The Broad Transcription Factor Links Hormonal Signaling, Gene Expression, and Cellular Morphogenesis Events During Drosophila Imaginal Disc Development

Imaginal disc morphogenesis during metamorphosis in Drosophila provides an ideal system for studying the hormonal control of morphogenesis. During metamorphosis, ecdysone signaling initiates a gene regulatory network.....

Abstract

Imaginal disc morphogenesis during metamorphosis in Drosophila melanogaster provides an excellent model to uncover molecular mechanisms by which hormonal signals effect physical changes during development. The broad (br) Z2 isoform encodes a transcription factor required for disc morphogenesis in response to 20-hydroxyecdysone, yet how it accomplishes this remains largely unknown. Here, we use functional studies of amorphic br⁵ mutants and a transcriptional target approach to identify processes driven by br and its regulatory targets in leg imaginal discs. br⁵ mutants fail to properly remodel their basal extracellular matrix (ECM) between 4 and 7 hr after puparium formation. Additionally, br⁵ mutant discs do not undergo the cell shape changes necessary for leg elongation and fail to elongate normally when exposed to the protease trypsin. RNA-sequencing of wild-type and br⁵ mutant leg discs identified 717 genes differentially regulated by br, including a large number of genes involved in glycolysis, and genes that encode proteins that interact with the ECM. RNA interference-based functional studies reveal that several of these genes are required for adult leg formation, particularly those involved in remodeling the ECM. Additionally, br Z2 expression is abruptly shut down at the onset of metamorphosis, and expressing it beyond this time results in failure of leg development during the late prepupal and pupal stages. Taken together, our results suggest that br Z2 is required to drive ECM remodeling, change cell shape, and maintain metabolic activity through the midprepupal stage, but must be switched off to allow expression of pupation genes.

TISSUE morphogenesis is required for the elaboration of the body axis and organs during metazoan development. In some contexts, hormonal signals provide temporal cues to coordinate these morphogenetic events. Imaginal disc morphogenesis in the fruit fly Drosophila melanogaster provides an excellent model to uncover molecular mechanisms by which hormonal signals are translated into the physical changes that occur during development. Imaginal discs are diploid tissues found within the larva that give rise to the adult integument during metamorphosis (Willis 1974; Csaba 1977; Fristrom 1988). Both classical experiments and more recent live imaging studies (De las Heras et al. 2018; Diaz-de-la-Loza et al. 2018) have revealed requirements for cell shape changes and rearrangements, as well as for remodeling of the extracellular matrix (ECM) in imaginal disc morphogenesis during metamorphosis. These experiments also demonstrate the key role that 20-hydroxyecdysone (hereafter referred to as ecdysone) plays in directing these processes. Ecdysone acts through a transcriptional cascade: the hormone binds to its heterodimeric receptor, which acts as a DNA-binding activator to drive transcription of early-response genes, including Eip74EF, Eip75B, and broad (br), which themselves encode DNA-binding proteins that activate late-response genes (Chao and Guild 1986; Feigl et al. 1989; Janknecht et al. 1989; Burtis et al. 1990; Segraves and Hogness 1990; DiBello et al. 1991; Yao et al. 1993; Crossgrove et al. 1996). Despite extensive research into Drosophila metamorphosis, parts of the pathway connecting the ecdysone cue to the effectors involved in imaginal disc morphogenesis have yet to be elucidated.

Of the ecdysone signaling early-response genes, br appears to have the most direct effects on imaginal disc morphogenesis (Kiss et al. 1988). br encodes four transcription factors, each carrying a unique zinc finger domain spliced to a common Broad-Complex, Tramtrack, and Bric a Brac/Pox virus and Zinc finger (BTB/POZ) domain (DiBello et al. 1991; von Kalm et al. 1994; Bayer et al. 1996) (Supplemental Material, Figure S1). These four isoforms, known as the Z1, Z2, Z3, and Z4 isoforms, have three genetically separate functions (br, reduced bristles on the palpus, and 2Bc), with the Z2 isoform performing the classic br function (Bayer et al. 1997). This function is critical for imaginal disc morphogenesis: animals lacking functional Br Z2 have discs that fail to elongate, and these animals die during the prepupal stage before head eversion (Kiss et al. 1988). The various isoforms of Br also add a layer of complexity to the regulation of metamorphic genes. br does not simply respond to the ecdysone cue through a general increase in transcription that subsequently effects the increased transcription of late-response genes; rather, the isoforms exhibit a dynamic pattern of expression that differs by tissue (Huet et al. 1993). In the imaginal discs, the expression of the Z2 transcript rises dramatically approximately 4 hr before pupariation, but greatly decreases in the hours after pupariation, while the expression of the Z1 transcript greatly increases between 2 and 4 hr after pupariation (Emery et al. 1994; Bayer et al. 1996). This late larval pulse of Z2 expression is critical for activation of late-response genes; however, the identities of these genes remain largely unknown. The illumination of the link between the ecdysone cue and morphogenetic effects in imaginal discs hinges upon the identification of these late-response genes and how they are regulated by ecdysone and br.

Previous screens using the hypomorphic br¹ allele identified a number of br-interacting genes, including Stubble (Sb), zipper, Rho1, Tropomyosin 1, blistered, and ImpE3 (Beaton et al. 1988; Ward et al. 2003), although the majority of these genes are not transcriptional targets of br (Ward et al. 2003). Nevertheless, the identity of these genes suggests roles for br in the major processes implicated in hormonal control of imaginal disc morphogenesis, namely cell shape changes/rearrangements and modification of the ECM. Here, we show that cell shape changes and cell rearrangements fail to occur normally in br⁵ mutant prepupae. In addition, consistent with previous work demonstrating that the ECM provides a constraining force and must be degraded to allow disc elongation (Pastor-Pareja and Xu 2011; Diaz-de-la-Loza et al. 2018), we show that the basal ECM protein Collagen IV is not substantially degraded in leg imaginal discs from amorphic br⁵ mutant animals as old as 8 hr after puparium formation (APF). In this study, we also use this allele to identify specific genes regulated by br in the leg discs at the onset of metamorphosis through an RNA-sequencing-based approach. This approach identified over 700 br-regulated genes, including genes with known metabolic and developmental functions, including ECM organization and modification. We tested a subset of these genes for roles in leg morphogenesis through RNA interference (RNAi) and found several that are necessary for proper development of the adult legs. These results demonstrate the value of a transcriptional target-based approach to identifying morphogenetic genes, and suggest that br regulates morphogenesis through the regulation of genes involved in multiple critical processes.

Materials and Methods

Fly stocks

All Drosophila stocks were maintained on media consisting of corn meal, sugar, yeast, and agar in incubators maintained at a constant temperature of 25° or at room temperature. w¹¹¹⁸, ybr⁵, br¹, distalless (dll)-Gal4, apterous (ap)-Gal4, P{UAS-Dcr-2.D}1, w¹¹¹⁸, Pgm1^LA00593, and the Transgenic RNAi Project lines (“short-hairpin” RNAi lines; https://fgr.hms.harvard.edu) were obtained from the Bloomington Drosophila Stock Center (Bloomington, IN). “Long-hairpin” RNAi lines were obtained from the Vienna Drosophila RNAi Center (Vienna, Austria; Dietzl et al. 2007). UAS-serp was obtained from Stefan Luschnig (University of Muster). UAS-Verm/CyO was obtained from Christos Samakovlis (Stockholm University). vkg-GFP (Flytrap; Buszczak et al. 2007) was obtained from Sally Horne-Badovinac (University of Chicago). w; hs-Z2 (CD5-4C); hs-Z2 (CD5-1) was obtained from Cindy Bayer (University of Central Florida). Loxl-1-RNAi (stock 11335R) was obtained from the National Institutes of Genetics Fly stock collection (Kyoto, Japan). dll-Gal4 was balanced with CyO, P{w⁺, Dfd-EYFP} (Le et al. 2006). w¹¹¹⁸ was used as the wild-type control, unless otherwise noted. All fly stocks and reagents are listed and described in the reagents table.

br isoforms

To clarify the relationship between the isoforms of the br gene listed on FlyBase and the Br protein isoforms bearing the Z1, Z2, Z3, and Z4 zinc finger domains, translated sequences of all 15 br isoforms listed on FlyBase were downloaded from FlyBase and aligned using Clustal Omega (Goujon et al. 2010; Sievers et al. 2011; Thurmond et al. 2019). The first 431 amino acids represent the Br “core” region and were shared among all isoforms; the remaining amino acids were compared to the published Z1, Z2, Z3, and Z4 zinc finger domain sequences to determine which zinc finger was carried by each FlyBase isoform (DiBello et al. 1991; Bayer et al. 1996).

