Structural analysis of pentacyclic triterpenes from the gum resin of Boswellia serrata by NMR spectroscopy

Article abstract of DOI:10.1002/mrc.1138

3α-Acetyl-β-boswellic acid (1), 3α-acetyl-α -boswellic acid (2), 3α-acetyl-9,11-dehydro-β-boswellic acid (3), 3α-acetyl-9,11-dehydro-α-boswellic acid (4) and 3α -acetyl-11-keto-β-boswellic acid (5) were isolated from the gum resin of Boswellia serrata. 1D and 2D NMR (COSY45, HMQC, HMBC, ROESY) spectra at 500 MHz were used for shift assignments and structure verification. All boswellic acids investigated share the cis conformation at ring D/E and the 3α orientation of the acetyl ester group. Owing to high-order spectra, NMR could not determine the exact conformation of H-20/H-30 of the β-boswellic acids. 3α-Acetyl-β-boswellic acid methyl ester (1′) was synthesized for experiments with a shift reagent, Eu(fod)3, that enhanced the resolution considerably. The oxygen atoms of the 3α-acetyl group form the apparent complex binding site for the shift reagent. Copyright

Full text of DOI:10.1002/mrc.1138

MAGNETIC RESONANCE IN CHEMISTRY
Magn. Reson. Chem. 2003; 41: 115–122
Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/mrc.1138
Structural analysis of pentacyclic triterpenes from the
gum resin of Boswellia serrata by NMR spectroscopy
Klaus Belsner,1 Berthold Bu¨ chele,1 Udo Werz,2 Tatiana Syrovets1 and Thomas Simmet1∗
1
Department of Pharmacology of Natural Products and Clinical Pharmacology, University of Ulm, D-89081 Ulm, Germany
Department of Organic Chemistry I, University of Ulm, D-89081 Ulm, Germany
2
Received 29 April 2002; Revised 1 October 2002; Accepted 9 October 2002
3a-Acetyl-b-boswellic acid (1), 3a-acetyl-a-boswellic acid (2), 3a-acetyl-9,11-dehydro-b-boswellic acid (3),
3a-acetyl-9,11-dehydro-a-boswellic acid (4) and 3a-acetyl-11-keto-b-boswellic acid (5) were isolated from
the gum resin of Boswellia serrata. 1D and 2D NMR (COSY45, HMQC, HMBC, ROESY) spectra at 500 MHz
were used for shift assignments and structure verification. All boswellic acids investigated share the cis
conformation at ring D/E and the 3a orientation of the acetyl ester group. Owing to high-order spectra, NMR
could not determine the exact conformation of H-20/H-30 of the b-boswellic acids. 3a-Acetyl-b-boswellic
acid methyl ester (1ꢀ) was synthesized for experiments with a shift reagent, Eu(fod)3, that enhanced the
resolution considerably. The oxygen atoms of the 3a-acetyl group form the apparent complex binding site
for the shift reagent. Copyright  2003 John Wiley & Sons, Ltd.
KEYWORDS: NMR; 1H NMR; 13C NMR; HMQC; HMBC; COSY; ROESY; shift reagent; pentacyclic triterpenes; boswellic acids
INTRODUCTION
EXPERIMENTAL
Samples
Extracts from the gum resin of Boswellia serrata may possess
therapeutic potential, as they have been claimed to possess
anti-inflammatory,1–3 antiviral4 and antitumor activities.5–9
The last effect may be due to inhibition of essential DNA
processing enzymes, the so-called topoisomerase I and II˛,
by boswellic acids.10,11 Boswellic acids (Fig. 1), putative
pharmacologically active principles of frankincense, were
first isolated in 1932.12 Structural details of 3˛-acetyl-ˇ-
boswellic acid, elucidated by traditional methods,13 have
recently been confirmed by x-ray crystallography.14 1D NMR
and GC/MS have also been applied.15,16 Unfortunately, mass
spectrometry yields ambiguous fragmentation patterns for ˛-
and ˇ-boswellic acids. Therefore, NMR spectroscopy seems
to be a more adequate method to resolve their molecular
configuration to be used in studies aimed at establishing
structure–activity relationships. To date, reported NMR data
are restricted solely to the equatorial orientation of H-3 and
3˛-Acetyl-ˇ-boswellic acid (1), 3˛-acetyl-˛-boswellic acid (2), 3˛-
acetyl-9,11-dehydro-ˇ-boswellic acid (3), 3˛-acetyl-9,11-dehydro-˛-
boswellic acid (4) and 3˛-acetyl-11-keto-ˇ-boswellic acid (5) were
extracted from the gum resin of Boswellia serrata.12 The raw
extract was purified by reversed-phase gradient HPLC to apparent
homogeneity.18 The purified compounds were subjected to mass and
NMR spectrometry, except boswellic acid (4), which was checked
only for its mass spectrum.
