2 G3, 2021, 11(12), jkab349 DOI: 10.1093/g3journal/jkab349 Advance Access Publication Date: 2 October 2021 Mutant Screen Report Genome-wide phenotypic RNAi screen in the Drosophila wing: phenotypic description of functional classes Ana López-Varea ,1 Patricia Vega-Cuesta ,1 Ana Ruiz-Gómez,1 Cristina M. Ostalé ,1 Cristina Molnar ,1,2 Covadonga F. Hevia ,1,† Mercedes Martı́n ,1,‡ Maria F. Organista,1 Jesus de Celis ,1 Joaquı́n Culı́,1 Nuria Esteban,1 and 1, Jose F. de Celis * 1 Centro de Biologı́a Molecular “Severo Ochoa,” CSIC and Universidad Autónoma de Madrid, Madrid 28049, Spain, and 2 IRB Barcelona, Barcelona 08028, Spain *Corresponding author: Email: jfdecelis@cbm.csic.es † Present address: Universidad Pompeu Fabra, Barcelona 08024, Spain. ‡ Present address: Universidad Miguel Hernández, Alicante 3202, Spain. Abstract The Drosophila genome contains approximately 14,000 protein-coding genes encoding all the necessary information to sustain cellular physiology, tissue organization, organism development, and behavior. In this manuscript, we describe in some detail the phenotypes in the adult fly wing generated after knockdown of approximately 80% of Drosophila genes. We combined this phenotypic description with a comprehensive molecular classification of the Drosophila proteins into classes that summarize the main expected or known biochemical/ functional aspect of each protein. This information, combined with mRNA expression levels and in situ expression patterns, provides a sim- plified atlas of the Drosophila genome, from housekeeping proteins to the components of the signaling pathways directing wing develop- ment, that might help to further understand the contribution of each gene group to wing formation. Keywords: phenotype; wing; screen; RNAi Introduction domains underlying pattern formation and cell differentiation by complex networks of interactions between transcription factors Large-scale genetic screens are instrumental to identify the set of and signaling pathways (Ostalé et al. 2018). The wing disc gener- genes affecting a particular process (St Johnston 2002). In addi- ates during metamorphosis a two-layered flat epithelial structure tion, the analysis of mutant phenotypes in individual tissues in which cells differentiate into cuticle or sensory organs in char- helps dissecting the mechanisms participating in its development acteristic patterns. In this manner, the adult wing provides a (Molnar et al. 2006; Cruz et al. 2009). Sequencing and annotation readout of the cellular and molecular mechanisms underlying of the Drosophila genome (Adams et al. 2000), in combination with the development of the wing imaginal disc. the development of techniques to generate loss-of-function con- The understanding of mutant phenotypes is critical to identify ditions using RNA interference now allow global descriptions of the function of a gene. In classical genetics, a mutant phenotype the phenotypic landscape, providing key information to under- precedes the knowledge of the biochemical characteristic of the stand the functional contribution of each gene or gene group to protein encoded by the affected gene. Consequently, “gene tissue development (Schnorrer et al. 2010; Valakh et al. 2012; functions” are assigned without any relation to the molecular Perkins et al. 2015; Ostalé et al. 2018). function of the corresponding protein (St Johnston 2002; The Drosophila wing has been the subject of a variety of genetic Kaufman 2017). However, since the sequencing and annotation screens aiming to define genes affecting its growth and the pat- of the Drosophila genome, we have access to relevant information terning of structures such as the wing margin or the wing veins about the biochemical function of gene products before any ge- (Bjorklund et al. 2006; Molnar et al. 2006; Bejarano et al. 2008; Cruz netic data are available. Thus, the molecular annotation of the et al. 2009). The wing develops from an epithelial tissue, the wing genome allows the correlation between mutant phenotypes and imaginal disc. The development of the wing disc involves regu- likely biochemical functions. lated cell proliferation that drives the tissue from an embryonic In order to correlate mutant phenotypes and biochemical primordium of 40 cells to the mature disc composed of approxi- functions, it is necessary to compile the available information mately 50,000 cells (Beira and Paro 2016; Ostalé et al. 2018). In ad- about a protein into informative terms. The most complete dition, the development of the disc involves a variety of cellular source of information about gene function is the gene ontology operations that are common to other multicellular developmen- (GO) database, which assigns to each gene product a set of defini- tal processes. They include the generation of gene expression tions related to its molecular function, subcellular localization, Received: June 06, 2021. Accepted: September 23, 2021 C The Author(s) 2021. Published by Oxford University Press on behalf of Genetics Society of America. V This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. 2 | G3, 2021, Vol. 11, No. 12 and biological function (Ashburner et al. 2000). GO annotations al- the molecular functions assigned for each gene and searched for low enrichment analysis on gene sets, identifying the terms un- phenotypic enrichment within each group. der- and over-represented in the set. As such, the use of GO enrichments is fundamental to extract meaningful biological in- formation from data sets obtained in, for example, differential Materials and methods gene expression experiments or genetic screens. In the GO data- Wing phenotypes base, each gene is described by a collection of identifiers, the We arranged the wing phenotypes of the Gal4/UAS-RNAi com- number of which depends on the current knowledge about the binations presented in the preceding manuscript (López-Varea gene (Ashburner et al. 2000). For example, for a gene extensively et al. 2021) into sets, each including the genes belonging to each studied such as Notch, GO assigns 130 terms. However, many of molecular group. Each mutant wing was described using a set these terms are related to the multiple development processes of terms that summarize its main phenotypic characteristics. where the function of Notch is required, and only two terms refer The abbreviations we used are listed in Supplementary Table to its participation in a signaling pathway. Furthermore, if we S2. All Gal4/UAS combinations were made at 25 C unless oth- compare the terms assigned to Delta (Dl; 61 terms) and Suppressor erwise stated, and we used the Gal4 lines salEPv-Gal4 (Cruz et al. of Hairless [Su(H); 30 terms], two other components of the Notch 2009), nub-Gal4 or sd-Gal4 (Calleja et al. 1996), and UAS-RNAi signaling pathway, with those of Notch, we find only 4 coinciden- lines obtained from the Vienna Drosophila Resource Center ces from 221 aggregated terms. Most terms are either unique (97) (VDRC), the National Institute of Genetics Fly Stock (NIG-Fly), or shared by two genes (57). For this reason, we generate a simpli- and the Drosophila Bloomington stock center (BDSC). The fied nomenclature compacting the different GO descriptions into grouping and phenotypic description for the entire set of a set of terms encompassing the key biological aspect of a gene Drosophila genes are presented in Supplementary Table S1. A product. This classification can be used for enrichment analysis list with all wing pictures included in this manuscript is pre- focusing on the most relevant aspect of each gene function. sented in Supplementary Table S3. Pictures of wings included The adult wing in flies raised under standard conditions has a in López-Varea et al. (2021) and Ostalé et al. (2021) are also in- constant size, morphology, and pattern of differentiated ele- cluded in Supplementary Table S3. All wing pictures we have ments. Changes in any of these characteristics are the constitu- were submitted to the Figshare repository: https://doi.org/10. 