants show phenotypes indicative of Hedgehog signaling dysfunction (Eguether et al., 2018; Eguether et al., 2014; Keady et al., 2012).
The Hedgehog pathway (Bangs and Anderson, 2017; Huangfu et al., 2003) plays pivotal roles in embryonic patterning, tissue homeostasis, and tumorigenesis (Choudhry et al., 2014; Emechebe et al., 2016; Gupta et al., 2010; Petrova and Joyner, 2014). In vertebrates, this pathway is organized in cilia, with the receptors and effectors dynamically redistributing during signaling. At the basal state, the Hedgehog receptor Ptch1 localizes to cilia and prevents the ciliary accumulation and activation of Smo (Rohatgi et al., 2007). In addition, Gpr161 localizes to cilia at the basal state and directs the processing of Gli3 into Gli3R via cAMP signaling (Mukhopadhyay et al., 2013). Once bound by Hedgehog ligand, Ptch1 leaves cilia and no longer suppresses Smo. Smo becomes activated, accumulates in cilia, and promotes the removal of Gpr161 from cilia (Corbit et al., 2005; Mukhopadhyay et al., 2013). The Gli transcription factors then concentrate at the ciliary tip and become activated before translocating into the nucleus to induce Hedgehog target genes (Haycraft et al., 2005).
The Hedgehog pathway is extensively regulated by post-translational modifications that control the activity of the components and their localization to subcompartments in the cell. For example, Smo becomes highly phosphorylated upon activation (Jia et al., 2004). Ubiquitination also modulates numerous steps in the Hedgehog pathway, including regulating levels and localization of Ptch1 and the processing of the Gli transcription factors (Hsia et al., 2015). Adding to this, we recently found that the ubiquitination of Smo regulates its ciliary localization by modulating its interaction with the IFT system (Desai et al., 2020). Covalent conjugation of ubiquitin (Ub), a small protein of 76 residues (8.6 kD), regulates the function and fate of many cellular proteins. Typically, the carboxy terminus of Ub is conjugated to the ε-amino group of a lysine or to the N-terminal residue of a substrate (Breitschopf et al., 1998). Ub itself contains seven lysines and an N-terminal methionine that can be further ubiquitinated to produce poly-Ub chains. These poly-Ub chains are named by the Ub residue that is extended (K6, K11, K27, K29, K33, K48, K63, M1; Komander and Rape, 2012). The type of poly-Ub chain determines the fate of the modified protein. The best understood are K48, which triggers degradation by the proteasome, and K63, which regulates protein complex formation. Ubiquitination is catalyzed by a cascade of three enzymes. First, Ub is activated by an E1 in an ATP-dependent process (Schulman and Harper, 2009). The activated Ub is then transferred to the active site of an E2 (van Wijk and Timmers, 2010). Finally, the Ub is covalently attached to substrates with the help of E3 ligases (Pickart, 2001). Ub can be removed from substrates by deubiquitinases, which is important for Ub homeostasis (Kimura and Tanaka, 2010). Enormous complexity exists in the Ub system, with the human genome encoding 2 E1 Ub–activating enzymes, >30 E2 Ub–conjugating enzymes, >600 E3 Ub ligases (Jin et al., 2007), and >100 deubiquitinases (Nijman et al., 2005).
Our recent work showed that Smo is ubiquitinated at the basal state, and this causes Smo to be removed from cilia in an IFT-dependent process (Desai et al., 2020). To identify the enzymes regulating Smo ubiquitination and ciliary localization, we performed a focused screen of Ub-related components that previously had been connected to cilia or Hedgehog signaling. Starting from 147 Ub-related components with connections to cilia or Hedgehog signaling, we identified 40 genes whose loss altered Hedgehog signaling as measured by a Gli-GFP reporter. 9 of these 40 caused aberrant Smo accumulation in cilia at the basal state. The nine included an E2-conjugating enzyme (Ube2l3), four E3 ligases (Wwp1, Arih2, Mgrn1, and Maea), a deubiquitinating enzyme (DUB; Bap1), and three adaptors (Kctd5, Skp1a, and Skp2). The E3 ligases appear to function in discrete steps where Arih2 regulates cellular levels of Smo, Wwp1 regulates ciliary levels directly, and Mgrn1 and Maea function in the cytoplasm. Our findings shed light on the long-standing enigma of how the ciliary levels of Smo are regulated by Hedgehog signaling.
