Differential heparan sulfate dependency of the Drosophila glypicans

Heparan sulfate proteoglycans (HSPGs) are composed of a core protein and glycosaminoglycan (GAG) chains and serve as coreceptors for many growth factors and morphogens. To understand the molecular mechanisms by which HSPGs regulate morphogen gradient formation and signaling, it is important to determine the relative contributions of the carbohydrate and protein moieties to the proteoglycan function. To address this question, we generated ΔGAG alleles for dally and dally-like protein (dlp), two Drosophila HSPGs of the glypican family, in which all GAG-attachment serine residues are substituted to alanine residues using CRISPR/Cas9 mutagenesis. In these alleles, the glypican core proteins are expressed from the endogenous loci with no GAG modification. Analyses of the dallyΔGAG allele defined Dally functions that do not require heparan sulfate (HS) chains and that need both core protein and HS chains. We found a new, dallyΔGAG-specific phenotype, the formation of a posterior ectopic vein, which we have never seen in the null mutants. Unlike dallyΔGAG, dlpΔGAG mutants do not show most of the dlp null mutant phenotypes, suggesting that HS chains are dispensable for these dlp functions. As an exception, HS is essentially required for Dlp's activity at the neuromuscular junction. Thus, Drosophila glypicans show strikingly different levels of HS dependency. The ΔGAG mutant alleles of the glypicans serve as new molecular genetic toolsets highly useful to address important biological questions, such as molecular mechanisms of morphogen gradient formation.

Precise control of cell fate decisions is a crucial step of tissue morphogenesis.The secreted signaling molecules called morphogens form concentration gradients in a developing tissue and specify the cell fate in a concentration-dependent manner.In the Drosophila developing wing, three major morphogens play key roles.Hedgehog (Hh) is expressed in the posterior compartment and induces expression of target genes in the anterior compartment.One of the Hh target genes is another morphogen, Decapentaplegic (Dpp; a Drosophila bone morphogenetic protein [BMP]).In addition, Wingless (Wg) is expressed in a stripe of cells at the dorsoventral compartment boundary and regulates patterning along the D-V axis.
Morphogen signaling pathways form multiple feedback loops to buffer against genetic and environmental perturbations, and HSPG coreceptors are major components of such feedback systems (19,20).For example, expression of Dally, which enhances Dpp signaling as a coreceptor, is repressed by Dpp signaling (6).Thus, Dally forms a negative feedback loop of this pathway.Similarly, dally expression is also controlled by Wg and Hh signaling, two additional pathways Dally regulates (21).The heparan sulfate (HS) modification machinery also plays a role in the morphogen feedback systems.Sulf1 is a secreted HS endosulfatase and negatively regulates Wg signaling by removing ligand-binding sites on HS (22).Expression of Sulf1 is induced by the Wg pathway itself.A similar phenomenon has also been reported for the Hh and Vein-epidermal growth factor receptor pathways (23,24).In addition, it is known that a loss of one type of HS sulfation is compensated for by an increase in sulfation at different positions (25)(26)(27).Thus, feedback loops of HSPGs and HS sulfation control provide compensatory effects to minimize consequences of a genetic change in morphogen signaling, contributing to the robustness of the system (9,20).
HSPGs are composed of HS chains that are attached to specific serine residues of the core proteins.It is clear that HS chains are critical for HSPG coreceptor activity because blocking HS biosynthesis inhibits morphogen signaling.Virtually all signaling events mediated by fibroblast growth factors, Dpp, Wg, and Unpaired (Upd; a ligand of the Jak/Stat pathway) are disrupted in HS-deficient animals, such as toutvelu (ttv; a Drosophila homologue of Ext proteins, which functions as a HS copolymerase) and sulfateless (sfl; the only Drosophila homologue of HS N-deacetylate/N-sulfotransferase, a key enzyme for HS modifications) (28)(29)(30).On the other hand, core protein structures of syndecans and glypicans are highly conserved during evolution (31), suggesting the importance of the core proteins in signaling activity.Biochemical and structural studies have demonstrated direct interactions between HSPG core proteins and morphogen ligands ( (15,16,(32)(33)(34)). Thus, the core protein plays a significant role and is not just as a carrier of HS chains.Therefore, it is important to understand the relative contribution of the sugar and protein moieties to proteoglycan functions.
