Calcium-Binding Protein Protocols Calcium-Binding Protein Protocols

290 Werner et al.
tion, using gradient echoes, yields improved signal-to-noise ratio and good water
suppression (36).
3.2. Data Processing and Analysis
1. All spectra recorded in one series are processed identically. Linear prediction (in
the 15N dimension) and resolution enhancement (in the 1H and 15N dimensions)
can be used to reduce overlap of adjacent peaks (see Note 6) (37,38). In the
cbEGF32–33 pair, the data were zero-filled to obtain a digitization of less than
about 4 Hz (in F2: 1H) and about 2 Hz per point (in F1: 15N). 2. The nonoverlapping peaks in the most intense [ 1H]– 15N autocorrelation spectrum
are picked and assigned and the intensities of these peaks are extracted from each
spectrum (see Note 7).
3. Peak height uncertainties are estimated, either from the baseline noise or from
the standard deviation of the difference in peak intensities of two spectra recorded
with the same relaxation delay (see Note 8).
4. The series of [ 1H]– 15N correlation peak intensities from the respective experiments
are used to obtain the relaxation time-constants of each residue by χ2 minimization
of monoexponential decay functions with initial intensity Io and decay
constants T1 or T2 as free parameters
I(t) = Ioe –t / T1,2 (6)
Monte Carlo simulations are performed to estimate the uncertainties of the free
parameters. The model is accepted if χ2 of the best fit is within a chosen (typically
the ninety-fifth) percentile of the χ2 distribution of the simulated data (see
Notes 7 and 9).
5. In the heteronuclear NOE experiments, errors of the peak intensities in spectra
with and without NOE are estimated from baseline noise for the respective spectrum.
In the presence of multiple pairs of heteronuclear NOE spectra averages
and standard deviations can be taken over the intensity ratios.
6. The NOE ratio is calculated as the ratio of the intensities in the spectrum with
saturation, Is, and without saturation, Ins, NOE = Is/Ins, and the uncertainty is
obtained by error propagation:
σNOE = Is / Ins √(σIs / Is) 2 + (σIns / Ins) 2 .
7. The rotational correlation time of the entire molecule is obtained from the average
of the T 1/T 2 ratios (28). In case substantial numbers of residues are affected
by exchange or fast motion, the average over a subset of residues representative
of overall diffusion is used (see also Subheading 3.3., step 2).
8. A plot of the T 1 vs the T 2 value of each residue overlaid on parametric curves of
T 1 and T 2 as functions of correlation time τ c and order parameter S 2 is a convenient
way of obtaining a qualitative assessment of the data (see Fig. 2). Residues
whose T 1 and T 2 values fall on a straight line through the origin, with slope T 1/T 2
can be described by a single correlation time. Residues with T 1 and T 2 values that
are shifted to the right (i.e., small T 2 values) outside the theoretical curves are