3˛-Acetyl-ˇ-boswellic acid (1): needles from methanol; m.p.
256–257 C (lit. 253 C); UV, ꢀmax 205 nm; EIMS, m/z 498 (MC) (7),
220 (11), 219 (21), 218 (100), 203 (15), 189 (11), 175 (7), 161 (9), 159 (5),
149 (6), 148 (6), 145 (6), 136 (9), 135 (11), 134 (7), 133 (13), 131 (5), 122
(13), 121 (13), 120 (5), 119 (16), 109 (12), 107 (15), 105 (15) (calc. for
C32H50O4: 498.75).
15
°
°
3˛-Acetyl-ˇ-boswellic acid methyl ester (10): amorphous powder
from methanol–water. A 20 mg amount of 1 was dissolved in
5 ml methanol, then 0.25 ml of water was added. Diazomethane
in diethyl ether was added dropwise until a distinct yellow color
remained. The solution was gently stirred in the dark for 30 min.
Remaining diazomethane was removed by flushing out the solvent
with nitrogen. The residue was dissolved in 25 ml diethyl ether and
washed with 5 ml dilute sodium hydroxide and water. The organic
phase was dried with anhydrous sodium sulfate. After evaporation
of the solvent, 15.1 mg (¾84%) of 10 were obtained.
1
identification of the olefinic H-12,17 while the crowded H-
3˛-Acetyl-˛-boswellic acid (2): needles from methanol; m.p.
spectrum between υ ¾2.2 and ¾0.8 ppm remains unresolved.
The goal of the present study was the detailed elucidation
of the structural configuration of boswellic acids to validate
reference materials and information required for further
molecular modeling studies aimed at the development of
novel lead compounds.
15
°
°
244 C with onset of decomposition, (lit. 245–248 C); UV, ꢀmax
205 nm; EIMS, identical with the ˇ-compound (1) (calc. for
C32H50O4: 498.75).
°
3˛-Acetyl-9,11-dehydro-ˇ-boswellic acid (3): m.p. 231–232 C
with onset of decomposition (lit.19 230 C); UV, ꢀmax 283 nm; EIMS,
°
m/z 498 (7), 497 (33) 496 (MC) (100), 452 (5), 436 (5), 421 (8), 392 (12),
283 (5), 256 (6), 255 (25), 218 (10), 203 (5), 185 (6), 173 (7), 171 (8), 159
(9), 157 (7), 147 (7), 145 (9), 143 (6), 133 (11), 123 (6), 119 (13), 109 (9),
107 (10), 105 (12) (calc. for C32H48O4: 496.74).
3˛-Acetyl-9,11-dehydro-˛-boswellic acid (4): isolated in minor
amounts, needles from methanol; UV, ꢀmax 283 nm; EIMS, m/z 498
(7), 497 (32), 496 (MC) (100), 421 (8), 392 (11), 255 (21), 173 (6), 171
(6), 157 (6), 147 (5), 145 (8), 143 (5), 133 (9), 131 (7), 121 (6), 119 (11),
109 (7), 107 (8), 105 (10) (calc. for C32H48O4: 496.74).
ŁCorrespondence to: Thomas Simmet, Department of
Pharmacology of Natural Products and Clinical Pharmacology,
University of Ulm, Helmholtzstr. 20, D-89081 Ulm, Germany.
E-mail: thomas.simmet@medizin.uni-ulm.de
Copyright  2003 John Wiley & Sons, Ltd.

116
K. Belsner et al.
pulse sequence a matrix of 256 ð 2048 (F1 ð F2) was recorded
and processed with sine or qsine along the F1 and F2 axes. The
HMQC spectral bandwidth was 2857.14 Hz in the F2 domain
and 9125 Hz for F1. Windows for HMBC were 2857.14 Hz
Oleanane Acids
R1
1
for H and 20 625 Hz for 13C, with a dataset of 228 ð 2048
(F1 ð F2).
For 2D spectra, 1H–1H-COSY45, inverse hydrogen
detected 1J(C,H) correlation, HMQC was optimized at 140 Hz
and HMBC at 7 Hz. The 2D ROESY experiment was con-
ducted with a spin lock period of 300 ms. Spectral windows
were 4032 and 1008 Hz with a dataset of 512 ð 2048 (F1 ð F2).