In Drosophila genetics, the Gal4/UAS system using a "UAS-PG ΔGAG " construct has been used as the major approach to address this question (12,(35)(36)(37).The UAS-PG ΔGAG constructs were made by substituting the glycosaminoglycan (GAG)-attachment serine residues with alanine residues so that HS cannot be added to the core protein.By overexpressing wildtype PG and PG ΔGAG with the Gal4/UAS system, the in vivo activities of the two constructs and their ability to rescue proteoglycan mutant phenotypes can be compared.Although this method has been helpful to determine the roles of HS chains and the core proteins, it has some limitations.Since there is no perfect Gal4 driver, the results tend to include unwanted effects of overexpression and ectopic expression.Given that morphogen signaling is a highly quantitative system, the Gal4-driven overexpression does not provide an appropriate platform to precisely assess the coreceptor activities.Namely, a previous work has shown that PG ΔGAG overexpression interferes with HS modification of other proteins, which makes it difficult to interpret data (35).Therefore, an ideal strategy to address this question is to make an endogenous ΔGAG allele by genome editing.
In this study, we generated and performed phenotypic analyses of ΔGAG alleles for dally and dlp.We found that these glypicans show strikingly different levels of HS dependency.Our analyses of double mutants, dally gem dlp ΔGAG and dally ΔGAG dlp ΔGAG , raised a possibility that the loss of HS chains of glypicans activates compensatory feedback mechanisms.

ΔGAG alleles of dally and dlp genes
Using CRISPR/Cas9 mutagenesis, we generated ΔGAG alleles for the two Drosophila glypicans, Dally and Dlp (Fig. 1A).In both loci, all GAG-attachment serine residues were substituted with alanine residues.The positions of the mutated residues are shown by arrowheads in Figure 1, and the specific serine residues altered in the entire amino acid sequences are shown in Fig. S1.In these alleles, the glypican core proteins are expressed from their endogenous loci with no GAG modification.In the Western blot analyses of wildtype extract, anti-Dally and anti-Dlp antibodies detect high-molecularweight proteins as smear bands.These smear bands are characteristic of proteoglycans reflecting varying lengths of individual GAG chains (Fig. 1B).These smear bands disappear in the ΔGAG mutant extracts, confirming the lack of GAG modification.
dally ΔGAG shows normal patterning of notal mechanosensory bristles and eyes In this study, we used dally gem as a null allele.dally gem mutants are semilethal, and approximately half of them die before the adult stage (38).dally ΔGAG shows a reduced lethality rate (10-20%) (Fig. 2A).The milder but non-wildtype phenotypes of dally ΔGAG are consistent with the idea that the core protein retains coreceptor activity in the absence of HS chains.We found that some dally adult phenotypes are absent in dally ΔGAG .For example, dally null mutants show the loss of anterior dorsocentral mechanosensory bristles, a specific sensory organ on the notum, with 100% penetrance due to reduced Dpp signaling in the developing notum (Fig. 2B, penetrance = 100%, n = 406 for females and 437 for males, (21)).We found that this phenotype completely disappeared in dally ΔGAG (Fig. 2B, penetrance = 0%, n = 852 for females and 812 for males).
Similarly, dally null alleles exhibit a fully penetrant rough eye phenotype (38).In contrast, dally ΔGAG shows normal patterning of the eye (Fig. 2C).These observations indicate that HS chains are dispensable for Dally function in notal sensory organ specification and eye morphogenesis.
dally ΔGAG shows reduced severity of wing vein V and wing margin patterning defects compared with dally null mutants Dally coreceptor is required for normal gradient formation of Dpp.In the absence of dally, the Dpp signal gradient steeply drops near the source of Dpp (6,7).As a result, the dally mutant wing is narrow in the AP axis and loses a posterior part of the longitudinal wing vein 5 (L5) ((38); Fig. 3A middle).The penetrance of this phenotype in dally gem is 100% for both sexes (Fig. 3B).We found that dally ΔGAG retains this phenotype with much reduced severity and lower penetrance (Fig. 3, A right  and B).
