Type-3 BRET, an Improved Competition-Based Bioluminescence Resonance Energy Transfer Assay

Abstract

We show that in conventional, competition-based bioluminescence resonance energy transfer (BRET) assays of membrane protein stoichiometry, the presence of competitors can alter tagged-protein density and artifactually reduce energy transfer efficiency. A well-characterized monomeric type I membrane protein, CD86, and two G protein-coupled receptors β₂AR and mCannR2, all of which behave as dimers in these conventional assays, exhibit monomeric behavior in an improved competition-based type-3 BRET assay designed to circumvent such artifacts.

Bioluminescence resonance energy transfer (BRET) is widely used to study protein oligomerization, particularly for membrane proteins. BRET works on the principle that excitation energy from a bioluminescent donor luciferase (Rluc) molecule can be transferred to a fluorescent acceptor (typically a derivative of green fluorescent protein; GFP) by nonradiative dipole-dipole coupling, such that emission is observed from the acceptor rather than the donor (1). The efficiency of transfer (BRET_eff) is dependent on the relative separation distance between the two molecules, making it informative with regard to the stoichiometry of proteins of interest attached to the donors and acceptors.

We have previously described two BRET-based assays that distinguish between monomers and dimers in situ (2). The type-1 BRET assay uses the relationship between BRET_eff and acceptor/donor ratio at a constant total protein density as an indicator of stoichiometry. The type-2 BRET assay relies on the systematic variation of protein expression levels, starting at very low expression levels, to identify constitutive dimers. An alternative approach, the BRET competition assay, is based on the principle that the extent of BRET between donor and acceptor molecules forming oligomeric complexes is sensitive to the presence of untagged competitor proteins. These compete with acceptor-tagged molecules for interaction with their donor-tagged counterparts, thereby reducing the BRET_eff for oligomers. In contrast, the presence of competitors has no effect on BRET_eff for monomeric, i.e., randomly interacting proteins (see Fig. S1, a and b, in the Supporting Material).

A key consideration in the use of BRET competition assays is the effect of competitor proteins on the total expression of their acceptor- and donor-tagged equivalents. The total expression levels of many membrane proteins are stringently regulated via a combination of translational regulation and recycling/degradation at the cell surface (3), and often partly regulated via negative feedback mechanisms. In the context of BRET experiments, increasing total protein density at the cell surface by coexpressing competitors could, in many cases, induce increased recycling of the proteins in question, regardless of whether they are tagged or untagged. This would lead to a reduction in the overall density of tagged proteins, which would in turn reduce the contribution to BRET_eff of nonspecific interactions, producing a false dimer signature for a monomer (see Fig. S1, c and d). This may also apply to the addition of closely related competitors during investigations of heterodimers, because they may also be subject to similar regulatory networks.

Although this issue has been appreciated for several years (e.g., Pfleger and Eidne (1)), the potential confounding effects of competitors on the expression of the donors and acceptors are not generally allowed-for in conventional BRET competition assays. In some cases, BRET_eff is measured at a single GFP/Rluc ratio in both the presence or absence of competitors (4,5), whereas elsewhere cells have been transfected with constant amounts of donor- and acceptor-encoding DNA along with increasing amounts of competitor DNA (6,7). Irrespective of the approach used, the decreases in BRET_eff observed upon introduction of competitor in these studies have invariably been interpreted as indicating oligomerization, conclusions that will be unsafe if the competitor alters the overall levels of expression of the tagged proteins.

Here, we show that the presence of competitor proteins can profoundly alter the expression of tagged type-I membrane (TM) proteins and G protein-coupled receptors (GPCRs) in conventional BRET competition assays. We go on to establish an improved competition-based assay that avoids this source of potential artifacts, which we call the type-3 BRET assay.

In experiments replicating conventional BRET competition assays, wherein BRET_eff is measured at single expression levels in the presence and absence of competitor, introduction of competitor proteins reduced the expression of a type I TM protein (CD86) as well as the GPCRs β₂AR (β₂ adrenergic receptor), mCannR2 (mouse cannabinoid receptor 2), and GABAbR2 (γ-amino butyric acid receptor 2) (Fig. 1). Flow cytometry analysis revealed that this was not the result of changes in transfection efficiency (see Fig. S2), and it is likely instead to be caused by intrinsic signaling processes regulating protein expression. The largest decreases in expression were observed for β₂AR and mCannR2, suggesting that the expression levels of these receptors are particularly sensitive to the presence of competitors, perhaps as the result of increased levels of basal, stochastic signaling (8) introduced to the cell by the unlabeled GPCRs.

Figure 1.

Introduction of competitor proteins can result in reduction of tagged protein expression. Expression values represent combined GFP and Rluc emission normalized to Rluc emission units using the GFP/Rluc ratio. Error bars represent mean ± SE. ^∗p < 0.05; ^∗∗p < 0.01; ^∗∗∗p < 0.005 (two-tailed t-test). (Dark red) Type-I protein dimer; (dark blue) type-I protein monomer; (pale blue/pale red) GPCR.

