HHMI Report 2005

 

Cell Signaling and Morphogenesis in Drosophila

 

Overview

The long-term interests of my laboratory are to identify, within a developmental context, the cellular components involved in the response of specific cells to extracellular signals; and to understand how these pathways are deployed during patterning and morphogenesis.  Our goal is to identify the parts responsible for the reception and integration of the signals, and then organize them into pathways and networks. Our experimental approach is mostly based on genetic methods, and over the years we have developed and/or improved a number of tools for functional genomic studies to make the process of gene discovery faster, easier, more reliable, and genome-wide. In addition, since cellular architecture provides an additional spatial and organizational layer of regulation, an important aspect of our research is to characterize how the subcellular localization of components of signal transduction pathways influence the output of signaling networks.

 

1. Gene manipulation methods

Since the realization, half a century ago, that genes encode the building blocks of cells, identifying their functions has become a priority in the life sciences. Linking genotype to phenotype has been the most rewarding approach to identify the function of genes and over the years many advances in the field have been made possible by the development of methods that allow precise spatial and temporal control of gene activity.

For example, a particular challenge in mosaic analyses is to easily detect marked clones in internal tissues. Towards this goal, we developed the "Positively Marked Mutant Lineages" (PMML) technique (Harrison and Perrimon, 1993) that allows one to generate clones of mutant cells that express either GFP or LacZ. More recently, we have improved this system based on the reconstitution of an actin-Gal4 gene following an FRT-mediated recombination event between two homologous chromosomes. It differs from a related technology developed by Lee and Luo (1999), which is based on the loss of the Gal80 repressor. In particular, it resolves the inherent problem associated with perdurance of the Gal80 repressor. Our method facilitates the detection of small marked clones of mutant cells, which is especially powerful for the analysis of stem cell lineages.

            As we have learned, in particular from studies in yeast, temperature sensitive (TS) mutations are a powerful tool to identify gene functions, especially in the context of those with pleiotropic roles. Because TS mutations are rare and difficult to recover, we designed a general method to generate TS proteins based on Conditional Splicing Inteins (Zeidler et al., 2004). Recently, in order to generate a collection of TS mutations that will be useful to study developmental processes, we have initiated a large-scale screen using a "gene trap TS Intein" Piggyback transposon. In addition, to add a different inducible level of control, we are considering developing Inteins that are controllable by a small molecule that could be fed to the flies.

 

2. Functional genomic screens using high-throughput RNAi screens

The availability of the Drosophila genome sequence in year 2000 (Adams et al., 2000) has provided us with an unprecedented resource for functional genomic studies. To address the issue that 75% of the genome is not yet functionally annotated, and to systematically analyze the functions of the ~14,000 predicted genes, we established a High-Throughput Screening (HTS) platform to conduct RNA interference (RNAi) screens in Drosophila tissue culture cells in the 384 well plate format. This approach is possible because, as first shown by Clemens et al. (2000), the simple addition of long, double-stranded RNAs (dsRNAs) to Drosophila cells in culture reduces or eliminates specifically the expression of target genes, and thus efficiently phenocopies loss-of-function mutations. Genome-wide RNAi screens allow, once an appropriate cell-based assay has been established, to literally identify most genes involved in the process under investigation in a matter of a few weeks. Results from the tissue culture screens are then followed up by in vivo validation using genetic approaches that include: existing mutations, P-element insertions and transgenic RNAi constructs.

The development of this platform has required the establishment of a screening infrastructure and the development of new protocols (Armknecht et al., 2005). Following our initial success with the technology (Kiger et al., 2003), we generated in collaboration with Renato Paro's group (Heidelberg Germany), the first Drosophila genome-wide library of 21,300 dsRNAs directed against all predicted open reading frames (ORFs) (Boutros et al., 2004). We then evaluated the robustness of this screening method in a diverse set of cell-based assays that are based on transcriptional reporter assays, protein modification (i.e., using a phospho-specific antibody) or high-content imaging. As part of these studies, we have developed protocols for the experimental and analytical optimization of the screens, including normalization strategies, identification of "off-target" effects, and statistical methods for plate analysis. In addition, with regards to the high-content imaging screens, we have made significant progress in developing automated image analysis algorithms and data mining tools. Importantly, we have applied this methodology to existing cell lines, but have also demonstrated that RNAi screens can be performed in primary cells established from embryos, thus expanding the repertoire of assays that can be developed and questions that can be addressed. We have also established at Harvard Medical School a Drosophila RNAi Screening Center (hrttp//flyrnai.org) that is open to the scientific community for performing screens based on the technology and "know how" that we developed. The center is very successful as, in the short time that it has been in operation, more than 28 screens have been conducted and 53 are in the pipeline.