Dally also regulates Wg-dependent specification of the sensory organs at the wing margin.The anterior wing margin bears rows of different types of sensory organs: thick and short mechanosensory bristles at the edge and thinner chemosensory bristles slightly posterior to the edge (39).The number of these bristles is substantially reduced in dally gem ((21); Fig. 3,  C and D).The bristle formation at the wing margin is partially restored in dally ΔGAG (Fig. 3, C and D).
Thus, dally ΔGAG shows reduced severity of wing vein V and wing margin patterning defects compared with dally null mutants.These observations suggest that both the core protein and the HS chains contribute to the Dally activity during the formation of the wing veins and the wing margin.
dally ΔGAG shows an increased penetrance of genitalia phenotypes compared with dally null mutants dally gem males show a complete lack of genitalia, called genitalia-missing, or "gem," phenotype (38).This phenotype was shown to be a Dpp-dependent specification event (40).The female genitalia are also sometimes reduced, but more often we observe a hyperplasia (Fig. 3E).In our culture condition, the penetrance of these genitalia phenotypes in dally null mutants are low (<10%, Fig. 3F).Interestingly, the penetrance of these phenotypes was modestly increased in dally ΔGAG , both in females and males (Fig. 3F).The increase in males but not in females was statistically significant (Fig. 3F).
We next asked if expression of wildtype dally can rescue the phenotypes of dally ΔGAG .Overexpression of dally using a strong driver (e.g., actin-Gal4) causes lethality and morphological abnormalities (41).We therefore chose to use a heatshock (hs)-dally transgene, which induces a modest level of dally expression without producing morphological defects at 29 C (42).With this condition, expression of wildtype dally almost completely rescued the lethality and the adult phenotypes of dally ΔGAG , including wing vein V and genitalia defects (Fig. S2).This finding showed that these phenotypes observed in dally ΔGAG are ascribed to reduced functioning of Dally.
dally ΔGAG shows ectopic wing vein formation Unexpectedly, we found a new, dally ΔGAG -specific phenotype, which has never been observed in the null mutants.dally ΔGAG shows ectopic wing vein formation at a specific region of the wing, which is posterior to the most posterior wing vein, L5 (Fig. 4A).The penetrance of this phenotype is approximately 60% in females (Fig. 4B).
Ectopic vein structures can be induced by ectopic activation of Dpp/BMP signaling.To examine the patterns of Dpp activity in developing dally ΔGAG wings, we stained dally ΔGAG pupal wing with anti-phosphorylated Mad (pMad) antibody at 24 h after puparium formation.At this stage, pMad staining mainly marks future longitudinal veins and crossveins (Fig. 4C).In the dally ΔGAG pupal wing, ectopic pMad signal was detected posterior to wing vein L5 (Fig. 4C), confirming that the ectopic vein phenotype in dally ΔGAG is associated with ectopic Dpp signaling.
Interestingly, we found that this phenotype is semidominant.dally ΔGAG /+ heterozygotes show a smaller ectopic vein at the same location (Fig. 4, D and F).Thus, the formation of this posterior ectopic vein is dally ΔGAG dosage dependent.The small ectopic vein reflects a delicate balance of ectopic Dpp signaling, and it is sensitive to a further change in the signaling dosage.For example, we found that this phenotype is sensitive to Dad.Dad is the Drosophila ortholog of inhibitory Smads.Vertebrate inhibitory Smads negatively regulates signaling of TGF-β family ligands by inhibiting phosphorylation of R-Smads (43)(44)(45).In Drosophila, heterozygosity of Dad slightly increases Dpp signaling but does not show any defect by itself (46,47).However, it significantly enhances the ectopic vein phenotype of dally ΔGAG (Fig. 4, E and F).
These results suggest that Dpp signaling is ectopically activated to induce extra venation in a region of a dally ΔGAG wing where its ligand concentration is very low.We next asked if the level of Dally core protein is altered in dally ΔGAG mutant cells.dally ΔGAG homozygous clones were induced in the wing disc, and Dally core-protein expression was examined using anti-Dally antibody.We found a higher level of Dally core protein in dally ΔGAG mutant cells compared with surrounding dally ΔGAG /+ and +/+ cells (Fig. 4, G-I").Quantification of anti-Dally antibody signal inside and outside of dally ΔGAG mutant clones is shown in Figure 4J.
