In terms of molecular size, NMR is the poor relation of X-ray crystallography. For natural-abundance proteins we can use only homonuclear techniques (1H-1H J values and NOEs), and we are limited to proteins of less than about 125 residues or a molecular weight of around 12,000 amu (12 kiloDaltons or kD). These are very small proteins, and in the early days of protein NMR there was quite a bit of competition for the relatively small number of biologically significant proteins that were within the reach of the technique, especially for those that did not already have an X-ray structure. A suitable NMR sample must also be soluble and monomeric at a concentration very high compared to physiological concentrations: a 1 mM sample with a volume of 0.5 mL is desirable, requiring a large amount of pure protein.

Why are we so limited in the size of molecules we can study? The enemy of NMR spectroscopy is overlap: we have a finite amount of "turf" to put our NMR peaks in—about 10 ppm for XH—and we have to fit all the resonances of the molecule into that space. For a specific type of proton with a particular relationship to electronegative atoms the space available is even less: for example, an Ha proton in an amino acid residue will fall between 4 and 5 ppm, with an extreme variation of 3-6 ppm. The "footprint" of the resonance is determined by a combination of the linewidth (width of each line in the multiplet) and the coupling constants (separation of those lines) and might typically be about 30 Hz for an Ha in a small protein. With a 600 MHz instrument, that corresponds to 0.05 ppm (30/600), so we might be able to "fit" 20 of these in each ppm of real estate for a total of 60 in the extreme "window" of 3-6 ppm for an Ha proton if we are extremely lucky and their chemical shifts are spread out evenly. More likely the majority will be near the center of this window and fewer will be near the edges. For a 100 residue protein, with 100 different Ha protons, there will be overlap. Using 2D NMR techniques is helpful but eventually even in two dimensions we will have overlap of crosspeaks, and unambiguous assignment of a resolved crosspeak still requires that the 1D chemical shift be unique.

What happens to these difficulties as we increase the molecular weight of the protein (Table 12.1)? Clearly, the number of resonances within the same chemical shift range will increase and inevitably there will be more overlap and ambiguity of assignments. For example, a NOESY crosspeak at F\ = 3.56 ppm and F2 = 9.28 ppm can be interpreted in nine different ways if there are three HN protons with chemical shift 9.28 ppm and three Ha protons with chemical shift 3.56 ppm. But this is not the only problem. As molecular weight increases, T2 gets shorter and linewidth increases. The FID decays faster and the Fourier transform of a fast-decaying signal is a broad Lorenztian peak. So the footprint of each resonance is now larger, and we can fit even fewer unique chemical-shift positions into the fixed range of chemical shifts for each type of proton (HN, Ha, aromatic, etc.). It's like

Table 12.1. Typical proteins and size limits for NMR


0 0

Post a comment