Sep-Oct 2007 13C-15N Connectivity Networks via Unsymmetrical
Indirect Covariance Processing of 1H-13C HSQC and
1H-15N IMPEACH Spectra
1219
Gary E. Martin,* Bruce D. Hilton, and Patrick A. Irish

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

Advanced Chemistry Development, Toronto, Ontario, M5C 1T4, Canada
Received April 12, 2007



Long-range 1H-15N heteronuclear shift correlation methods at natural abundance to facilitate the
elucidation of small molecule structures have assumed a role of growing importance over the past decade.
Recently, there has also been a high level of interest in the exploration of indirect covariance NMR
methods. From two coherence transfer experiments, A
B and A
C, it is possible to indirectly determine
BC. We have shown that unsymmetrical indirect covariance methods can be employed to indirectly
determine several types of hyphenated 2D NMR data from higher sensitivity experiments. Examples
include the calculation of hyphenated 2D NMR spectra such as 2D GHSQC-COSY and GHSQC-NOESY
from the discrete component 2D NMR experiments. We now wish to report the further extension of
unsymmetrical indirect covariance NMR methods for the combination of 1H-13C GHSQC and 1H-15N long-
range (GHMBC, IMPEACH-MBC, CIGAR-HMBC, etc.) heteronuclear chemical shift correlation spectra
to establish 13C-15N correlation pathways.


J. Heterocyclic Chem., 44, 1219 (2007).


Long-range 1H-15N heteronuclear chemical shift
correlation methods at natural abundance have been the
subject of several recent reviews [1-5]. The growing
importance of being able to access 1H-15N long-range
heteronuclear shift correlation data for the characterize-
ation of natural products and pharmaceuticals has fostered
two recent reports describing pulse sequences that allow
the simultaneous acquisition of long-range 1H-13C and 1H-
15N HMBC data [6,7]. Considerable recent attention has
also been focused on the development of indirect
covariance NMR methods, first reported by Brüschweiler
and co-workers [8-13]. Insofar as potential for small
molecule applications, to the authors, the most interesting
report was that describing indirect covariance methods,
which afford the capability of extracting carbon-carbon
connectivity information from a GHSQC-TOCSY
spectrum [11]. In their report, Brüschweiler and Zhang
commented that proton resonance overlap could lead to
artifacts in the calculated carbon-carbon correlation
spectra although they did not explore or document that
observation [11]. We subsequently described the analysis
of two types of artifacts observed in IDR-(Inverted Direct
Response)-GHSQC-TOCSY spectra with overlapped
proton resonances, which, in turn, prompted us to explore
the elimination of these artifacts via a method that we
have named unsymmetrical indirect covariance [14].
Subsequent work has shown that it is also possible to
mathematically combine various discretely acquired 2D
NMR spectra. The calculation of hyphenated 2D NMR
experiments from discretely acquired 2D spectra is based

