Jul-Aug 2008 Unsymmetrical Indirect Covariance Processing of Hyphenated and
Long-Range Heteronuclear 2D NMR Spectra – Enhanced
Visualization of 2JCH and 4JCH Correlation Responses
1109
Gary E. Martin,* and Bruce D. Hilton

Schering-Plough Research Institute,
Rapid Structure Characterization Laboratory
Pharmaceutical Sciences, Summit, NJ 07901

Kirill A. Blinov

Advanced Chemistry Development
Moscow Department, Moscow 117513
Russian Federation

and

Antony J. Williams

ChemZoo, Inc.
Wake Forest, NC 27587
Received October 8 2007

[31.57]
[47.85]
[77.53]
[60.10]
N
[42.44]
[169.42]
[26.80]
[60.14]
[51.92]
[142.11]
[132.67]
[116.19]
[128.48]
[123.33]
[123.24]
N
[42.72]
[50.67]
[52.13]
O
H
H H
H
[140.27]
[127.70]
[64.58]
O
H
70 60 50 40 30
F1 Chemical Shift (ppm)
50
100
150 F
2

C
h
e
m
i
c
a
l

S


Recent reports have demonstrated the unsymmetrical indirect covariance combination of discretely
acquired 2D NMR experiments into spectra that provide an alternative means of accessing the information
content of these spectra. The method can be thought of as being analogous to the Fourier transform
conversion of time domain data into the more readily interpreted frequency domain. Hyphenated 2D-NMR
spectra such as GHSQC-TOCSY, when available, provide an investigator with the means of sorting
proton-proton homonuclear connectivity networks as a function of the 13C chemical shift of the carbon
directly bound to the proton from which propagation begins. Long-range heteronuclear chemical shift
correlation experiments establish proton-carbon correlations via heteronuclear coupling pathways, most
commonly across three bonds (3JCH), but in more general terms across two (2JCH) to four bonds (4JCH). In
many instances 3JCH correlations dominate GHMBC spectra. We demonstrate in this report the improved
visualization of 2JCH and 4JCH correlations through the unsymmetrical indirect covariance processing of
GHSQC-TOCSY and GHMBC 2D spectra.


J. Heterocyclic Chem., 45, 1109 (2008).


INTRODUCTION
Recent investigations of the possibilities provided by
unsymmetrical indirect covariance (UIDC) processing
algorithms [1] have shown that it is possible to combine
1H-13C heteronuclear 2D NMR experiments with various
homonuclear 2D NMR experiments to produce the
equivalent of hyphenated 2D NMR spectra [2-4]. In a
similar fashion, 1H-13C direct and 1H-13C long-range
correlation experiments can be co-processed to afford the
equivalent of m,n-ADEQUATE spectra [5,6]. 1H-13C and
1H-15N heteronuclear 2D NMR spectra can also be co-

