Casein kinase 1 family proteins promote Slimb-dependent Expanded degradation

Hippo signalling integrates diverse stimuli related to epithelial architecture to regulate tissue growth and cell fate decisions. The Hippo kinase cascade represses the growth-promoting transcription co-activator Yorkie. The FERM protein Expanded is one of the main upstream Hippo signalling regulators in Drosophila as it promotes Hippo kinase signalling and directly inhibits Yorkie. To fulfil its function, Expanded is recruited to the plasma membrane by the polarity protein Crumbs. However, Crumbs-mediated recruitment also promotes Expanded turnover via a phosphodegron-mediated interaction with a Slimb/β-TrCP SCF E3 ligase complex. Here, we show that the Casein Kinase 1 (CKI) family is required for Expanded phosphorylation. CKI expression promotes Expanded phosphorylation and interaction with Slimb/β-TrCP. Conversely, CKI depletion in S2 cells impairs Expanded degradation downstream of Crumbs. In wing imaginal discs, CKI loss leads to elevated Expanded and Crumbs levels. Thus, phospho-dependent Expanded turnover ensures a tight coupling of Hippo pathway activity to epithelial architecture.


Introduction
The maintenance of epithelial tissue architecture through cell-cell and cell-extracellular matrix contacts, as well as cell polarity, is essential for organ function and size control (Genevet and Tapon, 2011;Low et al., 2014). The evolutionarily conserved Hippo (Hpo) pathway, a key signalling module that senses and responds to epithelial organisation, has emerged as a critical regulator of growth and epithelial integrity (Genevet and Tapon, 2011;Schroeder and Halder, 2012;Yu and Guan, 2013). At the core of Hpo signalling is a kinase cascade comprising Hpo and Warts (Wts), which promote the phosphorylation and inactivation of the pro-growth transcriptional co-activator Yorkie (Yki, YAP in mammals), thereby repressing tissue growth (Yu and Guan, 2013;Fulford et al., 2018;Hong and Guan, 2012). When Hpo signalling is inactive, Yki/YAP is able to enter the nucleus, associate with its transcription factor partner Scalloped (TEAD1-4 in mammals) and promote the expression of cell cycle regulators and apoptosis inhibitors, among others (Genevet and Tapon, 2011;Yu and Guan, 2013;Hong and Guan, 2012). To ensure that epithelial homeostasis is maintained, Yki/YAP also control the expression of Hpo pathway upstream regulators that dampen Yki/YAP activity as part of a negative feedback mechanism (Genevet and Tapon, 2011;Halder and Johnson, 2011).
Yki/YAP activity responds to epithelial organisation through the actin cytoskeleton (Gaspar and Tapon, 2014), basolateral polarity determinants (Chen et al., 2012;Cordenonsi et al., 2011;Grzeschik et al., 2010), adherens junction components, such as a-catenin (Schlegelmilch et al., 2011), and apical polarity proteins, such as Crumbs (Crb) (Grzeschik et al., 2010;Chen et al., 2010;Hafezi et al., 2012;Ling et al., 2010;Robinson et al., 2010;Varelas et al., 2010). While the vast majority of these inputs act through Hpo and/or Wts, some engage core kinase cascade-independent signalling (Genevet and Tapon, 2011;Yu and Guan, 2013). The FERM domain protein Expanded (Ex) is a key regulator of Yki function , which inhibits tissue growth using both Hpo-dependent and -independent mechanisms, in a manner analogous to mammalian Angiomotins (Moleirinho et al., 2014). Ex is an upstream activator of the Hpo core kinase cassette that forms a complex with Kibra and Merlin (Baumgartner et al., 2010;Genevet et al., 2010;McCartney et al., 2000;Yu et al., 2010). Ex has been proposed to promote core kinase activity by bridging Hpo association with the upstream kinase Tao-1 together with Schip1 (Genevet and Tapon, 2011;Chung et al., 2016), and by recruiting Wts to the apical membrane where it can be activated by Hpo (Sun et al., 2015). Ex also restrains tissue growth in a phosphorylation-independent manner by tethering Yki at the apical membrane via a direct PPxY:WW domainmediated interaction (Badouel et al., 2009;Oh et al., 2009), a process regulated by Ack-dependent tyrosine phosphorylation of Ex (Hu et al., 2016). Moreover, Ex has also been linked with F-actinmediated regulation of Yki function, by antagonising the action of Zyxin . Interestingly, ex is a Yki target gene and, therefore, it is a prime candidate to mediate the Hpo pathway feedback regulation that controls tissue homeostasis (Genevet and Tapon, 2011;Halder and Johnson, 2011).
Ex is a key link between epithelial polarity and Hpo signalling. This is controlled by the transmembrane protein Crb, which in addition to its recognised role in polarity, also regulates tissue growth by modulating the Notch and Hpo pathways (Grzeschik et al., 2010;Chen et al., 2010;Hafezi et al., 2012;Ling et al., 2010;Robinson et al., 2010;Richardson and Pichaud, 2010). Via its FERM-binding motif (FBM), Crb recruits Ex to the apical membrane, where it can promote inhibition of Yki (Genevet and Tapon, 2011;Chen et al., 2010;Hafezi et al., 2012;Ling et al., 2010;Robinson et al., 2010;Su et al., 2017). However, besides activating Hpo signalling through Ex, Crb also stimulates Ex phosphorylation and turnover of Ex protein (Genevet and Tapon, 2011;Grzeschik et al., 2010;Chen et al., 2010;Ling et al., 2010;Robinson et al., 2010;Ribeiro et al., 2014;Laprise, 2011). We have previously shown that ubiquitylation and degradation of Ex downstream of Crb is mediated by the phospho-dependent SCF Slimb/b-TrCP E3 ligase complex, which is also thought to regulate Ex levels independently of Crb (Ribeiro et al., 2014;Zhang et al., 2015). However, the identity of the kinase(s) that promote Ex degradation downstream of Crb is currently unknown.
Here, we identify the Casein Kinase 1 (CKI) family of protein kinases as regulators of Ex stability that function downstream of the polarity protein Crb. Depletion of CKI kinases suppresses Crbinduced Ex phosphorylation, ubiquitylation and degradation. Interestingly, CKI kinases regulate Ex in a partially redundant manner, which suggests that regulation of Ex stability is a key step in the regulation of Yki function and the maintenance of tissue homeostasis.

