Hedgehog mediated degradation of Ihog adhesion proteins modulates cell segregation in Drosophila wing imaginal discs

The Drosophila Hedgehog receptor functions to regulate the essential downstream pathway component, Smoothened, and to limit the range of signaling by sequestering Hedgehog protein signal within imaginal disc epithelium. Hedgehog receptor function requires both Patched and Ihog activity, the latter interchangeably encoded by interference hedgehog (ihog) or brother of ihog (boi). Here we show that Patched and Ihog activity are mutually required for receptor endocytosis and degradation, triggered by Hedgehog protein binding, and causing reduced levels of Ihog/Boi proteins in a stripe of cells at the anterior/posterior compartment boundary of the wing imaginal disc. This Ihog spatial discontinuity may contribute to classically defined cell segregation and lineage restriction at the anterior/posterior wing disc compartment boundary, as suggested by our observations that Ihog activity mediates aggregation of otherwise non-adherent cultured cells and that loss of Ihog activity disrupts wing disc cell segregation, even with downstream genetic rescue of Hedgehog signal response.

T he Hedgehog (Hh) signaling pathway patterns a variety of organs and tissues in metazoan embryos [1][2][3][4] . Post-embryonically, the Hh pathway also functions homeostatically in processes such as tissue maintenance and regeneration, acting on tissue stem or progenitor cells [5][6][7][8][9] . Defective Hh signaling during development thus leads to patterning defects such as disrupted segmentation in Drosophila 10 or holoprosencephaly 11 and other malformations in humans. Post-embryonic dysregulation of Hh pathway activity can result in proliferative conditions such as malignant tumors 12 or in tissue degeneration 13 .
The secreted Hh ligand is synthesized as a precursor protein that undergoes cleavage and cholesterol modification via autoprocessing, followed by further lipid modification through covalent attachment of palmitate 14 . Release of the mature, dually lipid-modified Hh ligand (HhNp) from cells then requires Dispatched 15,16 , and also involves other protein and lipoprotein components [17][18][19][20][21][22][23][24][25] . The Hh protein is typically released from a localized source, and can then induce concentration-dependent cellular differentiation or proliferation responses in cells of surrounding tissues and structures.
In insects and vertebrates, transcriptional upregulation of the receptor component, Patched (Ptc), is a direct response to Hh signaling [26][27][28][29][30] . In Drosophila wing imaginal discs, this response occurs exclusively in anterior (A) compartment cells, whereas Hh is produced and secreted exclusively in posterior (P) compartment cells, from where it spreads toward the A compartment 26,27,31 . Cells of the A and P compartments do not intermingle but remain segregated within the disc, separated by a smooth boundary that does not correspond to any morphological features [32][33][34] . This classically defined lineage restriction between cells of the A and P compartment depends largely on the response to the Hh signal exclusively in A cells and is postulated to result from differences in A and P cell affinities 35,36 . However, the identity of the gene(s) contributing to distinct A and P cell affinities is unknown.
Hh response does not occur in P compartment cells because critical components of the Hh pathway, such as the transcriptional effector Cubitus interruptus (Ci), are not expressed 37 . Cells of the A compartment in contrast express Ci and other pathway components, such as Ptc, which suppresses Smoothened (Smo) activity in the absence of Hh. In A compartment cells located close to the P compartment source of Hh protein, response to the Hh signal stabilizes and activates Smo 38 , and both suppresses formation of Ci repressor and stimulates formation of the activator form of Ci, thus triggering an increase in the transcription of target genes such as ptc and decapentaplegic (dpp) in a stripe of A cells at the A/P compartment boundary 26,27,31,39,40 . Moreover, this transcriptional stimulation of Ptc results in a second function of the Hh receptor beyond Smo regulation, namely, the sequestration of Hh and restriction of Hh movement by internalization of Hh through endosomes followed by lysosome degradation 40,41 .
Although Patched has long been known as an essential component of the Drosophila Hedgehog receptor 39,40 , more recent work shows that the Drosophila Hh receptor complex must also include Ihog (Interference Hedgehog) or its close relative Boi (Brother of Ihog) for Hh binding and biological response [42][43][44][45][46][47][48] , as well as for sequestration of the Hh protein to limit long-range signaling 42,43,49,50 . The Drosophila Ihog and Boi proteins, as well as the related mammalian proteins CAM-related/downregulated by oncogenes (Cdo) and Brother of CDO (Boc) 51 , are type I single-span transmembrane proteins with four or five extracellular immunoglobulin (Ig) domains, two or three extracellular repeats of fibronectin type III (FNIII) domains, and cytoplasmic sequences of unknown structure or function. Our previous biochemical and structural studies showed that the first FNIII domain (Fn1) of Ihog/Boi directly contacts HhN 45,46 , whereas Fn2, the second FNIII domain of Ihog/Boi, contacts Ptc 43 . The mammalian members of the Ihog family, Cdo and Boc, both contribute to Hh signaling 45,[52][53][54] by binding to mammalian Hh proteins via a non-orthologous FNIII repeat 45,52,55 .