6084/m9.figshare.16624645.v1; https://doi.org/10.6084/m9.fig ents of each mutant phenotype, and they manifest in different share.16624630.v1; https://doi.org/10.6084/m9.figshare.166246 combinations in each mutant wing. For the description of mutant 03.v1; and https://doi.org/10.6084/m9.figshare.16624591.v1 phenotypes, we develop a simplified terminology reflecting the morphological changes observed in each mutant wing such as its Molecular classification size [S and S(L)]), wing size and patterning of veins (S-P), differen- In this manuscript, we described in more detail the following mo- tiation of veins (V and Vþ), defects in the wing margin (WM), lecular classes defined in López-Varea et al. (2021): Ribosomal bi- defects in dorsoventral wing surfaces adhesion (WA), defects in ology (RIB), Cell adhesion genes (CA), Protein biology (PRO), RNA wing pigmentation (WP), changes in trichome size, number, po- biology (RNA), DNA biology (DNA), Cell signaling (SIG), Cellular larity, or spacing (cell differentiation; CD), changes in overall metabolism (MET), Cuticle proteins (CUT), Coding genes without wing shape (WS), and other wing defects including incomplete conserved sequence domains (CG), and Coding genes with con- unfolding of the wing surfaces (WF), appearance of necrotic served sequence domains CGh. The classes Immune responses patches and wing cuticle with abnormal appearance or lack of ri- (IMM) and Cell death (CD) are only included in Supplementary gidity (wing differentiation; WD). All cases in which the wing fails Table S1 and were not further analyzed because these genes to form were described as “nW” (no wing), phenotypes affecting rarely affect wing formation. Conversely, the classes the bristles of the wing margin as “Q” and phenotypes of ectopic Cytoskeleton (CYT), Transport across membranes (TRA), Protein bristle formation as “Qþ.” A list of all abbreviations can be found secretion (PTR), and Cell division (DIV) include many genes re- in Figure 1, Supplementary Table S2, and in the accompanying quired for wing formation and will be described elsewhere. In ad- manuscript (López-Varea et al. 2021). Some mutant phenotypes dition, we further classified the genes of each functional class are more informative than others with regard to the function of into 3–15 groups reflecting the most prevalent aspect of each the corresponding gene during wing development. In general, the gene (Table 1). For this classification, we used GO identifiers and phenotypes caused by the knockdown of genes with unknown de- IP domains as well as the Flybase gene group classification and velopmental roles can be interpreted by comparison to those gene summaries (Thurmond et al. 2019). In Table 1, we present a caused by perturbations in genes with known developmental list of all the groups and their abbreviations. roles. For example, we expect that knockdown of housekeeping genes involved in protein synthesis, modifications, trafficking, or Gene expression analyses and in situ degradation, in RNA processing or in general metabolic pathways hybridization data would reduce cell viability, and when this occurs throughout To define whether a gene is or not expressed in the wing disc, we imaginal development might lead to cell lethality and to a failure used RNA-seq data (Flegel et al. 2016; reads from run SRR3478156) to form a wing. In contrast, we expect that genes involved in the and microarray data (GeneChipTM Drosophila Genome 2.0 Array; modulation of the activity of signaling pathways would have phe- Affymetrix) obtained from mRNA extracted from dissected third notypes similar to perturbations in the main components of instar wing imaginal discs (Organista et al. 2015). The in situ these pathways (Molnar et al. 2011). In any case, a mutant pheno- hybridizations pictures correspond to a collection of 635 experi- type offers an entry point to direct further genetic and develop- ments published in Hevia et al. (2017), Molnar et al. (2012), and mental characterization of the corresponding gene. In this Organista et al. (2015). All the in situ hybridization experiments accompanying manuscript, we present and analyzed most of the were carried out in our laboratory. The expression patterns were 16 molecular classes into which we classify all the Drosophila classified as no expression (NE), ubiquitous expression (GEN), genes. For each class, we also defined several groups based on and restricted expression (PAT). Any expression pattern was A. López-Varea et al. | 3 Figure 1 Ribosomal class. (A) Distribution of the 235 genes included in the ribosomal class. Cytoplasmic large and small ribosomal subunits (CLRP and CSRP), Mitochondrial large and small ribosomal subunits (MLRP and MSRP) and Ribosomal-associated proteins (RAP). (B) Percentage of genes of the ribosomal class groups for which we tested its knockdown phenotype (Done; dark gray column), genes expressed in the wing disc (Exp; light gray column), and genes with a lethal or visible phenotype in knockdown conditions (Phe; striped column). (C) Frequency of mutant phenotypes observed in the CLRP, CSRP, MLRP, MSRP, and RAP groups: L (lethal), nW (failure to form the wing), S-P (changes in the size of the wing and relative positions of the veins), S (size of the wing), V (including ectopic or thicker veins and loss of veins), WA (failures in the adhesion between the dorsal and ventral wing surfaces), and WD (altered cuticular differentiation). The phenotypes WS (shape of the wing), WM (defects in the wing margin), CD (changes in cell size or trichome differentiation), WP (changes in wing pigmentation), and Q (differentiation of ectopic bristles in the wing and loss of bristles in the wing margin) were grouped as “Others” (OTH). (D–G) Expression pattern in the wing disc of the genes CG7354 (mRpS26), CG13880 (mRpL17), CG12275 (RpS10a), and CG5497 (mRpS28), showing robust and generalized expression. In situ hybridizations pictures correspond to experiments published as supplementary material in Molnar et al. (2012), Organista et al. (2015), and Hevia et al. (2017). (H–H’) Adult wings of UAS-Dicer2/þ; nub-Gal4/UAS-CG4338- RNAi (H) and UAS-Dicer2/þ; salEPv-Gal4/UAS-CG4338-RNAi, showing the transition from “nW” to “S” phenotype. (I, J) Adult wings of UAS-Dicer2/þ; nub- Gal4/UAS-betaNACtes3-RNAi (I) and UAS-Dicer2/þ; nub-Gal4/UAS-RpL37b-RNAi. (K–K’) Adult wings of UAS-Dicer2/þ; nub-Gal4/UAS-RpL27-RNAi (K) and UAS- Dicer2/þ; salEPv-Gal4/UAS-RpL27-RNAi (K’), showing the more frequently observed transition from a “nW” to a “S-P” phenotype. (L) Adult wing of UAS- Dicer2/þ; salEPv-Gal4/UAS-RpS18-RNAi genotype showing the more frequently found class of S-P phenotype. defined as “ubiquitous” when the staining occurs throughout the these classes several groups based in the known characteristics wing disc at similar levels that were consistently higher than of each gene (Table 1). This classification was based when possi- those observed in in situ experiments carried out with sense ble on the “gene group” annotation from Flybase (Thurmond et al. probes. A list with all wing imaginal disc in situ hybridization pic- 2019). tures included in this manuscript is presented in Supplementary Table S3. Ribosomal genes We include in this class all genes encoding components of the cy- toplasmic and mitochondrial large and small ribosomal proteins Results and discussion (54 genes for RIB-CLRP, 39 for RIB-CSRP, 47 for RIB-MLRP, and 28 We screened in the fly wing UAS-RNAi lines targeting 10,920 for RIB-MSRP; structural components of ribosomes; see Drosophila genes of which 7036 were estimated as being expressed Supplementary Table S1.1). We also included in this class, a in the wing disc. We also classified all Drosophila genes into 16 group of 67 genes annotated as ribosomal binding and assembly functional classes encompassing the more functionally relevant proteins, ribosomal biogenesis, and other genes encoding protein aspect of each gene. For this classification, we relied on available domains suggestive of a role in ribosome biology (Ribosomal-as- GO annotations, IP domains, and bibliographical searches. When sociated proteins; RIB-RAP, Figure 1A; Supplementary Table S1.1). a gene could not be classified unambiguously into only one func- Most of Ribosomal genes are expressed in the wing disc (95%), tional class, we selected one based on prior knowledge about the and this expression occurs at high levels (Figure 1B; gene function during imaginal development. In this accompany- Supplementary Table S1.1). In those cases for which we have in ing manuscript, we present a detailed description of 10 of these situ hybridization data, the expression was robust and ubiquitous functional classes, aiming to identify whether they have prefer- in the wing disc and other tissues (Figure 1, D–G). Mutations in ri- ential phenotypic characteristics. We also define within each of bosomal genes are lethal in homozygosis and they cause 4 | G3, 2021, Vol. 11, No. 12 Table 1 Molecular classes and groups Ribosome (RIB) RIB-CLRP RIB-CSRP RIB-MLRP RIB-MSRP RIB-RAP Cytoplasmic large Cytoplasmic small Mitochondrial large ri- Mitochondrial Ribosomal-associated ribosomal subunits ribosomal subunits bosomal subunits ribosomal subunits proteins Cell signaling (SIG) SIG-BMP/TGF SIG-HH SIG-HSW SIG-INR SIG-RTK BMP and TGFb Hedgehog signaling Hippo signaling Insulin receptor and Receptor tyrosine ki- TOR nase SIG-JAK SIG-JNK SIG-NOTCH SIG-TOLL SIG-WNT Jak-Stat signaling Jun Kinase signaling Notch signaling Toll