The screens that we have conducted to date fall under three categories: signal transduction, cell biology and differentiation, and host/pathogen interactions (Table 1). These are somewhat complementary as aspects of signal transduction involve cell biological processes (e.g., trafficking, cytoskeletal regulation), and host/pathogen interactions rely, in part, on processes such as trafficking. In addition, we have successfully conducted screens in primary cells. In particular, we have completed screens for axonal outgrowth and neuron viability (Sepp et al., in prep.) as well as muscle differentiation (Bai et al., in preparation).  Finally, in collaboration with Tim Mitchison's laboratory (HMS), we have demonstrated the power of comparative RNAi and compound screens to identify protein targets of small molecules (Eggert et al., 2004).

The goals behind the large-number of RNAi screens that we have performed were to: 1. determine how widely this new screening methodology could be applied and evaluate its robustness; and 2. generate a database of "RNAi signatures" of all Drosophila genes; a resource that can now be "mined" using clustering algorithms to establish correlations between groups of genes that allows discovery of their functions. Importantly, the diversity of the screens conducted and their quantitative natures have allowed us to obtain functional information for 8,500 of the 13,918 predicted ORFs (release 3.2), which is significant since previously functional information from other approaches was only available for 2,700 genes (W. Gelbart/Flybase, pers. comm.).

Altogether, these screens have identified hundreds of new genes involved in the various assays. We have used combinatorial RNAi to gain insight into the interaction and epistatic relationships between the factors identified.  In addition, we have validated some of the candidate genes either in the fly (Baeg et al., 2005), or with collaborators in other systems, such as in mammalian cells (Cherry et al., 2005; Nybakken et al., 2005; Phillips et al., 2005) and in Zebrafish (Dasgupta et al., 2005).

A number of important conclusions can be drawn from our RNAi screens. First, the quality of the data is impressive as the levels of false negatives and positives are very low (approx. 5 to 10%). Second, it is extremely powerful to conduct overlapping screens to quickly filter factors that are shared between pathways or processes from those that are unique. For example, in a comparative screen for host factors that are required either for entry or growth of the intracellular pathogens Listeria and Mycobacterium, we were able to identify factors that were required for both and others that were specific to each pathogen (Phillips et al., 2005; Agaisse et al., 2005). Third, the results from the screens can provide exquisite quantitative information, allowing us to probe pathways beyond the usual levels of sensitivity that can be reached using more classic genetic approaches, such as suppressor/enhancer and synthetic lethal screens. Thus, the RNAi methodology is emerging as a unique tool for "quantitative genetics", and will most likely play an important role in the identification of factors involved in  "genetic background" and disease susceptibility.

3. Future development and improvement of tools for functional genomic studies        

         The power of high-throughput RNAi screening in tissue culture has changed our primary approach to identify the functions of genes. Although the current methodology is robust and effective, it is by no means optimal. To fully exploit this unique tool, a number of technological advances need to be implemented. Because of the diversity in technical know-how required for these approaches, we will collaborate with groups that complement our own expertise. First, we envision that many RNAi screens in the future will be miniaturized and performed on glass slides. Thus, in collaboration with David Sabatini's laboratory (Whitehead), we are currently experimenting with the first generation of Drosophila RNAi chips (Wheeler et al., 2004), in which on a glass slide long dsRNAs are printed and cells are grown. The application of genome-wide RNAi to the cell microarray technology will reduce the cost of a screen by an estimated 30 times and will allow various multiplexing experiments.

