Regulation of morphogen pathways by a Drosophila chondroitin sulfate proteoglycan Windpipe

Heparan sulfate (HS) and chondroitin sulfate (CS) are the most evolutionarily conserved glycosaminoglycans (GAGs) that are found in diverse animal species, including Caenorhabditis elegans, Drosophila and mammals. HS and CS are long, unbranched polysaccharides, composed of repeating disaccharide units: GlcA–GlcNAc and GlcA–GalNAc, respectively. They exist as forms of proteoglycans (PGs) in which one or more GAG chains are covalently attached to specific serine residues on the core protein. Both types of PGs are found on the cell surface and in the extracellular matrix.

It has been well established that heparan sulfate proteoglycans (HSPGs) function as co-receptors for growth factor signaling, and regulating the distribution and reception of secreted signaling proteins (Esko and Selleck, 2002; Kirkpatrick and Selleck, 2007; Li and Kusche-Gullberg, 2016; Lindahl and Li, 2009; Xu and Esko, 2014). The list of ‘HS-dependent factors’, secreted ligands that require HSPG co-receptors for proper distribution and signaling, continues to grow. Interestingly, many of these factors function as morphogens: a special type of signaling molecules that direct different cell fates in a concentration-dependent manner. In vivo studies using the Drosophila model have shown that HSPGs regulate gradient formation and signaling of four key morphogen molecules: Decapentaplegic (Dpp; a Drosophila BMP), Wingless (Wg; a Drosophila Wnt), Hedgehog (Hh) and Unpaired (Upd, or Upd1; a ligand of the Jak/Stat pathway) (Nakato and Li, 2016). During development and homeostasis, these same molecules also function as ‘niche factors’ that control stem cell self-renewal and differentiation in the stem cell niches. Therefore, HSPG co-receptors play critical roles in orchestrating various stem cell behaviors (Bowden and Nakato, 2021). For example, Dally, a Drosophila HSPG of the glypican family, serves as a Dpp co-receptor and regulates Dpp gradient formation in the developing wing (Akiyama et al., 2008; Belenkaya et al., 2004; Fujise et al., 2003) as well as the female germline stem cell niche (Guo and Wang, 2009; Hayashi et al., 2009). Interestingly, as dally expression is repressed by Dpp signaling, Dally forms a negative feedback loop of this pathway (Fujise et al., 2003). Similarly, the expression of thickveins (tkv), encoding a Dpp receptor, is regulated by morphogen signaling itself (Lecuit and Cohen, 1998). Such multiple circuits of feedback loops are believed to contribute to the robustness of the morphogen systems (Eldar et al., 2003; Irons et al., 2010; Lander et al., 2007; Lei and Song, 2010; Nakato and Li, 2016). Expression of dally is also controlled by Wg and Hh signaling, two additional pathways that Dally regulates (Fujise et al., 2001). Thus, Dally acts as a molecular hub of morphogen feedback networks.

HSPG function is tightly regulated by its biosynthetic and post-biosynthetic modification events. In the Golgi apparatus, a series of modification steps add sulfate groups to specific ring positions of HS. The degree and patterns of sulfation have a major impact on the activity of HS (Xu and Esko, 2014). In addition to these reactions in the Golgi apparatus, HS structure is further modified on the cell surface by the extracellular endosulfatases, Sulfs, in a post-biosynthetic manner (Ai et al., 2003; Dhoot et al., 2001; Kamimura et al., 2006; Uchimura et al., 2006). Sulfs specifically remove sulfate groups at the 6-O position of glucosamine residues from highly sulfated regions of HS. In Drosophila, a single Sulf homolog, Sulf1, modulates FGF, Wg, Hh and Egfr signaling during development (Butchar et al., 2012; Dani et al., 2012; Kamimura et al., 2006; Kleinschmit et al., 2010, 2013; Takemura and Nakato, 2017; Wojcinski et al., 2011; You et al., 2011). In the developing wing, Sulf1 negatively regulates Wg signaling by removing ligand binding sites on HS (Kleinschmit et al., 2010). Importantly, expression of Sulf1 is induced by the Wg pathway itself. A similar phenomenon has been also reported for the Hh and Vein-Egfr pathways (Butchar et al., 2012; Wojcinski et al., 2011). Thus, like Dally, Sulf1 is another molecular hub of morphogen feedback circuits.

Compared to HS, much less is known regarding the role of CS in cell signaling (Cortes et al., 2009; Townley and Bülow, 2018). In Drosophila, only a few molecules have been shown to bear CS chains, which include Kon-tiki (Kon) (Losada-Perez et al., 2016; Perez-Moreno et al., 2014, 2017), Multiplexin (Mp) (Csordas et al., 2020; Harpaz et al., 2013; Momota et al., 2011) and Windpipe (Wdp) (Ren et al., 2015; Takemura et al., 2020). Given the structural similarities between CS and HS, chondroitin sulfate proteoglycans (CSPGs) might have modulatory, supportive and/or complementary functions to HSPGs. In fact, Wdp, a single-pass transmembrane CSPG with leucine-rich repeat motifs, modulates Upd and Hh signaling (Ren et al., 2015; Takemura et al., 2020). This raised the idea of a ‘dual PG co-receptor system’ in which HS-dependent pathways are also generally regulated by CS (Coles et al., 2011; Takemura et al., 2020). Interestingly, wdp expression is induced by Upd-Jak/Stat signaling in the midgut, forming a negative feedback loop in this pathway (Ren et al., 2015). This suggests a possibility that CSPGs might function together with HSPGs as morphogen feedback regulators. However, the functional relationship between HSPGs and CSPGs remains elusive.