Most dlp mutant phenotypes are absent in dlp ΔGAG allele Unlike dally ΔGAG , the dlp ΔGAG allele is surprisingly healthy.dlp ΔGAG homozygotes are viable and fertile and show no obvious morphological defect.Given that dlp null is a very sick mutant with 100% lethality at larval stage, this difference was striking.
Dlp is a Hh coreceptor, and the dlp mutant embryos show a segment polarity phenotype ((12); Fig. 5A).The segment polarity phenotype of dlp MH20 embryos shown in Figure 5A are relatively mild because they are only zygotically null and rescued by maternal supply.dlp ΔGAG embryos do not show this  phenotype despite the fact that these embryos are both maternally and zygotically homozygous with no rescue by wildtype dlp mRNA (Fig. 5A).
Dlp is known to show a biphasic effect on Wg signaling: it downregulates Wg signaling near the Wg-expressing cells but upregulates it where the ligand concentration is low (17,36,(48)(49)(50).Therefore, dlp knockdown shows ectopic mechanosensory bristles, slightly shifted from the edge due to increased Wg signaling at the wing margin, which is formed by high levels of Wg signaling (50); Fig. 5B).We observed a mild phenotype of the ectopic mechanosensory bristles in dlp ΔGAG , with a significantly lower expressivity than Bx>RNAi knockdown animals (Fig. 5, B and C).

dlp ΔGAG shows neuromuscular junction phenotypes
We previously showed that the trans coreceptor activity of glypicans needs GAG chains in vitro (10).Dlp functions in trans, with the most obvious example being its role at the neuromuscular junction (NMJ) during larval development.At the NMJ, Lar, a protein tyrosine phosphatase, functions on the presynaptic membrane to promote bouton formation (51).Two types of HSPGs on the postsynaptic membrane regulate Lar signaling in an opposing manner: Syndecan promotes Lar activity and Dlp suppresses it (52).Therefore, dlp mutants show an increased number of boutons (Fig. 5, D and E  Under starvation conditions, larvae show increases in locomotor activity and the number of synaptic boutons at the NMJ.Such increases in locomotor speed and bouton production depend on noncanonical BMP signaling (Fig. 5E; (53)).
We found that dlp ΔGAG failed to suppress bouton formation under normal conditions (Fig. 5, D and E).This is the same phenotype as dlp knockdown animals we have previously shown (53).Statistical analyses revealed no significant difference in the bouton numbers per area between dlp ΔGAG and dlp knockdown animals (Fig. S3).Furthermore, dlp ΔGAG failed to induce starvation-induced synaptic plasticity, resulting in the same number of boutons before and after 8-h food deprivation (Fig. 5E), which is also consistent with dlp knockdown (Fig. S3; (53)).These results indicated that HS chains of Dlp are indispensable at least for its NMJ functions.
dlp ΔGAG is lethal in the absence of dally Although we found important roles of Dlp HS chains at the NMJ, overall phenotypes of dlp ΔGAG are surprisingly mild.This prompted us to analyze whether Dally plays a role in this allele.To address this question, we generated the dally gem dlp ΔGAG double mutant.Remarkably, the double mutant showed 100% lethality (Fig. 6A).This suggests that most functions of Dlp HS chains can be complemented by Dally, except for their roles at the NMJ.
The lethality data also showed that dally gem dlp ΔGAG double mutants show higher levels of lethality than dally gem single mutants (Fig. 6A), indicating that Dlp HS chains function to support the survival of dally mutants.Thus, although previous studies indicated that Dally and Dlp have only limited functional redundancy in the wildtype background, they show close functional relationships when one of the glypicans is disrupted.

The phenotypes of glypican ΔGAG
We next recombined dally ΔGAG and dlp ΔGAG to generate the double mutant, or glypican ΔGAG .Based on the lethality data, they are more severe than dally ΔGAG (Fig. 6B), suggesting a partial functional redundancy between Dally and Dlp HS chains.Interestingly, although glypican ΔGAG is homozygous for dally ΔGAG , they do not show the posterior ectopic wing vein (Fig. 6C).Thus, Dlp HS chains not only weaken dally mutant phenotypes but also are required for the formation of the posterior ectopic wing vein in dally ΔGAG .
