Salvador–Warts–Hippo pathway regulates sensory organ development via caspase-dependent nonapoptotic signaling

The fundamental roles for the Salvador–Warts–Hippo (SWH) pathway are widely characterized in growth regulation and organ size control. However, the function of SWH pathway is less known in cell fate determination. Here we uncover a novel role of the SWH signaling pathway in determination of cell fate during neural precursor (sensory organ precursor, SOP) development. Inactivation of the SWH pathway in SOP of the wing imaginal discs affects caspase-dependent bristle patterning in an apoptosis-independent process. Such nonapoptotic functions of caspases have been implicated in inflammation, proliferation, cellular remodeling, and cell fate determination. Our data indicate an effect on the Wingless (Wg)/Wnt pathway. Previously, caspases were proposed to cleave and activate a negative regulator of Wg/Wnt signaling, Shaggy (Sgg)/GSK3β. Surprisingly, we found that a noncleavable form of Sgg encoded from the endogenous locus after CRISPR-Cas9 modification supported almost normal bristle patterning, indicating that Sgg might not be the main target of the caspase-dependent nonapoptotic process. Collectively, our results outline a new function of SWH signaling that crosstalks to caspase-dependent nonapoptotic signaling and Wg/Wnt signaling in neural precursor development, which might be implicated in neuronal pathogenesis.


Introduction
The Salvador-Warts-Hippo (SWH) pathway has been recognized as a significant regulator for growth control, tissue regeneration, and stem cell pluripotency 1,2 . It has also been found to play important roles in cancer metastasis [3][4][5] . Originally identified as a prominent regulator of organ size in Drosophila, this pathway is highly conserved from fly to mammals. The regulation of SWH pathway depends on its response to various upstream stimuli through intercellular junctions, including adhesion cues through cell-cell contact, polarity, extracellular signal, mechanical signals, and cellular stress. Core components of the SWH pathway comprise a kinase cascade, which is the main regulation modulating SWH signaling. The Ste20 family kinase Hpo (MST1/2 in mammals) forms a heterodimer with the adapter protein Sav (SAV or WW45 in mammals), thereby promoting their interaction with the serine/threonine kinase Wts (LATS1/2 in mammals). Hpo subsequently activates Wts activity via phosphorylation 6,7 . The activated Wts kinase operates with its cofactor Mats (MOB1A/B in mammals) to phosphorylate Yorkie (Yki; YAP and TAZ in vertebrates), which is a transcriptional coactivator and serves as the final effector of the Hippo signaling. Yki/YAP/TAZ lacks DNA-binding motif but binds to the promoters of target genes through interacting with Scalloped (Sd, TEAD1-4 in mammals) or other transcription factors 6 . By activating target gene expression, Yki/YAP/TAZ plays important roles in controlling cell growth, proliferation, and survival. Wts kinase inhibits transcriptional activity of Yki/YAP/TAZ through nuclear export, cytoplasmic retention, and protein degradation [8][9][10] . Phosphorylation-independent regulations also exist. Yki can directly bind to Hpo, Wts and the FERM-domain containing adapter protein expanded (Ex), and sequesters Yki in the cytoplasm 11,12 . In addition to intrinsically regulates Yki activity, Ex, as an apical junctions-localized protein, also transduces signaling cues through binding to the apical membrane protein Crumbs (Crb, CRB3 in mammals) [13][14][15][16] . In addition, ex is a downstream target gene of Yki, thereby forming a feedback regulatory loop of Hippo pathway 17 . It has been demonstrated that activation of the SWH pathway through elevating expanded (ex) levels is required to eliminate the inappropriately differentiating neurons during development 18,19 . However, whether SWH pathway has any roles in normal neurogenesis remained unclear.
Intriguingly, hypomorphic ex mutants often differentiate supernumerary sensory bristles 18 . Bristles are a component of the Drosophila peripheral nervous system and can be divided into macro-(large bristles) and microchaetae (small bristles) according to their size and position. Drosophila notum is a classical model to study pattern formation because each macrochaetae develop in precise positions and microchaetae appears in a characteristic density pattern 20 . Each of these external sensory organs comprises five cells (hair, socket, neuron, sheath cell, and glial cell) that are generated through asymmetric cell divisions of single sensory organ precursor (SOP) cell 21,22 . The accuracy of bristle patterns on the adult body depends on the correct SOP cell positioning. The phenotype of ex mutations promoted us to study in depth how ex mediates sensory organ development.
