Hedgehog Signaling

One hallmark ciliary signaling pathway that highlights the dynamics in PTMs is Hedgehog signaling in vertebrates (Figure 1; Gigante and Caspary, 2020). An elegantly orchestrated interplay of positive and negative regulators in Hedgehog signaling allows for the correct patterning of the developing embryo, in addition to maintaining adult tissue homeostasis (Shimada et al., 2019). Gradients of the hedgehog morphogens ultimately result in finely tuned levels of active GLI transcription factors that determine target gene expression (Briscoe and Novitch, 2008). In the absence of Hedgehog morphogens, their receptor Patched (PTCH1) localizes to the primary cilium (Rohatgi et al., 2007), while the key Hedgehog effector and G protein-coupled receptor (GPCR) Smoothened (SMO) surveys the primary cilium by shuttling in and out without appreciable local accumulation (Figure 1A; Kim et al., 2009; Goetz and Anderson, 2010). A second GPCR, the constitutively active GPR161, stimulates ciliary adenylyl cyclases to increase cAMP levels within cilia and thereby activates the cAMP-dependent protein kinase (PKA) (Mukhopadhyay et al., 2013). GPR161 also fulfills a second function in PKA signaling, as it serves as an atypical A-kinase anchoring protein (AKAP) that targets PKA to cilia (Bachmann et al., 2016). Here, it tethers to the cilia-resident PKA regulatory subunit RIα that senses ciliary cAMP (Mick et al., 2015). At high ciliary cAMP levels in unstimulated cells, PKA-RIα binds cAMP and releases the catalytic PKA-C subunit (Figure 1A; Taylor et al., 2004). Free PKA-C can phosphorylate and regulate target proteins, such as the GLI transcription factors that convey the signaling status to the nucleus (Tuson et al., 2011; Niewiadomski et al., 2014). PKA-mediated phosphorylation of GLIs is a pre-requisite for their proteolytical cleavage to yield repressor forms that block the transcription of target genes (Figure 1A). GLI transcription factors are precisely regulated by a variety of activating and deactivating PTMs, which include activating phosphorylations at the N-terminal repressor domain, and two clusters of PKA phosphorylation sites on the activator domain. PKA phosphorylation precedes further phosphorylation by CK1 and GSK3β, which in turn recruits the SCF E3 ubiquitin ligase that marks GLIs for proteolytic processing by the proteasome (Kong et al., 2019; Niewiadomski et al., 2019).

Upon Hedgehog ligand binding, PTCH1 exits the primary cilium and SMO is activated and retained in cilia, whereas GPR161 is removed (Rohatgi et al., 2007; Gigante and Caspary, 2020). As the adenylyl cyclase inhibitory SMO replaces the stimulating GPR161, ciliary cAMP decreases (Mukhopadhyay et al., 2013). Consequently, PKA activity ceases and the GLI transcription factors are no longer phosphorylated and further processed, such that they can function as activators to initiate target gene expression in the nucleus (Figure 1B).

One central element in Hedgehog signaling is the dynamic re-localization of the components involved. Similar to other cellular protein trafficking mechanisms, PTMs control the localization of Hedgehog signaling proteins, which is particularly well-studied for GPR161 (Mukhopadhyay et al., 2013; Pal et al., 2016). GPR161’s C-terminal tail not only contains the AKAP binding domain for PKA but also several protein kinase consensus sites, including one for PKA (Bachmann et al., 2016). Upon Hedgehog pathway activation, the C-terminal tail of GPR161 is phosphorylated by GRK2 and presumably PKA (Bachmann et al., 2016; Pal et al., 2016; May et al., 2021). GRK-mediated phosphorylation recruits the molecular sensor of activated GPCRs, specifically β-arrestin2, which is required for the removal of activated GPCRs from cilia (Figure 1B; Pal et al., 2016). Consequently, GPR161 exits cilia together with its binding partner PKA (May et al., 2021). Thereby, PKA activity in cilia is inhibited by two mechanisms: (i) reducing cAMP levels and (ii) removing PKA itself.

As exemplified by the GLI transcription factors and GPR161, specific phosphorylations are often catalyzed by individual kinases; however, our knowledge of specific protein phosphatases that antagonize these phosphorylations in cilia is still rudimentary. The protein phosphatases PP1 and PP2A have been reported to dephosphorylate SMO to dampen Hh signaling in Drosophila (Su et al., 2011; Liu et al., 2020). However, since primary cilia are dispensable for Drosophila Hh signal transduction, it remains unclear whether PP1 and PP2A also function within primary cilia. Mass spectrometric analyses have identified PP1 subunits in isolated Chlamydomonas cilia (Pazour et al., 2005) and PP2A subunits in primary cilia of kidney epithelial cells (Ishikawa et al., 2012), but these findings still await confirmation by independent methods. In contrast, lipid phosphatases, such as the inositol polyphosphate-5-phosphatase E, have been unambiguously shown to localize to primary cilia, where they modulate ciliary signal transduction by regulating protein trafficking (Chávez et al., 2015; Garcia-Gonzalo et al., 2015). Ciliary lipid phosphatase activities create a specific phosphatidylinositide phosphate environment that is required for efficient ciliary signaling.

