Extensive crosstalk of G protein-coupled receptors with the Hedgehog signalling pathway

The Hedgehog (Hh) signalling pathway is crucial for patterning of embryonic tissues (Briscoe and Thérond, 2013; Jiang and Hui, 2008). Hh ligands such as Sonic hedgehog (Shh) in vertebrates and Hh in Drosophila are morphogens, diffusing through tissues from localized sources and eliciting differential transcriptional responses in distant cells, dependent on ligand concentration and time of exposure (Briscoe and Small, 2015; Jiang and Hui, 2008). Some high-threshold Hh target genes, such as Nkx2.2 in the mouse neural tube or patched (ptc) in the Drosophila wing imaginal disc, are expressed only close to the source of Hh, where concentration is high and cells have been exposed for the longest time. In contrast, low-threshold targets, such as decapentaplegic (dpp) in flies, are expressed where ligand levels are lower (Placzek and Briscoe, 2018; Strigini and Cohen, 1997).

Hh concentration gradients are translated into activity of the cubitus interruptus (Ci) and Glioma-associated oncogene (GLI) family transcription factors in vertebrates and flies, respectively, through a conserved signalling pathway (Briscoe and Thérond, 2013; Hui and Angers, 2011). In the absence of Hh, activity of the atypical G protein-coupled receptor (GPCR) Smoothened (Smo) is suppressed by the Hh receptor Ptc (Denef et al., 2000). In this state, Ci/GLI proteins undergo partial proteolysis to truncated proteins (CiR/GLIR) that repress transcription of some target genes (Aza-Blanc et al., 1997; Méthot and Basler, 1999; Niewiadomski et al., 2014; Price and Kalderon, 1999; Wang et al., 2000). Hh binding to Ptc relieves its inhibition of Smo (Denef et al., 2000; Ingham et al., 2000), which undergoes extensive phosphorylation and adopts an active conformation (Apionishev et al., 2005; Chen et al., 2011; Jia et al., 2004; Zhang et al., 2004; Zhao et al., 2007). Smo signals through a conserved pathway to block CiR/GLIR formation and alleviate inhibition of full-length Ci/GLI proteins at multiple levels, enabling them to transcriptionally activate their targets (Briscoe and Thérond, 2013; Hui and Angers, 2011). The levels of Drosophila Smo phosphorylation and activity are correlated with Hh levels (Fan et al., 2012; Su et al., 2011), suggesting that Hh ligand concentration gradients are converted to intracellular signalling gradients that can explain, in part, the differential activation of low- and high-threshold target genes.

Recently, a unique mechanism for shaping Hh responses involving the second messenger 3′,5′-cyclic adenosine monophosphate (cAMP) was identified. In both flies and mammals, the level of cAMP in cells helps determine their sensitivity to Hh ligands (Praktiknjo et al., 2018; Pusapati et al., 2018). cAMP is produced from ATP by adenylyl cyclases whose activity is controlled by GPCRs signalling through heterotrimeric G proteins to stimulate (Gαs) or inhibit (Gαi, Gαo) cAMP synthesis (Wettschureck and Offermanns, 2005). In mammalian cells, activity of the GPCR Gpr161 stimulates cAMP production through Gαs (Mukhopadhyay et al., 2013). Mutation of Gpr161 decreases the amount of Shh needed for transcriptional activation of target genes and expands the distance over which medium- and high-threshold target genes are activated by Shh in the mouse neural tube (Mukhopadhyay et al., 2013; Pusapati et al., 2018), suggesting that cAMP desensitizes mammalian cells to Shh. This is consistent with reports that Gαs and the Gαs-coupled GPCRs PAC1 and Gpr17 suppress Shh target gene expression in mammalian cells (Cohen et al., 2010; He et al., 2014; Iglesias-Bartolome et al., 2015; Niewiadomski et al., 2013; Rao et al., 2016; Yatsuzuka et al., 2019), whereas Gαi and the Gαi-coupled GPCRs Cxcr4 and Gpr175 enhance their expression (Klein et al., 2001; Riobo et al., 2006; Singh et al., 2015). A similar mechanism appears to exist in Drosophila, where depletion of Gαi or Gαs increased or decreased, respectively, the range over which expression of some Hh target genes is activated in the developing wing (Praktiknjo et al., 2018). The effects of Gα depletion suggest that cAMP increases Hh sensitivity in Drosophila rather than decreasing it, consistent with effects observed when directly activating Gαi or Gαs in discs (Cheng et al., 2012). There is also evidence that cAMP can inhibit Hh target gene expression in flies (Ogden et al., 2008).

These effects of cAMP are likely due to changes in the activity of Protein kinase A (PKA), the main cellular target of cAMP. In flies and mammals, PKA phosphorylates and inhibits Ci/GLI proteins (Niewiadomski et al., 2014; Price and Kalderon, 1999; Wang et al., 1999) by a conserved mechanism (Marks and Kalderon, 2011), likely explaining the inhibitory effect of cAMP on Hh target gene expression in flies (Ogden et al., 2008). PKA also plays a unique positive role in the Drosophila Hh pathway, where it is the principal kinase responsible for phosphorylating Smo to activate signalling (Apionishev et al., 2005; Jia et al., 2004; Zhang et al., 2004). In fact, moderate overexpression of the PKA catalytic subunit hyperactivates Drosophila Smo and is sufficient to expand the domain over which Hh target genes are expressed (Jia et al., 2004).

Regulation of cAMP production in Hh-stimulated cells may be direct. Mammalian Smo can couple to Gαi (Qi et al., 2019; Riobo et al., 2006; Shen et al., 2013), and Drosophila Smo can activate Gα_i and Gα_s (Ogden et al., 2008; Praktiknjo et al., 2018). However, it is clear that other GPCRs are involved. A few have been identified in mammals, but it is not known whether these represent the exception or whether many GPCRs among the hundreds encoded in mammalian genomes have this capacity.

Drosophila represents a simpler system for studying GPCRs, with just 116 GPCRs encoded in the genome. These receptors have traditionally been classified into four families – the Rhodopsin-like Class A, Secretin-like Class B, Glutamate Class C and Atypical or Frizzled-like Class F (Table S1) – although more recent phylogenetic analyses reveal a more nuanced organization (Hanlon and Andrew, 2015). We assessed the extent of GPCR crosstalk with the Hh pathway by screening for GPCRs expressed in the Drosophila wing imaginal disc that influence Hh responses. We found that RNA interference (RNAi)-mediated depletion of more than 40% of GPCRs tested phenocopied either Gαs or Gαi depletion, diminishing or enhancing sensitivity to Hh, respectively. More in-depth analysis of one of these GPCRs, Mthl5, suggests that it attenuates Hh signalling by lowering the level of cAMP in cells. These results indicate that GPCR crosstalk with the Hh pathway in mammals is likely to be extensive rather than limited to just a few receptors.

