Primary cilia, A‐kinase anchoring proteins and constitutive activity at the orphan G protein‐coupled receptor GPR161: A tale about a tail

Primary cilia are non‐motile antennae‐like structures responsible for sensing environmental changes in most mammalian cells. Ciliary signalling is largely mediated by the Sonic Hedgehog (Shh) pathway, which acts as a master regulator of ciliary protein transit and is essential for normal embryonic development. One particularly important player in primary cilia is the orphan G protein‐coupled receptor, GPR161. In this review, we introduce GPR161 in the context of Shh signalling and describe the unique features on its C‐terminus such as PKA phosphorylation sites and an A‐kinase anchoring protein motif, which may influence the function of the receptor, cAMP compartmentalisation and/or trafficking within primary cilia. We discuss the recent putative pairing of GPR161 and spexin‐1, highlighting the additional steps needed before GPR161 could be considered ‘deorphanised’. Finally, we speculate that the marked constitutive activity and unconventional regulation of GPR161 may indicate that the receptor may not require an endogenous ligand.

Orphan G protein-coupled receptors (GPCRs) have been referred to as 'pharmacologically dark receptors' (Huang et al., 2015) because the absence of an identified endogenous ligand (hence their 'orphan' designation) has made it very difficult to discern much about their biological roles.One fascinating orphan receptor is GPR161, which is localised to primary cilia within the cell where it regulates local cAMP concentrations with exquisite control.GPR161 plays an essential role in development and its cellular localisation and activation is controlled by an influential morphogen, the secreted lipoglycoprotein Sonic Hedgehog (SHH).Before exploring the potential roles of GPR161 in cilia biology, we will first review the primary cilia itself.We will then discuss the latest research looking at GPR161 function and explore the evidence for its first purported ligand, spexin-1.The primary cilium is characterised by the presence of several structures: The central axoneme that is anchored by the basal body and contains microtubules arranged in a 9 + 0 configuration; the ciliary membrane, enriched in proteins and lipids that are distinct from those in the plasma membrane; transition fibres, which serve to regulate entry and exit into the primary cilium; and ciliary pocket, a site of clathrin-mediated endocytosis.Anterograde movement from the base to tip of the primary cilium is mediated via the intraflagellar transport (IFT)-B complex with a kinesin-2 motor protein, while retrograde movement from tip to base is mediated via the IFT-A complex with dynein-2 motor protein.(b) Canonical Shh signalling pathway.In the 'OFF' state, Patched 1 (PTCH1) and GPR161 are localised to the primary cilium membrane.GPR161 acts as a negative regulator of the pathway via its constitutive signalling in which protein kinase A (PKA) promotes GLI3 phosphorylation.Further processing of GLI3 via ubiquitination and limited proteolysis forms GLI3R and enables its activity in repressing the transcription of genes targeted by the Shh pathway.In the 'ON' state, the PTCH1 antagonist, SHH, triggers the entry of Smoothened (SMO) into the primary cilium and the exit of PTCH1 and GPR161 via ubiquitination and endocytosis, respectively.Active SMO promotes GLI2/3A formation via dissociation from Suppressor of Fused (SuFu), allowing for the transcription of Shh target genes.Created with BioRender.com. the human body, the primary cilium is a sensory organelle that acts as an 'antenna' for the cell to receive and transduce extracellular signals to intracellular responses (Wheway et al., 2018).During neural tube development, neuroepithelial cells project primary cilia at their apical surface into the central canal, where they detect morphogen gradients via receptors present in the ciliary membrane.As with all cilia, the primary cilium exists as an extension of the plasma membrane and is supported by a central axoneme at its core, which consists of several microtubules comprised of polymerised αand β-tubulin proteins (Wheway et al., 2018) (Figure 1a).It is akin to the motile cilia found in upper airway tracts and sperm flagellum; however, primary cilia diverge from motile cilia in their core configuration.Microtubules in motile cilia are arranged in a 9 + 2 circular configuration: 9 outer doublets and 2 inner singlets-the central singlets provide a scaffold on which motor proteins can travel to generate flexibility and motility.In the primary cilium, the microtubules are arranged in a 9 + 0 configuration-the absence of the two central singlet microtubules being the defining nonmotile characteristic (Satir et al., 2010).The axoneme is anchored by the structurally distinct basal body at the root of the primary cilium.Here, the ciliary membrane forms a deep groove called the ciliary pocket, in which Yshaped transition fibres connect the basal body to the membrane (Reiter et al., 2012;Wheway et al., 2018).Although the composition of these transition fibres is unknown, it is seemingly a docking site for motor proteins, thereby restricting the entry and exit of cargo in the primary cilium (Reiter et al., 2012).The ciliary pocket is also a site rich in clathrin, a critical component of clathrin-coated pits, and transferrin, an endosomal marker.The localisation of these two components at the ciliary pocket supports the idea that receptor endocytosis occurs at this site (Molla-Herman et al., 2010), rather than within the primary cilia, which do not appear to contain endocytic vesicles (Benmerah, 2013;Rohatgi & Snell, 2010).
Receptors present on the ciliary membrane sense morphogen gradients and other extracellular ligands to direct intracellular responses.
Trafficking of receptors into the primary cilium is an essential part of this signalling network and is mediated by various intraflagellar transport (IFT) motors and complexes-these allow bidirectional movement to and from the primary cilium (Hsiao et al., 2012) (Figure 1a).Rapid ligand binding and signalling within the primary cilium requires fast entry and exit of receptors, thus relying heavily on a highly coordinated system of IFT motor proteins and associated complexes such as Tulp3, BBSome complex and Rab proteins (Mukhopadhyay & Rohatgi, 2014;Rosenbaum & Witman, 2002;Shinde et al., 2020).
Anterograde movement of cargo towards the tip of the primary cilia is facilitated by the IFT-B complex accompanied by the kinesin-2 motor unit, while retrograde movement is facilitated by the IFT-A complex accompanied by the dynein-1b motor unit (Rosenbaum & Witman, 2002).Thus, primary cilia are highly dynamic and organised structures where protein localisation is tightly regulated.

