The NDNF-like factor Nord is a Hedgehog-induced extracellular BMP modulator that regulates Drosophila wing patterning and growth

Hedgehog (Hh) and Bone Morphogenetic Proteins (BMPs) pattern the developing Drosophila wing by functioning as short- and long-range morphogens, respectively. Here, we show that a previously unknown Hh-dependent mechanism fine-tunes the activity of BMPs. Through genome-wide expression profiling of the Drosophila wing imaginal discs, we identify nord as a novel target gene of the Hh signaling pathway. Nord is related to the vertebrate Neuron-Derived Neurotrophic Factor (NDNF) involved in congenital hypogonadotropic hypogonadism and several types of cancer. Loss- and gain-of-function analyses implicate Nord in the regulation of wing growth and proper crossvein patterning. At the molecular level, we present biochemical evidence that Nord is a secreted BMP-binding protein and localizes to the extracellular matrix. Nord binds to Decapentaplegic (Dpp) or the heterodimer Dpp-Glass-bottom boat (Gbb) to modulate their release and activity. Furthermore, we demonstrate that Nord is a dosage-dependent BMP modulator, where low levels of Nord promote and high levels inhibit BMP signaling. Taken together, we propose that Hh-induced Nord expression fine-tunes both the range and strength of BMP signaling in the developing Drosophila wing.


INTRODUCTION
Morphogens are conserved, secreted signaling molecules that pattern organs and tissues by eliciting graded responses from cells surrounding a localized source (Lawrence and Struhl, 1996;Wolpert, 1969). Secreted signaling proteins of the epidermal growth factor (EGF), fibroblast growth factor (FGF), Hedgehog (Hh), Notch, transforming growth factor β (TGF-β)/ bone morphogenetic protein (BMP), and Wnt/Wingless (Wg) families have been shown to act as morphogens either at a short range or over a long distance. How these morphogens and their corresponding signaling pathways coordinate to control patterning and growth in various developmental systems is a central question in developmental biology.
The developing appendages of Drosophila provide a valuable model system to genetically and molecularly identify morphogens and members of their corresponding signaling pathways. They have also assisted in revealing the fundamental features and interplay between different morphogens during pattern formation. The Drosophila Hh and BMP signals function as shortand long-range morphogens, respectively, and together organize patterning along the anteroposterior (A/P) axis of the wing imaginal discs (Lecuit et al., 1996;Mullor et al., 1997;Nellen et al., 1996;Strigini and Cohen, 1997;Zecca et al., 1995).
In the case of Hh signaling, the quiescent state of the pathway is maintained by the Hh receptor Patched (Ptc)-dependent inhibition of Smoothened (Smo) (Ingham et al., 2011). This inhibition is lifted by Hh binding to Ptc and its co-receptors (Allen et al., 2011;Izzi et al., 2011;McLellan et al., 2008;Zheng et al., 2010), which releases Smo from repression and thus activates intracellular signal transduction Lum and Beachy, 2004;Varjosalo and Taipale, 2008). In the wing discs, Hh is secreted in the posterior (P) compartment and spreads toward the anterior (A) compartment (Basler and Struhl, 1994;Capdevila et al., 1994;Tabata and Kornberg, 1994). Hh signaling does not occur in P compartment cells because they do not express critical components of the Hh pathway, such as the major transcriptional effector Ci (Eaton and Kornberg, 1990). Conversely, A compartment cells can receive and respond to Hh but are unable to produce Hh. Hh synthesized by P cells diffuses into the A compartment adjacent to the A/P boundary, where it is sequestered by the Ptc-Ihog/Boi receptor complex (Chen and Struhl, 1996;Zheng et al., 2010), thus forming a short-range gradient that activates the localized expression of a number of target genes, including ptc and decapentaplegic (dpp) (Basler and Struhl, 1994;Capdevila et al., 1994;Chen and Struhl, 1996;Ingham et al., 1991;Tabata and Kornberg, 1994).
Dpp is the homolog of vertebrate TGF-β superfamily ligands BMP2/4, which binds to the BMP type I receptor thickveins (Tkv) or Saxophone (Sax) and type II receptor Punt (Parker et al., 2004). Upon ligand binding, Tkv or Sax phosphorylates the transcription factor Mad (referred to as pMad; an indication of active Dpp signal transduction). Phosphorylated Mad then forms a trimeric complex with the co-Smad Medea, and is translocated to the nucleus where it regulates downstream target genes (Moustakas and Heldin, 2009). In contrast to the Hh signal, Dpp diffuses into both the A and P compartments and acts as a long-range morphogen Lawrence and Struhl, 1996). Dpp has multiple roles both in determining the wing size by promoting growth and epithelial morphogenesis during larval stages (Gibson and Perrimon, 2005;Schwank et al., 2008;Shen and Dahmann, 2005), and in patterning the longitudinal veins (LV) as well as the anterior and posterior crossveins (ACV and PCV) during pupal development (de Celis et al., 1996;Ralston and Blair, 2005;Serpe et al., 2005;Shimmi et al., 2005;Sotillos and De Celis, 2005;Vilmos et al., 2005).
The diverse signaling outputs of Dpp are influenced by the formation of ligand heterodimers and by their binding to various extracellular factors. In Drosophila, two other characterized BMP family ligands, the BMP5/6/7/8 homolog Glass bottom boat (Gbb), and the more distantly related Screw (Scw) (Arora et al., 1994;Padgett et al., 1987;Wharton et al., 1991), can form heterodimers with Dpp to augment the level and increase the range of BMP signaling in different cells and tissues. Other secreted or membrane-binding BMP-binding proteins, including Short gastrulation (Sog) Serpe et al., 2005), Twisted gastrulation (Tsg), Crossveinless (Cv) (Shimmi et al., 2005;Vilmos et al., 2005), Crossveinless-2 (Cv-2) (Conley et al., 2000;Serpe et al., 2008), Larval translucida (ltl) (Szuperak et al., 2011), Pentagone (Pent) (Vuilleumier et al., 2010), Kekkon5 (Kek5) (Evans et al., 2009), Dally and Dally-like protein (Dlp) (Akiyama et al., 2008;Belenkaya et al., 2004), have been identified as modulators that enhance or inhibit BMP signaling. Additionally, Tkv, the main type I Dpp receptor in Drosophila, critically affects Dpp tissue distribution through ligand trapping and internalization, and thus globally shapes the BMP activity gradient (Crickmore and Mann, 2006;Lecuit and Cohen, 1998;Tanimoto et al., 2000). Prominent among the many determinants that impact proper establishment and maintenance of the Dpp activity gradient is the action of the Hh signal, which simultaneously induces Dpp expression and lowers responsiveness to Dpp in the same cells at the A/P border by repressing transcription of the tkv receptor gene (Tanimoto et al., 2000). The coordinated activation and attenuation of BMP signaling activity by Hh illustrates the complex regulatory interactions of different morphogens in pattern and growth in developing systems.
Here, we show that the activity of BMPs is further fine-tuned by a previously unknown Hhdependent mechanism involving induction of nord, which encodes an NDNF-like factor.
Absence of nord results in reduced medial BMP signaling near the source of Dpp in the larval wing imaginal discs but increases long-range BMP activity surrounding the primordial PCV in the pupal wing. Thus, adult wings from nord mutant flies exhibit seemingly opposite phenotypes of reduced wing size and ectopic PCV. Expression and co-immunoprecipitation studies in cultured S2 cells demonstrate that Nord is secreted, associates with extracellular matrix, and binds to Dpp or the heterodimer of Dpp-Gbb. We further show that Nord is a dosage-depend biphasic BMP modulator, where low levels of Nord promote and high levels inhibit BMP signaling. Taken together, we propose that Hh-induced Nord expression modulates the distribution and level of BMP signaling in the developing Drosophila larval and pupal wing to obtain proper size and crossvein patterning. genes, with no previously characterized expression pattern or functional analysis ( Figure 1B-D).
We first verified the differential expression of nord across the Drosophila wing imaginal discs via quantitative reverse transcription PCR using RNA isolated from FACS sorted A, B and P cells ( Figure 1E). We also performed in situ hybridization to localize nord transcripts in the wing imaginal discs ( Figure 1F). Like the known Hh signaling target gene ptc ( Figure 1E, F), nord transcripts were absent from the P compartment and primarily detected in the A cells adjacent to the A/P compartment boundary ( Figure 1E, F). Of note, unlike ptc and most other Hh pathway target genes, nord expression was not detected in the central wing pouch ( Figure 1F).
Collectively, these results suggested that nord is a potential target gene of the Drosophila Hh signaling pathway.

