Rapid Communications
Using Unsymmetrical Indirect Covariance
Processing to Calculate GHSQC-COSY Spectra
Gary E. Martin,*,† Bruce D. Hilton,† Patrick A. Irish,†
Kirill A. Blinov,‡ and Antony J. Williams§
Schering-Plough Research Institute, Summit, New Jersey 07901,
AdVanced Chemistry DeVelopment, Moscow DiVision, Moscow,
Russian Federation, and AdVanced Chemistry DeVelopment,
Toronto, Ontario, Canada
ReceiVed May 10, 2007
Abstract: GHSQC-TOCSY experiments allow sorting of proton–pro-
ton connectivity information as a function of 13C chemical shift.
GHSQC-TOCSY is a relatively insensitive 2D NMR experiment. Given
two coherence transfer experiments, A f B and A f C, it is possible
to indirectly determine B T C. Unsymmetrical indirect covariance
processing of a 1H–13C GHSQC and a GCOSY spectrum afforded a
GHSQC-COSY spectrum, with an information content analogous to a
GHSQC-TOCSY experiment. However, GHSQC-TOCSY is of sig-
nificantly lower sensitivity and the data require considerably more time
to acquire than either of the component experiments. Investigators
needing access to GHSQC-TOCSY type data can, in principle, access
it from more readily acquired 2D NMR data. Strychnine (1) was used
as a model compound to illustrate this capability.
Two-dimensional NMR methods have unquestionably had a
major impact on the elucidation of complex chemical and natural
product structures.1,2 COSY and some form of heteronuclear
chemical shift correlation, most commonly HMQC among natural
products chemists (although HSQC is a highly preferable
choice),3 are routinely used to characterize structures. As
molecular structures increase in complexity, it becomes necessary
to resort to more sophisticated 2D NMR methods to disentangle
potentially overlapped connectivity information in complex NMR
spectra. One particularly useful experiment is GHSQC-TOCSY
(or alternately, GHMQC-TOCSY),3 which provides the ability
to extract proton–proton connectivity information from heavily
overlapped spectra by sorting proton–proton correlations as a
function of the 13C chemical shift of the directly bound carbon.4
Unfortunately, despite the utility of GHSQC-TOCSY, the
experiment suffers from a severe sensitivity penalty that has
undoubtedly limited the number of applications of the technique
in natural product structure elucidation.5–12
Recently, a new approach to handling NMR data, indirect
covariance NMR, has been developed.13 In an effort to suppress
processing artifacts in the conversion of IDR-GHSQC-TOCSY data
to a 13C–13C correlation spectrum, a further modification of the
indirect covariance method was developed.14 Unsymmetrical
indirect covariance processing of IDR-GHSQC-TOCSY data
eliminates one type of artifact response and renders a second type
diagonally asymmetric, allowing them to be eliminated from the
data presentation by the symmetrization algorithm applied to COSY
spectra. An alternative and far more interesting application of
unsymmetrical indirect covariance processing capabilities arises
when one considers using this approach to co-process discretely
acquired 2D NMR spectra. Given two coherence transfer experi-
ments,
A f B
and
A f C
where, for example the coherence transfer experiments might be
GCOSY and 1H–13C GHSQC, it is possible to use unsymmetrical
indirect covariance processing to indirectly determine
B T C.
For the present example case, the result would be a GHSQC-COSY
spectrum. We have previously shown using a 2 mg sample of the
small molecule autumnolide the considerable time/sensitivity sav-
ings that can be realized using this approach.15 GCOSY and a
1H–13C GHSQC spectra acquired in 10 and 60 min, respectively,
were co-processed using unsymmetrical indirect covariance pro-
cessing methods to afford a GHSQC-COSY spectrum with a S/N
ratio, based on projection through the F1 frequency domain, of 77:
1. In comparison, directly acquiring an 18 ms IDR-GHSQC-TOCSY
spectrum on the same 2 mg sample required 16 h to obtain a
spectrum with a S/N ratio, again based on projection through the
F1 frequency domain, of 8:1. This type of gain in apparent sensitivity
* To whom correspondence should be addressed. Tel: +908-473-5398.
Fax: +908-473-6559. E-mail: gary.martin@spcorp.com.
† Schering-Plough Research Institute.
‡ Advanced Chemistry Development, Moscow.
§ Advanced Chemistry Development, Toronto.
 Copyright 2007 by the American Chemical Society and the American Society of Pharmacognosy
Volume 70, Number 9 September 2007
10.1021/np070221j CCC: $37.00  2007 American Chemical Society and American Society of Pharmacognosy
Published on Web 08/11/2007

(S/N) is analogous to the results obtained following a two-
dimensional Fourier transformation.
To further illustrate the capabilities of unsymmetrical indirect
processing, we have used strychnine (1) as a model compound.