| Sonic Hedgehog signalling in the primary cilium
One of the major pathways in the vertebrate primary cilium is the Hedgehog (Hh) signalling pathway, of which the Sonic Hedgehog pathway (Shh) is crucial for the regulation of embryonic development.
Three main receptors facilitate the downstream effects of Shh signalling in primary cilia: Patched1 (PTCH1), a 12-transmembrane receptor; Smoothened (SMO), a seven transmembrane (7TM) Class F GPCR; and the Class A orphan GPCR, GPR161.There are two states of Shh signalling: 'OFF' and 'ON' (Figure 1b).In the 'OFF', or unstimulated basal state, there is an absence of the lipoglycoprotein SHH, the endogenous antagonist ligand of PTCH1.In this state, active PTCH1 and GPR161 receptors are localised to the ciliary membrane, where PTCH1 acts to repress the plasma membrane activity of SMO-the mechanism by which this occurs is currently unclear (Mukhopadhyay et al., 2013;Mukhopadhyay & Rohatgi, 2014;Rohatgi et al., 2007).
Several GLI transcriptional regulators (GLI1, GLI2 activator; GLI3 repressor) are present within cilia, however, in the basal state of the pathway, they are repressed by Suppressor of Fused (SuFu) in the cytoplasm.GLI3 is phosphorylated by protein kinase A (PKA) and further processed by ubiquitination and proteolysis events, triggered by downstream GPR161 signalling via the Gα s pathway (Mukhopadhyay et al., 2013).These steps convert GLI3 to its active repressor form, GLI3R, allowing translocation to the nucleus and transcriptional inhibition of Shh pathway target genes.For the 'ON' or active state, the antagonistic effects of SHH on PTCH1 release the inhibition of SMO, allowing SMO entry into the primary cilium, while PTCH1 and GPR161 are trafficked out of the cilium, presumably via ubiquitination (Kim et al., 2015;Shinde et al., 2020) or β-arrestin recruitment, respectively (Pal et al., 2016;Short, 2016).Within the primary cilium, SMO recruits the SuFu-GLI2/3 complex and promotes the dissociation of the SuFu-GLI complex, releasing full-length GLI2 and GLI3 to be converted to their activator forms, GLI2A and GLI3A.These activators then translocate to the nucleus to promote gene transcription.Downstream Shh target genes include PTCH1 and GLI1 for feedback loop regulation, and MYC and BCL2 for regulation of cell growth and survival (Mukhopadhyay et al., 2013).

| ORPHAN RECEPTOR GPR161: A KEY PLAYER IN PRIMARY CILIA
One of the major receptors involved in primary cilia signalling is the Class A 'orphan' GPCR, GPR161, so called because an endogenous ligand has not yet been ratified by the International Union of Basic and Clinical Pharmacology Nomenclature Committee.Structurally, this receptor is similar to other GPCRs in that it contains an extracellular N-terminus and intracellular C-terminal tail, three extracellular and intracellular loops (ECL and ICL, respectively), and a 7TM domain (Figure 2).Class A GPCRs contain several key motifs important for maintaining GPCR inactive and active state conformations and overall function (Nygaard et al., 2009;Rovati et al., 2007;Zhou et al., 2019).
These include the D(E)RY motif in TM3, N(D)PXXY motif in TM7, the CWXP motif in TM6, PIF motif formed by residues in TM3, TM5 and TM7, and the conserved D 2.50 residue in TM2, which is associated with a putative allosteric Na + binding pocket (Katritch et al., 2014;White et al., 2018).Of note, the highly evolutionarily conserved N(D) F I G U R E 2 Predicted secondary structure of human GPR161, from GPCRdb, highlighting key residues and motifs involved in receptor structure and function.Snake plot generated by GPCRdb (Kooistra et al., 2021) for human GPR161.Ballesteros and Weinstein numbering used for transmembrane domain residues.N-term: N-terminus; ICL: intracellular loop; ECL: extracellular loop; C-term: C-terminus.Note: the V158E mutation is a substitution mutation in the second intracellular loop of murine GPR161 which abolishes the constitutive activity of the receptor.While the mutation is referred to as V158E by the original authors (Mukhopadhyay et al., 2013), we calculate that it must be V129 3.54 in humans and V145 3.54 in mouse given the reported location three amino acids downstream of the DRY motif (DRYxxV).Residue 158 is an asparagine according to accession number NM_001081126.2.PXXY (where X denotes any amino acid) motif is present as HPXXY in GPR161, where the asparagine (N) amino acid is replaced by histidine (H).While the relevance of an asparagine to histidine change is unknown, one group has hypothesised that the same motif in the ATP receptor, P2Y 11 , may indicate engagement with alternative G protein activation mechanisms, however, this has not been validated experimentally (Zylberg et al., 2007).GPR161 has several predicted posttranslational modifications, as expected of a Class A GPCR, including glycosylation at amino acid residues 4 and 15, and a disulfide bond between residues Cys 100 and Cys 178 (between the top of TM3 to the middle of ECL2).No signal peptide has been proposed for GPR161.
Several phosphorylation sites have been predicted and mapped on GPR161 in ICL1, ICL3, and the C-terminal tail, notably the PKA phosphorylation motif, RXX(S/T), which is present in GPR161 as RRSS at residues 427-430 (https://www.phosphosite.org/uniprotAccAction?id=Q8N6U8).Interestingly, GPR161 has a long C-terminal tail which is proposed to have an A-kinaseanchoring protein (AKAP)-binding motif, the only GPCR to be identified as such (Bachmann et al., 2016).
The relevance of the AKAP motif and additional features highlighted in Figure 2 will be discussed in further detail below.

