Drosophila in vivo
RNAi techniques have been used to screen both the whole genome and subsets of genes. Published in vivo
screens are listed in Table 2
. In addition, publications that employed the NIG RNAi library are summarized on the NIG website [55
]. These screens are categorized into three classes. Class 1 encompasses screens that address questions that can only be addressed using in vivo
studies, for example, questions about development, behavior, and immunity; Class 2 encompasses screens that address questions that can be answered by both in vitro
and in vivo
approaches; and Class 3 encompasses screens for genes that have been pre-identified by an in vitro
screen and that require further validation by in vivo
methods. In the following section, we briefly summarize in vivo
screens that have been performed using Drosophila
6.1. Class 1: Screens to Address Questions that Can only Be Investigated Using an In Vivo Approach
Cronin et al
. assayed 10,689 different genes (78% of the Drosophila
genome) that affect susceptibility to intestinal Serratia marcescens
]. Of these, 8.3% (885 genes) were defined as hits; the majority (89.3%; 790 genes) were susceptibility candidates and 95 genes (10.7% of hits) were negative regulators. To determine whether the candidate genes functioned in the gut epithelium and/or in macrophage-like hemocytes, they performed tissue-specific knockdown experiments using different Gal4 drivers. Seventy-eight and 56 genes functioned were found to function only in the gut and hemocytes, respectively, and 79 functioned in both. Gene ontology (GO) enrichment analysis revealed a marked enhancement of genes associated with intracellular processes in the gut, such as endocytosis and exocytosis, proteolysis, vesicle-mediated transport, the stress response, immune system development, growth, stem cell division, and cell death. In hemocytes, genes involved in phagocytosis, including endocytosis, response to external stimuli, and vesicle trafficking, were enriched. They also performed analysis of the JAK-STAT pathway during S. marcescens
infection and found that JAK-STAT signaling enhanced epithelial cell death, and positively regulated the compensatory proliferation of intestinal cells. In summary, more than 800 genes were identified, many of which were of unknown function, demonstrating that host defense may involve many processes not limited to the classical innate immune response pathways.
The t(8:21)(q22;q22) translocation, which produces the abnormal fusion protein AML1-ETO in humans, is associated with acute myeloid leukemia (AML). However, AML1-ETO alone is not sufficient to cause leukemia in mouse; secondary mutations are required before AML1-ETO-expressing cells become leukemogenic. To identify suppressors of AML1-ETO, Osman et al
. chose to knockdown target genes in AML1-ETO-expressing cells using an in vivo
RNAi strategy [57
]. They screened UAS-IR transgenic lines, thereby targeting around 1,500 genes. Eight candidates were discovered. Among the candidates, they studied one gene, Drosophila
calpainB, and mammalian calpains in detail, which revealed that they were required for AML1-ETO stabilization. This study suggests that Drosophila
provides a promising, genetically tractable model to investigate the conserved basis of leukemogenesis.
Glycosylation has crucial regulatory roles in various biological processes. Yamamoto-Hino et al
. performed a primary screened of 6923 UAS-IR strains for genes involved in the glycosylation of a neural glycoprotein and identified 171 candidates [58
]. These were further validated by knockdown experiments, by using in silico
analysis and a secondary set of UAS-IR strains that targeted regions distinct from those of the primary strains. To identify additional glycosylation genes, they performed searches for genes that interacted with those identified in the primary screen using the yeast two-hybrid database, BIOGRID, and the genetic interaction data listed in FlyBase. After validation experiments, a total of 109 genes were identified as glycosylation-related, 95 of which were newly assigned to this category. The gene functional groups obtained included glycosylation reactions, transcription, RNA regulation, translation, intracellular trafficking, cytoskeletal regulation, signal transduction, protein degradation, mitochondrial function, and other functions. Further analysis revealed that 17 of the identified genes were specific for Drosophila
neural glycosylation. These data demonstrate that the use of interaction network databases and second target validation contributes to efficient screening and ultimately the production of conclusive results.
To identify melanotic tumor suppressor genes involved in blood cell homeostasis, Avet-Rochex et al
. screened 1,341 genes (approximately 10% of all Drosophila
genes) and validated the candidates obtained by re-screening independent secondary UAS-IR lines [59
]. Finally, they identified 59 genes previously unlinked to blood cell development or function in Drosophila
. The candidate genes were grouped into nine categories, namely protein synthesis, signaling, transcription, splicing, protein folding and stability, DNA replication/repair mitochondrial activity, membrane proteins targeting, and unknown function. The authors constructed an interaction network between the 59 genes using various databases (DroID, BioGrid, FlyBase), in addition to manual text-mining for each of the 59 genes and their mammalian or yeast orthologs. This approach allowed them to uncover several nodes of interaction among the genes, with 47 candidates linked to at least one other gene, suggesting that they function as complexes and/or in the same pathways.