Development of a Hedgehog reporter cell line
To identify genes involved in regulating the dynamics of Smo localization during Hedgehog signaling, we developed a cell line that monitors ciliary Smo and activation of Hedgehog-responsive genes. To do this, we first transfected immortalized mouse embryonic fibroblasts (MEFs) with a Smo-3xFlag construct (PD22) and cloned out a line (11479.6T PD22 clone 3, called “MEFSmo-3xFlag” hereafter) showing low ciliary Smo-3xFlag at the basal state and high ciliary Smo-3xFlag after pathway activation. This cell line was then transfected with a reporter construct (GP778) where nuclear localized GFP is driven from a promoter consisting of eight tandem Gli binding sites derived from the FOXA2 gene placed upstream of the chicken CryD1 minimal promoter (Fig. 1 A). Flow cytometry and single-cell cloning were used to identify a line, B13 (“GreenBomb” hereafter), where GFP fluorescence is enhanced ∼10-fold by addition of Smo agonist (SAG; Fig. 1, B–D; and Fig. S1, A–D). As predicted, knockout of Ift27 by CRISPR blocked induction of the GFP reporter by SAG, and knockout of Ptch1 caused GFP production without the need for pathway activation (Fig. 1 B). Characterization of GreenBomb cells indicated that they were similar to the parental MEF and MEFSmo-3xFlag cells with regard to percentage ciliation and cilia length (Fig. S1, E–G), and ciliary Smo-3xFlag was regulated normally (Fig. 1 D). These results indicate that GreenBomb is sensitive to perturbations of the Hedgehog pathway and can be used to characterize genes with unknown roles in ciliation and Hedgehog signaling.
Screening Ub-related genes for roles in Hedgehog signaling
To identify proteins regulating the dynamics of Hedgehog signaling, we identified Ub-related genes in published proteomic, CRISPR, and siRNA screens designed to identify genes related to Hedgehog signaling or cilia (a complete list of genes and sources can be found in Table S1; Breslow et al., 2018; Ishikawa et al., 2012; Mick et al., 2015; Pusapati et al., 2017). This list was supplemented with Ub-related genes suggested to play roles in Drosophila melanogaster Hedgehog signaling and ones predicted by UbiBrowser (http://ubibrowser.ncpsb.org.cn) to ubiquitinate Hedgehog components (Li et al., 2017). Finally, interacting proteins of the regulators we identified were also included. From these various sources, 147 genes were identified among 907 Ub-related genes (Fig. 1 E). Three gRNAs were synthesized for each gene and cloned in lentiCRISPR v2 or lentiCRISPR v2 PuroP93S. The clones were verified by sequence, packaged into lentiviral particles, and transduced into GreenBomb cells. After drug selection, the GreenBomb cells transfected with gRNAs were analyzed by flow cytometry with and without stimulation with SAG (Fig. 1, F and G). From this screen, we identified 18 genes whose loss reduced both basal Hedgehog signaling and signaling in response to SAG treatment (positive regulators); 20 genes whose loss activated the pathway without need for ligand (negative regulators); and two genes, Kctd5 and Ube2l3, that increased basal expression but reduced expression driven by SAG (Fig. 1 G and Table S2).
To determine the cellular localization of the Hedgehog regulators identified in this work, we cloned all except Kmtd2 and Usp34 and expressed epitope-tagged forms in fibroblasts. Kmtd2 and Usp34 had extremely large ORFs and resisted attempts to amplify either as whole cDNAs or in fragments. The collection of regulators has diverse distributions throughout the cell with Bfar, Cop1, Cul9, Fancg, Skp1a, Uba3, Ube2l3, and Wwp1 showing weak localization to cilia. In addition, Sytl4 localized to the ciliary pocket at the base of the cilium, and Fbxw11 localized to centriolar satellites (Table S3).
This collection of potential regulators was further screened to determine if their loss affected ciliation because this could have a major impact on Hedgehog signaling. As expected, knockout of Ift74, but not Bbs2, Bbs5, Ift27, or Ptch1, blocked ciliogenesis (Fig. 1 H). Importantly, none of the regulators affected ciliation except Kctd10 (Fig. 1 H), which is known to participate in CEP97 degradation (Nagai et al., 2018). This indicates that the reduced expression caused by the knockout of most positive regulators is not caused by the failure of ciliogenesis.
Screening the 40 regulators identified nine genes whose loss caused the accumulation of Smo in cilia when the Hedgehog pathway was at the basal state (Fig. 1 I). Of these nine genes, Ube2l3 encodes an E2 Ub–conjugating enzyme, and Arih2, Mgrn1, and Wwp1 encode E3 Ub ligases. Maea is a member of the RING family of E3 ligases but lacks a detectable RING-type zinc finger domain, raising questions of whether it retains E3 activity (Maitland et al., 2019). However, it is required for the ubiquitination of substrates as part of a complex (Lampert et al., 2018), so we refer to it as an E3 ligase even though it may function more like an adaptor. Bap1 is a deubiquitinase. The others, Kctd5, Skp1a, and Skp2, may act as substrate adaptors in E3 ligase complexes.
Arih2, Maea, Mgrn1, and Wwp1 regulate the ciliary localization of Smo
For further study of the E3 ligases that showed Smo accumulation in cilia, we isolated clones from each of the knockout populations. The clones behave like the uncloned populations as measured by flow cytometry of GFP expression from the reporter (Fig. 2 A) or by direct measures of Gli1 mRNA (Fig. S3 B). Similar to previous observations with Ift27−/− cells, clones of Arih2−/−, Mgrn1−/−, and Wwp1−/− had a high percentage of Smo-positive cilia in unstimulated cells, and no further increase was seen after treatment with sonic hedgehog (SHH)-conditioned medium (Fig. 2, B and C). Maea−/− cells had ∼26% Smo-positive cilia at the basal state, which increased to normal levels with SHH treatment (Fig. 2, B and C). Measurement of ciliary Smo intensity showed that SHH could further increase ciliary Smo levels in Arih2−/− and Maea−/− cells, but not Mgrn1−/− and Wwp1−/− cells (Fig. 2 D). Among the four E3 ligases, Arih2−/− cells had the highest ciliary Smo levels with or without SHH, whereas Maea−/− cells were the least affected (Fig. 2 D).