Although the requirement for Ihog/Boi for response to Hh has been amply confirmed [42][43][44]48 , some authors have been unable to observe a role for Ihog/Boi in Hh protein sequestration 56 . Here, we begin by confirming the role of Ihog/Boi in Hh sequestration under physiological conditions. We then explore the mechanism by which Ptc and Ihog/Boi jointly contribute to sequestration of the Hh protein ligand. We identify a post-transcriptional process in which reciprocal regulation of Ihog/Boi and Ptc controls their joint internalization and lysosome degradation upon Hh binding. Remarkably, despite spatially uniform transcription of ihog and boi genes, this Hh-induced receptor clearance results in reduced levels of Ihog/Boi protein in a stripe of cells at the A/P compartment boundary of the wing imaginal disc. Given that Ihog/ Boi proteins resemble typical cell adhesion molecules, we tested for activity in cell-cell adhesion and found that Ihog/Boi indeed mediate aggregation of otherwise non-adhesive cultured cells. In addition, we find that loss of Ihog activity can disrupt A/P cell segregation and lineage restriction, even with downstream genetic rescue of Hedgehog signal response.

Results
Ihog/Boi is absolutely required for Hh sequestration. Previously, we reported that Ihog/Boi-expression is required for sequestration of Hh to limit its range of action. In their original work defining the phenomenon of sequestration, Chen and Struhl 40 established that clones lacking smo function on the A side of the A/P boundary show increased expression of endogenous Hh target genes (such as ptc or dpp) in wild-type cells just anterior to the smo mutant clone, due to loss of Hh-induced ptc expression within the smo mutant clone. Chen and Struhl 40 also noted that upregulated expression of ptc through downstream pathway activation by additional mutation of the cAMPdependent protein kinase 1 (PKA-C1) within smo mutant clones restores sequestration of Hh, as indicated by lack of increased expression of endogenous Ptc in wild-type cells immediately anterior to the smo; pka-C1 clones. We confirm this finding ( Supplementary Fig. 1A), but also note that Ptc expression persists on the anterior side of pka-C1 clones that also lack Ihog/Boi, at an abnormally large distance from the Hh-expressing posterior cells ( Supplementary Fig. 1B, E, F). Taken together, these results confirm our previous conclusion that Ihog/Boiexpression is required for Hh sequestration to limit its range of activity, and that expression of the Ptc protein at physiological levels alone is not sufficient for this sequestration activity 43 (see Supplementary Discussion for more in depth discussion and for analysis of the work of Camp et al., who failed to observe a requirement of Ihog/Boi for sequestration). In the work presented below, we explore the mechanism underlying reciprocal regulation of Ptc and Ihog/Boi in Hh sequestration.
Ihog promotes Ptc-mediated Hh endocytosis in vitro. We previously demonstrated that Ihog/Boi proteins interchangeably function to interact directly with Ptc, and that Ihog/Boi activity is required for presentation of Ptc on the cell surface. We also found that Ihog/Boi activity and Ptc are both required for high-affinity Hh binding 43 . This crucial role of Ihog/Boi in binding of Ptc to Hh adequately explains why Ihog/Boi activity is indispensable in transducing the Hh signal. However, the process of Ptc-mediated Hh sequestration also requires elimination of Ptc-bound Hh from the cell surface by endocytosis 40,41 . We therefore reasoned that if Ihog/Boi activity is required for Hh sequestration, it may also be essential for Ptc-mediated Hh endocytosis.
To test this hypothesis, we first asked whether Ihog can promote Ptc-mediated Hh endocytosis in vitro. To this end, we transfected cultured Drosophila S2R+ cells with Ihog or Ptc alone, and in combination. After incubating the transfected cells with HhN-containing medium for 1 h at room temperature to allow for internalization, we immunostained the cells with a protocol allowing detection of both surface and internalized protein ("Methods"). Consistent with previous observations 43 , we found that the amount of surface HhN markedly increased upon cotransfection of Ptc and Ihog as compared to Ptc alone (compare Fig. 1f to d; quantification in Fig. 1g). Importantly, the presence of Ihog also significantly augmented internalization of HhN in Ptcexpressing cells (compare Fig. 1f to d; quantification in Fig. 1h). In this experiment, we also noticed that the Ihog protein alone mainly localized to the cell surface and, although bound to HhN, was not able to internalize HhN in the absence of Ptc ( Fig. 1b; quantification in Fig. 1g and h). Consistent with the overlapping role of Ihog and its close relative Boi as Hh co-receptors, we found co-expression of Boi with Ptc also increased HhN internalization in S2R+ cells. Altogether, our data suggest Ihog proteins are able to promote Ptc-mediated Hh internalization in vitro.