1110 G. Martin, B. Hilton, K. Blinov, and A. Williams Vol 45

processed to afford various types of 13C-15N heteronuclear
correlation spectra [7-9]. Most recently, we have
demonstrated the utility of indirect covariance processing
of GHSQC spectra as a means of identifying artifact
responses in various indirect and unsymmetrical indirect
covariance processed spectra [9,10]. Care should be
taken in segments of the spectra with overlapped
resonances due to the possibility of artifacts potentially
arising in association with those resonances.
Modern structure elucidation relies on the establishment
of atom-to-atom connectivity networks through the
identification of either homo- or heteronuclear coupling
pathways or, in some cases, via through-space dipole-
dipole interactions. In most instances homonuclear
correlations are observed using COSY or TOCSY
experiments across three or four bonds. Direct 1H-13C
heteronuclear correlation establishes proton-carbon
pairings; long-range 1H-13C correlation experiments
establish those correlations across typically two to four
bonds, with three-bond correlations being by far the most
prevalent. Generally, these experiments are employed to
assemble the carbon skeleton of a molecule being
characterized, and are supplemented, when necessary, by
direct or long-range 1H-15N heteronuclear shift correlation
data. In most cases these correlation data, when
interpreted in concert, allow an investigator to
successfully assemble the structure of an unknown
molecule. There are, however, cases in which access to
what would normally be weak long-range correlations can
be very beneficial. There also may be instances when a
correlation to a remote proton (e.g. 4JCH or >4JCH) will
resolve an ambiguity due to the overlap of less distant
proton resonances. Examples of both of these types of
problems are regularly encountered in the characterization
of complex natural product structures [11-13]. It is in this
vein that we were interested in exploring the possibility of
using unsymmetrical indirect covariance processing
methods to combine the proton-proton connectivity
networks that can be accessed in a GHSQC-TOCSY
experiment with long-range heteronuclear coupling
pathways from a GHMBC experiment.
RESULTS AND DISCUSSION
The processed IDR-(Inverted Direct Response)-GHS-
QC-TOCSY and GHMBC spectra were subjected to
unsymmetrical indirect covariance co-processing using
ACD/SpecManager. The resulting icv-(indirect covar-
iance calculated)-HSQC-TOCSY-HMBC spectrum der-
ived from the discretely acquired IDR-HSQC-TOCSY
and GHMBC spectra is shown in Figure 1. The assigned
13C chemical shifts of strychnine are shown by 1. Overlap
of the C8 and C16 carbon resonances at 60.1 ppm can
hamper the extraction of information from the HSQC-
TOCSY-HMBC spectrum shown in Figure 1 for these
positions as discussed below.


1

75 70 65 60 55 50 45 40 35 30 25
F1 Chemical Shift (ppm)
24
32
40
48
56
64
72
80
88
96
104
112
120
128
136
144
152
160
168
176
F
2

C
h
e
m
i
c
a
l

S
h
i
f
t

(
p
p
m
)

Figure 1. HSQC-TOCSY-HMBC spectrum calculated from a 16 ms
IDR-GHSQC-TOCSY spectrum and a 6 Hz optimized GHMBC
spectrum using unsymmetrical indirect covariance processing.

Connectivity information was individually extracted for
the aliphatic portion of strychnine from the icv-HSQC-
TOCSY-HMBC and GHMBC spectra as shown in
Figures 2 and 3, respectively. The overlap of C8 and C16
(60.10 and 60.14 ppm, respectively) precludes the
extraction of long-range correlation information for these
positions from the icv-HSQC-TOCSY-HMBC spectrum
shown in Figure 1. In contrast, the H8 and H16 proton
resonances are resolved at 500 MHz, allowing long-range
1H-13C connectivity information to be derived for these
positions from the 6 Hz optimized GHMBC spectrum.
The utility of an unsymmetrical indirect covariance co-
processed spectrum such as the icv-HSQC-TOCSY-
HMBC spectrum shown in Figure 1 derives from
information content less readily accessed from the

Jul-Aug 2008 Unsymmetrical Indirect Covariance Processing of 2D NMR Spectra 1111

discretely acquired spectra that were co-processed.
Conversely, a limitation is imposed when there is 13C
resonance overlap as in the case of the C8 and C16
resonances which both resonate at ~60.1 ppm. To
evaluate the correlation information obtained from the
icv-HSQC-TOCSY-HMBC spectrum shown in Figure 1,
the correlations shown in Figures 2 and 3 were compared.
Correlations contained in both the icv-HSQC-TOCSY-
HMBC and conventional HMBC spectra were eliminated,
and the remaining correlation information unique to the
unsymmetrical indirect covariance co-processed spectrum
is shown in Figure 4. In comparison, long-range 1H-13C
connectivity information contained only in the 6 Hz
optimized GHMBC spectrum is shown in Figure 5.