In the current study, to gain insights into the function of CSPGs in morphogen signaling and their relationship with HSPG co-receptors, we performed biochemical and genetic analyses of Wdp. We found that Wdp is a major CSPG in Drosophila, which bears 4-O-sulfated CS chains and regulates HS-dependent pathways. When we perturbed morphogen feedback loops by introducing a Sulf1 or dally mutation, wdp mutation resulted in severe morphological defects, including abnormal patterning of wing blade and hinge structures, egg retention in the ovary and a wing posture abnormality. Our results implicate Wdp as a novel player of the morphogen feedback buffering systems.

As the first step of biochemical analyses of Drosophila GAGs, crude HS and CS samples were prepared from adult flies and analyzed by anion exchange chromatography. The GAGs bound to a DEAE column were eluted with a 0–1.5 M NaCl gradient. The elution patterns showed two peaks both in HS and CS samples (Fig. 1A), suggesting that Drosophila GAGs can be largely separated into two groups with different charges. A similar pattern was also observed in the analysis of CSPGs. DEAE-column chromatography of CSPGs purified from adult flies using a stepwise elution with different salt concentrations (0.26 and 1.0 M NaCl) clearly separated the CSPG specimen into two fractions, fractions 1 and 2 (Fig. 1B).

To characterize the two fractions of Drosophila CSPGs, we analyzed the CS structures by disaccharide analysis. Briefly, CS was purified from fractions 1 and 2 as shown in Fig. 1B, and completely digested into disaccharides by chondroitinase ABC. The resultant disaccharide species were separated and quantified by reversed-phase ion-pair chromatography with a post-column detection system (Dejima et al., 2013b; Kamimura et al., 2006; Kleinschmit et al., 2010; Nakato et al., 2019; Toyoda et al., 2000). In Drosophila, two major CS disaccharide species, unsulfated (ΔDi-0S) and 4-O-sulfated (ΔDi-4S) disaccharide units, can be detected but 6-O sulfation of GalNAc residues is under the detection limit (Toyoda et al., 2000). We found that CS isolated from fraction 2 contains ΔDi-4S, but this disaccharide was not detectable in fraction 1 CSPGs (Fig. 1C). This observation indicates that there exists two groups of CSPGs in Drosophila: one bearing 4-O-sulfated CS and another with non-sulfated chondroitin. This is a unique feature in Drosophila CSPGs.

We found that a commercially available anti-CS antibody (LY111) detects Drosophila CS. This is one of few antibodies that can recognize Drosophila GAGs and offers a useful tool to study the biological functions of CS using this model organism. Immunoblot analysis of whole-protein extracts from wild-type adults using LY111 detects high-molecular-mass proteins (>200 kDa) as smear bands (Fig. 1D, left panel). These smear bands disappeared after the treatment of samples with chondroitinase ABC or chondroitinase ACII (Fig. 1A), confirming that the smear bands represent CSPGs.

We also performed an RNAi knockdown for Chsy, a Drosophila homologue of human ChSy-1, a key component of CS polymerases (Kitagawa et al., 2001; Mikami and Kitagawa, 2013; Sugahara et al., 2003) (Fig. 1D, middle panel). A UAS-Chsy RNAi transgene was driven using a ubiquitous actin-Gal4 driver. The LY111 signal was substantially reduced in Chsy knockdown animals (act>Chsy RNAi). Interestingly, the extract prepared from wdp mutants showed a significantly decreased amount of CS compared to wild type. In general, loss of a single PG core protein does not reduce GAG-positive bands from the whole animals or organs as detected bands represent the sum of sugar moieties of a large number of PG molecules. This result suggested that Wdp might be one of the major CSPGs in Drosophila. In fact, high-throughput expression analyses in FlyBase (see http://flybase.org/cgi-bin/rnaseqmapper.pl?dataset=celniker_wiggle&xfield1=FBgn0034718 and http://flybase.org/cgi-bin/rnaseqmapper.pl?dataset=tissues_stranded&xfield1=FB gn0034718) indicate that this gene is expressed at very high levels in many tissues and developmental stages.

Wdp was identified as a CSPG in our previous study via a glycoproteomic approach (Takemura et al., 2020). To confirm that Wdp is modified with CS by immunoblot analysis, we used a transgenic strain, wdp-HA (previously called wdp^KI.HA) (Takemura et al., 2020). In this strain, HA-epitope-tagged Wdp protein is expressed from its endogenous locus. Anti-HA antibody staining of ovary protein extract detected smear bands (Fig. 1D, right panel, wdp-HA). When we blocked CS biosynthesis by Chsy RNAi knockdown in wdp-HA animals (wdp-HA, act>Chsy RNAi), the smear bands were lost. Instead, a single band representing the Wdp core protein was detected. This result confirmed a CS modification of Wdp.

To determine the amount of CS and its disaccharide composition in wdp mutants, we performed disaccharide analyses. We found that the disaccharide composition of CS from wdp mutants was comparable to that of wild type (Fig. 1E; Table S2). However, the total amount of CS was reduced by approximately 19% in wdp mutants (Fig. 1F). The reduction of CS in wdp mutants is consistent with our western blot results and supports the idea that Wdp is a major CSPG in this animal.

