Metazoan Hedgehog (Hh) morphogens are essential regulators of growth and patterning at significant distances from their source, despite being produced as N-terminally palmitoylated and C-terminally cholesteroylated proteins, which firmly tethers them to the outer plasma membrane leaflet of producing cells and limits their spread. One mechanism to overcome this limitation is proteolytic processing of both lipidated terminal peptides, called shedding, but molecular target site requirements for effective Hh shedding remained undefined. In this work, by using Drosophila melanogaster as a model, we show that mutagenesis of the N-terminal Cardin–Weintraub (CW) motif inactivates recombinant Hh proteins to variable degrees and, if overexpressed in the same compartment, converts them into suppressors of endogenous Hh function. In vivo, additional removal of N-palmitate membrane anchors largely restored endogenous Hh function, supporting the hypothesis that proteolytic CW processing controls Hh solubilization. Importantly, we also observed that CW repositioning impairs anterior/posterior compartmental boundary maintenance in the third instar wing disc. This demonstrates that Hh shedding not only controls the differentiation of anterior cells, but also maintains the sharp physical segregation between these receiving cells and posterior Hh-producing cells.
Hedgehog morphogen family members are essential determinants of metazoan body plans. In Drosophila melanogaster, tissue patterning is achieved by only one Hedgehog protein (called Hh throughout this work), whereas vertebrates express three homologs: sonic hedgehog (Shh), desert hedgehog (Dhh) and Indian hedgehog (Ihh). Consistent with the essential signaling functions of Hh, its inactivation causes severe mispatterning during vertebrate and invertebrate development, and Shh misexpression can also contribute to various cancers (Briscoe and Thérond, 2013).
In Drosophila, Hh acts as a short-range signaling molecule during embryo development and later in the imaginal eye discs, and it also acts as a long-range signaling molecule in wing discs, which therefore serve as robust and reliable readouts for Hh biofunctions (Crozatier et al., 2004; Hartl and Scott, 2014). Notably, these Hh biofunctions depend on an unusual series of post-translational modifications. In a first step, 45 kDa Hh precursor biosynthesis and secretion into the endoplasmic reticulum is followed by its autocatalytic cleavage, which is coupled to the C-terminal covalent attachment of a cholesterol moiety to the 19-kDa Hh signaling domain (Lee et al., 1994; Porter et al., 1996). Next, a membrane-bound acyltransferase [called Skinny Hedgehog (Ski; also known as Rasp) in Drosophila or hedgehog acyltransferase (Hhat) in vertebrates] attaches a palmitate to conserved N-terminal cysteines (C85 in fly Hh, C25 in murine Shh) (Chamoun et al., 2001; Pepinsky et al., 1998). These lipidations firmly tether all secreted vertebrate and invertebrate Hhs to the outer plasma membrane leaflet of the cell that produces them; lipidations also determine the later signaling range and activity of Hhs, because Ski/Hhat inactivation or the replacement of the N-terminal cysteine acceptor with serine or alanine (C25→A/S in ShhC25A/S, C85→A/S in Drosophila HhC85A/S) abolishes most Hh biofunction in vitro and in vivo (Chamoun et al., 2001; Chen et al., 2004; Dawber et al., 2005; Goetz et al., 2006; Kohtz et al., 2001; Lee et al., 2001; Pepinsky et al., 1998). Finally, Hhs assemble into large clusters by using linear heparan sulfate (HS) chains of cell-surface proteoglycans as scaffolds (Vyas et al., 2008). These essential and highly conserved post-translational modifications raise the important questions of how membrane-tethered Hh clusters can reach target cells at a significant distance from the source and why palmitate is crucial for their unimpaired biofunction.
One widely accepted idea is that lipidated Hh moves on soluble lipophilic carriers called exosomes (Gradilla et al., 2014) or on filopodia-like extensions called cytonemes (Bischoff et al., 2013; Kornberg, 2014). Indeed, live-cell imaging of GFP-tagged morphogens in imaginal wing discs demonstrated that cytonemes can emanate from the anterior receiving compartment to take up Hh-GFP from posterior producing cells (Chen et al., 2017) or can emanate from the producing compartment to deliver the morphogen to receiving cells or cytonemes (Bischoff et al., 2013; Gradilla et al., 2014). Juxtamembrane proteolytic processing of both lipidated peptides, called shedding, has also been observed to release N- and C-terminally truncated yet bioactive Hh in vitro (Dierker et al., 2009; Jakobs et al., 2014; Ohlig et al., 2011). Although shedding provides a simple explanation of how lipidated Hh can act over long distances, this concept was initially greeted with skepticism because the functionally essential N-palmitate moiety gets removed in the process. This problem was later resolved by structural and biochemical assays showing that unprocessed N-terminal Hh peptides block binding sites of adjacent Hh molecules for the receptor Patched (Ptc), thereby rendering them inactive (Ohlig et al., 2011). One consequence of subsequent juxtamembrane cleavage, therefore, is that it removes the N-terminal inhibitory peptide during release and converts inactive Hh clusters into active soluble forms. The essential role of N-palmitate in this scenario is therefore indirect: with it, only quantitatively processed (i.e. fully activated) Hh clusters are released, and lack of N-palmitate (upon Hhat/ski inactivation or ShhC25A/S/HhC85A/S mutagenesis) results in the solubilization of C-terminally processed, but N-terminally unprocessed (i.e. inactive), clusters. Most recently, by uncovering a dominant-negative, cell-autonomous function of overexpressed HhC85A/S on endogenous Hh in the Drosophila melanogaster wing disc, we confirmed the above-described N-palmitate-dependent Hh processing and activation in vivo (Schurmann et al., 2018). We also observed that relocating a stretch of basic amino acids located close to the N-palmitate, called the Cardin–Weintraub (CW) motif, completely abolished palmitoylated Hh processing and bioactivation. Although both observations strongly supported that N-terminal peptide sequence or structure regulate Hh function in the Drosophila wing disc, a detailed molecular and functional characterization of the putative N-terminal sheddase target site, as well as an investigation on Hh shedding in developmental systems other than the wing disc, had not been conducted so far.