Crb promotes interaction of Ex with Casein kinase 1 family proteins
We have previously shown that Crb regulates Ex protein stability in a b-TrCP-dependent manner (Ribeiro et al., 2014). The Ex:Slimb (Slmb, Drosophila b-TrCP) interaction is mediated by a b-TrCP consensus sequence immediately following the Ex N-terminal FERM domain ( 452 TSGIVS 457 ). In agreement with the fact that b-TrCP targets substrates for ubiquitylation and degradation through the recognition of a phosphodegron, Ex is phosphorylated in Drosophila S2 cells in the presence of ectopic Crb (full-length or the intracellular domain, Crb intra ) (Ling et al., 2010;Robinson et al., 2010;Ribeiro et al., 2014). However, the kinase(s) involved in Ex degradation downstream of Crb are currently unknown. In our previous report, we used affinity purification coupled with mass spectrometry (AP-MS) to identify Slmb as an Ex interacting protein (Ribeiro et al., 2014). Upon re-analysis of our AP-MS data, we observed that Gilgamesh (Gish), the Drosophila orthologue of Casein kinase 1g, was specifically purified by an Ex truncation that fully recapitulates the Crb-mediated effect on Ex stability (Ex 1-468 ) ( Figure 1A) (Ribeiro et al., 2014). Importantly, Gish peptides were detected in the Ex 1-468 AP-MS only upon co-expression with wild-type (wt) Crb but not with a FERMbinding motif mutant version of Crb (DFBM) that cannot bind Ex or promote its depletion (B) and (C) Crb intra promotes Ex:Gish binding in a FBM-dependent manner. Reciprocal co-IPs were performed with FLAG-tagged NTAN or Ex 1-468 and HA-tagged Gish (B), or with FLAG-tagged NTAN or Gish isoI and V5-tagged Ex 1-468 CAAX (C), in the presence of Myc-tagged GFP, Crb intra or Crb DFBM . The expression and presence of copurified proteins were analysed by immunoblotting with the indicated antibodies. Asterisks denote non-specific Figure 1 continued on next page ( Figure 1A) (Ling et al., 2010;Robinson et al., 2010;Ribeiro et al., 2014). In contrast, Hpo interacted with Ex regardless of Crb presence, in agreement with previous reports ( Figure 1A) .
We validated our AP-MS data by performing co-immunoprecipitation (co-IP) analyses in S2 cell lysates. In reciprocal co-IP experiments, an interaction between Gish and Ex 1-468 was readily detected in S2 cells, specifically in the presence of wt Crb ( Figure 1B-C). The Casein kinase 1 (CKI) protein family includes several members (Jiang, 2017), with high homology in their kinase domains (Figure 1-figure supplement 1A-C), which are thought to, at least in part, share consensus sequences and targets (Knippschild et al., 2014;Schittek and Sinnberg, 2014;Venerando et al., 2014). The Drosophila gish (CKIg) locus produces multiple protein isoforms, one of which, Gish isoF , lacks a conserved C-terminal palmitoylation motif (Figure 1-figure supplement 1B,D), which would be predicted to affect its localisation (Davidson et al., 2005;Li et al., 2016). Therefore, we used two different isoforms of Gish that either contain (Gish isoI ) or lack (Gish isoF ) this palmitoylation sequence. Interestingly, only Gish isoI , but not Gish isoF , was able to interact with Ex ( Figure 1B) suggesting that CKI sub-cellular localisation may be critical to the regulation of Ex.

CKI kinases promote Ex phosphorylation and degradation in vitro
The CKI family commonly targets substrates for proteasomal degradation (Knippschild et al., 2014;Schittek and Sinnberg, 2014), prompting us to test if Gish could promote Ex degradation. To this end, we analysed Ex electrophoretic mobility shift and protein levels (indicative of phosphorylation and degradation, respectively) in S2 cells upon expression of different CKI kinases. Co-expression of Ex 1-468 and Gish isoI in the absence of Crb resulted in a mild depletion of Ex 1-468 protein levels ( Figure 2A), whereas Gish isoF had no effect. We reasoned that the mild effect of Gish isoI on Ex stability might be due to relatively low levels of Ex 1-468 reaching the membrane in the absence of Crb, which is not endogenously expressed in S2 cells. To mimic the apical membrane localisation of endogenous Ex, we generated an Ex variant containing a C-terminal CAAX sequence (Sotillos et al., 2004), which targets Ex to cellular membranes (Ex 1-468 CAAX ). Importantly, this variant remains responsive to Crb ( Figure 2B), and co-immunoprecipitates with Gish in a Crb-dependent manner ( Figure 1C). Ex 1-468 CAAX ran mostly as a single band in S2 cells in the presence of the Crb DFBM mutant and, similar to wt Ex 1-468 , as a higher mobility band or doublet in the presence of wt Crb (Figure 2A-C). An electrophoretic mobility shift and depletion was also observed when Ex 1-468 CAAX was co-expressed with Gish isoI , but not with a kinase-dead version of Gish (Gish KD ) ( Figure 2B and Figure 2-figure supplement 1A). Next, we tested whether two other CKI family members, CkIa and Dco (Drosophila CkId/e orthologue), had a similar effect on Ex and whether this effect was dependent on their kinase activity. Indeed, co-expression of Ex 1-468 CAAX or full-length (FL) Ex with wt Gish isoI , CkIa or Dco led to Ex mobility shift and depletion, although their kinase-deficient (KD) versions did not ( Figure 2C and Figure 2-figure supplement 1A-B). Together, these results suggest that multiple CKI family kinases can promote Ex phosphorylation and depletion.