Eu(fod)3 shift reagent was used to resolve the proton–carbon
interactions of 3˛-acetyl-ˇ-boswellic acid methyl ester (10).
Ac
HO
O
30
29
Molecular modeling
Ursane Acids
20
28
21
22
Structures were modeled using the InsightII/Discover
software (Accelrys, San Diego, CA, USA). The starting
structure sketched according to the NMR restrictions was
optimized with the discover force-field (CVFF91, conjugated
gradient) and then run through a molecular dynamics
calculation from 0 to 450 K in 10 fs (10 000 steps) and kept
there for 100 fs (100 000 steps). From a plot of potential energy
versus time, nine conformers with low potential energy were
picked and again optimized. In addition to the molecular
mechanics calculation, these structures were minimized with
MOPAC/AM120,21 (RHF; BFG-gradient 0.1). The resulting
conformers were compared with x-ray results for 3˛-acetyl-
ˇ-boswellic acid (1)14 by least-squares coordinate fitting.
The generated pdb files were visualized and printed with
Swiss-PdbViewer.22
19
12
26
R1
18
11
13
17
25
1
4
14
16
9
2
8
10
15
27
7
3
5
Ac
6
24
23
HO
O
Figure 1. Triterpenic acids from the gum resin of Boswellia
serrata. ˛-Boswellic acids are based on the oleanane and
ˇ-boswellic acids contain the ursane skeleton. The free acids
are present besides their 3˛-acetyl esters, investigated here.
3-˛-Acetyl-ˇ-boswellic acid (1), R1 D H2;
3˛-acetyl-˛-boswellic acid (2), R1 D H2;
RESULTS AND DISCUSSION
3˛-acetyl-9,11-dehydro-ˇ-boswellic acid (3) R1 D H;
3˛-acetyl-9,11-dehydro-ˇ-boswellic acid (4), R1 D H;
3˛-acetyl-11-keto-ˇ-boswellic acid (5), R1 D O. The
9,11-dehydroboswellic acids in gum resin extracts originate
from an 11-hydroxy precursor, depending on the isolation
procedure.19
Analysis of the EI fragmentation of ˛- and ˇ-boswellic
acid and their 3˛-acetyl esters reproduced the known
retro-Diels–Alder fission between carbons 9–11 and 8–14.
Unfortunately, the mass spectra of 1 and 2 with the
main fragment at m/z 218 are nearly identical. Even the
introduction of a conjugate double bond between C-9 and C-
11, which prohibits the retro-Diels–Alder fission, leads only
to a marginal mass spectrometric differentiation between the
˛- (4) and the ˇ-diene (3).
Table 1 shows the results of the NMR signal assignments
of compounds 1 and 2; the corresponding assignments for
compounds 3 and 5 are given in Table 2. In NMR analysis
of all monoolefinic boswellic acids, the 1D-1H signals of the
equatorial17 H-3 and olefinic H-12 appear as weakly resolved
multiplets at υ ¾5.26 and 5.17 ppm, respectively. The H-12
multiplet agrees with dihedral angles of H-11˛ and H-11ˇ
3˛-Acetyl-11-keto-ˇ-boswellic acid (5): small crystals upon
°
precipitation from methanol–water; m.p. 262–265 C with onset of
decomposition (lit.15 271 C). The difference from the reported value
°
might be due to entrapped traces of solvent in our preparation. In
any case, analysis by NMR, MS, HPLC and TLC did not reveal any
contamination. UV, ꢀmax 252 nm; EIMS, m/z 512 (MC) (14), 409 (5),
408 (15), 406 (6), 393 (5), 299 (6), 287 (6), 274 (20), 273 (100), 271 (5),
233 (15), 232 (70), 217 (5), 203 (5), 189 (6), 175 (8), 173 (5), 161 (11),
149 (8), 145 (5), 136 (5), 135 (33), 133 (9), 125 (5), 123 (6), 121 (11), 119
(13), 105 (15) (calc. for C32H48O5: 512.74).
NMR spectra
°
°
(48.0 and 68 ) calculated with AM1. Proton signals of rings B
to E are accumulated in the region between ¾2 and ¾0.8 ppm
and generally are not resolved at 500 MHz (Fig. 2).
After obtaining 13C- and DEPT-135- and, in some cases,
J-coupling resolved 13C spectra, the individual carbons were
assigned. Figure 3 shows the different NMR interactions
found for 3˛-acetyl-˛-boswellic acid (2) and 3˛-acetyl-ˇ-
boswellic acid (1).
