Nudicaulins, yellow flower pigments of papaver nudicaule: Revised constitution and assignment of absolute configuration

Article abstract of DOI:10.1021/ol303211w

The constitution of the aglycon of nudicaulin has been revised to be a pentacyclic indole alkaloid. The relative and absolute configurations of the two diastereomers, nudicaulins I (3a) and II (3b), have been assigned by NMR, conformational analyses, and interpretation of the experimental ECD spectra by quantum-chemical calculations.

Full text of DOI:10.1021/ol303211w

ORGANIC
LETTERS
XXXX
Vol. XX, No. XX
000–000
Nudicaulins, Yellow Flower Pigments
of Papaver nudicaule: Revised
Constitution and Assignment
of Absolute Configuration
†
‡
†
§
€
Evangelos C. Tatsis, Anu Schaumloffel, Anne C. Warskulat, Georges Massiot,
Bernd Schneider,*,† and Gerhard Bringmann*,‡
€
Max Planck Institute for Chemical Ecology, Hans-Knoll-Strasse 8, Beutenberg Campus,
07745 Jena, Germany, University of Wu€rzburg, Institute of Organic Chemistry,
Am Hubland, 97074 Wu€rzburg, Germany, and USR CNRS-Pierre Fabre No. 3388
ETaC, Center of Research and Development of Pierre Fabre, 3 avenue Hubert Curien,
31035 Toulouse Cedex 01, France
schneider@ice.mpg.de; bringman@chemie.uni-wuerzburg.de
Received November 21, 2012
ABSTRACT
The constitution of the aglycon of nudicaulin has been revised to be a pentacyclic indole alkaloid. The relative and absolute configurations of the
two diastereomers, nudicaulins I (3a) and II (3b), have been assigned by NMR, conformational analyses, and interpretation of the experimental ECD
spectra by quantum-chemical calculations.
In 1939, Price et al.1 isolated the pigment responsible for
the yellow color of Papaver nudicaule flower petals and
named it nudicaulin. The first investigations of the nudi-
caulin structure by elemental analysis, oxidative degrada-
tion, and acid hydrolysis suggested the presence of nitrogen
and a 4-hydroxyphenyl moiety in a diglucosidic compound.
In a later work, a triglucoside was claimed from enzymatic
degradation and the spectral properties indicated that nu-
dicaulin was neither a flavonoid nor a carotenoid nor a
betaxanthin.2 Based on NMR spectroscopy and mass
spectrometric analyses, Schliemann et al.3 proposed the
structure 1, with a pentacyclic indole alkaloid skeleton
(Figure 1) for the aglycon of the diastereomeric nudicaulins.
However, the N-containing six-membered ring e in this
postulated structure appeared to be severely distorted and
certainly highly strained. Moreover, the interruption of
the conjugated system is not consistent with the UV
absorption maximum at ca. 460 nm. Based on further
NMR experiments and chemical modifications, we de-
scribe here the revision of the previously postulated
constitution of the aglycon of these compounds as 2,
using nudicaulin I (3a) as a representative example
(Figure 1). The AX system of H-6 and H-8 (ring a), a
four-spin system of H-15ꢀH-18 (d), an AA0XX0 system of
H-20/60 and H-30/50 (f), and a singlet of H-3 were attrib-
uted to the aglycon. HSQC and HMBC correlations
(Figures S5ꢀS7) assigned all carbon atoms within the
† Max Planck Institute for Chemical Ecology.
‡ University of W€urzburg.
§ USR CNRS-Pierre Fabre No. 3388 ETaC.
(1) Price, J. R.; Robinson, R.; Scott-Moncrieff, R. J. Chem. Soc.
1939, 1465–1468.
(2) Harborne, J. B. Phytochemistry 1965, 4, 647–657.
