Parthenolide and its photochemically synthesized 1(10)Z isomer: Chemical reactivity and structure-activity relationship studies in human leucocyte chemotaxis

Article abstract of DOI:10.1016/S0968-0896(02)00553-9

The present study has achieved the photochemical conversion of a germacrolide into a melampolide. The investigation on their chemical properties allowed us to evaluate the minimum interatomic distance needed for transannular bridging of C₁₀ ring in germacrolides and to explain the regiochemical selectivity of electrophilic cyclizations. The antiinflammatory activity of parthenolide and its semisynthetic derivatives was evaluated by in vitro chemotaxis assay with human neutrophiles. These structure-activity relationship studies have led to hypothesize a new pharmacophore and have provided useful information for computationally designed drugs.

Full text of DOI:10.1016/S0968-0896(02)00553-9

Bioorganic & Medicinal Chemistry 11 (2003) 1503–1510
Parthenolide and Its Photochemically Synthesized 1(10)Z Isomer:
Chemical Reactivity and Structure–Activity Relationship Studies
in Human Leucocyte Chemotaxis
Hannes Neukirch,a Nicole C. Kaneider,b Christian J. Wiedermann,b
Antonio Guerrieroa and Michele D’Ambrosioa,*
a
`
Laboratorio di Chimica Bioorganica, Universita degli Studi di Trento, Via Sommarive 14, 38050 Povo (Trento), Italy
bDivision of General Internal Medicine, Department of Internal Medicine, University of Innsbruck,
Anichstrasse 35, 6020 Innsbruck, Austria
Received 10 July 2002; accepted 23 October 2002
Abstract—The present study has achieved the photochemical conversion of a germacrolide into a melampolide. The investigation on
their chemical properties allowed us to evaluate the minimum interatomic distance needed for transannular bridging of C10 ring in ger-
macrolides and to explain the regiochemical selectivity of electrophilic cyclizations. The antiinflammatory activity of parthenolide and its
semisynthetic derivatives was evaluated by in vitro chemotaxis assay with human neutrophiles. These structure–activity relationship
studies have led to hypothesize a new pharmacophore and have provided useful information for computationally designed drugs.
# 2003 Elsevier Science Ltd. All rights reserved.
Introduction
these four compounds share a similar spatial arrangement
but differ as to the chemical functionalities.
Parthenolide 1 (Fig. 1) has been revealed to be the
major active principle of European feverfew (Tanace-
tum parthenium) on account of numerous and diverse
assays. Despite the wide spectrum of biological properties
A subsequent investigation on the influence of chemical
functionalities and molecular conformations strengthened
that
1 exhibits and the intense pharmacological
research,1 the interrogatives about its mechanism of
action and the relationship between chemical structure
and biological activity are still left unsolved.
The exomethylene lactone functionality was immedi-
ately deemed to have fundamental importance because
of its exceptional reactivity with nucleophilic groups.2
Later, several sesquiterpene lactones (SL), tested as
witchweed germination stimulants, displayed that the
exomethylene lactone moiety did not have an essential
role but the spatial arrangement of the terpenoid skeleton
was more important than the presence of any specific
functional group in the molecule.3 For instance, 1 and
its 11,13-dihydro derivative 2 stimulated witchweed
germination at a comparable percentage and to a lesser
extent than the eudesmanolides santamarin and reynosin;
*Corresponding author. Tel.: +39-0461-881509; fax: +39-0461-
881696; e-mail: dambrosi@science.unitn.it
Figure 1. Chemical transformations of 1.
0968-0896/03/$ - see front matter # 2003 Elsevier Science Ltd. All rights reserved.
PII: S0968-0896(02)00553-9