Hh signalling controls the growth of the Drosophila wing along the anterior-posterior (A-P) axis and directly patterns the central region of the wing between longitudinal veins L3 and L4 (highlighted in Fig. 1A). We have previously shown that this depends on the activity of Gαs and Gαi, as depletion of either leads to a spectrum of defects associated with changes in Hh target gene expression (Praktiknjo et al., 2018). Wings in which Gαs was depleted by expressing a short RNA hairpin (using the nubbin-GAL4 driver) were smaller and not properly inflated (Fig. 1B) (Praktiknjo et al., 2018), a process known to require Gαs activation by the GPCR Rickets (Rk) (Baker and Truman, 2002; Kimura et al., 2004). Gαi-depleted wings showed an ectopic vein phenotype similar to that caused by dpp overexpression in the developing wing pouch (De Celis, 1998) and an increase in the area of the region bounded by wing veins L3 and L4 relative to total wing area (Fig. 1C, quantified in Fig. 1S) (Praktiknjo et al., 2018). This is a phenotype specifically associated with increased Hh signalling (Mullor et al., 1997).

These phenotypes suggested that some GPCRs are actively signalling through Gαs and Gαi in wing discs, and that their activity affects cell responses to Hh. We reasoned that reducing the levels of such GPCRs would produce a similar spectrum of phenotypes, depending on the Gα subunit to which they couple. To investigate this, we first carried out RNAseq analysis to identify which GPCRs are expressed in late third instar wing discs. We detected substantial expression of roughly one-quarter of the 116 GPCR-encoding genes, and at least some expression of more than half (Fig. 1D). To simplify our analysis, we excluded smo and the various frizzled homologues belonging to the atypical class F of the GPCR family (Hanlon and Andrew, 2015). We focused on 22 of the most highly expressed remaining GPCRs, most from the more classical Class A/B/C receptors (Table 1).

We expressed dsRNA transgenes targeting each of these GPCRs throughout the developing wing pouch and examined the effects on the adult wing. We observed a range of phenotypes upon depletion of 15 of the 22 GPCRs with at least one RNAi strain (Table S2). At the most extreme, depletion of the class A receptors Trapped in endoderm 1 (Tre1) or CCK-like receptor at 17D3 (CCKLR-17D3), or of the class C receptor CG32447 was lethal, with rare wingless escapers in the case of CCKLR-17D3 and CG32447.

Seven GPCRs produced depletion phenotypes more similar to Gαs. These included three Secretin-like class B adhesion GPCRs [the Methusaleh (Mth)-like family receptors Mthl6, Mthl8 and Mthl9], which caused a severe wing inflation defect (Fig. 1E-G), and four others (Rk, Mthl4, Stan and CG15744) that gave rise to uneven wings consistent with weaker inflation defects (Fig. 1H-L) and to a decrease in wing size (Fig. 1M). As mentioned above, Rk is known to stimulate cAMP production through Gαs.

On the other hand, depletion of four GPCRs (Mth, Mthl5, GABA-B-R2 and Cirl) caused Hh gain-of-function phenotypes similar to those observed upon Gαi depletion (Fig. 1N-R). Most notably, depletion of Mth, Mthl5 and GABA-B-R2 caused the appearance of extensive ectopic veins, and the ratio of L3-L4 to total wing area was significantly increased in all four cases (Fig. 1S). Interestingly, three of these GPCRs (or their orthologues) – GABA-B-R2, Mthl5 and Cirl – have been shown to couple to inhibitory Gα proteins to lower cellular cAMP levels (Franek et al., 1999; Patel et al., 2016; Scholz et al., 2017).

To confirm that these wing phenotypes were due to altered Hh signalling, we analysed Hh target gene expression earlier in imaginal discs. Hh is expressed by cells of the posterior compartment and diffuses into the anterior compartment to form a concentration gradient that stabilizes full-length Ci (Ci¹⁵⁵) in a central stripe of responding anterior cells, where it activates target gene expression (Fig. 2A). The low threshold target gene dpp is expressed in a broad stripe of cells, and the higher-threshold target ptc is expressed over a narrower domain. Quantification of fluorescence intensities along the A-P axis indicated that the domains of Ci stabilization and dpp and ptc expression are of roughly equal width in dorsal and ventral compartments (Fig. 2B). As we previously showed (Praktiknjo et al., 2018), Gαs depletion in the dorsal compartment of the wing disc (using the ap-GAL4 driver) dampened responsiveness to low levels of Hh. This was characterized by a narrowing of the range over which dpp was expressed in the dorsal compartment compared with the wild-type ventral compartment, with little effect on Ci stabilization or ptc expression (Fig. 2C,D).

To look at the effects of GPCR depletion, we started with the seven GPCRs whose depletion caused inflation defects. Of these, six had effects on dpp expression that phenocopied Gαs depletion. Depletion of Mthl4 (Fig. 2E,F), Mthl8 (Fig. 2G,H), Mthl9 (Fig. 2I,J), Rk (Fig. 2K,L), CG15744 (Fig. 2M,N) or Stan (Fig. 2O,P) decreased the range of dpp expression to varying degrees. The sole exception was Mthl6, the depletion of which had no effect (data not shown).

We also assessed the effects of depleting selected GPCRs on expression of the high-threshold Hh target gene engrailed (en). en is expressed throughout the posterior compartment in a Hh-independent manner, but in late third instar is upregulated in the first few rows of anterior boundary cells in response to high levels of Hh (Fig. S1), where it is thought to lower dpp expression (Raftery et al., 1991; Strigini and Cohen, 1997). En protein levels were roughly symmetrical between the dorsal and ventral compartments (Fig. S1A,B). Gαs-depleted Hh-responding cells activated en expression, although protein levels were somewhat (∼20%) lower than normal (Fig. S1C,D). Rk depletion had a similar effect (Fig. S1E,F). However, in both cases it was difficult to attribute this to a specific defect in Hh signalling, as Hh-independent posterior compartment En levels were, unexpectedly, also lower in dsRNA-expressing cells.

Depleting Gαi had the opposite effect of Gαs depletion. As we have shown previously (Praktiknjo et al., 2018), Gαi-depleted discs showed a Hh gain-of-function phenotype characterized by an increase in the range over which Ci stabilization and dpp (and to a lesser extent, ptc) expression were detected (Fig. 3A,B). Among the four GPCRs whose depletion caused Hh gain-of-function-like phenotypes, all [Mthl5 (Fig. 3C,D), GABA-B-R2 (Fig. 3E,F), Mth (Fig. 3G,H) and Cirl (Fig. 3I,J)] significantly increased the level and/or range of dpp expression when depleted, phenocopying Gαi depletion. The maximum level of ptc expression or the width of its expression domain were also significantly increased, and all but Cirl (which had the weakest adult phenotype) significantly increased the range and level of Ci stabilization.