Caspase activation has been implicated in SOP development through a caspase-dependent nonapoptotic machinery. This caspase-dependent machinery is thought to be required for cleavage and activation of a negative regulator of Wingless (Wg)/Wnt signaling, Shaggy (Sgg)/GSK3β, in SOP cell formation 23 . By studying how ex takes part in SOP development, we discovered a crosstalk between SWH pathway and caspase-dependent nonapoptotic signaling mediated through Wg pathway. Interestingly and unexpectedly, we found Sgg might not be the main target of the caspase-dependent nonapoptotic event.

Immunohistochemistry and histology
Preparation of wing discs for immunostaining and adult notum for light microscope was performed as described previously 30 . Confocal imaging was performed using Leica SP2, SP8 and Zeiss LSM 880 microscopy. Primary antibodies used were anti-Sens (guinea pig, a gift from H. Bellen); GFP (rat, NACALAI TESQUE# GF090R). Photograph of adult notum was carried out using Leica MZFLIII microscope and Nikon SMZ1500 microscope.
CRISPR/Cas9-based genome editing of sgg gene By using CRISPR/Cas9-mediated genome editing, mutagenesis of the corresponding genomic sequences in both 235th and 300th Asp residues of sgg-RD/RP/RQ isoforms were conducted in w 1118 flies. Two single guide RNAs (sgRNAs) were used to introduce double strand breaks near by the edited genomic region and followed by homology-directed repair (HDR). The HDR donor plasmid was designed to harbor a DNA cassette containing the upstream homology arm of sgg, 3XP3-ScarlessDsRed flanking with PiggyBac terminal repeats, and the downstream homology arm of sgg with D235G/D300G mutations, which was constructed into the pUC57-Kan vector. The sgRNA and HDR donor plasmids used for microinjection were purified using the Plasmid Midi-prep kit (Qiagene). After validation of the CRISPR-knockin sgg alleles by genomic PCR coupled with Sanger sequencing, the ScarlessDsRed selection marker was then excised by PiggyBac transposon. The genomic PCR coupled with Sanger sequencing was performed to confirm the precise excision of ScarlessDsRed.

Results
Ex is required to suppress extra macrochaete in the scutellum Reduced ex function in Drosophila by using transheterozygous ex mutants caused the appearance of ectopic macrochaete on the notum (Fig. 1b, c). Knockdown of ex in the scutellum, using the dpp-GAL4 driver, also resulted in the formation of extra macrochaete in 62.5% of flies (Fig. 1d). Compared with normal macrochaete, the extra macrochaete observed in ex mutants were occasionally thinner and shorter, but still contained socket cells of normal morphology (Fig. 1d'). These hypomorphic ex genotypes survived to adulthood without obvious growth defects in the scutellum ( Supplementary Fig. 1). To address whether the extra macrochaetae were produced from extra SOP cells, the SOP cells were visualized by Senseless (Sens) staining. Normally, two sets of SOPs (one anterior scutellar (aSC) and one posterior scutellar (pSC) bristles, respectively) exist on the scutellum of one wing imaginal disc, whereas more than two SOP cells were detected when ex was downregulated (Fig. 2). These results indicate the extra macrochaetae of ex mutants are derived from extra SOP cells, not caused by a defect in bristle differentiation or SOP asymmetric division.

Extra macrochaete formation requires Diap1 activity in an ex-dependent manner
Since Ex functions upstream of the core kinase cassette to regulate SWH pathway activation 17 , mutations of ex led to inhibition of SWH signaling pathway and activation of Yki activity. Consistent with loss-of-function phenotype of ex, ectopic macrochaetae often appeared on the scutellum of flies overexpressing wild-type Yki or the activated Yki S168A (Fig. 3b, c and Supplementary Fig. 2). These observations suggest that Yki activity is sufficient to induce extra bristle formation on scutellum. Moreover, overexpression of a Yki target gene, Diap1, caused extra scutellar bristles under the control of different GAL4 drivers (Fig. 3d, Supplementary Fig. 2 and 23 ). These data suggest that the SWH pathway is involved in ex-dependent bristle inhibition through modulating Diap1 activity. SWH activity was monitored by using ex intron 3 enhancer (referred to Ex Intron -GFP) reporter 18,19 , and also fj-lacZ, both of which report Yki activity. It was difficult to see overall changes in the notum region, although both reporters were present in the ectopic aSC SOP cell (Fig.  4). Collectively, our findings indicate that Yki activation contributes to ectopic macrochaetae formation.