Similar analyses of HS showed that HS disaccharide composition is unchanged in wdp mutants (Table S3). Interestingly, however, the total amount of HS was increased by approximately 44% in the mutant (Fig. 1G; Table S3). In many model systems, genetic manipulations that reduce HS result in increased production of CS (Bachvarova et al., 2020; Bai et al., 1999; Holmborn et al., 2012; Le Jan et al., 2012; Lidholt et al., 1992; Lin et al., 2000). Therefore, it is possible that the reduction of CS is compensated by the elevated synthesis of HS.

To analyze spatial distribution of CS in a tissue, we stained the wing discs using LY111. In wild type, the LY111 signal was detected uniformly throughout the wing disc (Fig. 2A). To ask whether the LY111 signal indeed reflects the distribution of CS, we blocked CS biosynthesis specifically in the posterior compartment of the wing disc. Expression of UAS-Chsy RNAi was driven by hh-Gal4, a posterior compartment-specific driver. We observed that RNAi knockdown of Chsy eliminated the LY111 signal (Fig. 2B,B′), confirming the specificity of LY111 staining in immunohistochemistry.

We next analyzed wdp mutant wing discs with the LY111 antibody. We observed a significant reduction in the LY111 signal intensity in the mutant discs compared to that in wild type (Fig. 2C). Quantification of the signal intensity confirmed that, consistent with our immunoblot analysis, the LY111 signal was significantly decreased in wdp mutants (Fig. 2D). This reduction was also confirmed by RNAi knockdown. Expression of UAS-wdp RNAi in the posterior compartment partially but significantly decreased the signal intensity (Fig. 2E,E′).

The anti-CS antibody (LY111) is believed to recognize highly sulfated structures of CS, such as CS-A units or 4-O-sulfated CS [GlcUAβ1–3GalNAc(4S)] (Deepa et al., 2007). Therefore, we asked whether the LY111 signal might be affected by blocking CS sulfation. CG31743 is orthologous to several human genes, including CHST11, and is predicted to encode a Drosophila homologue of CS 4-O sulfotransferase (C4ST). We found that RNAi knockdown of CG31743 abolished the epitope of this antibody, similarly to Chsy RNAi (Fig. 2F,F′). This result supports that LY111 detects a 4-O-sulfated CS in situ and confirms that CG31743 encodes a Drosophila C4ST. Hereafter, the gene CG31743 is referred to as C4ST.

It is worth noting that the LY111 signal was detected in a cell-autonomous manner in both RNAi knockdown treatments; GFP signals from gene-specific RNAi-expressing cells showed no overlap with LY111 signals (Fig. 2B,E,F). This is important information because it implies that major CSPGs in this tissue are either integral membrane PGs or secreted PGs that are embedded in the ECM in proximity to the expressing cells.

We next analyzed the distribution of CS in further detail, along the apicobasal axis of the wing epithelium. Our previous study has shown that Wdp is enriched in the basal membranes of epithelia, including wing cells (Takemura et al., 2020). As shown in Fig. 2, the LY111 signal largely overlaps with endogenously expressed epitope-tagged Wdp (Wdp–HA) (Fig. 2G,G″). In addition, the LY111 signal is also detected in the basement membrane (BM) layer, visualized by a protein trap line of Perlecan (trol-GFP) (Fig. 2G′,G″) (Medioni and Noselli, 2005). No significant signal was observed at the apical side of the epithelium in this organ. Thus, this observation showed that CSPGs are mainly localized in the basal membrane and the BM of the developing wing.

Our biochemical data as well as immunohistochemical observations suggest that CS chains of Wdp are 4-O sulfated. To determine the contribution of CS 4-O sulfation to Wdp function, we used RNAi knockdown of C4ST. Our previous study showed that wdp overexpression using Bx-Gal4 impairs Hh signaling, resulting in reduced central area of the wing between longitudinal wing veins L3 and L4 (Takemura et al., 2020) (Fig. 3A,B). Using this assay system, we addressed whether co-expression of UAS-C4ST RNAi with UAS-wdp affects Wdp activity as a negative regulator of Hh signaling.

We found that C4ST RNAi knockdown showed a statistically significant impairment of Wdp activity in inhibiting Hh signaling (Fig. 3C,D), demonstrating the importance of CS 4-O sulfation for the full activity of Wdp. Interestingly, C4ST RNAi did not completely rescue the L3–L4 area, interfering with the Wdp activity only partially. This might suggest a possibility that CS chains with no 4-O sulfation retain residual activity to modulate Hh signaling. However, although this UAS-C4ST RNAi construct efficiently reduced the LY111 signal in the wing disc (Fig. 2F), it is always possible that a partial effect is due to, at least partly, incomplete efficacy of RNAi. A future study with a C4ST null mutation will clarify this point.

Wdp was previously shown to regulate Jak/Stat and Hh signaling, two HS-dependent pathways (Ren et al., 2015; Takemura et al., 2020). These observations raised the question of whether this CSPG is a general regulator of morphogen pathways, also affecting signaling events of other HS-dependent factors, such as Dpp and Wg.