1220 G. Martin, B. Hilton, P. Irish, K. Blinov, and A. Williams Vol 44

on the premise that using coherence transfer experiments
of the type A
B and A
C, one can indirectly determine
BC. The first effort in this direction demonstrated the
combination of 1H-13C GHSQC and GHMBC spectra to
afford the equivalent of an m,n-ADEQUATE spectrum
[15]. Subsequent studies have demonstrated the calcula-
tion of GHSQC-COSY [16,17] and GHSQC-NOESY [18]
spectra from discretely acquired COSY, NOESY, and 1H-
13C GHSQC 2D NMR spectra. Recently, Kupe and
Freeman [7] have demonstrated the use of projection
reconstruction techniques to establish 15N-13C correlations
at natural abundance, using vitamin B-12 as a model
compound for their study, also noting in parallel that
indirect covariance methods can be used to obtain
homonuclear correlation spectra indirectly. Thus, we
would now like to demonstrate specifically, that
unsymmetrical indirect covariance NMR methods can
also be used to derive 15N-13C connectivity information
from discretely acquired 1H-13C GHSQC and 1H-15N
HMBC spectra.
For the present study, multiplicity-edited 1H-13C
GHSQC and 1H-15N IMPEACH-MBC (IMPEACH
hereafter) spectra were acquired using an ~5 mg sample
of (-) eburnamonine (1) dissolved in ~180 μL CDCl3 in a
3 mm NMR tube. The spectra were recorded at 600 MHz
using a Varian three channel spectrometer equipped with
a 3 mm gradient inverse detection probe at 26° C. The 1H-
13C GHSQC spectrum was recorded in 23 m as 1024 x 96
data points; the 1H-15N IMPEACH spectrum was recorded
in 16.2 h as 1024 x 66 data points. The spectra were
acquired with identical proton spectral widths and the data
were processed to yield data matrices that were identically
digitized with F1,F2 dimensions of 512 x 2048 points.
The processed 1H-13C GHSQC and 1H-15N IMPEACH
spectra were subjected to unsymmetrical indirect
covariance processing to yield the 13C-15N HSQC-
IMPEACH long-range correlation spectrum shown in
Figure 1. The 1H-15N IMPEACH spectrum is shown in
the top left panel and corresponds to the A
B coherence
transfer spectrum; the multiplicity-edited 1H-13C GHSQC
spectrum is shown in the bottom right panel and can be
considered the A
C coherence transfer experiment. The
GHSQC spectrum has been transposed to reflect the
eventual orientation of the 13C spectrum as the F2 axis in
the indirectly determined 13C-15N correlation spectrum
shown in the top right panel, which corresponds to the
BC coherence transfer spectrum. In the 13C-15N
correlation spectrum, 15N chemical shift information is
displayed in the F1 frequency domain while 13C chemical
shift information is presented in the F2 frequency domain.
13C-15N correlation responses in the spectrum shown in
Figure 1 arise from transfer between 13C and 15N via the
nJNH correlation response in the 1H-15N IMPEACH
spectrum, the 13C chemical shift information derives from
the chemical shift of the carbon directly bound to the
proton in question in the 1H-13C GHSQC spectrum. 2JNH
long-range correlations lead to 13C-15N direct correlation
responses in the spectrum shown in Figure 1. 3JNH and
4JNH long-range correlations in the 1H-15N GHMBC give
rise to 13C-15N responses corresponding to correlations via
two- and three-bonds, respectively.



13C-15N Heteronuclear correlation responses observed in
Figure 1 are shown on the structure above. The 15N neural
network shift calculations are shown parenthetically
(ACD/Labs, NNMR Predictor v10.02), accompanied by the
observed 15N shifts. Correlations plotted with red contours
in the 13C-15N correlation spectrum shown in Figure 1 arise
from correlations between methylene carbons and the
nitrogen; correlations plotted in black arise from correlations
from methine carbons to nitrogen (or methyl carbons,
although there are no methyl groups correlated to nitrogen
in the1H-15N IMPEACH spectrum of (-) eburnamonine).
The phase of the 13C resonance from the multiplicity-
editing of responses in the GHSQC spectrum is carried
forward into the 13C-15N HSQC-IMPEACH unsymmetrical
indirect covariance processed spectrum shown in the top
right panel in Figure 1. Presumably, methine and
methylene carbons with the same 13C chemical shift that
correlate to the same nitrogen could partially or completely
cancel, hence it may be useful to consider the acquisition of
both conventional and multiplicity-edited 1H-13C GHSQC
spectra when the data are acquired since these data can be
accumulated in a very reasonable period of time. For
weaker samples, the sensitivity loss associated with
multiplicity-editing (~20%) may obviate the acquisition of
a multiplicity-edited GHSQC spectrum in any case. It is
also interesting to note that in the 13C-15N HSQC-
IMPEACH correlation spectrum shown in Figure 1, the
C10 and C19 resonances, which have identical 13C
chemical shifts (~44.7 ppm), correlate to N1 and N4,
respectively, but do not produce artifacts as seen in the
case of IDR-GHSQC-TOCSY spectra with overlapping
proton resonances as in our previous work [14]. An
interesting corollary arises in the case of overlapped