         Image-based RNAi screens provide a wealth of information, particularly when many cellular parameters are scored in a single assay. In such screens, automation of image analyses creates a major challenge; as approximately 40,000 images need to be analyzed. To develop an integrated platform for automated image analysis for these screens, we have initiated a number of collaborations with a number of groups who are experts in computing and imaging.

         RNAi screens generate a large amount of data that need to be extensively "mined" to generate robust hypotheses for subsequent in vivo validation. The implementation of data mining is severely hampered by a shortage of developed and fully integrated tools. Our long term goal is to develop a set of "data mining tools" that will allow RNAi screeners to effectively extract information from data sets resulting from their screens. In particular, we will merge "RNAi signature" data sets with protein interaction, literature, and expression databases.

RNAi screens in tissue culture are attractive because they can be conducted relatively fast and also have technical features (i..e., multiplexing, subcellular resolution) that are not easily achievable in vivo. However, we have to be cautious that the findings made in culture are meaningful in vivo. Thus, in the future we will give priority to RNAi screens in primary cells, whose differentiation programs follow closely their in vivo differentiation (Bai et al., in preparation; Sepp et al., in preparation). In addition, to facilitate and increase the throughput of in vivo validation, we are currently working on transgenic RNAi methods to optimize the generation time and robustness of the technology. A major limitation of the technology resides in the reproducibility of the expression levels of the hairpin, an issue that we propose to resolve using the Phage phiC31 integrase method recently developed by Groth et al. (2004).

 

4. Network analysis of signal transduction pathways

         The canonical and linear view of signal transduction pathways has been extremely useful to identify components that biochemically interact. However, the inherent simplicity of this organizational view of pathway integrity is now hindering new conceptual advances in the field. For example, the concept of a linear flow of information is simplistic in light of the complexity of feedback loops, as we have identified for example in the context of the Tor/S6Kinase pathway (Kockel et al., 2005). In addition, the growing number of regulatory interactions between signaling cassettes, usually referred to as "cross-talk", challenge the view of modularity in signaling networks. Perhaps a more physiologically significant way to analyze signaling is to study the flow of information through interconnected signaling networks.

         The motivation behind our RNAi screens is not simply to facilitate functional gene discovery but to gain insights into the structure of signaling networks. Full genome RNAi screens allow us, for the first time, to obtain a global picture of the functions of all genes in a single process. In addition, and most importantly, many of the screens generate quantitative information on the contribution of each gene in a specific read out. The application of this "quantitative genetic" approach to signal transduction is likely to provide fundamental information on the organization and the flow of information through protein networks.     

         a. Identification of shared nodes between signaling pathways. As we have completed RNAi screens for most signal transduction pathways (Table 1), we can now integrate these data sets to identify points of connections or "nodes". Validation of these connecting nodes will be accomplished in vivo either in hemocytes insect blood cells, a system that we have extensively used (Boutros et al. 2002; Agaisse et al., 2003; Brueckner et al., 2004), or the imaginal disc. To facilitate these analyses, we are currently building a number of pathway sensing tools that will be used to readily assay the functions of these nodes.

         b. Organization and flow of information through protein networks. To gain insight into the global structure of signaling networks and specifically how information flows through these networks in response to extracellular signals, we propose, with our computational collaborators, to use gene expression profiles in RNAi treated cells to construct functional interactions. We have used this approach previously to identify a novel split and negative feedback loop downstream of the Tak kinase in response to LPS stimulation (Boutros et al., 2002). As a proof of principle, we will initially construct a network of the kinome by successively perturbing each kinase and protein phosphatase (PPase) by RNAi in S2 and Kc cells and capture gene expression with our home-made "designer chip". To evaluate the flow of information through the Kinase/PPase network, the results will be compared under different stimuli (LPS, Insulin) and at different time points. Altogether, this approach should allow us to gain knowledge not only on the organization of the signaling pathways but also on the quantitative flow of information through the network.