Depletion of Gαi, Mthl5 or Cirl all had similar effects on En. Anterior en was still activated in each case, but protein levels were lower (Fig. S1G-L). As with Gαs depletion, Hh-independent levels of En in the posterior compartment were also lower, confounding interpretation of the results. We noted that DAPI staining intensity was also noticeably lower in the dorsal compartment, consistent with reduced cellularity due to an elevated level of apoptosis in the depleted cells (see below and data not shown). When corrected for this decrease, En levels in these backgrounds were similar to those observed in Gαs- and Rk-depleted discs, despite the opposite effects of Gαs and Rk on dpp expression. This suggests that changes in en expression alone are unlikely to explain the expansion of dpp expression in Gαi/Mthl5/Cirl-depleted discs.

Taken together, our screen results indicate that 45% (10 of 22) of the expressed GPCRs that we tested are capable of crosstalking with the Hh pathway as part of their normal function. The similarity between GPCR and G-protein depletion phenotypes suggests that they do so by signalling through either Gαi or Gαs, as some GPCRs do in mammals.

We chose one little-characterized GPCR candidate for further testing. Mthl5 is an orphan GPCR belonging to the Mth-like GPCR family. This evolutionarily ancient family of proteins shows close resemblance to the Secretin family GPCRs, descended from the larger family of adhesion GPCRs (Langenhan et al., 2013; Patel et al., 2012). Mthl5 does not have a clear orthologue in mammals.

mthl5 is expressed at high levels in the developing Drosophila embryonic heart where it is required for normal morphogenesis of the aorta (Patel et al., 2016). To assess where mthl5 is expressed in discs, we generated an antiserum against the large extracellular N-terminal domain of the protein. Immunostaining with this antiserum revealed a ubiquitous distribution of the protein in wing discs (Fig. 4A), with Mthl5 protein accumulating preferentially in the apical region of cells (Fig. 4B). Mthl5 immunostaining was much reduced in mthl5-depleted disc cells (Fig. S2), confirming that the antiserum is specific and that the mthl5 dsRNA transgene is functional.

The mthl5 gene is situated on chromosome 3R in the intron of another gene (CG31368), transcribed in the opposite orientation (Fig. 4C). There are two isoforms (mthl5-RA and mthl5-RB) that differ by just three nucleotides (and one amino acid) due to alternate use of neighbouring splice sites in exon 2. The Gene Disruption Project generated a fly strain bearing a Minos element inserted in the coding region of mthl5 (Mi{ET1}CG31368^MB03076, which we re-designate mthl5^MB03076). The insertion in this strain is in exon 5, which is common to both mthl5 isoforms (Fig. 4C), a finding we confirmed by sequencing of genomic PCR products from the insertion flanks. In silico predictions showed that the mthl5^MB03076 mutation introduces a stop codon within predicted transmembrane domain 5 of the 7-transmembrane core of the protein, potentially leaving the long extracellular N terminus intact but deleting the short cytoplasmic C-terminal tail.

Less than one-third of mthl5^MB03076 homozygous mutants (55/176 expected) survived to adulthood. The majority died at third instar or pupal stages. Survival was also low (47/101 expected) for trans-heterozygotes of mthl5^MB03076 and Df(3R)BSC514, a chromosomal deficiency uncovering the mthl5 gene. Several defects were apparent in mthl5 mutant wing discs. Both mthl5^MB03076 and mthl5^MB03076/Df(3R)BSC514 mutant discs were on average about 30-50% larger than wild-type controls (Fig. 4D-G). The overgrowth was most noticeable in the pleura and hinge regions surrounding the pouch. In fact, the A-P length of the wing pouch in mthl5^MB03076 mutant discs was similar to wild type (Fig. 4H). Other morphological defects were frequently observed, including a cleft along the A-P boundary of the wing pouch in many mthl5^MB03076 mutant discs (arrow in Fig. 4E). Pyknotic nuclei were present on the basal side of the wing pouch, and activated Caspase 3 immunostaining indicated that apoptosis levels were high (Fig. 4I,J). We conclude that Mthl5 plays an important role in regulating growth and cell survival during wing disc development.

The pleural overgrowth and A-P boundary clefting in mthl5 mutant discs are reminiscent of phenotypes caused by increased Hh activity (Umemori et al., 2007). To determine if Hh signalling is upregulated in the mutants, as it is when Mthl5 is depleted, we analysed the adult wing phenotypes of escaper flies. Compared with control (w¹¹¹⁸) or mthl5^MB03076 heterozygotes (Fig. 5A,B), the wings of mthl5^MB03076 homozygous flies were somewhat variable in appearance, ranging from roughly normal (Fig. 5C) to distorted with a proportionately larger anterior compartment (Fig. 5D). Total wing area was 20% smaller than controls (Fig. S3A), consistent with the elevated levels of apoptosis in mutant discs. The mean L3-L4:total wing area ratio was not significantly different from wild type (although 15 of 29 wings had a ratio higher than the highest value observed in wild-type wings), and was slightly increased relative to mthl5^MB03076 heterozygotes (Fig. 5H). However, the size of the anterior compartment relative to the posterior compartment (A:P ratio) was significantly greater in mthl5^MB03076 homozygotes (Fig. 5I). This resembles the low-level Hh gain-of-function phenotype caused by moderate overexpression of Smo throughout the wing pouch, which induces low level dpp expression throughout the anterior compartment (Maier et al., 2014) and anterior compartment overgrowth (Fig. 5E,I), but does not increase pathway activity sufficiently to broaden ptc expression (Maier et al., 2014) or increase L3-L4 spacing (Fig. 5H). The phenotype of mthl5^MB03076/Df(3R)BSC514 wings was more severe. Wings were smaller than controls (Fig. S3A). Although, again, somewhat variable in appearance (Fig. 5F,G), the L3:L4 area was consistently and significantly larger, and the relative overgrowth of the anterior compartment was more extreme (Fig. 5H,I). The stronger phenotype in mthl5^MB03076/Df(3R)BSC514 animals suggests that mthl5^MB03076 is not a null allele.

To confirm that these changes are due to altered Hh signalling, we examined Hh target gene expression in mthl5 mutants. As in adult wings, the mthl5^MB03076 homozygous disc phenotype was variable. Nearly half of mthl5^MB03076 discs (14 of 32) showed clear signs of an expanded Hh pathway response. This was most apparent for dpp, which was expressed over a significantly wider anterior domain than in wild-type discs (Fig. 5J-L). There was also a statistically significant expansion of the anterior-most limit of Ci stabilization (Fig. 5M). These phenotypes were more extreme in mthl5^MB03076/Df(3R)BSC514 discs. Ci¹⁵⁵ often filled the anterior pouch (five out of 11 discs) (Fig. 5N), and in most cases the dpp and Ptc expression domains were substantially expanded (nine out of 11 and 11 of 13 discs examined, respectively) (Fig. 5N-P). Supporting the conclusion that target gene expansion in the mutants is due to loss of Mthl5 function, expression of a weaker Mthl5-GFP transgene (Mthl5-GFP^weak) that had little effect in a wild-type background significantly reduced the width of the dpp expression domain in mthl5 mutants (Fig. S3B-F).