The NMR spectra were recorded on a Bruker AMX 500
spectrometer operating at 500 and 125 MHz at 300 K.
Generally ¾15 mg of the respective pentacyclic triterpene
acid were dissolved in CDCl3 with TMS as internal standard
and transferred to a 5 mm, 400 MHz NMR tube (Kontes,
Vineland, USA). One-dimensional 1H and 13C NMR spectra
were recorded under standard conditions with an inverse
5 mm probe head. The spectral window for 1H and 13C
was 10 204 Hz and 31 250 Hz, respectively. Two-dimensional
spectra were recorded with the same probe. For the HMQC
The stereochemical uncertainty about a possible boat
conformation of ring A, which would also be in line with
Copyright  2003 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2003; 41: 115–122

NMR of pentacyclic triterpenes from Boswellia serrata 117
Table 1. 1H and 13C NMR assignments (υ in ppm, J in Hz) for 3˛-acetyl-ˇ-boswellic
acid (1) and 3˛-acetyl-˛-boswellic acid (2)a
1
2
υC
υH
υC
υH
10
20
1
170.34
21.31
34.59
170.37
21.31
34.39
2.10 (s)
2.06 (s)
¾1.48 (m, ˇ)
¾1.19 (m, ˛)
2.15 (bt, J D 15.3, ˛)
1.64 (bd, J D 15.4, ˇ)
5.30 (bs, ˇ)
¾1.16 (t, ˛)
¾1.43 (d, ˇ)
2
23.63
23.63
2.13 (bt, J D 14.4, ˛)
1.60 (d, J D 14.4, ˇ)
5.26 (bs, ˇ)
3
4
5
6
73.42
46.73
50.67
19.64
73.62
46.74
50.65
16.69
—
¾1.39 (d, J D 10.9, ˛)
1.81 (bt, J D 12.2, ˛)
1.71 (bd, J D 12.7, ˇ)
¾1.53 (m, ˛)
¾1.40 (bd, ˇ)
—
¾1.39b (d, ˛)
¾1.82 (m, ˛)
¾1.70 (m, ˇ)
7
33.10
32.82
1.49 (dt, J D 13.9, 3.2, ˛)
¾1.37 (d, ˇ)
8
9
10
11
12
13
14
15
40.07
46.85
37.43
23.44
124.59
139.56
42.30
26.57
39.85
46.81
37.46
23.57
121.89
¾1.57 (ddc, ˛)
¾1.63 (dd, ˛)
—
¾1.69 (m, ˛, ˇ)
5.15 (t, J D 3.5)
—
¾1.87 (m)
5.18 (t, J D 3.5)
¾139
—
41.94
26.13
¾1.83 (m, ˛)
¾1.00 (m, ˇ)
2.01(td, J D 13.3, 4.3, ˛)
¾0.85 (d, J D 13.3, ˇ)
—
¾1.77 (m, ˛)
¾0.98 (m, ˇ)
¾1.98 (˛)
16
28.16
27.04
¾0.80 (ˇ)
17
18
19
33.83
59.22
39.63
32.55
47.39
46.90
¾1.28 (˛)
1.93 (dd, J D 13.4, 4.0, ˛)
¾1.63 (m, ˛)
¾0.87 (m, ˛)
¾1.0 (m, ˇ)
20
21
39.78
31.29
¾0.9 (m, ˛)
¾1.37 (m)
¾1.23 (m)
¾1.40 (t, ˛)
¾1.25 (d, ˇ)
1.24 (s)
31.09
34.78
1.31 (m, ˛)
¾1.09 (m, ˇ)
¾1.42 (m, ˛)
¾1.20 (m, ˇ)
1.21 (s)
22
41.55
37.164
23
24
25
26
27
28
29
30
23.65
182.3
13.36
16.90
23.28
28.79
17.47
21.37
23.78
182.2
13.23
16.80
25.95
28.44
23.68
33.33
0.92 (s)
1.05 (s)
1.12 (s)
0.81 (s)
¾0.79 (d)
0.92 (d, J D 5.8)
0.87 (s)
0.98 (s)
1.17 (s)
0.82 (s)
0.85 (s)
0.85 (s)
a υ13C signals were assigned by C–H correlation and by HMBC. At position 11, ˛- and
ˇ-hydrogens completely overlapped and at position 21 the axial and equatorial hydrogen
orientation could not be identified by the coupling pattern.
b Approximate position according to 1J(C,H) correlation (d, ˛).