| Subcellular and tissue localisation of GPR161
GPR161 plays a crucial role in a number of processes during vertebrate embryonic development, such as left-right axis patterning, limb bud formation and differentiation of neural progenitor domains (Hwang & Mukhopadhyay, 2015;Mukhopadhyay & Rohatgi, 2014), which is driven by its negative regulation of the Shh pathway.
GPR161 is largely localised to the primary cilium of a cell where it appears to remain in a constitutively active state, that is, active in the absence of a bound ligand, via the Gα s -coupled signalling pathway.
Gα s -stimulated production of cAMP leads to activation of the catalytic subunits of PKA (PKA-C) and then ultimately to formation of the repressor form of GLI3, GLI3R by PKA-C mediated phosphorylation as well as ubiquitination and limited proteolysis (Mukhopadhyay et al., 2013).Following Shh pathway activation, GPR161 levels are reduced exclusively in the cilia, indicating that GPR161 is shuttled out of the primary cilium and prevented from interacting with ciliary downstream effector molecules, ultimately leading to attenuation of signalling (Mukhopadhyay et al., 2013).As shown by co-localisation with the endocytic marker, transferrin, small pools of GPR161 have also been localised to endocytic vesicles in NIH-3T3 cells (Mukhopadhyay et al., 2013) and mouse embryonic fibroblasts (Hwang et al., 2021) with stable expression of the receptor.In the latter experiment, cells were co-stained with acetylated tubulin and γtubulin, which are markers of the primary cilium axoneme and basal body (Hwang et al., 2021).This suggests that receptor internalisation appears to occur via recruitment of the scaffolding proteins β-arrestins (βarr1 and βarr2) (Hwang et al., 2018;Mukhopadhyay et al., 2013;Pal et al., 2016).At the very least, βarr2 is important in the internalisation of GPR161, as deletion of almost the entire C terminus of mouse GPR161 (truncation at position 376) led to complete loss of βarr2 recruitment, however, a smaller deletion that corresponds to the vacuolated lens (vl) mutant (truncation at position 401, described below) enhanced βarr2 recruitment (Pal et al., 2016).
Additionally, mutation of all putative phosphorylation sites to alanine in the 476-401 amino acid region of GPR161, resulted in both reduced βarr2 recruitment and removal of GPR161 from cilia (Pal et al., 2016).While βarr2, and not βarr1, is necessary for the ciliary trafficking of the somatostatin SST 3 receptor (Green Jill et al., 2015), there is some evidence for βarr1 involvement in GPR161 signal termination.For example, following SHH treatment, GPR161 depletion from cilia in murine embryonic fibroblasts was only lost in cells isolated from βarr1 and βarr2 double knockout mice, although direct BRET-based interaction studies showed that βarr2 was the dominant binding partner (Pal et al., 2016).Taken together, it can be concluded that the GPR161 C-terminus interacts with β-arrestins and that this most likely occurs before the truncation caused by the vl mutation.
RNA expression data from the Human Protein Atlas (proteinatlas.org) reveals that GPR161 is highly expressed in numerous human tissues, such as the endometrium and smooth muscle, although its role in these tissues is not clear.Reverse transcription PCR (RT-PCR) and digoxigenin-labelled riboprobes against murine Gpr161 revealed expression as early as E8.5 and E9.5 (Matteson et al., 2008;Mukhopadhyay et al., 2013).In situ hybridisation by the Millonig group suggested that GPR161 was primarily expressed in the lateral neural folds in early stages and in the developing hindlimbs in later stages of development (Matteson et al., 2008).However, this is contrary to findings where Gpr161 was detected all throughout the neural tube and minimally in the hindlimb (Mukhopadhyay et al., 2013).GPR161 is also present in several other structures, such as the central nervous system, retina, lens, spinal cord and dorsal ganglia (Matteson et al., 2008;Mukhopadhyay et al., 2013).Collectively, these data suggest that GPR161 has robust expression throughout embryonic development and neurogenesis, consistent with its reported phenotypes, below.

