Contemporary Methods in Structure Determination of Membrane Proteins by Solution NMR - NMR of Proteins and Small Biomolecules - page 151

structure determination. The introduction of transverse relaxation optimized spec-
troscopy (TROSY) by Wuthrich and co-workers [ 276 ] opened up a wealth of new
opportunities for solution NMR on large protein systems, including detergent-
solubilized membrane proteins.
First demonstrated for amide 1 H- 15 N[ 276 ] followed by aromatic 1 H- 13 C groups
[ 277 ], the basic idea exploits interference effects between the two dominant
mechanisms of relaxation, namely dipole-dipole coupling and chemical shift
anisotropy (CSA). The effect of this interference can be seen in an uncoupled
COSY spectrum, which for a directly bonded 1 H- 15 N pair gives rise to a doublet
in both the 1 H and 15 N dimensions, each separated by the one-bond coupling
constant 1 J HN . In this quartet, one peak will exhibit a narrower line width relative
to the other peaks as a result of partial cancellation of dipole-dipole and CSA
relaxation pathways. This effect is most pronounced at higher magnetic field
strengths, with maximal cancellation occurring at ~1 GHz proton frequency for
both 1 H and 15 N magnetization. In TROSY-type NMR pulse sequences, the coher-
ence that gives rise to the slow-relaxing component of the quartet is preserved while
the other components are either eliminated [ 276 ] or allowed to decay to negligible
levels during the course of the experiment [ 278 , 279 ]. Although all four coherences
would make contributions to the single 1 H- 15 N peak in the decoupled correlation
spectrum that is normally run for smaller proteins, interconversion between states
that undergo very fast vs slow relaxation give rise to magnetization losses that are
dominated by the rapidly decaying coherence. Consequently, for larger proteins
(~50 kDa and higher) the signal lost by discarding three parts of the quartet is
compensated by the increase in resolution and sensitivity that comes from avoiding
the more rapidly relaxing states during acquisition.
Numerous multidimensional triple-resonance NMR experiments have since
been designed that isolate this slow-relaxing component during evolution and
acquisition periods (reviewed in [ 280 ]). Additional strategies that take advantage
of the unique relaxation properties of slowly tumbling proteins have since been
described that provide additional sensitivity gains for large systems (e.g., polariza-
tion transfer schemes that use cross-correlated relaxation [ 281 , 282 ], or longitudinal
relaxation optimization [ 283 ]). This has allowed backbone assignments to be
obtained for many proteins exceeding 40 kDa, including a number of large mem-
brane protein-detergent complexes [ 110 , 284 ]. This has usually required the use of
deuterium-labeled samples to eliminate 1 H- 1 H dipole dipole interactions that also
contribute significantly to transverse relaxation rates. In some cases it can be
possible to skip deuterium labeling, particularly when only 2D 1 H- 15 N correlation
spectra are required, as was the case when screening sample conditions for DAGK
[ 285 ]. However, the ability to obtain backbone resonance assignments for typical
membrane protein samples requires that the TROSY effect be maximized through
the use of uniform deuterium incorporation and high spectrometer field strengths
(i.e., 700 MHz and greater). Even when these conditions are fulfilled, peak
intensities in the TROSY spectrum may be attenuated by microsecond to millisec-
ond timescale exchange processes. This has been commonly observed for mem-
brane proteins in detergent micelles [ 37 , 70 , 111 , 209 , 285 - 287 ], making the