signaling Wingless signaling SIG-GPCR SIG-NP SIG-OBP SIG-GR SIG-OTH GPCR signaling Neural Peptides Odorant-binding pro- Gustatory receptors Other signaling path- teins ways SIG-EC SIG-IPS SIG-JH Ecdysone receptor Inositol phosphate Juvenile hormone Transport across membranes (TRA) TRA-ABC TRA-ICC TRA-IPT TRA-LGIC TRA-NPC ATP-binding cassette Ion channels by con- Importins and expor- Ligand-gated ion chan- Nuclear pore complex transporter ductane tins nels TRA-OTH TRA-ATPASE TRA-SLC TRA-TSP VHA Other transporters H-ATPase Solute carriers Tetraspanins Vacuolar ATPASE Cell adhesion (CA) CA-AJ/SJ CA-CAM CA-ECM CA-IG Adherens and septate Cell adhesion mole- Extracellular matrix Immunoglobulin pro- junctions cules teins Cuticular proteins (CUT) CUT-CBDCP CUT-CO/VM CUT-CPF CUT-MU/Y Chitin binding domain Chorion and vitelline Cuticular protein fami- Mucins and yellow pro- membrane lies teins Metabolism (MET) MET-AA/NT/O MET-ATG MET-CTR MET-DTOX Amino acids, nucleoti- Autophagy Central metabolism Detoxification des and related mol- ecules MET-GLY MET-LIP MET-MTA MET-MIT Carbohydrates and gly- Lipid metabolism Metal metabolism Mitochondrial biology cobiology MET-OC MET-RDX MET-XEN One-carbon Oxidation-reduction Xenobiotic Protein biology (PRO) PRO-CHAP PRO-GLU PRO-KIN PRO-PHO Chaperons Sugar modifications Kinases Phosphatases PRO-PEP PRO-PTS PRO-MOD PRO-UBIT Proteases Proteosome Protein modifications Ubiquitin ligases and transferases RNA biology (RNA) RNA-MOD RNA-OTH RNA-BND RNA-ENZ mRNA modifications Other RNA molecules RNA-binding proteins RNA enzymes RNA-rRNA RNA-SP RNA-TNLF RNA-tRNA Ribosomal RNA mRNA splicing Translation initiation, Transfer RNA elongation, and ter- mination DNA biology (DNA) DNA-CHRO DNA-CMC DNA-ENZ Chromosomal struc- Chromatin modifying DNA enzymes ture proteins complexes DNA-REP DNA-GTF DNA-TF DNA replication General transcription Sequence-specific tran- factors scription factors Cytoskeleton (CYT) CYT-TUB CYT-MYO CYT-ACT Tubulin Myosin Actin Protein transport across membranes (PTR) PTR-EN PTR-ER PTR-GA Endocitosis Endoplasmic reticulum Golgi apparatus (continued) A. López-Varea et al. | 5 PTR-GPI/SRP PTR-LY/PE PTR-VT GPI anchor and signal Lysosome and Vesicle-mediated recognition particle Peroxisome transport Cell division (DIV) DIV-CC DIV-MIT DIV-MEI Cell cycle Mitosis Meiosis CG CG(h) CD IMM Gene with no homol- Gene with InterPro do- Cell death Immunology ogy main Name, abbreviation, and number of genes included in each functional/molecular class and name, abbreviation, and number of genes of each functional group included in each class. dominant Minute phenotypes consisting in the formation of bris- appearance of an inflated balloon, or only a fraction of the wing, tles of smaller size and developmental delay (Marygold et al. resulting in the formation of blisters with variable extension. The 2007). The knockdown phenotypes of ribosomal genes (UAS- region most frequently affected is centered along the posterior Dicer2/þ; nub-Gal4/UAS-RNAi combinations) mostly consist in late cross vein (Figure 2E). In general, these blistery wings appeared larval or early pupal lethality (68%; Figure 1, B and C). All combi- with variable frequency depending on the particular gene knock- nations of cytoplasmic ribosomal genes resulting in viable adults down. developed rudimentary wings or a total absence of the wing The last groups of genes we included in the cell adhesion class (Figure 1, C and H–L). When the knockdown was restricted to the encode cell adhesion molecules of the cadherin (Zartman et al. central region of the wing (UAS-Dicer2/þ; salEPv-Gal4/UAS-RNAi 2009), Fasciclin, Dscam, and Immunoglobulin families (Vogel combinations) this region generally failed to form et al. 2003), as well as a variety of other proteins with a GO de- (Supplementary Table S1; Figure 1K’), although we found some scription indicative of a role in cell adhesion (Supplementary cases in which the wing was fully formed but its size was reduced Table S1.2). These genes were included in two groups: cell adhe- (Figure 1, H–H’). In summary, genes encoding components of the sion molecules with immunoglobulin domains (CA-IG; 79 genes) ribosome are, as expected, expressed at high levels and ubiqui- and other cell adhesion molecules and proteins with predicted tously in the wing disc, and the most frequent phenotypes result- role in cell adhesion (CA-CAM; 98 genes). These genes are fre- ing from Ribosomal genes knockdowns (larval and pupal lethality quently not essential for wing development (CA-CAM and CA-IG and loss of wing tissue) are compatible with their requirements groups; Figure 2B), although a large fraction of CA-CAM genes are for epithelial cell viability. expressed in the wing disc (81%). In contrast, only 32% of genes encoding proteins with Immunoglobulin domains are expressed Cell adhesion genes in this tissue (Figure 2B). Furthermore, only 17% of these genes The regulation of cell-to-cell adhesion is critical for the genera- result in lethality or a visible phenotype (Figure 2B). The more fre- tion of tissue shape and the maintenance of epithelial integrity, quent phenotypes we observed in genes of these groups (cell ad- and in the imaginal epithelium plays key roles during signaling, hesion molecules and immunoglobulin-containing proteins) were cell division, apoptosis, and pupal differentiation (Gumbiner pupal lethality, wing size defects, altered vein differentiation, and 1996; Bökel and Brown 2002). We included in CA all genes encod- formation of wing blisters (Figure 2, C and E). The expression of ing cytoplasmic or membrane-associated proteins related to cell- cell adhesion genes occurs in a ubiquitous manner, but many to-cell adhesion and to cell to extracellular matrix adhesion. cases of increased expression in the interveins or the dorsal or Based on their structural and/or functional characteristics, we ventral wing surfaces are also observed (Figure 2D). In summary, further defined four large groups: adherens/septate junctions we find enrichment of two phenotypes for genes belonging to the (CA-AJ/SJ; 38 genes; Figure 2A), extracellular matrix (CA-ECM; 62 CA class: lethality in the case of CA-AJ/SJ and blisters in the case genes; Figure 2A), proteins containing immunoglobulin domains of CA-ECM, indicating requirements for cell viability and for the (CA-IG; 79 genes; Figure 2A), and other cell adhesion and attachment of the dorsal and ventral wing surfaces, respectively. adhesion-related molecules (CA-CAM; 98 genes; Figure 2A). The For the most part, the function of cell adhesion molecules con- group with a higher percentage of genes required for wing forma- taining immunoglobulin domains is dispensable during wing de- tion is those related to the formation of adherens and septate velopment. junctions (71% genes; Figure 2B). These genes are also generally expressed in the wing disc (97%; Figure 2B) and the phenotype of Protein biology their knockdown includes mostly larval or pupal lethality with a We include in this class a large number of genes (1689) encoding loss of wing tissue and defects in wing cuticle differentiation enzymes which activity results in post-translational modifica- (Figure 2, C and E; Supplementary Table S1.2). Genes encoding tions to other proteins (Supplementary Table S1.3). Most of these proteins related to the formation of the cell matrix or proteins re- proteins are poorly characterized, and, because they could affect quired for interactions with the cell matrix are also expressed in multiple targets, they contribute to a variety of cellular processes. the wing disc (69%; Figure 2B) and the phenotype more often We have considered as members of the Protein biology class the found after knockdown consisted in the formation of blisters in following groups: Peptidases (PRO-PEP; 642 genes; Figure 3A), the wing cuticle (“WA”; Figure 2, C and E; Supplementary Table Ubiquitin ligases and transferases (PRO-UBIT; 270 genes; S1.2). These blisters results from failures in the adhesion of the Figure 4A), Chaperones involved in protein folding (PRO-CHAP; dorsal and ventral wing surfaces during pupal development, once 134 genes; Figure 3A), proteins involved in glucids modifications the two wing surfaces are confronted to each other through their (PRO-GLU; 148 genes; Figure 3A), Kinases and Phosphatases that basal cell membranes (Fristrom et al. 1993; Bilousov et al. 2014). could not be easily ascribed to other functional classes based on The blister could affect the entire wing, giving the wing the their function (PRO-KIN and PRO-PHO; 137 and 109 genes, 6 | G3, 2021, Vol. 11, No. 12 Figure 2 Cell adhesion class. (A) Distribution of the 277 genes included in the Cell adhesion class into the groups Adherent and septate junctions (AJ/SJ), Cell