| GPR161 pathologies
GPR161 is involved in several pathologies, described both in humans and in animal models.For example, a homozygous missense mutation in the extracellular region of the GPR161 protein was implicated as a cause of pituitary stalk interruption syndrome (PSIS) in two related patients (Karaca et al., 2015).Meanwhile, loss of both copies of GPR161 (due to a copy-neutral loss of heterozygosity event and biallelic inactivation of the affected alleles) is thought to predispose carriers to the Shh subtype of medulloblastoma, which typically occurs in infants and sometimes children (Begemann et al., 2020;Dahlin et al., 2020).Supporting its role as a tumour suppressor, conditional deletion of Gpr161 in mouse neural stem cells leads the formation of tumours in the cerebellum that share histological features with human medulloblastoma (Shimada et al., 2018).Additionally, several rare variants of GPR161 have been identified in human spina bifida patientstwo of these variants are thought to disrupt both Shh and Wnt signalling pathways during neural tube differentiation (Kim et al., 2019).
Several animal models have been developed that highlight the importance of GPR161 in neural tube development and the Shh signalling pathway.For example, the vl mutation initially occurred spontaneously in the inbred C3H/HeSnJ mouse line (Matteson et al., 2008).The resulting phenotypes were studied extensively, and it was shown that the vl mutation caused two different phenotypes in mice.Approximately half of the vl/vl mouse embryos displayed lumbar-sacral spina bifida, while the other half displayed phenotypes such as thinning of the midline neuroepithelium and epidermis, dilation of the dorsal ventricle and the presence of ectopic neuroepithelial cells in the ventricle (Matteson et al., 2008).Additionally, $50% embryonic lethality was observed in the vl/vl mice, likely to be caused by the lumbar-sacral spina bifida, as the surviving vl/vl mice did not present with this deformity, but rather with congenital cataracts (Korstanje et al., 2008;Matteson et al., 2008).It is interesting to note these phenotypes did not appear in wild-type (WT) or heterozygous (+/vl) mice, and that the penetrance of the different vl phenotypes is affected by the genetic background of the animals (Korstanje et al., 2008;Matteson et al., 2008).Via positional cloning of the vl locus, the Millonig group reported over a decade later that GPR161 was responsible for the phenotypes in vl mutant mice (Matteson et al., 2008).The vl mutation is an 8 bp deletion predicted to cause a frameshift and premature stop codon at amino acid 386 in humans and 401 in mice (normally histidine, see Figure 2), leading to the loss of $70% of amino acids in the C-terminal tail of GPR161 (Matteson et al., 2008).Because the C-terminal tail of GPCRs is important for receptor endocytosis, it was predicted that deletion of most of the Cterminal tail would affect this mechanism by causing the loss of several putative phosphorylation sites, including the PKA phosphorylation motif at residue 430 (Matteson et al., 2008).Using immunofluorescence, the Millonig group showed that the vl mutation did indeed prevent GPR161 endocytosis, with the receptor colocalising with a plasma membrane-targeted GFP but not the FITC-labelled endosome marker, transferrin (Matteson et al., 2008).The effect of the vl mutation on trafficking within the primary cilium could not be measured as HEK293T cells are not ciliated under basal growth conditions; nor was a role for βarr explored.Subsequently, the previously described study by Pal et al. (2016) showed that a version of GPR161 that was truncated from the vl site onwards had paradoxically increased recruitment of βarr2 in their non-ciliated and ciliated cell models but could not exit cilia upon Shh pathway activation (Pal et al., 2016).
While deletion of a portion of the C-terminal tail of GPR161 causes different lens phenotypes with variable embryonic lethality, complete Gpr161 knockout via homologous recombination causes 100% lethality in mice at embryonic day 10.5 (Mukhopadhyay et al., 2013).These embryos displayed severe craniofacial defects that were detectable in the early stages of embryonic development, as well as unfused forebrain and midbrain regions on the head and a distinct lack of limb bud formation (Mukhopadhyay et al., 2013).Upon knockin of a GPR161 variant (Gpr161 mut1 ) in mice in which the receptor was unable to localise the primary cilium but was still signalling competent, craniofacial defects and limb bud formation were also observed (Hwang et al., 2021).In Gpr161 knockout embryos, there was an increase in the downstream expression of Ptch1 and Gli1, confirmed using RNA in situ hybridisation, where Ptch1 and Gli1 transcripts were expressed beyond the ventral region of the neural tube (Mukhopadhyay et al., 2013).Additionally, the expression of ventral cell-type-specific transcription factors such as FoxA2, Nkx2.2, Nkx6.1 and Olig2, was extended to the lateral and dorsal regions of the entire neural tube-this coincided with decreased expression of the predominantly dorsal cell-type-specific transcription factors Pax6 and Pax7 (Mukhopadhyay et al., 2013).Interestingly, a similar phenotype was recently observed in zebrafish, where the Shh target genes ptch2, gli1, nkx2.2b and nkx6.1 were up-regulated while pax7a was inhibited (Tschaikner et al., 2021).Overall, this data clearly demonstrates that Shh signalling is increased in Gpr161 knockout embryos, causing abnormal neural tube patterning and craniofacial defects (Mukhopadhyay et al., 2013).This may explain the differences in phenotypes between the Gpr161 knockout mice and vl hypomorphs-the former results in increased signalling of the Shh pathway while the latter prevents GPR161 internalisation but not its expression, which may mean some Shh opposition remains.Interestingly, the spina bifida phenotype that is present in vl/vl mice (which retain GPR161 in cilia) was also observed in Gpr161 mut1 mice (which are ciliary localisationdefective) (Hwang et al., 2021).Another reason why the vl mutation does not result in such detrimental phenotypes may be because GPR161 could regulate other pathways downstream of its activity, such as the canonical Wnt pathway or retinoic acid pathway (Li et al., 2015).It may also be because of the importance of GPR161 at different stages of neural development-that the vl mutation disrupts events that occur in late neural development (i.e., fusion of neural folds) may be a reason why there is lower embryonic lethality compared to the null allele (Li et al., 2015).While this hypothesis requires further investigation, both models highlight the tight regulation of the Shh/GPR161 pathway and its important role in the overall formation of the neural tube in embryonic development.
A cardiac looping phenotype has been reported in zebrafish, where in vivo knockdown of gpr161 was achieved using two different morpholino antisense oligonucleotides that targeted the 5 0 untranslated region (5 0 UTR) of zebrafish gpr161 mRNA to block its translation in early embryonic development (Leung et al., 2008).Injection of either gpr161 morpholino resulted in embryos with aberrant cardiac looping that was only partially restored via co-injection of morpholino-resistant gpr161 RNA (Leung et al., 2008).The group attributed this incomplete rescue of the cardiac looping phenotype to the lack of spatiotemporal control of the morpholino-resistant gpr161 RNA expression, given gpr161 expression would ordinarily be tightly controlled by the Shh signalling pathway.The group also reported an additional embryonic phenotype of aberrant left-right patterning, which is essential for positioning of organs in the developing embryo (Leung et al., 2008).Expression of genes that determine laterality were decreased in gpr161 knockdown embryos, particularly the expression of bmp4, a cardiac marker that is normally asymmetrically expressed in the left side of the developing heart (Leung et al., 2008;Smith et al., 2008).In the gpr161 knockdown zebrafish embryos, approximately half showed uniform expression on both sides of the developing heart (Leung et al., 2008).Recently however, this observation was challenged by the Stefan and Aanstad groups (Tschaikner et al., 2021), who noted that 100% of their gpr161 mutant zebrafish embryos showed laterality with the myocardium and that the defects observed by the Robishaw group (Leung et al., 2008) may be attributed to the off-target effects of the morpholino approach.In mice, knockdown of Gpr161 also did not result in aberrant cardiac looping, with cardiac contractions being observed up until death ($E10.0)(Mukhopadhyay et al., 2013).Indeed, there is clear evidence in published reports, casting doubt on the use of morpholino approaches to investigate animal development, as the off-target effects make it difficult to link phenotype to the original gene of interest (Eisen & Smith, 2008).Therefore, it is important to have appropriate controls in place or use more precise methods such as TALEN-mediated gene editing to ensure reproducible effects on phenotype specific to the gene of interest.