Ihog is essential for Ptc-mediated Hh endocytosis in vivo. Encouraged by the ability of Ihog/Boi to promote Ptc-mediated Hh internalization in S2R+ cells, we examined whether Ihog/Boi is required for Ptc-mediated Hh endocytosis in the wing imaginal disc. Hh expression in the disc is limited to cells of the posterior (P) compartment, which do not respond to the Hh signal. Target gene expression is thus restricted to a stripe of adjacent cells within the anterior (A) compartment. Given that the ptc gene not only encodes a component of the Hh receptor but is also a transcriptional target for Hh signaling, its expression serves as a reporter of Hh response and its highest level of expression is in a stripe of A cells immediately adjacent to the A/P compartment boundary (Fig. 2a)   Two-tailed unpaired Mann-Whitney U-test was used for statistical analysis. ns not significant. **P < 0.01. ***P < 0.001. ****P < 0.0001. j S2R+ cells transfected for expression of Ptc and/or Ihog were labeled by surface biotinylation. Immunoblots in the top panels show 5% of the biotinylated cell lysate, and the bottom panel shows the biotin-labeled proteins recovered by Streptavidin agarose beads. Co-expression with Ptc and treatment of HhN reduced levels of Ihog on the cell surface NATURE COMMUNICATIONS | DOI: 10.1038/s41467-017-01364-z ARTICLE and internalizes Hh ligand secreted from the P cells. Wild-type wing disc cells (with one or two wild-type alleles of ihog/boi) thus exhibit punctate Hh staining co-localized with Ptc in intracellular vesicles (arrow heads, Fig. 2a). A similar Hh protein distribution has been described previously 26,57 . In contrast, we found that removing all function of both ihog and boi eliminated punctate Hh staining in the A cells at the A/P boundary (Fig. 2b). As Ptc protein is not expressed in clones lacking both Ihog and Boi, the loss of internalized Hh we noted could be due to loss of Ptc 43 (Fig. 2b). To determine whether Ihog/Boi proteins are required for Hh endocytosis in the presence of Ptc, we activated ectopic Ptc expression by MARCM in wing disc cells and compared cells with and without Ihog/Boi function (Fig. 2c, d). We observed HhN puncta in boi/+; ihog clones expressing Ptc ( Fig. 2c; blue arrows), but not in Ptc-expressing cells lacking both Ihog and Boi (Fig. 2d). We note that the presence of a single wild-type ihog/boi allele sufficed to prevent a defect of Hh internalization, in either the presence or absence of ectopic Ptc expression (Fig. 2a, c), consistent with previous findings that a single wild-type allele of ihog/boi suffices to provide functional activity 43   . Each bar shows the mean ± s.d. of fluorescence from n = 20 cells, and representative images are shown in a-l. Two-tailed unpaired Mann-Whitney U-test was used for statistical analysis. ns not significant. **P < 0.01. ***P < 0.001. ****P < 0.0001 that expression of the Ptc protein alone is not sufficient for internalization. Consistent with the previous observations 42, 58 , we also found that Ihog and Boi are required to retain Hh on the surface of P compartment cells in the wing imaginal disc (Hh is present on the surface of cells within clones in Fig. 2e (boi/+; ihog); compare to clones in Fig. 2f (boi; ihog)). This activity is dosage-dependent, as reducing Ihog/Boi activity by removing a single wild-type boi allele caused a partial loss of Hh on the surface of P cells (compare cells within and outside of clones in Fig. 2e, f). As P cells do not express ptc 59, 60 , Hh retention on the surface of P cells is apparently Ptc-independent, and these observations are consistent with our in vitro data showing that HhN was retained by cells transfected to express only Ihog and not Ptc, and that surface-bound HhN was not internalized (Figs. 1a, 2e, and 2f). However, upon ectopic expression of Ptc in P compartment cells, Ptc and Hh were found in punctate structures in cells that contain both proteins 61 . Thus, we wondered whether Ihog and Boi are also required for the ectopic Ptc-mediated Hh internalization in P cells. To answer this question, we ectopically expressed Ptc in P cells by MARCM and found that Ptc induced Hh internalization in P cells with a single dose of wild-type boi indicated by the punctate co-localization of Ptc and Hh ((boi/+; ihog clones; Fig. 2g), but not in cells when both endogenous ihog and boi are removed (Fig. 2h). Our evidence thus shows that Ihog/Boi is required for Ptc-mediated Hh endocytosis in Hh-responding A cells, but also for endocytosis in Hh-producing P cells when Ptc is ectopically expressed.
Ihog mediates Hh-induced Ptc degradation. Ptc limits the Hh gradient by internalizing Hh, and targeting it for lysosomal To examine Ihog effects on subcellular localization of Ptc, we cotransfected S2R+ cultured cells for expression of Ptc and Ihog, alone or in combination, followed by incubation with medium containing HhN. When transfected alone, Ptc localized primarily in intracellular vesicles 43 , likely early endosomes as indicated by co-localization with Rab5 staining and absence of co-localization with lysosomal staining (Fig. 3c, i; quantification in Fig. 3m, n). In the absence of Ihog coexpression, the early endosome localization of Ptc was not markedly changed upon Hh treatment (Fig. 3d, j; Fig. 6 Hh/Ptc-dependent modulation of Ihog protein levels in wing imaginal discs. Wing imaginal discs were immunostained with antibodies against Ihog (red), Ptc (blue), and GFP (green) proteins. a, c Activation of UAS-driven Hh expression in clones marked by expression of GFP (green). Ptc protein levels are increased in Hh-expressing cells and in nearby cells within the A compartment (arrows in a), but not in the P compartment (arrows in c). Reduced Ihog (red) levels are consistently detected in cells with elevated Hh and Ptc expression (arrows in a). Note that the reduction of Ihog levels in A compartment extends beyond cells in the clone that express Hh (red lines in a), but coincides with cells showing Hh-induced expression of high levels of Ptc (yellow lines in a). b, d Activation of UAS-driven Ptc expression (blue) in clones marked by expression of GFP (green). Ihog is also slightly reduced in clones with elevated Ptc expression when located in the A compartment (arrow in b) far away from Hh-producing cells. Ihog protein levels are markedly reduced in the Ptc-expressing clones located in the P compartment where Hh is highly expressed (arrows in d). Scale bar, 50 µm. e-h Quantification of Ihog intensity in clones marked by GFP and different compartments determined by Ptc staining in the wing discs. Ihog intensity at different location is normalized by that of P compartment of the same wing disc. Each bar shows the mean ± s.d. of normalized Ihog intensity from n = 10 wing discs, and representative images are shown in a-d. Two-tailed unpaired Mann-Whitney U-test was used for statistical analysis. ns, not significant. **P < 0.01. ***P < 0.001. ****P < 0.0001 wild-type cells; Ptc protein similarly was undetectable in clones with a single dose of wild-type boi (boi/+; ihog clones; Fig. 4b). In cells lacking both Ihog and Boi (boi; ihog clones), however, low levels of Ptc protein produced by the tubP-Ptc transgene were detectable (Fig. 4c), indicating that Ihog/Boi function is required in vivo for Hh-induced Ptc degradation. As would be expected, Smo protein levels are markedly reduced in boi; ihog clones with stabilized Ptc proteins (compare Fig. 4f to e). No Ptc was detectable in control posterior boi; ihog clones lacking the tubP-Ptc transgene (Fig. 4a), and, correspondingly, no loss of Smo was detected (Fig. 4d).