To determine if these four E3 ligases affect Gpr161 dynamics in cilia, we stained cells for Gpr161 with and without SHH treatment. As previously reported, Gpr161 is present in cilia at the basal state but not after SHH treatment, and mutation of Ift27 prevents its removal after SHH treatment (Eguether et al., 2014; Mukhopadhyay et al., 2013). All four E3 mutants behaved like WT controls, indicating that they are not critical for Gpr161 dynamics (Fig. S2).
As previously described (Marada et al., 2015), Western blot analysis of Smo in control cells shows a band of ∼90 kD that is likely to be immature Smo transiting the ER and an unfocused band at ∼120 kD that is thought to be a post-ER glycosylated form. The loss of Ift27, Maea, Mgrn1, or Wwp1 did not significantly alter the amount of these two forms of Smo (Fig. 2, E and F). However, the loss of Arih2 dramatically increased the levels of these forms (Fig. 2, E and F), suggesting that Arih2 is responsible for controlling the cellular levels of Smo by regulating its degradation. The 150-kD band (marked with an asterisk in the figures) is a protein that binds nonspecifically to the Flag antibody and appears to be up-regulated in the Arih2−/−, Maea−/−, and Mgrn1−/− cells (Fig. 2, E and H).
The localization of Wwp1 is regulated by Hedgehog signaling
Of the four E3 ligases that are needed to maintain low ciliary Smo at the basal state, only Wwp1 localized to cilia (Fig. 3, A and B; and Table S3). Interestingly, the ciliary localization of Wwp1 is regulated by Hedgehog signaling, because Wwp1 is found in ∼9% of cilia at the basal state and nearly none after stimulation with SHH (Fig. 3, A and B). The localization of Wwp1 to cilia was lost in Ptch1-null cells (Fig. 3, A and B). The low percentage of positive cilia was unexpected because the loss of Wwp1 caused most cilia to accumulate Smo. It is possible that more cilia have Wwp1 present, but below our detection level. To increase the sensitivity of detection, we fused an Avi tag to the C-terminus of Wwp1 and coexpressed this with BirA targeted to cilia by fusion with Sstr3. The Avi tag is a short sequence that can be biotinylated by BirA (Beckett et al., 1999). By coexpressing this with cilia-targeted BirA, any Wwp1-Avi that enters the cilium will be biotinylated, which can be detected by fluorescent avidin. Using this compartmentalized labeling method, Wwp1 was found in ∼74% of cilia at the basal state and ∼11% after stimulation with SHH (Fig. 3, C and D). The localization of Wwp1 to cilia decreased to ∼7% in Ptch1-null cells (Fig. 3, C and D). The Smo accumulation phenotype in Wwp1−/− cells can be rescued by the reintroduction of Wwp1, but not by the Wwp1C886A active site mutant (Fig. 3, E and F), indicating that its E3 ligase activity is required for the ciliary removal of Smo. Moreover, Wwp1−/− cells are normal compared with the parental cells in terms of percentage ciliation and ciliary length (Fig. S3).
Wwp1 is one of the eight Nedd4 family HECT E3 ligases that bind to Ptch1 (Kim et al., 2015). Thus, Ptch1 may control Wwp1’s ciliary localization by providing a binding platform. Confirming published results, we found that Ptch1 binds to Wwp1 (Fig. 4 A). This binding is specific because Ptch1 did not bind Arih2, Maea, Mgrn1, Ube2l3, Kctd5, Bap1, Skp2, or Skp1a (Fig. S4). The WW domain of Wwp1 mediates the interaction with Ptch1 (Fig. 4, B and C). WW domains often interact with PY motifs, of which there are two in Ptch1 (Fig. 4 D). The deletion of either of the Ptch1 PY motifs does not affect the binding of Ptch1 to Wwp1, but the deletion of both PY motifs prevents its binding to Wwp1 (Fig. 4 E). Ptch1-knockout cells lost ciliary localization of Wwp1, but this could be rescued by reexpression of Ty1-tagged Ptch1 (Fig. 4, F and G; and Fig. S5). Consistent with the binding data, Wwp1 localized to cilia when either of the Ptch1 PY motifs was deleted individually, but the deletion of both Ptch1 PY motifs blocked the localization of Wwp1 to cilia (Fig. 4, F and G). The deletion of either PY motif caused a modest increase in Smo-positive cilia at the basal state, while the deletion of both motifs caused most cilia to be Smo positive (Fig. 4 H). Similar experiments by Kim et al. (2015) did not detect increased ciliary Smo levels at the basal state when the Ptch1 PY motifs were mutated, but the reason for the difference is not known. At the level of gene expression, the loss of either PY motif caused a small subset of cells to show increased basal signaling, while the loss of both PY motifs increased the number of cells with elevated basal signaling (Fig. 4 I). These data suggest that Wwp1 activity against Smo is regulated by controlling its ciliary localization through Ptch1 binding.
