Ongoing work in our laboratory can be subdivided into four categories. First, to facilitate Functional Genomic approaches in Drosophila, we develop, improve, and generate reagent resources and bioinformatics tools. These make gene discovery and identification of gene function both in vivo and in vitro/tissue culture faster, easier, more reliable, and genome-wide. Importantly, to maintain and build on the Drosophila community’s tradition of sharing, which was pivotal to establish Drosophila as one of the premier model systems, we make the methods and reagents that we develop immediately available to the community. Second, we apply these tools to tissue culture cells and in vivo to elucidate the organization of the core Cell Circuitry networks involved in signaling and development. Our approach, based on genome-wide RNAi screening, proteomic and computational analyses, is to identify the parts responsible for the reception and integration of the signals; organize them into pathways, protein complexes and networks; and then validate the findings in more complex in vivo biological systems. In addition, as a large fraction of the genome has no obvious phenotype when perturbed, we are performing combinatorial screens to reveal gene function. Third, as a model of Tissue Homeostasis, we study the mechanisms that control the proliferation and differentiation of Drosophila adult gut intestinal stem cells (ISCs) in both normal and injured conditions. Drosophila ISCs allow us to discover new genes important for gut homeostasis and to analyze crosstalk among signaling pathways. Fourth, as stability of large and complex biological systems depends on long distance communication between organs, we are dissecting the Inter-organ Communication Network, identifying new hormonal systems and determining their physiological roles and regulation.
Functional Genomics Resources
In vitro: In 2003, we established at Harvard Medical School the Drosophila RNAi Screening Center (http//flyrnai.org) to make available to the community our genome wide high-throughput RNAi screening reagents and methods. Using this platform, the functions of the ~15,000 predicted Drosophila genes can be systematically analyzed in various cell-based assays. To date more than 100 screens have been performed by our lab and others, underscoring the success of the center. Over the years, we have added a number of screening reagents and developed bioinformatics tools to improve the platform. We recently renamed the center “DRSC-Functional Genomics Resources” to better represent our current capabilities. We now have available for screening: Genome-wide RNAi libraries and Subset RNAi libraries (Kinase/Phosphatase; Ubiquitination; Transmembrane proteins; Transcription factors; RNA-binding proteins; Autophagy-related proteins; G-protein coupled receptors; Membrane-bound organelle; and Orthologs of human proteins for which there are FDA-approved drugs); Overexpression libraries (UAS-ORFs); and Reagents for both gain-of-function (UAS-miR) and loss-of-function (UAS-miR-sponges) miRNAs screens. Further, we have established new cell lines and methods for screening (primary muscle and neuronal cells, gene trap tagged cell lines, CRISPR mutant cell lines), and experimental and bioinformatics approaches methods for addressing off-target issues and other sources of false discovery. In addition, we have developed bioinformatics tools for reagent design, identification, and validation, as well as data analyses. These tools include RNAiCut for using biological criteria to deduce appropriate Z-score cut-off values for screen data analysis; SnapDragon for design of long or short double-stranded RNAs for cell based or in vivo RNAi, including off-target filtering; DIOPT for identifying orthologs; GLAD for view of gene lists (e.g. for screens of gene subsets); DGET for comparison of gene lists (e.g. screen hits) with existing cell, tissue, stage, and RNA-seq datasets; COMPLEAT for protein complex-based enrichment analysis of RNAi screen or other high-throughput datasets; Find CRISPR for identification of short guide RNAs (gRNAs) for S. pyogenes CRISPR-Cas9 approaches; and Evaluate CRISPR for filtering of gRNAs based on an algorithm predicting the likelihood of effectiveness of a gRNA.
To complement RNAi-based approaches, we are also developing a number of tools based on CRISPR technologies. For in vitrostudies, we are using CRISPR to mutate or engineer cell lines that can be used for screening. We have developed efficient protocols in Drosophila cells, which are particularly challenging as they are polyploid and difficult to grow following single-cell isolation. CRISPR-generated mutant cell lines, in combination with RNAi, provide a robust platform for combinatorial screening. We are also generating stable nuclease-dead Cas9 activator (dCas9a) cell lines that can be used, after transfection with gRNAs, to perform overexpression screens, thus complementing loss of function screens. We are also establishing a novel screening method for pooled CRISPR screens whereby gRNAs from a library are introduced into a specific docking site in cells, a phenotypic selection is then applied, and gRNAs enriched following selection are identified via next-generation sequencing.