c Multiplicity verified with the methyl ester spectrum and Eu(fod)3 shift reagent
application.
the observed results, can be excluded after calculation of
the respective heat of formation (ꢀ226.1 kcal molꢀ1) (AM1)
(1 kcal D 4.184 kJ). This conformation would rapidly change
to the corresponding chair with both H-3 and C-23 in an axial
position (heat of formation: ꢀ235.3 kcal molꢀ1). The methyl
groups 25, 26, 27 and 28 of 3˛-acetyl-ˇ-boswellic acid (1)
each show, in the COSY45 experiment, J coupling to their
respective axial hydrogen neighbors H-1, H-9, H-7, H-15 and
H-16 (Figs 2 and 3). HMBC allowed the final assignment of
all ring connections (Fig. 4). The conformation of rings A,
4
Copyright  2003 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2003; 41: 115–122

118
K. Belsner et al.
Table 2. 1H and 13C NMR assignments (υ in ppm, J in Hz) for
3˛-acetyl-9,11-dehydro-ˇ-boswellic acid (3) and
3˛-acetyl-11-keto-ˇ-boswellic acid (5) (the uncertain orientation at position 21
is similar to the situation with 1 and 2)
3
5
υC
υH
υC
υH
10
20
1
170.26
21.23
33.30
170.23
21.32
34.66
2.03
¾1.76 (bd, ˇ)
1.54 (bt, ˛)
¾2.19 (bt, ˛)
¾1.72 (bd, ˇ)
5.27 (bs, ˇ)
2.06 (s)
2.53 (bd, J D 13.2, ˇ)
¾1.19 (d, t, ˛)
2.22 (bt, J D 14, ˛)
1.58 (m, ˇ)
2
24.38
23.58
3
4
5
6
116.42
47.02
57.40
19.59
73.25
23.89
50.48
18.79
5.28 (bs, ˇ)
¾1.53 (J ¾ 13, ˛)
¾1.89 (m, ˛)
¾1.80 (m, ˇ)
¾1.68 (m, ˛)
¾1.38 (m, ß)
1.37 (bd, J D 12.1, ˛)
¾1.86 (m, ˛)
1.69 (d, J D 12.4, ˇ)
1.65 (m, ˛)
7
31.93
32.89
¾1.44 (d, ˇ)
8
9
10
11
12
13
14
15
43.40
152.6a
39.08
123.1
141.5
141.5
40.58
26.23
45.09
60.38
37.41
199.25
130.55
164.85
43.79
2.39 (s, ˛)
5.44 (d, J D 5.9)
5.62 (d, J D 5.9)
5.53 (s)
¾1.91 (m, ˛)
¾1.05 (m, ˇ)
¾2.00 (m, ˛)
¾0.92 (m, ˇ)
27.26
¾1.89 (m, ˛)
¾1.20 (d, J D 15.3 ˇ)
¾2.16 (m, ˛)
16
28.3
27.57
1.0 (bd, ˇ)
17
18
19
20
21
33.71
57.40
39.08
41.38
31.23
33.99
59.07
39.31b
39.34
30.94
¾1.47 (d, J ¾ 10.8)
¾1.30 (m, ˛)
0.9 (t, ˛)
¾1.52 (d, J D 11.3, ˛)
¾1.40 (m, ˛)
¾0.93 (m, ˛)
¾1.44 (m)
¾1.39 (m)
¾1.27 (m)
¾1.28 (m)
22
43.41
¾1.46 (m, ˇ)
¾1.31 (m, ˛)
1.27
40.95
¾1.47 (m, ˇ)
¾1.32 (m, ˛)
1.21 (s)
23
24
25
26
27
28
29
30
23.81
177.33a
23.42
21.80
17.42
28.72
17.42
21.51
23.89
181.7
13.27
18.42
20.55
28.86
17.42
21.13
1.10
1.17
0.92
0.84
0.79 (d, J D 6.5)
0.9 (bs)
1.12 (s)
1.17 (s)
1.32 (s)
0.81 (s)
0.78 (d, J D 6.5)
0.93c
a Measured in DMSO-d6, weak signal in CDCl3.
b The close proximity of the 13C signals 19 and 20 slightly obscures their exact
allocation.
c Signal shape not visible owing to complete overlapping with H-20.