Endocytosis and degradation of Ihog together with Hh/Ptc. Given the critical role of Ihog/Boi activity in Ptc-mediated Hh internalization and the occurrence of Hh-induced degradation of Ptc, we considered the possibility that Ihog itself may be internalized and degraded in a Ptc-dependent and Hh-dependent manner. Indeed, we noted by immunostaining that Ihog is internalized in S2R+ cultured cells co-expressing Ptc and treated with HhN (compare Fig. 1e, f; quantified in Fig. 1i), and furthermore that surface biotinylation of Ihog was reduced in cells co-expressing Ptc and treated with HhN (Fig. 1j). The internalized Ihog proteins were found both in Rab5-positive early endosomes ( Fig. 3f; quantified in Fig. 3o) and lysosomes ( Fig. 3l; quantified in Fig. 3p). Thus, Ihog protein itself was internalized with Ptc in a HhN-dependent manner. In the absence of externally added HhN, we also observed a greater internalization of Ihog in S2R+ cells co-expressing Ptc (compare Fig. 1e with a, Fig. 3e with a, and Fig. 3k with g). In addition, we noted that the levels of Ihog in the cell lysate also decreased upon co-expression with Ptc (Fig. 1j). Given that no endogenous Hh is expressed in these cells 64 , Ptc degradation associated with its normal turnover 65 may be responsible for causing both Ihog internalization and degradation.
Consistent with a role for Ptc in mediating Hh-dependent degradation of Ihog, antibody staining of the wing imaginal disc revealed that Ihog protein levels are inversely correlated with Ptc protein levels, namely, higher in the P compartment, lower in the A compartment, and at their lowest in a region coinciding with the stripe of high-level Ptc protein expression in A cells at the A/P compartment boundary 42 (Fig. 5a, e; Supplementary Fig. 3). This

ARTICLE
Ihog protein expression pattern is not due to re-localization of Ihog proteins in the columnar epithelium of the Drosophila wing disc, as the same expression pattern of Ihog protein is observed in each single focus plane from the apical to the basal surface through the entire columnar epithelium (Supplementary Fig. 4; Supplementary Movies 1-3). Moreover, this Ihog protein expression pattern is not due to Hh-induced downregulation of the ihog transcripts, as in situ hybridization revealed a uniform expression pattern of the ihog transcripts in the wing imaginal disc (Fig. 5i). To confirm that the reduced level of Ihog protein at the A/P compartment boundary is indeed due to a posttranscriptional phenomenon, we also generated a transgene encoding wild-type Ihog tagged with green fluorescent protein (GFP) under the control of the ubiquitous promoter rp49 (L>Ihog-GFP) 66 and quantified its abundance in various regions of the wing imaginal disc (Fig. 5c, g). Low levels of Ihog-GFP (<1.5-fold endogenous Ihog) are expressed ubiquitously throughout the organism from this transgene. However, in wing discs from flies with L>Ihog-GFP, the abundance of Ihog-GFP was decreased in A cells at the A/P compartment boundary, precisely in those cells where transcription of the endogenous ptc gene is induced (Fig. 5c, g). Similar to ihog, we also found that the boi transcripts in the wing imaginal disc were uniform (Fig. 5i) and that the expression levels of Boi-GFP protein were reduced in A cells at the A/P boundary upon introduction of a L>Boi-GFP transgene (Fig. 5d, h). Remarkably, this negative correlation between Ihog and Ptc protein levels becomes more evident in wing discs from boi mutant larvae (Fig. 5b,  To further investigate whether Ptc and Hh expression can indeed influence Ihog protein levels in vivo, we generated GFPmarked clones of cells expressing Hh from a Gal4-responsive promoter; as A cells are Hh-responsive, A clones responded to Hh stimulation by expressing high levels of Ptc, and these cells indeed displayed reduced Ihog protein levels (Fig. 6a, e). No change in Ihog protein levels was noted in similar P clones (Fig. 6c, g), where cells normally express Hh but lack Ci and consequently are not responsive to the Hh signal and therefore do not express Ptc. We next generated GFP-marked clones of cells expressing Ptc. We found that Ptc expression indeed reduced Ihog protein levels in the P compartment, where Hh but not Ptc is normally expressed (Fig. 6d, h). We noted that the reduction of Ihog levels in posterior cells clonally activated to express Ptc is restricted to cells within the clone (Fig. 6d, h), whereas Ihog reduction in the anterior extends beyond cells clonally activated for expression of Hh (Fig. 6a). We interpret this to be due to Hh stimulation and Ptc expression in cells beyond the boundary of the Hh-expressing clone. These data indicate that internalization and degradation of Ihog protein is triggered cell-autonomously by expression of Ptc, which can however be stimulated in a cell nonautonomous manner by the secreted Hh protein signal.
We also generated Ptc-expressing clones in the anterior, far from Hh-expressing posterior cells, by expression of the UAS-Ptc transgene (Fig. 6b, f). Consistent with our cultured cell studies (Fig. 1), we noticed Ihog levels were slightly reduced within these Ptc-expressing clones, far away from a source of Hh ligand, suggesting that some downregulation of Ihog proteins can be mediated by Ptc alone, in the absence of Hh ligand.