Wwp1 is not required for the removal of Ptch1 from cilia
Ptch1, the receptor for SHH, localizes in cilia at the basal state and is removed upon stimulation with SHH (Rohatgi et al., 2007). Because of Wwp1’s role in regulating Smo localization, it is important to determine if it is involved in regulating the dynamics of Ptch1. We were unable to obtain usable Ptch1 antibodies by injecting rabbits or from commercial sources, so we created a knock-in line where three copies of the Ty1 tag were inserted into the endogenous Ptch1 gene at its C terminus (Fig. 5, A–C). As expected, Ptch1-Ty1 localized to cilia at the basal state. Upon stimulation with SHH, Ptch1 levels dropped within 30 min and then started to increase after 2 h. Wwp1−/− cells responded like the controls, indicating that Wwp1 is not involved in regulating the dynamics of ciliary Ptch1 (Fig. 5, D and E).
Smo has K63-linked Ub chains
If the ciliary level of Smo is regulated by a specific type of poly-Ub linkage, DUBs that cleave the critical poly-Ub chain should increase basal ciliary Smo levels by interfering with Smo’s removal. To test this idea, we asked how the expression of Usp2, which is active against all types of poly-Ub linkages, affects ciliary Smo levels. Unfortunately, expression of full-length Usp2 or its catalytic domain (CD) tethered to Sstr3-Myc was toxic to cells. We overcame this by the use of a tetracycline (Tc)-inducible expression system to regulate expression of Usp2CD (Fig. 6 A). After induction of Sstr3-Usp2CD by Tc, Smo accumulated in Sstr3-Myc-Usp2CD–positive cilia, supporting the hypothesis that ubiquitinated Smo is removed from cilia at the basal state (Fig. 6 B).
To understand the type of poly-Ub linkage involved in the ciliary removal of Smo, we targeted the DUBs AMSH* (Stam2-AMSH fusion), Otub1, Usp13, and Yod1 to cilia by fusing them to the C terminus of Sstr3. Each of these deubiquitinases recognizes a subset of poly-Ub chains and cleaves them selectively (Fig. 6 C; Mevissen et al., 2013; Sato et al., 2008; Tan et al., 2017). None of the tested DUBs had an appreciable effect on the total level of Smo in the cell (Fig. 6 D). Of the tested DUBs, only AMSH* caused Smo accumulation in cilia (Fig. 6, E and F). The accumulation was not seen when the active site (AMSH*D348A) was mutated (Fig. 6, E and F; McCullough et al., 2004), supporting the idea that the DUB activity was responsible for the Smo accumulation. AMSH* is thought to remove K63-linked Ub chains, suggesting that this is the form that is actively targeted for the removal of Smo. Furthermore, this is consistent with Wwp1 being the critical ciliary E3 because Wwp1 catalyzes K63 polymerization (French et al., 2017).
The ability of Wwp1 to ubiquitinate Smo was examined by coexpressing Ty1-tagged Ub and Wwp1 along with Smo in HEK 293T cells. Smo was then captured by immunoprecipitation, and the ligated Ub was detected with an anti-Ty1 antibody. As controls, Wwp1 was replaced with Wwp1C886A, which is enzymatically dead, and Smo was replaced with SmonoK or Smoloop3KRR, which should lack the expected sites of Wwp1 ubiquitination (Desai et al., 2020). A large amount of Ub was incorporated into Smo when coexpressed with Wwp1, and the amount was reduced in cells expressing enzymatically dead Wwp1. Likewise, replacement of Smo with SmonoK or Smoloop3KRR reduced the incorporation of Ub by Wwp1 (Fig. 6 G). These data support a model where Wwp1 is the E3 that catalyzes K63 polymerization of Ub onto Smo at K444 and K448. This was further supported by the observation that there are dramatically fewer K63-positive cilia in Wwp1−/− cells than K63-positive cilia in Ift27−/− cells, even though they all have equivalent levels of ciliary Smo accumulation (Fig. 6 H).