In vivo: Because results from tissue culture screens need to be followed up with in vivo validation, and given the independent value of tools for in vivo genetic studies, we have improved methods for transgenic RNAi in Drosophila. We demonstrated that shRNAs are more efficient and specific than long dsRNA when expressed as transgenes. We have now generated a genome scale collection of 12,800 lines covering 73% of the genome in our optimized VALIUM vectors (83% of highly conserved genes). This collection is available to the community through the Bloomington Drosophila Stock Center (in 2015 alone, 78,000 TRiP stocks were distributed by BDSC to the community). These lines can be used to quickly validate results from genome-wide RNAi screens as well as to conduct low- or high-throughput genetic screens and other studies in vivo.
Recently, we optimized dCas9a to perform tissue-specific gain-of-function screens in vivo, and have started to generate a genome scale collection of transgenic lines, carrying U6-gRNAs positioned upstream of the target gene transcriptional start site (TSS). This collection will cover more than 10,000 genes. We also plan to generate a genome-wide collection of transgenic lines with gRNAs positioned downstream of the TSS. These lines can be combined with tissue-specific delivery of Cas9 to generate clones of cells (mosaics) or combined with germline expression of Cas9 to generate null mutations. In addition, these gRNA fly stock resources can be used for genome engineering of the promoter or coding regions.
We have initiated a number of projects based on these gRNA resources. First, we are systematically characterizing the function of small open reading frames (sORFs) to explore the poorly understood biological functions of small peptides; and second, we are targeting rate-limiting enzymes in metabolic pathways as a means of disrupting normal levels of various metabolites and perturbing specific metabolic processes. Finally, using our CRISPR platform, we are helping the Bellen lab to generate a collection of 5,000 CRIMIC lines that contain a MiMIC recombinational cassette element positioned in the first intron of each target gene. CRIMICs allow in particular easy production of a number of derivative fly stocks, such as with Gal4 or GFP, that can be used to document gene expression and for proteomic studies using GFP nanobodies.
In addition to these community resources, we plan to continue to develop a number of bioinformatics tools to facilitate Functional Genomics studies. In particular, we are developing the Molecular Interaction Search Tool (MIST) for visualization and analysis of network data; expanding our Drosophila Gene Expression Tool (DGET) to include new data sets and network functionality; and further improving upon our integrative DIOPT approach to identify orthologs. Finally, we have started incorporating DRSC tools into FlyBase to make them more visible to the fly community. As model organism databases (MODs) are planning to merge and integrate efforts, we expect that DRSC tools will also become widely used beyond Drosophila.
Building Networks: A simple way to view cell signaling is that specific receptors activate core pathways that function in almost all cells and tissues of the animal. In addition to the core pathways, a number of tissue-specific upstream regulators and downstream effectors of various pathways exist to specify the various biological outcomes. A more complete understanding of how various signaling pathways bring about different cell fates would benefit from identification of additional relevant components, including cell-specific components, and from network-level views of pathway activity. Over the past few years, we have used a number of “omics” approaches to expand our understanding of the structure signaling networks (including Mass Spec, RNAi, phosphoproteomics, and transcriptomics). Applying these approaches has allowed us to identify additional components of signaling networks such as the MAPK/ERK, AKT/PI3K, Hippo, and JAK/STAT pathways, as well as pathways controlling autophagy. A number of the new components have been validated in vivo. Given the number of candidates, we sought ways to integrate results. Most proteins are part of larger protein complexes, often located within specific sub-compartments of cells, which can be thought of as the functional units or molecular machines of the cell. To view pathway signaling in the context of protein complexes and sub-cellular locations, we developed a protein complex-based enrichment tool for analysis of ‘omics data (COMPLEAT), and applied the biotin proximity labeling APEX technology in vivo to identify proteins with shared ‘addresses’ within cells. These approaches have allowed us to produce a number of comprehensive lists of components of subcellular regions and organelles. As we are now gaining a better understanding of the structure of the subcellular proteome, we have initiated a large-scale protein-protein interaction study to define pair-wise interactions. In collaboration with the Vidal and Celniker’s labs, we are using yeast two-hybrid analysis to interrogate binary interactions among 10,000 Drosophila proteins. We anticipate that overlay of the protein complex and binary interaction maps will improve quality of those maps and contribute to our growing network-level understanding of protein interactions.