The fact that Ex and Gish interact, and that CKI kinases stimulate Ex degradation in a kinasedependent manner led us to examine if the CKIs act in the same pathway as Crb and Slmb. To test this, we co-expressed CkIa, Dco or Gish isoI in S2 cells with either wt Ex 1-468 CAAX or a variant carrying a mutation at S453 (Ex 1-468 S453A CAAX ), one of the crucial residues that mediates the interaction between Ex and Slmb (Ribeiro et al., 2014). All three kinases readily promoted the degradation of wt Ex 1-468 CAAX ( Figure 2D). However, the S453A mutant levels were largely unaffected by CKI isoforms, consistent with the notion that the CKI-mediated regulation of Ex stability is dependent on S2 cells were used to co-transfect V5-tagged Ex 1-468 with GFP and Crb intra , Crb DFBM , Gish isoI or Gish isoF , for 48 hr before lysis. Lysates were processed for immunoblot analysis using the indicated antibodies. Note that Gish isoI caused a more prominent depletion of Ex than Gish isoF , the CkIg isoform lacking a palmitoylation sequence. (B) FLAG-tagged Ex 1-468 CAAX was co-transfected with the indicated plasmids for 48 hr and lysates were processed for Western blotting analysis with the indicated antibodies. Kinase-deficient (KD) Gish isoI was unable to promote Ex 1-468 CAAX phosphorylation and depletion. (C) CKI kinases promote Ex phosphorylation and depletion in a kinase-dependent manner. S2 cells were transfected with the indicated plasmids for 48 hr before lysis. Immunoblot analysis with the indicated antibodies revealed that CkIa and Dco promote Ex phosphorylation and depletion in a kinase-dependent manner, as the kinase-dead (KD) versions did not cause a mobility shift in Ex 1-468 CAAX . (D) CKI kinases promote Ex degradation via S453. S2 cells were co-transfected with V5-tagged wt or S453A Ex 1-468 CAAX and HA-tagged CkIa, Dco or Gish isoI for 48 hr before lysis. Lysates were immunoblotted with the indicated antibodies. Unlike its wt counterpart, Ex 1-468 S453A CAAX was refractory to the action of CKI kinases and was not degraded in the presence of the kinases. (E) Gish expression promotes Ex:Slmb binding in the absence of Crb intra . Co-IPs were performed with FLAG-tagged NTAN or Slmb and either V5-tagged Ex 1-468 or Ex 1-468 CAAX , in the presence of Crb intra , Crb DFBM , Gish isoI or Gish isoF . Expression and presence of co-immunoprecipitated proteins was assessed by immunoblotting with the indicated antibodies. Note Figure 2 continued on next page the ability of Ex to interact with Slmb via the phosphodegron surrounding S453. Interestingly, expression of CkIa or Gish isoI resulted in the appearance of a slower-migrating Ex 1-468 S453A band ( Figure 2D). This suggests that, at least for CkIa and Gish, there may be alternative CKI phosphorylation sites in Ex 1-468 besides S453.
We also assessed whether CKI kinases could promote the Ex:Slmb interaction, which we have previously shown can be triggered by Crb over-expression (Ribeiro et al., 2014). We found that Gish isoI , but not Gish isoF , promoted the interaction between Ex 1-468 CAAX and Slmb in the absence of ectopic Crb ( Figure 2E), consistent with its ability to stimulate Ex mobility shift and depletion. CkIa and Dco were also able to induce binding between Ex 1-468 CAAX and Slmb in the absence of Crb (Figure 2figure supplement 1C). Dco consistently showed a weaker effect on Ex phosphorylation than the other two kinases. Our previous findings showed that the Crb-induced Ex:Slmb interaction relies on the region surrounding the b-TrCP consensus motif (aa 450-468 in Ex) (Ribeiro et al., 2014). Interestingly, an Ex truncation mutant lacking this domain (Ex 1-450 ) was refractory to degradation, and did not display a mobility shift following Gish isoI expression ( Figure 2-figure supplement 1D), suggesting that CKIs act upstream of Slmb to promote Ex phosphorylation and degradation.
b-TrCP substrates are often targeted for degradation by the sequential action of GSK3b and CKI kinases (e.g. b-catenin) (Gammons and Bienz, 2018). Thus, we tested if the Drosophila GSK3b orthologue Shaggy (Sgg) was involved in the regulation of Ex phosphorylation (Siegfried et al., 1992). However, we found that over-expression of Sgg did not promote Ex:Slmb binding (

CKI kinases promote Ex phosphorylation and degradation in vivo
Next, we wanted to validate our observations in developing Drosophila tissues. However, assessing Ex protein stability in vivo is complicated by the fact that ex is a Yki target gene  and, therefore, its steady state protein levels reflect not only direct posttranslational effects but also transcriptional inputs via the modulation of Hpo signalling. To directly study Ex stability in the absence of any confounding effects due to Yki-mediated transcriptional feedback, we generated transgenic flies carrying an Ex stability reporter construct consisting of Ex 1-468 fused to GFP, whose expression is controlled by the ubiquitin 63E promoter rather than its endogenous promoter (ubi-Ex 1-468 ::GFP). We also generated a mutant version that is refractory to Crb-induced degradation by mutating Ser 453, which mediates binding to Slmb (Ribeiro et al., 2014). As predicted, the ubi-Ex 1-468 ::GFP reporters resemble endogenous Ex in that they are normally localised at the apical cortex of wing imaginal disc cells ( Figure 3A (Chen et al., 2010;Hafezi et al., 2012;Ling et al., 2010;Robinson et al., 2010;Ribeiro et al., 2014). Loss of both crb and slmb led to accumulation of Ex 1-468 ::GFP in the cytoplasm ( Figure 3C-C'), suggesting that the Ex sensor can be degraded by that, similar to Crb intra , Gish isoI expression alone promoted the Ex:Slmb interaction. GFP and tubulin were used as transfection efficiency and loading controls, respectively. Asterisk denotes non-specific band (IgG heavy chain in IP panel). DOI: https://doi.org/10.7554/eLife.46592.004 The following source data and figure supplements are available for figure 2:   9H4-17 and crb 82-04 (C). Clones are marked by absence of RFP and highlighted by white dashed lines. The ubi-Ex 1-468 ::GFP reporter (green in A-C, grey in A'-C') is lost from the apical cortex of crb clones, accumulates apically in slmb clones, and accumulates in the cytoplasm of slmb, crb clones. DAPI staining (blue) marks nuclei. (D-G) Effect of Crb intra on the in vivo Ex protein stability reporter. Confocal micrographs of transverse sections of wing discs from wandering third instar larvae expressing a wt (D and E) or a S453A mutant version of the ubi-Ex 1-468 ::GFP reporter (F and G) (green in D-G and grey in D'-G'), in the absence Figure 3 continued on next page SCF Slimb/b-TrCP in the cytoplasm, as had been suggested for endogenous Ex (Zhang et al., 2015). When a Crb intra transgene was expressed in the posterior compartment using the hedgehog-Gal4 driver (hh-Gal4), ubi-Ex 1-468 ::GFP was lost from the apical surface and degraded ( Figure  . This re-localisation of Ex 1-468 S453A mirrored the localisation of over-expressed Crb intra , which localised to the apical cortex as well as basal membranes (Figure 3-figure supplement 1E). It is therefore possible that Ex 1-468 S453A is mislocalised to the basolateral membrane by ectopic Crb, where it cannot be degraded. A further advantage of using this N-terminal fragment as a reporter is that our constructs lack the Yki-binding PPxY motifs, and therefore, do not induce cell death and reduced tissue growth, as does full length Ex Badouel et al., 2009;Oh et al., 2009). Indeed, neither reporter affected wing size when expressed at low levels under the control of the ubi promoter (Figure 3-figure supplement 1F-I). In contrast, expression of full-length Ex under the control of en-Gal4 driver induced a reduction in tissue size (Figure 3-figure supplement 1J,L,P). This phenotype was partially rescued by co-expression of Crb intra , which we expect would promote increased turnover and degradation of Ex (Figure 3-figure supplement 1M,P). Interestingly, expression of the S453A mutant version of Ex resulted in a more severe undergrowth than wt Ex, and this was refractory to Crb intra co-expression, suggesting that Crb fails to regulate the levels of Ex when the residue that is recognised by Slmb/b-TrCP is mutated (Figure 3-figure supplement 1N-P).