The E2 Ube2l3 regulates ciliary and cell body levels of Smo
Our initial screen of Ub components with connections to cilia or Hedgehog contained 10 of the 37 putative E2-conjugating enzymes. Of these, knockout of Ube2g2 decreased SAG-induced expression, Ube2i increased basal expression, and Ube2l3 increased basal expression while decreasing SAG-induced expression (Table S2). Since we identified four E3 ligases from different families as critical to regulating ciliary Smo, we had concerns that an unexamined E2 might be important. Thus, we generated guides for the remaining 27 E2 Ub–conjugating enzymes (Table S4). Knockout of these E2s did not alter the output of GreenBomb (Table S2), indicating that one of the initial three is likely to be the critical E2. Of the initial three, only Ube2l3 knockout increased ciliary Smo at the basal level, so we focused on this enzyme. This E2 works with Arih2 and Wwp1 (French et al., 2017; Marteijn et al., 2009; Marteijn et al., 2005), consistent with a role in regulating ciliary Smo levels. Amplification of the ORF from fibroblast cDNA identified two splice variants (Fig. 7, A and B). The longer one encodes a 154-residue protein (NCBI accession no. NP_033482), while the shorter one encodes a 122-residue protein (RefSet accession no. XP_006522067). When epitope-tagged versions were expressed in fibroblasts, both isoforms localized to the cytoplasm but, interestingly, the short isoform also localized to cilia (Fig. 7 C). To understand the importance of the two splice variants to Smo localization, Ube2l3-knockout cells were generated (Fig. 7 D). Direct measurement of Gli1 mRNA showed that the loss of Ube2l3 increased basal expression and decreased SHH responses, similar to what was seen with the original reporter (Fig. 7 I). Mutant cells were rescued with the two isoforms as well as the enzyme-dead versions of each isoform. The parental Ube2l3 mutant cells accumulated Smo in cilia at the basal state (Fig. 7, E and F). This ciliary Smo accumulation could be rescued by the 122-residue isoform but not by the 154-residue isoform or by the enzyme-dead versions of either isoform (Fig. 7, E and F). Western blot analysis indicated that the loss of Ube2l3 increased the levels of both the ∼90 kD ER and the ∼120 kD post-ER forms of Smo (Fig. 7, G and H). The increased level of Smo was not due to increased transcription (Fig. 7, I and J). Reexpression of either the long or short form of Ube2l3, but not the enzyme-dead versions, restored the amount of Smo detected by Western blotting (Fig. 7 K). The observation that only the short form localizes to cilia and rescues the ciliary Smo accumulation phenotype while either isoform can rescue the accumulation of Smo in the cell body suggests that Ube2l3 needs to be localized to cilia to regulate ciliary Smo levels.
Our recent work demonstrated that ciliary Smo levels are regulated by ubiquitination, and we proposed a model where Smo becomes ubiquitinated if it diffuses into cilia when the pathway is inactive. This causes it to be removed from the cilium by the IFT/BBSome system, thus keeping ciliary Smo levels low at the basal state (Desai et al., 2020). In this work, we set out to identify the enzymes that control the ubiquitination state of Smo. We focused on Ub-related proteins that had been identified in large-scale ciliary screens and proteomic studies. From among 147 candidates, 40 genes were identified that altered Hedgehog signaling, 9 of which caused Smo to accumulate abnormally in cilia at the basal state. This collection included four E3 Ub ligases, one E2 Ub–activating enzyme, one DUB, and three adaptors. Previous work in flies indicated that cell surface Smo (which is analogous to the ciliary Smo levels in mammals) is regulated by ubiquitination. A variety of enzymes were implicated, including Cul4 (Li et al., 2018a), Herc4 (Jiang et al., 2019), Smurfs (Li et al., 2018b), Uchl5 (Zhou et al., 2018), and Usp8 (Li et al., 2012; Xia et al., 2012). Curiously, none of these enzymes showed a phenotype in our screen. This was unexpected because the core components of the pathway are conserved between Drosophila and vertebrates. However, ciliary localization is likely to require different regulatory mechanisms than bulk cell surface localization, and this probably explains the lack of conservation of regulatory proteins.
Of the E3 ligases, Wwp1 localized to cilia, suggesting that it is the key enzyme regulating ciliary Smo levels. In support of this idea, Wwp1 localization in cilia was Ptch1 dependent and regulated by Hedgehog ligand. The localization was facilitated by interactions between the PY motifs in Ptch1 and the WW domains in Wwp1. Wwp1 is not known for roles in cilia or in Hedgehog signaling but was found in a comparative proteomic study of cilia (Sigg et al., 2017) and was predicted by the UbiBrowser (Li et al., 2017) to ubiquitinate several Hedgehog components. Wwp1 is a HECT family ligase whose activity has been studied with regard to cancer, where its increased activity is correlated with aggressive tumors (Chen et al., 2007). Knockout mice have short tibias (Mouse Genome Informatics and the International Mouse Phenotyping Consortium) and are obese (Hoshino et al., 2020), two phenotypes that are commonly seen in cilia-defective mice. Wwp1 is reported to catalyze monoubiquitination and the extension of a poly-Ub chain through the rapid addition of two to four K63-linked subunits followed by slower addition of a variety of linkages onto the initial short K63 chain (French et al., 2017). The addition of K63 subunits is consistent with our finding that cilia-localized AMSH*, which catalyzes the removal of K63 poly-Ub, causes Smo to be retained in cilia in the absence of pathway activation. Recently, Shinde et al. (2020) also found that K63 Ub chains earmark Smo for BBSome-mediated removal from cilia.
The E3 ligase Arih2 appears to be critical in regulating the amount of Smo in the cell because its loss caused a dramatic increase in the overall level of Smo in addition to causing Smo accumulation in cilia. Loss of Arih2 in mice leads to embryonic lethality, while heterozygotes show a variety of cilia-related phenotypes, including polycystic kidney disease and bone morphology defects (Bult et al., 2019; Dickinson et al., 2016). Loss of Arih2 elevates total Smo levels suggests that Arih2 regulates the degradation of Smo, possibly through a quality control mechanism.
