Slow ring flips in aromatic cluster of GB1 studied by aromatic 13C relaxation dispersion methods

Abstract

Ring flips of phenylalanine and tyrosine are a hallmark of protein dynamics. They report on transient breathing motions of proteins. In addition, flip rates also depend on stabilizing interactions in the ground state, like aromatic stacking or cation-π interaction. So far, experimental studies of ring flips have almost exclusively been performed on aromatic rings without stabilizing interactions. Here we investigate ring flip dynamics of Phe and Tyr in the aromatic cluster in GB1. We found that all four residues of the cluster, Y3, F30, Y45 and F52, display slow ring flips. Interestingly, F52, the central residue of the cluster, which makes aromatic contacts with all three others, is flipping significantly faster, while the other rings are flipping with the same rates within margin of error. Determined activation enthalpies and activation volumes of these processes are in the same range of other reported ring flips of single aromatic rings. There is no correlation of the number of aromatic stacking interactions to the activation enthalpy, and no correlation of the ring's extent of burying to the activation volume. Because of these findings, we speculate that F52 is undergoing concerted ring flips with each of the other rings.

Keywords: Aromatic interaction; NMR spectroscopy; Protein breathing; Protein dynamics; Protein stability.

Fig. 1

Three-dimensional structure of GB1 (1pgb.pdb) shown as ribbon presentation. Phe and Tyr side-chains are shown colored in stick representation and are labeled accordingly

Fig. 2

Intensity of aromatic signals that can be affected by ring flips (Phe and Tyr δ and ε). Y3 is shown in blue, F30 in magenta, Y33 in grey, Y45 in cyan and F52 in red. Normalized relative intensities of δ (a) and ε (b) are plotted against the temperature. Intensities of − 5 °C and 200 MPa are plotted at − 15 °C, since going from 0.1 to 200 MPa has roughly the same effect on the rate of ring flips than lowering the temperature by 10 K. Here the intensities of the two individual signals (δ1 and δ2, or ε1 and ε2) are the same within the symbol size. In all other cases, only averaged signals δ* and ε* (or no signals) could be observed

Fig. 3

Region of a Tyr δ* (Y3 and Y33), b Tyr ε* (Y3 and Y33) and c Phe ε* (F30 and F52) in the aromatic ¹H¹³C-TROSY-HSQC of GB1 at 30 °C (red), 25 °C (orange), 20 °C (yellow), 10 °C (green) and ambient pressure. The spectrum at − 5 °C and 200 MPa is shown in blue, where split signals (δ1 and δ2, or ε1 and ε2, respectively) can be observed. Signals indicated as # are caused by sample impurities which can be detected at very high S/N experiments, which were needed for the − 5 °C and 200 MPa condition, where the split signals are still severely broadened

Fig. 4

Aromatic ¹³C L-TROSY-selected R_1ρ relaxation dispersions recorded on-resonance (tilt angle θ > 85°) at a static magnetic field-strength of 14.1 T. Dispersion profiles for Y3δ at 25 °C (a), F30δ at 35 °C (b), Y45ε at 20 °C (c) and F52ε at 10 °C (d) are shown. Data were fitted with fixed populations p₁ = p₂ = 0.5 and free (Y45) or fixed chemical shift differences ∆δ_disp derived from low temperature and high pressure spectra. The resulting flip rates are: (12 ± 2) × 10³ s⁻¹, (53 ± 4) × 10³ s⁻¹, (6 ± 2) × 10³ s⁻¹ and (4.8 ± 0.9) × 10³ s⁻¹, respectively

Fig. 5

Temperature dependence of flip rates. k_flip is plotted as a function of 1/T for F52 (red), Y3 (blue), Y45 (cyan) and F30 (magenta). The fits are displayed as solid lines, while the uncertainties of the fits are displayed as shaded areas in the appropriate colors. The data are represented using a logarithmic y-axis to show the expected linearity, but the fit was performed using non-linear regression of k_flip on T

Fig. 6

Pressure dependence of flip rates. k_flip is plotted as a function of pressure for F52 (20 °C, red), Y3 (30 °C, blue) and Y45 (30 °C, cyan). The fits are displayed as solid lines, while the uncertainties of the fits are displayed as shaded areas in the appropriate colors. The data are represented using a logarithmic y-axis to show the expected linearity, but the fit was performed using non-linear regression of k_flip on p

Fig. 7

Activation enthalpy of ring flips for certain scenarios. a Activation enthalpy of a ring without stabilizing contacts. b Activation enthalpy of a ring with stabilizing contacts, in this case a stacking ring (shown in red). The stabilization of the ground state is between 5 and 10 kJ mol⁻¹ (Burley and Petsko 1989). c Activation enthalpy of a ring with stabilizing contacts of a stacking ring (shown in red), both rings are undergoing concerted ring flips