When we used nubbin (nub)-Gal4 to drive UAS-wdp expression, we observed a few different phenotypes we did not see in Bx>wdp. One class of the new phenotypes was wing vein defects, including a spur of ectopic venation at the anterior and posterior crossveins (Fig. 4A,B), loss or reduction of crossveins (Fig. 4C), and the formation of extra vein materials, most frequently near the longitudinal wing vein 2 (L2) (Fig. 4D). These wing vein phenotypes are characteristic of Dpp signaling defects and commonly observed in HS-related mutants (Dejima et al., 2013b; Takeo et al., 2005).

In our previous study, we attempted to determine the role of Wdp in Dpp or Wg signaling by overexpressing UAS-wdp in the developing wing using apterous-Gal4 or hh-Gal4 (Takemura et al., 2020). However, as wdp overexpression in a large region of a tissue from an early developmental stage changed the shape of the tissue, this analysis was inconclusive. Therefore, we employed the flp-out technique to generate wdp-overexpressing clones in random locations of the wing disc (Bowden et al., 2022; Struhl and Basler, 1993). We stained these wing discs with an antibody specific to the phosphorylated form of the Mad protein (anti-pMad antibody), which serves as a direct readout of Dpp signaling (Fig. 4E–G′). We found that within the wdp-overexpressing clones induced in the central region of the wing discs where cells receive high levels of Dpp signaling, pMad levels were reduced compared to those in neighboring regions outside the clones (yellow brackets in Fig. 4F,G, quantification in Fig. 4H). This result indicated that Wdp downregulates Dpp signaling.

In addition to the vein phenotypes, wdp overexpression using nub-Gal4 resulted in wing notching, or a deletion of a part of the wing margin structure, indicative of the impairment of Wg signaling (Fig. 5A–D). To determine whether Wdp is involved in Wg signaling in vivo, we examined expression of Senseless (Sens), a high-threshold target of Wg signaling, in the wdp-overexpressing wing discs. In wild type, the Sens protein is expressed in two stripes of cells near the dorsoventral boundary of the wing disc (Fig. 5E). We found that Sens expression at the dorsoventral border was severely impaired by nub-Gal4-mediated overexpression of wdp (Fig. 5F). Sens expression in other regions, which is not dependent on Wg signaling, was not affected by Wdp. When wdp was overexpressed specifically in the dorsal compartment using the ap-Gal4 driver (ap>wdp), anti-Sens staining was diminished only in the dorsal row (‘D’), leaving the ventral row (‘V’) intact (Fig. 5G,G′, yellow arrowheads). This was also confirmed by staining ap>wdp discs with anti-Distal-less (Dll), a low-threshold target of Wg signaling. In wild type, Dll-positive cells were distributed evenly in the dorsal and ventral compartments (Fig. 5H). In ap>wdp discs, Dll expression was severely diminished in the dorsal compartment (Fig. 5I,J). Taken together, consistent with the adult wing phenotypes, these results showed that Wdp downregulates Wg signaling.

In a mammalian tissue model, Wnt-3a is known to bind to a highly sulfated structure of CS and 4-O sulfation, which affects Wnt-3a diffusion (Nadanaka et al., 2008, 2011). To determine whether Wdp, which bears 4-O-sulfated CS chains, affects the distribution of the Wg ligand, we stained wing discs bearing wdp flp-out clones with the anti-Wg antibody using the extracellular staining protocol (Kleinschmit et al., 2010; Strigini and Cohen, 2000). This protocol specifically visualizes the Wg ligand in the extracellular space. Surprisingly, we found a significant increase in the level of extracellular Wg protein within wdp flp-out clones (Fig. 5K–L′). Thus, wdp overexpression increases Wg ligand levels while downregulating its signaling.

To determine whether this function of Wdp requires CS chains, we generated flp-out clones overexpressing WdpΔGAG. In the wdpΔGAG construct, all serine residues required for CS attachment were substituted with alanine residues so that the core protein is not modified with CS (Takemura et al., 2020). We found that WdpΔGAG failed to increase the level of extracellular Wg protein (Fig. 5M,M′). These observations strongly suggest that Wdp overexpression sequesters the Wg ligand via CS chains and reduces the pool of ligand molecules available to form the functional ligand/receptor/co-receptor signaling complex, consistent with the idea that HS and CS competitively function to finetune Wg signaling.

We next examined whether Wdp interacts with Wg in vitro by coimmunoprecipitation experiments. We generated a construct for a secreted form (the extracellular domain) of Myc–Wdp (sec-Myc–Wdp) by deleting the transmembrane domain and the intracellular domain. Wg was expressed with or without sec-Myc–Wdp in S2 cells. We found that Wg was immunoprecipitated from conditioned medium with the anti-Myc antibody only in the presence of sec-Myc–Wdp (Fig. 5N). This result indicates that Wdp forms a complex with Wg, further supporting the idea that Wdp sequesters Wg. Taken together, these results showed that Wdp is a general regulator of morphogen signaling pathways that are known to be HS dependent.

Although wdp overexpression disrupts Dpp and Wg signaling, wdp null mutants do not show obvious defects on these pathways. It is well known that morphogen pathways are controlled by multiple circuits of feedback regulation to buffer against genetic and environmental perturbations, and thus are highly robust (Eldar et al., 2003; Fujise et al., 2003; Kleinschmit et al., 2010; Lander et al., 2007). We hypothesized that the loss of wdp is compensated by modulation of HS-related genes, masking wdp mutant phenotypes. To test this idea, we examined the genetic interactions between wdp and HS-related genes to determine what happens if this buffering system is compromised by breaking this feedback loop. We chose two HS-related genes for this analysis, Sulf1 and dally, as the protein products of both genes were previously shown to be molecular hubs extensively involved in morphogen feedback networks (Butchar et al., 2012; Fujise et al., 2003; Kleinschmit et al., 2010; Wojcinski et al., 2011; You et al., 2011).