Numerous previous studies have established the coreceptor functions of glypicans in morphogen signaling (9).On the other hand, only limited cases are known for nonglypican HSPGs as a coreceptor (9).It is worth noting that there is a large gap between the phenotypes of glypican ΔGAG and the HS biosynthetic gene mutants, such as ttv and sfl.Zygotic null mutants for these biosynthetic genes show 100% lethality during larval to pupal stages (28)(29)(30), while 40% of glypican ΔGAG mutants survive to the adult stages (Fig. 6B).This difference in the lethality levels between ttv (or sfl) and glypican ΔGAG suggested that nonglypican HSPGs play a role in morphogen signaling.To confirm this idea, we tested the effect of Sulf1 overexpression in the glypican ΔGAG background.Sulf1 inhibits multiple morphogen signaling pathways, including the Wg, Hh, and Dpp pathways (22,23).Overexpression of a Golgi-tethered form of Sulf1 (Sulf1-Golgi, or Sulf1(G)) in the developing wing by ap-Gal4, a dorsal compartment-specific Gal4 driver, decreases the proliferation of dorsal cells in a cell-autonomous manner (22).This results in a significant reduction in the dorsal cell/entire wing pouch area ratio (Fig. 6, D and E).A similar effect of Sulf1(G) expression was observed in glypican ΔGAG homozygous wing discs (Fig. 6, D  and E).This confirmed that morphogen signaling is mediated by HS chains on nonglypican HSPGs when glypican HS modification is impaired.
Previous studies have shown that a disruption of HS functions is compensated by various feedback systems, suggesting a possibility that such compensation systems function in ΔGAG alleles.For example, upregulation of nonglypican HSPGs may contribute to the mild phenotypes of the ΔGAG alleles.Therefore, we asked if Syndecan (Sdc) expression is altered in glypican ΔGAG .Anti-Sdc antibody staining of wing disc, NMJ, fat body, eye disc, and central nervous system showed no significant difference in Sdc levels and distribution between wildtype and glypican ΔGAG (Fig. 7).Thus, different mechanisms (e.g., other proteoglycans, HS sulfation) may be involved in ΔGAG mutants.

Discussion
Using glypican "ΔGAG" alleles in which all endogenous GAG-attachment serine residues are replaced with alanine residues, we found that levels of HS dependency of the two Drosophila glypicans, Dally and Dlp, are strikingly different.Regarding dally alleles, the dally ΔGAG allele generally shows less severe phenotypes compared with dally null mutants.We classified dally null mutant phenotypes into two classes: (1) HS chains being dispensable, which includes notal sensory bristle HS dependency of Drosophila glypicans formation and eye morphogenesis, and (2) both HS chains and core protein being required for proper development, which includes the formation of wing vein V, wing margin sensory bristles, and genitalia.The defective formation of anterior dorsocentral bristles, wing vein V, and genitalia in dally mutants is caused by reduced Dpp signaling.Therefore, the two classes were not simply dependent on the morphogen pathway that dally regulates but are likely to be context dependent.
In contrast to dally ΔGAG , we observed only limited morphological defects in dlp ΔGAG .This is a striking contrast to the zygotic null mutants of dlp, which are 100% lethal at larval stage (both maternal and zygotic null is 100% lethal at embryonic stage).We found that the dlp ΔGAG allele is healthy only due to the presence of Dally, as it is 100% lethal in the dally null background.
Although HS chains of Dlp are dispensable for most Dlp functions, analyses of the double mutants, dally gem dlp ΔGAG and dally ΔGAG dlp ΔGAG (glypican ΔGAG ), revealed additional activities of Dlp HS chains in these mutant backgrounds.They show the ability to (1) weaken dally mutant phenotypes, (2) complement the loss of Dally HS chains, and (3) contribute to the formation of the posterior ectopic wing vein in dally ΔGAG .Although there is very little functional redundancy between Dally and Dlp under normal environmental conditions, genetic manipulations inducing partially truncated coreceptor functions revealed intimate relationships between the glypicans.
