Review
Two-hybrid arrays

https://doi.org/10.1016/S1367-5931(01)00288-5Get rights and content

Abstract

The two-hybrid system is a genetic method for detecting protein–protein interactions. The assay can be applied to random libraries or arrays of colonies that express defined pairs of proteins. Arrays enable the testing of all possible protein pairs for interactions in a systematic fashion. The array format makes a large number of individual assays comparable and thus greatly simplifies the identification of false positives. Two-hybrid arrays have been used to study interactions among the proteins of yeast, hepatitis C virus, vaccinia virus, Drosophila, Caenorhabditis elegans, mouse and other species, and have already identified thousands of interactions.

Introduction

As protein interactions are an essential part of all life, numerous methods have been developed to study them [1]. The two-hybrid assay has proved to be one of the most efficient techniques for finding new interactions 2., 3., 4.. The procedure is simple, inexpensive, and has the important advantage of being unbiased (i.e. no previous knowledge about the interacting proteins is necessary for a screen to be performed). However, the system also has a reputation for producing a significant number of false positives that require cumbersome analysis to separate the ‘wheat’ of true interactions from the ‘chaff’ of false positives.

The advent of complete genome sequences has dramatically changed two-hybrid searches for interacting proteins. Two-hybrid screens of random libraries can be performed much more rapidly when inserts from positive transformants can be identified by sequencing just a few base pairs to identify the encoded proteins. With complete genome sequences at our fingertips, two-hybrid screens can be carried out without any sequencing when known proteins are tested for interactions. Many such individual tests have been published in the literature, but complete genome sequences allow them to be carried out systematically with complete families or functional groups of proteins. However, genome-wide screens have been done only with the whole protein complement of yeast and a few viruses. This review focuses on such genome-wide screens and highlights, from the large number of small-scale array experiments, just a few examples to illustrate their range of applications.

Section snippets

From two-hybrid assays to arrays

In the yeast two-hybrid system, interactions between two protein fusions are detected through protein–protein interaction-dependent reporter gene activation in vivo (Fig. 1). This procedure is typically carried out by screening a protein of interest against a random library of potential protein partners via a genetic selection (Fig. 1a). Plasmid DNA is recovered from cells expressing interacting proteins and gene identities are determined by DNA sequencing. However, two-hybrid screening can

Applications of small-scale two-hybrid arrays

Small-scale two-hybrid arrays (also called mini-arrays) can be used to study a wide range of biological questions (Table 1, Table 2). Finley and Brent [7] reported one of the first small-scale array experiments to study interactions among the members of a protein family, namely cyclin-dependent kinases, cyclins, and related proteins (cyclin-dependent kinase interactors, Cdis) from Drosophila and other species. These authors discovered 19 interactions in just 45 individual tests but plan to

Large-scale, but not genome-wide screens: mouse

Many companies and academic groups have started to work on human and mouse protein-interaction maps. However, few results have been published at this early stage. In a pilot study, Suzuki et al. [15••] tested 3500 mouse cDNAs for interactions using a new procedure (Fig. 2). Altogether, about 12 million protein pairs were tested and among them 145 interactions were found.

Viral genomes: hepatitis C virus and vaccinia

Surprisingly few viral genomes have been studied systematically for protein interactions, although their small size makes them ideal targets for such screens. Flajolet et al. [16] studied interactions among proteins of the hepatitis C virus (HCV). The HCV genome encodes a single polyprotein that is processed into about 10 mature proteins. In this study, all mature virus proteins were tested pairwise against each other. Surprisingly, no interactions were found this way. The authors conclude that

Yeast

The first array-based two-hybrid screen of a whole proteome was published in early 2000 [18••]. This paper describes screens of 192 bait proteins against the 6000 yeast prey proteins, resulting in 281 distinct protein pairs.

Ito et al. 19., 20••. described a similar strategy to search the yeast genome for protein interactions. First, they cloned all yeast ORFs in bait and prey vectors. However, instead of testing each bait clone against all prey clones, Ito et al. generated 62 pools with up to

Arrays and two-hybrid false positives

As with DNA microarrays, two-hybrid arrays allow a comparison of each individual assay with multiple identical assays. When the first array screens were done, it turned out that most positives are not reproducible when a screen is repeated [18••]. Although the molecular reasons for that are not really understood, simply repeating an array screen identifies those non-reproducible false positives (Fig. 3). In addition, a number of preys can be found multiple times with unrelated baits. Such

Conclusions

Despite its routine use in thousands of labs, the classical yeast two-hybrid system is certainly not perfect. False positives are still a major concern in conventional screens even though arrays and other modifications help to identify them. Another limitation is the possibility of bridging effects (i.e. endogenous proteins can act as bridging factors and therefore imply a direct interaction, although only an indirect interaction takes place). This affects all homologous systems including the

Acknowledgements

I would like to thank Stan Fields and Michael Pankratz for helpful comments on the manuscript.

References and recommended reading

Papers of particular interest, published within the annual period of review, have been highlighted as:

  • • of special interest

  • •• of outstanding interest

References (28)

  • R.L Finley et al.

    Interaction mating reveals binary and ternary connections between Drosophila cell cycle regulators

    Proc Natl Acad Sci USA

    (1994)
  • C.A Stanyon et al.

    Progress and potential of Drosophila protein interaction maps

    Pharmacogenomics

    (2000)
  • A.J Walhout et al.

    Protein interaction mapping in C. elegans using proteins involved in vulval development

    Science

    (2000)
  • B.L Drees et al.

    A protein interaction map for cell polarity development

    J Cell Biol

    (2001)
  • Cited by (66)

    • A yeast two-hybrid approach for probing protein-protein interactions at the centrosome

      2015, Methods in Cell Biology
      Citation Excerpt :

      This system can also be automated, allowing large sets of proteins to be systematically assayed. Finally, the array based format reduces the occurrence of certain types of false positives and makes the elimination of others easier due to the ease of retesting the interaction via a simple yeast mating (reviewed in Uetz, 2002). Several large scale Y2H studies have indicated that it is advantageous to examine subregions of proteins in addition to full-length proteins (Boxem et al., 2008; Flajolet et al., 2000; Flores et al., 1999; Formstecher et al., 2005; Fromont-Racine et al., 2000).

    • Cell growth and cell cycle in Saccharomyces cerevisiae: Basic regulatory design and protein-protein interaction network

      2012, Biotechnology Advances
      Citation Excerpt :

      We will refer to proteins belonging to the Growth and Cycle lists, as Growth and Cycle proteins, respectively. High-throughput techniques (e.g. yeast two-hybrid and mass spectrometry analyses) have been exploited to analyze macromolecular interactions in a number of organisms (Tyers and Mann, 2003), leading to the construction of large-scale interaction maps (Jeong et al., 2001; Uetz, 2002; Xenarios et al., 2000). The Growth interaction network, the Cycle interaction network, and the Growth & Cycle interaction network have been generated in this study, using the ad hoc data-base collecting function of the Agilent GenSpring Software; the resulting output list was used to construct the networks as reported in Table S2.

    • Networks and circuits in cell regulation

      2010, Biochemical and Biophysical Research Communications
    View all citing articles on Scopus
    View full text