Assigning stereochemistry with NOEs

Relative stereochemistry can often be determined by analysis of nuclear Overhauser effects or NOEs. I have seen many papers where the stereochemistry was defined based on the observation of one particular peak, but this may not be sufficient and some care must be taken when interpreting NOEs. Here I offer a few pointers on how best to interpret the crosspeaks in NOESY and ROESY spectra.

The nuclear Overhauser effect (NOE) is a through-space interaction over short distances, typically assumed to be less than 5 Å. NOESY and ROESY experiments correlate atoms via NOEs, so a crosspeak in a NOESY or ROESY spectrum indicates that the atoms involved are within 5 Å of each other. This seems simple enough, but there are some complications.

In addition to through-space interactions, NOESY and ROESY spectra often show zero-quantum artifacts for strongly coupled atoms, typically those with a significant J_HH coupling. This normally appears as DQF-COSY-like correlations. If the spectrum is processed in magnitude mode, then these zero-quantum artifacts will look identical to through-space interactions. It is possible to suppress zero-quantum artifacts through the combination of an adiabatic pulse and a weak gradient, but it doesn't eliminate them completely. The figure below shows the structure and atom numbering of 2β-chloro-3α bromocholestane, the H12-H11 crosspeaks of cholesterol acetate in a standard NOESY, and the H12-H11 crosspeaks of 2β-chloro-3α bromocholestane in a NOESY with zero quantum suppression. H12 and H11 in both molecules are in the same chemical environment and should produce the same NOEs. Obviously the NOESY with zero quantum suppression is a lot cleaner.

I recommend to always use phase sensitive processing, use pulse sequences with zero-quantum suppression whenever possible, and be particularly wary of peaks that could be due to atoms separated by two or three bonds.

Observation of an NOE indicates that the atoms involved are likely closer than 5 Å, but without careful calibration from multiple spectra it is difficult to say exactly how close. If one is looking at a single crosspeak does that mean the atoms are 2 Å apart or 5 Å? For this reason NOEs should be interpreted in pairs whenever possible. For example, the intensity of an NOE between atom A and atom B should be compared to the intensity of an NOE between atom A and atom C. Only if the NOEs differ in intensity can conclusions about the relative positions of the atoms be made. For example, in the expansion below on the left the cholestane shows NOEs from H2 to the adjacent diastereotopic methylene protons H1+ and H1- ("+" signifies the downfield peak and "-" the upfield one). However, the peaks have similar intensities so from these NOEs no conclusions can be drawn about the stereochemistry.

In the expansion on the right H1+ shows a sizeable NOE to H19 whereas the H1- NOE is much smaller. The difference in intensity allows H1+ to be identified as closer to H19 than H1-. Similarly, H12+ shows an NOE to H18 but H12- shows none at this contour level, allowing H12+ to be assigned as closer to H18 than H12-. Other NOEs show H12+ closer to H21 than H12-, while H12- is closer to H9 than H12+.

This information must be verified by comparison with a three dimensional model of the structure. Models can be physical, built from kits, or virtual ones constructed with chemical modelling software. Be careful with virtual structures, though, as not all packages generate reasonable conformations. Finally all of the NOEs must be consistent with the proposed stereochemistry. If some of the observations cannot be explained then the proposal needs to be revised. The figure below shows how the NOEs in the expansions above are consistent with the expected three dimensional structure of 2β-chloro-3α bromocholestane.

References

Thrippleton MJ, Keeler J.
Elimination of zero-quantum interference in two-dimensional NMR spectra.
Angew Chem Int Ed Engl. 2003 Aug 25;42(33):3938-41