Comparing the Affinity of GTPase-binding Proteins using Competition Assays

In this protocol we demonstrate a method for comparing the competition between GTPase-binding proteins. Such an approach is important for determining the binding capabilities of GTPases for two reasons: The fact that all interactions involve the same face of the GTPases means that binding events must be considered in the context of competitors, and the fact that the bound nucleotide must also be controlled means that conventional approaches such as immunoprecipitation are unsuitable for GTPase biochemistry. The assay relies on the use of purified proteins. Purified Rac1 immobilized on beads is used as the bait protein, and can be loaded with GDP, a non-hydrolyzable version of GTP or left nucleotide free, so that the signaling stage to be investigated can be controlled. The binding proteins to be investigated are purified from mammalian cells, to allow correct folding, by means of a GFP tag. Use of the same tag on both proteins is important because not only does it allow rapid purification and elution, but also allows detection of both competitors with the same antibody during elution. This means that the relative amounts of the two bound proteins can be determined accurately.


Introduction
The actin cytoskeleton that determines the shape, polarity and migratory properties of mammalian cells is regulated by the Rho-family of small GTPases. The Rho-family GTPases include RhoA that stimulates cytoskeletal contraction, Rac1 that stimulates actin branching and membrane protrusion, and Cdc42 that has similar effects on actin polymerization to Rac1 and causes the formation of filopodia 1,2 . GTPase signaling activity is determined by binding of a nucleotide, which controls the contraction and relaxation of the switch I and switch II loops that mediate the protein-protein interactions with both regulators and effectors. Guanosine 5'-triphosphate (GTP)-bound GTPases activate downstream effectors, whereas the Guanosine 5'-diphosphate (GDP)-bound form is inactive. In the cell, cycles of GTP hydrolysis and nucleotide exchange allow rapid turnover of GTPase signals that are necessary for cytoskeletal dynamics. Nucleotide turnover is regulated by three mechanisms. Guanine nucleotide exchange factors (GEFs) stabilize the nucleotide-free GTPase, catalyzing exchange of GDP for GTP, and thereby stimulating GTPase signaling activity 3,4 . GTPase-activating proteins (GAPs) catalyze hydrolysis of GTP to GDP, thereby inhibiting GTPase signaling activity 5 . Sequestering molecules such as regulator of chromatin condensation 2 (RCC2) and guanine nucleotide dissociation inhibitors (GDIs) obscure the switch loops and in the case of GDIs remove the GTPase from the membrane by interaction with the prenyl tail 6,7 . Each of the three classes of regulatory molecule interact with the switch loops, as do the downstream effectors and some trafficking regulators such as coronin-1C 7 . The purpose of this protocol is to measure competition for the switch I/II binding site between putative regulators and downstream signaling molecules. It should be noted that competition assays test binding to a shared binding site, so that this protocol is not suitable for testing interactions with other sites, such as binding of GDIs to the prenyl tail.
The subtlety of the conformation differences between active and inactive forms, combined with the labile nature of the bound nucleotide, has made study of GTPase-binding events difficult. The role of the bound nucleotide means that conventional binding assays such as immunoprecipitation or surface plasmon resonance are not well suited to investigation, as the nucleotide cannot be controlled. This obstacle is compounded by the overlap in the binding sites of GEFs, GAPs, effectors, sequestering molecules and trafficking molecules, which make binding data for a single interaction difficult to interpret in the context of the competition that will occur in the cell. Immunoprecipitation, in particular, is compromised by competition between binding partners, as under certain cellular conditions, one binding partner might be identified at the expense of all others, while under other conditions, another partner might dominate. The dynamic nature of GTPase signaling is essential to GTPase function and must be considered when analyzing the relationships between the binding interactions of different regulators. Indeed, we recently described a pathway that relied heavily on competitive binding. We identified coronin-1C as a trafficking molecule that bound to the switch loops of GDP-Rac1 In the cell, the large number of GTPase-binding proteins combined with the rapid nucleotide turnover makes such pathways difficult to interpret. The simplicity of this method in comparing only pairs of binding proteins and using carefully controlled nucleotide-loading conditions allows signaling pathways to be elucidated. However, the greatest strength of the protocol is also the greatest weakness as it is a simplification of the in vivo situation. Competition assays can be used to build a robust hypothesis, but this should then be tested in cells by knockdown experiments.
There are three features that must be considered when selecting the GFP-tagged GTPase-binding proteins to be used in the experiment. First, the fusion proteins must express well in mammalian cells, such as HEK293T, as competition assays require a reasonable amount of protein. Second, it must be possible to purify the recombinant protein without significant degradation, and where this is not possible, cloning of a GTPase-binding fragment should be considered. Third, the two GTPase-binding proteins must resolve from one another on SDS-PAGE to allow analysis in section 6.