Hh signaling regulates nord expression in the wing discs
To further analyze the expression pattern and investigate the function of Nord, we identified and characterized two Minos-Mediated Integration Cassette (MiMIC) lines from the Drosophila Gene Disruption Project (GDP) collection (Venken et al., 2011), one a gene-trap nord allele Mi{MIC}nord MI06414 and the second a protein-trap nord allele Mi{PT-GFSTF.2}nord  We also generated an additional protein-trap nord allele Mi{PT-RFPHA.2}nord  . The  Figure 2A). In these nord protein-trap alleles, the GFSTF or RFPHA tag was inserted in the appropriate orientation and reading frame of nord between amino acids 103R and 104F (Figure 2-figure supplement 2), which permitted visualization of the Nord protein localization in vivo. Immunostaining of wing imaginal discs from the nord protein trap alleles showed that Nord is expressed along the A/P boundary within the hinge and notum but avoids the central wing pouch ( Figure 2B, C), identical to the pattern of endogenous nord transcripts revealed by in situ hybridization ( Figure 1F).
The selective upregulation of nord in the wing imaginal disc A cells adjacent to the A/P boundary indicated that nord expression may be controlled by Hh signaling activity.
We next examined this issue directly by following endogenous Nord expression in the proteintrap nord allele Mi{PT-GFSTF.2}nord  (hereafter referred to as nord-GFP). We used the heat shock (hs)-Gal4 driver (Halfon et al., 1997)  To investigate whether Hh signaling activity can cell autonomously alter the expression pattern of nord, we performed clonal analyses using the flip-out technique to ectopically inhibit or activate the Hh signaling pathway (Ito et al., 1997). The expression of endogenous Nord was examined in the 3rd instar wing discs carrying the protein-trap nord allele Mi{PT-RFPHA.2}nord  (hereafter referred to as nord-RFP) and flip-out clones expressing UAS-mCD8-GFP alone or in combination with either UAS-Ptc or a constitutively active form of

Nord belongs to an evolutionarily conserved family of secreted proteins
The Drosophila nord gene is associated with only one protein-coding transcript and one polypeptide containing 587 amino acids. It has a single homolog in the mouse genome, called Neuron-derived neurotrophic factor (NDNF) (Kuang et al., 2010). Nord and NDNF belong to a family of evolutionarily conserved secreted proteins with a predicted signal peptide followed by one or two Fibronectin type III-like repeats (FN3) and a domain of unknown function (pfam10179: DUF2369). Analyses of the genome and EST sequences from various organisms suggest that nearly all bilaterian animals have either single or multiple orthologous genes for Nord/Ndnf ( Figure 3A). We cloned the full-length cDNAs from Drosophila Nord and NDNF from several different species into a mammalian expression vector and found that both Nord and various NDNF proteins were efficiently secreted into the medium after transient transfection into HEK 293T cells ( Figure 3B). We immunostained un-permeabilized HEK 293T cells that were cotransfected with plasmids expressing Myc-tagged Nord or various NDNFs together with cytoplasmic-localized GFP proteins, and detected secreted Nord/NDNF proteins on the surface of the transfected cells ( Figure 3C). Interestingly, we also found Nord/NDNF proteins spread a short distance from the source cells (marked by cytoplasmic GFP expression) and deposited on the surface of the surrounding cells (arrows, Figure 3C). Consistent with the observations in cultured cells, secreted Nord was detected outside of flip-out clones expressing a HA-tagged Nord ( Figure 3D). Furthermore, secreted Nord-GFP was detected in vivo when wing imaginal discs from 3 rd instar larvae carrying nord-GFP were immunostained in the absence of detergent ( Figure 3E). Together, these data demonstrated that Nord and its homolog NDNF belong to a family of secreted proteins.