In addition to Arih2 and Wwp1, our studies also identified the E3 ligases Mgrn1 and Maea as important to regulation of ciliary Smo levels. Mgrn1 loss was previously shown to increase ciliary Smo levels (Pusapati et al., 2017) and has been proposed to regulate the internalization of Smo (Kong et al., 2020). Maea was identified as part of a 10-subunit complex called the glucose-induced degradation–deficient/C-terminal to LisH (GID–CTLH) complex. This complex, which includes Maea, Ypel5, Wdr26, Rmnd5a, Ranbp9, Ranbp10, Armc8, Mkln1, Gid8, and Ttyh2, plays roles in the degradation of metabolic enzymes and cell proliferation (Lampert et al., 2018). Within the GID/CTLH complex, E3 activity is thought to come from the Rmnd5, and Maea is thought to play an adaptor role because it lacks critical residues for enzymatic activity (Maitland et al., 2019). Whether Maea regulates ciliary Smo levels as part of the GID/CTLH complex or in an alternative complex will require more experimentation.
The transfer of Ub to substrates by E3 ligases requires the activity of E2-conjugating enzymes. Screening the entire family of E2-conjugating enzymes revealed that loss of Ube2g2 decreased SAG-induced Hedgehog expression, loss of Ube2i increased basal expression, and loss of Ube2l3 increased basal expression while decreasing SAG-induced expression. Of these, only Ube2l3 localized to cilia. Ube2l3 can work with both Arih2 and Wwp1 (French et al., 2017; Marteijn et al., 2009; Marteijn et al., 2005). The loss of Ube2l3 caused a large increase in Smo as detected by Western blot analysis and increased the amount of Smo localized to cilia at the basal state. We identified two isoforms of Ube2l3, and reexpression of both of these rescued the increase in total Smo levels, but only the short isoform rescued the ciliary accumulation of Smo. This suggests that Ube2l3 must be localized to cilia to function in regulating the ciliary levels of Smo.
Our previous work suggested a model where ciliary Smo levels are regulated by ubiquitination at K444 and K448 in its intracellular loop 3 (Desai et al., 2020). Our current work indicates that ciliary Smo levels are kept low at the basal state by the action of the E3 ligase Wwp1 working together with the E2-conjugating enzyme Ube2l3. These two enzymes are localized to cilia and are expected to ubiquitinate Smo because Smo diffuses into cilia when the pathway is inactive. Pathway activation promotes the removal of Wwp1 from cilia, thus allowing any Smo that enters the cilium to remain and become activated by other mechanisms (Fig. 8). The regulation of cellular and ciliary Smo levels is complex and involves other E3 ligases, including Arih2 and Mgrn1. Arih2 localizes in the cytoplasm and is likely involved in regulating Smo degradation. Mgrn1 is thought to ubiquitinate Smo for its internalization and degradation (Kong et al., 2020). Together, our work outlines a complex pathway where ciliary Smo dynamics are regulated by coordinating the cytoplasmic and ciliary Smo levels during signaling.
Materials and methods
Plasmids were assembled by Gibson assembly (New England Biolabs) or the T5 exonuclease DNA assembly method (Xia et al., 2019) into the pHAGE lentiviral backbone (Wilson et al., 2008). All inserts are derived from the mouse unless otherwise stated. Inverse PCR, overlap extension PCR, and site-directed mutagenesis were performed as previously described (Lv et al., 2017). All inserts were fully sequenced and matched to the Ensembl reference sequence, National Center for Biotechnology Information reference sequence, or expected mutant forms. Plasmids are listed in Table S5, and SnapGene files will be provided upon request.
WT MEFs were derived from embryonic day 14 embryos and immortalized with simian virus 40 large T antigen. These cells were cultured in 95% DMEM (4.5 g/liter glucose), 5% FBS, 100 U/ml penicillin, and 100 µg/ml streptomycin (all from Gibco or Invitrogen). For the biotinylation of the Avi tag, 50 µM D-biotin was added for 24 h before fixing.
For SAG experiments, MEFs were plated at near-confluent densities and serum starved (same culture medium described above, but with 0.25% FBS) for 24 h before treatment to allow ciliation. SAG (Calbiochem) was used at 400 nM.
SHH-conditioned medium was generated from HEK 293T cells expressing Homo sapiens SHH (BL243), Xenopus tropicalis Scube2 (BL244), and Mus musculus Disp1 (BL323). Cells stably secreting SHH were grown to confluency in 90% DMEM (4.5 g/liter glucose), 10% FBS, 100 U/ml penicillin, and 100 µg/ml streptomycin. The medium was then replaced with low-serum medium (0.25% FBS) and grown for 24 h. Medium was collected, filtered, sterilized with a 0.45-µm filter (EMD Millipore), and titered for the ability to relocate Smo to cilia. Dilutions similar in effect to 400 nM SAG were used for experiments.
For inducible protein expression from the Tc-inducible promoter, MEFs were plated at 70% confluent densities and then serum starved for 24 h. Medium was then renewed with low-serum medium (0.25% FBS) containing 1 µg/ml Tc for another 24 h.
