Regulation of mammalian Gli proteins by Costal 2 and PKA in Drosophila reveals Hedgehog pathway conservation

Abstract

Hedgehog (Hh) signaling activates full-length Ci/Gli family transcription factors and prevents Ci/Gli proteolytic processing to repressor forms. In the absence of Hh, Ci/Gli processing is initiated by direct Pka phosphorylation. Despite those fundamental similarities between Drosophila and mammalian Hh pathways, the differential reliance on cilia and some key signal transduction components had suggested a major divergence in the mechanisms that regulate Ci/Gli protein activities, including the role of the kinesin-family protein Costal 2 (Cos2), which directs Ci processing in Drosophila. Here, we show that Cos2 binds to three regions of Gli1, just as for Ci, and that Cos2 functions to silence mammalian Gli1 in Drosophila in a Hh-regulated manner. Cos2 and the mammalian kinesin Kif7 can also direct Gli3 and Ci processing in fly, underscoring a fundamental conserved role for Cos2 family proteins in Hh signaling. We also show that direct PKA phosphorylation regulates the activity, rather than the proteolysis of Gli in Drosophilia, and we provide evidence for an analogous action of PKA on Ci.

Fig. 1.

Gli3 processing in Drosophila requires PKA and Cos2. (A-F) Wing discs with MARCM (Lee and Luo, 2001) clones (arrows) expressing Gli3 were marked by GFP expression (green) and tested for repression of hh-lacZ expression (red) in posterior compartment (right) cells. Clones also lacked (B) smo, (C) smo and pka or (E) smo and cos2 activities, or additionally expressed (D) activated mouse PKA catalytic subunit (mC*) or (F) wild-type Cos2. Repression of hh-lacZ in B,D,F indicates Gli3 processing to a repressor form.

Fig. 2.

PKA phosphorylation and Cos2 binding silence Gli1 in the absence of Hh signaling. (A-F) Wing discs expressing wild-type Gli1 or the indicated Gli1 variants ubiquitously using the C765-GAL4 driver. Gli1ΔPKA has S544A and S560A substitutions, GliΔCDN lacks residues 141-260, and Gli1ΔCORD lacks residues 601-750. Activation of the Hh target gene ptc-lacZ (red) was measured by staining with antibody to β-galactosidase in wing discs containing clones (arrows) marked by loss of (green) GFP and lacking (A,D-F) smo, (B) smo and pka, or (C) smo and cos2 activities. Gli1 activity is silenced within posterior (right) smo mutant clones (A) but is restored by loss of PKA (B), Cos2 (C), PKA sites S544 and S560 (D) or the CORD Cos2-binding domain (F).

Fig. 3.

Activity of Gli1 variants. (A-F) Wild-type Gli1 or variants were expressed throughout wing discs using C765-GAL4 to test induction of the Hh target gene ptc-lacZ (red) in anterior cells (left) and in posterior cells (right), which produce Hh. Gli1 lacking the SYGH motif (B) or residues 1-130 (C), both of which are implicated in Su(fu) binding, or lacking CDN domain residues 141-260 (E) induced ectopic ptc-lacZ expression only in posterior cells. Gli1 lacking PKA sites S544 and S560 or CORD region residues 601-750 also induced ptc-lacZ expression weakly (D) or strongly (F) in anterior cells.

Fig. 4.

Gli proteins have multiple Cos2-binding domains analogous to CDN, CORD and zinc-finger domains on Ci. (A) Schematic alignment of full-length Ci, Gli1 and Gli3 proteins using the conserved SYGH motif (blue), zinc fingers (turquoise) and PKA-nucleated phosphorylation region (red) as points of registration. Other elements, including CDN and CORD regions (green), cannot be aligned through primary amino acid sequence similarities. Lines indicate protein fragments fused to GST and found to bind Cos2 (black) or not to bind Cos2 (red). (B-E) Equal amounts of GST, GST-Gli1 or GST-Gli3 fragments (labeled by Gli residues included) were used to pull-down proteins from extracts of Kc cells expressing either Myc-tagged Cos2 or HA-tagged Kif7, and associated proteins were visualized on western blots with epitope tag antibodies. Either 2% or 5% of the extract used in each binding assay was also included on the western blots. (E) GST-Gli1 fusion protein was phosphorylated with the indicated kinase, or ATP was added to cell extracts (to 1 mM) where indicated, prior to mixing Kc cell extracts and GST proteins.

Fig. 5.

Kif7 can promote Ci and Gli3 processing but Gli1 proteins are not processed in Drosophila. (A-C) Wing discs with MARCM (Lee and Luo, 2001) smo cos2 mutant clones (arrows) expressing either (B) UAS-cos2 or (C) UAS-Kif7 were marked by GFP expression (green). Ci-155 levels, revealed by 2A1 antibody staining (red) were increased roughly to AP border levels in smo cos2 mutant clones unless rescued by UAS-cos2. (D-H) Wing discs with MARCM clones (arrows) expressing Ci or Gli proteins in the absence of smo activity were marked by GFP expression (green) and tested for repression of hh-lacZ (red) in posterior compartment (right) cells. (D-F) Clones also lacked cos2 activity but expressed Kif7. Kif7 allowed processing of (D) Gli3, (E) wild-type Ci and (F) Ci lacking CDN and CORD Cos2-binding domains, eliciting hh-lacZ repression in clones. (G) Wild-type Gli1 or (H) Gli1 with residues 1-524 substituted by residues 1-829 of Ci (Ci-Gli1) did not produce a repressor form.

Fig. 6.

Gli1 activity is regulated independently of proteolysis in Drosophila. (A-D) Wing discs expressing UAS-Gli1 using (A) en-GAL4, (B) C765-GAL4 or (C,D) MS1096-GAL4. (A-C) Gli1 protein, measured by C-terminal HA epitope tag staining (red), was confined to posterior cells (right) when expressed only in those cells (A) but was detected at similar levels in anterior (left) and posterior cells when GAL4 was expressed evenly across the AP axis (B,C). (D) Gli1 induced ptc-lacZ expression (red) more strongly in posterior ventral cells (upwards arrow) than in anterior dorsal wing pouch cells (downwards arrow), even though Gli1-HA protein levels were higher in the latter region (arrows in C). (E-L) Ci, Gli1 or Ci-Gli1 proteins were expressed ubiquitously using C765-GAL4 in wing discs containing (E,G,I,K) smo or (F,H,J,L) smo pka mutant clones (arrowheads), marked by loss of (green) GFP. (E,F) Full-length Ci protein detected by C-terminal Myc epitope staining (red) was reduced in posterior smo mutant clones but elevated by loss of PKA, but neither of these responses was seen for (G,H) Gli1, detected by C-terminal HA epitope staining (red) or for (I,J) Ci-Gli1, detected by 2A1 antibody (red), which also detects anterior Ci-155. (K,L) Induction of ptc-lacZ (red) by Ci-Gli1 was lost in posterior smo mutant clones and restored by loss of PKA.

Fig. 7.

PKA and Cos2 silence proteolysis-resistant Ci (Ci-S849A) in the absence of Su(fu). (A-C) Su(fu)^LP mutant wing discs ubiquitously expressing Ci-S849A, which is resistant to PKA- and Cos2-dependent processing, using the C765-GAL4 driver. Activation of ptc-lacZ (red) was (A) reduced in posterior (right) smo clones, but was restored in posterior (B) smo pka and (C) smo cos2 clones. Clones (arrows) are marked by loss of (green) GFP.