We then analysed whether CKI over-expression could modulate levels of the Ex reporter. Indeed, when compared to control wing discs, expression of CkIa or Gish using hh-Gal4 resulted in a significant decrease in the levels of the reporter in the posterior compartment ( Figure 3H-3K). This is consistent with our S2 cell results, and suggests that the CKI family kinases regulate Ex stability in vivo. However, this effect on Ex levels was not seen upon over-expression of either Dco or the poorly characterised CKI family member CG7094 ( Figure 3K), indicating that not all CKIs modulate Ex stability in vivo. Importantly, the effect of CKIs on Ex is not due to changes in Crb levels ( Figure 3-figure supplement 1Q), and is dependent on kinase activity, since a Gish isoI kinase-deficient construct could not alter reporter levels ( Figure 3K). In contrast, over-expression of CKIs did not alter the (D and F) or presence (E and G) of hh-Gal4-driven Crb intra over-expression. Ci immunostaining (red) indicates the anterior compartment, where hh is not expressed. DAPI nuclear staining is shown in blue. In the absence of Crb intra , both versions of the reporter localise at the apical surface. Expression of Crb intra causes depletion of the wt reporter, while it promotes mislocalisation of the S453A variant, in agreement with previously published data (Ribeiro et al., 2014). (H-K) Over-expression of CKI kinases promotes depletion of the Ex stability reporter. XY and transverse sections of third instar wing imaginal discs expressing ubi-Ex 1-468 ::GFP (green in H-J, grey in H'-J') and either no transgene (H), UAS-ckIa (I) or UAS-gish isoI (J) under the control of hh-Gal4. Ci immunostaining (anterior compartment, lacking hh expression) is shown in red. Expression of CkIa or Gish reduces the levels of the Ex in vivo stability reporter in vivo. (K) Quantification of the ratio between the levels of the ubi-Ex 1-468 ::GFP reporter in the posterior versus the anterior compartment in wing discs of the indicated genotypes (transgene expression driven by hh-Gal4 as in H-J). Data are shown in box (median and 25 th -75 th percentile) and whiskers (minimum to maximum) plots with all data points represented. n > 18 for all genotypes. Significance was assessed using a one-way ANOVA comparing all genotypes to w iso , with Dunnett's multiple comparisons test. **, p<0.01; ***, p<0.001, ****, p<0.0001. n.s. non-significant. In XY sections, ventral is up, whereas apical is up in transverse sections. Scale bars, 10 mm in A-C and 20 mm in D-J.

CKI loss-of-function promotes stabilisation of Ex in vitro
Next, we sought to analyse loss-of-function phenotypes of the CKI kinases. Upon treatment of S2 cells with gish RNAi, Crb intra -mediated depletion of both Ex 1-468 and Ex FL was strongly reduced ( Figure 4A-B and Figure 4-figure supplement 1A). We also attempted to deplete the other CKI kinases by RNAi. However, we were unable to generate a dsRNA that specifically targeted dco (Figure 4-figure supplement 1B). To overcome this, we used RNAi sequences that targeted only ckIa ("ckIa") or all CKIs (ckIa, dco and gish, termed "ckI pan ", previously used in Liu et al. (2002) (Figure 4-figure supplement 1B). In line with our previous data, knockdown of ckIa or gish alone resulted in robust stabilisation of Ex 1-468 in the presence of Crb intra ( Figure 4C). It is noteworthy that depleting all CKIs resulted in a more prominent stabilisation of Ex 1-468 , suggesting that CkIa and Gish work together to regulate Ex stability ( Figure 4C).
As the CKI kinases appear to act downstream of Crb to regulate Ex stability, we also assessed if their depletion could affect the ability of Crb to promote Ex:Slmb binding and, consequently, Slmbmediated Ex ubiquitylation and degradation. Indeed, the interaction between Ex and Slmb was abolished when gish or when all CKIs were depleted by RNAi ( Figure 4D). When we monitored Crbinduced Ex ubiquitylation, we observed that this was dramatically reduced when gish or all CKIs were depleted by RNAi ( Figure 4E-F). Indeed, knocking-down CKIs had a similar effect to depletion of slmb, part of the Ex E3 ligase complex activated downstream of Crb ( Figure 4E). These data suggest that CKI kinases act downstream of Crb to promote the interaction of Ex with Slmb, thereby stimulating Ex ubiquitylation and degradation.
Phosphorylation of Crb by aPKC has been proposed to influence its function and, therefore might contribute to the regulation of Ex (Ribeiro et al., 2014;Sotillos et al., 2004). To test this, we compared the effect of wt Crb to that of a variant of Crb containing point mutations in the putative aPKC phosphorylation sites (Crb 4A mut ). Similar to wt Crb, Crb 4A mut efficiently promoted Ex degradation ( Figure 4-figure supplement 1C), suggesting that aPKC phosphorylation is not required for the regulation of Ex downstream of Crb. A previous report has shown that, in addition to the N-terminal Slmb/b-TrCP degron, there is a C-terminal degron in Ex, which is inhibited by Wts-mediated phosphorylation (Zhang et al., 2015). To investigate whether Wts plays a role in Crb-mediated Ex degradation, we co-expressed Ex FL with either Crb FL or Crb intra , in the presence or absence of ectopic Wts (Figure 4-figure supplement 1D). Though our data confirm that Wts can stabilise Ex FL , we observed that the presence of Wts did not prevent the degradation of Ex mediated by the expression of Crb, indicating that the N-terminal degron supersedes the C-terminal one when Crb triggers Ex degradation (Figure 4-figure supplement 1D).