To determine the involvement of Diap1 in the ex mutant phenotype, Diap1 levels were manipulated in ex mutants. As expected, lowered Diap1 levels rescued the extra bristle phenotype of ex mutants while Diap1 transheterozygous mutants have normal patterning and bristle numbers on the notum (Fig. 3f-i). These data indicate that Diap1 is required for extra bristle formation in the absence of ex. The well-characterized function of Diap1 is its role as a caspase inhibitor. To determine the role of caspase activity during bristle determination, caspase activity was blocked by expressing the antiapoptotic proteins baculovirus p35 or dominant-negative Dronc under the control of dpp-GAL4 ( Fig. 3e and data not shown). Indeed, blockage of caspase activity led to ectopic bristle formation, which is consistent with previous reports 23 . Since many studies have shown that caspase activation negatively regulates macrochaetae development 23,31-33 , the involvement of Diap1 in bristle formation is likely to be through inhibition of caspase activity. Such caspase-dependent macrochaetae regulation has been shown to represent an apoptosis-independent   23,34 . Like previous authors, we also found no evidence for apoptosis in normal SOP patterning (data not shown).
Wg signaling modulates ex-dependent bristle phenotypes SOP cells of the macrochaetes arise from proneural clusters in the wing imaginal disc during larval stage. Spatial and temporal patterning of the proneural clusters are established by the expression of proneural genes achaete and scute, which is controlled by multiple cisregulatory elements distributed throughout the achaete and scute transcription units to permit the precision of SOP cell specification [35][36][37] . Among the critical regulators in modulating proneural gene expression, previous studies have identified Wg/Wnt signaling in particular as affected by nonapoptotic caspase activity during thoracic bristle patterning 23 . Sgg/GSK3β is a negative regulator of Wg/ Wnt signal transduction. ex and sgg mutations shared similar phenotypes in bristle formation (this study; 23,38 ). Independent evidence also points out that Wg signaling is altered in the absence of ex in eye discs 39 . Genetic analyses were performed to verify the involvement of Wg signaling cascade in ex-mediated SOP specification. Indeed, the extra bristle phenotype caused by downregulating ex was suppressed by removing one copy of Wg signaling components (Fig. 5a-c, e). Similar results were also found when dTcf was knocked down in dpp > ex RNAi flies (Fig. 5d). Overexpression of dominant-negative dTCF (dTCF[DN]) alone prevented macrochaetae formation (yellow bar in Fig. 5f-h). The penetrance of scutellar bristle loss was comparable in ex-knockdown flies with overexpressing dTCF[DN] transgene (magenta bar in Fig. 5f). Although the Notch pathway has been shown to play important role in bristle development 40 , E(spl) expression was not changed when ex was mutated (Supplementary Fig. 3), consistent with previous conclusions that ex mutations do not reduce Notch signaling 41 . These results suggest that Wg signaling pathway acts downstream of ex in SOP cell formation, and therefore that the SWH pathway may by the source of the nonapoptotic caspase activity that acts on Wg signaling.
A caspase-dependent cleavage resulting from caspase activation mediates nonapoptotic signaling in determining SOP cells. This caspase-dependent cleavage is thought to activate Sgg/GSK3β during SOP cell formation 23 . Previous studies have reported that caspasedependent cleavage occurs at the DEVD motif, which has been mapped to DEVD 235 and DEVD 300 of Sgg (Sgg46 isoform) protein 23,42 . Hence, we hypothesize that the caspase inhibitor Diap1 might be the critical effector connecting SWH pathway with Wg pathway through modulating kinase activity of Sgg to determine the correct number of SOP cells. To probe the involvement of Sgg in this regulation in depth, we generated an in vivo noncleavable form of Sgg (sgg D235G/D300G ) using CRISPR-Cas9 technique (thereafter referred to as CRISPR-sgg D235G/D300G , Fig. 6a). The sequence validation of CRISPR-sgg D235G/D300G was performed in both genomic DNA and cDNA ( Fig. 6b and Supplementary  Fig. 4a). If caspase cleavage of DEVD 235 or DEVD 300 was required to activate Sgg and inhibit Wg signaling during normal bristle development, then sgg D235G/D300G flies should have elevated Wg signaling and extra bristles, but extra macrochaetae were observed at only a low frequency and only in the first few generations. In case there might be selection for genetic modifiers suppressing the phenotype, we  Figure 3 (Fig 3_Sept07) with the correct Figure  3a in the attachment (and also sent you this updated file by email). Please make correction with the updated Figure 3.
