MAGNETIC RESONANCE IN CHEMISTRY
Magn. Reson. Chem. 2007; 45: 544–546
Published online 17 April 2007 in Wiley InterScience
(www.interscience.wiley.com) DOI: 10.1002/mrc.1998
The use of unsymmetrical indirect covariance NMR
methods to obtain the equivalent of HSQC-NOESY
data
Kirill A. Blinov,1 Antony J. Williams,2 Bruce D. Hilton,3 Patrick A. Irish3 and Gary E. Martin3∗
1 Advanced Chemistry Development, Moscow Department, 6 Akademik Bakulev Street, Moscow 117513, Russian Federation, Russia
2 Advanced Chemistry Development, 110 Yonge Street 14th Floor, Toronto, Ontario M5C 1T4, Canada
3 Pharmaceutical Sciences, Chemical & Physical Sciences, Schering-Plough Corporation, Summit, NJ 07901, USA
Received 7 December 2006; Revised 20 February 2007; Accepted 24 February 2007
We have recently demonstrated that unsymmetrical indirect covariance NMR methods can be used to
mathematically calculate the equivalent of low sensitivity, hyphenated NMR experiments by combining
data from a pair of higher sensitivity experiments. The present report demonstrates the application of this
method to the combination of HSQC and NOESY spectra to provide results comparable to HSQC-NOESY
data, albeit with greater sensitivity and with considerably less spectrometer time. Copyright  2007 John
Wiley & Sons, Ltd.
A number of reports have appeared in the recent liter-
ature by Bru¨schweiler and co-workers describing various
aspects of covariance NMR spectroscopy.1 – 6 In the area of
small molecule structure characterization, perhaps the most
applicable of these reports deal with the topic of indirect
covariance NMR spectroscopy.4 We have further investi-
gated the area of indirect covariance NMR, reporting the
detailed analysis of artifact peaks that can arise from pro-
ton overlap.7 Artifact analysis has led us to an investigation
of methods for the elimination of these artifacts resulting
in the development of unsymmetrical indirect covariance
methods. Exploring the capabilities of unsymmetrical indi-
rect covariance processing capabilities still further, we have
more recently demonstrated that it is possible to mathe-
matically combine discretely acquired 2D NMR data sets
to afford the calculated equivalent of much lower sensi-
tivity NMR experimental data. Examples reported thus far
have included the combination of HSQC and HMBC data
to yield the equivalent of m.n-ADEQUATE data,8 and the
combination of HSQC and COSY or TOCSY data to provide
the equivalent of HSQC-COSY or – TOCSY data sets.9 We
now wish to report a further extension of these methods to
the combination of HSQC and NOESY data to provide a
potential higher sensitivity alternative to the acquisition of
HSQC-NOESY data.
Hyphenated heteronuclear 2D NMR experiments, in gen-
eral, exhibit lower sensitivity than the individual component
2D-NMR experiments that are combined to give the hyphen-
ated technique. Examples include HSQC-COSY or – TOCSY,
-NOESY, and – ROESY. While there are a few very interesting
applications of both HSQC-NOESY and -ROESY experiments
ŁCorrespondence to: Gary E. Martin, Rapid Structure
Characterization Laboratory, Pharmaceutical Sciences, Chemical &
Physical Sciences, Schering-Plough Corporation, Mail stop
S7-D-1-1, 556 Morris Ave., Summit, NJ 07901, USA.
E-mail: gary.martin@spcorp.com
in the literature, there are undoubtedly fewer applications
reported than there would have been if these were higher
sensitivity experiments.10 – 14
In the interest of exploring the possibility of using unsym-
metrical indirect covariance processing for the mathematical
combination of HSQC and NOESY data to yield the equiva-
lent of an HSQC-NOESY experiment, a sample of the simple
molecule ibuprofen (1) was prepared. A set of spectra were
acquired that included a 1H reference spectrum, HSQC,
NOESY, and HSQC-NOESY spectra. All data were recorded
using a Varian three channel NMR spectrometer equipped
with a 3-mm gradient inverse-detection NMR probe. Mixing
times for the NOESY and HSQC-NOESY spectra were based
on a proton T1 measurement, and data were recorded for
mixing times of 450 and 900 ms The spectra were acquired
with identical spectral widths in F2. The HSQC data were
acquired in approximately 30 min. The NOESY spectra were
acquired in 3.75 and 6.8 h, respectively. The HSQC-NOESY
spectrum with a 450-ms mixing period was acquired in
44 h. The HSQC-NOESY spectrum is shown in the bot-
tom panel of Fig. 1. As expected, the NOESY responses
in the HSQC-NOESY spectrum were weak relative to the
direct correlation responses. The top panel of Fig. 1 shows
the calculated HSQC-NOESY spectrum derived from the
unsymmetrical indirect covariance processing of the HSQC
and 450 ms NOESY spectra. In general, the NOESY responses
in the calculated HSQC-NOESY spectrum are much stronger
relative to those in the bottom panel and the overall s/n ratio
is also considerably higher. The data in both panels in Fig. 1
were plotted with identical threshold values.
To compare the performance of the two experiments,
slices from both 2D contour plots shown in Fig. 1 are
presented in Fig. 2. Slices were extracted at the F1 chemical
shift of the sec-butyl methyl groups (¾22.5 ppm). The slice
from the 44 h-HSQC-NOESY spectrum is shown in the
bottom panel of Fig. 2; the corresponding slice from the
Copyright  2007 John Wiley & Sons, Ltd.