| CILIARY SIGNALLING AND THE ROLE OF CAMP COMPARTMENTALISATION
It is clear that the function of GPR161 is very tightly regulated within the cell, whether that be developmental regulation by SHH concentration gradients, by subcellular localisation within primary cilia, or by internalisation and removal from the ciliary surface.This intricate regulation is seemingly at odds with the observation that GPR161 is a constitutively active receptor, that is, one potential level of GPR161 regulation, the ability of a ligand to turn the receptor on, is paradoxically absent (as far as current evidence might suggest-more on this later).Instead, it seems that GPR161 has adapted other mechanisms to control signal transduction by manipulating cAMP compartmentalisation.

| Mechanisms of cAMP regulation
The second messenger cAMP is synthesised by membrane-localised adenylyl cyclases and then binds to and activates downstream effector molecules, including PKA, cyclic nucleotide-gated channels and the exchange proteins activated by cAMP (EPACs), to mediate signalling events (de Rooij et al., 1998;Nikolaev & Lohse, 2006;Zaccolo et al., 2021).In its inactive form, PKA is a tetrameric holoenzyme composed of two regulatory (R) and two catalytic (C, PKA-C) subunits, where the linker regions of the R subunits occlude the active site of the C subunits to prevent kinase function (Ramms et al., 2021).The C subunits are encoded by three genes producing Cα, Cβ and Cγ, while the R subunits are encoded by four genes producing RIα, RIβ, RIIα and RIIβ (Tschaikner et al., 2020).Different combinations of RI and RII type subunits and C subunits are responsible for targeting the PKA holoenzyme to specific subcellular locations, usually in association with AKAPs and in the presence of substrates or target proteins (Tschaikner et al., 2020;Zaccolo et al., 2021).The EPAC isoforms EPAC1 and EPAC2 are also directly activated by cAMP via a cAMP binding domain in a PKA-independent manner, and are similarly compartmentalised within the cell (Zaccolo et al., 2021).
The AKAP family of more than 50 scaffolding proteins is a key mediator of cAMP compartmentalisation, creating finely tuned cAMP microdomains by bringing various elements of cAMP signal transduction in close proximity (e.g., PKA regulatory subunits, adenylyl cyclases, EPACs and phosphodiesterases [PDEs]) (Wong & Scott, 2004).AKAPs contain a short 14-18 amino acid amphipathic α-helix that anchors the PKA holoenzyme via the N-terminal docking and dimerization (D/D) domains on the RI and RII type PKA subunits (Carr et al., 1991).At the target subcellular location, cAMP binds to the R subunits and induces a conformational change within the holoenzyme that removes the R subunit linker from the C subunit active site, releasing the active PKA-C to phosphorylate target proteins (Torres-Quesada et al., 2017).A final level of spatiotemporal control of cAMP signalling pools comes from the subcellular localisation and binding kinetics of various isoforms of cAMP-specific PDEs.These enzymes are responsible for the breakdown and hydrolysis of cAMP into ATP, with their cellular location thought to lead to microdomains: cAMP 'sinks' are formed where PDE concentrations are high, while cAMP 'hot spots' would exist where PDE expression is low or absent (Baillie, 2009).There are eight PDE families that act on cAMP that are further divided into subfamilies characterised by distinct tissue location and subcellular domains-they can also scaffold with AKAPs to form cAMP signalling complexes (Baillie, 2009;Epstein, 2017;Omori & Kotera, 2007;Tschaikner et al., 2020).For example, there are 4 subfamilies (A, B, C and D) within the PDE4 family with over twenty isoforms that contain N-terminal targeting sequences directing them to specific subcellular domains, such as PDE4A1 that is only localised to the plasma membrane, or PDE4D isoforms that are localised to the cytoplasm (Baillie, 2009;Omori & Kotera, 2007).Meanwhile, PDE4D5 associates with β-arrestin to terminate β 2 adrenoceptor signalling and promote receptor internalisation, thereby limiting further generation of cAMP (Baillie, 2009;Omori & Kotera, 2007).Importantly, there is evidence of a primary cilium-specific PDE isoform, with PDE4C found to localise to primary cilia in renal epithelial cells as part of a larger complex that includes AKAP150 and adenylyl cyclases 5 and 6 (Choi et al., 2011).

