Toward the creation of NMR databases in chiral solvents for assignments of relative and absolute stereochemistry: proof of concept.

Article abstract of DOI:10.1021/ol010108z

[structure: see text] An NMR database approach in a chiral solvent allows us to predict both relative and absolute stereochemistry of an unknown compound without degradation and/or derivatization. N,alpha-Dimethylbenzylamine (DMBA) is a suitable solvent for this purpose. Using the C.5-C.10 portion of oasomycin A, the feasibility and reliability of this approach is demonstrated.

Full text of DOI:10.1021/ol010108z

ORGANIC
LETTERS
2001
Vol. 3, No. 14
2245-2248
Toward the Creation of NMR Databases
in Chiral Solvents for Assignments of
Relative and Absolute Stereochemistry:
Proof of Concept
Yoshihisa Kobayashi, Nobuyuki Hayashi, Choon-Hong Tan, and Yoshito Kishi*
Department of Chemistry and Chemical Biology, HarVard UniVersity,
12 Oxford Street, Cambridge, Massachusetts 02138
kishi@chemistry.harVard.edu
Received May 23, 2001
ABSTRACT
An NMR database approach in a chiral solvent allows us to predict both relative and absolute stereochemistry of an unknown compound
without degradation and/or derivatization. N,r-Dimethylbenzylamine (DMBA) is a suitable solvent for this purpose. Using the C.5−C.10 portion
of oasomycin A, the feasibility and reliability of this approach is demonstrated.
Our recent endeavors toward creation of universal NMR
databases for stereochemical assignments have culminated
in the establishment of the complete structure of the
desertomycin/oasomycin class of natural products.1 In this
work, we treated structural motifs present in the antibiotics
separately from each other and compared their NMR profiles
with the corresponding NMR databases to predict their
relative stereochemistry and then confirmed the predicted
relative stereochemistry by enantioselective synthesis of each
segment. In our view, since the reliability of this approach
has been well demonstrated, confirmation of the predicted
relative stereochemistry by chemical synthesis is no longer
required. However, an enantioselective synthesis has pro-
vided us with information about the absolute configuration
of each segment, which was essential for us to correlate the
relative stereochemistry of one structural motif with the
others. In this context, we recognized the possibility that the
absolute, as well as relative, stereochemistry of a given
structural motif could be established through an NMR
database approach in a chiral solvent.2-4 In this and the
(2) Various methods have been used to establish the absolute configu-
ration of an unknown compound. In connection with the current work,
however, there is no known method to determine the absolute configuration
of an intact molecule by NMR spectroscopy. Among the methods combining
derivatization and NMR spectroscopy, the Kusumi-Kakisawa NMR
analysis of Mosher esters is perhaps most widely used: Ohtani, I.; Kusumi,
T.; Kashman, Y.; Kakisawa, H. J. Am. Chem. Soc. 1991, 113, 4092-4096.
(3) Four chiral solvents were reported for NMR discrimination of
enantiomers: see, (a) Pirkle, W. H. J. Am. Chem. Soc. 1966, 88, 1837. (b)
Pirkle, W. H.; Hoekstra, M. S. J. Magn. Reson. 1975, 18, 396-400 and
references therein.
(1) (a) Kobayashi, Y.; Lee, J.; Tezuka, K.; Kishi, Y. Org. Lett. 1999, 1,
2177-2180. (b) Lee, J.; Kobayashi, Y.; Tezuka, K.; Kishi, Y. Org. Lett.
1999, 1, 2181-2184. (c) Kobayashi, Y.; Tan, C.-H.; Kishi, Y. HelV. Chim.
Acta 2000, 83, 2562-2571. (d) Kobayashi, Y.; Tan, C.-H.; Kishi, Y. Angew.
Chem., Int. Ed. 2000, 39, 4279-4281. (e) Tan, C.-H.; Kobayashi, Y.; Kishi,
Y. Angew. Chem., Int. Ed. 2000, 39, 4282-4284. (f) Kobayashi, Y.; Tan,
C.-H.; Kishi, Y. J. Am. Chem. Soc. 2001, 123, 2076-2078.
(4) For reviews on subjects related to the current work, see: (a) Pirkle,
W. H.; Hoover, D. J. Top. Stereochem. 1982, 13, 263-331. (b) Parker, D.
Chem. ReV. 1991, 91, 1441-1457.
10.1021/ol010108z CCC: $20.00 © 2001 American Chemical Society
Published on Web 06/20/2001