It has been believed that the glypicans function as coreceptors for most morphogen ligands in Drosophila.When glypican HS modification is impaired, however, nonglypican HSPGs are able to function in morphogen signaling (Fig. 6, D  and E).As disruptions of HS functions are known to trigger various compensation systems to minimize their impact, it is possible that the mild phenotypes of the ΔGAG alleles are due to a compensatory feedback response.Possible mechanisms of such feedback mechanisms include, but are not limited to, the regulation of HSPG core-protein expression and/or distribution, HS sulfation, and HS chain length.No significant change in Sdc levels and distribution was observed in glypican ΔGAG (Fig. 7).This observation suggests that different mechanisms may contribute to the modest ΔGAG mutant phenotypes.Further studies are required to test this possibility and to determine the molecular nature of the unexpectedly mild phenotypes of the ΔGAG mutants.
In addition to known dally mutant phenotypes, dally ΔGAG shows the formation of an ectopic vein in the posterior region of the wing.This region of the pupal wing exhibits ectopic pMad, and the phenotype is enhanced by Dad.These findings suggest that Dpp signaling is ectopically activated to induce extra venation in a region where the ligand concentration is very low.It is unknown how this morphogen signaling is increased by the loss of Dally HS chains, but there are a few possibilities.First, the ligand gradient may be altered.Unlike dally null mutants, in which the Dpp gradient quickly drops near the source of Dpp (6,7), the gradient may be expanded in dally ΔGAG as Dally ΔGAG should have lower affinity, or be less "sticky," to Dpp.Therefore, the cells at the periphery may receive a higher level of Dpp, which can specify an ectopic vein fate.Second, as shown in Figure 4, G-I", dally ΔGAG mutant cells have an increased level of Dally core protein.Pentagone is a secreted factor that regulates the Dpp gradient by promoting Dally internalization and degradation (54).Recently, another soluble factor, Nord, was found to show a similar activity (55).Both proteins are heparin-binding factors and thus are likely to bind Dally through HS chains.If so, Dally ΔGAG may escape from Pent-and Nord-dependent degradation.It is also possible that the dally gene is transcriptionally upregulated in dally ΔGAG mutant cells.Thus, either altered local Dpp concentration or elevated levels of Dally core protein, or both as a combination, may lead to ectopic activation of Dpp signaling.
In the past studies addressing the relative contribution of the protein versus sugar moieties to the PG functions using conventional "ΔGAG" overexpression constructs, we tended to assume that "core-protein activity plus GAG activity equals PG activity" as the basis of the assessment.Our study suggests that this idea may be oversimplified.As HSPGs regulate both the signal reception on the cell surface (a cell autonomous function) and ligand distribution in a tissue (a nonautonomous function) (6), the consequence of a ΔGAG mutation is regionally different even in the same tissue.For example, dally ΔGAG shows multiple phenotypes caused by reduced Dpp-Dally signaling (e.g., wing vein V and genitalia defects) and also posterior wing vein phenotype caused by ectopic Dpp signaling.Interestingly, hs-dally did not enhance the ectopic wing vein phenotype of dally ΔGAG but partially suppressed it (Fig. S2C).This finding suggests that this phenotype is produced by a combined effect of cell autonomous and nonautonomous consequences from the lack of HS chains.Furthermore, it is worth noting that glypican ΔGAG provides a unique cellular condition: there are no HS attachment sites on the glypicans, which normally function as coreceptors for most morphogen signaling, but the HS biosynthetic machinery is intact.Altogether, ΔGAG mutants will be highly useful to address important biological questions, such as molecular mechanisms of morphogen gradient formation and scaling.

Generation of dally ΔGAG and dlp ΔGAG mutant alleles
In order to generate dally ΔGAG , five GAG-attachment serine residues at positions 549, 569, 573, 597, and 601 of the dally cDNA were substituted with alanine residues via CRISPR/ Cas9-mediated gene editing by injecting a donor DNA construct and a dally sgRNA.

Preparation of embryonic cuticles, wings, nota, and eyes
The standard embryo collection for preparation of cuticles consisted of a 4-h egg collection, followed by incubation for 20 h at 25 C (Tsuda et al., 1999).Embryos were fixed, and devitellinized cuticles were prepared with standard procedures.