| GPR161-A PKA target and AKAP
GPR161 is seemingly unique amongst GPCRs in that it is both a receptor and an AKAP.As shown in Figure 2, there is a short 20 amino acid long amphipathic α-helix in the receptor's C terminal tail spanning residues 459 to 478, which anchors the PKA holoenzyme via the RI type PKA subunits (Bachmann et al., 2016).RI type subunits typically have low affinity for RI-selective AKAPs, compared to RII type subunits, which traditionally have high affinity for their partner AKAPs (Torres-Quesada et al., 2017).Of note, GPR161 is one of only a handful of proteins with high affinity interactions between their AKAP motif and the RIα type PKA subunits (Bachmann et al., 2016;Torres-Quesada et al., 2017).Moreover, GPR161 is also a substrate for the very protein that it scaffolds, with an RRSS* PKA phosphorylation motif found between residues 427-430 (Figure 2).Direct evidence for an interaction between GPR161 and RIα was provided using an RLuc-based protein complementation assay in non-ciliated HEK293 cells, where perturbations of the amphipathic α-helix within the GPR161 C-terminus reduced complex formation (Bachmann et al., 2016).This interaction is further supported by the presence of PKA-RIα in the primary cilia of IMCD3 cells (Mick et al. 2015) and the concomitant exit of PKA-RIα and GPR161 from the cilia (May et al., 2021).Furthermore, alanine substitution of the putative PKA phosphorylation motif abolished PKA-specific phosphorylation of GPR161 in response to increased levels of cAMP after stimulation with a β-adrenoceptor agonist or the adenylyl cyclase activator, forskolin (Bachmann et al., 2016).Interestingly, the interaction between GPR161 and the RI subunit was not affected by mutation of the PKA phosphorylation site.The relevance of these interactions to primary cilia was investigated by mRNA injection of GPR161-mCherry and RIα-GFP mRNAs into zebrafish embryos.Alanine substitutions did not affect ciliary localisation of the receptor, however, mutation of GPR161 to a pseudo-phosphorylated form via aspartic acid substitutions decreased GPR161-mCherry in the cilia, implying that phosphorylation results in ciliary exit of GPR161 (Bachmann et al., 2016;May et al., 2021).GPR161 exits the cilia upon activation of the Shh pathway, likely to be mediated by G protein-regulated kinases (GRKs) and β-arrestins (Pal et al., 2016), and more recently, PKA-mediated phosphorylation of GPR161 has been proposed to contribute to the ciliary exit of GPR161 (May et al., 2021).Overall, the data from this study demonstrates that GPR161 acts as an AKAP for the type I PKA holoenzyme via the amphipathic α-helix in its C terminal tail, and its phosphorylation is mediated by the active PKA-C subunits.
Very recently, an exciting study revealed that SMO also interacts with PKA during Shh signal pathway activation (Happ et al., 2022).Typically, the catalytic activity of PKA is suppressed in the holoenzyme by the 'inhibitor sequence' in the regulatory subunit occupying the C subunit catalytic site; in type I holoenzymes this sequence acts as a pseudosubstrate (Ramms et al., 2021).Unlike GPR161, which binds to the RI subunit of PKA via its AKAP, SMO contains a protein kinase inhibitor (PKI) motif in its C-terminus that itself occupies the catalytic site of the C subunit to sterically inhibit its activity (Happ et al., 2022).Mutation of three key residues in the SMO PKI motif abolished co-localization between SMO and the PKA-C subunit in ciliated IMCD3 cells, prevented PKA-C downregulation in HEK293 cells, and rendered the receptor incapable of inducing GLI-mediated transcription in response to Shh pathway stimulation in Smo À/À mouse embryonic fibroblasts (Happ et al., 2022).
These paradoxical effects between the AKAP domain in GPR161 and the PKI motif in SMO are likely to contribute to the fine-tuned control of ciliary PKA and downstream effects on Shh signalling.

| Ciliary and extra-ciliary cAMP signals
As cAMP is a key modulator in the primary cilium, there is speculation whether cells can distinguish between cAMP produced in the cilium (ciliary cAMP) and the cell body (extra-ciliary cAMP) (Truong et al., 2021).Moreover, do cAMP pools within the primary cilia play a role in downstream signalling?A recent landmark paper by the Reiter group elegantly sought to answer these questions using optogenetic and chemogenetic tools in zebrafish that were designed to localise cAMP generation to either the cytoplasm or cilium (Truong et al., 2021).These tools showed that only cAMP generated within the primary cilium inhibited Hh pathway activation and Hh-dependent cell specification, both in vitro and in vivo.By extension, if ciliary cAMP generation inhibits Hh pathway activation, then agonism of cilialocalised Gα s -coupled GPCRs should do the same and, conversely, stimulation of Gα i -coupled receptors should enhance Hh signalling.Indeed, the authors showed that stimulation of the Gα i -coupled SST 3 receptor in ciliated mammalian cells activated Hh-mediated GLI1 expression.
Similarly, by using Designer Receptors Exclusively Activated by Designer Drugs (DREADDs) to control signalling of a Gα s -coupled GPCR with clozapine-N-oxide, the group was able to show that only the ciliary-localised DREADD and not plasma membrane-localised DREADD inhibited the Hh pathway (Truong et al., 2021).Finally, transgenic zebrafish embryos expressing bacterial photoactivatable adenylyl cyclase (bPac) constructs targeted to either the cytoplasm (Cyto-bPac) or cilia (Cilia-bPac), were used to show that blue light stimulation of the reporters generated cAMP in the cytoplasm and primary cilium at comparable levels.Activation of the Hh signalling pathway was shown via downstream Gli:mCherry reporter expression.In line with the idea that ciliary pools of cAMP negatively regulated the Hh pathway, Gli:mCherry expression in somites was reduced upon both forskolin treatment and blue light stimulation of Cilia-bPac, but not Cyto-bPac (Truong et al., 2021).The effects of the GPR161 mut1 variant, which is defective in terms of ciliary localisation but still signalling competent, showed a gradual increase in both Hh pathway activity and phenotype severity, observed in GPR161 mut1/mut1 , GPR161 mut1/ko and GPR161 ko/ko embryos (Hwang et al., 2021).This was attributed to the resulting ratio of Gli activator to repressor formed by the ciliary and extra-ciliary pools of GPR161, demonstrated by the gradual decrease in amount of Gli3R formation that caused the distinct phenotypes across different tissues (Hwang et al., 2021).This could be particularly relevant for GPR161 signalling in the primary cilium and may help elucidate exactly how GPR161 signals via Gα s and cAMP to repress the Shh signalling pathway.A diagram showing the current understanding of GPR161 signalling is presented in Figure 3.