Ihog mediates aggregation of Drosophila S2 cells. The reduced level of Ihog/Boi proteins at the A/P compartment boundary is intriguing in light of the long-known lineage restriction between A and P compartment cells [32][33][34] . Ihog/Boi proteins resemble cell adhesion molecules: both are type I single-span transmembrane proteins with 4 immunoglobulin and 2 fibronectin type III extracellular domains. However, whether Ihog/Boi proteins directly function in cell adhesion is not known. To address this possibility, we assessed the consequences of overexpressing Ihog or Boi in Drosophila S2 cells, which are inherently non-adhesive and have been used to demonstrate the adhesive properties of other membrane proteins 67 . Notably, we found that S2 cell cultures co-transfected with mCherry and either Ihog-expression or Boi-expression constructs formed multi-cellular aggregates (Fig. 7b-d), whereas S2 cell cultures transfected with mCherry alone did not (Fig. 7a, d). Within each experiment, only the cells transfected to express Ihog/Boi participated in aggregate formation and un-transfected cells were excluded (Fig. 7b-d), suggesting that aggregation is homophilic and requires expression of Ihog or Boi on each interacting cell. Ihog/Boi proteins, in addition to functioning as Hh receptor components 43 , thus also appear to function in cell adhesion.

Modulation of Ihog protein levels causes S2 cell segregation.
The expression levels of Ihog and Boi are lowest in anterior cells at the A/P compartment boundary as a result of Hh-triggered receptor internalization and degradation (Fig. 5). As a small difference in the abundance or activity of a single species of homophilic cell adhesion molecule between two interspersed populations of cells is predicted to cause segregation of the two populations 68 , it is conceivable that the modulation of the level of Ihog proteins at the A/P boundary may contribute to segregation of and lineage restriction between A and P cells. To simulate this situation, two populations of S2 cells were transfected separately ( Fig. 7e-h): one with constructs for expression of an Ihog-mCherry fusion and Ptc, to mimic A cells adjacent to the A/P compartment boundary (expressing high levels of pathway target genes, including Ptc), and the other with plasmids for expression of an Ihog-YFP fusion and Hh, to mimic Hh-secreting P cells. We note that as S2 cells lack endogenous expression of Ci, this experiment is not affected by transcriptional response but only by the role of Hh receptor binding in cell adhesion.
Forty-eight hours after transfection, cells were dissociated and mixed for 4 h to allow A-like and P-like cells to interact. Our expectation was that degradation of the Ihog/Ptc complex in Alike cells would be induced by Hh produced in P-like cells. We indeed found that P-like cells co-expressing Ihog-YFP and Hh (green) frequently aggregate only with themselves and sort out from the A-like cells co-expressing Ihog-mCherry and Ptc (red) (Fig. 7h, i). In parallel, we measured the expression level of both total (Supplementary Fig. 6A) and surface Ihog proteins ( Supplementary Fig. 6B-F) in the S2 cells used in cell-mixing assay before and 4 h after cell mixing. Consistent with what we presented above, the total Ihog protein is selectively reduced after mixing A-like and P-like cells. We indeed found that the Ihog protein localized on the surface of A-like cells was markedly reduced after mixing with P-like cells (Supplementary Fig. 6B, F). The significant difference in level of surface Ihog protein on Alike and P-like cells likely is the driving force for their segregation after mixing. Likewise, a greater decrease of the Ihog protein localized to the cell surface at the AP boundary was observed in vivo when we performed immunostaining using antibodies against the extracellular domain of Ihog in the absence of detergent (Supplementary Fig. 7). This result suggests that a difference in Ihog protein levels at the A/P boundary, especially at the cell surface, may suffice to cause differences in cell affinity in vivo, and could at least partially contribute to A/P cell segregation in wing discs (Fig. 7j).
Ihog proteins affect A/P cell segregation in vivo. Loss of Hh responsiveness in large clones of anterior cells results in invasion of posterior territory 35,36,40 , and similar behavior for large clones of cells lacking ihog/boi function has been previously reported 43,48 . Large clones of cells lacking ihog/boi and located near the A/P border thus were seen to reside within territory normally occupied by posterior compartment cells, even though expression of Ci clearly marked them as originating in the anterior compartment (Supplementary Fig. 8). To investigate whether this abnormal cell segregation behavior is due simply to loss of Hh signal transduction, we generated large clones of cells lacking not only Ihog/Boi but also PKA-C1; the latter causes Hh pathway activation through stabilizing effects on Ci, thus resulting in pathway activation in ihog/boi mutant clones [69][70][71][72][73] . We found that, although pka-C1 mutant A cells with functional Boi protein close to the AP boundary mix well with adjacent A cells (see rough boundary, Fig. 8a), the segregation defects of boi; ihog mutant A cells were not rescued by pka-C1 loss, as these triple mutant cell clones (boi; ihog; pka-C1) now form a smooth boundary with adjacent A cells (Fig. 8b, c).
One other unusual feature of these large triple mutant cell clones (boi; ihog; pka-C1) is their relative absence at more anterior locations in the A compartment. Sister clones (twin spots) derived from a single recombinant event are usually located adjacent or close to each other (85% in leg discs, higher in wing discs) 74 , regardless of their location within the disc. We noted this type of pairing where sister clones both retain boi + function (Fig. 8a, sister clones denoted by red and yellow asterisks). In contrast, ihog + clones arising in the A compartment away from the A/P boundary (marked by double GFP expression in the A compartment of the wing pouch-see yellow asterisks in Fig. 8c) were often found unaccompanied by sister clones when these sisters were expected to lack both ihog and boi function (boi; ihog; pka-C1; marked by absent GFP expression, red asterisks in Fig. 8c; see Supplementary Fig. 10A). Quantification shows a high statistical significance (p < 0.001) to the reduced frequency of pairing of such clones (Supplementary Fig. 10B).