The right wings from female flies were dehydrated in ethanol and subsequently with xylene (21,58).Adult cuticles of the notum were boiled in 2.5 N sodium hydroxide, washed in distilled water, and dehydrated in 2-propanol.The specimens were mounted in Canada balsam (Benz Microscope, BB0020).
Left eyes from adult males were imaged using a Nikon AZ100M microscope.Z-stacks were captured for each eye and compiled into a single focused image using NIS-Elements.
Anti-Dally polyclonal antibody was raised using a synthetic peptide, DSRAKDAVGGSTHQC (Biologica Co).For immunoblot analysis, protein samples were extracted from Drosophila adult whole body by SDS sample buffer.The peptide was injected into guinea pig, and the antibody was affinity purified.Mouse anti-Dlp (1:200, DSHB), guinea pig anti-Dally antibody (1:1000), and mouse anti-α-Tubulin (1:2,000, Sigma-Aldrich, DM1A) were used as primary antibodies.Signals were detected using HRP-conjugated secondary antibodies and Pierce ECL Western Blotting Substrate (Thermo Scientific).

NMJ analysis
Quantification of synaptic boutons of larvae was performed as described (53).Larval body wall muscles were dissected in Ca 2+ -free HL3 saline (70 mM NaCl, 5 mM KCl, 20 mM MgCl 2 , 10 mM NaHCO 3 , 5 mM trehalose, 115 mM sucrose, and 5 mM Hepes, pH 7.2) and fixed in Bouin's fixative (Sigma-Aldrich) at room temperature for 20 min.Tissues were incubated at 4 C overnight in primary antibody solutions and then at room temperature for 2 h in secondary antibody solutions.Alexa Fluor 647-conjugated AffiniPure Goat Anti-HRP and mouse anti-Dlg were used as the primary antibodies.The number of type I boutons, defined as axonal swelling, at muscles 6 and 7 in abdominal segment A3 were scored.The image of corresponding muscle was obtained and muscle surface area was calculated by ImageJ software.Bouton number was then normalized to muscle area.For starvation assay, we used larvae showing vigorous feeding in the food slurry that did not climb out of the food.After washing with water using a paintbrush, larvae were placed on 0.7% agar in vials and were left there for 8 h before dissection.GAG, glycosaminoglycan; Hh, Hedgehog; HSPG, heparan sulfate proteoglycan; NMJ, neuromuscular junction; Wg, Wingless.

Figure 1 .
Figure 1.ΔGAG alleles of dally and dlp genes.A, a schematic of the dally and dlp loci.Arrowheads indicate the position of serine residues that are substituted to alanine residues.Amino acid sequences of Dally ΔGAG and Dlp ΔGAG are shown in Fig. S1.B, immunoblot analysis of ΔGAG alleles.Protein extracts from wildtype, dally ΔGAG , and dlp ΔGAG adult flies were analyzed using anti-Dally (left) and anti-Dlp (right) antibodies (top panels).Anti-α-Tubulin antibody was used as an internal control (bottom panels).

Figure 2 .
Figure 2. Lethality and notal bristles of dally ΔGAG mutants.A, lethality of dally gem and dally ΔGAG alleles.The lethality rate of female (F) or male (M) homozygotes for dally gem and dally ΔGAG was calculated by four sets of independent experiments.n = 510 for dally gem females; n = 579 for dally gem males; n = 1228 for dally ΔGAG females; n = 1149 for dally ΔGAG males.Boxes indicate the 25 to 75th percentiles, and the median is marked with a line.The whiskers extend to the highest and lowest values within 1.5 times the interquartile range.B and C, adult notal bristles and eyes of dallyΔGAG mutants.Adult notum (B) and eyes (C) for wildtype (left), dally gem (middle), and dally ΔGAG (right) are shown.The penetrance of the loss of anterior dorsocentral bristles and rough eye phenotype in dally gem is 100% (n = 406 for females and 437 for males).These phenotypes are absent in dally ΔGAG (penetrance = 0%, n = 852 for females and 812 for males).Arrowheads mark anterior dorsocentral bristles.The scale bars represent 250 μm (B); 100 μm (C).*p < 0.05; **p < 0.01.

