An activity-based two-hybrid system for the selective identification of substrates of protein kinases

Abstract

The yeast two-hybrid (Y2H) 1 system [1] has revolutionized the identiWcation of protein–protein interactions and has had a tremendous impact on cell and developmental biology. In the Y2H system, two proteins are expressed as fusions with the DNA-binding domain (DB) of Gal4/LexA, the bait, and a transcription activation domain (AD), the prey. Interactions between the bait and the prey reconstitute a functional transcription factor that activates reporter genes that have been made responsive to binding by Gal4/ LexA (reviewed in Refs. [2,3]). Since its initial development, the Y2H system has undergone many modiWcations, each of which has enabled unique applications. These adaptations include an array of DB and AD fusion vectors for expression of varying levels or types of fusion proteins, the use of multiple baits and reporters that enable simultaneous tests for speciWcity, the one- and three-hybrid systems to identify DNA-binding proteins and ternary interactions, and so forth. Although these adaptations have reWned this approach, they do not enable discrimination of physiological targets of protein kinases that has generally required downstream biochemistry to conWrm whether the interacting protein is also a target for phosphorylation. This last step is often rate limiting because it requires phosphorylation of recombinant proteins or peptides by puriWed kinase(s). Our laboratory has employed the Y2H system extensively to identify physiological partners of CK2, a conserved Ser/Thr protein kinase from the fruit Xy, Drosophila [4–7]. Such screens, using the catalytic () subunit of CK2, have identiWed an array of developmentally important proteins. Analysis of the primary sequences of these proteins for the CK2 consensus site, (S/T)–(D/E/x)–x–(D/E), indicated that some of these were likely to be targets for phosphorylation, and this subsequently was conWrmed via biochemistry. One of these studies involved the interactions of CK2 with three conserved basic helix–loop–helix (bHLH) repressors—M8, M7, and M5—derived from the Enhancer of Split Complex (E(spl)C) [7]. These proteins interacted strongly with CK2 in the Y2H system and at the protein level, and they were phosphorylated at an invariant Ser residue. During attempts to map the phosphorylation site, we made the serendipitous Wnding that the CK2–M8 interaction was abolished upon phosphorylation. This result suggested that interaction and phosphorylation were coupled, thereby implicating the role of the active site of CK2. Although these studies suggested that the CK2– M8 interaction might reXect a prototypical enzyme–substrate interaction, it was not applicable to other proteins identiWed in our screens such as the regulatory subunit of CK2, CK2, and the ribosomal protein, rpL22, whose interaction and phosphorylation domains were separable [8,9]. Therefore, we hypothesized the presence of two classes of CK2 targets: those where interaction is independent of the active site (class 1, CK2 and rpL22) and those where the active site was required (class 2, M8/7/5). In addition, we reasoned that catalytically inactive CK2 might distinguish class 2 proteins from class 1 proteins. We found that this is indeed the case, and we tested three novel proteins using this simpliWed discriminatory assay.