1504
H. Neukirch et al. / Bioorg. Med. Chem. 11 (2003) 1503–1510
this conclusion.4 However three different skeletons were
selected so that no reasonable generalization emerged
from these biological assays.
We decided to verify the effectiveness of such research
procedure and in this study we have chosen partheno-
lide 1, an SL the pharmacologists are chiefly concerned
with, and the chemotaxis index bioassay.
Extremely interesting studies on the acid catalyzed
rearrangements of 4,5-epoxygermacrolides were even-
tually published (Fig. 2a and b);5 the authors suggested
a new pathway in their mechanism of biological action.
In fact the facile transannular cyclization to cis-guaianes
furnishes a carbocation at C-10 which becomes a second
alkylating site for nucleophilic groups in addition to the
exomethylene lactone functionality.
It is known that injured vascular endothelium shows an
increased release of chemokines, that attract neutro-
phils to sites of injury; interleukin-8 is among these
and is known to be the strongest chemoattractant for
these leukocytes.7 Neutrophils migrate into the vas-
cular intima and lead to further destruction of endo-
thelium by enhanced production and release of
superoxide anions.8 Activation and migration of
neutrophils play an important role in the development
of tissue injury and is a hallmark of every inflamma-
tory process.
This is an appealing hypothesis that has been supported
by results on the capability of many SL at inhibiting the
transcription factor NFkB.6 Despite several discrepancies,
the authors pointed out that the most active SL contain
two electrophilic sites: the exomethylene lactone and an
enone functionality. Parthenolide was considered to be
an exception being a highly active though monofunc-
tional compound. Notably, the most powerful group
comprises compounds of different skeletal types: guaia-
nolides, pseudoguaianolide, germacrolides, melampo-
lides, heliangolides, and 4,5-dihydrogermacranolides.
Chemistry
The 11,13-dihydroparthenolide 2 was easily obtained
through catalytic hydrogenation (Fig. 1) whereas a
derivative suitable for comparison with the (E)/(Z)
stereochemistry was identified in the geometrical isomer
1(10)-(Z)-parthenolide 3. We attempted to carry out the
1(10)-(E) to 1(10)-(Z) photoisomerization: parthenolide
and its 11,13-dihydro analogue 2 were exposed to
254 nm UV light in benzene for 20 h at rt furnishing the
respective cis-geometric isomers in 76% (3) and 100%
(4) yield (Fig. 1). The (Z)-configuration at the
C(1)–C(10) double bond in both 3 and 4 was proven by
the typical downfield shift of C(14) and by NOE
enhancement between H(1) and 3H(14). These represent
the first syntheses of 1(10)-(Z)-parthenolide and its
11,13-dihydro analogue and their full spectroscopic
characterizations. It has been reported that UV irradia-
Our impression is that metabolites have been chosen for
bioassays till now on the basis of their availability and
that a previous protocol ought to be revised.4 In order
to better understand the mode of action we believe that
SL should be selected on the basis of the following cri-
teria: (i) test a compound with an exomethylene lactone
functionality and its 11,13-dihydro analogue; (ii) exam-
ine the most stable molecular conformations and, if
necessary, compare the influence of slight modifications
from the adopted spatial arrangement; (iii) study their
chemical properties and the way they could be modified
by the introduction of new functional groups.
tion of dihydroparthenolide
2 in benzene gave
Figure 2. Reactions of 1, 2 and 3 with pTSA / MeOH.