Neely et al
. screened 7,061 evolutionarily conserved genes for potential developmental and adult heart function defects [60
]. They identified 498 loci that were classified into functional categories including signaling, ion transporter activity, metabolism and mitochondrial structure, development and morphogenesis, and transcriptional regulation. Among these, they confirmed that the CCR4-Not complex is involved in proper heart function in Drosophila
To identify novel genes regulating pain, Neely et al
. tested 16,051 elav-Gal4 > UAS-IR combinations targeting 11,664 different Drosophila
genes (82% of Drosophila
genome) for their contribution to noxious temperature avoidance [61
]. Positive hits were retested, and 622 specific transgenic UAS-IR strains, corresponding to 580 genes, were identified as candidate thermal nociception genes. Among these, the function of straightjacket, a member of the alpha2delta family of genes that function as subunits of voltage-gated calcium channels and control the function and development of synapses, was analyzed further. They also analyzed a mouse alpha2delta3 mutant model and showed that it also exhibited impaired behavioral heat pain sensitivity. These two mammalian studies by Neely et al
. also indicated that Drosophila in vivo
screens provide a promising system for investigation of regulatory systems conserved in metazoans.
Lesch et al
. screened 142 genes to identify loci required for normal wound closure in Drosophila
larval epidermis [62
]. They identified two categories of candidate genes; the first comprised genes encoding components of stress-activated protein kinase (SAPK) signaling pathways, such as the canonical Jun kinase relay and associated transcription factors (11 genes in total), and the second comprised genes involved in actin cytoskeletal remodeling, including Rho-like GTPases and loci involved in phagocytosis (10 genes in total).
Schnorrer et al
. screened 10,461 genes to find loci contributing to muscle morphogenesis and function [63
]. Of these, targeting of 2,785 resulted in defects. The results of this study were systematically compared with those of previous studies to identify muscle genes that had either been identified by expression profiling or through chromatin immunoprecipitation of Mef2-binding sites in embryos. More than half of the muscle-expressed genes and almost half of the Mef2 targets were functionally validated in this screen. A total of 30 genes were positive in all three datasets and 13 of these had no functional assignment before this study.
Pospisilik et al
. screened 10,489 genes using UAS-IR strains to identify candidate obesity genes, which resulted in the identification of 516 genes, 319 of which had human orthologs [64
]. GO based pathway analysis for biological processes revealed enrichment of gene sets involved in cell fate determination, cellular protein metabolic processes, signal transduction, intracellular transport, and regulation of smoothened signaling. A network interaction assembly, based on the results of yeast-two-hybrid analysis, text-mining, and pathway database information analysis of Drosophila
hits and their mammalian orthologs, revealed an interaction map that highlighted genes involved in development, nutrient transport, cell-cycle regulation, the proteasome, protein translation, and chromatin remodeling. The biological process “regulation of smoothened/Hedgehog signaling” was the top-scoring signal transduction pathway of all the annotated pathways in the primary screen. They further analyzed a fat-specific Sufu knockout mouse (Sufu is a potent endogenous inhibitor of Hedgehog signaling in mammals) and found that Hedgehog activation blocks white but not brown adipocyte differentiation in the mammalian model system. This mammalian study also revealed a signal conserved in metazoan obesity.
To identify genes that control the balance between neural stem cell (neuroblast) self-renewal and differentiation, Neumuller et al
. screened 12,314 individual genes (VDRC GD) by examining whether knockdown of each caused abnormalities in number, size or shape, or intracellular GFP fused to CD8 (CD8-GFP) accumulation of neuroblasts, ganglion mother cells, or intermediate neural progenitors [65
]. The quality of the result was evaluated by re-screening of a subset of the candidates using a second RNAi library, the KK library (described in Section 5
). The authors concluded that the reproducibility of the candidate genes is 78.5%. The candidate genes were clustered using databases containing data obtained from two-hybrid screens, biochemical analysis, interlog, text-mining, and genetic interactions between Drosophila
genes. Among the candidates, in addition to ribosome subunit genes, genes involved in splicing control and transcriptional elongation and chromatin remodeling were found to have important roles in neuroblast self-renewal and differentiation.
The distinct localization of membrane proteins with regard to cell polarity is crucial for the structure and function of various organs in multicellular organisms. To identify genes involved in the regulation of protein localization, Yano et al
. performed a large-scale screen using a Drosophila
RNAi library [66
photoreceptor cells have a morphologically distinct apico-basal polarity, along which Chaoptin (Chp), a glycosylphosphatidylinositol (GPI)-anchored membrane protein, and Na/K ATPase are localized to the apical and basolateral domains, respectively. By examining the subcellular localization of these proteins, they identified 106 genes whose knockdown resulted in their mislocalization. GO analysis revealed that the knockdown of proteasome components resulted in mislocalization of Chp to the basolateral plasma membrane, suggesting the direct or indirect involvement of the proteasome in the selective localization of Chp to the apical plasma membrane of Drosophila
To find regulators of neural progenitor self-renewal, Carney et al
. first performed microarray analysis to identify genes expressed in neuroblasts [67
]. They further selected 595 genes that had available UAS-IR strains and mammalian orthologs. By performing a neuroblast-specific, RNAi-based functional screen, 84 genes were identified to be required for proper maintenance of neuroblast numbers. These genes are excellent candidates for regulating neural progenitor self-renewal in Drosophila
and probably also in mammals.
