The complexities of G-protein-coupled receptor kinase function in Hedgehog signaling

Abstract

Hedgehog (Hh) signaling is essential for proper tissue patterning and maintenance and has a substantial impact on human disease. While many of the main components and mechanisms involved in transduction of the Hh signal have been identified, the details of how the pathway functions are continually being refined. One aspect that has attracted much attention recently is the involvement of G-protein-coupled receptor kinases (GRKs) in the pathway. These regulators of G-protein-coupled receptor (GPCR) signaling have an evolutionarily-conserved function in promoting high-threshold Hh target gene expression through regulation of Smoothened (Smo), a GPCR family member that activates intracellular Hh signaling. Several models of how GRKs impact on Smo to increase downstream signaling have been proposed. Recently, we demonstrated that these kinases have surprisingly complex and conflicting roles, acting to limit signaling through the pathway while also promoting Smo activity. In addition to the previously described direct effects of Gprk2 on Smo activation, Gprk2 also indirectly affects Hh signaling by controlling production of the second messenger cyclic AMP to influence Protein kinase A activity.

Keywords: :

• Drosophila
• G-protein-coupled receptor kinase
• cyclic AMP
• hedgehog
• protein kinase A
• smoothened

This article refers to:

The overall growth and patterning of embryonic tissues is largely governed by signaling pathways that are activated in response to secreted growth factors, or morphogens. The Hedgehog (Hh) proteins are one of the principal families of morphogens that are required for the normal development of most tissues in animals ranging from flies to humans.¹ The activity of Hh proteins is not limited to development, as they also play important roles in stem cell maintenance, and thus tissue homeostasis, in adults.² Not surprisingly given its prominence, misregulation of Hh signaling has wide-ranging biological consequences, and has been identified as the cause of a variety of congenital syndromes and cancers.^3,4

A number of conserved proteins required for transducing the signal downstream of Hh have been identified by genetic studies in model organisms, and these have been assembled into a core “canonical” pathway (reviewed in refs. 5,6). The functioning of this pathway is best understood in Drosophila. As in mammals, the key initiator of intracellular signaling in flies is Smoothened (Smo), a seven-pass transmembrane protein that, together with Frizzled proteins, forms a distinct class of the G-protein-coupled receptor (GPCR) protein family.⁷ Under basal conditions, Smo activity is kept low by the Hh receptor Patched. Hh binding relieves Patched-mediated repression of Smo. Smo in turn becomes highly phosphorylated, translocates to the cell surface, and signals through a complex of proteins that includes the serine/threonine kinase Fused (Fu), the scaffold protein Costal-2 (Cos2), and Suppressor of Fused (SuFu) to control the proteolytic processing and activity of the transcription factor Cubitus interruptus (Ci). Pathway activation leads to stabilization of Ci, reducing the levels of a proteolytically-cleaved 75 kDa transcriptional repressor form (Ci⁷⁵) and increasing the levels of a full-length transcriptional activator form (Ci¹⁵⁵), thereby promoting Hh target gene expression.

Like all morphogens, Hh proteins form concentration gradients in tissues centered on their sites of production, and differentially activate target genes in a concentration-dependent manner. This means that the ligand distribution gradient has to be converted to graded pathway activity in the field of target cells. This appears to be achieved in large part through graded activity of Smo. Like other GPCRs, Smo can exist in at least two conformational states with different signaling capacities.^8,9 Transition between these states is controlled by phosphorylation of the cytoplasmic tail of Smo. More extensive phosphorylation either increases the probability of Smo adopting an active conformation or induces a greater conformational shift, leading to higher levels of Smo activity and higher-threshold target gene activation.⁸ Smo phosphorylation is thus an important determinant of Hh activity.