Combinatorial screens: A striking finding from large-scale studies is that we could only functionally validate about half of the network components using single-gene loss of function studies, suggesting a high level of functional redundancy within networks. For example, using Mass Spec we generated an Insulin pathway interactome of about 560 proteins and only 50% were found to affect pathway output as measured using phospho-specific antibodies. To interrogate redundancy, we established a robust platform for performing combinatorial screens, including to identify synthetic interactions. We initially explored the used of combinatorial RNAi perturbations but realized that it was not an optimal approach. Instead, we found that CRISPR-generated mutant cell lines, in combination with RNAi, provide a robust platform for combinatorial screening. As a pilot experiment, we generated cell lines deficient for tumor suppressors of the Insulin pathway, TSC1, TSC2 and NF1. Next, we performed RNAi screens to identify synthetic lethal interactions, and successfully identified synthetic lethal interactions for both TSC and NF1. We subsequently showed in collaboration with the Manning lab that they were conserved in human cells. With our collaborators, we have now identified drugs approved for treatment of other diseases that target some of the candidate hits and are following up on the idea that they might be useful therapeutics.
We realized that in general, synthetic lethal screens have low reproducibility and high false negative rates and identified two contributing factors. First, synthetic lethal screens require a higher signal to noise ratio as compared to standard single-gene viability screens. Second, over 50% of identified synthetic lethal interactions implicate essential genes, which are not easily identified using existing screening reagents optimized for strong knockdown, such that the reagents impair viability in both mutant and wildtype cells. To address these issues, we have developed a method allowing simultaneous analysis of a range of knockdown efficiencies of a target gene in a cell population. This approach, called Variable Dose Analysis (VDA) exhibits reduced noise compared to previous synthetic lethal screening methods and initial results indicate improved identification of synthetic lethal interactions with essential genes. In the next few years, we plan to expand widely these synthetic screens to address fundamental questions of redundancy within signaling networks and exploit their use in drug discovery.
Drosophila Gut & Tissue Homeostasis
Precise regulation of epithelial stem cells is critical to maintain tissue integrity and prevent over-proliferation and cancer. Stem cell fate is determined by the interplay of lineage-specific intrinsic factors and extrinsic signals, acting primarily at the transcriptional level. Drosophila ISCs, which our lab and the Spradling lab identified in 2006, are an excellent basic model for epithelial stem cell fate as their modes of cell division and extrinsic signals are broadly conserved to mammals. The lineage of ISCs is simple, as these cells divide to produce enteroblasts (EB) that differentiate directly into enterocytes (EC) or enteroendocrine cells (EE) without further division. Since the characterization of ISCs, the major signals have been identified along with a few intrinsic transcription factors. We are using Drosophila ISCs to discover new genes important for gut homeostasis and analyze crosstalk among signaling pathways.