Reduced BRET_eff in the presence of competitor was observed for all proteins with the exception of the known monomer, CD2 (see Fig. S3). For the known dimers CD28 and CD80, this is likely caused by genuine displacement of the tagged proteins from BRET-productive homodimers by unlabeled competitors, because expression of these proteins was unaffected by the competitors (Fig. 1). However, the well-characterized monomer CD86 (2,9) also exhibited a reduction in BRET_eff (see Fig. S3), which is likely attributable to decreased nonspecific interactions within the plasma membrane as a result of its reduced protein density in the presence of the competitor. In contrast, CD2 exhibits no change in BRET_eff and its expression level is unaltered in the presence of competitor. It is therefore clear that conventional BRET competition assays are subject to the confounding effects of receptor expression changes in the presence of competitors, and therefore cannot be relied upon to correctly distinguish between monomers and dimers. Changes in BRET_eff observed in these assays could result from genuine competition for position in homodimers, or changes in nonspecific interactions arising due to alterations in protein density.

To allow for the effects of changing protein density, a new competition-based assay, called the type-3 BRET assay, was established. The basis of the new assay is that the relationship between BRET_eff and protein expression level will be unaffected by the presence of competitors in the case of monomers, but will change for dimers, such that lower BRET_eff will be observed at the same expression level (see Fig. S1 b). By measuring BRET_eff at systematically varied expression levels it should, in principle, be possible to distinguish between genuine and apparent dimers. Moreover, because a range of expression levels are used in this assay, it should be more easily interpretable than some conventional BRET competition assays in which attempts were made to achieve a single, constant protein density. To test this, BRET_eff was measured for C-terminally tagged CD2, CD86, CD28, and CD80 at different expression levels in human embryonic kidney 293T cells. The expression level was varied by incubating cells transfected simultaneously for various lengths of time posttransfection, as previously described for type-2 BRET assays (2). Transfection was performed using DNA encoding the untagged, GFP-tagged, and Rluc-tagged forms of the protein of interest in the ratio 26:12:1, because this equates to a 2:1 DNA ratio of untagged to tagged protein, and ensures that an excess of GFP-tagged over Rluc-tagged protein maximizes BRET_eff. For cells in which no competitor was expressed, DNA encoding the untagged protein was replaced with the same vector but encoding no protein, to ensure equivalent transfection conditions for all cells (see the Supporting Material).

In the new assay, both CD28 and CD80 gave significantly lower BRET_eff values at a given expression level in the presence of competitor (Fig. 2 a). In addition to this change in BRET_eff/expression relationship for CD28 and CD80, both proteins gave data with a lower BRET_eff y-intercept in the presence of competitor, consistent with an increase in the apparent monomeric behavior (2). In contrast, neither CD2 nor CD86 exhibited any change in BRET_eff/expression relationship in the presence of competitor (Fig. 2 b), consistent with the predicted behavior for monomers. In all cases, the GFP/Rluc ratio remained constant across varying protein densities independent of the presence of competitor (see Fig. S4).

Figure 2.

Application of type-3 BRET to type-I TM proteins and GPCRs. Reduced BRET_eff at a given expression level was observed for the dimers CD28, CD80, and GABAbR2 in the presence of competitor (A and C), but not for the monomers CD2 and CD86 (B). The GPCRs mCannR2 and β₂AR did not exhibit a change in BRET_eff/expression relationship upon addition of competitor (C). Data are combined from three independent experiments in all cases except CD86, which constitutes four experiments.

To determine whether the assay was capable of identifying GPCR dimers, it was applied to GABAbR2, a heterodimer subunit that homodimerizes when expressed alone (e.g., (2,10,11)). In the type-3 BRET assay, GABAbR2 yielded significant differences in the BRET_eff/expression relationship in the presence and absence of competitor (Fig. 2 c), consistent with dimeric behavior. Conversely, application of the type-3 assay to β₂AR and mCannR2 yielded data consistent with monomeric behavior (Fig. 2 c). This is in agreement with our previous observations that these GPCRs behave as monomers in both type-1 and type-2 assays (2). Possible explanations for the disagreement between these observations and previous BRET-based studies of β₂AR have been discussed elsewhere (2,12).

In conclusion, these data confirm that conventional BRET competition assays are prone to artifacts that are readily circumvented in type-3 BRET assays. These findings are particularly significant in relation to BRET-based studies of GPCR stoichiometry, because our data suggest that these types of receptors may be particularly susceptible to regulated expression in the presence of higher densities of receptors expressed as competitors. It is difficult to estimate the extent to which the interpretation of previous BRET competition studies of GPCRs and other membrane proteins may have been influenced by this effect, but it seems likely to have been a factor in at least some studies. Type-3 BRET assays ought to serve as a useful adjunct to existing BRET assays (2). It has the advantage over type-2 BRET assays, in which expression is varied for fixed ratios of acceptors and donors, of not being dependent on very low expression levels, making it the more flexible assay.