following Letters, we would like to report the realization of
this concept.
the 13C NMR spectra were recorded for the diastereomers
1a-h (optically active)1a of both (R)- and (S)-DMBA to
create the NMR database for the contiguous dipropionate
motif.6,7
The first step of our efforts was to identify a suitable chiral
NMR solvent that must meet several criteria, including, in
our view, the following: (1) a high capacity to discriminate
enantiomers, (2) a good capacity to dissolve organic com-
pounds, (3) a characteristic ideally similar to that of DMSO
or MeOH (vide infra), (4) chemical inertness, (5) a viscosity,
boiling point, and freezing point suitable for NMR experi-
ments, and (6) an easy access to its deuterated form. Using
the arbitrarily chosen substrate 1e (Figure 1), we conducted
Figure 1 shows the 13C NMR database in (R)- and (S)-
DMBA as a deviation in chemical shift for each carbon of
a given diastereomer from the average chemical shift of the
carbon in question. Each diastereomer exhibits an almost
identical NMR profile for (R)- and (S)-DMBA but shows
an NMR profile distinct and differing from the other
diastereomers, demonstrating that the database in (R)- and/
or (S)-DMBA can be used for prediction of the relative
stereochemistry of structural motifs in an intact form.
However, we should comment on the solvent-effect studies
conducted in the achiral solvent series; the overall NMR
profile for each diastereomer of 1a-h in DMSO was
virtually identical to that in methanol, whereas the overall
NMR profile in DMSO or methanol was significantly
different from that in chloroform.1a On the basis of this
observation, we concluded that DMSO or methanol is the
solvent of choice for the NMR database approach. Interest-
ingly, DMBA gives an overall NMR profile between DMSO
(or methanol) and chloroform, suggesting that DMSO and
methanol are superior to DMBA for the purpose of relative
stereochemistry assignment. On comparison of the graphs
in Figures 1 and 2, this situation becomes clearer; each
diastereomer 1a-h shows an NMR profile more distinct and
differing from the other diastereomers in DMSO than in
DMBA.8 Nonetheless, we should emphasize that DMBA is
still a viable solvent for the relative stereochemistry assign-
ment work.
Figure 1. Difference in carbon chemical shifts between the average
and the values for 1a-h (100 MHz): solid bar, (R)-DMBA; shaded
bar, (S)-DMBA. The x- and y-axes represent carbon number and
∆δ (δ1a-h - δave in ppm), respectively.
preliminary screenings of solvents, from which (R)- and (S)-
N,R-dimethylbenzylamines (PhCH(Me)NHMe, DMBA)
emerged as the chiral solvent of choice for the current work.5
The second step was to measure the 13C NMR spectra of all
diastereomers of a given NMR database, and we chose to
use the contiguous dipropionate NMR database to demon-
strate the feasibility and reliability of this approach. Thus,
Figure 2. Difference in carbon chemical shifts between the average
and the values for 1a-h (100 MHz, DMSO-d6). The x- and y-axes
represent carbon number and ∆δ (δ1a-h - δave in ppm), respectively.
(5) Over 20 chiral solvents were tested, including sulfoxides (CF3SOCD3,
PhSOCD3, etc.), alcohols (MeCH(OH)Et, PhCH(CF3)(OH), etc.), amide
(PHCH(Me)N(Me)CHO), and others.
2246
Org. Lett., Vol. 3, No. 14, 2001

