Monday, August 1, 2016

Sensitivity in HMBC and LR-HSQMBC experiments

Long range correlations are critical for structure elucidation. They are typically identified using HMBC experiments1, but a relatively recent alternative is the LR-HSQMBC experiment2. Recent attempts to confirm the structure of several challenging molecules in the Facility prompted an examination of the sensitivity of these two experiments to determine which is most likely to provide the crucial long range correlations.

HMBC and LR-HSQMBC experiments were recorded using two standard samples; RS2-153A, a small (244.3 Da) drug-like molecule; and cholesteryl acetate (428.7 Da). The experiments were run with the same number of scans (16) and t1 increments (256), and the polarization transfer delay optimized for couplings of 8 Hz. The spectra were processed identically using a squared sine bell in the direct 1H dimension and four-fold linear prediction and a squared cosine bell in the indirect 13C dimension. Sensitivity was measured by extracting 13C columns through the correlations, selecting a noise region of 5000 Hz and comparing this to the intensity of the signals. The magnitude of the coupling constants producing the correlations was obtained from a J-HMBC spectrum3, while a HETLOC4 was used to identify the sign.

The figure below compares the sensitivity (S/N) in the HMBC (blue squares) and LR-HSQMBC (red diamonds). The only cases where the LR-HSQMBC is more sensitive have coupling constants around 11.5 Hz and S/N near 150. These correlations involve two adjacent aromatic positions on an imidazole ring in RS2-153A that are likely to exhibit atypical relaxation and coupling. In every other case the HMBC is more sensitive than the LR-HSQMBC.

HMBC experiments are typically optimized to detect correlations of 8 Hz, but it is common practice to optimize for 4 Hz correlations when looking for missing, weak or longer correlations. To examine the effectiveness of optimizing for smaller couplings, HMBC spectra optimized for 6, 4 and 2 Hz were acquired, processed, and analyzed in the same way as the spectra described above.

The figure below plots the ratio of the S/N of the new experiments to the S/N in the 8 Hz HMBC. A S/N ratio greater than 100% indicates that optimizing for a smaller coupling increased the sensitivity. For most coupling constants changing the optimization does not improve the S/N significantly, and often reduces it. In particular, the 4 Hz optimized HMBC appeared to be of little benefit. The 2 Hz optimized HMBC, however, was of particular use for couplings in the range of 1-4 Hz, for which many correlations were more than twice as intense as in the 8 Hz HMBC.

The drawback of optimizing a HMBC for smaller couplings is that it increases the length of the polarization delay - from 62.5ms for 8 Hz optimization, to 250ms for 2 Hz optimization. With a longer delay relaxation can reduce the signal intensity. The data here, however, suggests that signal loss due to relaxation in the extended delay does not affect the sensitivity. For larger molecules with shorter T2s it may be more of a problem.

The results presented above suggest that the HMBC is better than the LR-HSQMBC in terms of sensitivity. It may be that in some cases the LR-HSQMBC provides extremely long range correlations that are not seen in the HMBC, but that was not the case with the two molecules examined here. When looking for long range correlations I would recommend first running an 8 Hz optimized HMBC, then trying a 2 Hz optimized HMBC if additional correlations are required.

1. D.O. Cicero, G. Barbato & R. Bazzo
Sensitivity Enhancement of a Two-Dimensional Experiment for the Measurement of Heteronuclear Long-Range Coupling Constants, by a New Scheme of Coherence Selection by Gradients.
J. Magn. Reson. 2001 148(1):209-213.

2. R.Thomas Williamson, Alexei V. Buevich, Gary E. Martin & Teodor Parella.
LR-HSQMBC: A Sensitive NMR Technique To Probe Very Long-Range Heteronuclear Coupling Pathways.
J. Org Chem. 2014 79:3887-38942.

3. A. Meissner & O.W. Soerensen.
Measurement of J(H,H) and long-range J(X,H) coupling constants in small molecules. Broadband XLOC and J-HMBC
Magn. Reson. Chem. 2001 39:49-52.

4. D. Uhrin, G. Batta, V.J. Hruby, P.N. Barlow & K.E. Koever.
Sensitivity- and Gradient-Enhanced Hetero (ω1) Half-Filtered TOCSY Experiment for Measuring Long-Range Heteronuclear Coupling Constants.
J. Magn. Reson. 1998 130:155-161


  1. Hi Brendan - your write-up of these experiments is excellent! The value of the 8 Hz and 2 Hz HMBC matches our experience (the 2 Hz optimization has helped us numerous times over the years). Funny about the LR-HSQMBC; I wonder if there is some further tweaking or optimization of this experiment (e.g. maybe the sine bell transformation, or the 8 Hz setting, is not optimal for this experiment?).

    1. Hi Bill

      Thanks for your comments. If you don't mind I will add them to the blog so other people can see the discussion as well.

      I am glad to hear that you had seen the 2Hz optimization work too. I was a bit surprised that the 4Hz optimization showed so little improvement when it seems to be such a commonly used tool.

      The LR-HSQMBC did not seem to respond so well to reducing the polarization delay. I did not see such an improvement at lower coupling constants as there was for the HMBC.

      I am not sure why this would be. It could be that the greater number of pulses means it is more sensitive to miscalibration. Maybe the creation of exclusively single quantum coherence doesn't capture as much signal. I don't think the processing is a problem though. I chose a sine bell transformation in the 1H dimension to maximize the resolution so that when extracting the 13C slices overlap from nearby resonances would be reduced. This window function reduces sensitivity but it was applied to all spectra. Maybe I do not have sufficient data yet. There are still correlations from the cholesteryl acetate spectra that I have not analysed yet.