Diagrama de flujo

The development of multi-dimensional NMR is crucial to the application of NMR to studies of protein. The first step in any biomolecular NMR studies is to assign the resonance frequency of particular nuclei. However, even in a small protein, there are hundreds of proton resonances which inevitably leads to severe overlap in the spectrum, making assignment impossible in one-dimensional (ID) spectroscopy. This problem of spectral overlap can be alleviated by dispersing the spectrum into a second dimension. Resonances that were previously overlapped may be resolved, thereby aiding the assignment of the resonance.

However, the conventional proton-based homonuclear NMR methods are limited by the size of the protein; at a higher molecular mass (>10kDa), the shorter proton relaxation times and the larger number of resonances lead to reduced magnetisation transfer efficiency, greater chemical shift overlap and degeneracy, and spin diffusion effects. These problems can be circumvented by making use of heteronuclear couplings in ^^N- and *^C-labelled proteins and further increasing the dimensionality. Even in a protein that is only enriched in *^N or the conventional homonuclear two dimensional (2D) experiments can be edited with respect to the ^^N or chemical shift, thereby simplifying the spectrum considerably. For a protein that is uniformly enriched in both in *^N and by using ^^C-edited and triple resonance NMR techniques, the spin system identification

and sequential assignment can be dramatically simplified. Heteronuclear NMR techniques also provide additional parameters such the heteronuclear scalar coupling constants and chemical shifts which are useful in the structural and biophysical analysis of protein. NMR methods making use of the ^^N and labelling are called triple resonance techniques as the resonances are always recorded together with the *^N and resonances. Such methodology allows the study and determination of protein structure up to -30 kDa. At higher molecular mass, however, the transverse relaxation time decreases with increase in molecular mass and correlation time, resulting in broadened peaks. This may be overcome by per-, selective or fractional deuteration of the protein, which can simplify the spectrum by ‘diluting’ the protons present, and enhance the NMR spectrum by significantly reducing the line-width by increasing the relaxation times (Sattler & Fesik 1996). Such advances in heteronuclear multidimensional NMR techniques have made possible the assignment of resonances and structural determination of larger proteins with increasing precision and accuracy, and a large number of structures of proteins and protein complexes with molecular mass 10-20 kDa, with some of 30 kDa and over, have been reported in the past few years (Garrett et al, 1999; Kelly et al, 1999). Backbone resonance assignments of a multimeric protein with molecular weight in excess of 60 kDa have also been reported (Shan et ai, 1996).

The multidimensional NMR experiments is made possible by a number of developments in the last twenty years or so: the progress in recombinant gene technology that permits ^^N and isotopically-labelled protein to be produced routinely by heterologous expression, the development of complex pulse sequences to transfer magnetisation between scalar J (through bond) and dipolar (through space) coupled spins, the availability of fast computing power and better spectrometers with improved radio frequency electronics, and the increased magnetic field strength as well as the development of commercially available hardware such as pulsed field gradients. Current research to further extend the capability of NMR is still proceeding apace, with novel applications found for protein NMR such as Structure- Activity Relationships (SAR) by NMR (Shuker et al, 1996), and new methodology such as the transverse relaxation optimised spectroscopy (TROSY) experiments (Pervushin et al, 1997) and application of liquid crystal technology (Tjandra & Bax 1997) being introduced. Together with the prospect of even higher magnetic field

strength spectrometers becoming available, the future of protein NMR spectroscopy looks to be highly promising.

2D NMR

The multidimensional experiments may be understood by first examining how 2D NMR experiment are constructed. The first papers published on 2D NMR were by Ernst and his co-workers (Aue et al, 1976; Kumar et al, 1975; Muller et al,

1975), inspired by the original idea of Jean Jeener (Jeener 1971). The basic concept is simple - in a basic 2D experiment, two pulses are separated by a time delay t\

which is steadily incremented when the pulse sequence is repeated during the course of the experiment, and the subsequent Fourier transform of not only the FIDs (time period ^2) after the second pulse, but also with respect to the time variable t\. The signal detected is therefore a function of two time variables and the resultant spectra would appear as diagonal and cross-peaks in a two-dimensional plot generated as a function of the two time periods.