 

5. Spatial cues and their roles in signal transduction and morphogenesis

            A major interest of our laboratory is to characterize how subcellular localization of signal transduction pathway components influence the output of signaling networks, and how the apico-basal and planar polarity of the cell influences this process. Ideally, for a specific pathway, we would like to know: 1. where the ligand is secreted and how it interacts with the extracellular matrix; 2. whether the transducing receptor is localized, and if it is, the biological significance of such compartmentalization; and 3. how the localization of cytoplasmic proteins in certain parts of the cells contribute to the overall signaling characteristics of the system, such as its kinetics and isolation from other pathways.  Finally, in the context of morphogenesis, we need to gain a better understanding of the connections between signaling pathways and dynamic states of cytoskeletal organization.

Despite major advances in recent years, we are still far from having a global understanding of cellular architecture and the spatial organization of signaling pathways. From our studies of the tumor suppressor genes scribble, disc large and l(2)giant larvae, we and others have proposed a model for the establishment of apico-basal polarity in epithelia (Bilder et al., 2000, 2003; Tanentzapf  and Tepass, 2003). These studies have provided a framework to further analyze the spatial subdivision of the plasma membrane and localized reception of extracellular signals, which is an area that we will investigate actively in the next five years.

         Signaling by the Wnt/Wingless (Wg), Hedgehog (Hh) and Decapentaplegic (Dpp) morphogen signaling pathways are critical to establishment of pattern in many tissues. In the past five years, we, and others, have documented the fundamental roles of  Heparan Heparan Sulfate Proteoglycans, and in particular members of the  Glypican family, in regulating the movement and distribution of morphogens through epithelia (Baeg, et al., 2001, 2004; Selva et al., 2001; LŸders et al., 2003; see recent review by Haecker et al., 2005). What still remains completely uncharacterized are the mechanism(s) that link these pathways to cytoskeletal reorganization. Towards this goal, we have recently examined the role of Dpp in imaginal discs, and shown that, unlike previously thought, Dpp does not regulate cell viability in the imaginal disc, but instead acts to reorganize the cytoskeleton most likely by acting on the organization of apical microtubules  (Gibson and Perrimon, 2005). The function of Dpp in imaginal discs is relevant to its function in the embryo where it is implicated in large-scale epithelial morphogenesis during dorsal closure (Stronach and Perrimon, 2001, 2002; Tan et al., 2003). One of our major effort in the next few years will be to identify the pathways that link the reception of morphogens, in particular Dpp, to the cytoskeleton.

         Finally, we have defined the basic rules that govern the maintenance of the hexagonal cellular arrays present in epithelia and uncovered a novel aspect of epithelial cell mitosis that we call "cytokinetic face formation", wherein polarized cell growth at the interface between nascent mitotic sisters underlies the growth and topological integrity of cell sheets.  Our recent identification of two mutations affecting cytokinetic face formation will now open up the problem of epithelial topology to conventional genetic analysis (Gibson et al., in preparation). With this framework of epithelial organization defined, we can now analyze the mechanisms by which the hexagonal shapes of epithelial cells are being modulated by extracellular signals and mechanical tension during morphogenesis. In particular, we will examine these issues in the context of germ band retraction that we have previously described in details (Schšck and Perrimon, 2002, 2003).

 

Table 1: RNAi screens completed to date or in progress

 

General cell biology

1. Cell morphology and adhesion                    Kiger et al. (2003).

2. Cell viability and growth                             Boutros et al. (2004).

3. Cytokinesis.                                                Eggert et al. (2004).

4. General secretion                                         Bard et al., In Preparation.

5. Axonal outgrowth                                       Sepp et al., In Preparation.

6. Myocyte differentiation                               Bai et al., In Preparation.

 

Signal Transduction

7. Wingless/Wnt pathway                               Dasgupta et al. (2005).

8. LRP6 signaling                                           Dasgupta et al., In Preparation.

9. Hedgehog pathway                                     Nybakken et al. (2005).

10. Small GTPases signaling                          Bradley et al., In Progress.

11. GEF signaling                                           Bakal et al., In Progress.        

12. JAK/STAT pathway                                 Baeg et al. (2005).

13. EGFR pathway                                         Friedman et al., In Preparation.

14. Insulin pathway                                         Kockel et al. (2005).

15. JNK pathway                                            Brouzes et al., In Progress.

16. PVR pathway                                            Brueckner et al., In Progress.

 