Among the Drosophila HS-modifying enzymes, Sulf1 is known to inhibit most, if not all, HS-dependent pathways, including Wg, Hh, BMP and Upd-Jak/Stat signaling, by removing the ligand binding sites on HS (Kleinschmit et al., 2010; Wojcinski et al., 2011). We first realized that wdp; Sulf1 double mutants were highly lethal. Although both wdp or Sulf1 single mutants were homozygous viable and fertile, the lethality of the double mutant was higher than 95%. In addition, adult survivors showed various morphological defects, including adult wing abnormalities. As reported before (Takemura et al., 2020), wdp mutant wings did not show any gross morphological defects (Fig. 6A,B). Similarly, Sulf1 mutant wings did not show any patterning defects, although the mutant wings were slightly larger than wild-type wings (Fig. 6C) (Dejima et al., 2013a). In contrast, in wdp; Sulf1 double-mutant survivors, the wing patterning was massively disrupted. Although the severity of the phenotypes varied between individuals, the penetrance of these defects was 100%. Interestingly, we found that two specific regions of the wing were affected. First, the medial–distal part of the posterior edge of the wing was deleted, abnormally pigmented and often had ectopic bristles (Fig. 6D–I, brackets). Second, similar defects, including pigmentation and ectopic bristles, were observed in the alula, a structure at the posterior hinge region (Fig. 6G–I, asterisks).

Given that the Hh pathway regulates patterning in the anterior compartments, these posterior defects of the wdp; Sulf1 double mutants cannot be explained by altered Hh signaling. The formation of alula is controlled by Dpp derived from the posterior compartment (Foronda et al., 2009). Therefore, we examined Dpp signaling in wdp; Sulf1 double-mutant wing discs using the anti-pMad antibody. We found that overall levels of pMad staining were higher in the double mutant compared to the wild type (Fig. 6J–L). Asterisks mark the region where cells receive posterior Dpp signaling, which directs alula formation. We observed modest overgrowth in this region of the double-mutant discs.

As wing notching and ectopic bristles are commonly observed with altered Wg signaling, we next asked whether this pathway is affected in the wdp; Sulf1 double-mutant wing discs. Anti-Sens antibody staining revealed two classes of phenotypes in the mutants (Fig. 6M–O). In the first group, the two rows of Sens-positive cells were more broadly spread in the posterior region of the double-mutant discs (Fig. 6N). The distance between the two rows was also larger. These observations suggest a broader Wg gradient. In the second group, the posterior Sens signals were lost and the disc was severely deformed, suggesting reduced Wg signaling (Fig. 6O). Thus, these discs appeared to show both upregulation and downregulation of Wg signaling. This is not uncommon in HS-related gene mutants in which the shape of a morphogen gradient is altered. In addition, the Wg pathway is known to trigger non-autonomous inhibitory signals (Piddini and Vincent, 2009). Thus, reduced Wg signaling in a cell can increase signaling dosage in the surrounding cells. We propose that altered signaling of both Dpp and Wg contribute to the wing-patterning defects of the wdp; Sulf1 double mutants.

In addition to a high level of lethality and abnormal wing morphology, wdp; Sulf1 adult female survivors were completely sterile. We therefore examined ovary morphology of these mutants. Young wild-type females have a pair of ovaries, each of which consists of 16–20 ovarioles, a string of progressively developing egg chambers (Fig. 7A). At the anterior tip of each ovariole is a structure called the germarium that contains the germline stem cells and follicle stem cells. At the posterior edge, ovarioles are connected to the oviduct through which mature eggs are transported to the uterus. During aging, the ovary reduces in size and oogenesis slows (Fig. 7B). We found that the overall morphology of the ovary from young wdp; Sulf1 double mutants was relatively normal (Fig. 7C). Surprisingly, however, the double-mutant ovaries from aged animals (day 21 after eclosion) were significantly larger compared to wild-type ovaries (Fig. 7D).

We next flattened the ovaries from day-21 females to visualize the composition of the egg chambers (Fig. 7E–H). Under light microscopy, oocytes in the egg chamber of stage 9 (as described in Prasad et al., 2007) and later were visualized as light gray (Fig. 7E). In addition, mature eggs (stage 14) could be recognized by a dark gray color and with fully developed dorsal appendages. We found that ovaries from aged single mutants of wdp as well as Sulf1 were smaller than wild-type ovaries (Fig. 7F,G). In striking contrast, flattened specimens of aged wdp; Sulf1 double mutants were much larger than those of wild type, with an abnormally higher number of mature eggs (Fig. 7H). Quantification of mature oocytes in an ovariole showed that this ‘egg retention’ phenotype is age dependent (Fig. 7J). Typically, an aged double-mutant ovariole contained three or more mature eggs at the posterior end (Fig. 7I), whereas a wild-type ovariole had one. Although the germarium existed, egg chambers with intermediate stages (stage 7–13 oocytes) were lost (Fig. 7I). In fact, this is a common characteristic of mutants that have egg-laying defects (Liao and Nässel, 2020; Monastirioti, 2003).