adhesion molecules excluding Immunoglobulin domain-containing proteins (CAM), Extracellular matrix adhesion (ECM), and Cell adhesion molecules containing Immunoglobulin domains (IG). (B) Percentages of genes for which we tested its knockdown phenotype (Done; dark gray column), of genes expressed in the wing disc (Exp; light gray column), and of genes with a lethal or visible phenotype in knockdown conditions (Phe; striped column) for the classes AJ/SJ, CA, ECM, and IG. (C) Frequency of lethal (L) and visible mutant phenotypes observed in the AJ/SJ, CA, ECM, and IG groups. Notice the high frequency of WA phenotypes in the ECM group compared to other groups. (D) Examples of expression patterns in the wing disc for different members of the CA class (name of each gene in the bottom right corner of each picture). The genes CG18657 and CG42677 show restricted expression to the ventral compartment and to the interveins, respectively. In situ hybridizations pictures correspond to experiments published as supplementary material in Molnar et al. (2012), Organista et al. (2015), and Hevia et al. (2017). (E) Wing phenotypes of representative examples of knockdowns of CA genes belonging to the AJ/SJ, CA, and ECM groups. All genotypes are from UAS-Dicer2/þ; nub-Gal4/UAS-RNAi flies. The name of each gene is given in the bottom-left corner of each wing picture. respectively; Figure 3A), genes encoding proteins involved in biology class; shown by group in Figure 3B), and is only higher other protein modifications such as palmitoylation, geranylation, than the average for this class or for the genome in the case of neddylation, arginylation, polyglycylation, sumoylation, and lip- components of the proteasome and protein folding groups (57% oylation, among others (PRO-MOD; 170 genes; Figure 3A) and and 41%; Supplementary Table S1.3, Figure 3, B and C, and components of the proteasome (PRO-PTS; 79 genes; Figure 3A). Supplementary Figure S1B). The fraction of genes with a mutant The Kinase and phosphatase groups are treated separately in the phenotype in knockdown conditions is particularly low in the accompanying manuscript (Ostalé et al. 2021). All other groups case of proteases (21%; Figure 3B) and protein kinases and phos- within the protein biology class behave in a similar manner. phatases (23% and 24%, respectively), possibly due to gene redun- Thus, a significant fraction of these genes (60%) are expressed in dancy or because these proteins are involved in biological the wing disc, ranging from 42% (Protein Peptidases; Figure 3B) to processes that are not critical for imaginal development. There 79% (components of the proteasome; Figure 3B). The fraction of are no clear enrichments in specific mutant phenotypes for any genes whose knockdown in the wing disc results in lethality or in group within the class (Figure 3, C and E). The main exceptions a mutant phenotype is low (29% for all members of the protein are lethality, which is more frequent in the proteasome and A. López-Varea et al. | 7 Figure 3 Protein class. (A) Distribution of the 1689 genes included in the Protein class into the groups peptidases (PEP), Ubiquitin ligases and transferases (UBIT), post-transcriptional protein modifications other than sugar, kinases, and phosphatases and ubiquitinylases (MOD), protein modifications by sugars (GLU), kinases (KIN), chaperons (CHAP), phosphatases (PHO), and components of the proteasome (PTS). (B) Percentages of genes for which we tested its knockdown phenotype (Done; dark gray column), of genes expressed in the wing disc (Exp; light gray column), and of genes with a lethal or visible phenotype in knockdown conditions (Phe; striped column) for the classes PEP, UBIT, MOD, GLU, KIN, CHAP, PHO, and PTS. (C) Frequency of lethal (L) and visible mutant phenotypes (colored columns) observed in the classes PEP, UBIT, MOD, GLU, KIN, CHAP, PHO, and PTS groups. Note the high frequency of lethal combinations in the CHAP and PTS classes and of wing size phenotypes in the KIN class compared to other groups. (D) Examples of expression patterns in the wing disc for different members of the Protein class (name of each gene in the bottom-left corner of each picture). The genes are ordered from not expressed in the wing disc (left) to ubiquitous and patterned expression (right). In situ hybridizations pictures correspond to experiments published as supplementary material in Molnar et al. (2012), Organista et al. (2015), and Hevia et al. (2017). (E) Representative wings of knockdowns for genes of the MOD (up), PEP, UBIT, GLU, and CHAP groups belonging to the Protein class. All genotypes are from UAS-Dicer2/þ; nub-Gal4/ UAS-RNAi flies. The gene names are displayed in the bottom-left corner of each wing picture. chaperon groups (57% and 42%; Figure 3, C and E and The expression of these genes visualized by in situ hybridiza- Supplementary Figure S1B), and wing adhesion phenotypes, tion includes not only multiple examples of ubiquitous expres- which are more frequent in the kinase group compared to the sion (Figure 3D) but also cases of restricted expression related other groups (Figure 3, C and E and Supplementary Figure S1B). to the development of vein and intervein territories (Figure 3D). 8 | G3, 2021, Vol. 11, No. 12 Figure 4 RNA class. (A) Distribution of the 851 genes included in the RNA class into the groups “Splicing” (SP), RNA binding (BND), mRNA processing (MOD), tRNA (tRNA), Translation initiation factors (TNLF), ribosomal RNA (rRNA), RNA enzymes (ENZ), and RNA molecules distinct to tRNA, mRNA, and rRNA (OTH). (B) Percentages of genes for which we tested its knockdown phenotype (Done; dark gray column), of genes expressed in the wing disc (Exp; light gray column), and of genes with a lethal or visible phenotype in knockdown conditions (Phe; striped column) for the groups SP, BND, MOD, tRNA, TNLF, rRNA, ENZ, and OTH. (C) Frequency of lethal (L) and visible mutant phenotypes observed for the groups SP, BND, MOD, tRNA, TNLF, rRNA, ENZ, and OTH. Note the high frequency of lethal combinations in the SP, tRNA, TNLF, ENZ, and rRNA groups and of wing size phenotypes in the BND, MOD, and TNLF groups compared to other groups. (D) Examples of expression patterns in the wing disc for different members of the RNA class (name of each gene in the bottom-left corner of each picture). Most genes are expressed in a ubiquitous manner in the wing disc. In situ hybridizations pictures correspond to experiments published as supplementary material in Molnar et al. (2012), Organista et al. (2015), and Hevia et al. (2017). (E) Representative wings of knockdowns of RNA genes belonging, from top to bottom, to the tRNA (up), MOD, ENZ, BND, rRNA, TNLF, and SP (bottom) groups. Genotypes for CG3689, CG9218, CG11342, CG8435, CG4152, and CG1796 genes are from UAS-Dicer2/þ; salEPv-Gal4/UAS-RNAi flies. The remaining are from UAS- Dicer2/þ; nub-Gal4/UAS-RNAi flies. The name of each particular gene is given in the bottom-left corner of each wing picture. In summary, only genes encoding components of the protea- RNA Biology some and genes encoding chaperons show a general require- We classified 851 genes as being fundamentally related to the ment for wing development, resulting in loss-of-function synthesis, processing, and modification of RNA molecules conditions in phenotypes compatible with reduced cell viabil- (Figure 4A; Supplementary Table S1.4). We further divided this ity. class into eight groups: genes involved in the synthesis and A. López-Varea et al. | 9 Figure 5 DNA class. (A) Distribution of the 1598 genes included in the DNA class into the groups “Chromosomal maintenance and structure” (CHRO), Chromatin modifying proteins and protein complexes (CMC), DNA modifying enzymes (ENZ), DNA replication (REP), General transcription factors (GTF), and sequence-specific transcription factors (TF). (B) Percentages of genes within each class for which we tested its knockdown phenotype (Done; dark gray column), of genes expressed in the wing disc (Exp; light gray column), and of genes with a lethal or visible phenotype in knockdown conditions (Phe; striped column). (C) Frequency of lethal (L) and visible mutant phenotypes observed in the classes CHRO, CMC, ENZ, REP, GTF, and TF. Notice the high frequency of lethal combinations in the GFT class, of wing veins effects in the CMC and ENZ classes. The increase in other phenotypes (OTH) observed in the REP class corresponds to cell differentiation phenotypes. (D) Examples of expression patterns in the wing disc for different members of the DNA class (name of each gene in the bottom-left corner of each picture). Most genes are expressed in a ubiquitous manner in the wing disc. In situ hybridizations pictures correspond to experiments published as supplementary