Identification of new factors: From a number of screens (receptome-wide RNAi screen, miRNA screens, etc.), we have identified and validated novel mechanisms of how stem cells adjust their rate of proliferation/differentiation to meet the demand for tissue regeneration. Arguably, our most novel finding is the identification of a number of ion channels that either directly or indirectly regulate stem cell proliferation. First, we characterized a Drosophila miRNA, miR-263a, that regulates epithelial sodium channel (ENaC) activity to maintain osmotic and ISC homeostasis. In the absence of miR-263a, the intraluminal surface of the intestine dehydrates, ECs swell, and ISCs overproliferate as a result of cytokine production for stressed ECs. Furthermore, dehydration of the intraluminal surface increases bacterial infection as evident by the increased expression of antimicrobial peptides. Strikingly, these phenotypes are reminiscent of the pathophysiology of cystic fibrosis (CF), as in CF the non-functional CF transmembrane conductance regulator (CFTR) increases ENaC activity resulting in chronic dehydration of the intraluminal surface liquid in organs such as kidney, colon, lung and sweat glands. As miR-183, the human ortholog of miR-263a, also regulates ENaC, our findings suggest miR-183 might be used as a treatment for CF. Second, we showed that the calcium channel TrpA1 is expressed in ISCs and is required for their proliferation. In response to tissue damage reagents such as paraquat and bleomycin, TrpA1 mediates calcium influx, which in turn activates Ras/MAPK pathway and induces stem cell proliferation. Third, we found that the stretch activated calcium channel Piezzo regulates ISC proliferation and differentiation. Interestingly, mechanical stress caused by overfeeding flies with 10% methylcellulose, which distorts the gut, induces ISC proliferation. We believe that this is reminiscent to changes in ISC proliferation observed in animals that feed infrequently, such as the Burmese python, where ISC proliferation is triggered following a large meal as a result of gut distortion.
Cross-talk between signaling pathways: In the past few years, work from our lab and others have documented that many signaling pathways mechanisms of how stem cells adjust their rate of proliferation/differentiation to meet the demand for tissue regeneration. Arguably, our most novel finding is the identification of a number of ion channels that either directly or indirectly regulate stem cell proliferation. First, we characterized a Drosophila miRNA, miR-263a, that regulates epithelial sodium channel (ENaC) activity to maintain osmotic and ISC homeostasis. In the absence of miR-263a, the intraluminal surface of the intestine dehydrates, ECs swell, and ISCs overproliferate as a result of cytokine production for stressed ECs. Furthermore, dehydration of the intraluminal surface increases bacterial infection as evident by the increased expression of antimicrobial peptides. Strikingly, these phenotypes are reminiscent of the pathophysiology of cystic fibrosis (CF), as in CF the non-functional CF transmembrane conductance regulator (CFTR) increases ENaC activity resulting in chronic dehydration of the intraluminal surface liquid in organs such as kidney, colon, lung and sweat glands. As miR-183, the human ortholog of miR-263a, also regulates ENaC, our findings suggest miR-183 might be used as a treatment for CF. Second, we showed that the calcium channel TrpA1 is expressed in ISCs and is required for their proliferation. In response to tissue damage reagents such as paraquat and bleomycin, TrpA1 mediates calcium influx, which in turn activates Ras/MAPK pathway and induces stem cell proliferation. Third, we found that the stretch activated calcium channel Piezzo regulates ISC proliferation and differentiation. Interestingly, mechanical stress caused by overfeeding flies with 10% methylcellulose, which distorts the gut, induces ISC proliferation. We believe that this is reminiscent to changes in ISC proliferation observed in animals that feed infrequently, such as the Burmese python, where ISC proliferation is triggered following a large meal as a result of gut distortion.
(Wnt/Wingless, Insulin Receptor, EGFR, INR, JNK, JAK/STAT, Notch pathways, Hippo) play a role in the division and differentiation of ISCs in the context of homeostasis and/or injury. The challenge is now to understand how these pathways intersect and cooperate, as it is not clear whether some pathways regulate the activity of others, or whether they act concomitently. We have taken a number of approaches to address the spatial and temporal regulation of signaling pathway activities in the gut epithelium. First, we developed a new method to visualize the dynamic activation of pathway activities. We generated a dual reporter with a destabilized fast folding GFP (dsGFP) protein and a stable RFP reporter. dsGFP, because of its short half-life, captures “real time” signaling, while stable RFP labels cells where the signaling activity occurred. To overcome the problem associated with the significant reduction of signal strength of dsGFP, we increased the posttranscriptional production of the fluorescent protein by including several translational enhancing elements. This dual reporter can be used in either fixed or live tissues to monitor spatio-temporal activity of signaling pathways at an unprecedented resolution. We are now using this tool to establish sensors of each signaling pathway involved in gut homeostasis, and will use these reagents to examine pathway cross-talks.