Therefore, in this work, we used site-directed mutagenesis to characterize N-terminal Hh cleavage in established and newly developed in vivo models of Hh signaling, while leaving Hh lipidation and multimerization unaffected. We determined that the CW motif serves as the preferred N-terminal cleavage site and found that removing, replacing or relocating this motif abolished most Hh function over the short range in eye discs and over the long range in wing discs. Moreover, we confirmed that impaired processing of overexpressed CW variants suppressed the activity of endogenous Hh that was ‘trapped’ in the same insoluble clusters. This dominant-negative effect of overexpressed CW variants resulted in severely impaired patterning of anterior wing structures or even in the complete lack of these structures in the most extreme cases. Importantly, ‘trapping’ of endogenous Hh in the wing disc also produced a tendency of anterior disc cells,which express the selector gene cubitus interruptus (ci; the protein product Ci being required for Hh reception; Alexandre et al., 1996), to extend into the posterior disc compartment, which expresses the selector gene engrailed (en; the protein product En being required for Hh production; Tabata et al., 1992). These findings support the suggestion that the wing disc boundary that segregates anterior Ci-expressing cells from posterior En-expressing cells into adjacent but immiscible cell populations depends on sustained local Hh signaling at the anterior/posterior (A/P) border (Blair and Ralston, 1997; Dahmann and Basler, 2000; Rodriguez and Basler, 1997; Rudolf et al., 2015), in contrast to being solely determined by selector gene expression as originally postulated. Our data therefore show that Hh shedding is absolutely required to maintain the A/P boundary in the Drosophila wing disc.
N-palmitate and CW amino acids regulate Hh solubilization in vitro
The polybasic CW motif (Cardin and Weintraub, 1989) is located eight or nine amino acids downstream of the conserved N-terminal acylation site in all Hh family members (Fig. 1A, Fig. S1A). One established role of this motif is to interact with sulfated, negatively charged HS chains (Farshi et al., 2011; Rubin et al., 2002) in flies (Bellaiche et al., 1998; Desbordes and Sanson, 2003) and in mice (Grobe et al., 2005). Notably, structural analysis of human SHH (pdb: 3m1n, Fig. 1A) revealed that its N-terminal peptide extends 30 Å away from the globular signaling domain (Pepinsky et al., 2000), thereby exposing it to HS at the cell surface or to cell surface proteases that release the morphogen (Ohlig et al., 2012). To confirm proteolytic CW targeting, we expressed control Drosophila Hh and recombinant Hh variants that have this site modified (Fig. 1A) from pUAST-rfA-attB vectors under the control of actin-Gal4 in Drosophila Schneider (S2) cells, and we determined the relative amounts of soluble protein versus the material at the cell surface (Movie 1). As shown in Fig. 1B, Hh and Hh3xA, the latter of which has basic CW residues 93, 95 and 97 replaced by alanines, were efficiently solubilized. By contrast, we observed less soluble HhΔCW (all CW residues deleted), HhΔCW-HA [CW residues replaced by a hemagglutinin (HA) tag] and HhHA1/HhHA2 (inserted HA tags upstream or downstream of the CW site) (Fig. 1B, Fig. S1B), as well as impaired release of similar SHH variants from human Bosc23 cells (Fig. S1C,D). To rule out that Hh release was affected because of its impaired multimerization or impaired interactions with HS (Vyas et al., 2008), we analyzed the former by size exclusion chromatography and the latter by HS affinity chromatography (Fig. 1C,D, Fig. S1E). We observed that HhNC85S (an artificially produced monomeric control protein that lacks both terminal lipids) and HhN (another control protein that lacks the C-cholesterol) did not cluster, as expected (Fig. 1C), in contrast to cholesteroylated Hh and Hh3xA. The binding of Hh/Hh3xA and HhN/HhN3xA to physiological HS isolated from fly larvae (Fig. 1D) and to a highly sulfated form of HS called heparin (Fig. S1E) was also similar, suggesting that the CW motif of Drosophila Hh is not essential for HS binding and HS-regulated multimerization, at least in vitro (Fig. S2). Instead, reduced solubilization of CW variant proteins from the surface of cells that express them (Fig. S3) suggests that this site may be a protease target required for Hh release (Jakobs et al., 2014; Ohlig et al., 2012; Ortmann et al., 2015) and that non-cleavable proteins remain permanently membrane tethered by their N-palmitate (Farshi et al., 2011). Consistent with this, the additional deletion of palmitate acceptor cysteines increased the solubilization of HhΔCW, HhΔCW-HA and HhHA1/HhHA2 (Fig. S1B), as well as that of Shh/ShhHA1 (Fig. S1C,D). From these data, we suggest that N-palmitate and CW positioning control Hh release in vitro.