GCOSY and 1H–13C GHSQC spectra were acquired with identical
1H (F2) spectral widths (which is no longer required in the
unsymmetrical covariance processing software, ACD/SpecManager
v10.02, but is useful in that it avoids interpolation required when
the F2 spectral widths are different). Spectral widths in the second
frequency domain (F1) were set as appropriate. The data were
processed to afford data matrices of 2048 × 512 points. The
GCOSY data were acquired as 1024 × 256 points and were zero-
filled to the final data matrix size. The GHSQC data were acquired
as 1024 × 96 data points. The GHSQC data were zero-filled in F2
and linear predicted to 192 points and then zero-filled to 512 points
in F1 during processing. Weighting functions were independently
optimized for the two experiments. For comparison, an IDR-
GHSQC-TOCSY spectrum with a 24 ms mixing time was acquired
and processed to afford a 2048 × 512 point final data matrix.
The aliphatic region of the unsymmetrical indirect covariance
processed GHSQC-COSY spectrum of strychnine (1) is shown in
Figure 1A; the corresponding region of the IDR-GHSQC-
TOCSY spectrum is shown in Figure 1B. Direct responses in the
latter are inverted (red contours). Relayed responses have positive
phase and are shown in black. Responses in the GHSQC-COSY
spectrum have phases corresponding to that of the multiplicity-
edited GHSQC spectrum used in the unsymmetrical indirect
covariance processing calculation. Methine-derived signals have
positive phase (black contours); methylene-derived signals have
negative phase (red contours). Projections through the F1
Figure 1. Unsymmetrical indirect covariance processed GHSQC-COSY spectrum calculated from a GCOSY and 1H–13C GHSQC spectrum
is shown in panel A. The component COSY and 1H–13C GHSQC spectra used to calculate the GHSQC-COSY spectra were acquired using
a Varian 500 MHz NMR spectrometer equipped with a 3 mm gradient inverse detection probe in 5 and 10 min, respectively. In contrast,
the 24 ms inverted direct response GHSQC-TOCSY data were acquired in 1.5 h. The two contour plots shown were prepared with identical
threshold levels. Comparison projections through the F1 frequency domain of the spectra above are shown in Figure 2.
1394 The Journal of Natural Products, 2007, Vol. 70, No. 9 Rapid Communications

frequency domain to compare the S/N performance of the two
experiments shown in Figure 1 are shown in Figure 2 and were
computed following the magnitude calculation of both 2D NMR
spectra.
In comparison, the GHSQC-COSY spectrum replicates the
information content of the 24 ms IDR-GHSQC-TOCSY spectrum.
All of the responses in the IDR-GHSQC-TOCSY are contained in
the calculated GHSQC-COSY spectrum; a number of the responses
were observed with considerably better peak intensity. There are
also legitimate responses contained in the GHSQC-COSY spectrum
that are not observed in the IDR-GHSQC-TOCSY spectrum either
because of the response being weak or because the mixing time
used was inappropriate for the correlation pathway in question. For
example, the H-12 resonance observed at 4.22/77.9 ppm in the IDR-
GHSQC-TOSCY spectrum exhibited TOCSY correlations only to
the H-11a and H-11b protons at ∼3.08 and ∼2.61 ppm, respectively.
On examination of the F2 slice at 77.9 ppm, a very weak correlation
response can be seen to the H-13 resonance at ∼1.18 ppm, although
this response is well below the threshold of the contour plot shown
in Figure 1B. In contrast, the H-12 resonance in the GHSQC-COSY
spectrum calculated using unsymmetrical indirect covariance
processing methods exhibited responses to the H-11a and H-11b
resonances, as well as readily observed correlations to H-13
resonating at ∼1.18 ppm and a long-range correlation across the
oxepin ether linkage to the H-23b proton resonating at ∼4.01 ppm.
It should also be noted that in cases of proton resonance overlap,
artifact responses can be generated during unsymmetrical indirect
covariance processing. Care must be exercised by an investigator
using these processing methods not to mistake an artifact response
due to resonance overlap as a legitimate correlation response.
The other comparison that should be noted is the relative S/N
ratio of the two experiments. The GCOSY and GHSQC spectra
were both acquired in 15 min in total in comparison to the 90 min
required to record the 24 ms IDR-GHSQC-TOCSY spectrum.
Projections through the F1 (13C) frequency domain of both spectra
are shown in Figure 2. The S/N ratio of the F1 projection for the
IDR-GHSQC-TOCSY spectrum was 40:1. In comparison the S/N
ratio of the F1 projection of the calculated GHSQC-COSY spectrum
was 144:1, a factor of 3.6 higher. When the difference in the S/N
ratio of the two experiments is considered in light of the acquisition
time for the data, a nearly 22-fold improvement is realized via the
unsymmetrical indirect covariance calculated GHSQC-COSY spec-
trum versus the acquisition of the 24 ms IDR-GHSQC-TOCSY
spectrum. This is not as high as was observed in our initial effort,
but it should be noted that the proton spectrum of strychnine (1) is
characterized by sharp multiplets without extensive coupling. A
proton spectrum with more extensively coupled multiplets, e.g.,
that of autumnolide used as a model compound in our previous
study,15 would require a correspondingly much longer data acquisi-
tion time to reach an acceptable S/N ratio in the GHSQC-TOCSY
spectrum.