Valakh et al
. knocked down 2,970 genes by neuron-specific RNAi in a search for genes involved in the formation, growth, and maintenance of the neuromuscular junction (NMJ) [68
]. Knockdown of 158 genes in post-mitotic neurons led to abnormalities in the neuromuscular system. Bioinformatics analysis demonstrated that genes with overlapping annotated functions were enriched within the hits for each phenotype, suggesting shared biological roles are important for this aspect of synaptic development. For example, genes for proteasome subunits and mitotic spindle organizers were enriched among those whose knockdown led to defects in synaptic apposition and NMJ stability. Their findings highlight the potential importance of proteasome function for active zone development and maintenance in the presynaptic compartment.
6.3. Class 3: Screens to Find True Positives among Candidate Genes Identified in In Vitro Screens
Infection of Drosophila
with pathogens is thought to lead to processing of the ligand of the Toll receptor, Spaetzle (Spz), by secreted serine proteases (SPs). However, knowledge of SPs acting upstream of Spz in regulating the Toll pathway is scarce. Kambris et al
. screened 75 distinct Drosophila
SP genes and identified five novel SPs [72
], including the Spz processing enzyme (SPE), which directly cleaves Spz [81
Saj et al
. performed a cell-based RNAi screen of 14,200 genes (BKN library) [82
] and identified 900 candidate Notch regulators [73
]. Subsequently, they used a Drosophila
RNAi library for in vivo
validation of the candidates in wing and eye development and confirmed 333 of 501 tested genes as Notch regulators. By mapping the phenotypic attributes of their data onto an interaction network, they identified another 68 relevant genes and found several modules of unexpected Notch regulatory activity. A total of 401 Notch regulators were compared to two other data sets of Notch regulators; one was obtained by studying genetic interactions and mutant phenotypes (Flybase) [83
] and the other was obtained from a whole genome in vivo
RNAi screen focused on genes involved in external sensory organ formation. There was a very limited overlap between the three data sets, which most likely reflected the differences in the approaches used to generate them.
Port et al
. performed a genome-wide RNAi screen to identify genes involved in the secretion of Wingless (Wg) [74
]. For the primary screen, Drosophila
S2R+ cell lines that stably expressed a Wg-Renilla luciferase (WgRluc) fusion protein or a secreted Renilla luciferase (sRluc) were established to assay specific defects in Wg secretion. A library of dsRNAs targeting more than 14,000 Drosophila
genes was screened in the generated S2R+ cell lines and 387 were assigned as primary hits. Following pre-exclusion of genes with well-documented functions unrelated to protein secretion, 115 were re-screened for eye and wing phenotypes using a Drosophila
RNAi library. A p24 protein, Éclair, and a protein termed Sorting nexin 3 (Snx3) were identified as hits in more than one assay.
To identify novel regulators of the Hedgehog (Hh) pathway among genes functioning in the UPS, Du et al
. first selected 248 UPS genes from the Drosophila
]. UAS-IR strains targeting 238 UPS genes were screened by examining adult wing blade phenotypes, the distribution patterns of full length Cubitus interruptus (CiFL), the transcription factors of the Hh pathway, and the expression of the dpp gene, a direct transcriptional target of Hh signaling, in the wing disc. Among these 238 genes, two novel loci (dUba3 and dUuc12) were found to be negative regulators of Hh signaling activity.
Aikin et al
. screened ~21,000 dsRNAs using S2 cells transfected with the Hh gene fused to Renilla luciferase (Hh-Ren) in a search for regulators of cholesterol-modified Hedgehog secretion [76
]. This identified 125 genes, which were then evaluated by secondary dsRNAs that did not overlap with those used in the primary screen. This led to the high confidence identification of 24 genes whose depletion significantly affected Hh-Ren secretion; 11 had previously been found in an RNAi screen for regulators of general protein secretion, while the others had no previously known role in secretion. Four genes (CG5964, CG8441, CG3305, and CG12693) were further analyzed using Drosophila
The various applications of RNAi libraries mentioned above indicate that they are a useful and straightforward system for performing genetic screens and in vivo validation. In addition, mammalian studies of genes identified in Drosophila (by Osman, Pospisilik and Neely) exemplify the power of Drosophila in vivo RNAi screens for the discovery of pivotal pathways conserved among metazoa.