G-Protein-Coupled Receptor Kinases: Conserved Regulators of Hh Signaling

In Drosophila, Protein kinase A (Pka) and Casein kinase I (CkI) are the main kinases responsible for activating Smo. In response to Hh, Smo is phosphorylated at three sets of clustered Pka and CkI sites.^10-12 Phosphorylation at the Pka sites is sufficient to drive low-threshold target gene expression, and primes Smo for further phosphorylation by CkI to promote high-threshold target gene expression.¹³ At least 18 additional Hh-responsive phosphorylation sites have been identified in Drosophila Smo, suggesting that additional kinases are involved in Smo regulation.¹¹ A growing body of evidence clearly implicates G-protein-coupled receptor kinases (GRKs) in this process in both flies and mammals. The mammalian GRKs are well-known regulators of GPCR signaling. The best-characterized function of these kinases is to dampen GPCR signaling in a process known as homologous desensitization.¹⁴ They do this by binding to and phosphorylating a wide variety of GPCRs specifically in their agonist-occupied activated conformations, creating high-affinity binding sites for β-arrestins. β-arrestin recruitment to phosphorylated receptors interferes with heterotrimeric G-protein coupling and promotes clathrin-dependent receptor internalization, effectively shutting off G-protein-dependent signaling and limiting the capacity for reactivation of the same receptor molecule. For some receptors, β-arrestin binding facilitates G-protein-independent signaling by acting as a scaffold for the assembly of signaling complexes.¹⁴

The first indication that GRKs participate in Hh signaling came from studies in mammalian cells. Smo internalization and phosphorylation, as well as activation of a Hh-responsive reporter gene, were found to require GRK2, one of seven mammalian GRK family kinases.^15,16 Further support came from the characterization of GRK2 morphant zebrafish and GRK2 knockout mouse embryos, which displayed phenotypes consistent with moderate impairment of Hh signaling and reduced Hh target gene expression.¹⁷ Several lines of evidence indicate that β-arrestins are also involved. β-arrestin is found in complex with active Smo.^16,18,19 Manipulation of β-arrestin levels has effects similar to GRK2 on Smo trafficking, and reduction of β-arrestin levels in cells and in vivo impairs Hh signaling.^15,16,18,20 GRK2 facilitates the association between Smo and β-arrestin,¹⁶ consistent with it regulating Smo in a similar manner as it does other GPCRs. This regulation is important for translocation of Smo to the primary cilium, its apparent site of action, upon pathway activation.^18,19,21

The role of GRKs in Hh signaling is conserved in lower organisms. Three separate studies have demonstrated that Gprk2, one of the two GRKs found in Drosophila, is required for efficient high-threshold target gene expression. Elimination of Gprk2 in the developing fly wing primordium (or imaginal disc), either by double-stranded RNA-mediated depletion²² or in genetic mutants,^23,24 leads to a moderate Hh loss-of-function phenotype, characterized by a reduction in the area of the central region of the adult wing (Fig. 1A-B). This phenotype results from the loss of high- but not low-threshold target gene expression (Fig. 1C-E), indicating that Gprk2 is required for high-level signaling. The requirement for Gprk2 is not unique to signaling in the developing wing, as mutant embryos derived from gprk2 homozygous mutant mothers show cuticle patterning defects typical of moderate Hh signaling impairment (Fig. 1H-I). Interestingly, gprk2 is a Hh target gene in flies, suggesting that it may participate in a feedback loop required to amplify pathway activity.²² At a cellular level, loss of Gprk2 leads to excessive accumulation of a hypophosphorylated form of Smo at the cell surface,²⁴ consistent with an impairment of receptor phosphorylation and internalization. Overexpression of Gprk2 has the opposite effect, downregulating endogenous Smo levels in cells in which Smo is strongly activated.²⁴ Perhaps surprisingly given the evidence in mammals, loss of the single β-arrestin (called Kurtz) in flies does not noticeably affect Hh signaling or Smo accumulation.^24,26 However, Kurtz can associate with and regulate Smo when overexpressed,^24,26,27 leaving open the possibility that the role of endogenous Kurtz is masked by functional redundancy.