[31.57]
[47.85]
[77.53]
[60.10]
N
[42.44]
[169.42]
[26.80]
[60.14]
[51.92]
[142.11]
[132.67]
[116.19]
[128.48]
[123.33]
[123.24]
N
[42.72]
[50.67]
[52.13]
O
H
H H
H
[140.27]
[127.70]
[64.58]
O
H


Figure 2. Long-range connectivity information extracted from the
aliphatic portion of the unsymmetrical indirect covariance processed
GHSQC-TOCSY-HMBC spectrum of strychnine (1) derived by co-
processing 16 msec IDR-GHSQC-TOCSY and 6 Hz optimized GHMBC
spectra. Correlations for the C8 (60.10 ppm; refer to structure 1 for
numbering scheme) and C16 (60.14 ppm) cannot be extracted from the
spectrum shown in Figure 1 because of the overlap of these resonances
at 125 MHz. (Correlation pathways are color-coded identically for
specific positions in Figures 2 and 3 to facilitate comparison.)

Inspection of the correlations unique to the icv-HSQC-
TOCSY-HMBC spectrum shown in Figure 4 shows that
the majority of the correlations not observed in the
GHMBC spectra are two- or four-bond correlations that
will generally be much weaker than the more prevalent
three-bond correlations typically observed in the GHMBC
spectrum. In this sense, the co-processed spectrum is
effectively enhancing the visibility of weaker correlation
responses relative to the noise threshold of the spectrum.
Two- and four-bond correlations can also be accessed in
accordion-optimized long-range heteronuclear shift
correlation experiments that have been reviewed [15]. In
contrast, apart from the correlations from the H8 and H16
resonances shown in Figure 5, there are only four other
correlations observed in the GHMBC spectrum that are
not observed in the unsymmetrical indirect covariance
processed HSQC-TOCSY-HMBC spectrum (Figures 1
and 2).
[31.57]
[47.85]
[77.53]
[60.10]
N
[42.44]
[169.42]
[26.80]
[60.14]
[51.92]
[142.11]
[132.67]
[116.19]
[128.48]
[123.33]
[123.24]
N
[42.72]
[50.67]
[52.13]
O
H
H H
H
[140.27]
[127.70]
[64.58]
O
H
weak
weak

Figure 3. Long-range 1H-13C correlations extracted from the 6 Hz
optimized GHMBC spectrum of strychnine. Since the H8 and H16
protons (refer to structure 1 for numbering scheme) are resolved, unlike
their corresponding carbons, which are overlapped, connectivity
information can be obtained for these positions from the GHMBC
spectrum. (Correlation pathways are color-coded identically for specific
positions in Figures 2 and 3 to facilitate comparison.)

[31.57]
[47.85]
[77.53]
[60.10]
N
[42.44]
[169.42]
[26.80]
[60.14]
[51.92]
[142.11]
[132.67]
[116.19]
[128.48]
[123.33]
[123.24]
N
[42.72]
[50.67]
[52.13]
O
H
H H
H
[140.27]
[127.70]
[64.58]
O
H
Figure 4. Correlations obtained from the unsymmetrical indirect
covariance processed HSQC-TOCSY-HMBC spectrum not observable
in the GHMBC spectrum at the common threshold level used to prepare
the plot shown in Figure 1. Correlation arrows are color coded for
individual positions in the structure in a manner identical to Figures 2
and 3 to facilitate direct comparison. New, long-range connectivity
information has not been created by unsymmetrical indirect covariance
co-processing of the IDR-GHSQC-TOCSY and GHMBC spectra –
rather, very weak correlations below the identical thresholds of the IDR-
GHSQC-TOCSY and GHMBC spectra are rendered more visible as a
result of the processing, making the utilization of this information viable
whereas the very weak correlations in the discrete experiment might be
considered suspect.