selected sgg D235G/D300G flies where proximal parts of the X chromosome were replaced with those from the FRT19 strain after meiotic recombination, and completely exchanged the autosomes by outcrossing at every generation. These flies, which should only be able to retain genetic modifiers on one section of the X chromosome, exhibited only three extra thoracic macrochaetae in 878 hemizygous sgg D235G/D300G males (Fig. 6c), a much lower frequency than observed when caspase activity was inhibited (Fig. 3d). Penetrance was similarly low in transheterozygous sgg 1 /CRISPR-sgg D235G/D300G females (Supplementary Fig. 4b). When ex was knocked down in the CRISPR-sgg D235G/D300G background, the frequency of extra scutellar macrochaetae was not affected by the sgg mutant background (Supplementary Fig. 4c, d). These observations suggest that Sgg is not the major substrate of caspase-dependent cleavage that affects numbers of macrochaetae in the scutellum. Intriguingly, overexpression of UAS-sgg D235G/D300G using Sca-Gal4 resulted in ectopic macrochaete in 19.3% of flies 23 . This much higher frequency than observed in CRISPR-sgg D235G/D300G flies suggests that overexpression of UAS-sgg D235G/D300G might have dominant negative effects. In sum, our data indicate that there is another target of caspases that affects Wg signaling. Our data cannot rule out some contribution of Sgg cleavage that is redundant with the other target(s).

Discussion
Unlike the well-known concept that the key roles for SWH pathway are in the regulation of cell proliferation and organ size, the present study reveals a novel function of SWH signaling in cell fate determination through nonapoptotic caspase signaling. As is wellknown, deregulation of SWH pathway leads to the activation of the Yki target gene, Diap1, which restrains caspase activity. Although this can regulate cell survival, here we report that SWH and Yki play a role in normal development in suppressing nonapoptotic caspase activity. In the Drosophila thorax, nonapoptotic caspase activity is needed to suppress activity of Wg signaling and restrain SOP cell specification 23 . This is here shown to depend on the SWH pathway, without which Diap1 expression is too high to permit normal patterning. While this paper was under review, another study reported that nonapoptotic caspase activity is also regulated by the SWH pathway, during tracheal development 43 . During SOP specification, caspase activity is transiently controlled by the turnover of Diap1. Diap1 degradation is triggered by the Drosophila IKK-related kinase (DmIKKε)dependent phosphorylation. In consistent with phenotype of high levels of Diap1, downregulation of DmIKKε led to extra macrochaetae formation 44 . Hence, the levels of Diap1 play determinant role in cell fate specification. Here we provide evidence that transcriptional regulation of Diap1 by Yki and the SWH pathway is also important for SOP cell determination, which is disrupted by hypomorphic mutations of ex, which affect the level of Yki activity.
