High-Quality Binary Interactome Mapping

Abstract

Physical interactions mediated by proteins are critical for most cellular functions and altogether form a complex macromolecular “interactome” network. Systematic mapping of protein–protein, protein–DNA, protein–RNA, and protein–metabolite interactions at the scale of the whole proteome can advance understanding of interactome networks with applications ranging from single protein functional characterization to discoveries on local and global systems properties. Since the early efforts at mapping protein–protein interactome networks a decade ago, the field has progressed rapidly giving rise to a growing number of interactome maps produced using high-throughput implementations of either binary protein–protein interaction assays or co-complex protein association methods. Although high-throughput methods are often thought to necessarily produce lower quality information than low-throughput experiments, we have recently demonstrated that proteome-scale interactome datasets can be produced with equal or superior quality than that observed in literature-curated datasets derived from large numbers of small-scale experiments. In addition to performing all experimental steps thoroughly and including all necessary controls and quality standards, careful verification of all interacting pairs and validation tests using independent, orthogonal assays are crucial to ensure the release of interactome maps of the highest possible quality. This chapter describes a high-quality, high-throughput binary protein–protein interactome mapping pipeline that includes these features.

Introduction

Interactions mediated by proteins and the complex “interactome” networks resulting from these interactions are essential for biological systems. Mapping protein–protein, protein–DNA, protein–RNA, and protein–metabolite interactions that form “interactome” networks is a major goal of functional genomics, proteomics, and systems biology (Vidal, 2005). Information obtained from large-scale efforts to identify protein interaction partners yields crucial biological insights throughout a range of applications. At the single protein level, interactome maps have helped assign functions to both uncharacterized and well-studied gene products (Oliver, 2000). At the systems level, interactome maps have enabled investigations of how regulatory circuits and global cellular network properties relate to biological functions (Han et al., 2004, Jeong et al., 2001, Milo et al., 2002, Yu et al., 2008).

The two major high-throughput strategies used so far to delineate protein–protein interactome networks are: (i) binary protein–protein interaction assays, which detect direct pairwise interactions, and (ii) affinity purification followed by mass spectrometry (AP–MS) approaches, which detect biochemically stable, copurifying protein complexes containing both direct and indirect protein associations. Classically, binary interaction assays have been based on the yeast two-hybrid (Y2H) system developed 20 years ago (Fields and Song, 1989), and which has been improved over time to increase efficiency and quality (Durfee et al., 1993, Gyuris et al., 1993, Vidal et al., 1996). Of late, alternative approaches have been developed to detect binary interactions, such as protein arrays, protein complementation assays, and the split ubiquitin method (Miller et al., 2005, Tarassov et al., 2008, Zhu et al., 2001).

Until recently high-throughput methods were regarded as more likely to produce lower quality information than low-throughput experiments. It has now been shown that highly reliable interactome datasets can be obtained at the scale of the whole proteome (Braun et al., 2009, Cusick et al., 2009, Simonis et al., 2009, Venkatesan et al., 2009) provided that all experimental steps are thorough and all necessary controls and quality standards are included. Lastly, careful verification of all candidate interactions and experimental validation using independent interaction assays are necessary to ensure the release of interactome maps of the highest possible quality.

Even when highly reliable, interactome maps should be considered as network models of interactions that can happen between all proteins encoded by the genome of an organism of interest. As such, they correspond to static representations of collapsed time-, space-, and condition-dependent interactions that dynamically regulate the behavior and developmental fate of diverse tissues. Thus, interactome maps should be used as static scaffold-like information from which the dynamic features of biologically relevant interactions, that is, those that do happen in vivo, can be modeled by integrating additional functional information such as transcriptional and phenotypic profiling data (Ge et al., 2001, Ge et al., 2003, Gunsalus et al., 2005, Vidal, 2001). Ultimately, novel potentially insightful interactions need to be evaluated for their biological significance using genetic experiments, where specific cis-acting interaction-defective alleles (IDAs) of one or both proteins or trans-acting disruptors are tested functionally (Endoh et al., 2002, Vidal & Endoh, 1999, Vidal et al., 1996).