Figure 1.gprk2 mutants show moderate Hh loss-of-function phenotypes. (A) Wild-type Drosophila wing. The region patterned by Hh signaling, bounded by the third and fourth wing veins, is shaded pink. (B) A gprk2 mutant wing, showing a reduction of the territory between veins three and four that is characteristic of moderate Hh signaling impairment. (C) Schematic diagram of anterior-posterior patterning of the wing by Hh. The wing disc is divided into two non-mixing cell compartments, anterior (A, shaded blue) and posterior (P, shaded purple), divided by a compartment boundary (dotted line). Hh is produced by P compartment cells, which do not express Ci and are thus unable to mount a transcriptional response. Instead, Hh diffuses into the Ci-expressing A compartment, inducing Hh target gene expression in a stripe of cells just A to the compartment boundary. High threshold target genes like ptc, which require high levels of Hh for activation, are expressed in a narrow stripe (green). Low threshold target genes like decapentaplegic (dpp) are expressed in a wider stripe of cells (red). (D) Wild-type wing disc, showing the typical pattern of expression of Ptc (green) and dpp (red, visualized by LacZ expression from an enhancer trap). (E) In a gprk mutant wing disc, high but not low threshold target gene expression is lost. (F) Wing disc in which an activated form of Gαs is expressed in the dorsal half (D, above dotted line) in an otherwise wild-type background. The resulting increase in cAMP production in dorsal cells has little effect on Ptc expression (compare with wild-type ventral cells, V), indicating that cAMP levels are not normally limiting for this response. (G) In a gprk2 mutant background, expression of activated Gαs in the dorsal half (D, above dotted line) specifically rescues ptc expression (compare with gprk2 mutant ventral cells, V). This indicates that cAMP levels are limiting for Hh responses in the absence of Gprk2. (H) Ventral cuticle of a wild-type embryo, showing the typical segmentally-repeated pattern of denticle belts. (I) In a gprk2 maternal and zygotic mutant embryo, the denticle belts are frequently fused (yellow arrowheads). This phenotype is similar to that observed with hypomorphic alleles of smo,²⁵ and indicates that Gprk2 is also required for normal Hh signaling in the embryonic epidermis.

How Do GRKs Affect Transduction of the Hh Signal?

The evidence from several systems linking GRKs and β-arrestins with Smo accumulation and trafficking strongly suggests that GRKs regulate Smo in a manner typical of other GPCRs. This implies that GRKs should function to desensitize, or shut off, Smo. With other GPCRs, this is manifested in several ways, including increased cell surface receptor levels and prolonged or excessive signaling responses in the absence of the regulating GRK.²⁸ In the case of Smo, cell surface protein levels are substantially increased in gprk2 mutant Hh-responding cells, which is usually a sign of high-level signaling.^29,30 Yet phenotypically, the loss of GRKs would appear to have the opposite effect, rendering Smo less active. Several explanations for this contradiction have been proposed. There is strong evidence that direct phosphorylation of Smo by GRKs contributes to the conformational shift associated with Smo activation in both flies and mammals. Thus the loss of GRKs can impair Hh responses by preventing Smo from reaching its most active state.^19,23 Gprk2 can also facilitate Smo dimerization in a manner independent of its kinase activity.²³ It has also been suggested that GRK-dependent β-arrestin-binding and/or subsequent internalization of Smo may be required for efficient signaling. This could reflect the need for receptor recycling or ciliary translocation (in mammals) for high-level signaling, or by analogy with other GPCRs, the participation of β-arrestins in assembly of a Smo-associated signaling complex (for example, with Fu and Cos2 in flies).^14,31-33

Our recent analysis of Gprk2 in flies³⁴ suggests that the role of GRKs in Hh signaling is more complex than envisioned. As summarized below, we found evidence that they act as both positive and negative regulators of signaling. Furthermore, they appear to have indirect effects on pathway activity in addition to the more direct effects on Smo noted above.

Gprk2 Suppresses Canonical Pathway Activation but Promotes Hh Target Gene Expression

We initially set out to assess how the loss of Gprk2 affects Smo activity. Rather than using Hh target gene expression as a readout, we focused instead on more proximal signaling events. In order for high-threshold target genes to be expressed, Ci¹⁵⁵ must be stabilized and its retention in the cytoplasm by SuFu and Cos2 must be neutralized to enable it to enter the nucleus.^35-37 Despite the loss of high-threshold target gene expression in gprk2 mutant wing discs, Smo-dependent stabilization and nuclear accumulation of Ci¹⁵⁵ in Hh-responding cells appeared normal, indicating that signaling from Smo to Ci was intact. This signal is transduced by Fu, Cos2, and SuFu, which all become phosphorylated in response to Smo activation.^38,39 Unexpectedly, Smo-dependent phosphorylation of all three proteins was increased in the absence of Gprk2. This indicates that total Smo activity is higher rather than lower without Gprk2, and is the first sign that Gprk2 acts negatively in the pathway to limit Smo activity, as predicted if it desensitizes Smo in the classical sense (Fig. 2A). Furthermore, it argues against the possibility that Gprk2/β-arrestin-dependent Smo trafficking or signaling complex formation is strictly required for activation of canonical signaling.