The CW motif is required for short-range Hh function in the Drosophila eye disc
The Drosophila eye is a honeycomb matrix of ∼750 photoreceptors (ommatidia) that develops in a wave of differentiation that moves from the posterior to the anterior of the eye disc, called the morphogenetic furrow. Cells located anterior to the furrow respond to Hh secreted by cells posterior to the furrow by eventually producing the same protein. This generates a cyclic short-range Hh signaling mode that progresses the furrow across the disc (Ma et al., 1993) and determines the number of ommatidia in the adult eye (Fig. 2A). Therefore, diminished Hh production in homozygous hhbar3 eye discs delays furrow progression and results in ∼68% smaller eyes (consisting of ∼500 ommatidia), which are smaller still if in trans with hhAC, a homozygous lethal null mutation (hhbar3/hhAC eyes consist of only ∼186±26 ommatidia/eye; n=10 male and 10 female eyes analyzed, P≤0.001, Fig. 2B) (Rogers et al., 2005). To rescue this phenotype, we used an established eye disc-specific GMR (glass multimer reporter)-Gal4 driver (Brand and Perrimon, 1993) to specifically express recombinant Hh and its CW variants from the same attP 51C landing site (Bateman et al., 2006) in hhbar3/hhAC eye discs. We found that GMR-Gal4-controlled Hh and Hh3xA expression restored most eye development (n=711±56 ommatidia/eye and n=614±29 ommatidia/eye, respectively) (Fig. 2C,D), whereas HhHA1 and HhΔCW were less effective (n=266±36 ommatidia/eye and 391±19 ommatidia/eye, respectively, n=20) (Fig. 2E,F). One important finding in this system was that the additional deletion of N-palmitate increased HhC85S;HA1 biofunction (n=440±26 ommatidia, n=20) over that of HhHA1, and HhC85S was also weakly active (n=383±31 ommatidia/eye, n=20) (Fig. 2G,H). This is remarkable, because non-palmitoylated Drosophila HhC85S was previously considered completely inactive (Dawber et al., 2005; Lee et al., 2001). No sex differences were observed (Fig. 2I). From these findings, we suggest that site-directed CW mutagenesis affects Hh short-range signaling in the eye disc to variable degrees, depending on the type of mutation, and that the additional deletion of palmitate increases the activity of HhHA1. This latter observation suggests that CW and palmitate biofunctions are somehow linked.
Presence and positioning of the CW motif affect Hh long-range signaling
We further analyzed this possible link in the Drosophila wing disc. Hh is produced in the posterior (P) wing disc compartment under the control of Engrailed (En in the P compartment is labeled green in Fig. 3A) (Tabata et al., 1992; Zecca et al., 1995) and spreads across the A/P boundary into the anterior (A) compartment. There, Hh binds to the receptor Ptc (Ingham et al., 1991), which releases the transmembrane protein Smoothened (Smo) from Ptc repression and patterns the central L3-L4 intervein region of the adult wing (Fig. 3A, bottom, labeled red) (Mullor et al., 1997; Strigini and Cohen, 1997). Consistent with this mechanism, en-Gal4 (Brand and Perrimon, 1993) induced expression of UAS-controlled recombinant Hh (Fig. 1A), expanded the L3-L4 intervein area, and, as a concomitant effect, reduced the L2-L3 intervein space (Fig. 3B, arrows) (Crozatier et al., 2004; Lee et al., 2001; Mullor et al., 1997; Strigini and Cohen, 1997). This phenotype is consistent with higher Hh concentrations required for Ptc activation and L3-L4 development than for the activation of Decapentaplegic (Dpp), which patterns the remainder of the wing (Hooper, 2003; Méthot and Basler, 1999; Mohler et al., 2000; Strigini and Cohen, 1997; Vervoort et al., 1999). Moreover, en-Gal4-regulated posterior overexpression of non-palmitoylated HhC85S under UAS control (en>HhC85S) apposed veins L3 and L4 (Fig. 3C, arrows) (Crozatier et al., 2004; Lee et al., 2001), supporting the conclusion that unprocessed HhC85S N-peptides block Ptc binding of endogenous Hh, as described previously (Schurmann et al., 2018) and as outlined in Fig. S4.