Fly staging, dissection, and photography of live leg imaginal discs

w¹¹¹⁸ and ybr⁵/Binsn flies were staged on food supplemented with 0.05% bromophenol blue, as described in Andres and Thummel (1994). w¹¹¹⁸ and ybr⁵/Y mutant animals were selected (mutant males were selected using the yellow marker) at −18 hr (blue gut larvae), −4 hr (white gut larvae), 0 hr (white prepupae), and +2 hr and +4 hr relative to puparium formation. Leg imaginal discs were dissected in phosphate-buffered saline (PBS), and then transferred to fresh PBS. Brightfield photomicrographs were captured within 5 min on a Nikon Eclipse 80i microscope equipped with a Photometrics CoolSNAP ES high performance digital CCD camera using a Plan APO ×10 (0.45 NA) objective. For the examination of collagen integrity in prepupal leg discs, we crossed ybr⁵/Binsn virgins to w¹¹¹⁸, vkg-GFP males, and crossed F₁ ybr⁵/w¹¹¹⁸; vkg-GFP/+ females to w¹¹¹⁸ males. We selected ybr⁵/Y; vkg-GFP/+ and w¹¹¹⁸/Y; vkg-GFP/+ white prepupae based upon the y phenotype and aged them at 25°. ybr⁵ prepupae were confirmed to contain the br⁵ allele based upon pupal morphology. We dissected leg imaginal discs at the indicate time points in PBS, mounted them live in mounting media (90% glycerol, 100 mM Tris pH 8.0, 0.5% n-propyl-gallate) and imaged them sequentially with brightfield microscopy and wide-field fluorescence microscopy on a Nikon Eclipse 80i microscope equipped with a Photometrics CoolSNAP ES high performance digital CCD camera, using a Plan APO ×20 (0.75 NA) objective. All images were captured with identical settings (light or fluorescence intensities, and exposure times). On average, three or four discs were imaged from each prepupa, and at least seven different prepupae were imaged at each time point. All digital images were cropped and adjusted for brightness and contrast in Adobe Photoshop (version CC 2018; Adobe, San Jose, CA) or ImageJ (version 1.51r; National Institutes of Health, Bethesda, MD), and figures were compiled using Adobe Illustrator (version CC 2018).

Trypsin experiments

w¹¹¹⁸ and br⁵ white prepupae were dissected in PBS to isolate three leg imaginal discs from the same animal. These imaginal discs were incubated on 3-well depression slides in either PBS, PBS plus 0.025% trypsin (25200-056; Gibco) or PBS plus 0.0025% trypsin for 15 min at room temperature, at which point they were immediately imaged by brightfield microscopy on a Nikon Eclipse 80i microscope equipped with a Photometrics CoolSNAP ES high performance digital CCD camera using a Plan APO ×10 (0.45 NA) objective. In total, 21 w¹¹¹⁸ and 17 br⁵ animals were dissected and imaged.

Immunostaining

Imaginal discs were hand dissected from w¹¹¹⁸ or br⁵ +4 hr prepupae in fresh PBS, and fixed immediately in 4% paraformaldehyde for 20 min. The following antibodies were used at the given dilutions: rat anti-DE-cadherin (clone DCAD2 from Developmental Studies Hybridoma Bank at the University of Iowa) at 1:25, rabbit anti-Vermiform (gift from Stefan Luschnig, University of Munster) at 1:1000 (Luschnig et al. 2006), Donkey anti-rabbit Cy3 (Jackson ImmunoResearch Laboratories, West Grove, PA) at 1:400 and donkey anti-rat Cy2 (Jackson ImmunoResearch Laboratories) at 1:400. Confocal images were acquired on a Leica SPE laser scanning confocal microscope using an ACS APO ×40 (1.15 NA) oil immersion lens. All digital images were cropped and adjusted for brightness and contrast in ImageJ (version 1.51s; National Institutes of Health). Figures were compiled using Adobe Illustrator (version CC 2018).

RNA-sequencing and data analysis

To control for potential genetic effects from the autosomes and Y chromosome, we crossed ybr⁵/Binsn females to w¹¹¹⁸ males and then crossed the resulting ybr⁵/w¹¹¹⁸ females with w¹¹¹⁸ males. This cross produced ybr⁵/Y and w¹¹¹⁸/Y males that, at a population level, had identical autosomes and Y chromosomes. Since we selected the br⁵ animals based upon the yellow cuticular phenotypes, we tested to make sure that y and br did not recombine apart to any significant degree. y and br are reported to map 0.2 cM apart on the X chromosome (Gatti and Baker 1989), and in two separate experiments we did not detect any recombination between these genes (n > 200). After RNA-sequencing, we identified the br⁵ mutation as a C to T transition at position 1654571 in GenBank AE014298 (X chromosome of D. melanogaster), converting a conserved histidine in the zinc finger of the Z2 isoform into a tyrosine. We found that all the reads through this interval had the mutation in the br⁵ samples, whereas none of the reads from w¹¹¹⁸ samples had the mutation.

Third instar ybr⁵/Y and w¹¹¹⁸/Y larvae were staged on food supplemented with 0.05% bromophenol blue. Blue gut larvae (−18 hr) and white prepupae (0 hr) were selected and leg imaginal discs were hand-dissected in PBS. Triplicate independent samples were obtained for ybr⁵/Y and w¹¹¹⁸/Y white prepupae and for w¹¹¹⁸/Y −18 hr larvae. Total RNA was isolated using TriPure (Roche, Indianapolis, IN) and then purified over RNAeasy columns (Qiagen, Valencia, CA). Approximately 5 μg of total purified RNA was obtained for each sample, and ∼1 μg of each sample was provided to the Genome Sequencing Core at the University of Kansas for library preparation using the TruSeq stranded mRNA kit (Illumina, San Diego, CA). Single read 100 (SR100) was performed on a single lane of an Illumina HiSeq 2500 (Genome Sequencing Core, University of Kansas).

The quality of the raw RNA-sequencing reads was visually confirmed using FastQC (version 0.11.5; Andrews 2010), and both low-quality data and adaptor sequences were removed using Trimmomatic (version 0.36; Bolger et al. 2014). The number of remaining reads per sample averaged 18.1 million (range = 12.6–22.8), and all reads were at least 50-nt in length. Filtered reads were mapped to the D. melanogaster reference genome (Release 5.3) using TopHat (version 2.1.1; Trapnell et al. 2009; Kim et al. 2013). Default parameters were employed, with the addition of the “–no-novel-juncs” flag to use the gene annotation as provided, and the “–library-type fr-firststrand” flag since the RNA-sequencing data derives from the Illumina TruSeq stranded mRNA kit. On average 86.2% of the reads mapped to the reference across samples (range = 83.1–88.5). Genotypes were compared, and differentially-expressed genes were identified using the Cufflinks pipeline (version 2.2.1; Trapnell et al. 2010; Roberts et al. 2011; Trapnell et al. 2013). Specifically, we employed the “cuffquant” and “cuffdiff” routines using default parameters, adding the “-b” flag to run a bias correction algorithm that improves expression estimates (Roberts et al. 2011).

Bioinformatic analyses

Gene ontology analysis was performed using the Gene Ontology Consortium’s (http://www.geneontology.org) gene ontology enrichment analysis tool (Gene Ontology Consortium 2015). To ensure we were considering genes showing large-scale induction or repression, significantly differentially expressed genes showing ≥1.5-fold change between w¹¹¹⁸ 0 hr and br⁵ 0 hr samples and at least 5 fragments per kilobase of exon per million fragments mapped (FPKM) in the sample showing higher expression (w¹¹¹⁸ for br-induced genes, br⁵ for br-repressed genes) were used as input into the tool. In a second experiment, significantly differentially expressed genes showing ≥1.5-fold change between w¹¹¹⁸ –18 and 0 hr samples and at least 5 FPKM in the sample showing higher expression (w¹¹¹⁸ 0 hr for developmentally induced genes and w¹¹¹⁸ –18 hr for developmentally repressed genes) were used. The same data sets were used for the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis using WebGestalt’s (http://www.webgestalt.org) overrepresentation analysis (Wang et al. 2017).