Nord is required for proper growth and crossvein patterning of the Drosophila wing
To investigate the function of Nord, we generated and characterized several nord loss-offunction alleles by using different genetic strategies. We first analyzed the phenotypes of two nord mutant alleles, nord 3D and nord 22A , created by the CRISPR/Cas9 strategy. Both alleles carry short deletions in the fourth coding exon that cause frameshifts and premature stop codons within the DUF2369 domain of Nord ( Figure 4A; Figure 4-figure supplement 1). Animals homozygous or trans-heterozygous for nord 22A and nord 3D are viable and fertile. As homozygous or trans-heterozygotes, they show no obvious defects in the shape of the wing blade. However, a consistent and significant decrease in wing size was observed in in both male and female flies homozygous flies are viable and fertile, and both male and female nord mutants showed a significant wing size reduction when compared to their wild-type counterpart ( Figure 4C).
Along with the growth defects in the wing, we also noted that 61% of females (59/97) and 21% (17/82) of males of the nord 22A allele showed ectopic venation in the vicinity of the posterior crossvein of one or both wings ( Figure 4D, E). The ectopic vein tissue was also found emanating from the PCV in the wing of adult flies homozygous for the nord MI06414 allele. Like nord 22A , the ectopic venation displayed higher penetrance in females than in males, with 86% of mutant females and 36% of mutant males showing additional vein material at one or both wings ( Figure 4D, E). This ectopic venation phenotype was not observed in nord 3D homozygotes.
To further address whether the ectopic venation phenotype is caused by loss-of-nord, we performed genetic complementation tests between nord MI06414 , nord 3D and nord 22A with various deficiency lines covering the nord locus ( . We next tested Df(2R)BSC155 in trans to either nord 22A and nord 3D and once again found that nord 22A exhibited high frequency ectopic venation ( Figure 4D, E; 69% females and 26% males), while nord 3D showed a modest interaction (14% females and <1% males). Similar to the homozygotes, both nord 22A and nord 3D in trans to Df(2R)BSC155 also produced smaller wings compared to controls (Figure 4-figure supplement 5). To examine whether this size difference was the result of reduced cell proliferation or a smaller cell size, we examined wing tricome density as a proxy for cell size (Brummel et al., 1999;Dobzhansky, 1929;Martin-Castellanos and Edgar, 2002). We found no difference for nord MI06414 / Df(2R)BSC155 or nord 22A / Df(2R)BSC155 compared to heterozygous (nord MI06414 / + or nord 22A / +) controls (Figure 4-figure supplement 6) suggesting that the alteration in wing size is caused by reduced cell proliferation. Additionally, both the ectopic venation phenotype and the variable penetrance in males and females were observed when nord was knocked down via wing-specific Gal4 (MS1096-Gal4) driven expression of UAS-nord-RNAi ( Figure 4D, E), which further validates that the observed phenotypes in the nord mutants are not indirect effects associated with loss-of-nord in tissues other than the wing discs. Together, these experiments indicate that Nord is required in wing imaginal discs for both proper growth and crossvein patterning and that the nord 3D allele may retain some function.

Spatial-temporal overlapping expression of nord and dpp in the developing Drosophila
wing Both wing growth and crossvein patterning require precisely controlled BMP signaling activity (Affolter and Basler, 2007;De Celis, 2003). During wing development, BMP signaling activity is an output of the combined action of two BMP ligands, the Drosophila BMP2/4 homologue Dpp and the BMP5/6/7/8 homologue Gbb (Bangi and Wharton, 2006). Gbb is broadly and uniformly expressed in the larval and pupal wing, while Dpp, a well-known target of Hh signaling, is expressed in a stripe of cells in the anterior compartment along the A/P compartmental boundary of the larval wing imaginal disc. From this source, Dpp protein is thought to spread and form a concentration gradient to control patterning and growth of the wing imaginal disc. In agreement with our finding that like dpp, nord is a target gene of the Hh signaling pathway in the wing imaginal discs (Figures 1, 2), we observed a spatial-temporal correspondence between nord and dpp expression. Both Nord protein indicated by the Nord-GFP fusion derived from the nord-GFP protein trap allele and Dpp precursor protein detected via an anti-Dpp prodomain antibody were present along the A/P boundary flanking the central wing pouch through the larval stage ( Figure 5A). In the pupal wing, both Nord and Dpp expression does not change during the first 8 hours after pupation (AP) ( Figure 5A). Subsequently, dpp disappears from the A/P boundary and commences expression in the differentiating longitudinal veins (deCelis, 1997;Yu et al., 1996), however, the Hh-dependent expression of nord along the A/P boundary remains for about 30 hours after pupation, and diminishes after Dpp expression was detectable in both the LV and PCV region ( Figure 5A; Figure  . We therefore asked whether elimination of Nord alters the level of phosphorylated Mad (pMad), the primary downstream signal transducer of BMP signaling in the wing disc. We quantified pMad signal intensity in nord mutant wing discs and compared it to that of wild-type controls. We found that the pMad intensity was slightly reduced in nord mutant wing discs, and this reduction of pMad levels was more evident in the A compartment where Nord is expressed ( Figure 5B, C; arrow). In contrast, by measuring the expression of high-threshold target gene ptc and low-threshold gene dpp, we found no obvious difference in the Hh signaling activity in the wild-type and nord mutant discs Of note, beside BMPs, Hh signaling also patterns the developing Drosophila wing. Hh shortrange activity is responsible for patterning the central L3-L4 region and determining the distance between L3 and L4 longitudinal veins (Mullor et al., 1997;Strigini and Cohen, 1997). When quantifying the distance between distal ends of longitudinal veins L3 and L4 (dL3-L4) and that of L2 and L5 (dL2-L5), we found no specific reduction in the L3-L4 region indicated by comparable or higher dL3-L4 /dL2-L5 ratio in the female and male nord mutant wings ( Figure 4B, C), which is consistent with the obviously reduced BMP, but not Hh, signaling activity during wing imaginal disc development ( Figure 5B, C). Together, these findings suggested a positive role of endogenous Nord in augmenting BMP signaling activity to promote wing growth.
During pupal wing development, BMP signaling is activated in the prospective CV regions prior to the appearance of other known vein promoting signals (O'Connor et al., 2006;Ralston and Blair, 2005), and abnormal BMP signaling can selectively affect the PCV and leave the longitudinal veins largely or entirely intact (deCelis, 1997;Haerry et al., 1998;Khalsa et al., 1998;Nguyen et al., 1998;Ralston and Blair, 2005;Ray and Wharton, 2001;Wharton et al., 1999;Yu et al., 1996). Therefore, we assessed the possible role of Nord in Dpp/BMP signal transduction during crossvein patterning. It is known that pMad becomes gradually refined to a narrow strip of precursor cells that form the future PCV during the first 24-28 hours AP (Conley et al., 2000;Shimmi et al., 2005). To examine whether the ectopic PCV in nord mutants is a direct consequence of enhanced BMP signaling, we quantified pMad signal intensity at the presumptive PCV region in pupal wings at varies time points after pupation. In contrast to the rather restricted pMad domain in wild-type pupal wings, we detected a broadened pMad domain around the presumptive PCV in nord mutants from 19-20 hours AP ( Figure 5D, E). Gradually, the ectopic pMad accumulation became an expanded patch adjacent to the presumptive PCV, indicating that BMP signaling was abnormally elevated in this region ( Figure 5D, E). In addition, ectopic pMad at PCV primordia was frequently observed when nord was knocked down via the wing-specific MS1096-Gal4 or A9-Gal4 driven expression of UAS-nord-RNAi throughout the wing discs ( Figure 4D