B, C and D with their attached methyl groups was proved
by ROESY. Figure 3 summarizes the important COSY and
ROESY effects. The orientation of H-18 and H-19, however,
remained problematic, because H-18 overlaps the signal
from H-20, which itself couples to H-19 and thus delivers a
high-order spectrum. Preliminary experiments with stepwise
addition of pyridine-d5 to the sample did not enhance the
resolution at this site. The pyridinium salts formed increased
the molecular charge and thus led to stronger shielding (data
not shown). To overcome this obstacle, experiments with the
methyl ester of 3˛-acetyl-ˇ-boswellic acid (10) and Eu(fod)3
were performed. Upon addition, Eu(fod)3 separates H-18
Copyright  2003 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2003; 41: 115–122

NMR of pentacyclic triterpenes from Boswellia serrata 119
Figure 2. COSY45 of 3˛-acetyl-ˇ-boswellic acid (1) in CDCl3, without shift reagent. The range with the olefinic hydrogen and H-3ˇ
is excluded. Ambiguities arise from the crowded area near H-21 and H-22 and by overlapping of H-20 and H-18. Typical patterns
are the cross peaks formed between H-1˛, H-6˛, H-7˛, and H-15˛ (4J-coupling).
and H-20. The large coupling constant of H-18/H-19 seen in
CH-COSY45 is in accordance with an axial orientation with
respect to ring E. However, the orientation of H-20 cannot
be deduced even at 1 : 1.44 molar ratio [metal ligand/spin
shift reagent, L/S], because the attached methyl-30 exactly
overlaps H-20 at all concentrations of Eu(fod)3, preserving a
high-order spectrum.
the Eu(fod)3 –triterpene complex has two potential chelating
centers, which exhibit most likely different affinities for the
lanthanide reagent. C-24 is much less shifted than C-3 and
C-20 (Fig. 5). Thus, the complex seems attached to the C-3
oxygen. An additional indicator for a Eu(fod)3 docking site
below or in the A–B–C–D plane is the evolution of the
H-11˛/ˇ signals. Starting as a non-resolved multiplet, the
signal splits into an axial and an equatorial group. The latter
experiences a stronger LIS than the axial hydrogen. Distances
calculated from AM1-computed models between all possible
positions at the periphery of ring A and the equatorial H-11
A strong ROE between H-18 and methyl-27 (without
shift reagent) may be the result of cross-relaxation mediated
1
1
˚
via H-19 (H-19—H-27 ¾2.17 A). Together with the H– H-
COSY45 4J long-range interaction from 16-˛ to methyl-28,
this confirms the cis orientation of rings D–E. As expected,
the shift of protons and carbons near to the putative complex
binding site at the carbonyl oxygen of the ester groups
exhibit the most pronounced shift. Figure 5(a) demonstrates
the Eu(fod)3-induced shift for the 13C of ring A.
˚
are ¾0.5–0.7 A lower than to the axial hydrogen.
The presence of Eu(fod)3 considerably simplified the
interpretation of the spectrum and allowed the determination
of several coupling constants, which otherwise cannot be
measured in the 1D-1H spectrum.
The high-frequency shifts are positive for all car-
bons of ring A, indicating preferentially pseudo contact
interaction. Considering the curve shape of the 13C shift,
Spectral comparison of 3˛-acetyl-˛-boswellic acid (2)
with the ˇ-acid revealed that the changes were restricted
to ring E. The 13C signals are shifted to higher frequencies
Copyright  2003 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2003; 41: 115–122

120
K. Belsner et al.
30
30
29
29
H
COSY
H
H
20
28
21
22
H
H
H
19
18
21
19
28
H
18
12
26
12
26
11
9
22
11
9
H
13
17
H
H
13 H
17
25
25
H
14
14
1
1
16
16
2
2
H
8
8
H
10
15
10
15
O
O
H
H
27
27
7
7
H
4
H
5
4
5
3
3
1'
1'
O
6
O
H
6
H
2'
2'
23
23
24
24
HO
HO
O
O
H
H
H
H
H
ROESY
H
H
H
H
H
H
H
H
H
H
H
H
O
H
O
H
H
H
H
H
1'
H
1'
H
O
O
2'
2'
H
H
H
HO
HO
O
O
Figure 3. Illustration of the 2D-NMR (COSY45, HMBC, ROESY) interactions of 3˛-acetyl-ˇ-boswellic acid (1) (left), and
3˛-acetyl-˛-boswellic acid (2) (right). The dotted line between methyl-28 and H-20 in the ROESY of 3˛-acetyl-ˇ-boswellic acid (1)
marks an interaction that suffers from overlap with the H-19 signal.
posed the greatest problem, but was finally achieved after 3J
coupling between H-18 and C-14 was identified in HMBC.
However, swapping between positions 8 and 14 cannot be
completely ruled out for this compound.
