Heteronuclear filters

Otting G, Wiithrich K. Heteronuclear filters in two-dimensional H H] NMR spectroscopy combined use with isotope labelling for studies of macromolecular conformation and intermolecular interactions. Q Rev Biophys 1990 23 39-96. 

X-ray crystallography, docking modes can be validated by various NMR techniques NOEs may be observed between the ligand and the receptor protein by heteronuclear-filtered NOE spectroscopy [51], chemical shift changes of protein resonances upon binding can be analyzed by simulation of shifts caused by ring currents and electrostatic effects [52], and saturation transfer difference measurements indicate which part of the ligand is in direct contact with the protein [52]. 

Traditional protein NMR spectroscopy of smaller proteins relies of 15N-filtered experiments, due to the relatively low expense of introducing 15N labels into proteins (compared to 13C) and the concomitant ability to use heteronuclear filtering to improve resolution in the H NMR dimension. Jelinek et al. were the first to demonstrate the ability to transfer this approach to peptides on TantaGel.80 They also showed the ability to detect pronounced peptide structure through the appearance of strong NOE correlations in 3H NOESY HRMAS spectra as shown in Fig. 8. This had important implications for the search of biological activity in peptides attached to supports, as the structure on the support may be different or more pronounced than in solution, if present at all in solution in peptides of this small size. 

The HMQC or HSQC sequences may be transformed into their ID equivalents by simply removing the incremental t time period (Fig. 6.22) so that the experiment becomes just a heteronuclear filter. Only magnetisation that has passed via the X spin will be observed in the final spectrum and again the suppression of all unwanted signals is greatly improved by the use of pulsed field gradients. The selective observation of C-labelled glycine in an aqueous mixture is illustrated in Fig. 6.23. 

When combining isotope filtering/editing with coherence transfer steps to multidimensional experiments, then further size restrictions apply. For example, isotope edited / filtered H TOCSY or COSY experiments are generally limited to systems of <10 kDa, because of their sensitivity to T2 relaxation. In larger systems, heteronuclear correlation spectroscopy can be used for the correspondingly labeled component, while structural information about both the labeled and unlabeled moiety can be extracted from isotope edi-ted/filtered NOESY spectra, respectively. 

Obviously, this approach cannot be used for selecting the nonisotope-labeled components. In the following we will consider isotope filtering/editing techniques that do not use heteronuclear chemical shift evolution. 

This scheme is applied in the so-called X-half-filter technique (Fig. 17.4a, c), with the only difference of an additional 90 (X) pulse with constant phase [16, 17]. Instead of generating heteronuclear multiple quantum coherence 2 Iy Sy, which cannot readily be detected (in case of selecting the 1H-X pairs), one now always ends up with proton antiphase coherence, but with the same phase alternation ... 

An alternative way of realizing an isotope filter is shown in Fig. 17.4b, where the 90° phase difference between the two proton magnetizations is exploited [18]. A second 90° j1 ) pulse (of same phase as the excitation pulse) at the end of the period r =l/2j leaves the heteronuclear antiphase magnetization of the X-bound protons unaffected, while the other protons are converted to z magnetization ... 

In most cases where long-range couplings are exploited to solve a structural problem corresponding information on heteronuclear one-bond interactions is needed as well. Instead of suppressing this valuable information with a low-pass filter as in the basic HMBC experiment, we additionally modified our pulse sequence - the second non-selective 90° pulse is... 

Selecting the C-bound protons before performing a homonuclear two-dimensional experiment enables to measure small heteronuclear coupling constants [16]. Such an experiment with a sample of natural isotopic abundance was first published by Otting and Wuthrich in 1990, where the half-filter element with spin-lock purge pulse was used to select the C-bound protons in a small protein in aqueous solution [6]. Later applications illustrated the usefulness of the same half-filter element with smaller molecules [17, 18]. 

The idea of back transformation of a three-dimensional NMR experiment involving heteronuclear 3H/X/Y out-and-back coherence transfer can in principle be carried to the extreme by fixing the mixing time in both indirect domains. Even if one-dimensional experiments of this kind fall short of providing any information on heteronuclear chemical shifts, they may still serve to obtain isotope-filtered 3H NMR spectra. A potential application of this technique is the detection of appropriately labelled metabolites in metabolism studies, and a one dimensional variant of the double INEPT 111/X/Y sequence has in fact been applied to pharmacokinetics studies of doubly 13C, 15N labelled metabolites.46 Even if the pulse scheme relied exclusively on phase-cycling for coherence selection, a suppression of matrix signals by a factor of 104 proved feasible, and it is easily conceivable that the performance can still be improved by the application of pulsed field gradients. 

Figure 15.1. (A) COSY, (B) TOCSY, (C) 1H-1T HSQC or HMQC, (D) dl- Y HMBC, for 4-oxopentanal. For clarity, only key assignments have been given as an example. Note that the double-ended arrows indicate how to interpret the spectra. In the case of COSY and TOCSY the information is represented as cross-peaks that are symmetrically oriented with respect to the central diagonal. In the single-bond correlation (HSQC/HMQC) a cross-peak represents in one dimension the carbon chemical shift and in the other dimension the proton chemical shift. Note there is no diagonal in heteronuclear NMR experiments. In the HMBC, lines are drawn vertically to connect the cross-peaks. In HMBC 2-4 bonds, H-13C correlations are often observed. Note that the 4-bond correlation is less common in NMR but has been included here as an example, and 1-bond correlation is commonly filtered from the HMBC experiment to improve detection limits for the weaker 2-4 bond correlations.

In order to determine couplings to nuclei in natural abundance, it is necessary to suppress the signals of protons that are not bonded to a magnetically active heteronucleus. An (iix hetero half-filter that selects for such nuclei in F, via the phase cycle was used for this purpose. Presaturation of the protons bonded to 12C by the BIRD pulse allows a rapid pulse-sequence (2 scans per second). The resulting 2D spectra are TOCSY spectra in which the cross peaks show the desired E. COSY pattern. From the results shown, the only limitation seems to be the resolution obtained, although the authors do not hesitate to use a third heteronuclear frequency domain for improvement. 

---