| DOES GPR161 HAVE AN ENDOGENOUS LIGAND?
Studies so far have relied upon GPR161 mutation, deletion, constitutive activity, or regulation by other ligands that activate cAMP/PKA.Arguably, development of GPR161 as a therapeutic target is unlikely to progress further until the endogenous ligand has been identifiedindeed, this is the ultimate prize for most orphan receptor pharmacologists.However, the unusual mode of regulation of GPR161 signalling, as demonstrated by its high constitutive activity, unique AKAP domain and it being a PKA phosphorylation target, does prompt the question-does GPR161 actually have or need an endogenous ligand?
Recently, a group in Copenhagen developed a sophisticated computational approach to the de-orphanisation of putative peptide receptors (Foster et al., 2019).Briefly, the team first identified 21 potential peptide GPCRs by looking at structural-and sequencebased similarities to known peptide receptors.They then mined the human proteome to predict potential precursor molecules to peptides, with the process whittling down >20,000 input proteins to $4800 proteins that were annotated as being secreted or possessing a signal peptide (of which only 120 were known GPCR ligands), then to $1400 proteins that were considered potential precursor molecules to peptides; these were filtered based on length (<750 amino acids) and the requirement that they have an unknown molecular function.
Peptide ligands that would require further, more complex maturation processes were excluded.Using these criteria, a final library of 218 peptides was formed, comprised of 120 new peptides from the filtered human proteome dataset, 49 known peptide ligands for GPCRs, 35 variants of known peptides and 14 known peptides for which no GPCR activity was shown (Foster et al., 2019).To reduce F I G U R E 3 Summary of the current understanding of GPR161 signalling.(a) GPR161 couples to Gα s G proteins (Gαβγ heterotrimers), binds to β-arrestins via its proximal C-terminus, and scaffolds inactive PKA holoenzymes via its C-terminal AKAP domain.(b) Constitutive signalling of GPR161 is mediated by the dissociation of the Gα s subunit from the Gβγ subunits and subsequent adenylyl cyclase (AC) stimulation of cAMP generation.(c) The AKAP domain on the C-terminal tail is an amphipathic helix which binds to type I PKA holoenzymes via their RI subunits (R; blue).cAMP generated by the receptor binds to the regulatory RI subunits, causing a conformational change that leads to the release of the catalytic subunits (PKA-C; pink) which can then phosphorylate their target substrates.Two of these substrates are GPR161 itself and GLI transcription factors.(d) Endocytosis of GPR161 occurs first by phosphorylation of its C-terminal tail by G protein receptor kinases (GRKs), followed by β-arrestin recruitment and, lastly, formation of clathrin-coated pits.Created with BioRender.com. the chance that a biological effect would be missed due to an inappropriate choice of endpoint, the 21 potential peptide receptors were screened using three main assays to explore G protein and β-arrestin signalling pathways: the dynamic mass redistribution assay that measures cellular changes in an agnostic manner, a time-resolved-FRET (TR-FRET) internalisation assay, and a β-arrestin recruitment assay ('PRESTO-Tango', as described by Kroeze et al., 2015).Additionally, assays measuring accumulation of inositol 1-phosphate and of cAMP, and Ca 2+ mobilisation, were used to further explore Gα q/11 , Gα s and Gα i/o -mediated signalling of potential positive pairs.An incidental observation to come from the PRESTO-Tango β-arrestin screen, which included all orphan GPCRs rather than the originally predicted 21 peptide receptors, was that one of the 218 peptides, spexin-1, activated GPR161 (Foster et al., 2019).In the assay, GPR161 appeared to recruit βarr2 in a concentration-dependent manner (pEC 50 , 6.57 ± 0.47) although the magnitude of effect was much smaller than other tested GPCRs (E max : GPR161-spexin-1, 1.4 arbitrary units [a.u.]; BB 3 -neuromedin B, 5.0 a.u.; GPR1-cholecystokinin, 15.0 a.u.) (Foster et al., 2019).Because GPR161 was not part of the original screen using dynamic mass redistribution or internalisation assays, the authors concluded that the spexin-1/GPR161 pairing was putative only and required further validation.Nevertheless, given that GPR161 is an orphan and the receptor has not previously been linked to any ligands at all, this pairing demands further investigation.

