The unfolded protein response selectively targets active smoothened mutants

Abstract

The Hedgehog signaling pathway, an essential regulator of developmental patterning, has been implicated in playing causative and survival roles in a range of human cancers. The signal-transducing component of the pathway, Smoothened, has revealed itself to be an efficacious therapeutic target in combating oncogenic signaling. However, therapeutic challenges remain in cases where tumors acquire resistance to Smoothened antagonists, and also in cases where signaling is driven by active Smoothened mutants that exhibit reduced sensitivity to these compounds. We previously demonstrated that active Smoothened mutants are subjected to prolonged endoplasmic reticulum (ER) retention, likely due to their mutations triggering conformation shifts that are detected by ER quality control. We attempted to exploit this biology and demonstrate that deregulated Hedgehog signaling driven by active Smoothened mutants is specifically attenuated by ER stressors that induce the unfolded protein response (UPR). Upon UPR induction, active Smoothened mutants are targeted by ER-associated degradation, resulting in attenuation of inappropriate pathway activity. Accordingly, we found that the UPR agonist thapsigargin attenuated mutant Smoothened-induced phenotypes in vivo in Drosophila melanogaster. Wild-type Smoothened and physiological Hedgehog patterning were not affected, suggesting that UPR modulation may provide a novel therapeutic window to be evaluated for targeting active Smoothened mutants in disease.

Fig 1

Active Smo mutants are temperature sensitive in vivo. (A and B) Growth at 29°C induces the UPR. RNA was harvested from w¹¹¹⁸ larvae and assessed for endogenous Xbp1 splicing by RT-PCR. The stress-induced Xbp1 splice variant was not observed in RNA from larvae grown at 22°C, but it was evident in RNA extracted from larvae grown at 29°C (A). The ER stress sensor UAS-Xbp1-GFP was expressed in salivary glands under the control of sgs3-Gal4 (sgs>Xbp1-GFP). Crosses were performed at 22°C or 29°C, as indicated. Individual salivary glands are outlined in white. Under conditions of low ER stress, Xbp1 was not in frame with GFP, and minimal GFP expression was observed (B′). Upon ER stress induction, Xbp1 was alternately spliced to place it in frame with GFP, resulting in a robust GFP signal (B). (C to F) Active Smo mutants are temperature sensitive. Wild-type, C320A, and C339A Myc-Smo proteins were expressed at 22°C (D to F) or 29°C (D′ to F′) under the control of the MS1096-Gal4 driver. Wild-type Myc-Smo did not induce a significant phenotype at 22°C and induced a mild phenotype at 29°C (D′, compared to panel D). Conversely, C320A and C339A Smo mutants induced mild phenotypes at 29°C and strong phenotypes at 22°C (E and F), suggesting that their activity is reduced under conditions of thermal stress. The MS1096 driver wing served as a control (C). (G) Expression of an active Smo mutant does not induce ER stress. Salivary glands from larvae expressing the Xpb1-GFP stress sensor alone (G′) or in combination with Myc-SmoC320A (G) under the control of sgs3-Gal4 were grown at 22°C on vehicle- or thapsigargin-treated food. Myc-SmoC320A (blue) did not induce the GFP stress sensor (G), but the potent ER stress-inducing compound thapsigargin did (G′). The filamentous actin stain phalloidin marks the plasma membrane (F-actin [magenta]). (H and H′). The Xbp1 mutation does not modify the Myc-SmoC320A phenotype. Wings from flies expressing Myc-SmoC320A under the control of the epithelial driver C765-Gal4 in the absence (H) or presence (H′) of the Xpb1 loss-of-function allele Xbp1^k13803 are shown. Introduction of the Xbp1^k13803 allele did not modify the SmoC320A-induced wing phenotype. For all conditions, representative salivary glands or wings are shown. Bars, 50 μm.

Fig 2

Active Smo mutants are temperature sensitive in vitro. (A) Cl8 cells were transfected with empty vector or wild-type, C320A, or C339A Myc-Smo expression vectors (pAc-myc-smo) in the presence of pAc-hh or empty vector and control or smo 5′UTR dsRNA, as indicated. Reporter gene activity was determined from cells cultured at 22°C or 29°C, as indicated. Whereas wild-type Myc-Smo rescued ptcΔ136-luciferase activity at both temperatures, C320A and C339A were compromised in their abilities to modulate reporter gene activity at the restrictive 29°C temperature. The control Hh response at each temperature was set to 100%. Reporter gene activity is shown as the percent activity relative to the control Hh response. Hh reporter gene activity was normalized against a pAc-Renilla control. Error bars indicate standard errors of the means (SEM). (B) Cl8 cells were transfected with hh or wild-type or mutant myc-smo expression vectors, as indicated. The Hh response for each temperature was set to 100%. Reporter activity induced by the indicated Myc-Smo protein in the wild-type smo background is shown relative to the Hh response. Hh reporter gene activity was normalized against a pAc-Renilla control. Error bars indicate SEM. (C) The heat shock response does not alter signaling by the active Smo mutant. Cl8 cells grown at 29°C were transfected with hh or myc-smoC339A expression vectors and treated with DMSO (vehicle) or the Hsp70 inhibitor VER, as indicated. Reporter activity induced by Myc-SmoC339A is shown relative to the Hh response, set to 100%. VER did not rescue C339A signaling at the restrictive temperature, indicating that attenuated mutant Smo signaling at 29°C was not due to the heat shock response. Hh reporter gene activity was normalized against a pAc-Renilla control. Error bars indicate SEM.