NMR methods to obtain the equivalent of HSQC-NOESY data 545
7 6 5 4 3 2 1
F2 Chemical Shift (ppm)
20
40
60
80
100
120
F1
C
he
m
ica
l S
hi
ft
(pp
m)
7 6 5 4 3 2 1
F2 Chemical Shift (ppm)
20
40
60
80
100
120
F1
C
he
m
ica
l S
hi
ft
(pp
m)
Figure 1. HSQC-NOESY spectrum of ibuprofen (1, bottom
panel) recorded with a 450 ms mixing time in 44.5 h at
500 MHz using a 2-mg sample of ibuprofen dissolved in
¾170 µl of deuteriochloroform in a 3-mm NMR tube. The
standard Varian gradient HSQC-NOESY pulse sequence was
employed without any modification. The HSQC-NOESY
spectrum calculated (top panel) using unsymmetrical indirect
covariance methods7 from a 30-min HSQC and a 3.75-h
NOESY spectrum with a mixing time of 450 ms. Both spectra
are plotted with identical scaling and thresholds.
calculated spectrum is shown in the top panel. The relative
s/n ratio difference in the two experiments is obvious in this
comparison. Both experiments were plotted with identical
scaling.
While there is no tracking in F1 in the calculated
spectrum in the top panel, minor phasing problems can
arise from incorrectly set phase in either the HSQC
Figure 2. Slice comparison. The bottom trace (A) shows the
slice taken at the chemical shift of the sec-butyl methyl group
(¾22.5 ppm) from the HSQC-NOESY spectrum shown in the
bottom panel of Fig. 1. The top trace (B) shows the identical
slice from the unsymmetrical indirect covariance calculated
HSQC-NOESY spectrum shown in the top panel of Fig. 1. Both
traces are truncated vertically at 5% of the height of the
sec-butyl methyl direct response. The measured s/n ratios for
the three NOESY correlations from the sec-butyl methyl
resonance observed in the spectrum are shown on both
traces. In addition to more than twice the s/n in the calculated
spectrum for two of the three responses, there is also
considerably less noise in the spectrum calculated using
unsymmetrical indirect covariance methods than is observed in
the trace from the 44-h HSQC-NOESY experiment.
or NOESY spectra prior to the unsymmetrical indirect
covariance calculation that cannot be corrected in the
present version of the software (Advanced Chemistry
Development (ACD)/SpecManager v10.02) following any
form of covariance processing. Consequently, considerable
care must be exercised in phasing both the HSQC and NOESY
spectra when they will be used for further, unsymmetrical
indirect covariance processing.
Overall, the HSQC-NOESY spectrum calculated using
unsymmetrical indirect covariance methods7 is identical in
Copyright  2007 John Wiley & Sons, Ltd. Magn. Reson. Chem. 2007; 45: 544–546
DOI: 10.1002/mrc

546 K. A. Blinov et al.
CH3
H3C
CH3
O
OH
1
information content to the acquired spectrum, despite the
fact that the former data were acquired in just over 4 h versus
the 44 h acquisition necessary to record the HSQC-NOESY
spectrum shown in the bottom panel of Fig. 1. As shown
by the slices taken at the F1 chemical shift (¾22.5 ppm)
of the sec-butyl methyl groups shown in Fig. 2, the s/n
ratio in the calculated spectrum is considerably higher. The
calculated spectrum is also free of the considerable noise in
the upfield region of the trace near the cross peaks from the
spectrum labeled 25 : 1 and 27 : 1, as well as the tracking in
the F1 frequency domain that is much more prevalent in the
bottom panel of Fig. 1. As has been shown previously, proton
resonance overlap may lead to artifacts in indirect covariance
spectra.7 The development of a method to circumvent the
problem of artifacts due to proton multiplet overlap is under
continued investigation, as is further investigation of what
may be possible using covariance, indirect covariance, and
unsymmetrical indirect covariance processing techniques.
REFERENCES
1. Bru¨schweiler R, Zhang F. J. Chem. Phys. 2004; 120: 5253.
2. Bru¨schweiler R. J. Chem. Phys. 2004; 121: 409.
3. Trbovic N, Smirnof S, Zhang F, Bru¨schweiler R. J. Magn. Reson.
2004; 171: 277.
4. Zhang F, Bru¨schweiler R. J. Am. Chem. Soc. 2004; 126: 123180.
5. Zhang F, Bru¨schweiler R. Chem. Phys. Chem. 2004; 5: 794.
6. Zhang F, Trbovic N, Wang J, Bru¨schweiler R. J. Magn. Reson.
2005; 174: 219.
7. Blinov KA, Larin NI, Kvasha MP, Moser A, Williams AJ,
Martin GE. Magn. Reson. Chem. 2005; 43: 999.
8. Blinov KA, Larin NI, Kvasha MP, Williams AJ, Zell M,
Martin GE. Magn. Reson. Chem. 2006; 44: 107.
9. Blinov KA, Larin NI, Williams AJ, Mills KA, Martin GE.
J. Heterocycl. Chem. 2006; 43: 163.
10. Kawabata J, Fukushi E, Mizutani J. J. Am. Chem. Soc. 1992; 114:
1115.
11. Castle LW, Johnston MD, Camoutsis CL, Castle RN. J. Heterocycl.
Chem. 1992; 29: 1869.
12. Wagner R, Berger S. Magn. Reson. Chem. 1997; 35: 199.
13. Shimbo K, Tsuda M, Fukushi E, Kawabata J, Kobayashi J-I.
Tetrahedron 2000; 56: 7923.
14. Gschwind RM, Xie X, Rajamohanan PR. Magn. Reson. Chem.
2004; 42: 308.
Copyright  2007 John Wiley & Sons, Ltd. Magn. Reson. Chem. 2007; 45: 544–546
DOI: 10.1002/mrc