Role of electrostatic interactions in transient encounter complexes in protein-protein association investigated by paramagnetic relaxation enhancement.

Protein-protein association can be viewed as a two-step process comprising the initial formation of an encounter complex ensemble followed by rearrangement, along a two-dimensional energy landscape, to form the final well-defined stereospecific complex (Figure 1).1-5 Theoretical work suggests that electrostatic interactions play an important role in encounter complex formation, thereby enhancing molecular association by permitting a reduced dimensionality search until the stringent orientational requirements for specific association are met.6 Recently, intermolecular paramagnetic relaxation enhancement (PRE) has been used to directly visualize an ensemble of lowly populated, highly transient encounter complexes in rapid exchange with the stereospecific complex.3-5 In this exchange regime, the observed intermolecular PREs are weighted population averages of the PREs of the species present, and depending on paramagnetic center-proton distances, species with an occupancy as low as 1% can be detected.3,4 For three relatively weak complexes (KD ∼ 1-20 μM) from the bacterial phosphotransferase system, the distribution of nonspecific encounter complexes appeared to be qualitatively correlated to the electrostatic surface potentials of the interacting proteins.4 In this paper, we extend our previous work on the complex of the N-terminal domain of enzyme I (EIN) and HPr4,7 to examine the ionic strength dependence of intermolecular PREs and provide direct experimental evidence that the interactions involved in the formation of shortlived encounter complexes are predominantly electrostatic in nature. NMR samples4 comprised 0.3 mM U-[2H/15N]-EIN and 0.5 mM HPr in 20 mM Tris-HCl buffer, pH 7.4, with the concentration of NaCl ranging from 0 to 0.5 M. HPr was paramagnetically labeled with EDTA-Mn2+ conjugated via a disulfide linkage to surfaceengineered cysteine residues at either position 5 (E5C) or 25 (E25C).4 Intermolecular backbone HN PRE rates, Γ2, for EIN were obtained by taking the difference in R2 relaxation rates of paramagnetic and diamagnetic samples, measured at 40 °C using a TROSY-based pulse scheme8 at a 1H frequency of 600 MHz. The paramagnetic label at E25C of HPr is located relatively close to the binding interface of the stereospecific EIN/HPr complex (Figure 2C), while the label at E5C is located on the opposite surface of HPr (Figure 2A). The observed Γ2 rates are weighted averages of the Γ2 rates for the stereospecific complex and the ensemble of nonspecific encounter complexes.4 The former is populated at g90%.4 To a good approximation, the PREs can be readily partitioned into those arising from the stereospecific complex and those from the ensemble of nonspecific encounter complexes using the following criteria: (1) residues with Γ2 > 5 s-1 for which there was no significant measurement error due to spectral overlap or line-broadening were selected for analysis; (2) observed Γ2 rates in agreement with those back-calculated on the basis of the stereospecific complex4 were attributed to the stereospecific complex (residues 115-125 for the EIN/E5C-HPr complex and residues 40-80 and 127-135 for the EIN/E25C-HPr complex), while the remainder were considered to originate primarily from the ensemble of nonspecific encounter complexes (residues 5891 for the EIN/HPr-E5C complex, and residues 89-92 and 107123 for the EIN/E25C-HPr complex) (Figure 2A,C). The effect of salt on the magnitude of the PREs is displayed in Figure 2B (EIN/E5C-HPr) and 2D (EIN/E25C-HPr) as correlation plots of the Γ2 rates at various salt concentrations versus the corresponding Γ2 rates at 0 M NaCl. For the PREs attributed to the stereospecific complex, only a weak salt dependence is observed (top panels): the data at 0.15, 0.3, and 0.5 M NaCl versus 0 M NaCl exhibit slopes of 0.93 ( 0.01, 0.89 ( 0.03, and 0.86 ( 0.03 for the EIN/E5C-HPr complex, and 0.92 ( 0.02, 0.91 ( 0.02, and 0.85 ( 0.03 for the EIN/E25C-HPr complex, respectively. (The corresponding correlation coefficients are 0.99, 0.92, 0.94, 0.97, 0.98, and 0.96.) For the PREs attributed to the ensemble of nonspecific encounter complexes, however, a much larger salt dependence is observed with values of the slopes for the 0.15, 0.3, and 0.5 M NaCl data versus 0 M NaCl data of 0.90 ( 0.03, 0.71 ( 0.02, and 0.63 ( 0.03, respectively, for the EIN/E5C-HPr complex, and 0.84 ( 0.05, 0.69 ( 0.04, and 0.60 ( 0.05, respectively, for the EIN/E25C-HPr complex. (The corresponding correlation coefficients are 0.98, 0.98, 0.96, 0.88, 0.86, and 0.79, respectively.) Importantly, the observed salt dependencies are the same within experimental error for the data obtained on both the EIN/E5C-HPr and EIN/E25C-HPr complexes, indicating that the data from the two samples are reporting on the same overall interactions. The decrease in magnitude of the PREs arising from the stereospecific complex as a function of salt is attributable to an increase in equilibrium dissociation constant, KD (largely due to a decrease in the association rate constant6d), and hence decrease in the population of stereospecific complex as the salt concentration increases, as confirmed by isothermal titration calorimetry (ITC) experiments (see Supporting Information for details). As is evident from Figure 3A, the log-log plot of KD versus NaCl concentration shows the expected linear dependence over the 0.15-2 M NaCl range.9 Under the conditions of the NMR experiments, this translates Figure 1. Protein-protein complex formation. The interaction surfaces involved in the stereospecific complex are shown in white and yellow. Published on Web 10/06/2007