dpp can be upregulated in apoptotic disc cells that are prevented from dying by inhibition of caspases, with dramatic effect on tissue growth (Perez-Garijo et al., 2004; Ryoo et al., 2004). However, without caspase inhibition, Dpp does not appear to play a substantial role in increasing proliferation and tissue growth in these conditions (Perez-Garijo et al., 2009). Nonetheless, we examined whether ectopic dpp expression in mthl5 mutant discs could be linked to high levels of apoptosis. dpp-LacZ expression was mainly restricted to anterior cells within the Hh-responding region, as normal, in mthl5^MB03076 discs (Fig. 5Q) where apoptosis levels are high, and in mthl5-depleted discs where apoptosis was much lower (Fig. S4A,B). In both cases, expression was detected in some basal pyknotic nuclei, the vast majority of which were located below the endogenous dpp expression domain. This suggests that they arose from the death of normal dpp-expressing cells rather than from induction of dpp expression in dying cells. Next, we tested whether blocking apoptosis by dorsal compartment expression of P35 could prevent the expansion of dpp expression in mthl5^MB03076/Df(3R)BSC514 mutants. Most of these discs were overgrown and disorganized, as seen in other contexts when widespread apoptosis is blocked (Martin et al., 2009). In the few discs that were interpretable, we observed Ci stabilization in a broad stripe of anterior cells up to three times the normal width (Fig. S4C). dpp expression was similarly expanded and mostly restricted to Ci¹⁵⁵-positive anterior compartment cells, indicative of ongoing and widespread Hh pathway activity. Together these results suggest that Hh signalling and not apoptosis is the main driver of expanded dpp expression when GPCRs like Mthl5 are depleted or mutated.

Growth of lateral regions of the wing disc such as the pleura is controlled in part by Dpp (Barrio and Milán, 2017). We were therefore interested to test whether the disc overgrowth phenotype in mthl5 mutants was due to increased Dpp signalling resulting from the anterior expansion of its expression domain. To examine dpp activity, we analysed the expression pattern of one of its high-threshold target genes, spalt-major (salm, visualized with a salm-LacZ enhancer trap) (de Celis et al., 1996). salm is normally expressed in anterior cells in the dpp expression domain and in a roughly symmetrical pattern extending towards the anterior and posterior margins of the wing pouch (Fig. 6A). In mthl5^MB03076 discs, the anterior compartment domain of spalt expression was significantly wider, whereas the posterior domain was unchanged (Fig. 6B,C). This is consistent with an anterior expansion of Dpp signalling activity. To test whether Dpp was driving disc overgrowth, we tried limiting Dpp pathway activity by lowering gene dose of thick veins (tkv), which encodes the Dpp receptor, using the tkv¹² null allele. Unexpectedly, tkv¹² heterozygous wing discs were larger than control w¹¹¹⁸ discs (Fig. 6D,E,H). However, introduction of one copy of tkv¹² significantly reduced the disc overgrowth phenotype of mthl5^MB03076 homozygotes, and improved wing disc morphology (Fig. 6F-H). These results suggest that anterior expansion of Hh pathway activity in the absence of Mthl5 indirectly drives disc overgrowth through a corresponding anterior expansion of Dpp signalling.

The simplest explanation for the similarity between mthl5 mutant and Gαi depletion phenotypes is that Mthl5 signals through Gαi to lower cAMP levels, decreasing sensitivity to Hh. Consistent with this, overexpression of a stronger-expressing Mthl5-GFP transgene caused a narrowing of the domain of dpp expression (Fig. 7A,B). We turned to Drosophila S2R+ cells to test this further. Transfection with an expression vector for Hh^N, a secreted and active form of the ligand, increased expression of the ptc-Luciferase (ptc-Luc) transcriptional reporter 13-fold compared with unstimulated cells (Fig. 7C). Co-expression of Mthl5-GFP significantly decreased Hh-induced ptc-Luc expression by almost 40% (Fig. 7C). Thus, in both flies and S2 cells, Mthl5 lowers Hh responsiveness.

To see whether Mthl5 affects cellular cAMP levels, we used the EPAC-BRET bioluminescence resonance energy transfer (BRET)-based cAMP biosensor, the BRET signal of which is inversely proportional to cAMP concentration (Cheng et al., 2012; Jiang et al., 2007). Compared with control cells, we observed a significant increase in net BRET signal (corresponding to a decrease in cAMP levels) in cells expressing Mthl5-GFP (Fig. 7D). Taken together, both loss- and gain-of-function results suggest that Mthl5 may normally limit Hh signalling by suppressing cAMP production.

To examine this further, we tested whether manipulations that lower cAMP levels would rescue the mthl5 mutant Hh target gene expression defect. We used two different approaches expected to specifically downregulate cAMP levels: expression of a mutant form of Gαi (Gαi^Q205L) that has a reduced ‘off’-rate, causing it to signal longer when activated (Schaefer et al., 2001); and expression of Dunce (Dnc), a Drosophila phosphodiesterase that hydrolyses cAMP (Davis and Kiger, 1981). As we previously reported (Cheng et al., 2012), expression of Gαi^Q205L did not alter Hh target gene expression in a wild-type background (Fig. 7E,F). However, in the mthl5^MB03076 background, Gαi^Q205L significantly reduced the width of the dpp expression domain (Fig. 7G,H). Similarly, Dnc expression on its own only weakly reduced dpp expression (Fig. 7I,J), but had a stronger effect in mthl5^MB03076 mutants (Fig. 7K,L). These genetic interactions suggest that the increase in Hh signalling in mthl5 mutants is due to an increase in cAMP levels.

As an independent confirmation, we tested the effect of limiting Pka activity on the mthl5 mutant phenotype. Wing discs from larvae heterozygous for mutant alleles of pka-C1 (pka-C1^76a3 and pka-C1^H2), encoding the Pka catalytic subunit were significantly larger than controls (Fig. 7M-O,S). However, when introduced into the mthl5^MB background, both alleles significantly suppressed disc overgrowth (Fig. 7P-S), supporting the interpretation that increased Pka activity due to elevated cAMP levels is the cause of the Hh signalling defects in mthl5 mutants.

PKA phosphorylation stabilizes Smo in Hh-responding cells. Therefore, we checked whether Smo protein levels are elevated in mthl5 mutant discs. In wild-type discs, Smo shows a complex pattern of post-translational regulation (Fig. 7T): protein levels are high in posterior compartment cells, where Ptc levels (and thus Smo inhibition) are low or absent; higher in anterior cells closest to the A-P boundary where Hh signalling (and Smo phosphorylation) is strongest; low in further anterior Hh-responding cells due to Gprk2-dependent internalization of activated Smo (Cheng et al., 2010); and lowest in far anterior cells beyond the range of Hh diffusion (Denef et al., 2000). In mthl5^MB03076 mutant discs with an expanded Hh response, we saw a corresponding anterior expansion of Smo stabilization (Fig. 7U,V). Conversely, expression of Mthl5-GFP blocked Smo stabilization in Hh-responding cells at the A-P boundary (Fig. 7W,X). These results suggest that Mthl5 impinges upon the pathway at or above the level of Smo, well upstream in the cellular response to Hh.