A recent study investigated the involvement of ubiquitin in Hedgehog signaling by fusing mono-ubiquitin to the C-terminus of SMO (Desai et al., 2020). The SMO-Ub fusion accumulated in cilia in the absence of stimulation in IFT and BBSome mutants but failed to accumulate in cilia after Hedgehog pathway activation in wild type cells (Desai et al., 2020). These findings indicate that ubiquitin is required for the removal of SMO from cilia by a process involving IFT and the BBSome. Moreover, β-arrestin2 was shown to mediate the ubiquitylation of GPR161 in response to Hedgehog pathway activation (Shinde et al., 2020), before GPR161 exits the primary cilium in a BBSome-dependent fashion (Ye et al., 2018). More evidence for the central role of ubiquitylation for cilia trafficking comes from mutational analysis of the Hedgehog receptor PTCH1. PTCH1 harbors two E3 ubiquitin ligase recognition motifs and remains in the cilium when both motifs are mutated, even upon stimulation with Hedgehog ligands (Kim J.C. et al., 2015). SMO has been reported to be a target of the ubiquitin ligase HERC4 (Jiang et al., 2019). Furthermore, ubiquitylation of SMO by a complex of the E3 ubiquitin ligase MGRN1 and the plasma membrane protein MEGF8 serves as a signal for proteasomal degradation (Kong et al., 2020). Yet, whether these ubiquitin ligases are directly involved in the IFT-dependent retrieval of SMO awaits experimental validation.

Molecular dissection of ubiquitylation may help to decipher the different functions of ubiquitin in regulating ciliary proteins. Upon Hedgehog pathway activation, specifically K63-linked ubiquitin chains increase in primary cilia upon GPCR activation or in BBSome mutants (Shinde et al., 2020). This suggests that K63 ubiquitin chains function as export signals for ciliary proteins, which are recognized by the BBSome (Desai et al., 2020; Shinde et al., 2020). In BBSome mutant mice, photoreceptor outer segments, which are uniquely modified cilia that harbor the entire signaling cascade for visual phototransduction, accumulate more than 100 proteins that are absent in wild types (Datta et al., 2015). Based on these findings, the BBSome has been proposed to mediate the removal of unwanted proteins from cilia and, therefore, may function as an important mediator of a ciliary protein quality control network (Shinde et al., 2020). Additional components, such as the AAA-ATPase VCP or the ubiquitin-regulatory X domain protein UBXN10, have been shown to localize to primary cilia (Mick et al., 2015; Raman et al., 2015). While data from trypanosomes indicate that the BBSome may directly recognize ubiquitin as it can be enriched on ubiquitin-agarose resin (Langousis et al., 2016), it does not contain canonical ubiquitin binding domains. How ubiquitylated proteins are recognized in cilia on a molecular level and what enzymatic activities regulate ubiquitylation within cilia remains to be established. The E3 ubiquitin ligase CBL is recruited to cilia in response to PDGFRα signaling (Schmid et al., 2018) and the deubiquitylase UBPY/USP8 has been reported to antagonize SMO ubiquitylation in Drosophila (Ma et al., 2016). These findings seem promising starting points for future studies elucidating the cilia-specific ubiquitylation network.

Cilium Disassembly

In contrast to cilia formation, we are just beginning to understand the molecular details of how cilia are dismantled to allow cell cycle re-entry (Liang et al., 2016; Breslow and Holland, 2019). NEK2, a kinase predominantly expressed in the S and G2 phases of the cell cycle, has been proposed to promote cilium disassembly (Figure 2D; Kim S. et al., 2015). Among several targets, NEK2 phosphorylates and stimulates the microtubule-depolymerizing KIF24 at the distal centriolar ends (Kim S. et al., 2015). This may not only block unwanted cilium assembly but also shift the balance toward disassembly when resting cells re-enter the cell cycle (Kim S. et al., 2015; Viol et al., 2020). In support of a central role for KIF24 in cilium disassembly, a recent study identified FLS2 as a CDK-like kinase that phosphorylates the KIF24 ortholog CrKIF13 in Chlamydomonas, allowing efficient cilia disassembly (Figure 2D; Zhao et al., 2020). In turn, phosphorylated FLS2 showed lower activity and appears to be dephosphorylated upon cilia disassembly when it enters cilia by binding to the IFT-B component IFT70 (Zhao et al., 2020). While the precise mechanisms of regulation still need to be established, it is tempting to speculate that dephosphorylation of FLS2 may not only alter its kinase activity but also unmask targeting signals for IFT. This suggests a mechanism by which an active kinase can be directed into cilia to promote disassembly by phosphorylating specific targets such as the microtubule-depolymerizing kinesin KIF24.

In cilia of Chlamydomonas, a ubiquitin conjugation system has been identified more than a decade ago (Huang et al., 2009), yet the involvement of ubiquitin in cilia disassembly has only recently been demonstrated. Despite a massive rise in the ubiquitin levels in shortening cilia, semi-quantitative mass spectrometric analysis of a temperature-sensitive Chlamydomonas model has only detected an increase in ubiquitylation of α-tubulin and ubiquitin itself (Wang et al., 2019). The study further revealed α-tubulin poly-ubiquitylation by K63 chains, which allows binding to the IFT-A subunit IFT139 for tubulin removal via retrograde IFT (Figure 2D; Wang et al., 2019). Intriguingly, the authors also observed an increase in K11 and K48 chains in response to cilia shortening. K11 chains are also assembled by the anaphase-promoting complex to drive proteasomal degradation of substrates during mitosis, suggesting potential mechanisms for cell cycle-dependent regulation (Matsumoto et al., 2010).