ATP measurement

Leg and wing imaginal discs were dissected from +4 hr w¹¹¹⁸ and br⁵ mutant prepupae in PBS and immediately homogenized in 100 μl of lysis buffer (6 M guanidine HCl, 100 mM Tris pH 7.8, 4 mM EDTA) on ice and frozen. A mixed leg and wing sample was used in this experiment to allow all of the dissections to be completed in one sitting with enough material for the subsequent analysis. We note that prior Northern blot analysis using a mixed sample of leg and wing imaginal disc material showed identical expression of ecdysone and Rho signaling pathway gene expression to those composed only of leg imaginal disc material, and that wing discs arrest development in br⁵ prepupa at the same stage that leg discs do (R. Ward, unpublished data). After all of the samples were collected, they were thawed to 4°, and a 10 μl aliquot was taken for protein measurement using a Bradford Assay (Bio-Rad, Hercules, CA). The remaining sample was boiled for 5 min, spun at 13,000 rpm in a refrigerated microcentrifuge for 3 min, and then 10 μl of the supernatant was diluted into 90 μl of dilution buffer (25 mM Tris pH 7.8, 100 μM EDTA). This sample was further diluted 10-fold in dilution buffer and 10 μl of this sample was used for ATP quantification using the ATP Determination Kit (A22066; Molecular Probes, Eugene, OR) according to the manufacturer’s protocol, using a BioTek Synergy HT plate reader. Each biological sample contained ∼13 wing discs and 35 leg discs. Triplicate samples were processed and three technical replicates of each biological sample was assayed, along with a dilution series of ATP from 1 nM to 1 μM. Statistical analysis was performed using a type 3 ANOVA with protein level and treatment as factors.

Northern blot analysis

Progeny from a cross of ybr⁵/Binsn X Binsn/Y were staged on standard Drosophila media supplemented with 0.05% bromophenol blue, as described in Andres and Thummel (1994). Total RNA was isolated by direct phenol extraction from leg imaginal discs dissected from staged Binsn/Y males. Approximately 9 μg of total RNA per sample were separated by formaldehyde agarose gel electrophoresis and transferred to a nylon membrane. The membrane was hybridized and stripped as described in Karim and Thummel (1991). Generation of probe fragments for br (br core and br-Z2) and rp49 is described in Andres and Thummel (1994). Specific probes were labeled by random priming of gel-purified fragments (Stratagene, St. Louis, MO).

Functional analyses

To amplify potential RNAi in long-hairpin lines, we crossed a UAS-Dicer transgene onto the Gal4 lines used to drive the expression of the gene-specific double-stranded RNA to produce P{UAS-Dcr-2.D}1, w¹¹¹⁸; Dll-GAL4/CyO, dfd-YFP and P{UAS-Dcr-2.D}1, w¹¹¹⁸; ap-GAL4/CyO, dfd-YFP. For consistency, we also used these recombined lines when crossing to short-hairpin RNAi lines. Virgin females of these genotypes were then mated to UAS-RNAi males. The vials were kept in incubators maintained at a constant temperature of 25°. The adults were transferred twice to new vials, and newly eclosing F₁ flies were separated by phenotype and examined for malformed third legs each day for a total of 8 days per vial. We considered an animal to be malformed if it displayed malformation in at least one leg, and defined a leg as malformed if any femur, tibia, or tarsal segment was bent, twisted, missing, or was excessively short and fat.

To control for background effects that may contribute to the appearance of malformed legs, we crossed UAS-Dcr1, w¹¹¹⁸; Dll-GAL4/CyO, dfd-YFP virgin females with males from the w¹¹¹⁸ line and a loxl1-RNAi line and examined the offspring for malformed legs. Curly-winged offspring from RNAi crosses, which carry the UAS-RNAi hairpin construct, but not the GAL4 driver, were also examined. A few of the RNAi constructs were carried over a balancer chromosome; only flies without the balancer were counted. Flies carrying malformed legs did not exceed 1% of all flies in either control cross, and exceeded 1% among curly-winged offspring from only two RNAi crosses. In neither of these two crosses (carrying RNAi hairpin constructs against CG7447 and E[spl]m4) did flies carrying malformed legs exceed 2% of all curly-winged flies, and neither of these crosses were among those in which a significant number of straight-winged flies carried malformed legs. For crosses in which we examined preadult stages expressing RNAi, individuals carrying the Dll-GAL4 driver were identified by the lack of dfd-YFP expression.

br-Z2 overexpression studies

Approximately 50 HS-br-Z2 or w¹¹¹⁸ late larvae were placed into an empty fly vial with a piece of moist Whatman paper in the bottom of the vial. The vial was heat shocked in a 37° incubator for 60 min, after which any animals that had pupariated were removed. The vial was moved to a 25° incubator for 6 additional hours, at which point all prepupae were removed to a food vial for further development (also at 25°). After a further 18 hr, the prepupae were scored to determine if they had pupated. The animals were then left at 25° and scored for eclosing 4 days later. In some experiments, HS-br-Z2 or w¹¹¹⁸ were collected as they pupariated, aged 4 hr at 25°, and dissected to examine their leg imaginal discs. Another subset was dissected at approximately 16 hr after pupariating to examine the terminal leg imaginal disc phenotypes. All dissections, microscopy, and image preparation were conducted as described above.

Adult specimen preparations

Adult legs were dissected from the third thoracic segment in PBS, cleared in 10% KOH overnight, and mounted in Euparal (Bioquip, Gardena, CA) on microscope slides. Images of adult leg cuticles were captured on a Photometrics CoolSNAP ES high performance digital CCD camera with a Nikon Eclipse 80i microscope. All digital images were cropped and adjusted for brightness and contrast in ImageJ (version 1.51r; National Institutes of Health) and figures were compiled using Adobe Illustrator (version CC 2018).

Data availability

Fly stocks are available upon request. Figure S1shows the gene structure and RNA isoforms of the broad locus. Figure S2 shows the results of luciferase ATP assays to quantify ATP levels between w¹¹¹⁸ and br⁵ +4 hr leg discs. Figure S3 compares +4 hr leg imaginal discs from w¹¹¹⁸, br⁵, and Dll > br-RNAi-expressing animals. Table S1 shows the genes that are differentially expressed between w¹¹¹⁸ and br⁵ 0 hr leg discs. Table S2 shows the gene ontology terms that are significantly over- and underrepresented in the br-induced and br-repressed gene sets. Table S3 shows the KEGG pathway analyses on the br-induced and br-repressed gene sets. Table S4 shows the genes that are differentially expressed between −18 and 0 hr in w¹¹¹⁸ leg discs. Table S5 shows the gene ontology terms that are significantly over- and underrepresented in the w¹¹¹⁸ developmentally regulated gene sets. The reagents table lists all the stocks and reagents used in this study. All RNA-sequencing data sets are available from the Gene Expression Omnibus. The project accession number is GSE140248, and the project title is “Genome-wide differences in RNA expression in Drosophila melanogaster leg imaginal discs based on time and presence/absence of broad-based gene regulation.” Supplemental material available at figshare: https://doi.org/10.25386/genetics.11729019.