Figure 3 .
Figure 3. Analyses of dally ΔGAG phenotypes.A, posterior portions of wing vein V of adult wings of wildtype (left), dally gem (middle), and dally ΔGAG (right) are shown.B, the penetrance of the wing vein V defect in females (F) and males (M) of dally gem and dally ΔGAG .The box plot summarizes the penetrance data from four sets of independent experiments.n = 104 for dally gem females; n = 142 for dally gem males; n = 376 for dally ΔGAG females; n = 337 for dally ΔGAG males.C, the anterior wing margin of wildtype (top), dally gem (middle), and dally ΔGAG (bottom).Chemosensory bristles are marked by arrows.D, the quantification of chemosensory bristles.n = 30 for wildtype; n = 30 for dally gem ; n = 30 for dally ΔGAG .E, genitalia defects of dally ΔGAG .Female (top) and male (bottom) genitalia are shown for wildtype (left panels) and dally ΔGAG (right panels).Arrows show hyperplasia of female genitalia and a complete loss of male genitalia.F, the penetrance of the genitalia defects in females (F) and males (M) of dally gem and dally ΔGAG .The box plot represents the penetrance data from four sets of independent experiments.n = 104 for dally gem females; n = 142 for dally gem males; n = 376 for dally ΔGAG females; n = 337 for dally ΔGAG males.The scale bars represent 100 μm (A); 30 μm (C); 200 μm (E).*p < 0.05; **p < 0.01; ***p < 0.001; n.s., not significant.

Figure 4 .
Figure 4. dally ΔGAG shows an ectopic wing vein phenotype.A, four examples of adult wings of dally ΔGAG .dally ΔGAG shows ectopic wing vein formation posterior to wing vein L5. Arrows point to the ectopic wing veins.B, the penetrance of the posterior wing vein defect.The box plot summarizes the penetrance data from four sets of independent experiments.This phenotype has never been observed in dally gem .n = 104 for dally gem females; n = 142 for dally gem males; n = 376 for dally ΔGAG females; n = 337 for dally ΔGAG males.C, pupal wings from wildtype (left) and dally ΔGAG (right) stained with anti-pMad antibody.Ectopic pMad signal was detected posterior to the wing vein L5 in dally ΔGAG pupal wing (asterisk, penetrance = 50%, n = 10 for wildtype; n = 10 for dally ΔGAG ).D, the posterior wing vein defect in dally ΔGAG heterozygotes (dally ΔGAG /+).The magnified image on the right is from the outlined inset from the left image.E, Dad, the inhibitory Smad, enhances the posterior wing vein defect of dally ΔGAG heterozygotes.Arrows indicate posterior ectopic veins.F, the ;(53)).In addition to the control of Lar activity during development, Dlp is required for starvation-induced synaptic plasticity.

Figure 5 .
Figure 5. Analyses of dlp ΔGAG phenotypes.A, embryonic cuticle is shown for wildtype, dlp MH20 (zygotic), and dlp ΔGAG .Two examples are shown for dlp MH20 mutants, which show a segment-polarity phenotype.B, ectopic mechanosensory bristles at the anterior wing margin in the wing of wildtype (top), Bx>dlp RNAi (middle), and dlp ΔGAG (bottom).Red arrows show the ectopic bristles.C, density plot showing the ratio of animals with the indicated number of the ectopic mechanosensory bristles for each genotype.n = 30 for wildtype; n = 29 for Bx>dlp RNAi; n = 30 for dlp ΔGAG .D, neuromuscular junction was stained with anti-HRP (axon) and anti-Dlg (postsynapse) antibodies.Signals for HRP and Dlg are shown in magenta and green, respectively.E, quantification of bouton number per muscle area (10,000 μm 2 ) in the neuromuscular junction of wildtype and dlp ΔGAG at 0 or 8 h of starvation.n = 35 for wildtype hour 0; n = 31 for wildtype hour 8; n = 26 for dlp ΔGAG hour 0; n = 21 for dlp ΔGAG hour 8.The scale bars represent 50 μm (A); 30 μm (B); 25 μm (D).**p < 0.01; ***p < 0.001; n.s.not significant.