Clonal analysis of CKI mutants
As CKI loss-of-function in S2 cells resulted in Ex stabilisation, we next aimed to validate these observations in vivo. When we assessed the levels of the ubi-Ex 1-468 ::GFP reporter in third instar wing imaginal discs carrying clones mutant for existing gish (gish KG03891 ) or dco (dco le88 ) alleles, we observed no differences from control cells, suggesting that these kinases are not essential for maintaining Ex levels at steady state in vivo ( Figure 5A and Figure 5-figure supplement 1A). We observed occasional gish KG03891 adult escapers, leading us to hypothesise that this allele is hypomorphic. Therefore, we generated a new gish mutant (bearing a premature termination codon at the N-terminal part of the kinase domain), as well as a mutant for the poorly characterised CKI family member CG7094, which is also expressed in imaginal discs (Brown et al., 2014) using CRISPR/Cas9 gene editing ( Figure 5-figure supplement 1B-C). However, when we generated FRT clones for these new alleles, neither caused a change in ubi-Ex 1-468 ::GFP levels compared to wt tissue ( Figure 5B and Figure 5-figure supplement 1D). In contrast, we observed that the ubi-Ex 1-468 :: GFP reporter was elevated in ckIa 8B12 clones ( Figure 5C). We noticed that apical levels of Crb were also increased in ckIa 8B12 clones ( Figure 5-figure supplement 1E). This was not due to a general increase in apical domain size, since levels of another apical protein, aPKC, did not increase in ckIa 8B12 clones ( Figure 5-figure supplement 1F-G).
control) or gish. Lysates were processed for immunoblot analysis using the indicated antibodies. In both cases, gish depletion blocked Ex degradation induced by Crb intra expression. (C) CKI kinase knockdown blocks Ex degradation induced by expression of Crb intra . S2 cells treated with dsRNA targeting lacZ or CKI kinases were co-transfected with FLAG-tagged Ex 1-468 and GFP, Crb intra or Crb DFBM . Immunoblot analysis of lysates using the Figure 4 continued on next page It is possible that the increase in Ex levels in ckIa 8B12 clones is the result of an increase in Crb within the clones, causing enhanced recruitment of Ex to the apical cortex, rather than an effect on Ex stability. Alternatively, increased Ex levels as a result of its stabilisation in ckIa 8B12 clones might in turn lead to increased membrane retention or stabilisation of Crb. If this were the case, increasing Ex dosage should increase Crb levels. Indeed, over-expression of UAS-Ex 1-468 led to a robust increase in endogenous Crb at the apical membrane ( Figure 5D). We had previously reported that clones mutant for the hypomorphic allele slmb 1 showed elevated levels of Ex, but not Crb (Ribeiro et al., 2014). However, upon careful examination, we were able to detect increased Crb levels in slmb 1 clones, albeit mainly when these clones were large ( Figure 5E). Furthermore, mutant clones for a null allele, slmb 9H4-17 , displayed a clear increase in Crb levels ( Figure 5F). Thus, both slmb and ckIa mutant clones display elevated levels of Ex and Crb.
Since our S2 cell data suggested that several members of the CKI family may regulate Ex redundantly ( Figure 4C), we generated triple mutant clones lacking CG7094, gish and dco, but these did not have altered levels of the Ex 1-468 ::GFP reporter ( Figure 5-figure supplement 1H). Triple mutant clones for ckIa, gish and dco had elevated Ex and Crb levels ( Figure 5-figure supplement  1I), but these were comparable to ckIa clones alone ( Figure 5C and In S2 cells, depletion of gish or ckIa resulted in Ex stabilisation in the presence of Crb ( Figure 4A-C). We hypothesised that over-expressing Crb in vivo would accelerate Ex turnover, thereby providing a sensitised background to examine the effects of CKI depletion. To circumvent the pleiotropic effects of long-term knockdown in imaginal disc epithelial cells, we expressed ckIa R-NAi or gish RNAi in the absence or presence of Crb intra using hh-Gal4, and controlled the expression temporally (for 24 hr or 48 hr) using tub-Gal80 ts (Figure 6). Quantifications for these data are presented in Figure 6  indicated antibodies revealed that depleting ckIa and gish alone or all CKIs (ckI pan ) dramatically impaired the ability of Crb intra to promote Ex 1-468 degradation. (D) Crb requires CKI kinase function to promote Ex:Slmb binding. Co-IPs were performed between FLAG-tagged GFP or Slmb DF-box and V5-tagged Ex 1-468 in the presence of Crb intra or Crb DFBM and depletion of lacZ or CKI kinases. Lysates were analysed by immunoblot using the indicated antibodies for detection of protein expression and co-purification. Note that depletion of all CKIs or gish prevented the Ex:Slmb interaction induced by Crb intra . (E) and (F) CKI kinases are required for Crb intra -induced Ex ubiquitylation. S2 cells were treated with the indicated dsRNAs for 24 hr before transfection with the indicated constructs. Following lysis under denaturing conditions, ubiquitylated proteins were isolated using anti-HA antibodies. The presence of Ex and Crb intra was assessed with the indicated antibodies. Knockdown of gish (E and F) or of all CKIs with ckI pan RNAi (F) significantly reduced Ex ubiquitylation, similar to depletion of slmb. GFP was used as transfection efficiency control. Tubulin was used as loading control. Open and full circles denote absence or presence of the indicated plasmid, respectively. Asterisks denote non-specific bands (IgG heavy chain in IP panels). DOI: https://doi.org/10.7554/eLife.46592.011 The following figure supplement is available for figure 4:   Figure 5 continued on next page significant rescue of levels and localisation of the ubi-Ex 1-468 ::GFP reporter compared to Crb intra alone ( Figure 6-figure supplement 1A-B). Thus, ckIa knockdown, but not gish knockdown, is able to inhibit Crb-mediated Ex degradation in vivo. dco RNAi had no effect on Ex::GFP reporter levels in the absence (Figure 6-figure supplement 1C), or presence of Crb intra (Figure 6-figure supplement 1D). We confirmed these findings using the MARCM technique, and observed that Crb intraexpressing, GFP-marked MARCM clones caused complete loss of apical ubi-Ex 1-468 ::mScarlet (Figure 6-figure supplement 1E,G). gish MARCM clones expressing Crb intra resembled Crb intra MARCM clones alone (loss of the apical Ex reporter) ( Figure 6-figure supplement 1F). In contrast, removing ckIa resulted in a partial rescue of Crb-induced Ex degradation, with a proportion of mutant cells retaining ubi-Ex 1-468 ::mScarlet apically ( Figure 6-figure supplement 1H). Together, these results suggest that, in vivo, CkIa regulates Ex stability by promoting Crb-induced turnover.