The two time variables t\ and are detected differently (Fig. 4.3). In a normal ID NMR experiment, the familiar sinusoidal decay of the FID is recorded directly as a series of data points separated by the dwell time, a duration which is set to be the reciprocal of the spectral width (SW) in order to satisfy the Nyquist criterion. These data points are therefore sampled at 1/SW intervals following excitation, with the total number of points collected N chosen to suit the required digital resolution and data size. In a 2D experiment, this is the acquisition time ti.

Notionally, this same set of data can be acquired indirectly - an TV number of experiments are performed, but with only one data point per experiment recorded, and in each successive experiment, the time (the t\ time delay) is steadily incremented by a 1/SW second delay before acquiring the single data point. The sum of these discrete data points from this rather more time-consuming set of experiments is equivalent to the directly acquired FID (Fig. 4.3). In reality, of course, the experiments are recorded not as a series of points, but as a series of ID experiments which, when Fourier transformed with respect to t\, produces the second dimension in a 2D frequency plot. This, in a 2D experiment, is essentially how the incremented delay t\ can generate another dimension, i.e. the indirect dimension that forms the

basis of multidimensional NMR. How this t\ time delay is inserted in a pulse

sequence and used to carry the information about a spin during a 2D NMR experiment is described below.

In d ire c t

D ire ct

+

+

FT

d

0

Figure 4.3: Direct and indirect method fo r the acquisition o f time domain signal. For directly detected dimensions o f multidimensional experiments a series o f points are collected during a single experiment, while in indirectly detected dimensions a single point is collected in each o f a series o f experiments.

In a simple ID ‘pulse-and-collect’ NMR experiment, there are two time period, i.e. the preparation and acquisition time. In the basic 2D NMR scheme, however, there are four time periods: preparation, evolution, mixing and detection (Fig. 4.4). The preparation period brings the sample to a state of thermal equilibrium

followed by the application of a pulse to bring the magnetisation onto the x-y plane. During the evolution period (the time period t\), the spins are allowed to precess or

‘evolve’ for a time f|, and the spins are ‘labelled’ according to the precessing frequency - by changing the period t\ , the phases of the spin at the beginning of the

mixing period are altered as a function of its frequency, thereby carrying over information about the spin. During the mixing period, the ‘coherence’ is transferred from one spin to another and correlation between two spins is achieved. The design of the mixing period determines, to a large extend, just what the particular 2D experiment achieves. The detection or acquisition period (time period r?) simply records each FIDs. Each successive FID records the history of the evolution of the spin system as a function of t\. The Fourier transformation of each t2 dimension

yields a set of directly detected ID spectra, and the intensities of the resonances are modulated sinusoidally as a function of the t\ duration as a result of phase changes at

the end of t\. The t\ dimension can then be further Fourier transformed to produce a

2D spectrum S(a)i,a)2) in which a ‘cross-peak’ appears at the intersection of the two nuclear signals.

Preparation Evolution Mixing D etection

Figure 4.4: The basic steps in a 2D NMR experiment.

The individual 2D NMR experiments are distinguished by the applied excitation and mixing procedures. A classic 2D experiment is the correlated spectroscopy (COSY) experiment, the homonuclear version of Jan teener’s original experiment. A simple COSY experiment has the pulse sequence of (90°)% - /i - (90°)x - Acquisition (f?), i.e. two 90° pulses applied along the x axis separated by a

time delay t\. Scalar coupled protons are correlated in this experiment and the proton resonances are dispersed into two dimensions, forming diagonal peaks and cross­ peaks. Other important homonuclear 2D NMR experiments include total correlation spectroscopy (TOCSY) which gives correlations to all the protons in an extended coupling network, and nuclear Overhauser effect spectroscopy (NOESY) that yields through space connectivities of protons by dipolar coupling. 2D NMR can also be used to resolve the proton resonances according to the chemical shifts of the directly bonded heteronuclei such as ^^N and as in the HSQC experiment. The HSQC is analogous to the COSY sequence but correlates two different nuclei. How this is done will be described in greater detail later.