Lentiviral vector–packaged pHAGE-derived plasmids (Wilson et al., 2008) were used for transfection. These vectors are packaged by a third-generation system comprising four distinct packaging vectors (Tat, Rev, Gag/Pol, and either vesicular stomatitis virus G or murine leukemia virus envelope) using HEK 293T cells as the host. DNA (plasmid of interest, 5 µg; Tat, 0.5 µg; Rev, 0.5 µg; Gag/Pol, 0.5 µg; vesicular stomatitis virus G or murine leukemia virus envelope, 1 µg) was delivered to the HEK cells as calcium phosphate precipitates. After 48 h, the supernatant was harvested, filtered through a 0.45-µm filter (EMD Millipore), and added to subconfluent cells. After 24 h, cells were selected with corresponding antibiotics: nourseothricin (Nat; 50 µg/ml), puromycin (Puro; 1 µg/ml), zeocin (Zeo; 500 µg/ml), or blasticidin (Bsd; 60 µg/ml for cytomegalovirus [CMV] promoter or 20 µg/ml for Gallus gallus CryD1 promoter).
gRNAs were selected from the Brie library (Doench et al., 2016) or designed using CHOPCHOP (Labun et al., 2019), E-CRISP (Heigwer et al., 2014), and the Dharmacon CRISPR Design Tool (https://dharmacon.horizondiscovery.com/gene-editing/crispr-cas9/crispr-design-tool/); corresponding oligonucleotides were cloned into lentiCRISPR v2 Puro (gift from Feng Zhang, Broad Institute, Cambridge, MA; Addgene plasmid 52961; Sanjana et al., 2014) or lentiCRISPR v2 PuroP93S (BL245) and screened by sequencing. lentiCRISPR v2 PuroP93S is similar to its parent except for a proline-to-serine mutation in the Puro N-acetyltransferase gene, which increases its resistance to Puro (https://www.addgene.org). The vectors were packaged into lentiviral particles and transfected into MEF cells. After selection, the pools were analyzed by flow cytometry. Individual cells were sorted into 96-well plates by flow cytometry or dilution cloning. Single mutant clones were identified with Sanger sequencing, GENEWIZ Amplicon-EZ sequencing, immunofluorescence, or immunoblotting. Sequencing results were analyzed with CRISPResso2, TIDE (Tracking of Indels by Decomposition), CRISP-ID, and SWS (Sanger sequencing and Microsoft Word wildcard searching) methods (Brinkman et al., 2014; Clement et al., 2019; Dehairs et al., 2016; Jie et al., 2017). Ty1 knock-in was achieved by using CRISPR-Cas9 genome editing. Cas9 and sgRNA were expressed from lentiCRISPR v2 Nat (OC166). The template for homology-directed repair was designed in Benchling (https://www.benchling.com/). Each homology arm was ∼800 bp. To enhance the editing efficiency, the target sequence of the gRNA and the protospacer adjacent motif sequences were added to both ends of the repair template (Zhang et al., 2017). Genome-edited cell lines were obtained and validated by Sanger sequencing and immunofluorescence. The following primers were used for detecting Ty1 knock-in of Ptch1: Ptch1KI_1, 5′-TGTGTTGCTCCGAGCTTTGT-3′; Ptch1KI_2, 5′-TCCATCCGGACTCCAGACAT-3′; Ptch1KI_3, 5′-GAACTTCAGCATCAAGCGGAT-3′; Ptch1KI_4, 5′-CATTTGCTTCTGGCCACACC-3′; Ptch1KI_5, 5′-AAAAGGTCCCCCAGGGTCTA-3′; Ptch1KI_6, 5′-ATGACCTCAGAAGCTGCCAC-3′. For the knock-in experiments, cells were transfected using a standard calcium phosphate method or Qiagen Effectene Transfection Reagent according to the manufacturer’s protocol.
For flow sorting, pelleted cells were resuspended in the corresponding media and sorted into tubes or 96-well plates with a BD FACS C-Aria II Cell Sorter (BSL-2+/BSC). For flow analysis, cells were seeded at nearly confluent densities in 12-well plates, serum starved for 24 h, and treated with or without SAG or SHH conditioned media for 48 h. Cells were trypsinized, pelleted, and then resuspended in 500 µl 1× PBS. Flow analysis was performed with a digital multiparameter flow cytometer analyzer (LSR II; BD Biosciences). Data were processed in FlowJo version 10. Ridgeline plots were drawn in RStudio with ggplot2 and ggridges (https://github.com/clauswilke/ggridges) in R 3.5.3.
Cells were fixed with 2% PFA for 15 min, permeabilized with 0.1% Triton X-100 for 2 min, and stained as described elsewhere (Follit et al., 2006). In some cases, fixed cells were treated with 0.05% SDS for 5 min before prehybridization to retrieve antigens. The primary and secondary antibodies are described in Table S6.