material in Molnar et al. (2012), Organista et al. (2015), and Hevia et al. (2017). (E) Representative wings of knockdowns of DNA genes belonging, from top to bottom, to the CHRO (up), CMC, ENZ, REP, and GTF (bottom) classes. All genotypes except His1:CG31617, Cap-G, His3:CG31613, and dmt are from UAS-Dicer2/þ; nub-Gal4/UAS-RNAi flies. His1:CG31617, Cap-G, His3:CG31613, and dmt are from UAS-Dicer2/þ; salEPv-Gal4/UAS-RNAi flies. The name of each gene is given in the bottom-left corner of each wing picture. maturation of ribosomal RNA (RNA-rRNA, 58 genes), genes re- of other RNA molecules such as piRNA, snRNA, siRNA, and lated to the synthesis and modification of tRNAs (RNA-tRNA; 118 snoRNA (RNA-OTH, 50 genes). genes), general translation initiation, elongation, and termination The RNA class shares several characteristics with the factors (RNA-TNLF, 83 genes), genes encoding members of the Ribosome class. Thus, RNA genes are generally expressed in the mRNA splicing machinery (RNA-SP; 244 genes), genes encoding wing disc (96%; Figure 4B), and they are mostly expressed at high enzymes catalyzing mRNA modifications (RNA-MOD; 99 genes), levels and in a ubiquitous manner (Figure 4D). They also include RNA enzymes (RNA-ENZ; 58 genes), genes encoding RNA-binding a large fraction of genes required for imaginal development (56% proteins (RNA-BND; 141 genes), and genes related to the biology aggregate; shown by group in Figure 5B). The fraction of genes 10 | G3, 2021, Vol. 11, No. 12 with a lethal or mutant phenotype is maximal for RNA-rRNA Figure 5, C and E) and defects in vein differentiation in the DNA- encoding genes (83%) and minimal for genes classified as RNA- CMC group, DNA-ENZ and DNA-TF groups (18%, 17%, and 16%, OTH (33%). The most frequently found phenotypes were lethality respectively; Figure 5, C and E). All other phenotypes appear with (48%) and wing size defects (16%) (Figure 4, C and E). The fre- similar frequencies in all different groups, varying from 1.6% to quency of other phenotypes was low in all groups (Figure 4C), al- 19% (Figure 5C). Most expression patterns observed for genes of though the RNA-tRNA class shows enrichment in the WA and the DNA class, excluding sequence-specific transcription factors, WD phenotypes (Supplementary Figure S1B). In summary, loss- were ubiquitous in the wing disc (Figure 5D). of-function conditions of genes included in the RNA class, mostly in the splicing, translation, and ribosomal RNA groups (RNA-SP, Cell signaling RNA-TNLF, and RNA-rRNA) share a large frequency of pheno- In this class, we included all components of the Ecdysone recep- types related to reduced cell viability (larval and pupal lethality, tor (SIG-EC; 20 genes), Juvenile hormone (SIG-JH; 6 genes), loss of wing tissue and strong defects in wing size and pattern) Hedgehog (SIG-HH; 19 genes), Hippo/Salvador/Warts (SIG-HSW; and reduced cell proliferation (reduced wing size). As expected 27 genes), Bone morphogenetic protein and Transforming growth these genes are mostly expressed ubiquitously and at high levels factor b (SIG-BMP/TGF; 39 genes), Insulin receptor and Inositol in the wing disc epithelium. signaling (SIG-INR/IPS; 72 genes), Jak-Stat (SIG-JAK; 15 genes), Jun Kinase (SIG-JNK; 28 genes), Notch (SIG-NOTCH; 46 genes), Toll DNA biology (SIG-TOLL; 39 genes), Wingless (SIG-WNT; 32 genes), and We classified 1598 genes as being involved in DNA biology Receptor tyrosine kinase (SIG-RTK; 84 genes) pathways (Figure 5A; Supplementary Table S1.5). This class was further di- (Figure 6A; Supplementary Table S1.6). We also include all Neural vided into six large groups including sequence-specific transcrip- peptides (SIG-NP; 45 genes), Odorant-binding proteins and gusta- tion factors (DNA-TF, 768 genes), general transcription factors tory receptors (SIG-OBP and SIG-GR, 128 and 60 genes, respec- (DNA-GTF, 168 genes), DNA replication (DNA-REP, 62 genes), tively), and a large collection of genes encoding proteins with the Chromatin modifying complexes (DNA-CMC, 246 genes), typical structure of G-protein coupled receptors (SIG-GPCR; 193 enzymes affecting DNA conformation and maintenance (DNA- genes). A small number of genes (26) encoding molecules related ENZ, 157 genes), and chromosomal structure and maintenance to TNF, Robo, Nitric oxide, Semaphorin, NFAT, and Ephrin signal- proteins (DNA-CHRO, 197 genes). For all these groups, we find ing were classified as “SIG-OTH” (Supplementary Table S1.6; similar frequencies of expressed genes (average 76%; Figure 5B) Figure 6A). and similar frequency of knockdowns with lethality or a mutant Annotated members of the canonical signaling pathways are phenotype (average 41%; Figure 5B). The only exception is the generally expressed in the wing disc, ranging from 80% for com- chromosomal genes class, which has much lower values of ponents of the Toll signaling pathway to 100% for members of expressed genes (47%) and genes with a mutant phenotype (18%, the Hippo pathway (Figure 6B). All these pathways have been Figure 5B). This last group includes all histone coding genes (114 thoroughly characterized in their functions during wing imaginal genes), which are present in multiple copies grouped in gene clus- development in the regulation of wing growth, vein patterning ters. In general, only one (His2A 1/20; His2B 1/23; His3 1/23 and and differentiation, and dorsoventral boundary formation His4 1/22) or two genes (His1 2/23) encoding each Histone type are (Molnar et al. 2011). As these signaling pathways are modules expressed in the wing disc. For these genes (His1:CG31617, formed by proteins linked by protein–protein interactions result- His2A:CG31618, His2B:CG17949, His3:CG31613, and His4:CG31611), ing in unique outputs, it is expected that knockdown of any com- the knockdown results in larval (His1:CG31617 and ponent of the pathway results in similar phenotypes varying only His2B:CG17949) or pupal (His2A:CG31618) lethality, as well as in their strength. For example, the knockdown of most members strong effects in wing size (His1:CG31617, His2A:CG31618, and of the BMP and TGFb pathways results in wings reduced in size His3:CG31613). The knockdowns of other genes encoding Histone and accompanied with loss and extra-vein phenotypes in the variants (His3.3A and His2Av) reduce wing size (Figure 5E). When case of BMP components (Figure 6, C and D). These phenotypes we exclude all Histone coding genes not expressed in the wing reveal the known functions of these pathways in promoting wing disc from the Chromosomal group, the overall expression and growth (BMP and TGF), vein patterning (BMP), and vein differenti- phenotype parameters are similar to those of other groups in the ation (BMP) (de Celis 1997; Brummel et al. 1999). The function of DNA class (not shown). the Hh signaling pathway is required to regulate the expression The larger group within the DNA class corresponds to tran- of the BMP ligand dpp and other genes in a central region of the scription factors and proteins with sequence-specific DNA-bind- wing disc abutting the anteroposterior compartment boundary ing domains that could contribute to the regulation of gene (Tabata and Kornberg 1994; Zecca et al. 1995). As expected, knock- expression (DNA-TF; 768 genes). Transcription factors play key down of Hh components results in the formation of smaller than roles during wing imaginal development, contributing to the sub- normal wings that lose some structures located between the division of the wing disc in domains of gene expression corre- veins L3 and L4 (Figure 6D). sponding to the future anatomical regions of the wing and thorax Members of the Notch pathway also share a similar set of phe- (Ostalé et al. 2018). For this reason, and because of the large num- notypes, consisting in vein thickening, loss of wing margin struc- ber of TF expressed in the disc, they will be analyzed separately tures, and reduced wing size (Figure 6, C and D and (Ostale et al. in preparation). The most frequent phenotypes we Supplementary Figure S1B). All these phenotypes are reminiscent found within the DNA class are pupal lethality, which is particu- of Notch loss-of-function alleles (de Celis and Garcı́a-Bellido larly prominent within the DNA-GTF (44%; Figure 5C), defects in 1994). Knockdown of components of the Epidermal growth factor wing size and pattern, including those cases in which the wing receptor pathway (RTK) results as expected in two opposite phe- fails to form in the DNA-GTF and DNA-REP groups (27% and 26%, notypes regarding vein formation. Thus, the knockdowns of respectively; Figure 5, C and E), reductions of wing size with mi- genes promoting ERK phosphorylation (EGFR, drk, Ras85D, dos, rol, nor or no effect on vein patterning in the DNA-CMC, DNA-REP