(3) Schliemann, W.; Schneider, B.; Wray, V.; Schmidt, J.; Nimtz, M.;
€
Porzel, A.; Bohm, H. Phytochemistry 2006, 67, 191–201.
r
10.1021/ol303211w
XXXX American Chemical Society

Figure 1. Constitutions of the aglycon core as previously published (1)3 and currently revised (2), and structures of the diastereomeric
nudicaulins I (3a) and II (3b). Red arrows indicate 1H,15N HMBC correlations.
substructures of rings a, d, and f, which were not subject
to revision, and established H-3/C-3 (δH 5.55/δC 48.9) as
a key position connecting these parts of the aglycon.
HMBC correlations of H-3 with C-4 (δ 101.7), C-5
(δ 156.0), and C-9 (δ 160.4) linked C-3 via a CꢀC bond
with C-4 to ring a. A correlation of H-3 with C-2 (δ 177.1)
and C-13 (δ 130.8) connected C-3 to the N-containing
ring c of the indole unit and also substantiated the
position of the nitrogen next to C-2, as also confirmed
by its chemical shift at a low field. Two additional HMBC
cross signals of H-3, one of them with C-12 (δ 165.0),
which is also connected through an HMBC correlation
with H-20/60 of ring f, and the other one with the acetal
carbon C-11 (δ 126.4), completed the assignment of the
carbon atoms of rings b and e. The position of the nitrogen
atom (δ 172.7) was established by 1H,15N-HMBC correla-
tions (Figure S14) of H-3 and H-18 with the indole
cross peaks with C-13 (δ 116.5), C-11 (δ 126.8), C-10
(δ 130.4), C-20/60 (δ 132.4), and C-2 (δ 140.5) (Figure S30).
1
1
In the Hꢀ H correlation spectrum optimized for long-
1
range couplings (1Hꢀ H lrCOSY) (Figure S28), the H-12
signal correlated with H-3 (δ 5.07) and H-20/60 (δ 7.10).
This showed that the quaternary C-12 of 3a had been
converted to a methine in dihydronudicaulin I (4a). H-3
still remained the key signal, now showing HMBC cross
peaks with C-2, C-4 (δ 108.3), C-5 (δ 155.1), C-9 (δ 161.2),
C-11, and C-13. The chemical shift of the acetal carbon
C-11 in 4a (δ 126.8) was the same as that in nudicaulin
I (3a) (δ 126.7). For complete 1H and 13C NMR data, see
Table S3. In summary, analysis of the dihydronudicaulins
(4a/b) further confirmed the constitution of the nudicaulin
aglycon (2) itself, especially the connection between the
indole unit and the rest of the molecule.
In addition, the nudicaulins (3a/b) were converted to
their O- and N-permethyl derivatives (5a/b) by applying a
1
nitrogen. The H,15N-HMBC correlations in combina-
1
Williamson ether synthesis (Figure 2b).4 By H NMR,
tion with a strong signal in the ROESY spectrum (Figure S13)
between H-15 and H-20/60 indicated a [20,30:1]-annelation
of the indole moiety to the central cyclopentene ring e.
The attachment of the carbohydrate units to the aglycon
through oxygen atoms and the interunit link between
the two glucose units A and B was established by means
of HMBC correlations of H-100 of Glc A with C-11,
H-10000 of Glc C with C-7, and H-1000 of Glc B with C-200
of Glc A. The number and attachment positions of the
sugar and acyl units of nudicaulins IꢀVIII3 remained
unaffected by the suggested structural revision of the
aglycon.
To further confirm the constitution of the aglycon 2, the
nudicaulins were structurally modified. By hydrogenation,
additional protons were introduced into the proton-defi-
cient region of the aglycon, especially in ring e (Figure 2a).