We next analyzed all of our CW variants in this system. With the exception of en>Hh3xA (Fig. 3D), all transgenic overexpressed palmitoylated proteins acted as dominant-negative suppressors of endogenous Hh function (Fig. 3 E-H; Fig. S5 details the observed phenotypic variability). Vein mispatterning phenotypes ranged from mild in en>HhΔCW;HA and en>HhΔCW wings (Fig. 3E,F), comparable to en>HhC85S phenotypes (Fig. 3C) (Crozatier et al., 2004; Lee et al., 2001), to severe in en>HhHA2 and extreme in en>HhHA1 wings (Fig. 3G,H). The en>HhHA1 phenotype is comparable to wing patterning phenotypes of flies expressing permanently membrane-tethered HhCD2 (Strigini and Cohen, 1997) or of those with abolished Hh signaling (Ascano and Robbins, 2004; Vervoort et al., 1999). Notably, additional C→S mutagenesis converted extreme en>HhHA2 and en>HhHA1 phenotypes into milder forms (Fig. 3I-L; quantified L3/L4 versus L2/L3 ratios are shown in Fig. 3M) and also rescued pharate lethality: whereas only 7% of imagos eclosed from 467 en>HhHA1 pharates at 25°C, 77% of imagos eclosed from 372 en>HhHA1;C85S pharates at the same temperature. Pharate lethality was also reduced in en>HhC85S;ΔCW- and en>HhC85S;ΔCW-HA-expressing lines compared with that in lines producing the corresponding palmitoylated proteins (Fig. 3N), and these findings were further confirmed by alternative posterior driver lines en(2)-Gal4 and hh-Gal4 (Fig. S6). We also confirmed that observed dominant-negative activities of transgenic overexpressed palmitoylated and non-palmitoylated CW variants on endogenous Hh function required direct physical interactions between both proteins, because recombinant protein overexpression under the control of ptc-Gal4 in a stripe of anterior cells (where endogenous Hh is not expressed) did not result in the same phenotypes (Fig. S7). From these experiments, we suggest that overexpressed palmitoylated Hh CW variants suppress Hh-regulated wing patterning as a consequence of their impaired processing and solubilization (Fig. 1): if overexpressed together with endogenous Hh in the same P cells, impaired processing of palmitoylated membrane anchors also prevents the solubilization of endogenous associated protein in the same cluster (even if these were dually processed). By removing the palmitate anchor in C85S recombinant proteins, their N-terminal processing is no longer required for protein solubilization. Therefore, extreme dominant-negative protein activities resulting from impaired cluster release are converted into milder forms resulting from partial blockade of Ptc-binding sites (Fig. S4) (Ohlig et al., 2011; Schurmann et al., 2018). This is exactly what we observed. Impaired palmitoylated Hh release was further supported by Gal4 driver lines 34B, 45433 and 45105, which overexpress recombinant Hh in anterior tissues or in tissues located outside of the wing disc proper (Fig. 4A,H,L) (Brand and Perrimon, 1993). Hh overexpression controlled by these driver lines results in phenotypes resembling a natural hh gain-of-function allele, hhMoonrat (Tabata and Kornberg, 1994), which is characterized by overgrowth of anterior wing tissue (Fig. 4B,I,M, arrows). With the exception of 45433- and 45105-controlled Hh3xA overexpression (Fig. 4J,N), we consistently observed biological inactivity of all palmitoylated CW variant proteins in these three systems (Fig. 4C-G,K,O), in accordance with their reduced solubilization.
Posterior expression of palmitoylated and non-palmitoylated CW variants impairs Hh target gene transcription
We next confirmed inhibited Hh signaling at the molecular level, using wing-disc Hh target gene expression as a readout (Chen and Struhl, 1996; Strigini and Cohen, 1997). In the normal situation, Hh is produced in the posterior compartment and transports into the anterior compartment to induce secondary En expression in a thin stripe at the A/P compartment boundary (Fig. 5A). Therefore, in addition to driving posterior Hh expression, en also represents an anterior high-threshold target gene (Guillen et al., 1995; Strigini and Cohen, 1997). In addition, the Hh receptor Ptc is upregulated in a ten-cell-wide stripe anterior to the A/P border by high/medium Hh concentrations, and Dpp is expressed in a 12- to 15-cell-wide stripe in response to low Hh concentrations (Strigini and Cohen, 1997). En, Ptc (Fig. 5) and Dpp (Fig. S8) expression can therefore serve as direct readouts for Hh biofunction in the wing disc. Indeed, compared with the wild-type situation (Fig. 5B-B″), en>Hh overexpression expanded anterior PtcLacZ and En production areas (Fig. 5C-C″), as expected from the resulting expansion of L3/L4 intervein tissue (Fig. 3B), and HhC85S overexpression reduced PtcLacZ expression (Fig. 5D-D″), in line with the observed wing phenotypes (Fig. 3C). Notably, additional removal of the first 15 N-terminal amino acids (HhC85S;Δ86-100) restored most PtcLacZ expression (Fig. 5E-E″), confirming the concept that unprocessed N-terminal peptides inhibit Hh function (Fig. S4). PtcLacZ expression was reduced in response to en>Hh3xA expression (Fig. 5F-F″), and the intensity and range of En and PtcLacZ signals in response to posterior en>HhΔCW and en>HhC85S;ΔCW expression were reduced even further (Fig. 5G-H″). As expected from the most strongly impaired wing development (Fig. 3H), overexpressed HhHA1 completely suppressed the PtcLacZ signal in the pouch. Unexpectedly, it also restricted En expression to the most posterior domain (Fig. 5I-I″, asterisk). Deletion of N-palmitate in en>HhC85S;HA1 increased En and PtcLacZ expression (Fig. 5J-J″), again consistent with partially restored wing development (Fig. 3L). Posterior En expression in en>HhHA2 discs was also restricted posteriorly, and PtcLacZ was expressed in a thin stripe at a significant distance from the posterior En expression domain (Fig. 5K-K″, asterisk). In contrast, HhC85S;HA2-expressing discs (Fig. 5L-L″) were similar to HhC85S- and HhC85S;ΔCW-expressing discs (Fig. 5D-D″,H-H″; quantified PtcLacZ signals are shown in Fig. 5M and a complete set of averaged PtcLacZ expression profiles is shown in Fig. S9). Finally, although the low-threshold Hh target gene dpp remained mostly unimpaired (Fig. S8), consistent with only moderately affected development of the Dpp-regulated L2/L3 intervein spaces (Fig. 3D-L), we also noticed an apparent HhHA1/2-induced migration of the posterior En-expression boundary away from the stripe of DppLacZ expression (Fig. S8J-K″, asterisks).