Several GPCRs have been implicated in the vertebrate Hh signalling pathway. They share an ability to modulate cAMP levels through Gαs and Gαi, which can alter sensitivity of cells to Hh ligand (Pusapati et al., 2018). Cells can express a large number of GPCRs, suggesting there could be widespread crosstalk between GPCRs and the Hh pathway. However, it remained an unresolved issue whether the ability to affect Hh signalling is restricted to only a few specific GPCRs or is a more general feature of these receptors. Our screen to test the ability of GPCRs expressed in the Drosophila wing disc to influence Hh responses indicates that the latter is true. We found that many GPCRs can crosstalk with the Hh pathway – some 45% of those tested in this system. Manipulation of these GPCRs expanded or restricted the domain of Hh responses, consistent with them working to enhance or diminish sensitivity to Hh rather than being strictly required components of the pathway. Analysis of mutants for one specific GPCR identified in the screen, Mthl5, provided evidence that this effect on Hh signalling involves modulation of cAMP levels (and thus PKA activity), a factor with a positive effect on Smo activity in Drosophila.

The best-characterized example of a GPCR influencing Hh signalling is GPR161 in mammals. In Gpr161 mutant mice, the range of expression of the Shh target genes Nkx2.2 and Foxa2 in the neural tube is expanded dorsally (Mukhopadhyay et al., 2013). This Hh gain-of-function phenotype can be recapitulated in Gpr161 mutant cultured neural precursor cells, and is accompanied by a three- to fourfold increase in sensitivity to Shh ligand (Pusapati et al., 2018). GPR161 overexpression had the opposite effect, suppressing the response to Shh in cells (Pusapati et al., 2018). This appears to be linked to the ability of GPR161 to activate Gαs and stimulate cAMP production, thereby promoting formation of the GLI3R transcriptional repressor. In this context, GPR161 acts as an attenuator of signalling by modulating how easily activated the pathway is, rather than as a direct inhibitor of signalling.

We saw similar effects from manipulating GPCRs in flies. This was true at the level of Hh-dependent adult wing phenotypes and readouts of Hh pathway activity in discs, including both low (Ci¹⁵⁵ stabilization and dpp expression) and intermediate/high (expression of ptc, stabilization of Smo) threshold responses. Most GPCRs fell neatly into two categories. The first consists of four GPCRs – Mthl5, GABA-B-R2, Cirl and Mth. The G protein selectivity of three of these is known, with all coupling to Gαi or Gαo to lower cAMP levels (Franek et al., 1999; Patel et al., 2016; Scholz et al., 2017; Wojcik and Neff, 1984). Depletion of these GPCRs generally expanded the range of Hh responses. This phenocopied Gαi depletion and resembled the effects seen in Gpr161 mutant mouse embryos. The second category contained six GPCRs – Mthl4, Mthl8, Mthl9, Rk, Stan and CG15744. The selectivity of only one of these receptors is known: Rk stimulates cAMP production through Gαs (Baker and Truman, 2002; Kimura et al., 2004). Depletion of these receptors phenocopied Gαs depletion, narrowing the range of dpp expression. The match between GPCR and Gα depletion phenotypes for those GPCRs with known coupling specificity provides support for the specificity of our screen results. For both categories of GPCRs, the effects fit nicely with the expectations of attenuators and enhancers of Hh responsiveness, respectively. Loss of Gαi-like receptors, such as Mthl5, allowed cells further from the source of Hh, and thus exposed to lower levels of Hh, to initiate a signalling response, while loss of Gαs-like GPCRs shrank the effective range.

Of the readouts we used, expression of dpp produced the most-striking effects, but is not exclusive to the Hh pathway. For example, apoptotic disc cells can upregulate dpp in some circumstances (Perez-Garijo et al., 2004; Ryoo et al., 2004), whereas En can repress dpp expression (Raftery et al., 1991; Strigini and Cohen, 1997). This raised the possibility that GPCRs could be acting on dpp through a different mechanism. With regard to apoptosis, we did not observe any correlation between apoptotic cells and dpp expression in mthl5-depleted or mutant discs, and the dpp expression domain was expanded even in conditions where apoptosis was comparatively low or suppressed. Furthermore, an apoptosis-based mechanism would not explain why dpp expression is more restricted in Mthl5-overexpressing discs, or when Gαs or GPCRs like Rk are depleted. Apoptosis thus seems unlikely to be the major driver of the changes we observed. Although failure to activate anterior en expression could upregulate dpp expression, we observed substantial anterior En expression after depletion of Gαi, Mthl5 or Cirl. Although the levels were slightly lower than normal for unknown reasons, a comparable decrease in anterior En levels was not sufficient to trigger expansion of dpp expression in Gαs- or Rk-depleted discs. Although we cannot rule out the possibility that a reduction in anterior En levels could contribute to expansion of dpp expression in Gαi-like GPCR-depleted discs, it seems unlikely to be the main cause.

The interpretation that Hh signalling itself is affected is best supported for the Gαi-like GPCRs. Unlike dpp expression, stabilization of Smo and Ci¹⁵⁵, and expression of Ptc are Hh-specific readouts. In particular, stabilization of Smo and Ci¹⁵⁵ are proximal readouts of cytoplasmic Hh signalling. Their concomitant upregulation in Gαi-like GPCR-depleted or mutant cells, specifically those within range of the A-P boundary (and a similar effect in discs in which PKA activity is moderately upregulated; Jia et al., 2004) is a good indicator of increased Hh pathway activity. As stabilization of Ci¹⁵⁵ is sufficient to de-repress dpp and enable its expression in response to Hh (Méthot and Basler, 1999), increased Hh-dependent Ci stabilization (and possibly Ci activation) is the most likely explanation for the observed expansion of dpp expression. However, given the stronger effects on dpp expression than on other Hh pathway readouts, it remains possible that GPCRs affect dpp through another mechanism. We also cannot rule out the possibility that GPCRs affect hh expression, rather than ligand sensitivity. This is not the case in mice (Pusapati et al., 2018), and the fact that Mthl5 reduced responsiveness to transfected Hh in S2 cells (as did Gαs depletion in our previous work; Praktiknjo et al., 2018) suggests that the effects we see occur downstream of Hh.

One of the main differences in the effects of GPCR signalling between different systems is that cAMP attenuates Hh pathway activity in mammals but seems to predominantly enhance it in flies. In both systems, cAMP/PKA negatively regulates Hh target gene expression by promoting conversion of Ci/GLI proteins to their repressor forms. However, in Drosophila PKA also acts as the gatekeeper for pathway activation by phosphorylating Smo. Although its inhibitory effect on Ci can be inferred in extreme conditions in flies (such as pka-C1 mutant clones; Li et al., 1995; Ohlmeyer and Kalderon, 1998), the activating function of PKA seems to predominate at physiological levels of cAMP/PKA signalling. This may be because Ci is protected from increased PKA activity as Smo activity increases in fly cells. Specifically, PKA switches from Ci- to Smo-containing complexes in response to Hh, and active Smo promotes release of Ci from the complex that converts it to CiR (Ranieri et al., 2014; Robbins et al., 1997; Ruel et al., 2007).