The successful reaction was confirmed by HRMS (found
for 4a: 874.27482; calcd: 874.27642). The only UV max-
imum at 255 nm suggested an interuption of the conju-
gated double-bond system. Thus the 1H NMR spectrum of
dihydronudicaulin I (4a) displayed aglycon signals shifted
to a higher field as compared to the parent compound 3a,
consistent with a reduced number of conjugated double
bonds. In addition, a new singlet at δ 4.99 appeared in the
1H NMR spectrum, which correlated with δC 55.0 (C-12)
in the HSQC spectrum (Figure S29) and showed HMBC
HSQC (Figure S38), and HMBC spectra (Figure S39) the
new O- and N-methyl signals in the aglycon part of the
permethylated nudicaulin were assigned to the positions
that correspond to hydroxy and amino groups in the un-
derivatized structures. In the NMR spectra of permethy-
lated nudicaulin I (5a), three methyl signals were observed
that were categorized as O-CH3 (δH 4.01/δC 56.1; δH 4.05/
δC 56.6) and N-CH3 (δH 4.35/δC 35.7) according to their
chemical-shift values. HMBC correlations assigned the
O-CH3 groups to C-40 (δ 168.2) in ring f and C-5 (δ 157.9)
in ring a, while the N-CH3 group, as expected, correlated
with C-2 (δ 177.6) and C-19 (δ 149.6). Hence, permethyla-
tion evidenced the position of the free hydroxy groups and
of the nitrogen atom, thus, finally revising the previously
postulated3 constitution of the nudicaulin aglycon as
12-(4-hydroxyphenyl)-3,11-dihydrobenzo-furo[20,30:1]-
cyclopenta[1,2-b]indole-5,7,11-triol (2).
The diastereomeric nudicaulins I (3a) and II (3b) contain
two stereogenic centers in the aglycon. This made an in-
vestigation of their relative and absolute configurations
necessary. A strong cross signal in the ROESY spectrum
(Figure S13) between H-3 and H-100 of Glc A of nudicaulin
I (3a) showed that H-3 and the Glc A were oriented to the
(4) Ciucanu, I.; Kerek, F. Carbohydr. Res. 1984, 131, 209–217.
Org. Lett., Vol. XX, No. XX, XXXX
B

For the elucidation of the absolute configuration of the
nudicaulins 3a/b, electronic circular-dichroism (ECD) spec-
troscopy in combination with quantum-chemical ECD
calculations5 was the method of choice. A first analysis of
the experimental UV absorptions of 3a and 3b measured
in two different solvent systems revealed a strong solvent
dependence of the first UV band. In the MeCNꢀH2O
mixture, this absorption appeared with high intensity
around 460 nm, possessing a broad blue-shifted shoulder,
while two signals of low intensity around 460 and 385 nm
were observed in MeOH (Figure S43).
First excited-state calculations for the cis-configured
nudicaulins I and II were performed with time-dependent
(TD) CAM-B3LYP, including CPCM for solvent effects
of MeOH. However, the calculated spectra did not show
sufficiently good agreement with the experimental ones
(Figure S46). A reason for this could be interactions be-
tween the solvent and the glucose moieties that might
influence the chiroptical properties.
Figure 2. Modification of nudicaulins I and II (3a/b) and ‘new’
HMBC interactions (in red). (a) Catalytic hydrogenation.
(b) Permethylation. (c) Enzymatic hydrolysis by a pectinase.
Therefore, an arbitrarily chosen number of water mole-
cules (12) were added to the structure of the two cis-
diastereomers, (3S,11R)-3 and (3R,11S)-3, using the
CAST program.6 The conformations thus identified were
further optimized with B97-D/SVP and then submitted to
semiempirical ZINDO/S-CI UV calculations. The results
were compared with the UV curves calculated for identical
conformations, without water molecules. The ones includ-
ing water possessed a red-shifted, intensified first absorp-
tion band, thus corroborating the substantial effect of the
solvent (Figure S45, Table S5). These findings showed that
within TDDFT the CPCM approach alone would possibly
not be sufficient to describe the excited-state properties of
the authentic nudicaulins I and II properly, even though
the effects seemed smaller in the experiment when using
methanol. In addition to the ZINDO/S-CI tests, the
analysis of the excited states of the TDDFT computations
showed that within the aglycon core electronic transitions
with charge-transfer (CT) character occurred within the
same wavelength region that was affected by the solvent.