An increase in Ex stability at the apical plasma membrane would be expected to lead to decreased Yki activity upon ckIa loss (Fulford et al., 2018). In fact, slmb depletion in the posterior wing disc leads to downregulation of the Yki transcriptional reporter ex-lacZ and increased Yki nuclear exclusion (Zhang et al., 2015). Levels of both ex-lacZ and diap1-GFP3.5, a Yki-responsive fragment of the diap1 promoter (Zhang et al., 2008) were markedly decreased in ckIa mutant wing disc clones ( Figure 7A,C), while ex-lacZ was not affected in gish mutant tissue ( Figure 7B). Furthermore, we used a Yki-GFP knock-in line (Fletcher et al., 2018) to show that Yki is excluded from the nucleus in ckIa mutant cells compared with neighbouring wild type tissue ( Figure 7D). Finally, overexpression of CkIa, but not Gish, in the posterior compartment of the wing disc increased Yki activity as measured by ex-lacZ levels ( Figure 7E-H). Thus, loss of ckIa leads to Ex stabilisation correlated with decreased Yki activity, while its overexpression has the opposite effect, consistent with a role for CkIa in promoting Yki activity via the control of Ex stability.

Discussion
Ex was one of the first identified upstream regulators of the Hippo pathway  and functions as a growth suppressor in Drosophila (Fulford et al., 2018). In agreement with its key role in growth control, it is increasingly evident that Ex is tightly regulated, both at the transcriptional level via Yki itself  and through its subcellular localisation and stability (Chen et al., 2010;Ling et al., 2010;Robinson et al., 2010;Su et al., 2017;Ribeiro et al., 2014;Zhang et al., 2015;Ma et al., 2017). Crb recruits Ex to its site of activity at the apical plasma membrane, but also limits its apical levels by triggering Ex turnover via Slmb/b-TrCP, ensuring the fine-tuning of Yki activity (Chen et al., 2010;Ling et al., 2010;Robinson et al., 2010;Ribeiro et al., 2014;Zhang et al., 2015). Slmb can also promote the degradation of Ex in the cytoplasm in a Crb-independent manner (Zhang et al., 2015, Figure 3C and C'). Another ubiquitin ligase, Plenty of SH3s (POSH) has also been implicated in Ex degradation in parallel to Slmb (Ma et al., 2018). Since our data indicate that Ex levels are not sensitive to Slmb in the eye imaginal disc ( Figure 5-figure supplement 2), it is possible POSH is the dominant Ex regulator in this tissue. Interestingly, recent work indicates that the Fat atypical cadherin, a major upstream branch of Hippo signalling (Bennett and Harvey, 2006;Cho et al., 2006;Silva et al., 2006;Willecke et al., 2006) also influences Ex stability by apically recruiting the adaptor Dlish/Vamana (Misra and Irvine, 2016;Wang et al., 2019;Zhang et al., 2016). Dlish binds the Ex C-terminus via its SH3 domain and  Figure 6 continued on next page promotes the Slmb:Ex association, thereby increasing Ex turnover (Wang et al., 2019). Thus, Ex functions as a hub in Hippo signalling that receives input from both the Crb and the Fat branches of upstream signalling.
However, recognition by Slmb/b-TrCP requires substrate phosphorylation, which had remained an unexplored aspect of Ex regulation. Here, we provide evidence that the CKI family of kinases and, in particular, CkIa and Gish, act downstream of Crb to promote phosphorylation of Ex, in turn allowing binding to Slmb/b-TrCP and Ex ubiquitylation and degradation. In conditions whereby Crbmediated turnover of Ex is active, such as through Crb intra expression, loss of ckIa (and gish in cell culture) inhibited Ex degradation ( Figure 4A-C and Figure 6C-D). In the absence of CKI function, Crb is unable to trigger Ex:Slmb binding and thus can no longer induce Ex degradation ( Figure 4D-F). In agreement with the requirement for Slmb/b-TrCP function, Ex degradation is dependent on phosphorylation and the presence of a b-TrCP consensus site ( 452 TSGIVS 457 ), which is phosphorylated to allow the interaction to occur (Ribeiro et al., 2014). Our data indicate that CKIs mediate Ex phosphorylation downstream of Crb, as the effect of CKIs on Ex stability is dependent on their kinase activity ( Figure 2B-C, Figure 2-figure supplement 1A-B and Figure 3K) and mutation of a conserved residue in the Slmb/b-TrCP consensus (S453A) rendered Ex insensitive to degradation induced by Crb, CKIs or Slmb (Ribeiro et al., 2014) (Figures 3G and 2D). Interestingly, CkIa and Gish were still able to induce a mobility shift in Ex when the S453 residue was mutated ( Figure 2D), suggesting that other residues besides S453 may be targets of CKI-mediated phosphorylation. An Ex truncation lacking residues 450-468 (Ex 1-450 ) fails to undergo a mobility shift in the presence of Crb intra or Gish (Figure 2-figure supplement 1D). This may indicate that the additional residues targeted by CKI are in this region, or simply that CKI needs these residues to dock onto Ex.
It remains to be determined whether phosphorylation at additional sites besides S453 is required to promote Ex:Slmb binding, and whether CKIs are part of a phospho-priming mechanism that involves additional kinases, as commonly seen for other CKI targets, such as b-catenin downstream of Wnt signalling (Liu et al., 2002;Amit et al., 2002;Jiang and Struhl, 1998;Stamos and Weis, 2013;Winston et al., 1999). In the case of b-catenin, CK1a acts as a priming kinase for GSK3b/Sgg; however, our data suggest that Sgg does not affect Ex stability (Figure 2-figure supplement 1E-G). Wts/LATS is another candidate, since mammalian LATS1/2 act as priming kinases for CK1d/e to target YAP for degradation by SCF b-TrCP (Zhao et al., 2010). However, Wts has been suggested to stabilise Ex through phosphorylation at S1116, rather than promote its degradation (Zhang et al., 2015). Furthermore, Crb-dependent Ex degradation appears to override Wts-mediated stabilisation (Zhang et al., 2015 and Figure 4-figure supplement 1D). The potential involvement of other kinases in Ex/Slmb binding therefore remains an open question.