Fig 3

Active Smo mutants are retained in the ER. (A to C) Wild-type (A), C320A (B), and C339A (C) Myc-Smo proteins were expressed in S2 cells at permissive (A to C) or restrictive (A′ to C′) temperatures. Myc-Smo was visualized by indirect immunofluorescence. Smo (Myc) is red, phalloidin (PM marker) is blue, and Cal-GFP-KDEL (ER marker) is green. Wild-type Smo was vesicular at both temperatures (A and A′). Mutants overlapped with the ER marker at both temperatures (B and C). Bar, 5 μm. (D) Activated Smo mutants are not post-ER glycosylated. Whole-cell lysates from Cl8 cells expressing Hh and wild-type, C320A, or C339A Myc-Smo proteins at 22°C were treated with λ-phosphatase and the indicated deglycosylating enzymes. Post-ER-phosphorylated and glycosylated forms of wild-type Smo were detected (lanes 1 to 3 compared to lane 4, fully collapsed). C320A and C339A Smo proteins fully collapsed upon endo H and PNGase treatments, indicative of them being retained in the ER (lanes 5 to 12). Kinesin served as a loading control.

Fig 4

Active Smo mutants are destabilized at the restrictive temperature. pAc-myc-smo vectors encoding wild-type (A and D), C320A (B and E), and C339A (C and F) Myc-Smo proteins were cotransfected into S2 cells with pAc-GFP at permissive (22°C) and restrictive (29°C) temperatures. Cells were stained for Smo by indirect immunofluorescence using anti-Myc (magenta) and imaged by confocal microscopy. Whereas wild-type Myc-Smo (A and A′) and GFP (D to F and D′ to F′) stability and expression were not significantly affected by temperature, both active mutants were destabilized at the restrictive temperature (B and C compared to B′ and C′). Multiple fields of cells were examined over two independent experiments. Representative fields are shown. Bar, 50 μm. (G) The Smo protein is destabilized at the restrictive temperature. Western blot analysis of whole-cell lysates from Cl8 cells expressing wild-type or C339A Myc-Smo proteins revealed C339A protein levels were decreased at 29°C. Wild-type Smo was not destabilized at 29°C. Kinesin (Kin) was used as the loading control.

Fig 5

Murine Smo mutants are ER localized and temperature sensitive. (A and A′) Mammalian mSmo mutants are not post-ER glycosylated. Lysates prepared from NIH 3T3 cells expressing wild-type, C299A, and C318A mSmo proteins were treated with vehicle (−), endo H, PNGase, O-glycosidase, and/or λ-phosphatase as indicated (+) and analyzed by Western blotting. The post-ER glycosylated form, present only with wild-type mSmo, was not affected by endo H (post-ER; A and A′) but was affected by PNGase and O-glycosidase (deglycosylated; A and A′). The ER-localized forms of the wild type and each of the active mutants were sensitive to endo H, demonstrating a faster mobility after treatment (deglycosylated; A and A′). Endo H- and PNGase-treated C299A and C318A had identical mobilities, suggesting that they lack post-ER glycosylation (deglycosylated; A). λ-Phosphatase did not affect mobility, indicating that wild-type mSmo is not phosphorylated in the absence of Shh (A′, lane 10). Tubulin was the loading control. (B) Oncogenic mSmoM2 is largely ER retained. Lysates from NIH 3T3 cells expressing wild-type or M2 mSmo proteins were treated with deglycosylating agents as described for panel A. A significant pool of endo H-resistant post-ER protein was evident for wild-type mSmo. The bulk of mSmoM2 was endo H sensitive. The post-ER pool of mSmoM2 was modest but detectable (lane 6, post-ER label). Tubulin was used as the loading control. (B′) Indirect immunofluorescence of mSmoM2 in NIH 3T3 cells demonstrated that, whereas a pool of mSmoM2 (green) was detected in the primary cilium (ciliary slice [arrow]), the bulk of the protein colocalized with the ER-resident protein GRP94 (ER slice [red]). 4′,6-diamidino-2-phenylindole DAPI (blue) marks the nucleus. Bar, 20 μm. (C and C′) Active mSmo mutants are temperature sensitive. (C) NIH 3T3 cells were grown at 37°C for ∼44 h and then maintained at 37°C or shifted to 40°C for an additional 4 h prior to lysis, as indicated. Induction of the ER stress sensor CHOP was assessed by Western blotting of whole-cell lysates. Tubulin served as a loading control. (C′) NIH 3T3 cells expressing wild-type, C318A, or M2 mSmo proteins were cultured as for panel C. Whereas ER (white arrowhead) and post-ER (black arrowhead) forms of wild type Smo were not significantly affected by temperature shift, both of the mutants were destabilized at the high temperature. Destabilization of the ER-resident forms of wild type and both mutants at 40°C was attenuated by treating cells with the proteasome inhibitor MG132, suggesting that they are cleared by ERAD. The post-ER form of wild type Smo was unaffected by MG132 treatment. Tubulin was the loading control (con). (D and D′). ERAD attenuation stabilizes mutant mSmo proteins. NIH 3T3 cells expressing wild-type, C318A, or M2 mSmo proteins were transfected with control or Hrd1 siRNA as indicated. Cells were shifted to 40°C for 4 h prior to lysis. Hrd1 knockdown stabilized ER-retained mutant mSmo proteins (D, lanes 1 to 6, compared to lanes 7 to 12) and rescued mSmoM2-mediated induction of endogenous Gli1 (D′, lanes 1 and 2 compared to lane 4). Vec (D′, lane 5) is the empty vector control lysate. Tubulin was the loading control.