H. Neukirch et al. / Bioorg. Med. Chem. 11 (2003) 1503–1510
1505
11,13-dihydromicheliolide (Fig. 1),9 so we looked for
the reasons of such a diverse outcome. Since Govin-
dachari et al. purified the product by sublimation at
110 ꢀC, we decided to examine thermal as well as acid-
base chemical properties of 1.
The parthenolide was warmed in toluene at 110 ꢀC
under N2 for 7 h and gave decomposition to unknown
products in only 2%, whereas stirring in MeOH/TMG
(tetramethyl-guanidine) led exclusively to the Michael
type product 5 by addition of methanol to the
11,13-double bond (Fig. 1). The structure follows from
the observed parent peak at m/z=280, the dis-
appearance of the olefinic protons of the exomethylene
moiety and the appearance of a singlet signal at
3.38 ppm, in accordance with a methoxy substituent at
C(13). Parthenolide 1 was dissolved in MeOH/H+ at rt
and resulted in the formation of compounds shown in
Figure 2a. The reaction mixture was purified by repe-
ated HPLC and the products were characterized by MS
and NMR spectroscopy. Compound 6 (a bicyclo[5.3.0]-
decane skeleton) exhibited a parent peak at m/z=280,
Figure 3. Formation rates of 6 and 10 from acid catalyzed reactions of
1 and 2 (see Experimental).
H(3)–H(15) was established by a COSY experiment.
Compound 13 showed two exomethylenic protons at
C(15) instead of the methyl signal of the starting com-
pound. Both 12 and 13 showed parent peaks at
m/z=248. The parent peak at m/z=280 of compound
14 indicated addition of methanol; the connectivity and
the relative configurations were established by COSY
and NOE difference spectroscopy.
1
the H and 13C NMR spectra were compatible with a
guaianolide structure bearing hydroxy and methoxy
groups at quaternary carbons. Compound 7 and 8 were
identified by comparison of their spectral data (1H, 13C
NMR and MS) with reported values.10 The four mem-
bered ring of compound 9 (a bicyclo[6.2.0]decane skele-
ton) was established by COSY maps and COLOC
correlations of H(1) with C(2), C(4), C(5), C(10) and of
3H(15) with C(1), C(3), C(5). The stereochemistry of the
ring fusion was unambiguously established as trans, due
to NOE enhanced signals of: Hb(2) on irradiation at
3H(14), Hb(2) and H(6) on irradiation at 3H(15), H(1)
on irradiation at the methoxy protons.
Pharmacology
The chemotaxis chamber consists of two parts: a lower
part, which contains the chemoattractants and is cov-
ered with a nitrocellulose filter, and an upper part. In
this system, a concentration gradient of the chemo-
attractant, which is put into wells of the lower part, is
built, and cells, which are added into the wells of the
upper part of the chamber, migrate toward the chemo-
attractant gradient. Chemotaxis12 (migration in the
presence of an attractant gradient) can be distinguished
from chemokinesis (spontaneous migration in the
absence of attractants) when quantifying stained nitro-
cellulose filters. If substances in the lower compartment
of the chamber are chemotactic for cells, these migrate
with higher frequency and speed, increasing the distance
of migration into the filter micropores with a lead made
by a few cells that can be microscopically measured,
whereas cells migrating randomly in the absence of
chemotactic gradients most often appear as a dense
layer of cells just beneath the filter surface. Data are
expressed as chemotaxis index, which is the ratio
between the distance of migration toward IL-8 and that
toward medium. The chemotaxis index of IL-8 has been
set as reference (100%).
Our experiments prove that the 11,13-dihydro-
micheliolide (Fig. 1) can only be obtained through acid
catalyzed rearrangement, it is therefore necessary to
suppose that either the solvent or the glassware the
CIBA researchers used, contained acidic impurities. It
should be added that nucleophiles bound to C(13) or
C(10) in base or acid conditions respectively.
Compounds 10 and 11 (Fig. 2b) were generated from
11,13-dihydroparthenolide 2 and identified by spectral
data. The set of products obtained by treatment of both
1 and 2 with H+/MeOH parallels the results from a
study on the lipiferolide.11 The measure of relative con-
centrations for 6 and 10 allowed us to establish that their
rate of formation are comparable and only slightly affec-
ted by the exomethylene lactone functionality (Fig. 3).
Compound 3 was also dissolved in H+/MeOH but it was
consumed in three weeks; the reaction mixture was pur-
ified by reversed phase HPLC and three main products
12, 13 and 14 were isolated: they all derived from exclu-
sive opening of the epoxide ring without any transannular
cyclization (Fig. 2c). All three products were seen to pos-
sess the (Z)-configuration at C(1)–C(10), as was con-
firmed by one-dimensional NOE difference spectroscopy.
Results
By reason of our observed diverse reactivity between 1
and 3, it is necessary to examine the reactions involving
the carbonium ions generated from 1(10)-(E)- or
1(10)-(Z)-parthenolide and to carefully analyze the
In compound 12 two olefinic protons were found at 5.26
and 5.51 ppm and the connectivity H(14)–H(1)–H(2)–