After molecular modeling and minimization, the NMR-
derived structures were nearly identical with the published
crystal structures. Comparison of the ring atom positions of
the two published x-ray structures left a slight difference
for C-19, C-21, C-29 and C-30, whereas C-18, C-20 and C-22
behave oppositely. In the H spectrum, the most important
1
difference is the disappearance of the splitting of H-30 and the
collapse of the methyl groups 29 and 30 to one group, which
introduces some ambiguity into the interpretation. However,
this loss is compensated for by enhanced separation of
other protons in ring E. H-18 moves from υˇ ¾1.28 to υ˛
1.93 ppm, and the respective double doublet J D 13.4, 4.0 Hz
underlines the orientation of H-18. In contrast to the weakly
defined chair conformation of ring E in ˇ-boswellic acids,
the multiplet of H-21 υˇ ¾1.37 ! υ˛ 1.31 here resembles a
double triplet, and 4J coupling observed in COSY45 from
H-30 to H-19 and H-21, respectively, gives the impression of
a relatively stable chair conformation of ring E.
In the NMR spectra of 3˛-acetyl-11-keto-ˇ-boswellic acid
(5), the 11-keto group isolates H-9 and induces a high-
frequency shift of C-9 to 60.38 ppm and H-9 to 2.39 ppm. The
strong ROEs with H-1˛, H-5˛, H-7˛, H-12 and H-27 prove
the configuration as a ˇ-boswellic acid. The addition of a keto
group to the ursane skeleton also reduces shielding at ring E.
However, this is mainly restricted to H-18 (d, J D 11.3 Hz).
The H-18 doublet is again an indicator for the axial position
of H-19, but the exact assignment of C-29 and C-30 remained
impossible, as already stated for 1.
˚
of 0.08 A r.m.s. The selected conformers converged during
minimization to highly similar final structures. Their ring
˚
carbons perfectly fitted (r.m.s. of 0.12 and 0.1 A) to the
corresponding carbons of the two x-ray conformers. The
differences between the NMR and x-ray structures are minor
and are restricted to the ester group in the 3˛-position. The
result of the fitting experiments for the nine conformers with
one crystal structure (blue) is shown in Fig. 6.
CONCLUSIONS
Taken together, high-field NMR delivers the same minimized
structure of 3˛-acetyl-ˇ-boswellic acid (1) as already obtained
with x-ray crystallography.14 The corresponding ˛-boswellic
acids have the same backbone and are therefore highly
similar with the exception of C-20. The results are consistent
with the known stereochemistry of the boswellic acids as
In NMR analysis of 3˛-acetyl-9,11-dehydro-ˇ-boswellic
acid (3), the assignment of C-8 and C-14 of boswellic acid (3)
Copyright  2003 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2003; 41: 115–122

NMR of pentacyclic triterpenes from Boswellia serrata 121
Figure 4. HMBC optimized at 7 Hz of 3˛-acetyl-ˇ-boswellic acid (1). Solvent, CDCl3; without addition of shift reagent. The attached
proton test (APT) is plotted on the y-axis. Typical for boswellic acids is the square formed by the cross peaks C-8/H-27, C-14/H-27,
C-14/H-26 and C-8/H-26 and the intense cross peaks of H-9.
∆δ 13C of Ring A Atoms
∆δ 13C of Ring A Substituent Atoms
8
6
4
2
0
8
6
4
2
0
C''
C1
C2
C3
C4
C5
C10
C1'
C2'
C24
C23
0.0
0.5
1.0
1.5
0
1
molar ratio
(a)
molar ratio
(b)
Figure 5. High-frequency shift for the 13C resonance frequencies induced by Eu(fod)3. (a) Carbon atoms of ring A; (b) carbon atoms
of the substituent of ring A; the strong effect on C-10 and C-20 indicates involvement of the acetoxy group into the complex formation
with Eu(fod)3.
derived from the absolute stereochemical data of the amyrin
skeleton.23–25
angle of ring A to E. Thus, the application of a shift reagent
established the orientation of H-18 and also confirmed some
of the more difficult interpretations of the other proton
signals, but it was not able to prove the conformation at
C-20. Typical signal patterns of boswellic acids have been
identified. HMBC reveals a rectangle formed by the cross
peaks of C-8/H-27, C-14/H-27, C-14/H-26 and C-8/H-26.