Triple mutant clones appear not to have general survival or growth defects, as wing discs containing unpaired clones of double-intensity GFP expression invariably also contain triple mutant clones at the A/P boundary (red asterisks, Fig. 8c). Moreover, regardless of the number of orphan mutant clones present in the A compartment of the wing pouch (yellow asterisks, Fig. 8c), we generally observe a single triple mutant clone straddling the AP boundary (red asterisk, Fig. 8c; Supplementary Fig. 11). The observed mismatch in location and numbers of triple mutant and their sister clones suggests that the triple mutant cells do not mix well with surrounding wildtype cells, and instead appear to preferentially associate with each other and localize to the A/P boundary. In addition, the separation of triple mutant clones from their sister clones was only observed when large mosaic clones were induced very early during development, which indicates that the clonal segregation is a fairly slow process. Nevertheless, the predominant location of triple mutant clones at the A/P boundary suggests a preference of these clones for the location in the disc having the lowest level of Ihog/Boi protein.
As another approach to test whether pathway activation can reverse the segregation defects of ihog/boi double mutant clones we examined the effects of MARCM 75 -activated expression of a constitutively active form of Smo (UAS-SmoGlu) 76 in large A clones near the A/P compartment boundary of the wing pouch. We found that pathway activation mediated by UAS-SmoGlu can rescue the abnormal segregation of smo mutant A clones (compare Supplementary Fig. 11A, B), but not of large boi; ihog mutant A clones near the A/P compartment boundary (compare Note that boi; ihog mutant cells form a sharp boundary (yellow line) with adjacent A cells (yellow asterisks). c Note that boi; ihog pka-C1 triple mutant clones are absent in the A compartment away from the A/P boundary of the wing pouch even with the presence of three wild-type sister clones (yellow asterisks); these triple mutant clones typically accumulate in a large clone at the A/P boundary (see text). The boi; ihog pka-C1 triple mutant clone originating in the posterior compartment of the wing pouch is indicated with a green asterisk and its wild-type sister clone is indicated by a blue asterisk. Scale bar, 50 µm. d Schematic summary of Ihog protein activity. In the absence of Hh, Ihog/Boi interacts with Ptc to form a receptor that is presented on the cell surface 43 . When Hh ligand is present, it binds this receptor and blocks the inhibition of Smo by Ptc, thus permitting activation of downstream pathway components. Binding of the Hh ligand induces ligand:receptor internalization and degradation of the ligand: receptor complex, preventing further action of Hh on distant cells. Ligandinduced receptor clearance also creates a discontinuity in the Ihog protein levels at the A/P compartment boundary, which may contribute to cell segregation and lineage restriction at the compartment boundary by reducing cell adhesion Supplementary Fig. 11C, D-F). Most of the clones we examined were large, but we did note that smaller clones generated later during development are more likely to respect the A/P boundary, regardless of downstream pathway activation (see for example, Fig. 2b). The variation in segregation behavior related to clone size might explain why a previous study that examined mostly smaller clones failed to detect abnormal A/P segregation of clones lacking Ihog/Boi 56 . Although the MARCM system does not permit simultaneous visualization of twin spot sister clones, we did find that boi; ihog; UAS-SmoGlu clones are less frequently localized to the anterior compartment of the wing pouch, consistent with the abnormal segregation of sister boi; ihog; pka-C1 mutant clones from ihog + clones noted above. The inability of downstream pathway activation to rescue abnormal segregation behavior of ihog; boi clones at the A/P boundary (whether by pka-C1 loss or MARCM-induced expression of UAS-SmoGlu), the dissociation of boi; ihog mutant clones from their ihog + sisters at a fivefold higher frequency than that previously reported for separation of wild-type twin spots, and the apparent migration and fusion of triple mutant clones at the A/P boundary together suggest that segregation defects of triple mutant cells are not due to a failure to transduce the Hh signal and that Ihog/Boi proteins likely contribute to A/P cell segregation in wing discs independently of their function as Hh receptor components ("Discussion").

Discussion
The Hh receptor component Ptc has two distinct functions: one in signaling via regulation of Smo and the other in Hh sequestration 40 . Previously, we demonstrated that Ihog/Boi interacts directly with Ptc, is required for presentation of Ptc on the cell surface, and that Ihog and Ptc are both required for high-affinity Hh binding 43 . This crucial role of Ihog/Boi in binding of Ptc to Hh suffices to explain why Ihog/Boi is indispensable for transduction of the Hh signal. With regard to Hh sequestration, restricting the range of Hh action in Drosophila wing imaginal discs requires depletion of Hh produced in P cells by the A cells immediately adjacent to the A/P boundary. These A cells not only must bind Hh but also eliminate it from the cell surface to prevent its further action upon dissociation 40,41 . Here, we provide in vitro (Fig. 1) and in vivo (Fig. 2) evidence to show that Ihog/ Boi is essential for Ptc-mediated Hh endocytosis, as well as for degradation of Ptc following Hh binding. Taken together, our evidence shows that the Ihog/Boi co-receptor, like Ptc itself, has dual roles in transducing and sequestering the Hh signal. Notably, Ihog/Boi activity in A cells: (i) enhances Hh binding and pathway activation, thus increasing the levels of Ptc expression that aid in adsorption of the Hh ligand 43 ; and (ii) aids in Ptc-mediated Hh internalization and degradation upon binding, making the process of sequestration irreversible.