The unsymmetrical indirect covariance processed icv-
HSQC-TOCSY-HMBC spectrum shown in Figure 1
should not be misconstrued as providing connectivity
information that is not contained in the discretely acquired
spectra that have been processed together. Rather, as
demonstrated in several instances with calculated icv-

1112 G. Martin, B. Hilton, K. Blinov, and A. Williams Vol 45

HSQC-COSY spectra [2,3] the co-processing algorithm
provides higher relative s/n ratios in the same fashion as
Fourier transforming a normal 2D NMR spectrum [16].
[31.57]
[47.85]
[77.53]
[60.10]
N
[42.44]
[169.42]
[26.80]
[60.14]
[51.92]
[142.11]
[132.67]
[116.19]
[128.48]
[123.33]
[123.24]
N
[42.72]
[50.67]
[52.13]
O
H
H H
H
[140.27]
[127.70]
[64.58]
O
H
weak
very weak

Figure 5. Long-range 1H-13C connectivity information unique to the 6
Hz optimized GHMBC spectrum. While the correlations for H8 (C8,
60.10 ppm; refer to structure 1 for numbering scheme) and H16 (C16,
60.14 ppm) may well be contained in the HSQC-TOCSY-HMBC
spectrum, these correlations cannot be reliably differentiated and used
because of the overlap of their respective carbon resonances. In contrast,
these correlations are available in the GHMBC spectrum and
compliment the information content of the HSQC-TOCSY-HMBC plot.

In this sense, legitimate connectivity information
contained in the spectra being co-processed that is weak
or near the 2D contour plot threshold is observed after co-
processing with higher apparent s/n allowing this
information to be reliably used rather than question the
legitimacy of the response(s) in question.
CONCULSIONS
Using unsymmetrical indirect covariance processing to
co-process IDR-GHSQC-TOCSY and GHMBC spectra to
produce an icv-HSQC-TOCSY-HMBC spectrum leads to
the reliable visualization of weaker two- and four-bond
correlations that would be expected to be contained in the
discretely acquired GHMBC spectrum used in the co-
processing. In most instances, it should be possible to
solve a structure from the available GHSQC-TOCSY and
GHMBC spectra without resorting to the use of
unsymmetrical indirect covariance processing. There
may, however, be instances when there are proton
resonance overlaps where having the connectivity
information spread as a function of 13C shifts on both axes
will be beneficial. Alternatively, in those cases where
reliable access to two- and or four-bond correlations is
necessary to confirm a structure, the icv-HSQC-TOCSY-
HMBC presentation may be a useful adjunct to the
interpretation of the spectra and the assignment of a given
structure.
EXPERIMENTAL
All of the NMR data employed in this study were acquired
using a sample of ~5 mg of strychnine dissolved in ~180 μL
deuterochloroform (CIL) which was then transferred to a 3 mm
NMR tube (Wilmad) using a flexible Teflon needle and a gas-
tight syringe (Hamilton). Data were acquired using a Varian
500 MHz two channel NMR instrument equipped with a Varian
3 mm gradient inverse detection probe; the sample temperature
was regulated at 26 °C. The pulse sequences for the exper-
iments performed were used directly from the vendor-supplied
pulse sequence library without further modification. The
GHSQC-TOCSY data were acquired with inversion of the direct
responses (IDR) and a 16 ms mixing time [13]. The long-range
delay in the GHMBC experiment was optimized for 6 Hz.
2D NMR spectra were acquired with identical spectral widths
in the F2 frequency domain although this is not specifically
required for the ACD/SpecManager v10.02 software used in this
study to perform the co-processing of the data. The spectral
width in the F1 frequency domain was set as appropriate for each
experiment. Weighting functions were optimized for each
experiment and the data were processed via a combination of
linear prediction and zero-filling in the second frequency
domain to afford data matrices for both experiments of 2048 x
512 points. The processed spectra were subjected to
unsymmetrical indirect covariance processing using ACD/
SpecManager software v10.02. The processing was done on a
Dell laptop computer with a 1.7 GHz processor and 1 Gb of
RAM. The unsymmetrical indirect covariance processing
required ~ 5 sec.
REFERENCES AND NOTES
[1] Blinov, K. A.; Larin, N. I.; Kvasha, M. P.; Moser, A.;
Williams, A. J.; Martin, G. E. Magn. Reson. Chem. 2005, 43, 999.
[2] Blinov, K. A.; Larin, N. I.; Williams, A. J.; Mills, K. A.,
Martin, G. E. J. Heterocycl. Chem., 2006, 43, 163.
[3] Martin, G. E.; Hilton, B. D.; Irish, P. A.; Blinov, K. A.;
Williams, A. J. J. Nat. Prod., 2007, 70, 1393.
[4] Blinov, K. A.; Williams, A. J. Hilton, B. D.; Irish, P. A.;
Martin, G. E. Magn. Reson. Chem., 2007, 45, 544.
[5] Blinov, K. A.; Larin, N. I.; Williams, A. J.; Zell, M.; Martin,
G. E. Magn. Reson. Chem., 2006, 44, 107.
[6] Schoefberger, W. ;Smreki, V.; Viki-Topi, D.; Müller, N.
Magn. Reson. Chem., 2007, 45, 583.
[7] Martin, G. E.; Hilton, B. D.; Irish, P. A.; Blinov, K. A.;
Williams, A. J., Magn. Reson. Chem., 2007, 45, 624.
[8] Martin, G. E.; Hilton, B. D.; Irish, P. A.; Blinov, K. A.;
Williams, A. J. J. Heterocycl. Chem., 2007, 44, 1219.
[9] Martin, G. E.; Hilton, B. D.; Blinov, K. A.; Williams, A. J. J.
Nat. Prod., 2008, 46, 138.
[10] Martin, G. E.; Hilton, B. D.; Blinov, K. A.; Williams, A. J.
Magn. Reson. Chem. 2007, 45, in press.
[11] 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.
[12] Martin, G. E.; Hadden, C. E.; Russell, D. J.; Kaluzny, B. D.;
Guido, J. E.; Duholke, W. K.; Stiemsma, B. A.; Thamann, T. J.; Crouch,
R. C.; Blinov, K. A.; Elyashberg, M.; Martirosian, E. R.; Molodtsov, S.
G.; Williams, A. J.; Schiff, P. L., Jr. J. Heterocycl. Chem., 2002, 39,
1241.
[13] Blinov, K.; Elyashberg, M.; Martirosian, E. R.; Molodtsov, S.
G.; Williams, A. J.; Tackie, A. N.; Sharaf, M. H. M.; Schiff, P. L., Jr.;
Crouch, R. C.; Martin, G. E.; Hadden, C. E.; Guido, J. E.; Mills, K. A.;

