NMR Spectroscopy of supramolecular chemistry on protein surfaces

Abstract

As one of the few analytical methods that offer atomic resolution, NMR spectroscopy is a valuable tool to study the interaction of proteins with their interaction partners, both biomolecules and synthetic ligands. In recent years, the focus in chemistry has kept expanding from targeting small binding pockets in proteins to recognizing patches on protein surfaces, mostly via supramolecular chemistry, with the goal to modulate protein-protein interactions. Here we present NMR methods that have been applied to characterize these molecular interactions and discuss the challenges of this endeavor.

Keywords: NMR; molecular recognition; protein ligand interaction; protein surfaces; supramolecular chemistry.

Figure 1

Ligands targeting charged areas on protein surfaces discussed in this review. The protein shown as example is hPin1 (pdb 1NMV, [75]). The guanidiniocarbonylpyrrole motif (GCP, top left) recognizes the carboxylate groups of aspartate or glutamate residues by forming an extended H-bonding network [–46]. Supramolecular tweezers (top center) thread lysine or arginine side chains into their aromatic cavities [–18]. Calixarenes (top right, [–40]), Ru^II(bpy)₃ complexes (bottom left, [–63]) and porphyrins (bottom center, [–60]) can be functionalized with either multiple acidic or basic groups to target charged areas of either polarity on a protein surface. Cucurbiturils (bottom right, [–51]) recognize methylated lysines and arginines by binding their methylated head groups inside the macrocycle.

Figure 2

¹H NMR titration of lysine with tweezers. All signals show chemical shift perturbations and different degrees of line broadening. The protons that are located right inside the tweezer cavity upon binding experience the largest effects. The asterisk marks impurities in the tweezers stock solution. Reprinted (reproduced) with permission from [80], copyright (2017) Wiley-VCH Verlag GmbH & Co. KGaA.

Figure 3

¹H,¹⁵N-HSQC Titration of full-length hPin1 with supramolecular tweezers (original data). (a) Spectra for different tweezer concentrations are overlaid and color-coded. Ligand binding causes a concentration-dependent shifting and/or broadening of the signals. At a 1:1 ratio of ligand to protein (red spectrum), many signals broaden. This can indicate the formation of soluble aggregates. (b) The expansion shows the area marked by the dashed box. Residue Q94 experiences a shifting of the signal upon tweezer binding (chemical shift perturbation). Residues R21, N30, K77 and Q129 show strong line broadening due to chemical exchange, which leads to a vanishing of the signals at low tweezer concentrations. Residue E35 shows both chemical shift perturbation and line broadening. Residue E87 is not affected by tweezers. (c) NMR Structure of hPin1 (pdb 1NMV, [75]) with lysines highlighted in cyan and arginines highlighted in magenta.

Figure 4

Relative signal intensities can be used to identify ligand binding sites (schematic representation of a hypothetical titration). (a) Binding of the ligand results in a reduction of the relative signal intensities I/I₀ (protein with ligand vs. protein alone) of residues in proximity of the binding site due to line broadening. (b) The stabilizing effect of a ligand on a protein–peptide interaction in a ternary complex results in an additional decrease of relative signal intensities for the protein residues affected (red), compared to the binary complex of protein and peptide in the absence of a ligand (blue).

Figure 5

Schematic ¹H,¹⁵N-HSQC spectrum of tauF4 (chemical shifts from BMRB # 17945, [109]) with and without specific ¹⁵N-lysine labeling. (a) Uniformly ¹⁵N-labeled protein. The amide NH of all residues (except Pro) yield a signal, resulting in signal overlap. (b) Selective ¹⁵N-lysine labeling. Only the amide NH of lysine residues are visible. The specific labeling significantly reduces signal overlap and thus makes it easier to track the shifting of single resonances upon ligand binding.

Figure 6

H2(C)N spectra specific for arginine (a) and lysine (b) residues of the hPin1-WW domain at different tweezers concentrations (color-coded). The signals for R14 and R21 are both split and overlap. Upon tweezer binding, line broadening and thus reduced signal intensities are observed. (c) Plotting the relative signal intensities for each signal as a function of tweezers concentration (given in equivalents) reveals a distinct binding order. R17 is the preferred binding site, while R36 and the N-terminus are not bound at all. (d) Structure of the hPin1-WW domain with lysines highlighted in cyan and arginines highlighted in magenta. Reprinted (reproduced) with permission from [80], copyright (2017) Wiley-VCH Verlag GmbH & Co. KGaA.