The Drosophila GSK3β ortholog, Sgg, was identified as a potential substrate for caspase-dependent cleavage. One isoform, Sgg46, is inactive but can be cleaved into the active isoform Sgg10, which negatively regulates Wg signaling through the phosphorylation and degradation of Arm 23 , and also directly phosphorylate Scute and its activator Pannier in SOP cell specification 45 . Caspases potentially have hundreds of substrates, but Sgg46 was believed to be significant for Wg signaling and SOP patterning because overexpression of a form with mutated caspase sites, SggD235G/D300G, phenocopied blockade of nonapoptotic caspases by p35 overexpression 23 . In fact, SggD235G/D300G overexpression was quantitatively less effective than completely blocking caspases with p35, but this was attributed to the simultaneous presence of wildtype Sgg46 encoded by the endogenous locus 23 . It was presumed that the overexpressed, SggD235G/D300G protein behaved as a competitive inhibitor of Sgg46 cleavage. If this model was correct, we would expect that modifying the endogenous sgg locus to encode Sgg D235G/D300G (which would not affect other, shorter isoforms of Sgg, Fig. 6a), should more completely prevent cleavage of Sgg46 and more completely block the nonapoptotic caspase regulation of Wg signaling and SOP patterning, resulting in many extra macrochaetae, comparable with ectopic expression of Diap1 or p35. In b Sanger sequencing of CRISPR-sgg D235G/D300G genomic DNA. Note that the corresponding sequences of DEVD235th and DEVD300th residues are mutated from GAT to GGT (glycine). c Bristle patterning of CRISPR-sgg D235G/D300G flies. Note the ectopic scutellar bristles (indicated by red arrow) are found in a very low penetrance. d Model of integrated pathways in regulating SOP specification. Components involving in signal transduction of SWH and Wg pathways are indicated in blue and purple, respectively. The crosstalk between these two pathways is nonapoptotic caspase activity contrast to this expectation, we found almost no phenotypic effect of the endogenous sgg D235G/D300G mutant. This is not consistent with the model that Sgg46 is the main target of nonapoptotic caspase signaling in bristle patterning. Although we cannot exclude that Sgg46 might be activated by cleavage at another site, this would not explain why Sgg46 could not be activated when SggD235G/D300G protein was overexpressed. Therefore, we conclude that nonapoptotic caspases regulate one or more other substrates that are critical for Wg signaling and SOP patterning, and that SggD235G/D300G overexpression is a competitive inhibitor of cleavage of these other substrates.
A recent study has reported that the unconventional myosin Crinkled acts as an adapter to facilitate Sgg46 cleavage and activation by Dronc 46 . Aside from our finding that the caspase cleavage sites of Sgg46 are largely dispensable for bristle patterning, the model that Sgg46 is a Dronc target also does not fit with the observation that bristle patterning is disrupted by p35 overexpression 23 , since p35 does not inhibit Dronc. The ectopic p35 phenotype strongly suggests that the major regulators of bristle patterning are substrates of p35dependent effector caspases, not direct Dronc targets. These data could explain how SggD235G/D300G overexpression is dominant negative, however, if SggD235G/ D300G inhibits Dronc and Crinkled function, leading to deficient nonapoptotic signaling by downstream effector caspases.
Our main conclusion is that, in addition to its previously known roles, the SWH pathway is important for regulating nonapoptotic caspase signaling, presumably through Yki control of Diap1 transcription. In the Drosophila thorax this is required to restrain Wg signaling and bristle patterning (Fig. 6d). It is possible that this nonapoptotic caspase signaling might underlie the crosstalk between SWH and Wg signaling in other tissues, such as the eye and wing, the molecular basis of which has so far remained unclear. At one time it was thought that wg was a transcriptional target of Yki but it is now thought this reflects enhanced wg autoregulation when Wg signaling is elevated 39,47 . Crosstalk between SWH and Wnt signaling appears to be conserved, having also been reported in vertebrates [48][49][50][51][52][53][54] . The mechanisms that have been suggested in vertebrates are not yet known to involve nonapoptotic caspase signaling, however.
It is striking that in both Drosophila and in mammals, SWH signaling and nonapoptotic caspase signaling both are implicated in neuronal morphogenesis. SWH signaling affects synapse development and dendrite morphogenesis [55][56][57][58][59] , while nonapoptotic roles of caspases remodel Drosophila dendritic arborization neurons and regulate axon degeneration in mammals. Defects of caspase-dependent nonapoptotic signaling affect plasticity and result in disease such as Alzheimer's disease 60,61 . Wnt signaling is also involved in neuronal development and has been associated with neurological diseases including Alzheimer's disease, Parkinson's disease, schizophrenia, and autism [62][63][64][65][66] . Crosstalk between SWH signaling, Wnt signaling and caspase-dependent nonapoptotic signaling may contribute to the molecular mechanisms of neuronal pathogenesis. This crosstalk may occur in multiple processes during development, in light of the finding that DrICE has a nonapoptotic function, which acts downstream of the SWH signaling to regulate endocytic trafficking during tracheal morphogenesis 43 .