Results

Phenotypic characterization of br⁵ mutant leg imaginal discs

To identify genes regulated by br during metamorphosis using an RNA-sequencing approach, we wanted to use a null br allele. The br⁵ allele has been reported to be amorphic, with homozygous mutants failing to develop past the early prepupal stage (Kiss et al. 1988). Consistent with this notion, br⁵ encodes a protein with a His⁴⁹²–Tyr substitution in the conserved Z2 zinc finger domain, suggesting that it has defective DNA binding properties (L. von Kalm, personal communication; confirmed in the RNA-sequencing; see Materials and Methods). We previously used live imaging to show that br⁵ hemizygous mutants failed to complete metamorphosis (Ward et al. 2003). To more closely examine how imaginal disc development is disrupted in these animals, we dissected and compared leg discs from br⁵ and w¹¹¹⁸ larvae and prepupae (Figure 1). br⁵ leg discs resemble w¹¹¹⁸ leg discs during larval stages (through 4 hr before pupariation), consisting of a columnar epithelium with folds that make three to four concentric circles. The discs bulge from the central circles (which develop into the distal-most leg segments), and are covered by the peripodial epithelium. By the prepupal stage, leg discs begin to differ between the two genotypes. Both br⁵ and w¹¹¹⁸ discs begin to elongate in a telescoping fashion at the onset of metamorphosis (0 hr), beginning with the centermost regions. However, by 2 hr after pupariation, elongation of br⁵ discs clearly lags behind that of w¹¹¹⁸ discs. w¹¹¹⁸ discs continue to elongate and by 4 hr after pupariation, the future five tarsal segments and distal tibia are clearly identifiable underneath the peripodial epithelium. br⁵ mutant discs, on the other hand, form shorter, wider structures with deeper folds between the tarsal segments. br⁵ mutant discs show only limited elongation over subsequent hours and fail to evert to the outside of the prepupa. This phenotype is completely penetrant (n > 1000 leg discs examined). Notably, imaginal discs can be found inside degenerating br⁵ mutant late prepupae that look similar to +4 hr mutant discs, including having an intact peripodial epithelium (n > 20 animals; Figure 1K).

Figure 1.

Broad is required for normal prepupal leg imaginal disc development. Brightfield photomicrographs of leg imaginal discs from w¹¹¹⁸ (A–E) and br⁵ (F–K) mutant larvae and prepupae. Times given are relative to puparium formation. Note that leg imaginal discs are similar between w¹¹¹⁸ and br⁵ during larval time points (−18 and −4 hr), but are noticeably different starting at 0 hr. Although w¹¹¹⁸ discs elongate from the center (distal tarsal segment; indicated by asterisk) and are segmented by +4 hr, br⁵ discs show limited elongation and have wider tarsal segments. Intact leg imaginal discs can be found inside dead br⁵ prepupae as late as 17 hr after pupariation. These late imaginal discs have not elongated much past the +4 stage and often have an intact peripodial epithelium (arrow). Bar, 100 μm.

The disruption of prepupal elongation in br⁵ mutants, coupled with the presence of the peripodial epithelium in late-staged prepupae (which is known to be degraded by matrix metalloproteinases; Proag et al. 2019), motivated us to examine ECM breakdown in wild-type and br⁵ mutant discs. We therefore crossed a GFP-tagged version of Collagen IV (Vkg-GFP) into w¹¹¹⁸ and br⁵ and examined their leg discs at various time points during metamorphosis (Figure 2). At early stages (0 and 2 hr after pupariation), wild-type and br⁵ leg discs closely resemble one another, with fibrous cylindrical or nearly conical basal ECM structures. By 4 hr after pupariation, some wild-type discs exhibit degradation of the basal ECM. This degradation takes the form of a “clearing” of a central channel along the length of the disc. In total, 37.5% of wild-type discs (n = 32, from 12 animals) showed this clearing at 4 hr after pupariation, 66.7% (n = 21, from 12 animals) at 6 hr, and 100% (n = 12, from 8 animals) at 7 hr (Figure 2K). In contrast, most br⁵ mutant discs retained the fibrous ECM appearance as late as 8 hr after pupariation (Figure 2, J and K). Only 12.0% of discs (n = 25, from 7 animals) showed partial clearing; the other discs showed no clearing.

Figure 2.

broad is required for efficient degradation of the basal ECM in prepupal leg imaginal discs. Wide-field fluorescence (A–J) and brightfield (A’–J’) photomicrographs (at the same focal plane) of leg imaginal discs from w¹¹¹⁸ (A–E) and br⁵ mutant (F–J) prepupae expressing Viking-GFP (Collagen IV). Ages of the animals relative to puparium formation are indicated above the figures. Note that Viking-GFP forms cables of ECM lining the lumen at the basal side of elongating leg discs, which are gradually cleared from +4 to +7 hr APF in w¹¹¹⁸ leg discs (white arrows in C, D, E, and J; black arrows in C’, D’, E’, and J’ represent matching location in brightfield images). Similar clearing is not observed in the br⁵ mutant discs and persists at least through 8 hr APF (J) in many animals. Bar, 100 μm. (K) Graph showing percentage of discs showing partial or total clearing of GFP expression in the channel.

Since br⁵ mutant leg imaginal discs show reduced elongation and defective proteolysis of the ECM, we wondered whether an exogenous protease could restore normal elongation to the br⁵ discs. We therefore dissected three leg imaginal discs from individual w¹¹¹⁸ or br⁵ animals at the onset of metamorphosis and treated one with 0% trypsin (PBS control), another with 0.0025% trypsin, and the third with 0.025% trypsin for 15 min at room temperature. Leg imaginal discs from 18 of the 21 w¹¹¹⁸ animals had clearly elongated after 15 min in 0.0025% trypsin (Figure 3B). The tarsal segments in these discs were still obviously segmented, although in some cases the depth of the folds at these segment boundaries was reduced relative to the PBS control discs. w¹¹¹⁸ leg discs incubated for 15 min in 0.025% trypsin showed even greater elongation (in 100% of the discs), but only with noticeable tarsal segment boundaries in two of the discs, whereas the remainder had elongated discs with the tarsal segments appearing as one long continuous segment (Figure 3C). In the br⁵ mutant leg discs, there was a clear difference in the morphology of the leg discs at the onset of the experiment (which remained unchanged through 15 min in PBS; data not shown) compared to the w¹¹¹⁸ discs, in which the distal tarsal segment was rounder and the folds between the tarsal segments were noticeably deeper (Figure 3D vs. Figure 3A). A total of 10 of the 17 discs treated with 0.0025% trypsin showed some elongation of the tarsal segments, but no shallowing of the segment boundaries (Figure 3, E and G). Similarly, 12 of the 17 discs treated with 0.025% trypsin showed an increase in the degree of elongation (Figure 3G). The morphology of these discs was clearly distinct from w¹¹¹⁸ treated similarly. The absolute level of elongation in br⁵ was less than in w¹¹¹⁸, and more interestingly, the br⁵ discs did not eliminate the folds between segments, but rather deepened those folds (Figure 3F). This morphology was observed in 14 of the 17 discs examined (Figure 3G).

Figure 3.

br⁵ leg discs show aberrant elongation upon application of trypsin, and retain anisometric cell shapes through 4 hr after pupariation. (A–F) Brightfield photomicrographs of leg imaginal discs from a representative w¹¹¹⁸ (A–C) and br⁵ (D–F) mutant 0 hr prepupa incubated for 15 min in PBS (A and D) or 0.0025% trypsin (B and E) or 0.025% trypsin (C and F). All three discs are from the same animal. The w¹¹¹⁸ discs elongate in the lower dose of trypsin, and elongate more in the higher dose of trypsin to the point where the segmental folds are pulled smooth. In contrast, br⁵ mutant discs show less elongation in trypsin overall, and deeper folds in the presumptive tarsal segments (arrows). (G) Quantification of the elongation defects and aberrantly folded leg discs from w¹¹¹⁸ and br⁵ mutant prepupae. (H and I) Confocal optical sections of tarsal segments from leg imaginal discs from w¹¹¹⁸ (H) and br⁵ (I) mutant +4 hr leg imaginal discs stained with antibodies against DE-Cadherin (distal tarsal segments are to the right in each image). Note that in the w¹¹¹⁸ discs, epithelial cells in the tarsal segments are mostly isometric, whereas cells in the br⁵ tarsal segments are anisometric with longer circumferential cell lengths (arrows). Bar, 100 μm.