Nord is a biphasic modulator of BMP signaling during wing growth
To better understand the role of Nord in modulating BMP signaling in vivo, we generated transgenic flies carrying a Gal4-inducible UAS-Nord transgene. Under control of the ubiquitous wing blade driver nub-Gal4, ectopic Nord expression resulted in reduced range and level of the pMad gradient in the 3 rd instar wing imaginal discs ( Figure 6A, B, 25°C), and accordingly decreased wing size in both adult males and females ( Figure 6C, D, 25°C; Figure 6-figure supplement 1). In flies, minimal Gal4 activity is present at 16°C, while 29°C provides a balance between maximal Gal4 activity with minimal effects on fertility and viability due to growth at high temperature (Duffy, 2002). Taking advantage of the temperature dependent nature of Gal4 activity in Drosophila, we compared the dosage effect of Nord on pMad intensity in larvae raised at two different temperatures, 25°C and 29°C. Indeed, in animals raised at 29°C expressing higher levels of exogenous Nord, we detected both a much reduced pMad gradient in the larval wing discs and more severely decreased wing size in the adult flies ( Figure 6A-D; Figure 6figure supplement 1). In contrast, Hh signaling activity is relatively normal in the wing discs expressing ectopic Nord. Although the Ptc expressing domain became narrower, we did not notice any obvious decrease in the levels of Ptc expression. (Figure 6A, B). Consistently, ectopic Nord expression did not cause any specific reduction in the distance of L3 and L4 based on the dL3-L4 /dL2-L5 ratio ( Figure 6D), suggesting the wing growth defect caused by ectopic Nord is unlikely due to inhibition of Hh signaling. Interestingly, partial L5 and PCV loss was also noticed in the wings with more dramatic size reduction ( Figure 6C, arrowheads and arrows) and the frequency of disrupted L5 and PCV were more dramatic when the flies were raised at higher temperatures and expressed higher levels of ectopic Nord ( Figure 6C). Therefore, our observations indicate that ectopic Nord attenuates BMP signaling leading to inhibition of wing growth and vein patterning. Along with the positive role of endogenous Nord in enhancing BMP signaling to promote wing growth, we propose a model that Nord has both positive and negative effects in modulating BMP signaling activity, where low (endogenous) levels of Nord enhance and high (ectopic) levels of Nord inhibit BMP signaling.

Nord is a biphasic modulator of BMP signaling during PCV patterning
To further examine the model that Nord is a biphasic modulator of BMP signaling, we sought to manipulate Nord levels during pupal wing development in the posterior compartment and examine the effects on PCV formation since with this structure it is possible to assay both positive and negative roles by looking for ectopic versus loss of crossvein formation. Accordingly, we used both hh-Gal4 and en-Gal4 to drive different levels of ectopic Nord in the P compartment of the wing disc, including the PCV primordia where Nord is not normally expressed. To avoid the influence of prior larval stage Nord expression on the role of BMP signaling specifically during PCV pupal development, we used Gal4 together with tub-Gal80 ts (a temperature-sensitive version of Gal80) to temporally control UAS-Nord expression. At low temperature (18°C) Gal80 ts represses the function of Gal4 bound to a UAS sequence but is unable to do so at the restrictive temperatures (29°C) (McGuire et al., 2003). We performed temperature-shift experiments (from 18°C to 29°C) to initiate ectopic Nord expression at different times during pupal development and characterized the impact on PCV patterning in the resulting adult wings. Consistent with a previous report (Roberts 1998), the length of pupal period became shorter after the temperature was raised from 18°C to 29°C, due to temperaturedependent effects on growth rate and we found that shifting the temperature to 29°C right after pupation, leads to eclosion ~ 84 -96 hours later ( Figure 7A). We found that activation of UAS-Nord expression by temperature shifts at 78 hours or earlier before eclosion caused the most severe PCV phenotype, but activation at 66 hours before eclosion or later resulted in essentially normal PCV ( The critical time window in which the primordial PCV responds to exogenously expressed Nord (66 -78 hours before eclosion at 29°C) coincides with the stage during which BMP signaling induces PCV formation. We next tested whether the PCV phenotypes caused by ectopic Nord were correlated with alterations of BMP signaling activity. As shown in Figure 7D, adult PCV defects and pMad patterns in pupal wings showed a correlation with the level of ectopic Nord, where the temperature shift occurred 12 hours after the start of pupal development. In pupal wings expressing a low level of exogenous Nord (1x UAS-Nord), ectopic pMad and ectopic crossveins were detected, while moderate overexpression (2x UAS-Nord) lead to a mixed phenotype of partial pMad and crossvein vein loss or wider pMad and ectopic veins. When high Nord levels (3x UAS-Nord) were expressed in the pupal wings, pMad and the PCV were largely absent. Although the temporal resolution is somewhat limited, the results clearly indicate that the level of Nord influences the outcome of PCV patterning during the early pupal development where a lower level of exogenous Nord resulted in enhanced BMP signaling and ectopic PCV, while higher levels of exogenous Nord inhibited BMP signaling and caused disrupted or depleted PCVs. Taken together, these results demonstrated that Nord is a dosage-depend biphasic modulator of BMP signaling both in wing growth and crossvein patterning.

Nord binds to Dpp and interferes with BMP signaling in vitro
The activity of BMPs are modulated by a large variety of binding proteins that can either enhance or inhibit their signaling in a context dependent manner (Chang, 2016;Umulis et al., 2009). Given the spatial-temporal overlapping expression of nord and dpp and the dosagedepend biphasic modulation of Nord on BMP signaling in wing growth and crossvein patterning, we assessed whether Nord modulates BMP signaling via binding to either of the two BMP ligands, Dpp and Gbb, that are expressed in the developing Drosophila wing. We turned to an in vitro model to examine possible interactions of Nord with Dpp and Gbb by carrying out coimmunoprecipitation assays. We added a GFP tag to the C-terminus of Nord and expressed the fusion protein in Drosophila S2 cells. The conditioned medium from Nord-GFP-expressing cells was collected and mixed with medium from cells transfected with Flag-tagged Dpp and HAtagged Gbb alone or in combination. The mixed-media were then incubated with anti-FLAG or anti-HA antibody coupled beads to precipitate the BMP ligands. We found that Nord coprecipitated with Dpp and, to a lesser extent, with Gbb ( Figure 8A). Additionally, we observed an increased level of Nord proteins co-precipitated with Dpp or Gbb when Dpp and Gbb were co-expressed ( Figure 8A), indicating Nord may have a higher affinity for Dpp-Gbb heterodimers formed in cells co-expressing Dpp and Gbb.
We next collected conditioned medium from S2 cells expressing Dpp or Dpp-Gbb with or without co-expressed Nord (source cell) and compared the levels of processed Dpp or Dpp-Gbb within the different conditioned media. We found that far less cleaved Dpp protein was released into the conditioned medium when Nord was co-expressed ( Figure  Upon incubation of the Mad-S2 responding cells with conditioned medium collected from the source cells, BMP signaling activity was monitored by measuring the pMad signal intensity. Conditioned medium containing either Dpp or Dpp-Gbb, but not that containing Nord alone, was able to induce Mad phosphorylation ( Figure 8C). In agreement with the dramatically reduced amount of Dpp or Dpp-Gbb ligands ( Figure 8B), the conditioned medium collected from source cells co-expressing Nord showed lower pMad signal intensity compared to that collected from source cells lacking Nord co-expression ( Figure 8C, D). Of note, Nord is not a membranetethered protein ( Figure 3A), but we observed noticeable amounts of Nord deposited on the surface of source cells, as well as the surrounding extracellular matrix ( Figure 3C, D). Thus, the matrix-associated Nord may sequester Dpp and Dpp-Gbb ligands to the source cells and thereby reduce ligand level in the media, which in turn lead to decreased BMP signaling activity in the responding cells ( Figure 8C, D).