En retraction and Ci expansion in cells located posterior to the stripe of PtcLacZ expression
We investigated the unexpected apparent posterior shift in the boundary that separates En-expressing from non-expressing cells and that generates an area of En-OFF cells located posterior to the stripes of PtcLacZ (Fig. 5K-K″, asterisk) and DppLacZ (Fig. S8J-K″, asterisks) expression. As described previously, in the wild-type situation, posterior disc cells express En to drive Hh production but also to block the expression of Dpp directly and indirectly by the suppression of ci transcription (Sanicola et al., 1995; Schwartz et al., 1995; Tabata et al., 1992). In effect, this mechanism restricts ci/dpp transcription to anterior cells (Fig. 6A-A‴), and the inactivation of En and its paralog Invected (Inv) in the posterior compartment can result in Ci and Dpp misexpression and even in complete P-to-A transformations (Schwartz et al., 1995; Tabata et al., 1995). Consistent with this, we observed that in HhHA2-expressing discs, anterior Ci staining expanded into the new stripe of En-OFF cells located posterior to the PtcLacZ expression domain (Fig. 6B-C‴, asterisks), and these cells also expressed low levels of DppLacZ (Fig. S8K-K″). This suggested that posterior En-expressing cells may somehow have changed their identity into more anterior types (Schwartz et al., 1995; Tabata et al., 1995), or that anterior cells had invaded and displaced posterior En-ON cells (in violation of the so-called selector affinity model, which states that the En-ON or En-OFF state of a cell selects whether it adopts P- or A-type identity, respectively, and that this ‘preset’ identity invariably locks the cell into the respective compartment; Morata and Lawrence, 1975). Note that similar analysis of wing discs overexpressing HhHA1 was not conducted because the complete lack of PtcLacZ expression in the disc proper (Fig. 5I-I″) prevented unambiguous identification of A/P border cells in this tissue.
Hh co-expression does not rescue dominant-negative HhHA1 activity
To confirm that CW variant Hh proteins render associated but otherwise functional Hh inactive, we macroscopically analyzed wings of single and compound fly lines overexpressing CW variant proteins from the attP 51C landing site on chromosome 2 and Hh from one specific attP2 landing site on chromosome 3 (Fig. 7A). We expected that co-expressed recombinant Hh would compensate for most or all dominant-negative wing patterning but would probably not rescue patterning defects caused by invariantly membrane-tethered transgenic HhHA1. In this case, we expected that dominant-negative function of overexpressed HhHA1 would be reversed only by the additional removal of its N-palmitate. In this assay, we found that Hh control expression from chromosome 2 or from chromosome 3 expanded the L3-L4 area (Fig. 7B,C) and that their combined expression expanded this area even further (Fig. 7D). No such additional L3-L4 expansion was observed upon en>Hh3xA;Hh co-expression (Fig. 7E). Moreover, en-Gal4-controlled Hh co-expression reversed HhHA2 and HhΔCW patterning defects (Fig. 3) and turned them into gain-of-function phenotypes (Fig. 7F,G), with the exception of en>HhHA1;Hh wings that still lacked all L3-L4 intervein tissue (Fig. 7H). Finally, the additional deletion of N-palmitate reverted HhHA1-induced dominant-negative wing mispatterning into a gain-of-function phenotype (HhC85S;HA1, Fig. 7I), consistent with our expectations. Taken together, the results from the quantification of wing patterning phenotypes (Fig. 7J) support the conclusion that impaired proteolytic processing of palmitoylated N-terminal peptides inhibits Hh release in vivo. They also support the idea that N-palmitate serves to firmly anchor unprocessed proteins to the membrane of the producing cell and that increasing the relative amounts of bioactive Hh can restore wing development to variable degrees. Although inserted HA tags downstream of the CW site block Hh processing to a very strong degree, the replacement of basic CW residues with neutral alanines abolishes recombinant Hh biofunction in some (but not all) developing tissues yet does not suppress co-expressed endogenous or exogenous Hh biofunction, possibly because it can be easily cleaved (Fig. 1, Fig. S1).
Decreased target gene expression underlies unrestored wing development in flies co-expressing Hh and HhHA1
Finally, we confirmed that Hh signaling activities in compound wing discs were inhibited (Fig. 8). We found that en-Gal4-controlled posterior Hh overexpression from one or both chromosomes expanded the stripe of PtcLacZ expression (Fig. 8A-C″). This is consistent with the strong gain-of-function phenotypes observed in adult wings of these genotypes (Fig. 7). By contrast, en-Gal4-controlled expression of Hh3xA, HhΔCW and HhHA2 reduced PtcLacZ reporter expression to only a limited degree (Fig. 8D-F″), unlike HhHA1 (Fig. 8G-G″), which suppressed PtcLacZ expression to a greater degree (Fig. 8H). This experiment also confirms that anterior cells are unimpaired in their ability to receive and to respond to solubilized mixed morphogen clusters.