Raf, csw, ksr, and cnk) cause loss of veins and reduced wing size, groups, and DNA-ENZ groups (26%, 23%, and 22%, respectively; and in strong conditions lead to the total absence of the wing A. López-Varea et al. | 11 Figure 6 Signaling class. (A) Distribution of the 880 genes included in the Signaling class into the groups Gustatory receptors, Neural peptides, and odorant-binding proteins (GR/NP/OBP), G-Protein coupled receptors (GPCR), Receptor tyrosine kinase pathways (RTK), Insulin receptor and Inositol phosphate signaling pathways (INR), Notch signaling pathway (NOTCH), BMP and TGFb signaling pathways (BMP/TGF), Toll receptors signaling pathway (TOLL), Wnt signaling pathway (WNT), Hippo/Salvador/Warts signaling pathways (HIPPO), Jun Kinase signaling pathway (JNK), ecdysone and Juvenile hormone signaling (EC/JH), Hedgehog signaling (HH), JAK/STAT signaling (JAK), and other signaling molecules (OTH). (B) Percentages of genes for which we tested its knockdown phenotype (Done; dark gray column), of genes expressed in the wing disc (Exp; light gray column), and of genes with a lethal or visible phenotype in knockdown conditions (Phe; striped column) for the groups shown in (A). (C) Frequency of lethal (L) and visible mutant phenotypes observed in the classes RTK, INR/IPS, NOTCH, BMP/TGF, TOLL, WNT, HSW, JNK, HH, and JAK. Note the high frequency of size defects in the INR/IPS and HIPPO classes, of vein defects in the RTK and BMP/TGF classes and of loss of wing tissue or strong defects in wing size and pattern in the HH and JNK classes. The increase in other phenotypes (OTH) observed in the NOTCH class corresponds to wing margin defects. (D) Representative wings of knockdowns Signaling class genes belonging, from top to bottom, to the RTK, INR/IPS, NOTCH, BMP, TGF, HSW, and HH classes. Wings corresponding to drk, Ras85D, dos, rl, csw, ksr, and cnk are from UAS-Dicer2/þ; salEPv-Gal4/UAS-RNAi flies and the rest from UAS-Dicer2/þ; nub-Gal4/UAS-RNAi flies. The name of each particular gene is given in the bottom-left corner of each wing picture. (Figure 6D). These phenotypes are compatible with the known et al. 2012), result in the formation of ectopic veins (Figure 6D). function of EGFR signaling promoting vein differentiation and Knockdown of different members of the InR signaling pathway wing growth (Sturtevant et al. 1993; Simcox 1997). On the other affect cell size and wing size, without any effect on the pattern of hand, knockdown of pathway members that antagonize ERK veins (Figure 6, C and D). These phenotypes are compatible with phosphorylation [tai and MKP3; see Molnar and de Celis (2013) the known function of InR signaling to promote cell growth in and Ruiz-Gómez et al. (2005), respectively] or that repress gene ex- imaginal cells (Edgar 2006). The Hippo pathway is required to pro- pression in the absence of pathway activation (cic see Jiménez mote cell proliferation in all imaginal discs (Halder and Johnson 12 | G3, 2021, Vol. 11, No. 12 2011). Most components of the pathway act to promote the phos- of most member of this pathway occurs at high levels in the wing phorylation of the transcriptional coactivator Yorkie, causing its disc (Figure 6B). inactivation by sequestering this protein in the cytoplasm Other genes we classified as members of the signaling class (Halder and Johnson 2011). Accordingly, knockdown of Yorkie up- were GPCR (193 genes) and gustatory receptors, neural peptides, stream components results in the formation of larger and and odorant-binding proteins (233 genes). This is a very numer- rounder wings missing the cross veins (Figure 6D). The expression ous group (426 genes; Supplementary Table S1.6), but only a Figure 7 Metabolism class. (A) Distribution of the 1631 genes included in the Metabolism class into the groups Lipid metabolism (LIP), amino acid, nucleotides, and other small molecules metabolism (AA/NT/O), Detoxifying enzymes (DTOX), Mitochondrial biology and mitochondrial respiratory chain (MIT), Sugars metabolism (GLY), Oxido-reduction metabolism (RDX), and other metabolic processes including Central metabolism, autophagy, one-carbon metabolism, metal metabolism, and xenobiotic metabolism (OTH). (B) Percentages of genes within each group for which we tested the knockdown phenotype (Done; gray column), of genes expressed in the wing disc (Exp; light gray column), and of genes with a lethal or visible phenotype in knockdown conditions (Phe; striped column). (C) Frequency of lethal (L) and visible mutant phenotypes observed in the groups LIP, AA/NT/O, DTOX, MIT, GLY, OTH, and RDX. (D) Examples of expression patterns in the wing disc for different members of the Metabolism class (name of each gene in the bottom-left corner of each picture). Most genes are expressed in a ubiquitous manner in the wing disc. In situ hybridizations pictures correspond to experiments published as supplementary material in Molnar et al. (2012), Organista et al. (2015), and Hevia et al. (2017). (E) Representative wings of knockdowns of Metabolism genes belonging, from top to bottom, to the AA/NT (top), DTOX, GLY, and LIP (bottom) classes. Wings corresponding to Lsd- 2, Gbs-70E, and Hmgcr are from UAS-Dicer2/þ; salEPv-Gal4/UAS-RNAi flies and the remaining from UAS-Dicer2/þ; nub-Gal4/UAS-RNAi flies. The name of each gene is given in the bottom-left corner of each wing picture. A. López-Varea et al. | 13 small fraction of these genes are expressed in the wing disc (37% pathway. The phenotypes for components of each pathway vary for SIG-GPCR and 16% for SIG-GR, SIG-NP, and SIG-OBP; mostly in strength and are very much compatible with the known Figure 6B). The number of knockdowns resulting in lethality or a requirements of these pathways during wing development. mutant phenotype was significantly lower than the average for Perhaps not surprisingly, the majority of GPCR receptors, neural the signaling class (15% and 23% vs 36%). We also noticed that peptides, and odorant-binding proteins are not required for wing the majority (14/32 for SIG-GPCR, 10/10 for SIG-GR, 1/2 SIG-NP, development. and 13/15 for SIG-OBP) of knockdowns resulting in a mutant phe- notype correspond to genes that were classified as not expressed Metabolism in the wing disc, suggesting that these phenotypes may corre- In “Metabolism” we include all proteins involved in the synthesis spond to off-target effects. In contrast, the number of possible and degradation of lipids, sugars, nucleotides, and aminoacids off-target effects for members of the canonical signaling path- (anabolism and catabolism) and also proteins involved in detoxi- ways was very low (14/176). In summary, we find a set of coher- fication, xenobiotic processing, autophagy, and maintenance of ent phenotypes for the different components of each signaling the RedOx balance. We also included in this class all genes Figure 8 Cuticle class. (A) Distribution of the 387 genes included in the “Cuticle” class into the groups “Cuticular Protein Families” (CPF), “Chitin binding domain containing proteins” (CBDCP), Mucins, Glue proteins, Chitin deacetylase-like family and yellow family (MU/Y), and Chorion and vitelline membrane proteins (CO/VM). (B) Percentages of genes within each group for which we tested the knockdown phenotype (Done; dark gray column), of genes expressed in the wing disc (Exp; light gray column), and of genes with a lethal or visible phenotype in knockdown conditions (Phe; striped column). (C) Frequency of lethal (L) and visible mutant phenotypes observed in the CPF, CBDCP, Mu/Y, and CO/VM groups. Note the high frequency of “Size” phenotypes in the CPF group. (D) Examples of expression patterns in the wing disc for different members of the Cuticle class (name of each gene in the bottom-left corner of each picture). In situ hybridizations pictures correspond to experiments published as supplementary material in Molnar et al. (2012), Organista et al. (2015), and Hevia et al. (2017). (E) Representative wings of knockdowns of Cuticle genes belonging, from top to bottom, to the CBDCP, CO/VM, and CPF groups. All wings correspond to UAS-Dicer2/þ; nub-Gal4/UAS-RNAi flies. The name of each gene is given in the bottom-left corner of each wing picture. 