Most confocal images were acquired at room temperature with an inverted microscope (TE-2000E2; Nikon) equipped with a Solamere Technology modified spinning disk confocal scan head (CSU10; Yokogawa). Three-image Z stacks were acquired at 0.2-micron intervals and converted to single planes by maximum projection with MetaMorph software (MDS Analytical Technologies). Confocal images were obtained at room temperature with an LSM910 microscope equipped with a 63× objective and converted to a maximum projection with ZEN 3.1 blue edition (Zeiss). Widefield fluorescence images were captured at room temperature using a forward-looking infrared camera on a Zeiss Axiovert 200M microscope equipped with a 100× Zeiss objective using μManager (Vale Lab, University of California, San Francisco). Images were processed in ZEN 3.1 blue edition (Zeiss) or Adobe Photoshop CS6 with brightness and contrast adjusted.
Protein and mRNA analysis
For Western blots, cells were pelleted and lysed directly with denaturing gel loading buffer (Tris-HCl 125 mM, pH 6.8, glycerol 20% vol/vol, SDS 4% vol/vol, β-mercaptoethanol 10% vol/vol, and bromophenol blue). The primary and secondary antibodies are described in Table S6. Western blots were developed by chemiluminescence (Pierce Super Signal West Dura; Thermo Fisher Scientific) and imaged using an Amersham Imager 600 imager (GE Healthcare Life Sciences). Bands were quantified with Gel-Pro Analyzer 4 (Meyer Instruments).
For immunoprecipitations, cells were serum starved for 48 h, and proteins were extracted with lysis buffer (20 mM Hepes, pH 7.5, 50 mM KCl, and 1 mM MgCl2) with 0.5% digitonin and protease inhibitor (cOmplete EDTA-Free; Roche). Insoluble components were removed by centrifugation at 20,000 g. Primary antibodies preadsorbed to protein G-sepharose beads (GE Healthcare) were added to the cell extract, and the mixture was incubated for 2 h at 4°C. After centrifugation, beads were washed with lysis buffer supplemented with 0.1% digitonin before elution in denaturing gel loading buffer for SDS-PAGE electrophoresis and Western blot analysis.
Smo ubiquitination assay
HEK 293T cells were plated at 60% confluent density onto a 10-cm plate. After 24 h, the cells were transfected with 5 µg of each vector using calcium phosphate transfection. 24 h after transfection, cells were treated with 1 µM MG132 for 4 h to block proteasomal degradation. Cells were lysed and Smo was captured as described in the immunoprecipitation method in Protein and mRNA analysis.
Statistical results were obtained from at least three independent experiments. Statistical differences between groups were tested by one-way ANOVA, two-way ANOVA, or repeated measures ANOVA in GraphPad Prism 7.04. Differences between groups were considered statistically significant if P < 0.05. Otherwise, nonsignificance was labeled on the figures. Statistical significance is denoted as *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001. Error bars indicate SD.
Online supplemental material
Fig. S1 shows the development of a Hedgehog reporter line. Fig. S2 shows that Arih2, Maea, Mgrn1, or Wwp1 do not regulate the ciliary localization of Gpr161. Fig. S3 shows that Wwp1 knockout does not affect ciliogenesis. Fig. S4 shows that Arih2, Maea, Mgrn1, Ube2l3, Kctd5, Bap1, Skp2, and Skp1a do not interact with Ptch1. Fig. S5 shows that knockout of Ptch1 in GreenBomb activates the Hedgehog pathway. Table S1 provides a complete list of candidate genes and sources. Table S2 lists flow cytometry and other data from the screen. Table S3 describes subcellular localization of the Hedgehog regulators. Table S4 lists E2 ubiquitin-conjugating enzymes screened. Table S5 lists plasmids used in the study. Table S6 lists the primary and secondary antibodies and detection reagents used in the study. Table S7 lists RT-PCR and quantitative RT-PCR primers used in the study.
All data are available in the main text or the supplemental materials.
We thank Dr. Carol E. Schrader and the staff of the University of Massachusetts Medical School Flow Cytometry Core for assistance during this project. We thank Dr. Xiaoou Zhang for analyzing Amplicon-EZ next-generation sequencing data.
This work was supported by the National Institutes of Health (grant GM060992 to G.J. Pazour). Flow cytometry resources were supported by National Institutes of Health grant S10 1S10OD028576 to Carol Schrader, Ph.D., University of Massachusetts Medical School.
The authors declare no competing financial interests.
Author contributions: Conceptualization: B. Lv, O.A. Cabrera, and G.J. Pazour; methodology: B. Lv, O.A. Cabrera, M.W. Stuck, P.B. Desai, and G.J. Pazour; validation verification: B. Lv, M.W. Stuck, P.B. Desai, and G.J. Pazour; formal analysis: B. Lv, M.W. Stuck, P.B. Desai, and G.J. Pazour; investigation: B. Lv, O.A. Cabrera, M.W. Stuck, P.B. Desai, and G.J. Pazour; writing – original draft preparation: B. Lv and G.J. Pazour; writing – review and editing: B. Lv, O.A. Cabrera, M.W. Stuck, P.B. Desai, and G.J. Pazour; visualization preparation: B. Lv; supervision: G.J. Pazour; funding acquisition: G.J. Pazour.