Coupled essential roles of N-terminal Hh palmitoylation and processing in anterior receiving cells
In this study, we analyzed the consequences of recombinant Hh overexpression in posterior compartments of the Drosophila eye and wing disc. In both systems, the targeted deletion, replacement or relocation of the conserved Hh CW motif reduced recombinant Hh activity (Chamoun et al., 2001; Lee et al., 2001; Pepinsky et al., 1998) (Fig. 9A-C), and even converted most recombinant proteins, if overexpressed under en-Gal4 control, into dominant-negative suppressors of endogenous Hh function (Fig. 9D). This was not due to impaired interactions with the Ptc receptor, because Hh3xA was active if expressed in the eye disc or under the control of 45433-Gal4 and 45105-Gal4 in anterior wing disc tissues. Moreover, HhHA1 and HhHA2 both retained this site, yet acted most strongly on associated co-expressed Hh. Instead, we propose that inhibited N-terminal processing reduced the release and activity of recombinant CW variants (Fig. 9C) to variable degrees, and, by the same logic, also the release of physically associated endogenous Hh (Fig. 9D). Consistent with this, palmitate removal enhanced the solubilization of overexpressed, processing-impaired Hh variants from Bosc23 and S2 cells and reduced their dominant-negative activities in the developing wing disc. This suggests that N-palmitate serves to tether unprocessed Hh to the cell surface until regulated release is required (Chamoun et al., 2001; Konitsiotis et al., 2014).
We note that the in vivo strategy chosen in our work, that is, site-directed alterations in the putative Hh processing site (Ohlig et al., 2012) rather than the knockout or knock down of candidate proteases, avoids complex phenotypes, lethality, or no phenotypes caused by multiple affected substrates, essential other substrates, or compensatory mechanisms (Edwards et al., 2008). The same strategy was used by others to inactivate membrane molecules with similar mutations, such as interleukin-6 (Baran et al., 2013), T-cell immunoglobulin and mucin domain 3 (Möller-Hackbarth et al., 2013), tumor necrosis factor-α (Etzerodt et al., 2014), growth hormone binding protein receptor (GHR) (Wang et al., 2002), Notch (Brou et al., 2000) and L-selectin (CD62L) (Walcheck et al., 2003; Zhao et al., 2001). In these molecules, as well as vertebrate Shh (Ohlig et al., 2012), the sheddase cleavage site is located about 8-13 amino acids away from the plasma membrane (Dierker et al., 2009; Ohlig et al., 2012), consistent with the idea that Hh CW sites represent such cleavage sites (Fig. 9A,B). Notably, the positively charged Hh CW motif has also been demonstrated to bind to negatively charged HS (Farshi et al., 2011; Rubin et al., 2002), suggesting a possible mode of Hh release regulation. Supporting this possibility, replacing basic CW arginines with neutral alanines increases Hh3xA processing and solubilization, again resembling increased release of GHR and CD62 variants with the same alanine mutations (Wang et al., 2002; Zhao et al., 2001).
Possible mediators and subcellular sites of Hh release
Two important questions arise from our model. The first question is, which protease(s) release Hh in vivo? So far, it has been shown that members of the a disintegrin and metalloprotease (ADAM) family can contribute to the release of recombinant and endogenous Shh in vitro (Damhofer et al., 2015; Dierker et al., 2009; Ohlig et al., 2011), but it is still unclear whether ADAMs can directly act on Hh substrates in vivo. The second question is, which subcellular sites can release Hh? Most recently, one possible site has been suggested by the group of I. Guerrero (Gonzalez-Mendez et al., 2017). This group has previously shown that lipidated Hh transports on filopodia-like cellular extensions called cytonemes (Kornberg, 2014), and they also noticed that their restricted length requires cytonemes that emanate from the Hh-producing posterior compartment to somehow relay their Hh cargo to receiving cytonemes or to Ptc receptors on receiving anterior cells to allow for Hh long-range transport. As one way to achieve Hh relay at these sites, the authors suggested that Hh proteolysis may transfer the protein at specified ‘cytoneme synapses’ (Gonzalez-Mendez et al., 2017). We additionally note that Shh proteolysis may also act at the embryonic midline to control axon guidance at the optic chiasm (Peng et al., 2018). Finally, lack of evidence for lipid-dependent Shh binding to its receptor Ptc (Gong et al., 2018) is consistent with our model.
Hh control of compartmental lineage restriction
The ‘selector affinity’ model of tissue compartmentalization suggests that P versus A identities in thoracic Drosophila discs are determined by two transcription factors: En and Ci. However, it has already been shown that in addition to their selector function, En is also induced by Hh in a small stripe anterior to the A/P border during late third instar, in turn slightly shifting the En-ON border anteriorly (Guillen et al., 1995; Strigini and Cohen, 1997). In this work, we show that when HhHA1/2 is expressed under en-Gal4 control in the posterior wing disc compartment, both proteins consistently induce an unexpected posterior retraction of the En-ON boundary away from the stripe of PtcLacZ expression at the A/P border. Moreover, this new posterior stripe of En-OFF cells expressed the anterior compartment marker Ci. Although the mechanism behind HhHA1/2-induced En restriction is not clear, two possible explanations seem plausible. One is that the status of en as a selector gene applies only to the most posterior half of that compartment and that En expression in the anterior half is subjected to some hypothetical feedback control. Because the En protein suppresses ci transcription (Schwartz et al., 1995) and because cells with deleted En/Inv function start to produce Ci and change to an anterior type (Tabata et al., 1995), subsequent Ci expression in the anterior half of the posterior en>HhHA1/2 disc compartment, as observed in our work, could be explained. However, P-to-A conversion is normally associated with pattern duplications in adult wings (Tabata et al., 1995), which we never observed in en>HhHA1/2 flies, and is also incompatible with high-threshold PtcLacZ expression far away from posterior Hh-producing En-ON cells. An alternative explanation for our observations is the previously published ‘border guard’ (Rodriguez and Basler, 1997) or ‘signaling affinity boundary’ (Blair and Ralston, 1997) model. Here, the main affinity difference responsible for maintaining the physical segregation between A and P cells is not solely a heritable autonomous property under the control of En or Ci (as stated by the selector affinity model), but depends on local Hh signaling at the A/P border to indirectly elicit a Ci-mediated transcriptional response in A-type cells (Dahmann and Basler, 2000). This, in turn, is necessary and sufficient to increase mechanical tension along the A/P boundary and to control cell segregation (Rudolf et al., 2015). Supporting this concept, anterior cells that lack Smo receptors, and thus the ability to receive and transduce the Hh signal (Alcedo et al., 1996), no longer obey their A-lineage restriction, intermix with P-compartment cells, and displace the A/P lineage boundary, notably without changing their Ci expression (meaning that Ci expression remained ON and En remained OFF, although cells had moved into the posterior compartment) (Blair and Ralston, 1997; Rodriguez and Basler, 1997). Notably, clones of Smo-OFF cells generated straight boundaries with their neighboring cells and formed vein tissue in the position of the normally posterior L4, as also observed in our study (Fig. 3H). From these published data, together with our observations, we conclude that the posterior migration of anterior Smo-OFF cells may relate to, or even explain, the cellular behavior of en>HhHA1/2 discs that fail to release (most) Hh and thereby deprive anterior cells from the signal. Our data support the border guard/signaling affinity models by showing that unimpaired N-terminal Hh processing is not only essential for its anterior patterning function, but may also contribute to stably maintained compartment boundaries in vivo.
