Solution structure
NMR spectra were acquired at 27 °C on Bruker Avance 800 and DRX500 spectrometers, equipped with triple-resonance (1H/15N/13C) cryoprobes. All data were acquired using a sample containing 2 mM 15N/13C-labeled protein, 25 mM sodium phosphate, pH 6.0, 10 mM sodium chloride, and 10% 2H2O and comprised the following: 2D: [15N–1H] heteronuclear single quantum coherence (HSQC), [13C–1H] HSQC covering full 13C spectral width, constant-time [13C–1H] HSQC covering only aliphatic 13C region, [1H–1H] NOE spectroscopy (NOESY) experiments (τm = 50, 100, and 150 ms), [1H–1H] NOESY experiments filtered to remove 15N-coupled signals in F2 (τm = 50 and 150 ms); 3D data sets: CBCANH, CBCACONH, HBHANH, HBHACONH, [1H–13C–1H] HCCH–total correlated spectroscopy, [1H–13C–1H] HCCH–correlated spectroscopy, 15N NOESY–HSQC (τm = 50 and 150 ms), and 13C NOESY–HSQC (τm = 50 and 150 ms), with separate data sets acquired for 13C aliphatic and aromatic spectral regions. 1H, 15N, and 13C chemical shifts were calibrated using sodium 3,3,3-trimethylsilylpropionate as external 1H reference.39
NOE distance restraints were derived from analysis of all of the data from NOE-based experiments. Cross-peak intensities were measured using the program SPARKY40 and grouped into four categories. The strongest dαN (i, i + 1) were used to set the upper limit for the category “very strong” (0–2.3 Å), strong dNN (i, i + 1) connectivities in α-helices defined the category “strong” (0–2.8 Å), dαN (i, i + 3) cross peaks in helices defined the category “medium” (0–3.5 Å), and all remaining peaks were classified as “weak” (0–5 Å). Lower bounds for all NOE restraints were set to zero,41 and no multiplicity corrections were required since r− 6 summation was used for restraints involving groups of equivalent or nonstereoassigned spins.33,42
Structures were calculated from polypeptide chains with randomized ϕ and ψ torsion angles using a two-stage simulated annealing protocol within the program XPLOR, essentially as described elsewhere,43,44 but employing larger numbers of cycles as follows: first-stage calculations comprised Powell energy minimization (500 steps), dynamics at 1000 K (25,000 steps), increase of the van der Waals force constant and tilting of the NOE potential function asymptote (4000 steps), switching to a square-well NOE function then cooling to 300 K in 2000 step cycles, and final Powell minimization (1000 steps). Second-stage calculations used Powell minimization (500 steps), increasing dihedral force constant during 4000 step cycles of dynamics at 1000 K (with a strong van der Waals force constant and square-well NOE potential function), cooling to 300 K in 1000 step cycles, and 2000 steps of final Powell minimization.
The program CLUSTERPOSE was used to calculate the mean rmsd of ensembles to their mean structure.45,46