Given that trypsin was not able to elongate br⁵ leg discs to the extent of that observed in w¹¹¹⁸ discs, and that br⁵ discs display aberrant morphology as early as the onset of metamorphosis, we suspected that the underlying cell shape changes and/or rearrangements that normally occur in wild-type animals do not occur in these mutant animals. To address this question, we examined cell shapes in the distal tarsal segments of +4 hr prepupal leg imaginal discs from w¹¹¹⁸ and br⁵ mutant animals using antibodies against DE-Cadherin. Although epidermal cells from w¹¹¹⁸ +4 hr leg discs are generally isometric, being equally long in the proximal-distal and circumferential dimensions (Figure 3H), epidermal cells from br⁵ +4 hr leg discs show considerable anisometry, with longer cell dimensions in the circumferential axis than the proximal-distal axis (Figure 3I). This phenotype is completely penetrant (based upon five experiments with leg discs from >30 mutant animals). The br⁵ mutant distal tarsal segments are also wider at +4 hr than w¹¹¹⁸ leg discs (Figure 1E vs. Figure 1J), suggesting a defect in cell rearrangements that are known to occur in wild type (Condic et al. 1991), although the defect in segmentation in the br⁵ discs make this difficult to quantify.

Identification and bioinformatic characterization of genes regulated by br in prepupal leg discs

To better understand the role of broad in regulating imaginal leg morphogenesis, we used RNA-sequencing to identify genes that are differentially expressed between w¹¹¹⁸ and br⁵. We dissected leg discs from these animals at the onset of metamorphosis (0 hr), the time we first see differences in the tissues between genotypes. We generated libraries from three biological replicates per genotype and sequenced them on an Illumina HiSeq 2500. We then mapped reads to the reference genome (release 5.3) with TopHat, and used the Cufflinks pipeline to identify a total of 717 genes that were significantly differentially expressed between the two genotypes (at a false discovery rate of 5%) (Figure 4) (Trapnell et al. 2009; Trapnell et al. 2010; Roberts et al. 2011; Kim et al. 2013; Trapnell et al. 2013). About two-thirds of these genes (66.9%) were more highly expressed in the br⁵ genotype. We refer to these genes as br-repressed genes (and genes that were more expressed in w¹¹¹⁸ as br-induced genes).

Figure 4.

Over 700 genes are significantly differentially expressed in the absence of functional br-Z2. (A) Volcano plot of genes differentially regulated in br⁵ mutant leg imaginal discs relative to w¹¹¹⁸ controls. Red denotes genes with large-scale induction or repression that were used in the gene ontology and KEGG analyses, while yellow denotes genes that were significantly differentially regulated but not included in the analyses (fold change <1.5 or FPKM <5). (B) Overlap of br-regulated genes with developmentally regulated genes. Most br-induced genes were developmentally upregulated (62.0% were higher in 0 hr w¹¹¹⁸ prepupae than in −18 hr w¹¹¹⁸ larvae; only 7.6% were significantly higher at −18 hr), while a comparatively smaller portion of the br-repressed genes were developmentally downregulated (23.3%, compared to 26.3% that were developmentally upregulated).

To investigate which functions might be regulated by br, we identified gene ontology terms that were significantly overrepresented in our data sets using the Gene Ontology Consortium’s gene ontology enrichment analysis tool (Gene Ontology Consortium 2015) (Table S1). To ensure we were considering genes showing large-scale induction or repression, we used br-induced and br-repressed genes with fold changes ≥1.5 and FPKM ≥5 in at least one of the samples, leaving 155 br-induced genes and 356 br-repressed genes. Significantly overrepresented gene ontology terms for this set of genes included “Notch signaling,” “Molting cycle, chitin-based cuticle,” and “Developmental process”. Metabolic gene ontology terms for this set of genes included “glycolytic process,” “glycogen metabolic process,” and “carbohydrate biosynthetic process” (Table 1 and Table S2). KEGG pathway analysis of the same set of genes using WebGestalt (Wang et al. 2017) revealed significant overrepresentation of genes involved in glycolysis/gluconeogenesis, and several sugar metabolism pathways for br-induced genes (Table S3). Many br-induced genes have predicted cell signaling, developmental, or metabolic functions.

Table 1. Selected significantly enriched biological process GO terms for br-induced genes.

GO term	No. of genes	Expected	Fold change	Significance
Single-organism carbohydrate catabolic process	9	0.27	32.93	3.91E-08
ATP generation from ADP	8	0.23	35.37	3.14E-07
Glycolytic process	8	0.23	35.37	3.14E-07
ADP metabolic process	8	0.25	32.65	5.86E-07
Pyruvate metabolic process	8	0.33	24.25	5.88E-06
Notch signaling pathway	8	0.52	15.43	1.87E-04
Molting cycle, chitin-based cuticle	8	0.91	8.75	1.26E-02
Developmental process	55	31.49	1.75	1.39E-02
Anatomical structure development	53	30.32	1.75	2.30E-02
Organ morphogenesis	22	7.98	2.76	4.11E-02

Shown above are selected gene ontology (GO) terms representative of those found to be significantly overrepresented among br-induced genes. See Table S2 for the full list.

Since br serves as an early-response gene to the late larval ecdysone pulse (Chao and Guild 1986; DiBello et al. 1991), we expected that br-regulated genes would essentially represent a subset of genes regulated by ecdysone. To test this idea, we conducted a second set of RNA-sequencing analyses comparing genes expressed in mid-third instar (−18 hr) w¹¹¹⁸ leg imaginal discs to those expressed in w¹¹¹⁸ leg imaginal discs at the onset of metamorphosis (0 hr). A total of 1186 genes are significantly upregulated at the 0 hr stage relative to the −18 hr stage (developmentally induced genes), while 1193 genes are significantly downregulated (developmentally repressed genes) (Table S4). We anticipated that these developmentally regulated genes would serve as a proxy for ecdysone-regulated genes; consistent with this expectation, many known ecdysone-regulated genes are found in this data set, including Eip74EF, E23, and Cyp18a1 (Janknecht et al. 1989; Burtis et al. 1990; Hurban and Thummel 1993; Bassett et al. 1997; Hock et al. 2000). Significantly enriched gene ontology terms in the developmentally induced genes include numerous terms related to developmental processes, such as “imaginal disc eversion” and “regulation of tube size”, while terms relating to mitosis and DNA replication are significantly overrepresented in the developmentally repressed genes (Table S5). Of the 237 genes induced by br, 147 (62.0%) were upregulated at the 0 hr stage in w¹¹¹⁸ flies and only 18 (7.6%) were downregulated at the 0 hr stage (Figure 4B). In contrast with the predictions of the hierarchical model of ecdysone-driven gene regulation, more br-repressed genes were upregulated in 0 hr prepupae than downregulated, although the difference was not significant [126 genes and 112 genes (26.3% and 23.3%), respectively] (Figure 4B).

The absence of metabolic gene ontology terms among those enriched in the developmentally induced genes suggests that wild-type discs persist in their metabolic activity even as the larval tissues are downregulating metabolism (Merkey et al. 2011), and that this maintenance of metabolism is dependent upon br function. This notion is consistent with the observation that br mutant leg discs appear to arrest development at about 4 hr APF (Figure 1). To address this experimentally, we quantified ATP levels in +4 hr w¹¹¹⁸ and br⁵ mutant imaginal discs. To ensure sufficient sample size, we used mixed leg and wing imaginal discs. To our surprise, we found that ATP levels were significantly higher in br⁵ discs (P = 0.03, Figure S2).

Change in br expression in imaginal discs features an isoform switch at metamorphosis

Although the large proportion of br-regulated genes that are temporally regulated at metamorphosis is generally consistent with the hierarchical model of br gene regulation, in which br is induced by ecdysone and in turn induces the late response genes, we also found many br-regulated genes that do not appear to be developmentally regulated (Figure 4B). We therefore suspected a more complex relationship between ecdysone-regulated and br-regulated genes. Given that previous work has shown differential expression of br isoforms in different tissues (Huet et al. 1993), we wanted to investigate the expression of br in imaginal discs relative to the late larval ecdysone pulse. We therefore performed a northern blot analysis using total RNA isolated from wild-type (Binsn/Y) leg discs at six time points from 18 hr before pupariation to 6 hr after pupariation (Figure 5A), and probed the samples for expression of all four br isoforms (using a probe specific to the core BTB/POZ domain) and the br-Z2 isoform. It is noteworthy that the br Z2 isoform is clearly expressed in leg imaginal discs as early as 18 hr before puparium formation, and that its expression tails off just after the onset of metamorphosis and is gone by 4 hr APF. In contrast, using the br-core probe, we note that the br-Z1 isoform (∼4.4 kb transcript) shows minimal expression 18 hr before pupariation, and highest expression between 2 and 4 hr after pupariation in leg imaginal discs. This pattern of gene expression was confirmed with five additional imaginal disc Northern blots, four of which included both leg and wing samples, and closely matches what has been reported for whole animals at these time points (Andres and Thummel 1994).

