Introduction

Biochemical pathways are really systems of dynamically assembling and disassembling protein complexes, and thus, much of modern biological research is concerned with how, when, and where proteins interact with other proteins involved in biochemical processes. The demand for simple approaches to study protein-protein interactions, particularly on a large-scale, has grown recently with the progress in genome projects, since the associating of unknown with known gene products provides one crucial way of establishing the function of a gene. It was with this challenge in mind that our laboratory developed protein-fragment complementation assays (PCAs). In this strategy, two proteins of interest (proteins A and B) are fused to complementary fragments of a reporter protein (an enzyme, fluorescent protein, etc.). If the proteins A and B interact, the reporter fragments are brought together, fold into the native structure of the reporter and reconstitute its activity (Figure 1). PCA reporter proteins have been chosen as those producing a variety of detectable activities, including fluorescent, luminescent, and colorimetric signals, as well as simple survival selection assays (1,2,3,4,5,6,7,8,9,10,11,12,13,14). We have demonstrated that the PCA strategy has the following capabilities: (i) it allows detection of protein-protein interactions in vivo and in vitro in any cell type; (ii) it allows detection of protein-protein interactions in appropriate subcellular compartments or organelles; (iii) it allows detection of interactions that are specifically induced in response to developmental, nutritional, environmental, or hormone-induced signals; (iv) it allows monitoring of kinetic and equilibrium aspects of protein assembly in cells; and (v) it allows screening for novel protein-protein interactions in any cell type (2,3,6,9)(15,16,17,18,19).

Figure 1. Protein complex dynamics can be studied by fusing proteins of interest (proteins A and B represented by two spheres) to complementary fragments of a reporter protein (Click to enlarge)

Principle

We demonstrated the principle of PCA starting with the enzyme dihydrofolate reductase (DHFR) as a reporter (1). It was obvious that if the folding of the enzyme from its fragments (as detected by reconstitution of activity) was absolutely dependent on the binding together of the interacting proteins, then the system described is, in fact, a detector of the interactions. We and others have since demonstrated that this principle can be generalized to a number of enzymes including Gaussia and Renilla luciferases, TEM β-lactamase, as well as green fluorescent protein (GFP) and its variants (1,2,3,4,5,6,7,8,9,10,11,12,13,14). A crucial feature of PCA fragments is that they are designed not to fold spontaneously without being brought into close proximity by the interaction of the proteins to which they are fused (1,20). If spontaneous folding occurred, PCA simply would not work. Spontaneous folding would lead to a false positive signal, a situation that would hopelessly confound the interpretation of library screens in vivo (anticipated to be an important application). In contrast to PCA, there are assay systems based on β-galactosidase and split inteins that resemble PCA, but which are conceptually and practically different (21,22). In both cases, well-known naturally occurring and spontaneously associating subunits of the enzymes are fused to interacting proteins. The central problem here is that subunits, even if weakly associating, are always capable of doing so to some extent, meaning that there is a constant background of spontaneous assembly.

Limitations

The PCA strategy is general, in the sense that it is not restricted to a single enzyme reporter, and it has been devised in several different forms, each of which is best suited to address a specific question. For instance, simple survival-selection PCAs, such as those based on DHFR, are most useful for library selection, while luminescence or fluorescence readout PCAs are best for studies of the spatial and temporal dynamics of protein complexes. Because the fusion proteins can be expressed in cells that are relevant models for studying a specific biochemical pathway, they are likely in their native biological state including the correct posttranslational modifications (obviously the PCA fragments themselves must not interfere with targeting or modification of the proteins, and this must be tested).

Among the simplest and therefore most popular PCAs are those based on fluorescent proteins (such as GFP and variants), because signal is provided by the intrinsic fluorophore (7,8,9)(14,15,17,23). However, fluorescent proteins must be expressed at high levels to assure that signal is above background cellular fluorescence, and fluorescent protein PCAs have been demonstrated to be irreversible, which can be useful (trapping and visualizing rare complexes) but could also lead to misinterpretation of turnover or localization of interacting proteins (8,23,24). On the other hand, PCAs based on DHFR and β-lactamase as reporters have been demonstrated, based on indirect evidence, to be reversible following disruption of interactions, while a PCA based on Gaussia luciferase has been directly shown to be reversible (2,3,6). Reversibility of PCA thus allows for the detection of kinetic and equilibrium aspects of protein complex assembly and disassembly in living cells (Figure 2).