We next immunostained mutant ovarioles with anti-Vas (germline cells) and anti-Fasciclin 3 antibodies (follicle cells) and observed them by confocal microscopy. In old wild-type females, ovarioles show normal progression of oogenesis in ordered egg chambers (Fig. 7K). Fig. 7L shows an example of an ovariole from an aged wdp; Sulf1 double mutant, in which the organization and morphology of the egg chambers were massively disrupted. The lack of intermediate-stage egg chambers and accumulation of mature eggs were also confirmed (data not shown). These observations indicate that simultaneous loss of wdp and Sulf1 results in the failure of ovulation – the transport of mature eggs from the ovary to the oviduct – leading to the swollen-ovary phenotype.

We next analyzed the genetic interactions between wdp and dally, another feedback hub of HS-dependent morphogen pathways. We first found that wdp significantly enhanced some, but not all, dally mutant phenotypes. For example, 93.5% of males of wdp; dally mutants showed a complete lack of external genitalia (Fig. 8A,B), whereas only 3.3% of dally mutant males showed this phenotype. Also, the expressivity of a wing vein defect of dally mutants was strongly enhanced in the double mutants (Fig. 8C–E). dally mutants showed an incomplete longitudinal wing vein V, lacking its most distal portion (Nakato et al., 1995) (Fig. 8D). In wdp; dally mutant wings, the deletion of wing vein V extended into the proximal region and it often failed to reach the posterior cross vein (Fig. 8E). Interestingly, however, the penetrance of the wing-notching phenotype of dally mutants was not affected by the wdp mutation (data not shown).

In addition to the effects on known dally phenotypes, we also found that wdp; dally double-mutant adults showed a defect in resting wing posture, or the ‘outstretched wing’ phenotype. This phenotype has never been reported in any HS-related gene mutants. In the double mutants, wings were held out from the body at a 45–90° angle from the longitudinal body axis, whereas in the single mutants, the wings were held over the abdomen (Fig. 8F–I). It is worth mentioning that classical mutants showing the same phenotype include dpp^d-ho and upd^os-s (Fig. 8J,K) (St Johnston et al., 1990; Wang et al., 2014). These are hypomorphic alleles of dpp and upd, respectively, two genes encoding morphogen ligands that use Dally as a co-receptor (Fujise et al., 2003; Hayashi et al., 2012).

This observation indicated that HS-dependent morphogen pathways are required for normal wing posture. Outstretched wing posture can be caused by altered flight muscle function and physiology (Everetts et al., 2021). Therefore, we hypothesized that the morphogen pathways function for flight muscle development. To test this idea, we inhibited HS biosynthesis in the developing muscles by RNAi knockdown of one of two HS biosynthetic genes, tout-velu (ttv, encoding a HS co-polymerase) and sulfateless (sfl, encoding a N-deacetylase/N-sulfotransferase). Expression of either UAS-ttv RNAi or UAS-sfl RNAi by the mef2-Gal4 driver recapitulated the outstretched-wing phenotype (Fig. 8L,M). Taken together, our results show that in the absence of Sulf1 or dally, wdp mutation led to a high level of lethality and morphological defects that were normally rescued by the feedback buffering system.

Morphogens are a class of signaling molecules that form concentration gradients in a developmental field and specify different cell fates in a concentration-dependent fashion. Many of these pathways can become oncogenic when hyperactivated. Therefore, the signaling dosage of these pathways has to be tightly controlled during development for proper patterning as well as cancer prevention. One of the key features of the morphogen system is its robustness: multiple circuits of feedback regulation buffer against genetic and environmental perturbations. HSPG co-receptors play critical roles in quantitative control of morphogen signaling output as well as feedback control (Nakato and Li, 2016). On the other hand, the functions of CSPGs in morphogen signaling of the genetically tractable model organism Drosophila are largely unknown.

Our study showed that Drosophila CSPGs are classified into two groups: PGs with or without 4-O-sulfated CS. Wdp is a major 4-O-sulfated CSPG and regulates Dpp and Wg signaling, two major pathways regulated by HSPGs. We found that wdp overexpression increased the extracellular level of the Wg ligand, which resulted in reduced expression of downstream targets of Wg signaling. This finding suggests that Wdp downregulates Wg signaling by sequestering the Wg protein. Thus, in this context, CS appears to compete with HS to control the amount of the ligand available to activate the receptors.

Despite the obvious phenotypes of wdp-overexpressing animals, the effect of wdp null mutation in Dpp- or Wg-dependent specification events was not evident. However, in the absence of Sulf1 or Dally, two known molecular hubs of morphogen feedback networks, a wdp mutation produced synthetic lethality and various morphological and physiological phenotypes. These data indicate that the feedback systems of morphogen-HSPG signaling provide buffering effects and can compensate for the lack of the CSPG Wdp. Thus, Wdp is not only a general regulator of HS-dependent pathways but also a novel component of the morphogen feedback regulatory network. The identification of Wdp as a specific CSPG molecule that regulates all four key HS-dependent pathways suggests that HS-dependent factors might be generally controlled by both HS and CS (Coles et al., 2011; Takemura et al., 2020). CSPGs appear to provide additional layers of morphogen regulation, which is likely to finetune signaling dosage and provide the robustness of cell signaling as well as developmental programs.