Biphasic modulation of BMP signaling by Nord in vitro
In vivo, our loss-and gain-of-function analyses suggested that Nord is a dosage-depend biphasic modulator of BMP signaling in wing growth and crossvein patterning. In the in vitro signaling assay, ectopically expressing Nord in the source cells led to reduced BMP ligand release and thus decreased the signaling activity (pMad/Mad) from an equal volume of conditioned medium ( Figure 8D). However, when the signaling activity was further normalized to the ligand amount ( Figure 8B), the relative ligand activity [(pMad/Mad)/BMPs] from the medium without Nord is lower than that with Nord ( Figure 8E), suggesting that while ectopic Nord expressed in the source cells reduced the levels of released ligand, that soluble ligand appears to have a higher signaling activity perhaps as a result of an association with Nord ( Figure 8E). Notably, the highest level of Nord supplied from conditional medium failed to completely inhibit the free rDpp induced BMP signaling. We next transiently transfected Mad-S2 cells for expression of Nord-GFP. Similar to Nord supplied from conditioned medium, we found that transient expression of Nord-GFP in the Mad-S2 cells partially inhibited the average Mad phosphorylation induced by exogenous rDpp ( Figure 8H). Remarkably, immunofluorescent

DISCUSSION
In Drosophila, the short-range morphogen Hh and the long-range morphogen BMP function together to organize wing patterning (Lecuit et al., 1996;Mullor et al., 1997;Nellen et al., 1996;Strigini and Cohen, 1997;Zecca et al., 1995). It has been previously shown that the Hh signal shapes the activity gradient of BMP by both inducing expression of Dpp, and simultaneously downregulating the Dpp receptor Tkv resulting in lower responsiveness to Dpp in cells at the A/P compartment border (Tanimoto et al., 2000). In this study, we show that the activity of BMP is further fine-tuned by another previously unknown Hh-dependent mechanism. Using a genome-wide expression profiling of the Drosophila wing imaginal discs, we identify nord, which encodes a secreted protein localized to the extracellular matrix, as a novel target gene of the Hh signaling pathway (Figures 1-3 (Figures 6, 7). At the molecular level, we showed that Nord is a BMP binding protein that directly enhances or inhibits BMP signaling in cultured S2 cells (Figure 8). Combining the genetic and biochemical evidence, we propose that Nord-mediated binding of Dpp and Dpp-Gbb not only enhances the local BMP signaling activity by augmenting ligand concentration near the Nord/Dpp secreting cells, but also reduces unwanted BMP signals in a nonautonomous fashion by impeding the mobilization of the long-range BMP signals provided by Dpp and Dpp/Gbb homo and heterodimers respectively, from the A/P boundary or the differentiating LV4 (Figure 9). Loss-of-nord thus leads to reduced local BMP signaling (medially) flanking the A/P boundary and increased long-range BMP activities surrounding the primordial PCV, giving rise to the seemingly opposite phenotypes of reduced wing size and ectopic PCV ( Figure 4). Taken together, we propose that Hh-induced Nord expression provides an exquisite regulation of the strength and range of BMP signaling in the developing Drosophila wing.

Nord is a novel, multi-functional BMP binding protein
The activity of TGF-β type factors, including the BMP subfamily, are modulated by a large variety of binding proteins that can either enhance or inhibit their signaling in a context dependent manner (Chang, 2016;Umulis et al., 2009). These modulator proteins vary broadly in structure, location, and mechanism of action. Well known extracellular and freely diffusible proteins include Noggin, Tsg, Follistatin, the CR (cysteine-rich) domain containing proteins such as Chordin/Sog, and the Can family named after two founding members, Dan and Cerberus (Chang, 2016). With the exception of Tsg, all of these factors appear to behave as antagonists, where BMP binding prevents association of the ligand with the receptor complex.
The other broad category of BMP binding proteins includes membrane-bound or matrixassociated proteins and, in contrast to the highly diffusible class of BMP binding factors, these proteins often act as either agonists or antagonists depending on context. These proteins are also structurally diverse but to date, none contains FN3 or DUF2369 domains that are characteristic of Nord and NDNF, its vertebrate counterpart. From a mechanistic point of view, perhaps the two most instructive Drosophila members of this class of modulators are the heparan sulfate proteoglycan (HSPG) Dally and the CR containing protein Cv-2. HSPGs are well characterized as modulators of growth factor signaling (Nakato and Li, 2016). In the case of FGFs, HSPGs act as true co-receptors in which they form a tripartite complex with ligand and FGFR, the signaling receptor (DiGabriele et al., 1998;Schlessinger et al., 2000). However, they can also mediate signaling in other ways. Analysis of dally loss-of-function clones in imaginal discs demonstrate that it has both cell autonomous and non-autonomous effects with respect to BMP signaling (Akiyama et al., 2008;Belenkaya et al., 2004;Fujise et al., 2003). In general, low levels tend to promote signaling while high doses attenuate signaling. Many models have been put forth to explain these opposing effects and often come down to balancing ligand sequestration and diffusion properties. For instance, in the absence of HSPGs, Dpp may more freely diffuse away from the disc epithelial cell surface. In this case HSPG acts to enhance signaling by keeping Dpp tethered to the cell surface where it can engage it's signaling receptors. On the other hand, a high level of HSPG may compete with signaling receptors for BMP binding and thereby reduce signal (Nakato and Li, 2016).
The situation with respect to signal modulation becomes even more complex for factors such as Nord that bind both HSPGs (Akiyama et al. 2021 accompanying manuscript) and BMPs (this report). An instructive example to consider is Cv-2, a secreted factor that, like Nord, binds both to HSPGs and BMPs and is also induced by BMP signaling (Serpe et al. 2008). Like Dally, Cv-2 also has dose dependent effects on signaling in wing imaginal discs, where low levels enhance while high levels inhibit BMP signaling. By virtue of being bound to HSPGs it may simply function as an additional tethering molecule that keeps BMPs localized near the cell surface. However, Cv-2 has the unique property that it is also able to bind TKV, a Drosophila BMPR type I receptor (Serpe et al., 2008). This has led to speculation that it could act as an exchange factor that aids in handing off a BMP ligand from the HSPG pool to the type I receptor. Mathematical modeling showed that this mechanism can produce a biphasic signal depending on affinities of the various BMP binding proteins involved and their concentrations (Serpe et al., 2008).
In the case of Nord, its mechanism of action is likely compatible with a variety of these and/or alternative models. While we have shown that Nord is a BMP binding protein and Akiyama et al. (2021) have shown that it also binds HSPGs, it is not clear whether the BMP and HSPG binding sites overlap or are distinct and where they are positioned relative to the FN3 and DUF2369 domains. This is an important issue to consider with respect to the two CRISPR mutants that we generated which truncate Nord within the DUF2369 domain. Interestingly, the nord 3D allele appears to retain some function since it does not generate ectopic crossveins as do the nord MI06414 or nord 22A alleles, yet nord 3D still produces small wings in transheterozygous combination with a deficiency or nord 22A , consistent with having lost the BMP growth promoting ability. The discrepancy in crossvein patterning between the different nord alleles may be explained by a difference in residual function of the various truncated Nord protein products ( Figure 4). Because the nord MI06414 allele yields a much shorter predicted Nord peptide compared to the two CRISPR alleles, it is likely to behave as a protein null with a stronger phenotype. The two nord CRISPR alleles, although similar in the sequence deleted from the Cterminus, differ in how many non-nord encoded amino acids occur between the frameshift and the stop codon. The nord 22A allele has an additional 14 amino acids relative to nord 3D . Perhaps this extension of the truncated fragment destabilizes or interferes with residual function found in the nord 3D allele. Additional biochemical studies defining the BMP and HSPG binding sites, the stability of truncated Nord fragments, and whether Nord can also associate with either the type I or II receptors will aid in formulating a more precise mechanistic model.

