Commonly used methods for investigating protein interactions, like fluorescence emission titration (a method that is generally used to obtain information about stoichiometry and binding constants) and fluorescence resonance energy transfer (FRET)–based experiments, were found to be inappropriate. This failure was due to the inherent tendency of MIA protein to form aggregates (8); moreover, a FRET experiment with an N-methylanthraniloyl–labeled AR54 derivative failed due to spectral overlap. Since dynamic light scattering (DLS), nuclear magnetic resonance (NMR), and isothermal calorimetry (ITC) were not sensitive enough to detect binding events at physiologically relevant concentrations, we decided to employ FP for elucidating the interaction of MIA protein with AR54.
FP experiments with a carboxyfluorescein-labeled derivative were compromised by nonspecific interactions of MIA protein and the respective control protein with the fluorophore. We therefore re-evaluated the choice of our assay format and decided to establish a new FP-based assay in which the protein, rather than the inhibitor, was labeled. In this scenario, we envisioned that the change in molecular weight resulting from binding would be observable only if the labeled protein of interest (MIA protein) was bound by an inhibitory compound (AR54) immobilized to a well plate (Figure 1). The FP signal should decrease after competitive displacement of labeled MIA protein from immobilized AR54 by an inhibitory compound.
As a label, we chose the luminescent Ru(bpy)3 [tris(2,2’-bipyridine)ruthenium (II)] complex due to its sufficiently long lifetime. To ensure that Ru(bpy)3 does not affect binding properties of MIA protein, we performed Boyden Chamber invasion experiments, where Mel Im cells were treated with Ru(bpy)3-labeled MIA protein [MIA-Ru(bpy)3] and, in comparison, with unlabeled MIA protein. Non-modified MIA protein reduces cell invasion by ~40–50% in this in vitro model because MIA protein specifically interferes with attachment of melanoma cells to matrigel (7). We found that unlabeled and labeled MIA protein behave identically, confirming that Ru(bpy)3-labeled MIA protein is functionally active (data not shown).Binding of MIA-Ru(bpy)3 to AR54, 30 kDa and 70 kDa fibronectin fragments
First, we measured the FP signal of MIA-Ru(bpy)3 in a well coated with AR54-biotin compared with an uncoated well. The significant increase in FP in the well coated with AR54-biotin was attributed to the severely restricted rotational mobility of MIA-Ru(bpy)3 bound to the immobilized AR54-biotin (Figure 2). In order to assess whether we could displace MIA-Ru(bpy)3 from the immobilized AR54-biotin, we treated this complex with 7.8 µM AR54 in solution. In this case, the FP of MIA-Ru(bpy)3 was almost identical to MIA-Ru(bpy)3 free in solution (in a well not coated with AR54-biotin). This demonstrates that the molecular mobility is unhindered and that the binding is reversible.
The interaction of MIA protein with fibronectin has been described previously (6). In order to test our assay with this known interaction partner, we applied 30-kDa and 70-kDa proteolytic fragments of human fibronectin, as shown in Figure 3A. As expected, FP decreased upon addition of the fibronectin fragments. Together with the AR54 results, this finding demonstrates that our HTFP assay is capable of detecting protein interactions with a small peptide as well as a 70 kDa protein.
Next, we performed a titration of MIA-Ru(bpy)3 with the 30-kDa fibronectin fragment to demonstrate that our assay is also capable of determining binding constants. As presented in Figure 3B, we determined a dissociation constant (Kd) value of 33 nM.Buffer additives and detergent controls
To assess the suitability of the HTFP assay as a screening platform for the identification of potential MIA protein inhibitors, we investigated the influence of various buffer additives and detergents commonly used in molecular biology. As expected, the addition of 0.1% Triton X-100 or 0.1% 2-mercaptoethanol disrupted the interaction of MIA-Ru(bpy)3 and AR54-biotin (data not shown). DMSO, which is often used in inhibitor screenings for dissolving compound libraries, could be tolerated for concentrations of up to 2.5%, but the addition of 50 mM EDTA induced a significant decrease in FP signal (data not shown). This can be explained by a photoinduced redox reaction involving the luminescent label Ru(bpy)3 and EDTA (24). Consistent with this proposed mechanism, a similar decrease was also observed in the absence of AR54-biotin.