An unexpected aspect of our results is the fact that, in cultured S2 cells, Gish (CKIg) depletion has a strong effect on Ex stability ( Figure 4A-C), while CkIa appears to be the dominant player in the wing imaginal disc, both in the presence ( Figure 6C-D) and absence ( Figure 5C) of Crb intra expression. This is surprising, since Crb-induced Ex phosphorylation and ubiquitylation presumably occurs at the apical plasma membrane, where Gish is known to be localised (Morin et al., 2001). This is unlikely to be due to a lack of Gish activity, since it is implicated in several signalling pathways in the wing disc (Li et al., 2016;Gault et al., 2012). However, CKI isoforms are known to exist in distinct subcellular pools associated with components of the signalling pathways they regulate (Jiang, 2017;Knippschild et al., 2014). It is therefore possible that recruitment of CkIa to Crb/Ex may involve an adaptor protein expressed in the wing disc, as is the case for CkIe, which requires the DEAD-box  RNA helicase DDX3 to phosphorylate Dishevelled2 in the mammalian Wnt pathway (Cruciat et al., 2013).
As Ex apical accumulation is predicted to antagonise Yki, we measured several readouts of Ykidependent transcription and found that Yki activity is indeed decreased upon ckIa depletion and increased by CkIa overexpression, while Gish manipulations had no such effect ( Figure 7A-C,E-H). ckIa mutations would therefore be predicted to reduce tissue growth by reducing Yki activity. However, we observed that ckIa mutant clones are generally overgrown, round, and tend to delaminate from the wing disc epithelium, so that they often do not contribute to the adult wing ( Figure 5C and Figure 7A,C). This overgrowth phenotype is due to a strong elevation of Wingless (Wg) signalling in ckIa mutant tissue (Legent et al., 2012) and is very similar to the loss of two Wg antagonists, axin and APC1/2 (Adenomatous Polyposis Coli) (Muñoz-Descalzo et al., 2011). Thus, it is likely that the decreased growth occurring as a result of partial loss of Yki activity in ckIa mutant clones is compensated by a strong increase in Wg signalling.
Our data also revealed that loss of either ckIa or slmb function results in an increase in the apical levels of Crb as well as Ex ( Figure 5E-F and Figure 5-figure supplement 1E). This is not due to a general increase in apical domain size, since aPKC is not affected ( Figure 5-figure supplement  1F-G). These observations raise two possibilities; either that Crb stability is regulated by the same mechanism as Ex, or that Crb and Ex can affect each other's apical localisation. It seems unlikely that Crb is co-degraded with Ex in a Slmb/b-TrCP-dependent manner, since Crb lacks clear b-TrCP and CKI consensus sequences in its intracellular domain. Moreover, it has been previously shown that Crb levels increase in ex mutant clones (Chen et al., 2010), a result not easily reconcilable with codegradation. Instead, we favour the second possibility, in which excess Ex accumulation leads to increased Crb apical levels. Consistent with this idea, over-expression of Ex 1-468 leads to a robust increase in apical Crb ( Figure 5D). Apical Crb endocytosis occurs through docking of the AP-2 (adaptor-protein 2) complex member a-adaptin, which competes with Stardust to bind the Crb intracellular domain (Lin et al., 2015). It is therefore possible that excess Ex interferes with Crb recognition by a-adaptin, thereby decreasing Crb internalisation rates in ckIa or slmb mutant clones.
CRB3, the mammalian orthologue of Crb, can regulate Hippo signalling by mechanisms analogous to those described in Drosophila. CRB3 interacts with two proposed orthologues of Ex, Angiomotin (AMOT), and FRMD6, in a density-dependent manner (Varelas et al., 2010;Angus et al., 2012;Mao et al., 2018;Moleirinho et al., 2013). It would therefore be interesting to test if the stability of either AMOT or FRMD6 is regulated by CRB3, CKI and b-TrCP. In summary, our data support a model whereby the apical polarity determinant Crb promotes Ex phosphorylation by CKI family kinases, which in turn stimulates Slmb/b-TrCP-mediated ubiquitylation and degradation. We propose that this mechanism facilitates precise and dynamic control of Ex protein levels at the apical membrane, which is crucial for the regulation of Hippo signalling.  (green in C, C'' and D, grey in C', (C''' and D'). Clones are marked by absence of RFP (red) and highlighted by white dashed lines; DAPI (blue) stains nuclei. Reporter gene expression is drastically reduced in ckIa 8B12 (A, C), but not gish 17 (B), mutant clones. Yki:: GFP appears excluded from the nucleus of ckIa 8B12 mutant cells (D). Scale bars 20 mm. (E-H) Overexpression of CkIa, but not Gish isoI , results in upregulation of ex-lacZ. Maximum intensity projections of z-stacks of the pouch region of wing imaginal discs from third instar larvae overexpressing no transgene (E), UAS-CkIa (F), or UAS-Gish isoI (G) under the control of hh-Gal4. Crosses were raised at 25˚C and larvae were dissected at wandering L3 stage. ex-lacZ expression was detected by immunostaining for b-galactosidase (red in E-G, grey in E'-G'); the posterior compartment is marked by expression of GFP (green); DAPI (blue) stains nuclei. (H) Quantification of the posterior to anterior ratio of ex-lacZ signal intensity in the pouch region; CkIa expression significantly upregulates this Yki target gene (p=0.0001, one-way ANOVA comparing all means to hh >control, with correction for multiple comparisons; n 8 for all genotypes), while Gish isoI does not (p=0.4808). Scale bars 20 mm. DOI: https://doi.org/10.7554/eLife.46592.020 The following source data is available for figure 7: Source data 1. Source data for quantification of relative levels of the Ex::GFPin vivoreporter in wing imaginal discs expressing RNAi targeting CKI kinases alone or in combination with expression of Crbintra . DOI: https://doi.org/10.7554/eLife.46592.021

Drosophila cell culture, expression constructs and chemical treatments