As Wdp affects multiple HS-dependent factors, wdp mutations offer an interesting opportunity to genetically analyze the HS-CS relationship. We found that double mutations in wdp in combination with an HS-related gene exhibited three novel, unique phenotypes that have never been observed in single mutants of genes encoding HSPGs and HS biosynthetic enzymes: (1) deletion, pigmentation and ectopic bristle formation in specific regions of the wing; (2) egg retention in the ovary; and (3) an outstretched wing phenotype. There are a few possible mechanisms by which ovulation is impaired in aged wdp; Sulf1 double mutants. Ovulation is controlled by octopaminergic neural signaling, which activates the contraction of ovary and oviduct muscles to push a mature egg from the posterior end of the ovary into the oviduct (Lim et al., 2014; Monastirioti, 2003; Sun and Spradling, 2013). Therefore, simultaneous loss of wdp and Sulf1 might disrupt a step in this pathway or normal muscle development in these organs. Alternatively, ovulation might be impaired by physical disruption in wdp; Sulf1, such as the failure of the formation of a tubular structure that connects the ovary and oviduct (Deady et al., 2015; Kiss et al., 2019). Drosophila ovulation and flight muscle development will be additional useful model systems to study the functions of HS and CS.

As both Wdp and Sulf1 have inhibitory activities on HS-dependent pathways (Kleinschmit et al., 2010; Takemura et al., 2020), it was reasonable to observe that they genetically enhanced each other. In this regard, it is interesting that wdp enhanced a specific set of dally mutant phenotypes. Dally acts as a co-receptor for morphogen ligands to promote signaling. At the same time, however, it limits their diffusion (Fujise et al., 2003). It is possible that wdp enhanced dally by aggravating the gradient formation and local availability of morphogen ligands.

In vertebrates, CSPGs are well established as major structural components of connective tissues, including cartilage, and support their mechanical cushioning properties. Vertebrate CSPGs are also known to regulate cell signaling by binding to growth factor ligands (Li et al., 2011; Mizumoto et al., 2015; Nadanaka et al., 2008; Whalen et al., 2013) or activating cell surface receptors (Izumikawa et al., 2014; Mikami et al., 2009; Mizumoto et al., 2012, 2015). CS, like HS, is evolutionarily old and shared by more primitive animal species that have no cartilage, bone or skin. It is intriguing to know how CS emerged during evolution and what its original roles were. One possibility is that HS and CS were partners as signaling regulators in an ancestral species. Further studies of CSPGs in various invertebrate model organisms such as Drosophila will help provide insight into these questions.

The following fly strains were used in this study: Oregon-R, wdp^KO (Takemura et al., 2020), wdp-HA (Takemura et al., 2020), Sulf1^ΔP1 (Kleinschmit et al., 2010), dally^gem (Nakato et al., 1995; Tsuda et al., 1999), trol-GFP (Medioni and Noselli, 2005), dpp^d-ho [Bloomington Drosophila Stock Center (BDSC) #308], upd^os-s (BDSC #79), hsp70-flp (BDSC #8862), ap-Gal4, hh-Gal4, mef2-Gal4 (BDSC #27390), Bx-GAL4 (BDSC #8860), UAS-GFP (BDSC #1521), nub-GAL4 (BDSC #25754), Act5C-Gal4 (BDSC #3954), Act5C>CD2>Gal4 flp-out cassette (Fujise et al., 2003), UAS-wdp (Takemura et al., 2020), UAS-wdp RNAi (TRiP.HM05118, BDSC #28907), UAS-Chsy RNAi [GD14159, Vienna Drosophila Resource Center (VDRC) #29084], UAS-C4ST RNAi (UAS-CG31743.IR.Y) (Yamamoto-Hino et al., 2015), UAS-ttv RNAi (GD1993, VDRC #4871) and UAS-sfl RNAi (HMS00543, BDSC #34601). The genotypes of fly strains used for the data shown in the figures are listed in Table S1.

Flies were raised on a standard cornmeal fly medium at 25°C unless otherwise indicated. For the Dpp signaling assay, flp-out clones overexpressing wdp were generated as previously described (Bowden et al., 2022; Struhl and Basler, 1993) in wing discs bearing an Act5C>CD2>Gal4 transgene cassette, hsp70-flp, UAS-GFP and UAS-wdp. The FLP expression from hsp70-flp was induced by heat-shock treatment of larvae at 37°C for 30 min at 30–40 h after egg laying. For the Wg signaling assay, we overexpressed wdp with ap-Gal4 during third larval instar stage by a temperature shift.

The right wings from female flies were dehydrated in ethanol and subsequently with xylene (Fujise et al., 2001; Takemura et al., 2020). The specimens were mounted in Canada balsam (Benz Microscope, BB0020).