Is Nord structure and function conserved across species?
Nord shows some sequence similarity to the neuron derived neurotrophic factor NDNF family of proteins ( Figure 3A). Based on a very recent study, like many other neurotrophic factors, NDNF arose in the ancestor of bilaterians or even later (Heger et al., 2020). In agreement, by analyzing the genome and EST sequences from various organisms, we found that nearly all bilaterian animals have either single or multiple orthologous genes for Nord/Ndnf ( Figure 3A). Of note, we did not identify any Ndnf homologues in the flatworm Planarian, but these factors are highly conserved across vertebrates (Kuang et al., 2010). All vertebrate family members contain a signal peptide, two FN3-like repeats and a domain of unknown function (DUF2369) that is now referred to as the NDNF domain. The NDNF domain partially overlaps with the first FN3 but shows some additional conservation that extends between the two FN3 domains ( Figure 3A).
The FN3 module is quite diverse in sequence but is thought to exhibit a common fold that is used as an interaction surface or spacer (Campbell and Spitzfaden, 1994;Koide et al., 1998).
The function of the NDNF domain is not clear but it may also provide a protein interaction surface.
Although the vertebrate NDNFs are highly conserved throughout the entire protein length, the C. elegans and Drosophila relatives are quite divergent in primary sequence and show little conservation beyond a few key residues that define the second FN3 and NDNF domains (Kuang et al., 2010). Notably, the Drosophila protein is missing the first FN3 domain and therefore it is not clear the extent to which they may exhibit functional conservation. Ironically, the original human NDNF clone was identified on the basis of domain structure conservation with Drosophila Nord which was identified via enhancer trapping to be a gene expressed in mushroom bodies and whose loss leads to defects in olfactory learning and memory (Dubnau et al., 2003). Unfortunately, that particular LacZ enhancer trap line which disrupted the nord locus is no longer available. The use of our new alleles should prove helpful for either confirming or eliminating the involvement of Nord as a modulator of learning and memory and/or other neuronal functions in larva and adult Drosophila.
In the mouse, NDNF is highly expressed in many neurons of the brain and spinal cord (Boyle et al., 2011;Kuang et al., 2010;Schuman et al., 2019). Studies using cultured mouse hippocampal neurons revealed that it promotes neuron migration and neurite outgrowth and hence its name (Kuang et al., 2010). In later studies NDNF was also found to be up-regulated in mouse endothelial cells in response to hind-limb ischemia, where it promotes endothelial cell and cardiomyocyte survival (Ohashi et al., 2014). Additionally, recent studies have shown that NDNF expression is significantly down regulated in human lung adenocarcinoma (LUAD) and renal cell carcinoma (RCC), indicating NDNF may also provide a beneficial function as a tumor suppressor (Xia et al., 2019;Zhang et al., 2019).
Taken together, these studies have suggested some possible functions for vertebrate NDNF.
However, they have primarily relied on in vitro cell culture models and only recently have in vivo loss-of-function studies been reported (Messina et al., 2020). Remarkably, NDNF mutants were discovered in the genomes of several probands with Congenital Hypogonadotropic Hypogonadism (CHH), a rare genetic disorder that is characterized by absence of puberty, infertility and anosmia (loss of smell) (Boehm et al., 2015;Lima Amato et al., 2017). This phenotype is very similar to that produced by loss of the anos1 which also encodes a FN3 superfamily member and is responsible for Kallmann syndrome, a condition that similarly presents with CHH and anosmia due to lack of proper GnRH and olfactory neuron migration (Stamou and Georgopoulos, 2018). These migration defects also occur in zebrafish with NDNF mutations (Messina et al. 2020).
The results of our study on the function of Drosophila Nord raises the issue of whether any of the ascribed vertebrate NDNF functions could involve alterations in BMP signaling. In the case of angiogenesis and EMT, BMPs, as well as other TGF-β family members, participate at many levels (Goumans et al., 2018;Kahata et al., 2018). At present, however, no involvement of BMP or TGF-β signaling has been implicated in migration of the GnRH neurons, although BMP signaling does define neurogenic permissive areas in which the olfactory placode forms (Forni

Drosophila maintenance
Animals were grown on standard food containing molasses at room temperature unless otherwise indicated. The hs-FLP and actin>y+>Gal4 (Ito et al., 1997)  Drosophila stocks used in this study are listed in the Key Resources Table. Dissociation and sorting of imaginal disc cells Wing imaginal discs were dissected from wandering third instar larvae of the genotypes hh-Gal4; UAS-mCD8-GFP or ptc-Gal4; UAS-mCD8-GFP. Discs were stored in Schneider's Drosophila Media (21720, Invitrogen) plus 10% FBS (10438, Invitrogen) on ice for less than two hours prior to cell dissociation. Discs were washed twice with 1 ml cell dissociation buffer (Sigma, C-1544). Elastase (Sigma, E-0258) was diluted to 0.4 mg/ml in fresh cell dissociation buffer once discs were ready. Discs were incubated for 20 min at room temperature in 0.4 mg/ml elastase with stirring by a magnetic micro stirring bar. Undissociated tissue was spun out, cell viability was measured (>80%), and cells were immediately isolated using the BD FACSAria system. Dead cells labeled with propidium iodide (P3566, Invitrogen) were excluded during FACS, and purity of sorted cells was greater than 99% by post-sorting FACS analysis. Total RNA was extracted from sorted cells (RNeasy, Qiagen) and stored at −80°C. Quality was assessed with the Agilent Bioanalyzer 2100 (RIN > 7.0). transformed. Probe sets were mapped to genes using the Drosophila_2.na30.annot.xml annotation file, available from the Affymetrix website. 14,448 of 18,952 (76.2%) probe sets map to gene isoforms-13,016 (90.1%) of which correspond to unique genes (some genes are mapped by ≥1 probe set). Probe sets mapping to the same gene were not combined to minimize technical artifacts. Genes (probe sets) whose expression is not only higher in A cells than P cells (FoldA/P > 1.2), but also higher in the A/P boundary adjacent cells than general A cells (FoldB/A > 1.5) were selected as potential Hh-induced target genes (Figure 1). A total of 61 probe sets (59 unique genes) were identified as potential Hh signaling target genes (Supplementary file 1). The heatmap was generated in R with the ggplot2 package and the genes were ordered by the FoldB/A change.