| What do we know about spexin-1?
Spexin-1 (neuropeptide Q, NPQ; NWTPQAMLYLKGAQ) is a 14 amino acid neuropeptide derived from pre-pro-spexin, a precursor responsible for generation of at least one other peptide, known as spexin-2 (NPQ53-70) (Mirabeau et al., 2007;Sonmez et al., 2009;Toll et al., 2012).Spexin-1 was identified by two separate groups using different hidden Markov model-based computational approaches and shown to be highly conserved in mammals, birds and fish (Mirabeau et al., 2007;Sonmez et al., 2009), though not in rats due to an amino acid mutation located in the C-terminal cleavage site immediately distal to the spexin-1 sequence (Toll et al., 2012).In line with the trafficking of secreted peptide hormones such as insulin, spexin-1 was localised by immunocytochemistry to secretory granules in a transfected rat pancreatic cell line, and spexin mRNA was expressed in the submucosal layer of the mouse oesophagus and stomach fundus (Mirabeau et al., 2007).The same group also demonstrated that spexin-1 induced concentration-dependent contractions in rat stomach muscle (EC 50 , 0.75 μM) (Mirabeau et al., 2007).Since its discovery, spexin-1 has been implicated in a variety of physiological pathways -obesity and glucose homeostasis, regulation of food intake and energy expenditure, lipid metabolism, cardiovascular and renal homeostasis, gastrointestinal function and inflammation-in a number of species including humans, mice, rats, goldfish and zebrafish (Al-Daghri et al., 2018;Ani et al., 2017;Behrooz et al., 2020;Kakarala & Jamil, 2014;Kołodziejski et al., 2018;Kumar et al., 2016;Lin, Huang, et al., 2018;Lin, Zhao, et al., 2018;Porzionato et al., 2010;Walewski et al., 2014;Zheng et al., 2017).However, these observations are yet to translate to the development of novel therapeutic agents, or even a consensus in the field as to its genuine physiological role.
Of the potential roles of spexin-1, perhaps the most therapeutically attractive is its reported role in cardiometabolic disease, which is consistent with its expression pattern in humans.The spexin-1 precursor, prepro-spexin, was identified by Northern blotting to be restricted to brain, kidney and pancreas (Sonmez et al., 2009), although a more recent study employing conventional RT-PCR, qRT-PCR and immunohistochemistry staining found more widespread distribution of spexin-1 including in the pancreas, adrenal gland, skin, stomach, small intestine and liver (Gu et al., 2015).However, the human evidence for a cardiometabolic role of spexin-1 is largely derived from correlative studies.For example, several studies have found that spexin-1 is decreased in serum from patients with Type 1 and Type 2 diabetes mellitus, gestational diabetes mellitus and obesity (Gu et al., 2015;Kołodziejski et al., 2018;Kumar et al., 2016), and gene expression profiling of obese patients showed that spexin-1 mRNA was significantly decreased in adipose tissue compared to non-obese patients (Walewski et al., 2014).Some cross-sectional studies have shown no correlation between serum spexin-1 and fasting blood glucose, triglyceride, low-density lipoprotein cholesterol (Kumar et al., 2016), body composition or fitness levels (Hodges et al., 2018), while others have shown a negative correlation between serum spexin-1 and age, body mass index, fasting blood glucose and triglycerides (Gu et al., 2015;Lin, Huang, et al., 2018).Many of these studies are limited by small sample sizes and lack of diversity-for example, a longitudinal study investigating the effect of gastric bypass surgery on spexin-1 levels included data from only white female patients, limiting the generalisations of the study (Kumar et al., 2018).Thus, while the link between spexin-1 and cardiometabolic disease features is promising, direct measures such as knock-out mice, or phenotypes amongst humans with gene variants, would be more compelling (to our knowledge there is not a published mouse knockout model, although there has been a TALEN spexin-1 zebrafish knockout that showed a hyperphagic phenotype; Zheng et al., 2017).
It seems counterintuitive that galanin and spexin-1 have opposing effects yet share a common receptor, suggesting that another receptor such as GPR161 might be the endogenous target of spexin-1.
However, the pharmacological evidence for spexin-1 activity at the galanin receptors is growing.In the original study that paired spexin-1 with the galanin receptor family, spexin-1 treatment of HEK293 cells that were modified to signal via a SRE-luciferase reporter resulted in increased reporter activity via GAL 2 and GAL 3 , but not at GAL 1 or kisspeptin receptors-spexin-1 was actually more potent than galanin (Kim et al., 2014).This interaction was supported by competitive radioligand binding assays in which spexin-1 displaced galanin and spexin-1 radioligands at GAL 2 and GAL 3 receptors (Kim et al., 2014).
Interestingly, spexin-1 appears to modify food intake only via GAL 3 receptors (Lv et al., 2020).Selective antagonists of GAL 2 or GAL 3 receptors (M871 and SNAP37889, respectively) and spexin-1 were in mice fed a normal diet.SNAP37889, but not M871, coinjection significantly blocked spexin-1-induced decreased food intake, suggesting that this effect is exerted by GAL 3 receptors.This effect is not entirely in line with the fact that GAL 2 and GAL 3 receptors have overlapping expression (Marcos & Coveñas, 2021), with effects thought to occur through both spexin-1 and galanin (Mills et al., 2021).Thus, more work is needed to tease apart the contributions of the different ligands to galanin receptor physiology.An alternative explanation for the differential signalling outcomes of spexin-1 and galanin at the galanin receptors is biased agonism, which is the propensity of different ligands at the same receptor to exert different signalling outcomes.In a NanoBiT β-arrestin recruitment assay, spexin-1 was a GAL 2 partial agonist for both βarr1 and βarr2, with an EC 50 of $300 nM (Reyes-Alcaraz et al., 2018), which is very similar to that reported in the PRESTO-Tango β-arrestin recruitment assay (EC 50, 269 nM) (Foster et al., 2019).Ligand bias was evident by the changed rank order of potencies for galanin, spexin-1 and the spexinderived galanin agonist, Fmoc-dA4-dQ14, in inositol phosphate accumulation, SRE-luciferase and pERK1/2 assays (Reyes-Alcaraz et al., 2018).
By further testing the ligands in either wild type, Gα q/11 -or β-arrestin 1/2-knockout cell lines, the authors confirmed that relative to the reference ligand galanin, spexin was biased towards each G proteindependent assay tested, when compared to β-arrestin pathways.Finally, β-arrestin conformational biosensors were used to reveal that galanin and spexin-1 engaged different β-arrestin conformations upon binding (Reyes-Alcaraz et al., 2018), providing good evidence that spexin-1 is indeed biased towards G protein-mediated signalling at GAL 2 receptors.

| UNRESOLVED QUESTIONS REGARDING GPR161 AND SPEXIN-1
Over recent decades, numerous orphan receptor/endogenous ligand pairings have been claimed, with very few going on to be ratified by the International Union of Basic and Clinical Pharmacology Nomenclature Committee (Davenport et al., 2013).There are many reasons why the proposed partners do not stand the test of time-the original papers may have only investigated one or a few concentrations over a narrow range, thereby missing obvious non-specific effects; knockout mice may not have been available until after the original deorphanisation publication; the apparent specificity of the pairing may actually represent an indirect effect of the orphan receptor on the true ligand target; the activity is genuine but outside the concentrations reached within the body, and many other possible causes.For these reasons, the proposed pairing of spexin-1 with GPR161 must be treated with caution and warrants further careful investigation in parallel experiments where GPR161 is either present or absent.This is also likely why Foster et al. (2019) were careful to label the pairing as 'indicative' only and noted that because the receptor wasn't part of their original study aims, they did not follow up the finding in that paper (Foster et al., 2019).Certainly, from a coexpression point of view there is only moderate overlap in expression of spexin-1 and GPR161, both in specific tissues and over the course of development.We also note that it is likely that GPR161 is not a conventional GPCR.For instance, GPR161 appears to have very high constitutive activity, it regulates compartmentalised cAMP via a unique AKAP in its C-terminal tail while also being a PKA substrate, and its signalling appears to be terminated by removal from primary cilia.For these reasons we would argue that GPR161 does not necessarily have an endogenous ligand, although that will be difficult to prove as the absence of evidence is not evidence of absence.

| Nomenclature of Targets and Ligands
Key protein targets and ligands in this article are hyperlinked to corresponding entries in http://www.guidetopharmacology.org and are permanently archived in the Concise Guide to PHARMACOLOGY 2021/22 (Alexander, Christopoulos et al., 2021;Alexander, Fabbro et al., 2021;Alexander, Kelly et al., 2021).

2
| THE PRIMARY CILIUM 2.1 | Structure and role of the primary cilium in neural development Non-motile primary cilia are key cellular structures that modulate neural development in the growing embryo.Found on almost every cell type in F I G U R E 1 Diagram showing the structure of primary cilia and canonical Shh signalling.(a)