An RNAi screening platform to identify secretion machinery in mammalian cells

Abstract

Integrative approaches to study protein function in a cellular context are a vital aspect of understanding human disease. Genome sequencing projects provide the basic catalogue of information with which to unravel gene function, but more systematic applications of this resource are now necessary. Here, we describe and test a platform with which it is possible to rapidly use RNA interference in cultured mammalian cells to probe for proteins involved in constitutive protein secretion. Synthetic small interfering RNA molecules are arrayed in chambered slides, then incubated with cells and an assay for secretion performed. Automated microscopy is used to acquire images from the experiments, and automated single-cell analysis rapidly provides reliable quantitative data. In test arrays of 92 siRNA spots targeting 37 prospective membrane traffic proteins, our approach identifies 7 of these as being important for the correct delivery of a secretion marker to the cell surface. Correlating these findings with other screens and bioinformatic information makes these candidates highly likely to be novel membrane traffic machinery components.

Introduction

Two recent events provide the cornerstone of an emerging strategy as to how we might systematically study cell function. The first is the sequencing of the human genome and its continually improving annotation, thereby providing the basic catalogue of all human proteins (Rual et al., 2004). The second key event is the ability to specifically and rapidly silence individual genes and accordingly downregulate their corresponding proteins in mammalian cells through the use of RNA interference (RNAi) (reviewed by Novina and Sharp, 2004). Although targeted gene ablation has long been possible in model organisms such as yeast, fly and mouse, RNAi provides access to study gene function in mammalian cell culture systems within a time scale of days.

One cellular process of great importance is protein secretion. The transport of newly synthesised proteins from the endoplasmic reticulum (ER) via the Golgi stack and endosomal system to the cell surface is an essential aspect of normal cell function (reviewed by Behnia and Munro, 2005). Proteins and lipids need to be correctly delivered to the plasma membrane to exert their function, in addition to providing a counterbalance to the internalisation of molecules through endocytic processes. Furthermore, the cargoes themselves differ greatly between cell types, and each other, as they include soluble molecules, transmembrane proteins and macromolecules of the extracellular matrix. As such, the intracellular machinery required to undertake this task is immense, and has not yet been comprehensively identified in mammalian cells. Integrative approaches are therefore required if a complete picture of trafficking pathways and their components is to emerge (reviewed by Quenneville and Conibear, 2006).

The first large-scale application of RNAi to study membrane traffic in mammalian cells was recently reported by the laboratory of Marino Zerial (Pelkmans et al., 2005). They used synthetic small interfering RNA (siRNA) oligonucleotides to downregulate almost 600 human kinase proteins, and then assay the cells for defects in the endocytosis of viruses utilising either the clathrin- or caveolae-mediated route. Their primary screen measured the translation of viral genes, and these data were then supported by fluorescence microscopy analysis of fluorescent transferrin uptake. This work therefore indicates the feasibility of using a microscopy approach to systematically analyse protein function with respect to membrane traffic. Automated microscopy is an essential aspect of performing such studies with a key requisite being that the system is sufficiently modular to not only allow appropriate information to be gathered from a single type of primary screen, but also from additional and potentially more complex secondary screens (Liebel et al., 2003). More recently, an automated platform for high content screening of cells using time lapse microscopy has also been described, again incorporating RNAi as the method to probe protein function (Neumann et al., 2006).

To date, no complete genome-wide screens have been reported in mammalian cell culture systems, although we are undoubtedly close to such events (Moffat and Sabatini, 2006). This is most likely due to the ever changing catalogue of mammalian open reading frames (ORFs) and therefore as a consequence the lack of comprehensive RNAi tools. This is in contrast to the situation in other organisms and cell systems, most notably cultured S2 cells from Drosophila. Recently, a double stranded RNA (dsRNA) downregulation approach was used in these cells to screen for protein secretion and Golgi complex organisation (Bard et al., 2006). Surprisingly, from over 1100 initial hits, only a total of 130 proteins were considered to be specific regulators of secretion. Since the primary assay in this work employed chemiluminescent detection of secreted horseradish peroxidase, and not high content microscopy analysis, further evaluation of the experimentally derived 1100 hits was not possible. Therefore, the final candidates were selected by excluding those initial hits that were predicted to operate in alternative cellular mechanisms, indicating that their effect on secretion was only indirect.

In order to undertake a complete and systematic assessment of the proteins required for secretory protein transport from the ER to the cell surface in mammalian cells, a platform needs to be established that combines a robust, reproducible and quantitative secretion assay with automated high content image acquisition and analysis. In this work, we describe such a platform that uses reverse transfection of cells on siRNA arrays (Erfle et al., 2004, Neumann et al., 2006) to downregulate target genes in a highly parallel manner. Cells on these arrays are then monitored for constitutive ER-to-plasma membrane transport using a well established temperature sensitive membrane transport marker. We demonstrate the feasibility of the approach by targeting well established transport proteins and 37 novel human genes which encode proteins that localise to membranes of the secretory pathway or the cytoskeleton (Simpson et al., 2000 and http://gfp-cdna.embl.de). Our approach identifies seven new proteins that are essential for the ER-to-plasma membrane transport of the secretory marker used.