Since more advanced methods to correctly account for
these properties would have led to an immense increase in
computational costs, the compound had to be simplified
synthetically, at least to minimize the solvent effects.
Attempts to remove all three glucose units were unsuc-
cessful, since, apparently, Glc A is needed to prevent de-
composition of the chemically unstable aglycon core.
Therefore, nudicaulins I and II could only be partially
deglucosylated to give their monoglucosidic forms 6a/b
(Figure 2c), using a commercial pectinase. Compared to
the nudicaulins I and II, the NMR spectra of the mono-
glucosides 6a and 6b (Figures S22, S23, S25) were con-
siderably simplified in the carbohydrate regions. The
ROESY spectra of the two diastereomeric monoglucosides
(Figures S24, S26) confirmed the results that had been
obtained for nudicaulins I (3a) and II (3b), indicating an
unchanged aglycon structure after hydrolysis.
same side of the aglycon skeleton, i.e., cis to each other
(Figure 1). Interestingly, nudicaulin II (3b) (Figure S20)
showed such a ROESY cross signal, too, which was also in
agreement with a relative cis-configuration, but it was far
less intense. Relative to the integral intensities of the ROE
signals between H-20/60 and H-30/50, which were used as a
reference (100%), the integral intensity of the H-3 ꢀ H-100
cross signal of nudicaulin I (3a) was 32% and that of
nudicaulin II (3b) was 19%. A conformational search with
B97D/SVP resultedin a minimumstructurewithadistance
00
˚
d(H-3,H-1 ) of 2.4 A for the (3S,11R)-3 diastereomer (cis),
which was consistent with the strong ROE of nudicaulin I
00
˚
(3a), while the longer distance d(H-3,H-1 ), 3.8 A, for the
lowest-energy conformer of the (3R,11S)-3 diastereomer
(also cis) was in accordance with the weaker ROE of
nudicaulin II (3b). For (3R,11R)-3 a0s0 a likewise imaginable
˚
trans-isomer, a distance d(H-3,H-1 ) of about 4.5 A was
predicted, from which no ROE signal would be expected,
because the aglycon core was in between and, therefore,
could be excluded (Figure S42).
In the ROESY spectra of dihydronudicaulins I (4a) and
II (4b) (Figures S31 and S36), ROE signals were observed
between H-3 and H-100 of glucose A, between H-12 and
H-100 of glucose A, between H-12 and H-20/60, and between
H-3 and H-20/60. No ROE signal was observed between
H-3 and H-12 suggesting an opposite orientation with
respect to the plane formed by the indole moiety and ring e
(corresponds to trans-orientation of Glc A and H-12 and
trans-orientation of the p-hydroxyphenyl ring and the ben-
zofuran moiety). According to these data, the hydrogenation
had occurred with a high degree of stereoselectivity.
(5) For selected reviews, see: (a) Warnke, I.; Furche, F. WIREs
Comp. Mol. Sc. 2012, 2, 150–166. (b) Pescitelli, G.; Di Bari, L.; Berova,
N. Chem. Soc. Rev. 2011, 40, 4603–4625. (c) Autschbach, J. Chirality
2009, 21, E116–E152. (d) Bringmann, G.; Bruhn, T.; Maksimenka, K.;
Hemberger, Y. Eur. J. Org. Chem. 2009, 2717–2727.
(6) Grebner, C.; Becker, J.; Stepanenko, S.; Engels, B. J. Comput.
Chem. 2011, 32, 2245–2253.
Org. Lett., Vol. XX, No. XX, XXXX
C

The offline CD spectra of 6a and 6b were also recorded
in methanol, yielding nearly mirror-imaged curves, in which
the first Cotton effect now occurred around 385 nm, instead
of 460 nm in compounds 3a/3b (Figure S44), showing that
a comparison of the spectra calculated for a simplified
structure with the experimental ones of the authentic nudi-
caulins would definitely not have been possible in this case.