Figure 5.

Downregulation of br expression at the onset of metamorphosis is required for normal late prepupal and pupal development. (A) Northern blot analysis of total RNA extracted from leg imaginal discs from third instar larvae (−18 hr and −4 hr relative to pupariation) and prepupae (0, +2, +4, and +6 hr) showing expression of all br isoforms (br-core) and just br-Z2. Expression of the Z2 isoform is substantially reduced by 2 hr after pupariation and is absent by 4 hr after pupariation. Note that the Z1 isoform (∼4.4 kb, indicated by asterisk) is upregulated at the onset of metamorphosis (0 hr) and is strongly expressed by 2 hr after pupariation in leg discs. rp49 was used as a control for loading and RNA transfer. (B) Photomicrographs of w¹¹¹⁸ (left side) and HS-brZ2 (right side) early pupae ∼20–24 hr after pupariation that had been heat shocked at 37° for 60 min within 6 hr before pupariation. The HS-brZ2 pupae has short legs (arrows indicate the distal tips of legs in the two animals). (C) Quantification of pupation and eclosion in w¹¹¹⁸ and hs-br Z2 animals that had been heat shocked within 6 hr of pupariation. (D) Brightfield photomicrograph of a leg disc from a HS-brZ2 dead prepupae ∼20–24 hr after pupariation that had been heat shocked at 37° for 60 min within 6 hr before pupariation. Note that this animal had not pupated and the leg disc was found inside the degenerating prepupae. This terminal phenotype is similar to what is seen in a wild-type prepupae at about 6 or 7 hr after pupariation. (E) Brightfield photomicrograph of leg discs from w¹¹¹⁸ (left side) and HS-brZ2 (right side) 4 hr after pupariation that had been heat shocked at 37° for 60 min within 6 hr before pupariation. Note that the HS-brZ2 leg disc looks similar to that from w¹¹¹⁸. Bar, 100 μm.

Since transcription of the Z2 isoform of br appears to stop at or near the onset of metamorphosis, we suspected that derepression of some Br-repressed genes may be critical for leg morphogenesis. We therefore attempted to disrupt this process by misexpressing the br Z2 isoform during the prepupal stage. To accomplish this, we heat shocked late larvae carrying a heat-inducible br Z2 construct (Crossgrove et al. 1996), collected all animals that pupariated within 6 hr after the heat shock, and followed the animals through prepupal and pupal development. In six separate experiments totaling 99 animals, only 50% of the HS-brZ2 animals pupated, and of those only 4 developed to late pupae (none eclosed), with the remaining dying as early pupae (Figure 5, B and C). Most of those that pupated had short legs and wings (Figure 5B) or poorly everted heads. In contrast 98% of w¹¹¹⁸ animals treated in the same way pupated (n = 106), and developed to late pupae (Figure 5C). In a separate experiment of 80 w¹¹¹⁸ animals treated in the same way 73% eclosed (Figure 5C). Dissection of leg imaginal discs from the dead hs-brZ2 prepupae revealed variable terminal leg phenotypes, but most appeared to arrest at stages comparable to wild-type legs between 4 and 7 hr APF (Figure 5D). To confirm that the first several hours of prepupal development occurred normally in these animals, we repeated the experiment, collected animals as they pupariated, aged them 4 hr, and then dissected and imaged the leg imaginal discs. All the legs appeared to have elongated normally at +4 hr regardless of how many hours (0–5 hr) before pupariation that they were heat shocked (Figure 5E). Taken together, these results suggest that either the continued expression of some br-induced genes or failure to derepress some br-repressed genes are detrimental to later prepupal and pupal leg development.

Functional analysis of br-induced genes in leg morphogenesis

Given the important role that br plays in leg disc morphogenesis, we expected that some of the genes induced by br in the leg discs at metamorphosis would also be critical for this process. To test this, we induced RNAi using Distal-less (Dll)-GAL4. Dll-GAL4 is expressed in the distal tibia and tarsal segments during late larval and prepupal time points (Cohen 1993; Ward et al. 2003). We chose br-target genes that fell into a number of different categories. We tested genes with “ATP generation from ADP” and “Generation of precursor metabolites and energy” gene ontology annotations. We tested genes with peptidase or peptidase inhibition functions due to our observation that basal ECM degradation is perturbed in br⁵ mutant leg discs (Figure 3), and we were particularly interested in Stubble (Sb) because it has been shown in previous screens to have an enhancer of br effect (Beaton et al. 1988; Gotwals and Fristrom 1991; Ward et al. 2003). We tested members of the Enhancer of split complex due to the enrichment of the “Notch signaling” GO term among br-induced genes. We also tested a collection of genes that have chitin-related functions since the gene ontology enrichment showed an overrepresentation of these genes in the br-induced set. Finally, we also tested several genes in the Ecdysone-induced 71E cluster (Eig71E), since 6 of the 11 Eig71E genes were found to be br-induced.

In control experiments, we drove the expression of br-RNAi in distal leg segments and observed no adults; +4 hr leg discs from Dll > br-RNAi animals revealed defects similar to that observed in br⁵ animals (Figure S3, B and C). Specifically, both br⁵ and Dll > br-RNAi leg imaginal discs show incomplete elongation with deeper folds between segments, although discs in which br was knocked down via RNAi show slightly greater elongation than br⁵ legs. No additional elongation was observed in Dll > br-RNAi leg discs beyond 4 hr APF (data not shown). Distal leg expression of Sb-RNAi resulted in 18% of the animals displaying a malformed leg phenotype characterized by shorter, fatter distal tibia and tarsal segments (Figure 6B). This malformed phenotype is similar to that observed in a hypomorphic Sb allele (Figure 6C). These experiments validate the use of Dll-GAL4 to reduce gene expression of br-induced genes during late larval and prepupal leg morphogenesis.

Figure 6.

Driving RNAi against some br-induced genes in distal leg segments leads to malformed adult legs. Brightfield photomicrographs of adult legs from w¹¹¹⁸ (A), Dll > Sb-RNAi (B), Sb^Ebr228 (C), Dll > dy-RNAi (D), Dll > CG9416-RNAi (E), and Dll > serp-RNAi (F). Note that the Dll > Sb-RNAi (B) has a shorter and fatter distal tibia and tarsal segment similar to that observed in the homozygous sbd mutant (brackets) (C). Malformed phenotypes include bends or kinks in the tibia (arrows), and misshapen tarsal segments (asterisks). Dll > serp-RNAi flies (F) show a high penetrance of missing distal tarsal segments. Bar, 200 μm.

We next drove RNAi against 36 br-induced genes using both long-hairpin and short-hairpin UAS-RNAi lines with the Dll-GAL4 driver. A total of 41 RNAi lines were tested, and phenotypes similar to those reported in screens by Ward et al., including bent tibias and poorly proportioned tarsals (Ward et al. 2003), were observed in several of the lines (Table 2 and Figure 6). In total, five lines exhibited malformed legs at a rate of >4%, including two lines from the protease and protease inhibitor group, CG9416 and Sb, and two lines knocking down chitin-related genes, dusky (dy) and serp. The final line carried a construct against grainy head (grh). In all cases, straight winged flies that carried the GAL4 driver showed significantly more leg defects than curly-winged flies from the same cross that did not carry the GAL4 driver (P < 0.001; Table 2). The phenotypes observed in each of these crosses differed (Figure 6): in those from the dy line, flies exhibited slightly curved tibias, while flies from the CG9416 line exhibited more severely kinked tibias. serp line flies frequently showed bent or missing tarsal segments, while the tarsal segments in Sb and grh flies were often misshapen or misproportioned (Figure 6B and data not shown). Variation in phenotype penetrance and expressivity also existed between crosses with lines carrying constructs against the same genes (Table 2). For example, bent tibias were observed in 38% of flies expressing a short-hairpin construct against CG9416, while flies expressing a long-hairpin construct did not display this phenotype.

Table 2. Frequency of leg defects in flies expressing RNAi against br-induced genes.