Ihog/Boi proteins contribute both to presentation of Ptc on the surface 43 and to Hh-dependent Ptc degradation (Figs. [3][4][5]. We note that Ihog itself is also internalized and degraded in a Hhdependent manner, and that this activity is critically dependent upon co-expression of Ptc (Figs. 1, 3, 6, and 7). Ihog and Ptc thus interact physically, and Ihog and Ptc proteins are mutually required for surface presentation of the Hh receptor, for highaffinity binding 43 and for Hh-dependent receptor internalization and degradation. These activities are summarized schematically in Fig. 8. On this basis and because of the absolute requirement for Ihog/Boi in all Hh biological responses and Hh sequestration, we consider Ihog to be an integral component of the Drosophila Hh receptor.
Despite a direct interaction between Ihog and Ptc and their joint contributions to formation, surface presentation, and ligand-dependent internalization of an active receptor, it is curious that these two pathway components have opposing effects on pathway activity. Ihog proteins thus are indispensable for positive response to Hh ligand, whereas Ptc in the absence of Hh ligand suffices to suppress Smo and pathway activity (Fig. 8). These two components are also distinctly regulated, with Ptc transcription increasing upon Hh stimulation, whereas Ihog protein levels decrease post-transcriptionally due to Hh-dependent and Ptcdependent internalization and degradation; the distinct levels of Ihog/Boi protein across the wing disc thus is wholly predictable from the expression of Ptc and Hh, whether in their normal patterns or as altered by experimental manipulation.
With regard to cell segregation and lineage restriction at the A/ P compartment boundary, the current view is that response to Hh signaling induces a change in adhesive properties of anterior cells abutting the compartment boundary, and that this difference leads to the segregation of A cells and P cells, and consequent creation of a lineage restriction. Differences in adhesiveness between cells could result from the presence of distinct types of cell adhesion molecules, or from different levels of the same cell adhesion molecule. However, despite extensive previous efforts to identify cell adhesion molecules that specify distinct A and P cell affinities, including systematic analyses of Drosophila cadherinlike proteins and screens for mutations that affect cell segregation 77,78 , the identity of these hypothesized molecules remains unknown 77 .
Our finding that Ihog/Boi-expression can mediate aggregation of S2 cultured cells and that expression of Ihog/Ptc and Ihog/Hh in S2 cells to simulate A and P cell-surface character can cause segregation of these A-and P-like cells in vitro suggest that Ihog/ Boi proteins may contribute to A/P cell segregation in vivo. The in vivo role of Ihog/Boi proteins in cell adhesion indeed is supported by our analysis of the boi; ihog mutant clones with pathway activation by pka-C1 loss or MARCM-induced expression of UAS-SmoGlu originating near the A/P boundary. In these clones, the expression of downstream Hh pathway target genes induced by loss of pka-C1 or by MARCM-induced expression of UAS-SmoGlu fails to rescue the cell segregation defects of boi; ihog mutant A cells, which remain segregated from adjacent A cells (Fig. 8b, c; Supplementary Fig. 11D-F). In addition, boi; ihog; pka-C1 triple mutant clones originating in the A compartment, away from the A/P boundary, preferentially segregate away from their wild-type sister clones and appear to localize at the A/P boundary (Fig. 8c). This behavior contrasts sharply with that of wild-type clones 74 , and was not observed when either ihog/boi or pka-C1 were mutated independently ( Fig. 8a and Supplementary  Fig. 9; see also refs. 43,[69][70][71][72]. If surface expression of Ihog/Boi specifies critical cell adhesion properties, why do boi; ihog pka-C1 clones segregate away from their sister clones in the anterior whereas boi; ihog clones do not? We would speculate that this behavioral difference may be due to loss of pka-C1 function, which enhances cell proliferation [69][70][71][72] . This proliferative difference could facilitate the initiation of movement for these triple mutant clones (Ganhui Lan, 2017, personal communication), whose eventual relocation to the A/P boundary may then be accounted for by the low level of Ihog/Boi adhesion proteins on cells at this location. Alternatively, it is possible that other cell adhesion molecules may contribute to cell segregation behavior, as posterior cells lacking Ihog/Boi function fail to invade the anterior compartment ( Supplementary  Fig. 12). Indeed, such a functionally overlapping role would be consistent with the suggestion of Vegh and Basler 77 that multiple cell adhesion molecules may be involved in segregation of A/P compartment cells. Further, multiple factors in addition to cell adhesion differences may contribute to the boundary characteristics of clones, including internal forces mediated by cytoskeletal elements or by cell-cell adhesion, orientation of cell division and differential proliferation rates that generate patterns of mechanical tension [79][80][81][82] . It nevertheless remains striking that the Ihog/ Boi proteins, shown here to mediate cell adhesion in vitro and to affect cell segregation under defined circumstances in vivo, show their greatest difference in expression between A cells and P cells at the compartment boundary of the wing imaginal disc.

Methods
Constructs. The L>Ihog-GFP and L>Boi-GFP transgenes were constructed by cloning Ihog/Boi-GFP fusion between the rp49 promoter and the hsp70 3′UTR in a pCasper vector 66 . Expression constructs of Ihog/Boi, Ptc variants used in Drosophila cell culture were fused in frame with a C-terminal HA tag, 3xMyc tag, GFP tag or mCherry tag, and cloned into the pAcSV plasmid. Constructs with Actin-Gal4 in pCasper vector and Rab5-GFP in the pUAST vector (a gift from M. Scott) were used to express Rab5 in S2R + cells.