Work involved the use of the Drosophila cell line Schneider S2 (RRID:CVCL_Z232). Cells were obtained from the ATCC and mycoplasma testing revealed no contaminations. Drosophila S2 cells were grown in Drosophila Schneider's medium (Thermo Fisher Scientific) supplemented with 10% (v/ v) FBS, 50 mg/mL penicillin and 50 mg/mL streptomycin. Expression plasmids were transfected using Effectene transfection reagent (QIAGEN). Expression plasmids were generated using Gateway technology (Thermo Fisher Scientific). ORFs were PCR amplified from cDNA clones obtained from the Drosophila Genomics Resource Center (DGRC, https://dgrc.cgb.indiana.edu/vectors/Overview) and cloned into Entry vectors (pDONR207, pDONR-Zeo). Vectors from the Drosophila Gateway Vector Collection and an in-house V5 tag expression vector were used as destination vectors (Ribeiro et al., 2014). All Entry vectors were verified by sequencing. Point mutations were generated using the Quikchange Site-Directed Mutagenesis kit (Agilent). The Ex full length, Ex 1-468 and Crb intra plasmids were previously described (Ling et al., 2010;Genevet et al., 2010;Ribeiro et al., 2014). The WT and mutant versions of Crb differ by specific amino acid substitutions in its intracellular portion (Y 10 P 12 E 16 to A 10 A 12 A 16 ) (Ling et al., 2010). The NTAN-FLAG (N-terminal amidohydrolase 1, an enzyme from the N-end rule pathway that associates with a small number of proteins in Drosophila S2 cells), HA-ubiquitin and Crb 4A mut plasmids were kind gifts from M. Ditzel (University of Edinburgh, Edinburgh), P. Meier (Institute for Cancer Research, London) and B. Thompson (Francis Crick Institute, London), respectively. mScarlet (85042; RRID:Addgene_85042) and pCFD3 (49410; RRID: Addgene_49410) plasmids were obtained from Addgene. Where indicated, proteasome inhibition was achieved by treating cells with 50 mM MG132 (Cambridge Bioscience) and 50 mM calpain inhibitor I (Ac-LLnL-CHO or LLnL) (Sigma) for 4 hr before cell lysis or with 5 mM MG132 overnight.

Immunoprecipitation and immunoblot analysis
For purification of FLAG-tagged proteins, cells were lysed in lysis buffer (50 mM Tris pH 7.5, 150 mM NaCl, 1% Triton X-100, 10% (v/v) glycerol, and 1 mM EDTA), to which 0.1M NaF, phosphatase inhibitors 2 and 3 (Sigma) and protease inhibitor cocktail (Complete, Roche) were added. Cell extracts were spun at 17,000 g for 10 min at 4˚C. FLAG-tagged proteins were purified using anti-FLAG M2 Affinity agarose gel (Sigma) for >1 hr at 4˚C. FLAG immunoprecipitates were then washed four times with lysis buffer before elution using 150 ng/ml 3x FLAG peptide for 15-30 min at 4˚C. For isolation of ubiquitylated proteins, cells were collected by centrifugation and washed with cold PBS. 10% of cell material was lysed as described above. The remaining 90% was lysed with boiling 1% SDS-PBS for 5 min. Following quick vortexing, samples were incubated for 5 min at 100˚C before 5fold dilution using 0.5% BSA-1% Triton X-100-PBS. DNA was sheared by sonication and cell extracts were cleared by centrifugation at 17,000 g for 10 min at 4˚C. Samples were diluted 2-fold with 0.5% BSA-1% Triton-X-100-PBS and incubated overnight at 4˚C with monoclonal anti-HA agarose beads (Sigma) using Bio-Spin Columns (Bio-Rad). Following incubation, samples were washed with 0.5% by the Fly Facility of the Department of Genetics, University of Cambridge. crb 82-04 was obtained from Duojia Pan (UT Southwestern). slmb 9H4-17 was obtained from Daniel St Johnston (Gurdon Institute, UK). slmb 1 was obtained from Daniel Kalderon (Columbia University). dco le88 was obtained from Tomas Dolezal (University of South Bohemia). ckIa 8B12 was obtained from Jessica Treisman (NYU Skirball Institute). gish K03891 was obtained from Bloomington (BL-13263).
All crosses were raised at 25˚C unless otherwise stated. Genotypes were as follows: Figure 3A, Figure   Significance was calculated by unpaired t-test. For Figure 6-figure supplement 1B, Ex 1-468 ::GFP ratio was calculated as aforementioned and significance was calculated using a two-way ANOVA (comparing all genotypes to Crb intra within each timepoint) with Dunnett's multiple comparisons test. For Figure 7H, P/A ratio of ex-lacZ intensity was calculated by manually drawing around each compartment of the pouch in maximum intensity projections, then measuring the mean grey pixel value in Fiji. Significance was calculated using a one-way ANOVA comparing all means to hh >control, with correction for multiple comparisons; n 8 for all genotypes.

Analysis of genetic interactions in Drosophila adult wings
For analysis of genetic interactions in the Drosophila wing, flies with the genotypes of interest were collected and preserved in 70% EtOH for at least 24 hr. Wings were removed in 100% isopropanol, mounted in microscope slides using Euparal (Anglian Lepidopterist Supplies) as mounting medium and baked at 65˚C for at least 5 hr. Adult wing images were captured using a Pannoramic 250 Flash High Throughput Scanner (3DHISTECH) and extracted using the Pannoramic Viewer software (3DHISTECH). Wing area was quantified using ImageJ (the alula and costal cell of the wing were both excluded from the analysis). Images were processed using Adobe Photoshop (RRID:SCR_ 014199).

Mass spectrometry analysis
AP-MS experiments followed a GeLC MS/MS approach. Gel lanes were fragmented into eight equally sized pieces and subjected to in-gel trypsin digestion using a Perkin Elmer Janus Automated Workstation. Peptide mixtures were acidified to 0.1% TFA and injected onto a nanoACQUITY UPLC (Waters Corporation) coupled to an LTQ-Orbitap XL (Thermo Fisher Scientific) via an Advion Biosciences Nanomate. Peptides were eluted over a 30 min gradient (5-40% ACN). Mascot distiller was used to extract peak lists, which were searched with Mascot v.2.4.1 (Matrix Science; RRID:SCR_ 014322) against the Drosophila melanogaster Uniprot reference proteome. Methionine oxidation was entered as a variable modification and search tolerances were 5 ppm and 0.8 Da for peptides and fragments, respectively. Individual lane searches were combined and results compiled in Scaffold 4.0.3 (RRID:SCR_014345).