There are a number of potential caveats to the experiment that need to be considered, and possibly addressed: Possible denaturation of purified GTPase-binding proteins during the acid elution step or steric hindrance by the GFP tag. In our hands, these have not been a problem, but must be tested. The purified proteins can be tested in functional assays 10 . Commercial kits now exist for testing the activity of GEFs or GAPs without the need for isotope-labeled nucleotides. Sequestering proteins, by their nature protect GTPases from GEF or GAP activity, so can be used as competitive inhibitors in the commercial GEF or GAP assays, as we did in our recent publication 7 .
The relevant feature of proteins that traffic GTPase are the capacity to bind the GTPase, and this can be tested easily in a pull down assay. An alternative approach to testing protein integrity that is applicable to all binding proteins is to titrate protein eluted from GFP-trap beads with glycine with the same protein removed from GFP-trap beads by enzymatic cleavage. The experiment would be analyzed by probing both the GFP-tagged and cleaved protein with an antibody against the protein itself. If the protein is undamaged by elution, equilibrium should be achieved at a 1:1 ratio. This approach would also indicate whether the presence of the GFP tag itself compromises the binding properties of the candidate protein, though this does require the production of a construct with an enzymatic cleavage site between the tag and the binding protein. Whether the protein is compromised by the tag or the elution step, the problem could be addressed by modifying the protocol to use an alternative purification method. Rather than GFP, binding proteins could be His-tagged, purified using Ni-NTA and analyzed using an antibody against the His-tag. The important feature is that both binding proteins must share a common tag although, if necessary, two tags could be added to a protein, one for purification and the other for detection.
The protocol is designed to investigate competition between interactions with the switch I/II domains. Although the majority of GTPase interactions are mediated by this motif, there are some exceptions, most notably the interactions of GDIs that bind to the prenyl tail, as well as obscuring the switch domains. In principle, the protocol could be adapted to use GTPase purified from mammalian cells, so that the GTPase is prenylated, however, the presence of multiple binding sites or allosteric effects complicate the interpretation of competition-binding data. Further problems associated with such a modification are that GDIs co-purify with GTPase from mammalian cells, compromising the purity of the isolated proteins and the hydrophobic nature of the prenyl groups means that prenylated GTPases are associated with either GDI or lipid membrane and such factors would need to be considered in the experiment.
The amount of GST-Rac1 being used in the assay. The constant GTPase binding protein must be at a greater concentration than the Rac1, or when the competitor is added, it will simply bind to free Rac1. It will be immediately obvious if this has happened as binding of the competitor, without a loss of the constant protein, will be detected in much the same way as when the two competing proteins bind to one another as shown in Figure 3B. As an additional control (Step 5.3), a binding reaction containing double the amount of constant binding protein and no variable binding protein should be included (Step 5.3). If the Rac1 in the titration experiment is saturated, doubling the amount of constant binding protein will have no effect on the output. The volumes suggested in the protocol should be appropriate, but the amount of Rac1 can be easily reduced. If binding of the competitor without loss of the constant binding partner is observed, reducing the amount of Rac1 should be attempted before trying to map binding sites to avoid ternary complex formation.
Non-specific interaction of GTPase-binding proteins with the GST or bead, as well as specifically with Rac1. This problem would be manifested by residual binding of the constant GTPase-binding protein, even when the variable GTPase-binding protein has reached a plateau at high concentration. Identification of this issue will be aided by conducting reciprocal experiments where the constant and variable GTPase-binding proteins are swapped. Reciprocal experiments will also greatly improve the accuracy of the estimate of equilibrium point, so should always be included. In cases of non-specific binding, the relative concentrations at which equilibrium is achieved can still be calculated by comparing band intensity between the maxima and minima for each protein, or by measuring the extent of non-specific binding by using GST beads as bait, rather than GST-Rac1.
Pull down assays using different nucleotide-loading conditions should be used to complement the competition assay described in this protocol. Determining the nucleotide preference of partners is important for both understanding the competition events and understanding the signaling pathway that the GTPase-binding protein is involved in. In Figure 2B we analyze binding of proteins with established preference for GTP-loaded or nucleotide-free GTPase as a means to validate nucleotide loading. However, it is sensible to investigate the effect of nucleotide loading on each of the competitors as well. If the hypothetical competitors show different preferences, competition will make less of a contribution to the signaling pathway, and indeed nucleotide turnover is likely to be the mechanism that directs exchange of the binding proteins.