Jul-Aug 2008 Unsymmetrical Indirect Covariance Processing of 2D NMR Spectra 1113

Magn. Reson. Chem., 2003, 41, 577.
[14] Martin, G. E.; Spitzer, T. D.; Crouch, R. C.; Luo, J.-K.;
Castle, R. N. J. Heterocycl. Chem., 1992, 29, 557.
[15] Martin, G. E. in Ann. Rep. NMR Spectrosc. Webb, G. A., Ed.,
Academic Press, New York, 2002, vol. 46, pp. 37-100.
[16] In several reports (see refs. 2-4) it has been feasible to
perform the experiment comparable to the indirect covariance or
unsymmetrical indirect covariance processed data allowing direct s/n
comparison of the experimental and calculated spectra to be made. In
contrast, in the present case, there is no experimental equivalent to the
spectrum derived by unsymmetrical indirect covariance processing
shown in Figure 1. Consequently, we can only express the results of
unsymmetrical indirect covariance processing vs. the initial GHMBC

spectrum in relative terms. The weaker 2JCH and 4JCH correlations are
present in the GHMBC spectrum but were below the threshold used to
prepare contour plots of those data. In contrast, these responses are
observed above the threshold used to prepare the contour plot shown in
Figure 1. A direct comparison would entail plotting 1H spectral slices
from the GHMBC spectrum at a given 13C shift vs. 13C “spectra” at a
given 13C chemical shift from the spectrum shown in Figure 1, which the
authors did not feel was a legitimate compareison.