We have compared the unsymmetrical indirect covariance
calculated GHSQC-COSY and IDR-GHSQC-TOCSY of several
complex alkaloids. In each case, the calculated GHSQC-COSY
spectrum reproduced the information content of the hyphenated 2D-
NMR spectrum. Unsymmetrical indirect covariance processing
methods can also be used to mathematically combine other pairs
of coherence transfer 2D NMR experiments, including 1H–13C
GHSQC and GHMBC,16 1H–13C GHSQC and 1H–1H NOESY,17
and 1H–13C GHSQC and 1H–15N GHMBC or 1H–15N IMPEACH-
MBC spectra to afford 13C–15N GHSQC-GHMBC18 and 13C–15N
GHSQC-IMEACH19 heteronuclear shift correlation plots, respec-
tively. 13C–15N correlation has no experimental equivalent at natural
abundance but has also recently been demonstrated by Kupcˇe and
Freeman using projection reconstruction methods.20 We are cur-
rently exploring the use of unsymmetrical indirect covariance
processing to solve chemical structure problems, which will be the
subject of forthcoming reports.
Figure 2. Projections through the F1 (13C) frequency domain in the GHSQC-COSY spectrum calculated using unsymmetrical indirect
covariance processing (A) and the 24 ms GHSQC-TOCSY spectrum after magnitude calculation (B). The S/N ratio of the two projec-
tions was 144:1 and 40:1, respectively. Both spectra shown in Figure 1 were magnitude calculated prior to plotting the projections shown
above.
Rapid Communications The Journal of Natural Products, 2007, Vol. 70, No. 9 1395

References and Notes
(1) Reynolds, W. F. J. Nat. Prod. 2002, 65, 221–244.
(2) Fukushi, E. Biosci. Biotech. Biochem. 2006, 70, 1803–1812.
(3) Hadden, C. E.; Martin, G. E.; Luo, J.-K.; Castle, R. N. J. Heterocycl.
Chem. 1999, 36, 533–539.
(4) Hadden, C. E.; Duholke, W. K.; Guido, J. E.; Robins, R. H.; Martin,
G. E.; Sharaf, M. H. M.; Schiff, P. L., Jr. J. Heterocycl. Chem. 1999,
36, 525–531.
(5) Crouch, R. C.; Martin, G. E. J. Heterocycl. Chem. 1995, 32, 1665–
1669.
(6) Martin, G. E. J. Heterocycl. Chem. 1997, 34, 695–699.
(7) McCoy, M. C.; Faulkner, D. J. J. Nat. Prod. 2001, 64, 1087–1089.
(8) MacMillan, J. B.; Ernst-Russell, M. B.; de Ropp, J. S.; Molinski, T. F.
J. Org. Chem. 2002, 67, 8210–8215.
(9) MacMillan, J. B.; Molinski, T. F. Org. Lett. 2002, 4, 1535–1538.
(10) Sandler, J. S.; Colin, P. L.; Hooper, J. N. A.; Faulkner, D. J. J. Nat.
Prod. 2002, 65, 1258–1261.
(11) Kogan, N. M.; Rabinowitz, R.; Levi, P.; Gibson, D.; Sandor, P.;
Schlesinger, M.; Mechoulam, R. J. Med. Chem. 2004, 47, 3800–3806.
(12) Ratnayake, A. S.; Bugni, T. S.; Feng, X.; Harper, M. K.; Skalicky,
J. J.; Mohammed, K. A.; Andjelic, C. D.; Barrows, L. R.; Ireland,
C. M. J. Nat. Prod. 2006, 69, 1582–1586.
(13) Zhang, F.; Brüschweiler, R. J. Am. Chem. Soc. 2004, 126, 13180–
13181.
(14) Blinov, K. A.; Larin, N. I.; Kvasha, M. P.; Moser, A.; Williams, A. J.;
Martin, G. E. Magn. Reson. Chem. 2005, 43, 999–1007.
(15) Blinov, K. A.; Larin, N. I.; Williams, A. J.; Mills, K. A.; Martin, G. E.
J. Heterocycl. Chem. 2006, 43, 163–166.
(16) Blinov, K. A.; Larin, N. I.; Williams, A. J.; Zell, M.; Martin, G. E.
Magn. Reson. Chem. 2006, 44, 107–109.
(17) Blinov, K. A.; Williams, A. J.; Hilton, B. D.; Irish, P. A.; Martin,
G. E. Magn. Reson. Chem. 2007, 45, 544–546.
(18) Martin, G. E.; Irish, P. A.; Hilton, B. D.; Blinov, K. A.; Williams,
A. J. Magn. Reson. Chem. 2007, 45, 624–627.
(19) Martin, G. E.; Hilton, B. D.; Irish, P. A.; Blinov, K. A.; Williams,
A. J. J. Heterocycl. Chem. 2007, 44 in press.
(20) Kupcˇe, E.; Freeman, R. Magn. Reson. Chem. 2007, 45, 103–105.
NP070221J
1396 The Journal of Natural Products, 2007, Vol. 70, No. 9 Rapid Communications