Still, the agreement of the spectra of the monoglucosides
obtained with CAM-B3LYP was not truly improved as
compared to the experimental ones (Figure S47). Using the
double-hybrid functional B2GP-PLYP (with the COSMO
solvent field for methanol), the energies and rotatory
strengths of the first excitations were still not quite well
predicted in the case of the (3R,11S)-configured mono-
glucoside 6 (Figure S48). By contrast, B2GP-PLYP com-
putations of the hypotheticꢀsince chemically unstableꢀ
aglycon 2 provided a much better match with the experi-
mental CD data of the monoglucosides 6a/b (Figure S49).
To understand this discrepancy, coupled-cluster calcu-
lations with RICC2/def2-SV(P) were carried out for the
first 25 excitations of the (3R,11R)-configured aglycon 27
and, exemplarily, for the first five excited states of the
(3R,11S)-configured monoglucoside 6. The main objective
was to determine the correct energetic positions of the CT
transitions and to validate the impact of the remaining glucose
moiety within the molecule upon excitation. Comparison of
the RICC2 excitations, however, did not show any substantial
differences between 6 and 2. The B2GP-PLYP spectra ob-
tained for the aglycon also seemed comparable with RICC2,
although the first excited state had a clearly different main
configuration (Table S6). Deviating from this, however, was
the presence of the glucose unit in the B2GP-PLYP calcula-
tions leading to stronger changes of the rotatory-strength
values within the first three excited states, of which the first
one showed significant CT character (Figure S50), which
obviously occurred at a wrong position and falsified the sign
and intensity of the overall CD effect between 360 and
390 nm. Thus, already one glucose unit significantly ham-
pered the applicability of TD methods, making the coupled-
cluster approach mandatory for this kind of structure.
To conclude, the comparison of the experimental CD
spectra of the monoglucosides of nudicaulins I (6a) and II
(6b) with the ones computed for the cis-aglycon 2 with
RICC2/def2-SV(P) allowed an unambiguous attribution
of the absolute configuration (Figure 3). The CD curve
calculated for (3S,11S)-configured 2 was in perfect agree-
ment with the experimental spectrum of the monogluco-
side of nudicaulin I (6a). This determined 6a to have
the (3S,11R)-configuration.7 Analogously, the spectra of
(3R,11R)-configured 2 and the monoglucosidic nudicaulin
Figure 3. Absolute configurations of 6a/b by comparison of the
exptl CD curves with calcd curves for the aglycon model 2, using
RICC2/def2-SV(P) (UV shift = 29 nm, bandwidth σ = 0.40 eV).
II (6b) provided a clear match, assigning the latter to
possess the (3R,11S)-configuration.
From these results, the absolute configurations of the
parent compounds nudicaulins I (3a) and II (3b) were
eventually deduced to be (3S,11R) and (3R,11S), respec-
tively. This structural assignment became possible only by
the experimental decrease of the strong solvent effects of
the glucose moieties, which in turn allowed the application
of accurate higher-level quantum-chemical methods to
predict the excited-state properties. The occurrence of dia-
stereomeric pairs of nudicaulins with racemic (or scalemic)
aglycon parts raises a question regarding the biosynthesis,
which remains to be investigated.
Acknowledgment. This work was supported by an FP7
Marie Curie Intra-European Fellowship from E.U. to
E.C.T. (BIOSYN-NUDICAUL project, Contract Number
221274). We are grateful to B. Amslinger (University of
€
Wurzburg) for the CD measurements.
Supporting Information Available. Experimental pro-
1
cedures; tables with H and 13C NMR data; MS and
UVꢀvis data; NMR spectra of compounds 3ꢀ6; compu-
tational details. This material is available free of charge
via the Internet at http://pubs.acs.org.
(7) For the formal reason of the CIP convention, the altered priorities
at C-11 lead to a change of the stereodescriptor.
The authors declare no competing financial interest.
D
Org. Lett., Vol. XX, No. XX, XXXX