Second, we identified 53 transcription factors, specifically expressed or enriched in the ISCs, and are characterizing how they cooperate to integrate signal inputs. We are currently focusing on transcritption factors with mammalian orthologs implicated in epithelial homeostasis or cancer, and are using targeted DamID (TADA) to characterize their targets. Transcription factor Dam fusions of the major proliferative signaling pathways will be used to address questions about crosstalk and target co-regulation.
Inter-organ Communication Network
Hormonal regulation of physiology: Drosophila has emerged in recent years as a prime model to dissect the intricate interactions between organs and the role hormones play in coordinating the state of one organ/tissue with others. For many years most hormonal studies focused on Insulin signaling, regulated by Dilps, and to some extent on AKH/Glucagon. However, in recent years a number of additional hormonal systems have been uncovered. In particular, our lab identified Upd2 as theDrosophila Leptin ortholog and characterized the mechanism by which Upd2 senses fat in adipose tissues. In addition, we showed that Drosophila muscles, depending on their physiological states, produce a number of systemic factors, such as ImpL2/IGFBP, Myostatin/GDF11, and Activin-beta that affect the physiology of other tissues. We also showed that ImpL2, which is produced from gut tumors, triggers systemic organ wasting reminiscent to cachexia by downregulating systemic Insulin levels. We are currently exploiting this “cachexia” model to characterize the underlying molecular mechanisms of muscle wasting, focusing on the roles of the ubiquitin-proteasome system (UPS) and the autophagic-lysosome pathway in muscle atrophy. In addition, we are interested in identifying additional factors derived from gut tumors important for organ wasting.
Skeletal muscles are the primary site of nutrient utilization in multicellular organisms and hence play an important role in nutrient homeostasis and manifestation of metabolic disorders like obesity. Skeletal muscles often show opposite responses to signals such as exercise and high calorie food. However, how these tissues directly communicate with each other is not well understood. To understand the nature of the cross talk between these tissues we are identifying myokines and adipokines, and characterizing how their productions are regulated by diet, muscle activity, and other hormonal systems. Muscle activity in particular can augment muscle glucose uptake, lipolysis, lipid oxidation, and systemic anti-inflammatory signals. These physiological changes can significantly improve lipid clearance, adipose tissue function and peripheral insulin sensitivity. To study the effect of muscle activity, we have established an optogenetic muscle activity assay to characterize the molecular mechanisms that mediate muscle-activity induced physiological changes both in the muscle and in distant tissues.
Screen for new hormones: Many connections between organs remain to be investigated and no systematic large-scale methods or screens to specifically identify long-distance communication factors have been performed. Using transgenic RNAi reagents, we can design screens to identify genes involved in “organ communication.” With this approach, we uncover communication between tissues by simply inducing expression of RNAi transgenes in Tissue A and examining the effect in Tissue B.
We have made significant progress identifying organ communication factors that regulate various physiological states. We performed a screen of more than ~1000 putative secreted factors in fat body, glia, muscles, or gut, and looked for phenotypes in other organs. Using this approach, we have identified many evolutionary conserved factors including proteases and growth factors that affect growth, metabolism, and physiology.
Although the organ communication screens have led to the identification of many interesting candidates, most of them encode fly genes of unknown function that have not previously been characterized genetically. Our current challenge is to determine if the candidates identified in the screens act directly on distant tissues or via other factors. In addition to performing hemolymph/blood proteomics that detected about 700 putative proteins in circulation, we developed a novel proteomic method to discover organ communication factors. This approach led us so far to identify 56 and 127 proteins trafficking from muscle to head and fat body to muscles, respectively. A number of these proteins (or their orthologs) have been found in the human and fly blood, are predicted to be secreted, and/or are found in mammalian in vitroadipocyte and myocyte secretomes. We are currently following up on a number of promising candidates. For example, we are characterizing an evolutionary conserved secreted factor that traffics from the fat body to muscle. RNAi knockdown of this gene in the fat body results in flies with poor climbing ability, increased mitochondria, decreased actin fiber width, and abnormal neuromuscular junctions in muscles.
Beyond identifying the pathways regulated by new hormones and their roles, the next step will be to understand whether they act acutely or chronically, and how they cooperate with other signaling systems in physiological regulation.