Cell immunostaining. Forty-eight hours after transfection, cells were incubated 1 h with conditioned medium containing HhN protein or control conditioned medium. Cells were then washed twice with PBS, fixed in 4% formaldehyde (Ted Pella) in PBS, blocked by 1.5% normal goat serum (NGS) in PBS, incubated with primary antibody in PBS containing 1.5% NGS for overnight at 4°(to stain surface Ihog/ Ptc/HhN), permeabilized by 0.3% Triton X-100/PBS for 15 min at room temperature, incubated with primary antibody in PBS containing 1.5% NGS and 0.3% Triton X-100 for 1 h at room temperature (to stain intracellular Ihog/Ptc/HhN), washed three times with 0.3% Triton X-100/PBS, incubated with secondary antibody and washed with 0.3% Triton X-100/PBS. Lysosome staining. S2R+ cells transfected with Ihog-YFP and Ptc-mCherry were incubated in HhN-conditioned medium or control medium for 1 h, live stained with 50 mM LysoTracker Deep Red (Thermo Fisher L12492) in media for 1 h at room temperature, washed with PBS for three times, and imaged with a confocal microscope.
Cell-surface biotinylation. Cells were incubated in HhN-conditioned medium or control medium for 1 h, washed with PBS and incubated for 30 min in PBS containing 1 mg/ml Sulfo-NHS-LC-Biotin (Pierce, 21335). The reaction was quenched by washing the cells with 100 mM glycine in PBS three times. Cells were then lysed in 0.5% Digitonin (50 mM Tris-HCl pH = 7.5, 150 mM NaCl, and protease inhibitors) for 30 min. The lysate was clarified by centrifugation, and biotinylated protein was recovered by binding to Streptavidin-Sepharose beads (GE healthcare, 17-5113-01) for 2 h at room temperature. Beads were washed, and protein was recovered in SDS-PAGE sample buffer. Proteins samples were resolved by SDS-PAGE and transferred to PVDF membranes (Millipore) for western blot analysis. Uncropped western blot images can be found in Supplementary Fig. 13.
RNAi. Primer pairs with each primer encoding a T7 promoter sequence and genespecific sequence were used to amplify PCR sequence product from cDNAs to generate templates for in vitro dsRNA synthesis using T7 RNA polymerase (Ambion). See Supplementary Table 1 for a list of primers used. S2R+ cells were bathed in 15 μg/ml of dsRNA in serum-free medium for 1 h followed by addition of 10% FBS and transfection with 0.5 g of dsRNA using Effectene (Qiagen). Cells were incubated for an additional 2 days to allow protein turnover prior to treatment with control or HhN-containing medium.
In situ hybridization. RNA probes were created from in vitro transcription of PCR products carrying the T7 RNA polymerase recognition sequence at one end and synthesized by using a digoxigenin (Dig)-labeling kit (Roche). Wing discs of L3 larvae were hybridized with probes overnight at 56°C using standard procedures and visualized using anti-Dig-AP (1:1,000; Roche). Primers used for generating PCR templates are listed in Supplementary Table 2.
Drosophila strains. Mutant and transgenic strains are listed in Supplementary  Table 3.
Genotype of larvae for generating mosaic clones. Wing imaginal disc clones were generated by FRT-FLP recombination. Fly embryos or first-second instar larvae were heat shocked for 1 h at 37°C to produce large or smaller mosaic clones 40,75 . The genotypes of larvae for generating mosaic clones are listed in Supplementary Table 4.
Image collection and quantification of fluorescence intensity. To assess protein amounts in different subcellular domains, images used for comparison within an experiment were obtained with identical settings on a Zeiss Spinning Disc microscope and then used for quantification without any manipulation. A mask was constructed by manually outlining the cells in the image taken in the Rab5 or Lyso channel, for example. This mask was then applied to the image taken in the channel for the protein of interest, and the fluorescence was measured. Local background correction was performed by moving the mask to measure fluorescence at a representative nearby region and then subtracting this value from that of interest. All bars represent mean ± s.d. from 20 individual cells. Statistical analysis was performed using GraphPad Prism software.
To assess protein amounts in different wing imaginal discs, all wing discs from third instar larvae were dissected, fixed, stained in the presence or absence of detergent, and imaged in parallel using the same confocal microscope settings. To quantify Ihog, Ptc, and GFP-staining intensities, a rectangle was selected and centered at the AP boundary across both ventral and dorsal compartments. Average pixel intensity was determined using the Plot Profile function of ImageJ and plotted using GraphPad Prism software.
Cell aggregation assay. S2 cells were transfected separately with plasmids expressing desired proteins. Forty-eight hours after transfection, S2 cells were washed with PBS and dissociated by 0.05% trypsin treatment for 5 min at 25°C. The dissociated cells were resuspended in the medium with 10% fetal bovine serum, and differentially transfected cells were mixed in 1.5 ml eppendorf tubes. After rotating at room temperature for 4 h to allow aggregation to occur, cells were transferred into the wells of culture slides for live imaging by confocal microscopy.
Statistical analyses. All data in column graphs are shown as mean values with s.d. Statistical analyses were performed with Mann-Whitney's U-test as indicated in each figure legend. The sample sizes were set based on the variability of each assay and are listed in the figure legends. Independent experiments were performed as indicated to guarantee reproducibility of findings.
Data availability. The authors declare that all data supporting the findings of this study are available within the article and its Supplementary Information files or from the corresponding authors upon reasonable request.