Nuclear magnetic resonance relaxation analysis is a powerful
method
for studying the internal mobility of proteins. We have developed
for analysis the slowly relaxing local structure (SRLS) approach.
SRLS is general in its nature in several respects, including the tensorial
representation of the physical quantities comprising the dynamic model.
By controlling tensor symmetry, a broad range of systems can be treated
with physical relevance, typically with data-fitting techniques. In
simple limits, SRLS yields the traditional model-free (MF) method.
In the present context, MF simplicity means featuring the highest
possible tensor symmetry. This renders MF-based data-fitting susceptible
to the usage of fit parameters, yielding physically ill-defined results.
A typical candidate is the R
ex term, devised
to represent ms-μs motions but often invoked
by the fitting scheme just to improve the statistics. Here, we consider
two such cases using the N–H bond as probe and the proteins
p50 and human neutrophil gelatinase-associated lipocalin as paradigm
systems. We illustrate the harm caused by the physically unjustified
involvement of R
ex in MF-based 15N relaxation analysis. Then, we show that forgoing the usage of R
ex, SRLS analysis of the very same experimental
data provides interesting new information.