MATERIALS AND METHODS
The following Gal4 driver lines were used: Ptc-Gal4 (ptc>): w[*]; P(w[+mW.hs]=GawB)ptc[559.1], Bloomington stock #2017; En-Gal4e16E (en>): P(en2.4-Gal4)e16E, FlyBase ID FBrf0098595; Hh-Gal4 (hh>): w[*];; P(w[+mC]=Gal4)hh[Gal4], Bloomington stock #67046; 34B-Gal4 (34B>): y1w[*];; P(w[+mW.hs]=GawB)34B, Bloomington stock #1967; GMR-Gal4 (GMR>): GMR17G12 (GMR45433-Gal4): P(y[+t7.7]w[+mC]=GMR17G12-Gal4)attP2, Bloomington stock #45433 (discontinued but available from our lab); GMR18D10 (GMR45105-Gal4): P(y[+t7.7]w [+mC]=GMR18D10-Gal4)attP2, Bloomington stock #45105 (discontinued but available from our lab). Other flies used were hhbar3, FlyBase ID FBal0031487. Driver lines were crossed with flies homozygous for UAS-hh or variants thereof and kept at 25°C unless otherwise noted. Correct protein processing and secretion of pUAST-attP-hh constructs was first confirmed by Drosophila S2 cell expression. Ectopic Hh expression in the morphogenetic furrow of the eye disc was conducted by crossing the following fly lines: UAS-Hh*/CyOWeeP;hhAC/Tm6B and GMR-Gal4/GMR-Gal4;hhbar3/Tm6B. Resulting UAS-Hh*/GMR-Gal4;hhbar3/hhAC flies were analyzed with a Nikon SMZ25 microscope. w−;+/+;hhbar3/hhAC flies served as negative controls; white1118 flies served as positive controls. All constructs were inserted into the 51C1 landing site (BestGene) mediated by germline-specific PhiC31 integrase (Bateman et al., 2006). PtcLacZ reporter flies were kindly provided by Jianhang Jia (Markey Cancer Center, and Department of Molecular and Cellular Biochemistry, University of Kentucky College of Medicine, Lexington, USA). The w−;P(en2.4-Gal4)e16E/CyOGFP;dppLacZ;Tm6B enhancer trap line and the w−;P(en2.4-Gal4)e16E.UAS-Cherry;Tm2;Tm6B driver/reporter line were kindly provided by the Klämbt lab (University of Münster, Germany).
At least five wing discs per genotype/experiment were fixed in 4% paraformaldehyde (PFA) for 1 h at 4°C, permeabilized in 1% Triton X-100 and incubated in blocking solution containing 5% goat serum and 1% Triton X-100 in PBS for 1 h at room temperature. Discs were transferred into the same blocking solution additionally containing primary antibody at the concentrations given below and incubated at 4°C overnight. Discs were washed three times in PBS then incubated in secondary antibody in blocking solution for 2 h at room temperature. Discs were again washed three times in PBS and mounted in Vectashield (H-1000, Vector Laboratories). Samples were stained with anti-β-galactosidase antibodies (Cappel, MP Biomedicals, 08559761, 1:50, overnight incubation), Cy3-conjugated goat anti-rabbit antibodies (Jackson ImmunoResearch, 111-165-144, 1:600), anti-Engrailed (en 4D9, DSHB, 1:50, overnight incubation), anti-Ci (2A1, DSHB, 1:50 of a 32 µg/ml stock, overnight incubation) and Alexa488-conjugated donkey anti-mouse antibodies (Thermo Fisher Scientific, A21202, 1:600). Discs were immunolabeled using the same antibody batch and dilution, always following the same above-described procedure. Images were taken on a LSM 700 Zeiss confocal microscope with ZEN software, always with the same settings. Maximum intensity projections are shown. To generate plot profiles with ImageJ, identical areas covering most of the wing disc proper were analyzed, as defined by the region of interest (ROI) manager, and the plot histogram copied to MS Excel. Because Ptc is expressed in the anterior compartment only, the averaged anti-β-Gal antibody signal in the posterior compartment was regarded as noise and subtracted from the profile. Three wing discs for each genotype were analyzed. Anterior plot intensities were not normalized to show unaltered differential Hh variant activities.