Immunostaining of the wing discs and ovaries was performed as previously described (Hayashi et al., 2009, 2012; Takemura et al., 2020). The primary antibodies used were as follows: rat anti-HA 3F10 (1:200, Roche, 11867423001), rabbit anti-HA C29F4 (1:1000, Cell Signaling Technology, 3724), rabbit anti-pSmad3 (1:1000, Epitomics, 1880-1), guinea pig anti-Sens (1:1000, a gift from Hugo Bellen, Baylor College of Medicine, TX, USA), mouse anti-Dll (1:400, a gift from Dianne Duncan, Washington University in St. Louis, MO, USA), mouse anti-CS-A (1:100, Tokyo Chemical Industry, LY111), mouse anti-Fasciclin III 7G10 [1:50, Developmental Studies Hybridoma Bank (DSHB)] and rabbit anti-Vas (1:500, a gift from Satoru Kobayashi, University of Tsukuba, Tokyo, Japan). The secondary antibodies used were Alexa Fluor 488, 568 or 633 conjugated (1:500, Thermo FisherScientific). Extracellular labelling of Wg protein was performed as described previously (Kleinschmit et al., 2010; Strigini and Cohen, 2000) using the anti-Wg antibody (4D4, DSHB) at a 1:3 dilution. Images were obtained using a Zeiss 710 laser scanning confocal microscope.

For immunoblot analysis, protein samples were extracted from Drosophila adult whole body (for CS detection) or adult ovaries (for Wdp–HA detection) by SDS sample buffer. Mouse anti-CS A (1:1000, Tokyo Chemical Industry, LY111), rat anti-HA antibody (3F10) (1:2000, Roche, 11867423001), and mouse anti-α-tubulin (1:2000, Sigma-Aldrich, DM1A) were used as primary antibodies. Signals were detected using HRP-conjugated secondary antibodies (goat anti-mouse IgG Fc-HRP and goat anti-rat IgG-HRP obtained from SouthernBiotech, Birmingham, AL) and Pierce ECL Western Blotting Substrate (Thermo Fisher Scientific). For blot transparency, original immunoblots are given in Fig. S1.

For coimmunoprecipitation experiments, we generated a construct for a secreted form of Myc–Wdp (sec-Myc–Wdp) by deleting the transmembrane domain and the intracellular domain (A493–H719). Drosophila Dmel2 tissue culture cells were transfected with pMT-Wg (Kleinschmit et al., 2010) and/or pAW-sec-Myc–Wdp (this study). After incubation at 25°C for 72 h, 1 ml of each conditioned medium was incubated with anti-cMyc monoclonal antibody-agarose beads (Sigma-Aldrich) overnight at 4°C, washed, eluted with 6 M urea and analyzed by immunoblotting.

To isolate Drosophila GAGs, approximately 1.0 g of lyophilized adult flies was defatted with acetone and then extracted with 0.5% SDS, 0.1 M NaOH and 0.8% NaBH₄ as previously described (Toyoda et al., 2000). The crude GAGs were applied to a HiPrep DEAE 16/10 column [16 mm internal diameter, 100 mm length; GE Healthcare (Uppsala, Sweden)] equilibrated with 25 mM phosphate buffer (pH 6.0) and elution was performed with a 0–1.5 M NaCl gradient in the same buffer at a flow rate of 1.0 ml/min.

To isolate Drosophila CSPGs, approximately 1.2 g of lyophilized adult flies was defatted with acetone. The samples were treated with 4 M guanidinium chloride, 0.05 M phosphate buffer, pH 6.0, and 1% Triton X-100, containing proteinase inhibitors (cOmplete™ ULTRA Tablets, Mini, EDTA-free, EASYpack obtained from Roche), for 2.5 h at room temperature with constant stirring. The extract was centrifuged at 20,000 g for 10 min to remove insoluble materials. The crude CSPG fractions were dialyzed into distilled water and then into 25 mM phosphate buffer, pH 6.0. The resulting solution was separated by anion-exchange chromatography using a Hi Trap DEAE FF (16 mm×50 mm) column (GE Healthcare) at a flow rate of 2 ml/min. The column was equilibrated with 25 mM phosphate buffer (pH 6.0), 0.5 M urea and 0.02 M NaCl, and eluted stepwise with increasing concentrations of NaCl at 0.26 M and 1.0 M. The eluents were monitored at 280 nm. The fractions (0.26 M and 1 M NaCl) were desalted and dissolved in 4 ml of water.

Disaccharide composition analysis was carried out as previously described (Dejima et al., 2013b; Kamimura et al., 2006; Kleinschmit et al., 2010; Nakato et al., 2019; Toyoda et al., 2000). Briefly, a 20 µl portion of the sample solution was incubated with 5 µl of 0.2 M Tris-acetate buffer (pH 8.0) and 10 µl of an aqueous solution containing chondroitinase ABC or ACII [1 mIU; chondroitinase ABC (EC 4.2.2.4) and chondroitinase ACII (EC 4.2.2.5) were obtained from Seikagaku, Tokyo, Japan] at 37°C overnight. The resulting disaccharide species were separated using reversed-phase ion-pair chromatography [Docosil C22 (4.6×150 mm; particle size, 5 μm) was obtained from Senshu Scientific, Tokyo, Japan]. The effluent was monitored fluorometrically for post-column detection of CS or HS disaccharides (Toyoda et al., 2000).

We thank Hugo Bellen, Dianne Duncan, Satoru Kobayashi, the Developmental Studies Hybridoma Bank, the Bloomington Drosophila Stock Center [National Institutes of Health (NIH) P40OD018537], the Vienna Drosophila Resource Center, the Transgenic RNAi Project at Harvard Medical School [NIH/National Institute of General Medical Sciences (NIGMS) R01-GM08947], and the Drosophila Genomics Resource Center (NIH 2P40OD010949) for antibodies and fly strains. We are grateful to Melanie LeMinh for her support of Drosophila genetic experiments.