The evidence suggests that GPCRs are not core components involved in transduction of the Hh signal. Instead, their activity feeds in to shape the response to Hh in tissues, modulating basal cellular cAMP levels to alter thresholds for Hh pathway activation and output (Pusapati et al., 2018). In vertebrate cells, this is thought to be through effects on GLI regulation. The higher the basal cellular levels of cAMP and PKA activity are, the more likely that GLI proteins are converted to their repressor forms. For a given level of target gene expression, overcoming this increased activation threshold requires stronger Hh pathway activity, meaning exposure to a higher concentration of Shh. Ligand sensitivity is thus attenuated by cAMP, which narrows the range over which target genes are expressed. However, in Drosophila cells the higher basal cellular levels of cAMP and PKA activity are, the lower the barrier to Smo activation. The pathway should thus be activated at lower ligand concentrations, which would expand the range over which target genes are expressed. It makes sense that modulating ligand sensitivity in this way would have the strongest effects in cells exposed to the lowest and most-limiting Hh levels. Consistent with this, the strongest changes we observed in both Gαs- and Gαi-like GPCR-depleted discs were in dpp-expressing cells farther from the source of Hh.

We identified the little-characterized GPCR Mthl5 as a novel attenuator of Hh signalling in discs. mthl5 mutants showed clear signs of Hh pathway upregulation, including anterior compartment overgrowth and increased L3-L4 spacing in wings, and anterior expansion of Smo and Ci¹⁵⁵ protein stabilization and Hh target gene expression. Overexpression of Mthl5 had the opposite effect, restricting the domain of dpp expression. Mthl5 likely crosstalks with the Hh pathway by coupling to inhibitory Gα subunits and decreasing cellular cAMP levels. This is supported by the similarity of mthl5 and gαi depletion phenotypes, the ability of Mthl5 expression to lower cAMP levels and ptc-reporter activity in S2 cells, and the partial rescue of mthl5 mutant defects by lowering cAMP or PKA levels.

The anterior expansion of dpp expression in mthl5 mutants caused a corresponding shift in Dpp signalling activity, as evidenced by an increase in the width of the anterior expression domain of salm. Cells in both the wing pouch and the pleura are dependent on low-level activation of Dpp signalling for growth (Barrio and Milán, 2017). We imagine that the anterior expansion of dpp expression could lead to excessive proliferation of cells in the anterior lateral and hinge region of the discs, which could in part account for overgrowth in mthl5 mutant discs, although this remains to be confirmed.

Our results demonstrate that crosstalk of GPCR signalling with Hh signalling is conserved from flies to vertebrates, and suggest that there may be many more vertebrate GPCRs that are capable of doing this. The cilium, which plays a central organizing role in vertebrate Hh signalling (Bangs and Anderson, 2017), may act as an insulator against this sort of crosstalk by compartmentalizing cAMP responses for some GPCRs (Marley et al., 2013). Although it remains to be tested, this crosstalk could be a specific mechanism for linking the myriad stimuli that activate GPCRs – mechanical force for Cirl (Scholz et al., 2017), planar polarity cues for Stan (Usui et al., 1999), a nutrient-regulated peptide for Mth (Delanoue et al., 2016) and a glycoprotein hormone marking developmental timing for Rk (Loveall and Deitcher, 2010) – to developmental patterning and tissue homeostasis. Further work will be needed to determine the links between the physiological regulators of these GPCRs and Hh pathway output.

For crosses involving RNAi transgene expression, RNAi transgenic males were crossed to w,UAS-Dcr;nub-Gal4 females (for adult wings) or w,UAS-Dcr;ap-Gal4,dpp¹⁰⁶³⁸/CyO,Kr-Gal4,UAS-GFP females (for discs) at 25°C for 2 to 3 days and the resulting offspring were transferred to 27°C until dissection or hatching. All other crosses were carried out at 25°C. For crosses where we analysed growth (Figs 4D-F, 5A-G, 6D-G, 7M-R) or directly compared Hh responses in mthl5 mutants and wild-type discs (Figs 5J,K,O,P, 6A,B, 7T,U), crosses were handled in parallel batches and flipped regularly to avoid crowding of the larvae. The following strains were used: UAS-Dcr-2.D, UAS-P35, nub-GAL4, ap-GAL4, dpp¹⁰⁶³⁸ (dpp-LacZ), salm⁰³⁶⁰² (salm-LacZ), mthl5^MB03076, Df(3R)BSC514, UAS-Dnc, tkv¹², pka-C1^H2 and pka-C1^76a3 [all obtained from the Bloomington Drosophila Stock Centre (BDSC)]; UAS-gαi.dsRNA and UAS-Gαi^Q205L (Schaefer et al., 2001) [generously provided by J. Knoblich (Institute of Molecular Biotechnology, Vienna, Austria)]. A list of transgenic RNAi strains obtained from the Vienna Drosophila Resource Centre (indicated by a ‘V’) or from the Transgenic RNAi Project via the BDSC (indicated by ‘TRiP’) and used in this study can be found in Table S2.

The mthl5^MB03076 strain contains an insertion of a Minos element in coding exon 5 of the mthl5 gene, at position 3R:11,884,499. We validated the insertion by amplifying both the 5′ and 3′ flanking regions, using the following primer pairs: Minos 5′ (5′-cttcatctttcaggaggcttg-3′)+mthl5 exon 6 (5′-gacattactaacagcatcagcg-3′); Minos 3′ (5′-ggagttcttcgcccaccc-3′) plus mthl5 exon 4 (5′-gctccgcaaaatcttaaaccc-3′). The resultant PCR products were cloned into pGEM according to manufacturer's instructions and sequenced to confirm the correct insertion site. The w¹¹¹⁸;mthl5^MB03076/TM3 stock we obtained from Bloomington had a background lethal mutation on the third chromosome, which we removed by recombination.

To generate Mthl5-GFP expression vectors, the Mthl5 open reading frame was PCR amplified from cDNA in two overlapping fragments. The 5′ fragment from nucleotides 1 to 738 of the coding sequence was amplified to introduce an EcoRI site immediately upstream of the start codon (primers 5′-gaattcatgctcgtaaaaacgcttgg-3′ and 5′-cgttgtcacaatgttaccaacc-3′). The 3′ fragment from nucleotides 721 to 1488 of the coding sequence was amplified to replace the stop codon with a NotII site (primers 5′-cctactctcagagatctggttgg-3′ and 5′-gcggccgcgtaatcgttgccgttcatataa-3′). Each PCR product was cloned into pGEM. The 5′ fragment was then excised using EcoRI and BglII, which cuts mthl5 at coding sequence nucleotide 732. The 3′ fragment was excised using BglII and NotI. These two fragments were then cloned together with a NotI-KpnI fragment encoding GFP into pMT.puro for expression in cell culture. The full-length sequence was then excised as an EcoRI-KpnI fragment and transferred into pUAST. Transgenic flies carrying pUAST/Mthl5-GFP were generated by BestGene.