Figure 3 shows the 13C chemical shift differences between
the spectra of 1a-h in (R)- and (S)-DMBA. Under the
(R)- and (S)-DMBA,12 and the preparation included in the
Supporting Information meets with our needs. We were
pleased to observe that (R)- and (S)-DMBA-d13 could easily
dissolve oasomycin A13 (>10 mg/0.5 mL) but disappointed
to see that the NMR signals were very broad in DMBA.
Since aggregation appeared to be the probable cause for the
observed signal broadening, we tested (R)- and (S)-DMBA-
d13 containing 9.1% DMSO-d6 (DMBA/DMSO ) 10:1) and
were delighted to observe sharp, well-resolved signals.14,15
We then examined the capacity to discriminate enantiomers
and found that although the degree of chemical shift
differences decreases with an increase in DMSO content,
(R)- and (S)-DMBA-d13 containing 9.1% DMSO-d6 satis-
factorily met with our needs (Figure 4).
Figure 4. Difference in carbon chemical shifts for C.5 and C.7 of
1e (100 MHz) in DMBA containing DMSO-d6. The x- and y-axes
represent DMSO-d6 content (v/v %) in (R)- or (S)-DMBA and
|∆δR-S| in ppm, respectively.
Figure 3. Difference in carbon chemical shifts of 1a-h (100 MHz)
between (R)- and (S)-DMBA. The x- and y-axes represent carbon
number and δR - δS in ppm, respectively.
To correlate the NMR profile of the C.5-C.10 segment
of oasomycin A (Figure 5) with that of 1a-h, it was
conditions of measurement, a chemical shift difference of
0.010 ppm is reliably detected.9 The 13C chemical shift
differences observed in (R)- and (S)-DMBA are well beyond
this limit for 1a-h. Thus, the absolute configuration of a
given dipropionate motif can be deduced through comparison
of the 13C NMR profile in (R)-DMBA with that in (S)-
DMBA, in reference to the corresponding diastereomer
among 1a-h.
To assess the feasibility and reliability of the NMR
database approach in a chiral solvent, we chose first to use
the C.5-C.10 portion of the desertomycin/oasomycin class
of natural products.10,11 For this purpose, we needed to
develop a practical and scalable synthesis of perdeuterated
Figure 5. Structure of oasomycin A.
(6) A Varian Mercury 400 spectrometer (100 MHz) was used to collect
all data for 1a-h in DMBA, with acetone-d6 as an external reference (δ
29.8) and a lock-signal and with readout of NMR spectra being adjusted to
0.001 ppm/point (sw ) 23980.8, fn ) 524288).
(7) The chemical shift assignments were established via the same
procedure as the one used for oasomycin A. The details are included in the
Supporting Information.
necessary to establish the 13C chemical shift assignment for
the antibiotic in (R)- or (S)-DMBA-d13 containing 9.1%
DMSO-d6. Given the complete chemical shift assignment
(8) In DMBA, the overall profile of 1b and 1f is similar to that of 1d
and 1h, respectively. In DMSO, however, the overall profile of 1b is notably
different from that of 1d. In addition, the difference between 1f and 1h is
more visible in DMSO than in DMBA.
(9) Half-bandwidth was 0.008 ppm in nondeuterated DMBA.
(10) Desertomycins: (a) Bax, A.; Aszalos, A.; Dinya, Z.; Sudo, K. J.
Am. Chem. Soc. 1986, 108, 8056-8063. (b) Dinya, Z.; Sztaricskai, F.;
Horvath, E.; Schaag, J. B. Rapid Commun. Mass Spectrom. 1996, 10, 1439-
1448 and references therein.
(11) Oasomycins: (a) Grabley, S.; Kretzschmar, G.; Mayer, M.; Philipps,
S.; Thiericke, R.; Wink, J.; Zeeck, A. Liebigs Ann. Chem. 1993, 573-579.
(b) Mayer, M.; Thiericke, R. J. Chem. Soc., Perkin Trans. 1 1993, 2525-
2531.
(12) Although perdeuterated phenethylamine is known (Ohhara, T.;
Harada, J.; Ohashi, Y.; Tanaka, I.; Kumazawa, S.; Niimura, N. Acta
Crystallogr. 2000, B56, 245-253.), we prefer the method included in the
Supporting Information in terms of scalability, practicability, and cost.
(13) We thank Dr. Gerhard Kretzschmar for a sample of oasomycin A.
(14) A Varian INOVA 500 spectrometer (125 MHz) at Eisai Research
Institute of Boston was used to record all NMR spectra of oasomycin A in
(R)- and (S)-DMBA-d13, with DMSO-d6 as an internal reference (δ 39.5).
Eight out of 55 carbon signals of oasomycin A are hidden under the solvent
peaks (see the spectrum included in the Supporting Information).
(15) Enantiomeric discrimination of 1e was observed in DMBA contain-
ing 9.1% of the following cosolvents: CD3OD, CDCl3, CD3CN, C5D5N,
(CD3)2CO, and C6D6.
Org. Lett., Vol. 3, No. 14, 2001
2247

