Isotopic labeling of proteins has enabled the development of series of new heteronuclear experiments, which are more suitable for coupling constant measurements than the corresponding homonuclear experiments. There are several experimental approaches, which can be divided roughly into four groups. The first group of experiments relies on the measurement of coupling constants directly from in- or antiphase splittings. The second group consists of experiments that use double- and zero-quantum (DQ/ZQ) spectroscopy for determination of couplings from DQ/ZQ splittings. The third group of experiments is designed for the creation of an E.COSY pattern, from which the small coupling constant of interest can be measured with the aid of a larger coupling. The fourth group is based on the quantitative J-correlation principle, where the intensity of two cross-peaks, or diagonal peak and cross-peak are compared. Recently introduced spin-state-selective filtering can be combined with J-resolved, DQ/ZQ, and E.COSY experiments to assist in the measurement of couplings. The following sections outline the basic principles, advantages, drawbacks, and limitations of the aforementioned approaches.
The simplest way to construct a multi-dimensional NMR experiment with coupling constant information is to record a two-dimensional correlation map in which the coupling constant of interest can be measured either from the indirectly or the directly detected dimension from an in- or antiphase splitting. In protein NMR spectroscopy, this approach is successful in the determination of large one-bond couplings between, for example, 13Ca and 1Hα, and 15N and 1HN. The J-resolved spectroscopy may fail if the coupling of interest is small compared with the line width, which is often the case with structurally important 3J-couplings. To determine these relatively small couplings from J-resolved experiments, special means have to be employed. These are discussed in some detail in Section 5.1.1.
In addition to simplicity, an obvious advantage of J-resolved experiments is that they are not as prone to systematic errors as experiments in which couplings are extracted from cross-peak intensities. Since the coupling constant of interest is measured from the frequency difference between the corresponding resonances, instead of the relative intensities, J-resolved methods are not apt to errors influencing signal intensity. Furthermore, systematic errors that do occur originate from the same spin operator(s) throughout the experiment. Regrettably, J-resolved experiments are sensitive to the effects of differential relaxation and cross-correlation effects, whenever separation of multiplet components is small compared with line width. Effects of cross-correlation can, however, be recognized from a small asymmetry in the line shape.
The most serious drawback of the J-resolved experiments is their inherent nature to increase spectral crowding. For instance, a typical 25 kDa protein has approximately 200 main chain NH cross-peaks that can be well dispersed in the usual 15N-HSQC spectrum at high magnetic field. However, if the proton coupled 15N-HSQC spectrum is recorded, the number of cross-peaks increases to 400, and consequently, it is more than likely that the resulting spectrum will exhibit serious overlap. Fortunately, it is possible to measure J-couplings without additional cross-peaks due to J-splitting by utilizing spin-state-selection (vide infra).