Target gene	TRiP or VDRC		Line no.		% Carrying dll > Gal4	% With malformed legs	Note
Protease and protease inhibitors
CG3355	TRiP		52897		48.9%	0.0%
CG5639	TRiP		57376		48.2%	0.0%
	VDRC		1306		50.0%	1.2%
CG8170	TRiP		61889		36.2%	0.0%
CG9416	TRiP		57563		28.6%	37.5%	Bent tibias; 0 out of 80 flies without dll > Gal4 had malformed legs
						n = 32
	VDRC		10064		45.9%	2.6%
Stubble	TRiP		42647		30.4%	23.0%	Bent femurs, tibias, and tarsals; 2 out of 159 flies without dll > Gal4 had malformed legs
						n = 61
Serine Protease Immune Response Integrator	TRiP		42882		47.1%	0.0%
Notch pathway genes
Enhancer of split m4	TRiP		61261		41.3%	1.5%
Enhancer of split m7	TRiP		29327		51.6%	1.3%
Enhancer of split m8	TRiP		26322		39.4%	1.2%
	VDRC		37686		45.6%	0.0%
Enhancer of split mα	VDRC		35886		0.0%	N/A	No adult flies—larvae failed to molt
Enhancer of split mγ	TRiP		51762		48.0%	0.0%
Scabrous	TRiP		56928		54.3%	0.0%
Eig71E genes
Ecdysone-induced gene 71Ec	TRiP		57384		45.9%	0.0%
Ecdysone-induced gene 71Ed	TRiP		56952		52.4%	0.0%
Ecdysone-induced gene 71Ef	TRiP		55930		46.1%	0.0%
Ecdysone-induced gene 71Eg	TRiP		56953		47.9%	0.9%
Chitin-related genes
Dusky	TRiP		34382		55.3%	0.0%	All flies had slightly curved tibias, but not severe enough to be counted as malformed
	VDRC		3299		41.5%	100.0%	Bent tibias; 0 out of 62 flies without dll > Gal4 had malformed legs
						n = 44
Imaginal disc growth factor 4	TRiP		55381		45.8%	0.0%
Serpentine	TRiP		65104		15.4%	18.2%	Bent or broken tarsals; 0 out of 192 flies without dll > Gal4 had malformed legs
						n = 35
Vermiform	TRiP		57188		0.0%	N/A	Died as prepupae
Metabolic genes
Aldolase 1	TRiP		65884		25.0%	0.0%
Enolase	TRiP		26300		1.4%	0.0%
Glyceraldehyde 3 phosphate dehydrogenase 2	TRiP		26302		0.0%	N/A	Died as pupae/some failed to evert legs
Glycogen phosphorylase	TRiP		33634		40.3%	1.9%
Phosphofructokinase	TRiP		34336		46.3%	0.0%
Phosphoglycerate kinase	TRiP		33633		43.2%	0.0%
Phosphoglyceromutase 78	TRiP		26303		49.1%	0.9%
Phosphoglucose mutase 1	TRiP		34345		12.9%	0.0%
Triose phosphate isomerase	TRiP		51829		47.7%	0.0%
Miscellaneous
Amalgam	TRiP		33416		14.8%	0.0%
	VDRC		22944		45.2%	0.0%
CG5758	TRiP		57808		33.3%	0.0%
Slowdown	VDRC		106464		45.4%	3.6%
CG10960	TRiP		34598		45.8%	0.0%
CG12026	VDRC		42480		42.7%	6.6%
Grainy head	TRiP		42611		37.8%	6.6%	Short, fat tarsals; 0 out of 175 flies without dll > Gal4 had malformed legs
						n = 106
Ecdysone-inducible gene E3	VDRC		16402		40.0%	0.0%

Lines listed include both TRiP and VDRC (long hairpin) lines, with stock numbers shown. Lines exhibiting a significant number of malformed legs are indicated in bold. Percentages of adult flies carrying the Gal4 driver (determined by absence of curly wings) is indicated; values found by chi-square test to be significantly lower than 50% are underlined. All RNAi lines were crossed with virgin females carrying the Dll-Gal4 driver and a UAS-Dicer construct.

Several lines produced fewer of the straight-winged RNAi-expressing flies than expected, suggesting some lethal effect. In total, 16 crosses targeting br-induced genes resulted in significantly <50% of straight wing flies (Table 2). These include four crosses (CG9416, grh, Sb, and serp) that exhibited malformed legs at a rate of >4%, as well as three crosses {vermiform (verm), Enhancer of split mα [E(spl)mα], and Glyceraldehyde 3 phosphate dehydrogenase 2 (Gapdh2)} that produced no RNAi-expressing straight wing flies that could be scored for the malformed leg phenotype. Individuals expressing RNAi against E(spl)mα failed to undergo larval molts, appearing to remain as first instar larvae for several days before dying (data not shown). In contrast, individuals expressing RNAi against verm or Gapdh2 pupariated. Among offspring of the verm RNAi cross, the majority of the animals pupated, but without everting their legs. After about a day following pupation, there was noticeable necrotic tissue in the areas where the legs, the antennae, and wing margins form in wild-type pupae (Figure 7B). A similar phenotype was observed in offspring of the Gapdh2 RNAi cross, but at a relatively low penetrance (data not shown). When we dissected +4 hr leg discs from Dll > verm-RNAi prepupae, we found that they exhibited an elongation defect that was more severe than what was observed in br⁵ mutants or br RNAi knockdown flies (compare Figure 7D to Figure S3, B and C). To investigate whether the majority of the br⁵ mutant phenotype is due to loss of verm, we drove UAS-verm in br RNAi imaginal discs (using Dll-GAL4). Neither expression of verm (nor its fellow chitin deacetylase, serp), was able to rescue br RNAi flies to adulthood (7 independent experiments with UAS-verm with n = 409 adults eclosing and 4 independent experiments with UAS-serp with n = 291 adults eclosing). We then examined the +4 hr leg imaginal discs from animals expressing UAS-serp in Dll > br-RNAi and found no rescue of the prepupal leg defect, although there is clear evidence of Serp expression (n = 54 legs from 14 animals; Figure S3).

Figure 7.

Distal leg expression of verm-RNAi leads to severe defects in leg morphogenesis. (A and B) Color photomicrographs of Dll-Gal4 (A) and Dll > verm-RNAi (B) midstage pupae. Note the lack of elongated legs in the Dll > verm-RNAi animal that instead shows necrotic tissue in the areas where the legs should evert from (arrows). (C and D) Brightfield photomicrographs of leg imaginal disc from Dll-Gal4 (C) and Dll > verm-RNAi (D) +4 hr prepupae showing little distal elongation and almost no segmentation in the verm-RNAi disc. Bar, 100 μm.

Another metabolism gene, Phosphoglucose mutase 1 (Pgm1), was of interest because only 13% of expected Dll > Pgm1-RNAi animals eclosed (Table 2), and the cytological location of Pgm1 in 72D8 on the third chromosome overlapped a br¹-interacting loci from our previous screen (Ward et al. 2003). We examined Dll > Pgm1-RNAi prepupae and pupae and observed that approximately 25% failed to pupate; however, those that did pupate had normal looking legs (data not shown). We then crossed flies carrying the lethal Pgm1^LA00593 allele (genotype Pgm1^LA00593/TM3 Sb Ser) to br¹ females and observed that 21.4% of the br¹/Y; Pgm1^LA00593/+ adults exhibited malformed legs (n = 103).