The following genotypes are shown in each figure: Fig. 1A, w,UAS-Dcr/w;nub-GAL4/+; Fig. 1B, w,UAS-Dcr/w;nub-GAL4/+;UAS-gαs-IR^{TRiP.HMC03106}/+; Fig. 1C, w,UAS-Dcr/w;nub-GAL4/UAS-gαi-IR; Fig. 1E, w,UAS-Dcr/w;nub-GAL4/UAS-mthl6-IR^V108048; Fig. 1F, w,UAS-Dcr/w;nub-GAL4/UAS-mthl8-IR^V100246; Fig. 1G, w,UAS-Dcr/w;nub-GAL4/UAS-mthl9-IR^V108967; Fig. 1H, w,UAS-Dcr/w;nub-GAL4/UAS-rk-IR^V105360; Fig. 1I, w,UAS-Dcr/w;;nub-GAL4/UAS-rk-IR^V905; Fig. 1J, w,UAS-Dcr/w;nub-GAL4/+;UAS-cg15744-IR^V4801/+; Fig. 1K, w,UAS-Dcr/w;nub-GAL4/UAS-mthl4-IR^V50752; Fig. 1L, w,UAS-Dcr/w;nub-GAL4/UAS-stan-IR^V107993; Fig. 1N, w,UAS-Dcr/w;nub-GAL4/UAS-mthl5-IR^V101593; Fig. 1O, w,UAS-Dcr/w;nub-GAL4/UAS-gaba-b-r2-IR^V1785; Fig. 1P, w,UAS-Dcr/w;nub-GAL4/+;UAS-gaba-b-r2-IR^TRiP.JF02779/+; Fig. 1Q, w,UAS-Dcr/w;nub-GAL4/UAS-cirl-IR^V100749; Fig. 1R, w,UAS-Dcr/w;nub-GAL4/UAS-mth-IR^V102303; Fig. 2A, w,UAS-Dcr/w;ap-GAL4,dpp¹⁰⁶³⁴⁶/+; Fig. 2C, w,UAS-Dcr/w;ap-GAL4,dpp¹⁰⁶³⁴⁶/+;UAS-gαs-IR^{TRiP.HMC03106}/+; Fig. 2E, w,UAS-Dcr/w;ap-GAL4,dpp¹⁰⁶³⁴⁶/UAS-mthl4-IR^V50752; Fig. 2G, w,UAS-Dcr/w;ap-GAL4,dpp¹⁰⁶³⁴⁶/UAS-mthl8-IR^V100246; Fig. 2I, w,UAS-Dcr/w;ap-GAL4,dpp¹⁰⁶³⁴⁶/UAS-mthl9-IR^V108967; Fig. 2K, w,UAS-Dcr/w;ap-GAL4,dpp¹⁰⁶³⁴⁶/UAS-rk-IR^V905; Fig. 2M, w,UAS-Dcr/w;ap-GAL4,dpp¹⁰⁶³⁴⁶/+;UAS-cg15744-IR^V4801/+; Fig. 2O, w,UAS-Dcr/w;ap-GAL4,dpp¹⁰⁶³⁴⁶/UAS-stan-IR^V107993; Fig. 3A, w,UAS-Dcr/w;ap-GAL4,dpp¹⁰⁶³⁴⁶/UAS-gαi-IR; Fig. 3C, w,UAS-Dcr/w;ap-GAL4,dpp¹⁰⁶³⁴⁶/UAS-mthl5-IR^V101593; Fig. 3E, w,UAS-Dcr/w;ap-GAL4,dpp¹⁰⁶³⁴⁶/UAS-gaba-b-r2-IR^V1785; Fig. 3G, w,UAS-Dcr/w;ap-GAL4,dpp¹⁰⁶³⁴⁶/UAS-mth-IR^V102303; Fig. 3I, w,UAS-Dcr/w;ap-GAL4,dpp¹⁰⁶³⁴⁶/UAS-cirl-IR^V100749; Figs 4A,B,D, 5A,O, 6D, 7M,T, w¹¹¹⁸; Figs 4E,I, 5C,D, 6F, 7P,U, w;;mthl5^MB03076/mthl5^MB03076; Figs 4F,J, 5F,G,P, w;;mthl5^MB03076/Df(3R)BSC514; Fig. 5B, w;;mthl5^MB03076/+; Fig. 5E, w,UAS-Dcr/w;nub-GAL4/UAS-smo3′UTR-dsRNA;UAS-Smo-GFP/+; Fig. 5J, w;dpp¹⁰⁶³⁴⁶/+; Fig. 5K,Q, w;dpp¹⁰⁶³⁴⁶/+;mthl5^MB03076/mthl5^MB03076; Fig. 5N, w;dpp¹⁰⁶³⁴⁶/+;mthl5^MB03076/Df(3R)BSC514; Fig. 6A, w;salm⁰³⁶⁰²/+; Fig. 6B, w;salm⁰³⁶⁰²/+;mthl5^MB03076/mthl5^MB03076; Fig. 6E, w;tkv¹²,FRT40/+; Fig. 6G, w;tkv¹²,FRT40/+;mthl5^MB03076/mthl5^MB03076; Fig. 7A,W, w;ap-GAL4,dpp¹⁰⁶³⁴⁶/+;UAS-Mthl5-GFP/+; Fig. 7E, w;ap-GAL4,dpp¹⁰⁶³⁴⁶/UAS-Gαi^Q205L; Fig. 7G, w;ap-GAL4,dpp¹⁰⁶³⁴⁶/UAS-Gαi^Q205L;mthl5^MB03076/mthl5^MB03076; Fig. 7I, w;ap-GAL4,dpp¹⁰⁶³⁴⁶/UAS-Dnc; Fig. 7K, w;ap-GAL4,dpp¹⁰⁶³⁴⁶/UAS-Dnc;mthl5^MB03076/mthl5^MB03076; Fig. 7N, w;pka-C1^76a3/+; Fig. 7O, w;pka-C1^H2/+; Fig. 7Q, w;pka-C1^76a3/+;mthl5^MB03076/mthl5^MB03076; and Fig. 7R, w;pka-C1^H2/+; mthl5^MB03076/mthl5^MB03076.

A rabbit polyclonal antiserum against the Mthl5 extracellular N terminus was generated at Genscript. A bacterially expressed fusion protein consisting of Mthl5 amino acids 24-219 (lacking the N-terminal signal peptide) fused to a 6xHis tag was used as the immunogen.

