N uclear magnetic resonance (NM R) spectroscopy is applicable to atomic nuclei
that possess a magnetic moment. In biology, these are typically nuclei w ith odd-
num bered m asses {e.g. ’H, and '^N). Such a nucleus m ay be regarded as a spinning,
field along its spinning axis. If an external magnetic field is placed around the nucleus,
the nucleus will rotate around an axis that is either parallel or antiparallel to the direction
o f the applied field and thus corresponding to upper and lower energy levels. The
energy difference between these two spin states is characteristic o f the particular type o f
nucleus and the strength o f the applied m agnetic field. The application o f radio
frequency electrom agnetic radiation to an ensemble o f spins at the appropriate frequency
w ill induce resonance between the two spin states, leading to the generation o f
detectable transverse m agnetization (M acArthur et a l, 1994; Evans, 1995). The real
value in N M R lies in the fact that the magnetic field experienced by the nucleus is not
the applied field because the applied external magnetic field is m odified by the fields o f
neighbouring atoms. This effect is known as the chemical shift and gives rise to the
different positions o f N M R signals in the spectrum that are characteristic o f the
environment o f a specific proton (M acArthur et a l, 1994; Evans, 1995).
In conventional pulse Fourier experim ents, the sample is subjected to a short radio pulse whose frequency is centred in and covers the spectral region o f interest. The
generated output signal (or free induction decay) contains oscillating signals from all resonances as a function o f time t. This is then converted by a Fourier transform ation
to give a one-dim ensional spectrum o f signal intensity as a function o f resonance
frequency (Chang, 1981; Evans, 1995; W illard et a l , 1981). By repeating the
experim ent m any times, in which an additional increm entable tim e interval t, is set
between the first pulse and a second short pulse, the data can be processed to obtain the
N M R spectrum as a function o f two frequencies (j. e. a tw o-dim ensional N M R spectrum;
W illard ef a/., 1981; Evans, 1995).
Two-dim ensional N M R experiments are used for structure determination, and
different tw o-dim ensional experiments reveal different interactions betw een hydrogens that are spatially close to each other. The tw o m ajor tw o-dim ensional experiments are
the C O SY experiment, which gives signals that correspond to hydrogen atoms that are
covalently connected through one or two other atoms (known as J-couplings), and the
NO ESY experim ent, w hich gives signals that correspond to hydrogen atom s that are
X -R a y
C rystallization and derivative preparation D ata co llectio n and processing
i
Location o f heavy atom s (or m olecular replacem ent) C alculation o f phases and electron-density m apping
:
C hain tracin g and interpretation
i
M odel b u ild in g R e fin em en t o f model
PROS
• W ell-establised technique.
• M ore m athem atically d irect im age construction. • M ore o bjective interpretation o f data.
• R a w -d ata pro cessin g h ig h ly autom ated.
• Q u a lity indicators available (resolution, R-factor). • M utants, different ligands and hom ologous structures (as low as 2 5 % seq u en ce identity) readily com pared by difference Fourier m ethods.
• L arge m olecules and assem blies can be determ ined, e.g. virus particles.
• S urface w ater m olecules relatively well defined. • Produces a single m odel th at is easy to visualize and interpret.
CONS
• Protein has to form stab le crystals th at d iffract well. • N e e d heavy-atom derivatives th a t form isom orphous crystals.
•C ry s ta l production can be d ifficu lt and tim e c o n su m in g and often im possible.
• U nnatural, nonphysiological environm ent.
• D ifficulty in ap p o rtio n in g u n certain ty betw een sta tic and dynam ic disorder.
• S urface residues m ay be influenced b y crystal packing. • M ay n o t w holly represent structure as it exists in solution. • Less useful for large flexible m o d u lar pro tein s, e.g com plem ent factors B , I and H.
• M odel represents a tim e-averaged structure w here d etails o f m o b ility are unresolved.
N M R
S am ple p rep aratio n w ith possible isotope labelling D ata collection (N O E S Y and C O S Y spectra)
i
S equential assignm ents
A nalysis and q u an tific atio n o f N O E peak intensities and conversion to ap p ro x im ate p roton-proton separations G eneration o f m odels co n siste n t w ith the N O E -derived
seperations
G eneration o f m odels co n sisten t w ith the N O E -derived sep aratio n s and to rsio n -an g le ranges from coupling constants,
u sually by d istan c e geom etry and sim ulated annealing algorithm s
M odel im p ro v em en t by inclusion o f initially unassigned N O E d istan ces and stereo sp ccific assignm ents
P R O S
• C lo s e r to biological conditions.
• C a n provide inform ation on d ynam ics and identify individual side-chain m otion or d isorder (especially in loops)
• S econdary stru ctu re can be d erived from lim ited experim ental data.
• F re e from artefacts re su ltin g from crystallization.
• C a n be used to m o n ito r co n fo rm atio n al change on ligand binding.
• G o o d for c h eck in g the correct fold o f m utants. •I d e a l for sm all dom ains.
• Solution co n d itio n s can be explicitly chosen and readily changed, e.g. pH , tem p eratu re, etc.
•U s e fu l for p ro tein -fo ld in g studies.
therefore dan g er o f C O N S
•R e q u ire s co ncentrated solution aggregation or oligom erisation.
•C u rre n tly lim ited to determ in atio n o f relatively sm all proteins (O O K D a )
• L a c k o fe sta b lise d quality in d icato rs o f d a ta and m odel, such as resolution and R -facto r
• A w eaker and m ore su b je ctiv e interpretation o f the experim ental d a ta th an in X -ray crystallography
• S urface residues generally less w ell defined th a n in X -ray crystallography
• T h e distin ctio n betw een flex ib ility and lack o f d a ta is not alw ays easy
•P ro d u c e s an en sem b le o f 10-15 p o ssib le structures rather than one m odel ie a tim e average stru ctu re
•C o n fo rm a tio n a l v ariab ility can m ak e interpretation difficult •C o m p le te stru ctu re d eterm in atio n req u ired if hom ology is less th an - 6 0 % se q u en ce identity
Table 2.1. Stages in the determ ination o f a protein structure by X-ray crystallography
and N M R and their characteristics. (Based on M acA rthur et al., 1994, reproduced from
Overhauser effects). The signals from these experim ents are assigned to the protein
sequence in order to obtain distance ‘constraints’ from specific hydrogen atoms in one
residue to hydrogen atoms in a second residue. A series o f restrained m inim ised models
are constructed that are consistent with these distance constraints and can be used to
study low m olecular weight (M^) proteins (up to 12 kDa) to atomic resolutions. N M R
is the only high resolution technique to determine the three dim ensional structure in
solution in an environm ent near to physiological. Problem s in the separate resolution
o f N M R signals arise with larger proteins. However, new techniques including three-
dimensional N M R and the isotopic labelling o f protein samples have been used to
determ ine structural information on proteins o f up to 30 kD a (Fesik et al., 1989).
N M R is ideal for large multidom ain proteins which can be studied by isolating each
folded domain independently. N M R structures o f several SCR dom ains have been
elucidated (Chapter 4). N M R is therefore o f use in studying m ulti-dom ain proteins consisting o f small domains o f less than 100 residues. As o f 17 D ecem ber 2002; 2,450
N M R structures had been deposited in the Protein D ata Bank.