Figure 2. Multiple effects of Gprk2 on the Hh signaling pathway. (A) Gprk2 “desensitizes” Smo like a typical GPCR, phosphorylating the active form of Smo (Smo**) to promote its internalization and degradation. When Gprk2 is missing active Smo accumulates, leading to increased canonical pathway activation. (B) Gprk2 directly promotes Smo activation by phosphorylating Pka- and CkI-phosphorylated Smo, increasing its conformational shift and enhancing its activity.²³ Loss of Gprk2 thus prevents the full activation of Smo required for high-threshold target gene expression. (C) Gprk2 indirectly promotes Smo activation via Pka. Gprk2 activity helps to maintain high cellular cAMP levels, most likely through its role in limiting GPCR signaling. Loss of Gprk2 leads to a decrease in cAMP production and reduced Pka activity. This in turn decreases Pka-dependent Smo phosphorylation and activation, leading to a loss of high threshold target gene expression.

If GRKs are shutting off Smo signaling, why is high threshold target gene expression lost in their absence? One answer, mentioned above, is that they play dual roles when it comes to Smo. In addition to desensitizing Smo, direct GRK-mediated phosphorylation is required for Smo to reach its most active state (with activity assessed by target gene expression) (Fig. 2B).^19,23 Taking this into consideration, a likely explanation for the increase in canonical pathway activity observed in the absence of Gprk2 is that it results from the ectopic accumulation of a sub-maximally activated form of Smo. Although total pathway activity is higher because of the increase in cell-surface Smo levels and/or the duration of its activity in gprk2 mutant cells, the reduced per-molecule activity of Smo may affect the dynamics of activation in a way that prevents high-threshold target gene expression. In this interpretation, the indicators of pathway activity that increase in the absence of Gprk2 may effectively be measures of low-threshold signaling.

Connecting GRKs, Second Messenger Regulation and Hh Signaling

Direct GRK-mediated activation of Smo provides a straightforward explanation for the effects of these kinases on Hh target gene expression. However, we found that the situation is more complicated, with the loss of Gprk2 also indirectly impacting on the Hh pathway through altered second messenger signaling. This is not entirely unexpected. In both flies and mammals, a relatively small number of GRKs (two and seven, respectively) is tasked with regulating a huge complement of receptors—more than one hundred in flies and one thousand in mammals,^7,40 the majority of which are thought likely to be GRK substrates.⁴¹ In mammals, a single GRK can regulate many receptors, and one receptor can be regulated by several kinases.⁴¹ The situation is presumably similar in Drosophila. Indeed, we have observed some functional redundancy between Gprk1 and Gprk2 in flies, at least during development, suggesting that they regulate some common targets.²⁴ Given this level of interconnectedness, the removal of a single GRK has the potential to simultaneously affect the activity of multiple GPCRs, and by extension second messenger dynamics, within a single cell.

Does this actually occur? Although the effects of GRK elimination are to some extent mitigated by functional redundancy, selective defects in receptor desensitization and second messenger production are observed in the absence of individual GRKs, both in cell culture models and in mouse knockouts.²⁸ For example, in addition to impairing Hh signaling, depletion of GRK2 in mammalian cells also elevates production of the second messengers cyclic AMP (cAMP) and diacylglycerol in response to activation of the V2 Vasopressin and Angiotensin II Type 1_A GPCRs, respectively,^42,43 whereas some other GPCR responses are normal.⁴⁴ Similarly, elimination of GRK6 causes exaggerated second messenger responses upon activation of a subset of GPCRs,^44-46 without affecting others.⁴² Because second messenger-dependent kinases also feed back to regulate diverse GPCR responses through heterologous desensitization,⁴⁷ there is also the potential for misregulation of GPCRs controlled by a given GRK to indirectly affect the activity of GPCRs not regulated by that GRK. Thus, individual GRKs can and do control a complex set of receptor responses, with consequent effects on second messenger production.