Discussion

In this study, we have demonstrated that Br is required for many of the tissue-level events required for leg morphogenesis during the first several hours of metamorphosis in Drosophila, and have identified a collection of genes regulated by br that may help explain this developmental process. Many previous studies have demonstrated that leg morphogenesis requires ecdysone signaling (Mandaron 1971; Fristrom et al. 1973), cell shape changes and rearrangements (Condic et al. 1991), and proteolysis of the ECM (Diaz-de-la-Loza et al. 2018). Here, we show that leg discs in br⁵ mutant animals fail to fully degrade their ECM (Figure 2), and that even if the ECM is degraded by exogenous trypsin, they fail to fully elongate (Figure 3), suggesting that the underlying cell shape changes and rearrangements are also defective. Anisometric cell shapes and wider tarsal segments in +4 hr leg discs from br⁵ mutant animals support this suggestion (Figure 3). Consistent with all of these observations, RNA-sequencing comparisons between br⁵ and w¹¹¹⁸ leg imaginal discs at the onset of morphogenesis indicate that br regulates the expression of several potential proteases and chitin-modifying genes. RNAi-based functional analyses confirm a requirement for some of these br-induced genes in leg imaginal disc morphogenesis (Figures 6 and 7). Furthermore, bioinformatic analyses of br-regulated genes suggests that br is required to maintain robust metabolism in imaginal discs when glycolysis is otherwise downregulated in the larval tissues.

br-dependent ECM remodeling plays a critical role in leg development

Our results suggest that br plays a central role in remodeling the ECM in leg imaginal discs during metamorphosis. We show that Collagen IV is not cleared from the basal ECM in br⁵ mutant prepupal leg discs (see Figure 2). In addition, four of the six genes that resulted in malformed adult legs when downregulated by RNAi (dusky, serpentine, Stubble, and vermiform) have been proposed to interact with the apical ECM, while another (grainy head) is known to regulate ECM genes (Pare et al. 2012; Yao et al. 2017). Stubble encodes a serine protease whose endopeptidase domain is required for normal leg development (Appel et al. 1993; Hammonds and Fristrom 2006). Stubble protein has been shown to be required for degradation of the Dumpy protein, an apical ECM component that links the epithelial cells to the cuticle (Diaz-de-la-Loza et al. 2018). In the absence of this degradation, imaginal discs fail to elongate. serpentine and vermiform were identified as genes encoding chitin deacetylases involved in organization of the larval tracheal apical ECM (Luschnig et al. 2006). These genes are also found in the larval cuticle, where they are apically secreted and necessary for the formation of a stable matrix (Pesch et al. 2015; Pesch et al. 2016). The dusky gene encodes a zona pellucida protein that is expressed in epithelial tissues across Drosophila development, most notably in the developing wing disc during wing elongation (Roch et al. 2003; Jazwinska and Affolter 2004; Ren et al. 2005). Zona pellucida proteins are membrane-anchored proteins that interact with the ECM; a study of embryonic denticles showed localization to specific apical subcellular domains, where the proteins organize and modify the ECM to control cell shape changes (Fernandes et al. 2010). grainy head encodes a transcription factor that regulates genes involved in cuticle formation (Bray et al. 1989; Dynlacht et al. 1989; Bray and Kafatos 1991). grainy head mutants display defective ECM phenotypes in the head skeleton (Nusslein-Volhard et al. 1984; Bray and Kafatos 1991), trachea (Hemphala et al. 2003), and in wound repair (Mace et al. 2005). Chromatin immunoprecipitation data suggests that Grainy head also regulates some of the other leg morphogenesis genes identified in this study, including CG9416, serpentine, vermiform, and Stubble (Pare et al. 2012; Yao et al. 2017).

The connection that most of these genes have with the ECM is consistent with the known role of the ECM in imaginal disc elongation; the ECM provides a constraining force to imaginal discs (Pastor-Pareja and Xu 2011), and both the apical and basal ECMs must be degraded before elongation (Diaz-de-la-Loza et al. 2018). Our results suggest that br provides a key regulatory mechanism for these processes, and that they underlie a large part of the br⁵ phenotype. br does not appear to be wholly responsible for the proteolysis of the ECM at metamorphosis, however. Notably absent from our list of br-regulated genes are the matrix metalloprotease genes Mmp1 and Mmp2, which are required for breakdown of the basal ECM during disc eversion (Srivastava et al. 2007), although both of these genes are developmentally upregulated at the onset of metamorphosis in leg discs (Table S4).

Br appears to play a role in maintaining metabolic activity during metamorphosis

The amorphic br⁵ allele leads to failure of imaginal disc development, and, as expected, several development-related biological process gene ontology terms were enriched in br-induced genes, including “developmental process,” “anatomical structure development,” and “organ morphogenesis.” Somewhat surprisingly, the enriched gene ontology terms also included glycolysis-related metabolic terms, including “pyruvate metabolic process,” “ATP generation from ADP,” and “glycolytic process.” In support of these findings, KEGG pathway analysis identified glycolysis/gluconeogenesis as the most significantly enriched pathway from these genes. More interestingly, neither analysis revealed enrichment of glycolysis-related metabolic genes in the comparison of w¹¹¹⁸ leg discs from −18 to 0 hr (neither up- nor downregulated at 0 hr). This finding contrasts with previous microarray studies in both D. melanogaster and the silkworm Bombyx mori showing downregulation of glycolytic and other metabolic genes in whole animals at pupariation (White et al. 1999; Tian et al. 2010). Taken together, these results suggest that while larval tissues are shutting down metabolism (Merkey et al. 2011), the imaginal discs retain metabolic activity throughout the transition to prepupa, and that this metabolic program requires Br. This energy requirement could be needed for the cell shape changes and rearrangements that drive leg morphogenesis. Consistent with this idea, knockdown of the glycolysis gene Gapdh2 results in failure of leg discs to elongate and evert in some pupae, and knockdown of any of four other metabolism genes appears to result in larval or pupal lethality (Table 2). One of these genes, Phosphoglucose mutase 1 also dominantly enhanced the malformed leg phenotype associated with br¹, similar to what we previously observed with the overlapping deficiency Df(3L)brm11 (Ward et al. 2003). Given these observations, we were surprised to find ATP levels were higher in br⁵ mutant discs than in w¹¹¹⁸ discs at 4 hr after pupariation (Figure S2), but this result may instead reflect the reduced energy consumption of br⁵ mutant leg discs that have already arrested morphogenesis by 4 hr APF.

broad is dynamically regulated before and during metamorphosis

The Broad-Complex was initially described as an early-response gene in the ecdysone-induced transcriptional cascade (Chao and Guild 1986; DiBello et al. 1991), but subsequent experiments suggested that the regulation of br by ecdysone both in whole animals and in imaginal discs is more complex and involves an isoform switch from Z2 to Z1 at pupariation. Specifically, studies in the salivary glands and whole animals showed a switch to Z1 upon pupariation (Andres et al. 1993; Huet et al. 1993), and imaginal discs cultured in the presence of ecdysone resulted in persistence of the Z1 transcript even after the other br isoforms were no longer detected (Bayer et al. 1996). Our Northern blot analysis confirms this isoform switch at the onset of metamorphosis in imaginal discs, and also reveals that br Z2 is expressed earlier in the discs than would be predicted by whole animal blots (Figure 5A; Andres and Thummel 1994). This latter observation is somewhat paradoxical considering that the br-induced genes are a reasonably good subset of the developmentally induced genes revealed by RNA-sequencing comparisons of leg discs from −18 to 0 hr (Figure 4). This raises the possibility that many of these br-induced genes are also regulated in a temporal fashion, perhaps by ecdysone/EcR/Usp directly. Future experiments should be aimed at exploring the more complex regulation of these genes, particularly those that we have identified as playing important roles in driving leg morphogenesis. It is also interesting that Br appears to be responsible for controlling early prepupal developmental events (proteolysis of the ECM, cell shape changes and rearrangements) when the Z2 isoform is ostensibly “off.” We believe that this is driven by perdurance of the Br protein. In support of this, Emery et al. (1994) demonstrated that levels of the Br-Z2 protein remained strong in the imaginal discs until between 4 and 8 hr after pupariation, despite the decrease in transcription by 0 hr. Finally, it is worth noting that we identified a large number of br-repressed genes in our RNA-sequencing experiment, raising the possibility that derepression of br-repressed genes may also contribute to normal leg morphogenesis. Our functional studies, in which we extended br-Z2 expression through a heat-inducible promoter, resulted in complete lethality, with half of the animals dying prior to pupation. The terminal leg phenotypes in the prepupal lethal animals revealed that development was normal until 4–7 hr after pupariation. For those animals that pupated, their legs and wings were short, the heads were often malformed, and they died soon after pupation. Taken together, these findings are consistent with a model in which perdurance of Br is sufficient to positively regulate genes needed for early prepupal leg development, at which point reduced Br results in derepression of a second set of genes that positively affects later prepupal and pupal leg development.