Cloning and expression of recombinant proteins
hh cDNA (nucleotides 1-1416, corresponding to amino acids 1-471 of D. melanogaster Hh) and hhN cDNA (nucleotides 1-771, corresponding to amino acids 1-257) were inserted into pENTR, sequenced, and cloned into pUAST for protein expression in S2 cells or the generation of genetically modified flies. Mutations were introduced by QuikChange Lightning site-directed mutagenesis (Stratagene). Primer sequences are listed in supplementary Materials and Methods. S2 cells (CVCL_Z232) were cultured in Schneider's medium (Invitrogen) supplemented with 10% fetal calf serum and 100 μg/ml penicillin/streptomycin. The cells were obtained from C. Klämbt, and tested negative for mycoplasma. S2 cells were transfected via Effectene (Qiagen) with constructs encoding Hh, HhN or their N-terminal variants, together with a vector encoding actin-Gal4, and cultured for 48 h in Schneider's medium before protein was harvested from the supernatant via a heparin-agarose affinity pull-down. Proteins were analyzed by 15% SDS-PAGE and western blotting with polyvinylidene difluoride membranes. Blotted proteins were detected by anti-Hh (d300 rabbit IgG, Santa Cruz Biotechnology, sc-25759, 1:2000). Incubation with peroxidase-conjugated donkey anti-rabbit IgG (Dianova) was followed by chemiluminescent detection (Pierce). Shh constructs were generated from murine cDNA (NM_009170) by PCR and expressed in Bosc23 cells (RRID: CVCL_4401), as previously described (Schurmann et al., 2018) The cells were obtained from D. Robbins (University of Miami, USA), and tested negative for mycoplasma. Genetic characteristics were authenticated by PCR-single-locus-technology (Eurofins). Shh-containing media were ultracentrifuged for 30 min at 125,000 g, and the proteins were TCA precipitated and analyzed by 15% SDS-PAGE and western blotting with polyvinylidene difluoride membranes. Blotted proteins were detected by anti-Shh antibodies (goat IgG; R&D Systems, AF464, 1:1000). Incubation with peroxidase-conjugated donkey anti-goat IgG (Dianova, 705-035-003, 1:5000) was followed by chemiluminescent detection (Pierce).
Gel filtration analysis was performed on an Äkta protein purifier (GE Healthcare) on a Superdex200 10/300 GL column (Pharmacia) equilibrated with PBS at 4°C. Eluted fractions were TCA precipitated and analyzed by SDS-PAGE. Signals were quantified using ImageJ. One gram (wet weight) of D. melanogaster first to third instar embryos was homogenized and digested overnight in 320 mM NaCl and 100 mM sodium acetate (pH 5.5) containing 1 mg/ml pronase at 40°C. The digested samples were diluted 1:3 in water and 2.5 ml aliquots were applied to DEAE Sephacel columns. HS was eluted and applied to PD-10 (Sephadex G25) columns (GE Healthcare), lyophilized, re-dissolved in 20 µl water, digested with chondroitinase ABC overnight as indicated, and again purified by DEAE chromatography. Samples were diluted and again applied to PD-10 columns prior to lyophilization. β-Elimination of peptides was omitted from this purification protocol to allow for HS coupling to NHS-activated Sepharose via the attached peptides. We confirmed efficient HS coupling to NHS-activated Hi-Trap fast protein liquid chromatography (FPLC) columns by using soluble alkaline phosphatase-coupled fibroblast growth factor 8 and vascular endothelial growth factor as previously described (Farshi et al., 2011). HS binding of Hh and HhN and their variants was then determined by FPLC (Äkta protein purifier). Samples were applied to the columns in the absence of salt, and bound material was eluted with a linear 0-1 M NaCl gradient in 0.1 M phosphate buffer (pH 7.0). Eluted fractions were quantified as described above. HhN and Hh binding to heparin columns (GE Healthcare) was carried out with the same protocol, except for elution in a linear 0-1.5 M NaCl gradient in 0.1 M sodium phosphate buffer (pH 7.0).
Bioanalytical and statistical analysis
All statistical analyses were performed in GraphPad Prism using one-way analysis of variance tests (parametric, post-test as indicated, confidence interval 95%). For wing quantifications, wings were analyzed for each data set and ratios between L3-L4 intervein areas and L2-L3 intervein areas determined. Ommatidia from male and female flies were counted and statistically analyzed the same way.
The excellent technical and organizational assistance of S. Kupich and R. Schulz is gratefully acknowledged.
Conceptualization: D.M., G.S., C.K., K.G.; Methodology: D.M., G.S., C.K., K.G.; Validation: P.K., D.M., K.G.; Formal analysis: P.K., D.M., K.G.; Investigation: P.K., D.M., G.S., S.S., S.B.; Resources: G.S., S.S., S.B., K.G.; Writing - original draft: P.K., K.G.; Writing - review & editing: P.K., D.M., K.G.; Visualization: P.K., D.M.; Supervision: C.K., K.G.; Project administration: C.K., K.G.; Funding acquisition: K.G.
This work was financed with the support of Deutsche Forschungsgemeinschaft (German Research Council; GR1748/4-1, GR1748/5-1 and CiM FF-2015-02).
The authors declare no competing or financial interests.