for this class of natural products in DMSO-d6,10a,11a we made
the chemical shift assignment through recording the NMR
spectrum in (R)-DMBA-d13 containing 9.1% (DMBA/DMSO
) 10:1), 33.3% (2:1), and 50% (1:1) DMSO-d6 and cor-
relating each signal in DMBA-d13/DMSO-d6 (9.1% v/v) with
that in DMSO-d6 (100%). Figure 6 shows the methyl group
region for illustration of this procedure.
Figure 6. Correlation of carbon chemical shifts of the methyl
groups of oasomycin A between DMSO-d6 and (R)-DMBA-d13.
The x- and y-axes represent the 9-18 ppm region and DMSO-d6
content (v/v %) in (R)-DMBA-d13, respectively. The numbers 47-
55 correspond to the numbers assigned to the methyl groups of
oasomycin A, cf. Figure 5.
Figure 7. Difference between adjusted carbon chemical shifts of
oasomycin A and those for 1a-h (∆δ ) δ1a-h - δoasomycin A, 100
MHz, (R)-DMBA-d13/DMSO-d6 (9.1% v/v)). The x- and y-axes
represent carbon number of oasomycin A and ∆δ in ppm,
respectively.
With the complete chemical shift assignment established,
we followed the previous procedure to correlate the 13C NMR
profile of each of the diastereomers 1a-h with that of the
C.5-C.10 portion of oasomycin A.1b The adjusted 13C NMR
profiles thus obtained (Figure 7) correctly predict that the
diastereomer 1e represents the relative stereochemistry of
the C.5-C.10 portion of oasomycin A.1b
The absolute configuration of the C.5-C.10 portion of
oasomycin A was then established through comparison of
its 13C NMR profile in (R)- and (S)-DMBA-d13 with that of
the diastereomer 1e representing the stereochemistry of this
portion of the antibiotic. The 13C chemical shift differences
(δR - δS) observed for C.7, C.9, C.10, C.48, and C.49 were
+0.002, -0.110, -0.050, -0.029, and -0.030 ppm, respec-
tively.14 The corresponding carbons of 1e gave ∆(δR - δS)
) +0.034, -0.039, -0.069, -0.015, and -0.027 ppm in
DMBA/DMSO-d6 (9.1% v/v). On the basis of this experi-
ment, the absolute configuration of the C.5-C.10 portion
of oasomycin A was concluded to be the same as the absolute
configuration of 1e. This conclusion agrees with the absolute
configuration previously established via the enantioselective
synthesis of the C.3-C.12 degradation product.1b
the NMR database approach in a chiral solvent. This
approach allows us to predict the absolute and relative
stereochemistry of a given, unknown compound without
degradation and/or derivatization work. In the following
Letter,16 we will present further extension of this approach.
Acknowledgment. We are most grateful to Dr. Yuan
Wang at Eisai Research Institute of Boston for NMR
measurements of oasomycin A. Financial support from the
National Institutes of Health (NS 12108) is gratefully
acknowledged.
Supporting Information Available: Chemical shift as-
signments and NMR databases of 1a-h in the chiral
solvents, NMR spectrum and chemical shift assignments of
oasomycin A in the chiral solvents, procedure for correlating
the NMR profile of the C.5-C.10 portion of oasomycin A
with those of 1a-h, and experimental procedures for the
synthesis of (R)- and (S)-DMBA-d13. This material is
available free of charge via the Internet at http://pubs.acs.org.
OL010108Z
In conclusion, using the C.5-C.10 portion of oasomycin
A, we have demonstrated the feasibility and reliability of
(16) Hayashi, N.; Kobayashi, Y.; Kishi, Y. Org. Lett. 2001, 3, 2249-
2252.
2248
Org. Lett., Vol. 3, No. 14, 2001