Host/Pathogen interaction

17. Host factors required for Drosophila C virus pathogenesis.         Cherry et al. (2005).

18. Host factors required for VSV virus pathogenesis.                      Cherry et al., In Prep.

19. Host factors required for growth of Mycobacterium.                    Phillips et al. (2005).

20. Host factors required for growth of Listeria.                                Agaisse et al. (2005).

 

 

REFERENCES

Adams, M. D., Celniker, S. E. et al. (2000) The genome sequence of Drosophila melanogaster. Science 287, 2185-95.

Agaisse, H., Petersen, U-M., Boutros, M., Mathey-Prevot, B. and Perrimon, N. (2003) Signaling Role of Hemocytes in Drosophila JAK/STAT-Dependent Response to Septic Injury. Developmental Cell 5, 441-450

Agaisse, H., Burrack, L., Philips, J., Rubin, E., Perrimon, N. and Higgins, D. E.  (2005) Genome-wide RNAi Screen Identifies Host Factors Required for Intracellular Bacterial Infection. Science. In Press..

Armknecht, S., Boutros, M., Kiger, K., Nybakken, K., Mathey-Prevot, B. and Perrimon, N. (2005) High-throughput RNA interference screens in Drosophila tissue culture cells. Methods in Enzymology.  392, 55-73.

Baeg, G. H., Lin, X., Khare, N., Baumgartner, S. and Perrimon, N. (2001) Heparan sulfate proteoglycans are critical for the organization of the extracellular distribution of Wingless. Development 128, 87-94.

Baeg, G. H., Selva, E., Goodman, R., Dasgupta, R. and Perrimon, N. (2004) The Wingless morphogen gradient is established by the cooperative action of Frizzled and Heparan Sulfate Proteoglycans receptors. Dev Biol. 276. 89-100.

Baeg, G. H., Zhou, R. and Perrimon, N. (2005) Genome-wide RNAi analysis of JAK/STAT signaling components in Drosophila. Genes and Development. In Press.

Bilder, D., Li, M. and Perrimon, N. (2000) Cooperative regulation of cell polarity and growth by Drosophila tumor suppressors. Science 289, 113-116.

Bilder, D., Schober, M. and Perrimon, N. (2003) Integrated activity of PDZ protein complexes regulates epithelial polarity. Nature Cell Biology 5, 53-58.

Boutros, M., Agaisse, H. and Perrimon, N. (2002) Sequential activation of signaling pathways during innate immune responses in Drosophila. Developmental Cell 3, 711-722.

Boutros, M., Kiger, A. A., Armknecht, S., Kerr, K. Hild, M., Koch, B., Haas, S. A., Heidelberg Fly Array Consortium., Paro, R., and Perrimon, N. (2004) Genome-Wide RNAi Analysis of Growth and Viability in Drosophila Cells. Science 303. 832-835.

BrŸckner, K.,  Kockel, L., Duchek, P.,  Luque, C. M., R¿rth, P., and Perrimon, N. (2004) The PDGF/VEGF Receptor Controls Blood Cell Survival in Drosophila. Developmental Cell 7. 73-84.

Cherry, S., Doukas. T., Armknecht, S., Whelan, S., Wang, H., Sarnow, P. and Perrimon, N. (2005) Genome-wide RNAi screen reveals a specific sensitivity of IRES-containing RNA viruses to host translation inhibition. Genes and Development 19. 445-452.

Clemens, J., Worby, C., Simonson-Leff, N., Muda, M., Maehama, T., Hemmings, B. and Dixon, J. (2000). Use of double-stranded RNA interference in Drosophila cell lines to dissect signal transduction pathways. Proc Natl Acad Sci U S A. 97, 6499-503.

Dasgupta,, R., Kaykas, A., Moon, R. T. and Perrimon,  N. (2005). Functional genomic analysis of the Wingless/Wnt-Wingless signaling pathway.Ó Science 308. 826-832.