14 | G3, 2021, Vol. 11, No. 12 related to mitochondrial biology. The metabolism class is large into four additional groups: “Autophagy,” including peptidases re- (1631 genes; Figure 7A and Supplementary Table S1.7) and was lated with autophagy and proteins related to autophagosome as- subdivided into several groups: “Amino acids, Nucleotides and sembly (MET-ATG; 27 genes); “One-carbon metabolism” (MET- other related molecules” (MET-AA/NT/O; 278 genes), including OC; 13 genes); “Metal metabolism” (MET-MTA; 35 genes); and the biosynthesis of amino acids and molecules derived from “Xenobiotic metabolism” (MET-XEN, 42 genes). amino acids, biosynthesis of purine and pyrimidine nucleotides Within the MET class, we found expression in the wing imagi- and molecules derived from nucleotides and Nitrogen metabo- nal disc in the range of 55–82% of genes (Figure 7B), and a low lism; Oxido-reduction metabolism (MET-RDX; 94 genes); percentage of genes resulting in a mutant phenotype in knock- “Carbohydrates and glycobiology” (MET-GLY; 200 genes), includ- down conditions (28% in average), with the exception of genes in- ing disaccharides and polysaccharides digestion, cellular trans- volved in mitochondrial biology, for which this percentage raises port of monosaccharides, Glycolysis and gluconeogenesis to 54% (Figure 7B). Mutant phenotypes were similarly distributed pathways, Pyruvate degradation, pentose phosphate pathway, in all different subgroups (Figure 7C), with a greater prevalence of and Glycoconjugates metabolism; Lipid metabolism (MET-LIP, pupal lethal phenotypes in the mitochondrial biology group 489 genes) including Lipid transport, fatty acids biosynthesis, tri- (Figure 7C). For members of the different groups included in the acyclglycerols metabolism, membrane phospholipids metabo- Metabolism class, we found a range of visible phenotypes that lism and cholesterol, steroid and isoprenoids metabolism; were also observed in other molecular classes (Figure 7E). In “Detoxification” (MET-DTOX; 208 genes), including Cytochromes some cases, these phenotypes were reminiscent of Notch signal- involved in oxide-reduction reactions, oxidative stress and detox- ing insufficiency (Lipid class; Figure 7E). In general, the expres- ification of endogenous and xenobiotic compounds, central me- sion of these genes visualized by in situ hybridization was tabolism, including tricarboxylic acid cycle (MET-CTR; 38 genes), ubiquitous in the wing disc (Figure 7D). and “Mitochondrial biology” (MET-MIT; 207 genes), including re- spiratory chain complexes components, electron carriers, and Cuticle other mitochondrial proteins (Supplementary Table S1.7; We include all structural and related constituents of external Figure 7A). The remaining 155 genes (“MET-OTH”) were classified membranes such as the adult cuticle, eggshell, and gut Figure 9 Expression analysis of the CG and CGh classes. (A) Percentage in the genome of members of the functional classes CG (CG, green sector, 2084 genes) and CGh (CGh, blue sector, 1675 genes). (B) Percentage of CG (green columns) and CGh (blue columns) genes analyzed in knockdown conditions (Done), expressed in the wing disc (Exp), and resulting in lethality or a mutant phenotype (Phe). The same values are indicated as a reference for the entire genome (gray columns). (C) Percentage of CG (CG, green column, 44 genes) and CGh genes (CGh, blue columns, 58 genes), and all functional classes grouped together (Genome, gray columns; 562 genes) expressed in a ubiquitous manner (GEN), not expressed (NE), and expressed in a spatial pattern (PAT). (D) Examples of expression patterns in the wing disc for different members of the CG and CGh classes. The name of each gene is indicated in the bottom-left corner of each picture. In situ hybridizations pictures correspond to experiments published as supplementary material in Molnar et al. (2012), Organista et al. (2015), and Hevia et al. (2017). A. López-Varea et al. | 15 peritrophic matrix (Figure 8). The “Cuticle” group includes 387 ubiquitous expression and also cases of patterned expression re- genes of which only 30% are expressed in the wing disc (Figure 8). lated to the developing wing veins (Figure 8D). Most visible phe- This group was further divided into four groups (Figure 8A): notypes correspond to genes belonging to the CUT-CPF group “Cuticular Protein Families,” which are the structural proteins of (59% of described phenotypes; examples in Figure 8E), which the adult cuticle (CUT-CPF; 183 genes), “Chitin binding domain were very prominent in phenotypes affecting the size of the wing containing proteins” (CUT-CBDCP; 116 genes), Chorion and vitel- (Figure 8, C and E). The groups CUT-CBDCP and CUT-MU/CUT-Y line membrane proteins (CUT-CO and CUT-VM; 39 genes), and a exhibit wing differentiation phenotypes with frequencies higher fourth group including the Mucins, Glue proteins, and yellow than other cuticle genes (Figure 8C and Supplementary Figure family (CUT-MU and CUT-Y; 49 genes). As expected, these groups S1B). have low frequencies of genes expressed in the wing disc (from 18% for CUT-COP/CUT-VM to 45% for CUT-MU/CUT-Y genes; CG and CGh Figure 8B), and low frequencies of genes with a knockdown caus- The CG class, containing genes for which there is a complete lack ing lethality (from 9% for CUT-CBDCP to 13% for CUT-MU/CUT-Y of functional information, and the CGh class including genes genes; Figure 8, B and C) or a mutant phenotype (from 10% for with at least one InterPro (IP) domain, are among the most abun- CUT-MU and CUT-Y to 26% for CUT-CPF genes; Figure 8, B and C). dant in the Genome, including 2084 (CG) and 1675 (CGh) genes For genes of this class expressed in the wing disc, we found (Supplementary Table S1). The most frequently found conserved Figure 10 Phenotypic analysis of the CG and CGh classes. (A) Frequency of knockdown phenotypes for genes belonging to the CG class (green columns) and CGh class (blue columns) compared to the genome (gray columns). Lethal combinations (lethal), lack of wing (nW) defects in wing size and pattern (S-P), defects in wing size (S), differentiation of ectopic or thicker veins (Vþ), loss of veins (V), appearance of wing blisters (WA), wing differentiation defects (WD), loss of wing margin structures (WM), alterations in wing shape (WS), defects in cuticle pigmentation (WP), appearance of extra bristles in the wing surface or loss of bristles in the wing margin (Q) and defects in trichome differentiation and cell size (CD). (B) Representative wings of knockdowns of CG (upper rows, green code) and CGh (lower rows, blue code) belonging, from left to right, to the phenotypic classes nW/S-P, S, V (Vþ and V), wing adhesion (WA) and wing margin (WM). Wings corresponding to CG8675 and dy are from UAS-Dicer2/þ; salEPv-Gal4/UAS-RNAi flies, the remaining wings from UAS-Dicer2/þ; nub-Gal4/UAS-RNAi flies. The name of each particular gene is given in the bottom-left corner of each wing picture. 16 | G3, 2021, Vol. 11, No. 12 domains in the CGh class, found in more than 20 genes, were and Santander Foundations. The authors affirm that all data nec- “Protein of unknown function DUF1091” (IPR010512; 91 genes), essary for confirming the conclusions of the article are present “Leucine-rich repeat” (IPR001611; 35 genes), “Protein of unknown within the article, figures, tables, and supplementary informa- function DM4/12” (IPR006631; 30 genes), “Haemolymph juvenile tion. hormone binding” (IPR010562; 26 genes), “EF-hand domain pair” (IPR011992; 23 genes), and “Alkaline-phosphatase-like, Core do- main superfamily” (IPR017850; 22 genes). The CG and CGh genes Funding include 15% and 12% of the fly genome (Figure 9A), but only a This research was supported by Secretarı́a de Estado de small fraction of CG genes (30%) were considered as expressed in Investigación, Desarrollo e Innovación, Grant/Award Number the wing disc (Figure 9B). For those genes that were expressed, PGC2018-094476-B-I00. The founders have no role in the research the more prevalent type of spatial pattern visualized by in situ hy- design, execution, analysis, interpretation, and reporting. bridization was ubiquitous (Figure 9, C and D), but many instan- ces of genes expressed in restricted patterns were also identified (Figure 9, C and D). The frequency of different phenotypes we Conflicts of interest found in the CG and CGh classes were similar to each other The authors declare that there is no conflict of interest. (Figure 10A). For most phenotypes, we found a lower frequency compared with the genome, the only remarkable exception being defects in dorsoventral adhesion (WA) (Figure 10, A and B). Many Literature cited phenotypes affecting the formation of the veins and wing margin, Adams MD, Celniker SE, Holt RA, Evans CA, Gocayne JD, et al. 2000. as well as reductions in the size of the wing (Figure 10B), suggest The genome sequence of Drosophila melanogaster. Science. 287: that numerous Drosophila genes without clear orthologs in other 2185–2195. noninsect species play important roles during wing development. Ashburner M, Ball CA, Blake JA, Botstein D, Butler H, et al. 2000. Gene ontology: tool for the unification of biology. Nat Genet. 25:25–29. Concluding remarks Beira JV, Paro R. 2016. The legacy of Drosophila imaginal discs. Chromosome. 