NMR Lineshape Analysis of Intrinsically Disordered Protein Interactions

Abstract

Interactions of intrinsically disordered proteins are central to their cellular functions, and solution-state NMR spectroscopy provides a powerful tool for characterizing both structural and mechanistic aspects of such interactions. Here we focus on the analysis of IDP interactions using NMR titration measurements. Changes in resonance lineshapes in two-dimensional NMR spectra upon titration with a ligand contain rich information on structural changes in the protein and the thermodynamics and kinetics of the interaction, as well as on the microscopic association mechanism. Here we present protocols for the optimal design of titration experiments, data acquisition, and data analysis by two-dimensional lineshape fitting using the TITAN software package.

Keywords: Binding; IDP; Kinetics; Nuclear magnetic resonance; Titrations.

Fig. 1

(a) Definition of the exchange rate, k _ex, and frequency differences, Δω ^H and Δω ^N, for a protein-ligand interaction observed in a two-dimensional heteronuclear correlation experiment. (b) ¹H, ¹⁵N-HSQC spectra and projected ¹H 1D cross sections for a simulated protein-ligand interaction (700 MHz, 1 mM protein concentration, K _d 2 μM, Δω ^H 4400 s^–1 [1 ppm], Δω ^N 220 s^–1 [0.5 ppm]) illustrating lineshapes that may arise under various exchange regimes, as indicated. Contour levels are constant across all spectra. Adapted from [40]

Fig. 2

Screenshot of the main TITAN interface. The workflow, indicated by arrows, is progressively enabled as the user proceeds through the analysis

Fig. 3

Selection of a binding model. The two-state binding model selected here is suitable for simple protein-ligand interactions, but a variety of additional models are available as shown here and described in Table 1

Fig. 4

(a) Titration setup dialog. Protein and ligand concentrations (as required by the particular binding model selected) must be specified together with the acquisition parameters ns (number of scans) and rg (receiver gain). Spectrum noise levels are required to accurately weight residuals between spectra and are calculated automatically upon importing data. (b) For convenience, data for this dialog may be copied and pasted from a suitable Excel spreadsheet

Fig. 5

Pulse sequence setup dialog. All fields should be parsed automatically from NMRPipe input files, except those highlighted yellow, for which appropriate values should be set by the user

Fig. 6

Screenshots of the ROI editor illustrating the setup of a new spin system. (a) When first opened, the editor shows an overlay of all spectra in the left-hand panel which may be zoomed and panned using the controls in the toolbar (toolbar not shown). (b) ROIs are marked out as a series of points in the right-hand panel. (c) Final state of the editor after selecting initial estimates of the free and bound state peak positions

Fig. 7

Setting up spin groups for the fitting of overlapping resonances. (a) Two residues, D223 and F229, have been defined with overlapping ROIs within the ROI editor. (b) The overlapping residues have been associated with a common “spin group” within the spin system editor (red circle). The spin group is an arbitrary text label. ROIs for residues within the same spin group will be merged, and (overlapping) resonances therein will be fitted simultaneously

Fig. 8

Screenshot of the model parameters editor. Initial values for parameters such as K _d and k _off can be specified, together with the allowable parameter range (e.g., which may be constrained on the basis of prior knowledge). Fitting of particular parameters may be activated and deactivated using the checkboxes

Fig. 9

Visualization of fitting results. (a) Overlaid contour plots of observed and fitted spectra (blue and red, respectively). (b) Three-dimensional views of observed and fitted spectra (gray and magenta, respectively)

Fig. 10

Density plot of the parameter covariance matrix derived from bootstrap error analysis. Parameter IDs are listed within the fitting output and can be explored interactively using the mouse cursor

Fig. 11

Optimal selection of ROIs. (a) Recommended setup: ROIs extend approximately two to three linewidths from the center of resonances and contain the entire region of the spectrum within which resonances are observed across the titration. Note that ROIs are identical for all spectra; hence only a single boundary can be observed. (b) Not recommended: individual ROIs do not encircle the entire region within which resonances are observed across the titration. (c) Not recommended: too large a selection, resulting in slow fitting, increased noise, and overlap with an adjacent residue. (d) Not recommended: too tight a selection, limiting accuracy when fitting linewidths

Fig. 12

Setting up shared parameters within the spin system editor. The example shown here is a symmetric dimer, defined by linking all properties of the asymmetric dimer states D1 and D2. For each of the D1 and D2 states, the direct and indirect chemical shifts, dI and dS, and the direct and indirect linewidths, R2I and R2S, are assigned to the global parameters “shared 1” to “shared 4” as indicated