Eggert, U. S., Kiger, A. A., Richter, C., Perlman, Z. E., Perrimon,N., Mitchison, T. J. and Field, C. M. (2004) Parallel chemical genetic and genome-wide RNAi screens identify cytokinesis inhibitors and targets. PLOS Biology 2. e379.

Gibson, M. C. and Perrimon, N. (2005) Extrusion and death of DPP/BMP- compromised epithelial cells in the developing  Drosophila wing imaginal epithelium. Science 307. 1785-1789.

Groth A. C., Fish, M., Nusse, R. and Calos M. P. (2004) Construction of transgenic Drosophila by using the site-specific integrase from phage phiC31. Genetics. 166, 1775-82.

Harrison, D. and Perrimon, N. (1993) Simple and efficient generation of marked clones in Drosophila. Current Biology. 3, 424-433.

Haecker, U., Nybakken, K. and Perrimon, N. (2005) Heparan Sulfate Proteoglycans: The sweet side of Morphogens. Nature Review in Cell Biology. In Press.

Kiger, A. A., Baum, B., Jones, S., Jones, M. R., Coulson, A., Echeverri, C. and Perrimon, N. (2003) A functional genomic analysis of cell morphology using RNA-interference. Journal of Biology 2 (4).

Kockel, L., Kerr, K., Melnick, M. and Perrimon, N. (2005) A phospho-specific RNAi screen for the Insulin pathway. Submitted.

Lee, T. and Luo, L. (1999) Mosaic analysis with a repressible cell marker for studies of gene function in neuronal morphogenesis. Neuron. 22, 451-61.

LŸders, F., Segawa, H., Stein, D., Selva, E. M., Perrimon, N., Turco, S. J. and HŠcker, U. (2003) slalom encodes an adenosine 3'-phosphate 5'-phosphosulfate transporter essential for development in Drosophila. EMBO J. 22, 3635-3644.

Nybakken, K., Vokes, S. S., Lin, T-Y., McMahon, A. P. and Perrimon, N. (2005) Genome-wide RNA interference screening identifies many new genes with roles in Hedgehog signaling. Submitted.

Philips, J., Rubin, E. and Perrimon, N. (2005) Genome-wide RNAi screen in flies identifies CD36 family member required for mycobacterial infection. Science. In Press.

Schšck, F., and Perrimon, N. (2002) Cellular processes associated with germ band retraction in Drosophila. Developmental Biology 248, 29-39.

Schšck, F. and Perrimon, N. (2003) Retraction of the Drosophila germ band requires cell-matrix interaction. Genes and Development 17, 597-602.

Selva, E., Hong, K., Baeg, G. H., Beverley, S. M., Turco, S. J., Perrimon, N. and Haecker, U. (2001) Dual Role of the fringe connection gene in both Heparan Sulfate and fringe-dependent signaling events. Nature Cell Biology 3, 809-815.

Stronach, B. E. and Perrimon, N. (2001) Investigation of leading edge formation at the interface of amnioserosa and dorsal ectoderm in the Drosophila embryo. Development 128, 2905-2913

Stronach, B. E. and Perrimon, N. (2002) Activation of the JNK pathway during dorsal closure in Drosophila requires the mixed lineage kinase, slipper. Genes and Development 16, 377-387

Tan, C., Stronach, B. and Perrimon, N. (2003) Roles of Myosin Phosphatase during Drosophila development. Development 130, 671-681.

Tanentzapf ., G. and Tepass U. (2003) Interactions between the crumbs, lethal giant larvae and bazooka pathways in epithelial polarization. Nat Cell Biol. 1, 46-52.

Wheeler, D. B., Bailey, S. N., Guertin, D. A., Carpenter, A. E., Higgins, C. O. and Sabatini, D. M. (2004) RNAi living-cell microarrays for loss-of-function screens in Drosophila melanogaster cells. Nature Methods 1, 127-132.

Zeidler, M., Tan, C., Bellaiche, Y., Cherry, S., HŠder, S., Gayko, U. and Perrimon, N. (2004) Temperature-sensitive control of protein activity by conditionally splicing inteins. Nature Biotechnology 22. 871-879.