125:573–592. We present here a simplified classification of Drosophila protein- Bejarano F, Luque CM, Herranz H, Sorrosal G, Rafel N, et al. 2008. A coding genes that aims to contain in only two terms the main fea- gain-of-function suppressor screen for genes involved in dorsal–- tures of each protein. We have used this classification to incorpo- ventral boundary formation in the Drosophila wing. Genetics. 178: rate data from mRNA expression and from the results of an RNAi 307–323. phenotypic screen in the wing. As expected, no functional class Bilousov O, Koval A, Keshelava A, Katanaev VL. 2014. Identification or group within a class could be subscribed to a particular pheno- of novel elements of the Drosophila blisterome sheds light on po- type. However, some phenotypes are enriched in particular func- tential pathological mechanisms of several human diseases. tional classes, including cell viability phenotypes in the PLoS One. 9:e101133. Ribosome, Protein transport and RNA classes, wing vein pattern- Bjorklund M, Taipale M, Varjosalo M, Saharinen J, Lahdenpera J, et al. ing in the Cell signaling class and failures in dorsoventral apposi- tion in the cell adhesion class (Supplementary Figure S1). We also 2006. Identification of pathways regulating cell size and present a phenotypic description for a large set of genes for which cell-cycle progression by RNAi. Nature. 439:1009–1013. there is very little functional information (CG and CGh classes). Bökel C, Brown NH. 2002. Integrins in development: moving on, Although these phenotypes are not predictive of molecular or bi- responding to, and sticking to the extracellular matrix. Dev Cell. ological function in the adult Drosophila wing, we expect that they 3:311–321. open new entry points to undertake the functional analysis of the Brummel T, Abdollah S, Haerry TE, Shimell MJ, Merriam J, et al. 1999. corresponding genes. The Drosophila activin receptor Baboon signals through dSmad2 and controls cell proliferation but not patterning during larval development. Genes Dev. 13:98–111. Data availability Calleja M, Moreno E, Pelaz S, Morata G. 1996. Visualization of gene The data underlying this article are available in the article and in expression in living adult Drosophila. Science. 274:252–255. its online supplementary material. All wing pictures we have Cruz C, Glavic A, Casado M, de Celis JF. 2009. A gain-of-function were submitted to the Figshare repository: https://doi.org/10. screen identifying genes required for growth and pattern forma- 6084/m9.figshare.16624645.v1; https://doi.org/10.6084/m9.fig tion of the Drosophila melanogaster wing. Genetics. 183:1005–1026. share.16624630.v1; https://doi.org/10.6084/m9.figshare.16624603. de Celis JF. 1997. Expression and function of decapentaplegic and thick v1; and https://doi.org/10.6084/m9.figshare.16624591.v1 veins during the differentiation of the veins in the Drosophila wing. Development. 124:1007–1018. de Celis JF, Garcı́a-Bellido A. 1994. Roles of the Notch gene in Acknowledgments Drosophila wing morphogenesis. Mech Dev. 46:109–122. We thank NIG-FLY, Bloomington Stock Center, and VDRC Stock Edgar BA. 2006. How flies get their size: genetics meets physiology. Center for providing the tools necessary for this work. This re- Nat Rev Genet. 7:907–916. search was supported by Secretarı́a de Estado de Investigación, Flegel K, Grushko O, Bolin K, Griggs E, Buttitta L. 2016. Roles for the Desarrollo e Innovación, Grant/Award Number PGC2018-094476- histone modifying and exchange complex NuA4 in cell cycle pro- B-I00. We thank two anonymous referees for their criticism. We gression in Drosophila melanogaster. Genetics. 203:1265–1281. would also like to acknowledge the support from the Drosophila Fristrom D, Wilcox M, Fristrom J. 1993. The distribution of PS integ- transgenesis and confocal microscopy CBMSO scientific services. rins, laminin A and F-actin during key stages in Drosophila wing The CBMSO enjoys institutional support from the Ramón Areces development. Development. 117:509–523. A. López-Varea et al. | 17 Gumbiner BM. 1996. Cell adhesion: the molecular basis of tissue ar- Genetics and Therapeutics. IntechOpen. 5:93–109. doi: chitecture and morphogenesis. Cell. 84:345–357. 10.5772/intechopen.72758. Halder G, Johnson RL. 2011. Hippo signaling: growth control and be- Ostalé C.M., Esteban N, López-Varea A and Celis J.F. 2021. yond. Development. 138:9–22. Functional requirements of protein kinases and phosphatases in Hevia CF, López-Varea A, Esteban N, de Celis JF. 2017. A search for the development of the Drosophila melanogaster wing. G3. doi: genes mediating the growth-promoting function of TGFb in the 10.1093/g3journal/jkab348 Drosophila melanogaster wing disc. Genetics. 206:231–249. Perkins LA, Holderbaum L, Tao R, Hu Y, Sopko R, et al. 2015. The Jiménez G, Shvartsman SY, Paroush Z. 2012. The Capicua repress- transgenic RNAi project at Harvard Medical School: resources or—a general sensor of RTK signaling in development and dis- and validation. Genetics. 201:843–852. ease. J Cell Sci. 125:1383–1391. Ruiz-Gómez A, López-Varea A, Molnar C, De La Calle-Mustienes E, Kaufman TC. 2017. A short history and description of Drosophila mela- Ruiz-Gómez M, et al. 2005. Conserved cross-interactions in nogaster classical genetics: chromosome aberrations, forward Drosophila and Xenopus between Ras/MAPK signaling and the genetic screens, and the nature of mutations. Genetics. 206: dual-specificity phosphatase MKP3. Dev Dyn. 232:695–708. 665–689. Schnorrer F, Schönbauer C, Langer CCH, Dietzl G, Novatchkova M, et López-Varea A, Ostalé CM, Vega-Cuesta P, Ruiz-Gómez A, Organista al. 2010. Systematic genetic analysis of muscle morphogenesis MF, et al. 2021. Genome-wide phenotypic RNAi screen in the and function in Drosophila. Nature. 464:287–291. Drosophila wing: global parameters. G3. doi:10.1093/g3journal/- Simcox A. 1997. Differential requirement for EGF-like ligands in jkab351. Drosophila wing development. Mech Dev. 62:41–50. Marygold SJ, Roote J, Reuter G, Lambertsson A, Ashburner M, et al. St Johnston D. 2002. The art and design of genetic screens: Drosophila 2007. The ribosomal protein genes and Minute loci of Drosophila melanogaster. Nat Rev Genet. 3:176–188. melanogaster. Genome Biol. 8:R216. Sturtevant MA, Roark M, Bier E. 1993. The Drosophila rhomboid gene Molnar C, Casado M, López-Varea A, Cruz C, de Celis JF. 2012. mediates the localized formation of wing veins and interacts ge- Genetic annotation of gain-of-function screens using RNA inter- netically with components of the EGF-R signaling pathway. ference and in situ hybridization of candidate genes in the Genes Dev. 7:961–973. Drosophila wing. Genetics. 192:741–752. Tabata T, Kornberg TB. 1994. Hedgehog is a signaling protein with a Molnar C, de Celis JF. 2013. Tay Bridge is a negative regulator of EGFR key role in patterning Drosophila imaginal discs. Cell. 76:89–102. signalling and interacts with Erk and Mkp3 in the Drosophila mela- Thurmond J, Goodman JL, Strelets VB, Attrill H, Gramates LS, et al.; nogaster wing. PLoS Genet. 9:e1003982. Molnar C, López-Varea A, Hernández R, de Celis JF. 2006. A the FlyBase Consortium. 2019. FlyBase 2.0: the next generation. gain-of-function screen identifying genes required for vein for- Nucleic Acids Res. 47:D759–D765. mation in the Drosophila melanogaster wing. Genetics. 174: Valakh V, Naylor SA, Berns DS, DiAntonio A. 2012. A large-scale 1635–1659. RNAi screen identifies functional classes of genes shaping synap- Molnar C, Resnik-Docampo MFM, Martin MFC, de Celis JF. 2011. tic development and maintenance. Dev Biol. 366:163–171. Signalling pathways in development and human disease: a Vogel C, Teichmann SA, Chothia C. 2003. The immunoglobulin su- Drosophila wing perspective. In: Dr. Dijana Plaseska-Karanfilska perfamily in Drosophila melanogaster and Caenorhabditis elegans (Ed.), Human Genetic Diseases. InTech. 1–39. and the evolution of complexity. Development. 130:6317–6328. Organista MF, Martı́n M, de Celis JM, Barrio R, López-Varea A, et al. Zartman JJ, Kanodia JS, Yakoby N, Schafer X, Watson C, et al. 2009. 2015. The spalt transcription factors generate the transcriptional Expression patterns of cadherin genes in Drosophila oogenesis. landscape of the Drosophila melanogaster wing pouch central re- Gene Expr Patterns. 9:31–36. gion. PLoS Genet. 11:e1005370. Zecca M, Basler K, Struhl G. 1995. Sequential organizing activities of Ostalé C, Ruiz-Gomez A, Vega P, Ruiz-Losada M, Estella C, et al. 2018. engrailed, hedgehog and decapentaplegic in the Drosophila wing. Drosophila imaginal discs as a playground for genetic analysis: Development. 121:2265–2278. concepts, techniques and expectations for biomedical research. In: Drosophila melanogaster—Model for Recent Advances in Communicating editor: B. Oliver