Fig 6

Small-molecule UPR modulators target oncogenic Smo. (A) Active mSmo mutants are destabilized by thapsigargin. NIH 3T3 cells expressing wild-type, C318A, or M2 mSmo proteins were treated with 1 μM thapsigargin (Thaps, +) or vehicle control (−), as indicated for 4 h prior to lysis. Western blotting of whole-cell lysates revealed C318A and M2 Smo proteins were destabilized in response to drug treatment. Wild-type mSmo was not significantly affected. Tubulin was the loading control. (B and C). Thapsigargin attenuates mSmoM2-induced pathway activity. RNA was harvested from NIH 3T3 cells expressing either wild-type or M2 mSmo proteins. qPCR analysis revealed that thapsigargin treatment resulted in specific attenuation of mSmoM2-induced gli1 transcript (B) and protein levels (C). smo expression increased modestly in response to drug treatment (B′). For qPCR analysis, expression is shown as the fold induction over the wild-type mSmo vehicle control. Expression was normalized to the GAPDH reference gene. Error bars indicate standard errors of the means. For protein analysis (C), tubulin served as the loading control, and results with CHOP indicate ER stress induction. (D) UPR modulators destabilize mSmoM2. NIH 3T3 cells expressing mSmoM2 protein were treated with HSP90 inhibitors 17-AAG (50 μM and 100 μM) and SNX-2112 (25 μM and 50 μM) and the proteasome inhibitor bortezomib (25 μM and 50 μM) for 4 h prior to lysis. DMSO was the vehicle control. Western blotting of whole-cell lysates revealed mSmoM2 protein to be destabilized in response to HSP-90 inhibitors but not in response to bortezomib. Tubulin was the loading control. CHOP results indicate ER stress.

Fig 7

Thapsigargin attenuates ectopic signaling by an active Smo mutant in vivo. (A) Transgene expression is unaffected by thapsigargin. MS1096-GAL4, UAS-EGFP (MS1096>EGFP) larvae were grown on media containing vehicle or thapsigargin, as indicated. Wing imaginal discs from third-instar larvae under both conditions demonstrated comparable GFP expression (green). 4′,6-Diamidino-2-phenylindole (DAPI; magenta) marks the nuclei. (B to E) Thapsigargin prevents Myc-SmoC320A-induced Hh gain-of-function wing phenotypes. Larvae expressing wild-type or C320A UAS-myc-smo under the control of MS1096-Gal4 or C765-Gal4 were grown at 22°C on food containing vehicle (C to E) or thapsigargin (C′ to E′). Representative wings from adult flies are shown. The MS1096-Gal4 driver wing served as a control (B). Thapsigargin did not affect wings expressing wild-type Smo (C and C′). The C320A-induced phenotype was significantly attenuated by drug, allowing for development of near-normal adult wings (D′and E′, compared to D and E). (F and G) SmoC320A-induced downstream pathway activity is attenuated by thapsigargin. Wing imaginal discs were dissected from myc-smoC320A-expressing third-instar larvae grown on vehicle-containing (F and G) or thapsigargin-containing (G′) food at 22°C. UAS-myc-smoC320A was expressed by using the dorsal compartment wing driver Apterous (Ap)-Gal4. Wing discs were immunostained for Myc-SmoC320A (anti-Smo; blue), full-length Ci (green), or β-galactosidase (dpp-lacZ; magenta). Note the significantly reduced SmoC320A protein level, reduced Ci stabilization, and decreased β-galactosidase signals in thapsigargin-treated discs (G compared to G′; wing pouch is shown). The Ap-Gal4 driver wing disc (F) served as a control. For all wing disc images, discs are shown with dorsal side up and the anterior to the left. Bar, 100 μm. (H and H′) Thapsigargin does not attenuate signaling induced by aberrant Hh ligand production. Hh^mrt larvae were grown at 22°C on food containing vehicle (H) or thapsigargin (H′). Thapsigargin did not alter the gain-of-function phenotype induced by the Hh^mrt allele. Representative wings from adult flies are shown.