The acetyl ester oxygen seems to be the primary
attachment site for a lanthanide shift reagent. Eu(fod)3
enhances the NMR spectrum’s resolution of rings A, B, C
and D considerably. However, the resolution of ring E is not
satisfactory, which is most likely caused by the decreasing
Copyright  2003 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2003; 41: 115–122

122
K. Belsner et al.
2. Sander O, Herborn G, Rau R. Z. Rheumatol. 1998; 57: 11.
3. Gerhardt H, Seifert F, Buvari P, Vogelsang H, Repges R.
Z. Gastroenterology 2001; 39: 11.
4. Hussein G, Miyashiro H, Nakamura N, Hattori M, Kakiuchi N,
Shimotohno K. Phytother. Res. 2000; 14: 510.
5. Heldt MR, Syrovets T, Winking M, Sailer ER, Safayhi H,
Ammon HPT, Simmet Th. Naunyn-Schmiedeberg’s Arch.
Pharmacol. 1997; 355(Suppl.): 30.
6. Hoernlein RF, Orlikowsky Th, Zehrer C, Niethammer D,
Sailer ER, Simmet Th, Dannecker GE, Ammon HPT. J. Pharmacol.
Exp. Ther. 1999; 288: 613.
7. Bu¨chele B, Syrovets T, Zugmaier W, Belsner K, Simmet Th.
Naunyn-Schmiedeberg’s Arch. Pharmacol. 2000; 362(Suppl.): 101.
8. Syrovets T, Schu¨le A, Banuta C, Heinz S, Simmet Th. Naunyn-
Schmiedeberg’s Arch. Pharmacol. 2001; 363(Suppl.): 22.
9. Belsner K, Bu¨chele B, Syrovets T, Simmet Th. Naunyn-
Schmiedeberg’s Arch. Pharmacol. 2001; 363(Suppl.): 44.
10. Syrovets T, Bu¨chele B, Gedig E, Slupsky JR, Simmet Th. Mol.
Pharmacol. 2000; 58: 71.
Figure 6. Nine conformers with low potential energy were
randomly selected from a molecular dynamics simulation at
11. Winking M, Sarikaya S, Rahmanian A, Jodicke A, Bo¨ker DK. J.
Neurooncol. 2000; 46: 97.
12. Winterstein A, Stein G. Hoppe-Seyler’s Z. Physiol. Chem. 1932;
208: 9.
13. Allan GG. Phytochemistry 1968; 7: 963.
14. Rajnikant V, Gupta VK, Rangari VD, Bapat SR, Agarwal RB,
Gupta R. Cryst. Res. Technol. 2001; 36: 93.
15. Pardhy RS, Bhattacharyya SC. Indian J. Chem. 1978; 16B: 176.
16. Hairfield EM, Hairfield HH. Jr. J. Chromatogr. Sci. 1989; 27: 127.
17. Schweizer S, Eichele K, Ammon HPT, Safayhi H. Planta Med.
2000; 66: 781.
°
˚
450 C and optimized with Mopac/AM1 to 0.1 A precision. The
picture is the result of an overlay of the conformers ring atoms
to the corresponding x-ray structure (blue). The latter was used
without optimization. Differences are restricted to ring A
substituents, mainly the ester group, which disappear after
minimization of the crystal structure. For clarity, hydrogen
atoms have been excluded.
In contrast to the x-ray structure, ring E of the ˇ-boswellic
acids seems to be rather mobile as judged from the difficulty
in identifying axial and equatorial positions at ring E. In the
context of medicinal chemistry, the rigid backbone should
facilitate molecular docking experiments.
18. Buchele B, Zugmaier W, Simmet T. J. Chromatogr. B submitted.
¨
19. Schweizer S, vonBrocke AFW, Boden SE, Bayer E, Ammon HPT,
Safayhi H. J. Nat. Prod. 2000; 63: 1058.
20. Stewart JJP. J. Comput.-Aided Mol. Des. 1990; 4: 1.
21. Dewar MJS, Zoebisch EG, Healy EF, Stewart JJP. J. Am. Chem.
Soc. 1985; 107: 3902.
22. Guex N, Peitsch MC. Electrophoresis 1997; 18: 2714.
23. Ruzicka L, Wirz W. Helv. Chim. Acta. 1939; 22: 948.
24. Corey EJ, Cantrall EW. J. Am. Chem. Soc. 1959; 78: 1745.
25. Corey EJ, Ursprung JJ. J. Am. Chem. Soc. 1956; 78: 183.
REFERENCES
1. Ammon HPT, Mack T, Singh G, Safayhi H. Planta Med. 1991; 57:
203.
Copyright  2003 John Wiley & Sons, Ltd.
Magn. Reson. Chem. 2003; 41: 115–122