Wing discs were dissected in PBS from third instar larvae of the genotype w;apGAL4/+ and transferred individually in a minimal volume of PBS using forceps to a microcentrifuge tube on ice. After 20 min of collecting, the tubes were snap-frozen and stored at −70°C. Total RNA was isolated from 100-120 imaginal discs per sample using TRIzol reagent (ThermoFisher Scientific). RNA was quantified using a Nanodrop spectrophotometer and quality was assessed using a 2100 Bioanalyzer instrument (Agilent Technologies). Three independent RNA preparations were submitted to the IRCM Molecular Biology Core Facility for library preparation. Libraries were pair-end sequenced at Genome Quebec using a HiSeq2000 platform. Bioinformatic analysis was performed by the IRCM Bioinformatics Core Facility. Raw reads (26-28 million per sample) were filtered for quality (using FastQC, phred33 score>30) and aligned to the dm3 reference genome using TopHat2, yielding per-gene sequence read counts.

Wandering third instar larvae were dissected, and the imaginal discs fixed and processed for immunostaining as previously described (Maier et al., 2014). For comparisons of Hh responses in mthl5 mutants and wild-type discs (Figs 5J,K,O,P, 6A,B, 7T,U), crosses for comparison were handled, dissected and processed for immunostaining in parallel batches, and discs imaged using the same confocal settings. Antibodies used for immunostaining were: rabbit anti-β-galactosidase (A11132, 1:1000; Thermo Fisher Scientific), rat anti-Ci¹⁵⁵ [2A1, 1:50; Developmental Studies Hybridoma Bank (DSHB), Iowa City, IA, USA], mouse anti-Ptc (Apa-1, 1:1000; DSHB), rabbit anti-cleaved caspase 3 (Asp 175, 1:200; Cell Signaling Technologies), mouse anti-Wg (4D4, 1:25; DSHB) and rabbit anti-Mthl5 (1:1000; this study; Fig. S2). Image stacks of the discs were taken using a Zeiss LSM700 confocal microscope. Quantification of fluorescence intensity was carried out using Image J. For dorsal versus ventral compartment comparisons, measurements were made in equal-sized rectangles spanning the A-P boundary in dorsal and ventral regions of each wing disc image using the Plot Profile function. For Ci¹⁵⁵ and Hh target genes, dorsal and ventral data were normalized by dividing each value by the average of the 10 highest intensity values from the ventral (wild-type) compartment within the same disc, effectively converting them to the percentage of maximum wild-type response. For En and Smo, data were normalized to the mean fluorescence intensity in the wild-type ventral posterior compartment. For comparisons between mthl5 mutants and wild-type discs (Figs 5J,K,O,P, 6A,B, 7T,U), measurements were made in a single box of uniform size spanning the D-V boundary and data were normalized to the average maximal mean fluorescence intensity in the wild-type discs (for target genes and Ci¹⁵⁵) or the average posterior compartment fluorescence (for Smo). Data from five discs were arranged to align the A-P boundaries and pixel-by-pixel data averaged to yield a mean intensity plot±s.d. along the A-P axis. Quantification of the total size of the imaginal disc was performed using ImageJ by tracing the outlines of DAPI-stained discs using the Polygon tool and measuring the outlined area with the Measurement tool. To measure wing pouch size, a line spanning the folds at the posterior and anterior limits of the pouch was drawn at the midline of the wing pouch using the Straight line tool, and its length measured with the Measurement tool. For estimating the anterior and posterior extents of salm expression, a straight line spanning the A-P boundary to the anterior or posterior limit of salm-LacZ expression was drawn along the midline of the pouch and measured. These values were divided by the total pouch length to correct for any small differences in disc size, although similar results were obtained without normalization. Statistical significance was determined using an unpaired two-tailed Student's t-test.

Adult flies were collected and stored in 50% ethanol/50% glycerol. Before mounting, they were rinsed with water. Wings were then dissected in water and transferred into a drop of Faure's solution on a glass slide and cover-slipped. A weight was placed on the cover slip to flatten the wings and they were left to dry overnight. Wings were imaged with a Leica DM2000 light microscope. Wings areas were measured by outlining the appropriate region (whole wing, L3-L4 area, from L4 to the anterior margin to approximate anterior compartment and from L4 to the posterior margin to approximate posterior compartment) using the Polygon selection tool of Image J. The area of the resulting polygon was calculated using the Measure tool.

EPAC-BRET assays were performed as previously described (Maier et al., 2014). Briefly, 1×10⁶ S2R+ cells [obtained from the Drosophila Genomics Resource Center (DGRC)] were plated in 0.5 ml of serum-free Schneider's medium in 24-well plates on day 1. A few hours later, they were transfected with 100 ng of pMT.puro/Mthl5-GFP or empty pMT.puro vector (control), together with 250 ng of the EPAC-cAMP biosensor-encoding plasmid pMT.puro/GFP10-EPAC-RLucII_T781A,F782A. Transgene expression was induced on day 2 by adding CuSO₄ to a concentration of 0.5 mM. On day 4, the cells were harvested, washed once with PBS and split into four wells of a white walled, clear-bottom 96-well plate. Substrate was added and emissions at 400 nm (RLucII, donor) and 515 nm (GFP10, acceptor) measured and corrected for background as described previously (Cheng et al., 2012). The net BRET signal was calculated as the ratio of background corrected GFP10: RLucII emission. Quadruplicate readings for each condition were averaged, and the experiment performed eight times. Statistical significance of the difference in the compiled experimental averages for each condition was assessed using a ratio-paired t-test in GraphPad Prism. This test controls for inter-experimental variability in the magnitude of the raw reads. Data were then normalized to the mock-transfected control (=100%) within each experiment to generate graphs.

ptc-Luciferase reporter assays were performed as previously described (Maier et al., 2014). Briefly, cells were plated as above and co-transfected with 100 ng pMT/Ci, 75 ng pGL.basic/ptcD136-luc (Chen et al., 1999), 75 ng pRL/CMV and 100 ng of each additional expression plasmid (pMT.puro/Hh^N, pMT.puro/Mthl5-GFP or pMT.puro vector as control). Cells were induced and split as above. On day 4, luciferase activity measurements were performed as described previously (Maier et al., 2014). Quadruplicate readings for each condition were averaged, and the experiment performed six times. Statistical significance of the difference in the compiled experimental averages for each condition was assessed using a ratio-paired t-test in GraphPad Prism. Data were then normalized to the mean Hh response (+Hh condition=100%) within each experiment to generate graphs.

In silico prediction of the transmembrane helices in Mthl5 protein was performed using the TMHMM Server v. 2.0 (http://www.cbs.dtu.dk/services/TMHMM/).

We thank Karen Oh for expert technical assistance, Dhara Patel for her early contribution to the project and Dr Virginie Calderon of the IRCM Bioinformatics Core Facility for preparing the heat map of GPCR expression. Fly stocks obtained from the Bloomington Drosophila Stock Center (supported by NIH grant P40OD018537) were used in this study, as were antibodies from the Developmental Studies Hybridoma Bank (created by the NICHD, National Institutes of Health, and maintained at The University of Iowa, Department of Biology, Iowa City, IA 52242) and cells from the DGRC (Drosophila Genomics Resource Center; supported by NIH grant 2P40OD010949).

The peer review history is available online at https://journals.biologists.com/dev/article-lookup/148/7/dev189258/