This is particularly relevant in the case of Hh signaling, where the cAMP-regulated kinase Pka plays an integral role in the pathway. Pka actually has dual functions, acting as both a silencer and activator of the pathway. On one hand, Pka phosphorylation promotes the proteolytic cleavage of Ci and one of its mammalian orthologs, Gli3, to their shorter transcriptional repressor forms,^48,49 and also directly inhibits the transcriptional activator functions of full-length Ci and Gli proteins,⁵⁰ both of which serve to keep target genes off in the absence of Hh. Consequently, strong inhibition or elimination of Pka leads to constitutive Ci/Gli activation and ectopic target gene activation in both Drosophila and vertebrates.^51-54 On the other hand, Pka activation can induce ectopic target gene expression and gain-of-function phenotypes in flies and vertebrates,^10,55-57 whereas Pka inhibition can block Hh-induced activation of target genes,^10,57 indicating that Pka also has a positive function. Smo is one identified target of positive regulation by Pka in flies, through its C-terminus.^10-12 Comparable Pka sites are not present in vertebrate Smo proteins, and there is no evidence that they are directly regulated by Pka, leaving the identity of the positive target in vertebrates unknown. The existence of another target is further supported by evidence from Drosophila embryos, where Pka appears to stimulate Hh target gene expression not only through Smo, but through another unknown target acting in parallel to Smo and Ci regulation as well.⁵⁸

The fact that Pka regulates critical pathway components with such opposite outcomes would suggest that its activity must be tightly controlled and perhaps itself regulated by Hh in order for the pathway to function properly. However, the evidence in this regard is equivocal. Smo has long been suspected to function as a GPCR, and there is evidence that in some systems it can couple to Gα_i, a heterotrimeric G-protein α-subunit that inhibits adenylate cyclase activity to suppress cAMP production⁵⁹ (further reviewed in Ref.33). In other systems this seems not to be the case. Furthermore, Hh responsiveness in Pka mutant cells can be restored by reintroduction of a cAMP-insensitive form of the kinase, which has been interpreted as meaning that Pka activity need not be regulated by Hh.^53,54 Whether or not it is directly regulated by Hh, it is clear that Pka activity needs to fall within a limited range to allow optimal functioning of the pathway. In Drosophila, no Pka activity leads to constitutive Ci activation,^53,54 too little limits Smo activity,¹⁰ a moderate increase enhances Hh responsiveness,^10,55 and too much inhibits target gene activation.^54,58

cAMP Misregulation Impairs Hh Target Gene Expression in gprk2 Mutants

Like its mammalian counterparts, Gprk2 has been implicated in second messenger regulation in Drosophila. In analyzing animals homozygous for a mutant allele of gprk2 that disrupts expression in the ovary, Lanutti et al.⁶⁰ found that the loss of Gprk2 was associated with a 70% drop in ovarian cAMP levels. By genetically manipulating enzymes involved in cAMP metabolism to increase or further decrease cAMP levels, they also showed that the reduction in cAMP levels was a significant contributor to the germline defects and reduced fertility of the mutant flies. This implies that Gprk2 affects egg production indirectly, through its effects on second messenger signaling. Using null alleles of gprk2, we found that cAMP levels in gprk2 homozygous mutant larvae were similarly reduced compared with wild-type animals. This seems to be a more general effect, as it could be reproduced by Gprk2 depletion in cultured cells. We therefore hypothesized that the low levels of cAMP/Pka activity may be limiting for efficient functioning of the Hh pathway in gprk2 mutants.

To test this hypothesis, we examined the effects of manipulating cAMP levels and Pka activity by expressing transgenic regulators of each on Hh target gene expression in the mutants. Lowering the already low levels of cAMP or Pka activity in the gprk2 mutants further impaired Hh-dependent high-threshold target gene expression in gprk2 mutant flies, but not wild-type flies. Conversely, increasing cAMP levels or Pka activity efficiently and specifically rescued Hh-dependent expression of high-threshold target genes to apparently normal levels in the absence of Gprk2 (Fig. 1F-G). Both results indicate that the low levels of cAMP in gprk2 mutants are limiting for Hh target gene expression. Thus, as with its role in egg production, Gprk2 has indirect effects on the Hh pathway through second messenger signaling. Although Smo can itself signal through Gα_i to lower cAMP levels, double knockdown experiments suggest that the cAMP defect in gprk2 mutants is independent of Smo activity. The changes in cAMP regulation could result from a broader misregulation of GPCR signaling in the gprk2 mutants.