Total RNA was extracted from FACS-sorted A, B and P cells (see details in Dissociation and sorting of imaginal disc cells). Possible contamination of genomic DNA was excluded by
treatments of DNAse I (AM2222, Thermo Fisher Scientific). RNA was reverse-transcribed to cDNA using Maxima Reverse Transcriptase (EP0742, Thermo Fisher Scientific) with random hexamers. All samples within an experiment were reverse-transcribed at the same time; the resulting cDNA was stored in aliquots at -80°C until used. cDNA was PCR-amplified using SYBR Green Supermix (1708880; Bio-Rad). qPCR was carried out with an ABI PRISM Sequence Detection System (Applied Biosystems). Reactions were run in triplicate in 3 independent experiments. Expression data were normalized to the geometric mean of the housekeeping gene pkg and were analyzed using the 2-ΔΔCT method. The primer sequences are provided in Key Resources Table.

In situ hybridization
In situ hybridization of wing discs was performed as previously described (Hsia et al., 2017).
Briefly, RNA probes were created from in vitro transcription of PCR products carrying the T7 RNA polymerase recognition sequence at one end and synthesized by using a digoxigenin (Dig)-labeling kit (Roche). Wing discs of L3 larvae were hybridized with probes overnight at 56 °C using standard procedures and visualized using anti-Dig-AP (1:1,000; Roche). Primers used for generating PCR templates are listed in Key Resources Table.

Generation of Nord Crispr alleles
The nord 3D and nord 22A alleles were generated using the Crispr/Cas9 system. The following guides 5'-GGACCTGTTCGGAATCCACC-3' and 5'-GGGTGAGGTTCTGTCTACCC-3 were separately cloned into the BbsI site of pU6-BbsI-chiRNA plasmid (obtained from Addgene) and both were simultaneously injected by Best Gene into w 1118 ; PBac{y[+mDint2]=vas-Cas9}VK00027 on chromosome 3 (Bloomington Stock Center #51324). G0 flies were crossed to a balancer stock (w; Pin/Cyo Star ) and then individual males were crossed to w; Gla, Bc/CyO{GFP} to establish stocks. DNA from homozygous adults was amplified by PCR using primers that flanked the two Crispr target sites and sequenced. The nord 22A allele was a five bp deletion generated at guide sequence 2 site, while the nord 3D allele was an 11bp deletion generated at the guide 1 site.

Cell culture and transfection
Drosophila S2 cells (S2-DGRC) were obtained directly from the Drosophila Genomics Resource Center (DGRC) and cultured in Drosophila Schneider's medium supplemented with 10% of fetal bovine serum (Omega Scientific) and 1% Penicillin-Streptomycin-Glutamine (Thermo Fisher) at 25°C in a humidified incubator. S2 cells stably expressing the FLAG-Mad transgene (Mad-S2) cells were generated by co-transfecting pBRAcpA-FLAG-Mad (Jensen et al., 2009) and pCoBlast, and then followed by selection with 12.5 µg/ml blasticidin. HEK 293 cells were obtained directly from ATCC and cultured in Dulbecco's Minimal Essential Medium with 10% fetal bovine serum (Omega Scientific) and 1% Penicillin-Streptomycin-Glutamine (Thermo Fisher) at 37°C in a humidified incubator with 5% CO2. Transfection was performed with FuGENE 6 transfection reagent (Promega). All the cell lines were regularly confirmed to be free of contamination (e.g. mycoplasma) through PCR-based tests as recommended by the NIH.

Constructs
The coding sequence of Nord or Ndnf from various species were fused in frame with C-terminal GFP tag or HisMyc tag, and cloned into the pAcSV vector for expression in the Drosophila S2 cells, or into the pcDNA3.1 vector for expression in the HEK 293 cells. The coding sequence of Nord was fused with a C-terminal HA-GFP tag and cloned into the pUAST vector to generate the transgenic UAS-Nord line. Constructs expressing Dpp-FLAG, Gbb-FLAG, Gbb-HA, and FLAG-Mad were generated using PCR methods to tag and amplify the gene of interest from a full-length cDNA and then cloned into the S2 cell constitutive expression vector pBRAcpA (Cerbas et al.1994). The sequences are provided in Supplementary file 3.

Imaginal discs and pupal wing immunostaining and imaging
Wing discs from 3 rd instar larvae were dissected, fixed in 4% formaldehyde in PBS, blocked and permeabilized by 5% normal goat serum (NGS) & 0.3% Triton X-100 in PBS, incubated with primary antibody in PBS containing 5% NGS and 0.3% Triton X-100 overnight at 4°C, washed 3 times with 0.3% Triton X-100/PBS, incubated with secondary antibody, and washed with 0.3% Triton X-100/PBS. To selectively stain the secreted Nord (Figure 3), the above immunostaining procedure was carried out in the absence of Triton X-100 (PBS alone for blocking and antibody incubation buffer; 0.01% Tween-20/ in PBS for washing buffer). Pupal wings were collected and pre-fixed as previously described (Classen et al., 2008), then followed by the procedure described above for immunostaining of the larval wing discs. The stained larval wing discs or pupal wings were mounted and imaged with a ZEISS spinning disc confocal microscope.

Image collection and quantification of fluorescence intensity
To compare the expression profile of pMad, Dpp, or Ptc in different genotypes, we used wing imaginal discs at the same developmental stage, which were dissected from wandering larvae, corresponding to 1-6 h before the entry into the pupal stage. Larvae from the control and corresponding experimental group were raised at the same temperature and density. Wing discs were dissected, fixed, immunostained, and mounted by following the same protocol. All images were taken using the same confocal microscope settings. The pixel intensities of pMad, Dpp, or Ptc were obtained within a fixed rectangular region across both ventral and dorsal compartments using the Plot Profile function of Fiji. Then, the average pixel intensities from multiple discs were plotted using the GraphPad Prism software. The number of wing imaginal discs used in each experiment was provided in the corresponding figure legend.

