Using LANL's GFP to determine protein-protein interactions
Typical protein complementation assays (PCA) work by spliting a reporter molecule such as a fluorescent protein or enzyme into two pieces (see Michnick's work). Alone, the pieces cannot fold and are inactive. However, if the pieces are attached to interacting proteins, their interaction forces the two pieces of the reporter molecule together and its activity is restored. This sounds appealing in principle but is difficult in practice. Key to understanding the limitations is understanding that typical reporter fragments generated by splitting a reporter molecule, such as GFP, are relatively large, have poor solubility, and can perturb the normal behavior of the attached test proteins (described by Lynne Regan et al.). In the typical case for GFP, the protein is split into two large pieces that cannot spontaneously fold and aggregate on their own (see Fig. 1, #1 and #2).
Chaperones try to refold them, but each piece unfolds and aggregates before it "finds" its compliment (Fig. 1, #3), and the GFP fragments are not usually close enough together to recombine (Figure 1, #4). As a result, very little of the active, fluorescent GFP forms (Figure 1, #5). Many researchers hope that by attaching the GFP fragments to potentially interacting "passenger" proteins (Fig. 2, #1), they will overcome the solubility issues with the GFP fragments. Unfortunately, this is not the case. With interacting passenger proteins attached to the cumbersome GFP fragments, folding interference leads even more strongly to misfolding, not only of the GFP pieces but also the fused passengers. (Fig. 2, #2).
In a small number of cases, chaperones will succeed in refolding the passenger domain(s), but the process is very inefficient and depends entirely on the host cell's folding machinery (Fig. 2, #3). As a result, a small percentage of the correctly folded proteins can interact, which brings the two fragments of GFP close enough to recombine (Fig. 2, #5). Since folded GFP is stable, it remains folded and eventually (slowly) the GFP builds up in a way consistent with Le Chatelier’s Principle and can be detected (Fig. 2, #6).
The general idea of the LANL approach is presented in Figure 3. Initial work was done to express GFP as two domains (Fig. 3, #1). The first domain,
a hairpin structure, interacts instantly with the second expressed domain, the GFP strand 1-9 "detector" (Fig. 3, #2) to form a fully functional GFP (Fig. 3, #3). Taking the next step, expressing the first domain as two separate strands (Fig. 4, #5) does not create a funtional GFP even with the expression of the "detector" (Fig. 3, #5).
The LANL approach is applied to protein-protein interactions by doing the following. Strand 10 of GFP(S10) is attached to the first test protein, say protein A. Strand 11 of GFP (S11) is attached to the second test protein, call it protein B in this example (Fig. 4, #1). Interaction of protein A with protein B is detected using GFP strands 1-9, the "GFP 1-9 detector." This system works by reducing entropy. Normally, S10 and S11 don't interact by themselves and the so-called 'three-body interaction' between S10, S11, and 1-9 is not energetically favorable and cannot happen.
If test proteins A and B interact with each other, they bring the attached S10 and S11 close enough together to interact with the GFP 1-9 detector (Fig. 4, #2). Essentially, the interaction of protein A and protein B converts the energetically unfavorable three-body problem
into an energetically favorable two-body problem (Fig. 4, #3).
A researcher can do this by following these simple steps. First, a vector is created that includes short nucleotide section of GFP called a "microdomain" tag (strand 10 of GFP, or S10) appended onto the nucleotide sequence of the first protein of interest (protein X), to make protein X-S10. The vector also includes a different nucleotide section of GFP (strand 11 of GFP, or S11) appended onto the nucleotide sequence of another protein (protein Y). Protein Y is expressed from the same plasmid at a second ribosome binding site to give protein Y-S11. The vector is transformed into a host cell, for example E. coli, containing a second plasmid expressing the GFP 1-9 detector. Determing whether protein X is interacting with protein Y is easily done by expressing the tagged proteins at the same time as the GFP 1-9 detector. If fluorescence is observed, the two proteins interact.
This same transformation can be used to determine the solubility of the interacting proteins by staggering their expression. This is done in a similar way as LANL's solubility assay. By inducing expression of the tagged proteins first, shutting their expression off, and then inducing the expression of the GFP 1-9 detector, observing fluorescence indicates that the proteins not only interact, but they are soluble as well.
Expanding this approach, it would be easy to create a library that contains thousands of target proteins to determine if they interact with thousands of other proteins.
To learn more, read about published results in our Technical Library.