Further support for an indirect effect comes from an analysis of Smo itself. Smo is less highly phosphorylated in the absence of Gprk2.²⁴ We found that this Gprk2-dependent effect on Smo phosphorylation persists when the direct Gprk2 phosphoacceptor sites in Smo²³ are eliminated. While there may be additional unmapped Gprk2 phosphorylation sites in Smo, it could also be that another kinase phosphorylates Smo in a Gprk2-dependent manner. Two pieces of evidence suggest that this latter explanation is at least partly correct and that the kinase involved is Pka, as our model would predict. First, Pka activation largely restores Smo phosphorylation in the absence of Gprk2. Second, a Pka phosphomimetic form of Smo rescues Hh target gene expression in gprk2 mutants. Interestingly, the phosphomimetic Smo is less effective than cAMP in rescuing target gene expression, suggesting that Pka may act through a target in addition to Smo in wing discs, as it does in embryos.⁵⁸ Taken together, our results suggest that Gprk2 helps maintain cellular cAMP levels high enough to permit full phosphorylation and activation of Smo, and possibly another target, by Pka (Fig. 2C). How this indirect function of Gprk2 fits with its direct effects on Smo phosphorylation and dimerization²³ is not clear. While the pathway can clearly function in the absence of Gprk2 if Pka activity is high enough, both mechanisms may be important for efficient Hh signaling under normal circumstances.

Implications for cAMP and PKA Function in Hh Signaling

Our study reveals an interesting connection between endogenous cAMP levels and pathway function. Despite the fact that Pka phosphorylation activates Smo, under normal circumstances moderate activation of Pka has only a small effect in extending the range of high threshold target gene expression in the wing disc.¹⁰ This is true even when controlling for the negative effect of Pka on Ci.⁵⁸ This is consistent with our observation that manipulating cAMP production has little effect on high-threshold target gene activation in wild-type discs³⁴ (though lowering cAMP levels can de-repress low-threshold targets⁵⁹). Thus Pka activity, and by extension cAMP levels, are not major limiting factors for Hh pathway activation in the wing disc. In contrast, Pka activity is limiting for pathway activation in the embryonic epidermis, as expression of constitutively active Pka leads to a dramatic expansion in Hh target gene expression due to dual effects on Smo and at least one other target.^55,58 The functioning of the pathway in the embryonic epidermis in some ways resembles the situation in gprk2 mutant wing discs in this regard. One explanation for the observed difference in pathway function between discs and embryos could thus be that Gprk2 normally participates in Hh signaling only in discs. However, the fact that gprk2 maternal/zygotic mutant embryos show patterning defects typical of reduced Hh signaling (Fig. 1H and I) suggests that this is unlikely. An alternate explanation could be that baseline cAMP levels are higher in wild-type wing discs than they are in embryonic epidermal cells or gprk2 mutant discs, so that cAMP levels limit signaling only in the latter. Differences in the complement of GPCRs that are expressed and active in a given tissue could contribute to such intrinsic differences in cAMP levels. While this is speculative, there is evidence that regulation of cAMP levels by different GPCRs can influence Hh pathway-dependent responses in the vertebrate nervous system.^61,62 If cellular cAMP levels do set thresholds for Hh pathway activation, then cell- or tissue-type-dependent differences in cAMP levels could provide a simple explanation for the often-conflicting results concerning the importance of heterotrimeric G-proteins in functioning of the pathway.

Abbreviations:
cAMP	=	3′-5′-cyclic adenosine monophosphate
Ci	=	Cubitus interruptus
CkI	=	Casein kinase I
Cos2	=	Constal-2
Fu	=	Fused
GPCR	=	G-protein-coupled receptor
Gprk2	=	G-protein-coupled receptor kinase 2
GRK	=	G-protein-coupled receptor kinase
Hh	=	Hedgehog
Pka	=	Protein kinase A
Smo	=	Smoothened
SuFu	=	Suppressor of Fused

Acknowledgments

This work was supported by funding from the Canadian Institutes of Health Research.