Cell immunostaining
Forty-eight hours after transfection, NIH 293 cells were washed twice with PBS, fixed in 4% formaldehyde in PBS, blocked by 1.5% normal goat serum (NGS) in PBS, incubated with the primary antibody in PBS containing 1.5% NGS for overnight at 4° (to stain surface Nord or Ndnf), washed with 0.01% Tween-20/1XPBS, incubated the with secondary antibody and washed with 0.01% Tween-20/PBS. Mad-S2 cells immunostaining was carried out through similar procedures but in the presence of 0.1% Triton X-100 during blocking, antibody incubation, and washing steps. The stained cells were mounted and imaged with a ZEISS spinning disc confocal microscope.

Immunoprecipitation Assay
S2 cells were separately transfected to express Nord with a C-terminal GFP tag, or the FLAGtagged Dpp, HA-tagged Gbb alone or Dpp-FLAG/Gbb-HA in combination. >72 hours after transfection, conditioned medium from transfected cells were collected. The medium containing Nord or the BMP ligands were mixed, followed by incubation with anti-FLAG or anti-HA antibody coupled beads (Anti-M2 Affinity Matrix from Sigma; Anti-HA Affinity Matrix from Roche) overnight at 4 °C. Precipitated proteins were analyzed by Western blotting using anti-GFP, anti-HA, and anti-FLAG antibodies. Beads were washed five times with washing buffer (50 mM Tris-HCl at pH 6.8, 150 mM NaCl, and 1% NP40). Proteins bound to the beads were recovered in the SDS-PAGE sample buffer. Procedures from medium collection were carried out at 4°C or on ice. Proteins samples were resolved by SDS-PAGE and transferred to PVDF membranes (Millipore) for Western blot analysis.

S2 cell-based BMP signaling assay
The S2 cell-based BMP signaling assay was adopted from assays as previously described (Shimmi and O'connor, 2003). Recombinant Dpp (159-DP-020, R&D Systems) was diluted in the culture medium according to the manufacturer's recommendations. To prepare BMP ligands secreting source cells, Briefly, Drosophila S2 cells were transfected with plasmids to express Dpp-FLAG or/and Gbb-FLAG with or without Nord-HA-GFP. >72 hours after transfection, the conditioned medium was collected, and the cells were lysed in lysis buffer (50 mM Tris-HCl at pH 6.8, 150 mM NaCl, 1% NP40, and protease inhibitors). S2 cells stably expressing the FLAG-Mad transgene (Mad-S2) cells were generated and used as BMP responding cells.
Alternatively, we also transiently transfected S2 cells or Mad-S2 cells with plasmids to express GFP or Nord-GFP as described in the figure legends. >48 hours after transfection, the responding cells were incubated with the conditioned medium collected from the source cells for 1 hour. After incubation, the responding cells were washed and then lysed in the lysis buffer.
Both the conditioned medium and the lysates were clarified by centrifugation, and proteins were recovered directly in the SDS-PAGE sample buffer. Proteins were separated by SDS-PAGE under reducing conditions and then transferred onto PVDF membranes (Millipore) for Western blot analysis.

Wing size and wing Trichome measurements
Adult wings were dissected from animals in 100% ethanol and mounted in 1:1 Wintergreen oil-Canadian balsam medium. The wings were imaged using a 4X objective and area was measured using Fiji (ImageJ). Measurements were taken from the end of the costa on the anterior portion of the wing hinge to the end of the alula on the posterior. Area in pixels squared was converted to millimeters squared with a calibration value determined using a hemocytometer under the 4X objective. Wing trichomes from the dorsal wing surface were imaged with a 40X objective in the region between veins 4 and 5 just distal to the PCV (Supplementary file 4). Trichomes were counted manually within the imaged area (37500 m 2 ).

Statistical analysis
All data in column graphs are shown as mean values with SD and plotted using GraphPad Prism software. As described in the figure legends, one-way ANOVA followed by Sidak's multiple comparison test, unpaired two-tailed t-test, or two-sided Fisher's exact test was used for statistical analysis. The sample sizes were set based on the variability of each assay and are listed in the Figure legends. Independent experiments were performed as indicated to guarantee reproducibility of findings. Differences were considered statistically significant when P < 0.01.

CONFLICT OF INTEREST STATEMENT
The authors declare that they have no conflicts of interest with the contents of this article.    between distal ends of longitudinal veins L2 and L5 (dL2-L5), L3 and L4 (dL3-L4) and the ratio of dL3-L4 /dL2-L5. Each bar shows the mean ± SD. from n=20 wings. All flies were grown at 25°C.
One-way ANOVA followed by Sidak's multiple comparison test or unpaired two-tailed t-test were used for statistical analysis. ***P < 0.001, ****P < 0.0001, ns: not significant. au: arbitrary units.   Unpaired two-tailed t-test were used for statistical analysis. ****P < 0.0001, ns: not significant. genotypes. The animals were grown at 18°C till prepupa stage, and then transferred to 29°C around 78, 72 or 66 (± 1) hours before eclosion. n>30 wings for each genotype at a given temperature shift time point. (D) Representative pupal wing from larvae that carry the indicated transgenes under the control of tub-Gal80 ts together with en-Gal4. The animals were grown at 18°C till prepupa stage, and then transferred to 29°C at 12 hours before dissection. The collected pupal wings were immunostained for anti-pMad (red), anti-Ptc (blue) and anti-GFP (Nord-GFP, green). Yellow arrowhead indicates ectopic pMad around the primordial PCV; yellow arrow indicates reduced pMad signal around the primordial PCV. Scale bar, 100 μm.      shows the mean ± SD. from n>14 adults wings, and representative images were shown in panel A. All flies were grown at 25°C. One-way ANOVA followed by Sidak's multiple comparison test was used for statistical analysis. ****P < 0.0001, ns: not significant.  Each bar shows the mean ± SD. from n>10 adults wings. All flies were grown at 25°C. The unpaired two-tailed t-test was used for statistical analysis. **P < 0.01, ****P < 0.0001, ns: not significant.  were collected at indicated time points after pupation (AP), and were immunostained for GFP (Nord, green) and pMad (red). Note that Nord is expressed in the early pupal wing during the pMad refinement at the PCV primordia. The yellow arrows indicate the PCV primordia. Scale bar, 100 μm.    temperatures. Each box shows the mean ± SD. from n=30 adults wings, and representative images were shown in Figure 7C. Two-tailed unpaired t-test was used for statistical analysis. ****P < 0.0001.

Supplementary file 1. Genes that are differentially expressed in A, B and P cells
Genes (probe sets) whose expression is not only higher in A cells than P cells (FoldA/P > 1.2), but also higher in the A/P boundary adjacent B cells than general A cells (FoldB/P > 1.5) were selected as potential Hh-induced target genes. A total of 61 probe sets (59 unique genes) were identified as potential Hh signaling target genes. Wing trichomes from the dorsal wing surface were imaged with a 40X objective in the region between veins 4 and 5 just distal to the PCV. Trichomes were counted manually within the imaged area (37500 m 2 ). Scale bar, 50 μm.