1506
H. Neukirch et al. / Bioorg. Med. Chem. 11 (2003) 1503–1510
of any acid catalyst. Thus the natural parthenolide can
be placed in the germacrolide subgroup. Besides MM
calculations allowed to measure 292 pm for the
C(1)–C(5) distance.
Mechanistically the transannular cyclization starts with
the electrophilic opening of the epoxide ring and the
generation of a carbocation susceptible to cyclize: the
3
ary C(4) cation would give rise to a [6.2.0]- opposed to a
[5.3.0]-decane ring system whereas the 2ary C(5) cation
may form a [5.3.0]- opposed to a [4.4.0]-decane skeleton.
Compound 1 yielded solely the rearrangement products
derived from the final C(10) 3ary carbocation: [6.2.0] and
[5.3.0] respectively (Fig. 5). Therefore, the stability order
of carbocations may be thought as the main driving
force in determining the regiochemistry of transannular
cyclizations. To the best of our knowledge, this feature
has never been outlined before.
Changing to 1(10)-(Z)-parthenolide 3, MM calculations
determined the preferred conformation (Fig. 4) that is
typical of the melampolide subgroup: there is an anti
arrangement, with C(14) below and C(15) above the
medium ring. Here the interatomic distance C(1)–C(5)
amounts to 333 pm and this prevents the transannular
cyclization from occurring. The acid catalyzed opening
of the epoxide ring in 3 led to the exclusive formation of
the 3ary carbocation at C(4), then the positive charge is
neutralized by losing H+ or adding methanol (Fig. 2c).
Figure 4. MM-calculated lowest energy conformers of 1 and 3 and
their C(1)–C(5) interatomic distances.
stereoelectronic requirements implied in the subsequent
rearrangements.
In order to investigate the anti-inflammatory effect in
vitro, human neutrophils were exposed to compounds 1,
2, 3, 4, 6, 7, 8, 9 or medium as control. Compound 4
diminished the chemotactic response slightly, com-
pound 3 was totally ineffective (Fig. 6). The others, as a
whole, decreased chemotaxis significantly, in detail the
effect of 6 at the lowest concentration (10 pg/mL) was
comparable to 1 at the highest concentration (107 pg/
mL). A linear dose response could be seen for compounds
1, 2, 6 and 9; compounds 7 and 8 did not show a concen-
tration-dependent response and might therefore act via
non-receptor dependent mechanisms. This however
remains speculative and will need further studies. In
previous studies homologous deactivation, that means
migration toward IL-8 after pre-treatment with IL-8,
The electrophilic transannular cyclization reaction is
strictly dependent on the spatial arrangement of the
medium cycle substrate. In general, every one of the
four subgroups of germacranolide sesquiterpenes shows
a specific conformation which is determined by the
(E)/(Z) combination of the two double bonds; the
germacrolide subgroup comprises the molecules having
the 1(10)-(E) and 4-(E) stereochemistry. In the main
conformation, the double bonds are approximately per-
pendicular to the plane of the medium ring and adopt a
crossed orientation with the two methyl groups in the
position above the plane (Fig. 4). The 4,5-epoxide
function doesn’t modify the disposition of surrounding
atoms, rather it becomes the preferred site for the attack
Figure 5. Mechanisms of acid-catalyzed transannular cyclization of 1. The formation of cation at C(5) is preferred with respect to C(4); the path-
ways a are exclusively observed whilst the pathways b do not occur. IUPAC notations for bicyclic decane ring systems are reported in square
brackets.

H. Neukirch et al. / Bioorg. Med. Chem. 11 (2003) 1503–1510
1507
Figure 6. Chemotactic response of 1, 2, 3, 4, 6, 7, 8 and 9 at four different concentrations. Data are given as percentual valuesÆstandard error of the
mean (SEM). Statistical analysis: Mann–Whitney U (*P <0.05; **P <0.001) after Kruskal–Wallis (P <0.01); n=6 for 1, 2, 3 and 4, n=3 for 6, 7, 8
and 9.
was 52.6%.13 The active compounds were therefore
comparable in their ability to reduce chemotactic
response.
In the absence of any knowledge about the receptor,
pharmacophores may provide important information in
the drug design process. To this end, parthenolide
appears to be misleading because it might be a pro-drug
which exerts its anti-inflammatory properties by means
of a metabolic activation. The chemical features in a
proper spatial arrangement, forming the pharmaco-
phore unit, should be searched in the bicyclic core of a
sesquiterpene compound. In fact, the modest structural
changes among 6, 7 and 8 have marked influence on the
migration of neutrophils. The biological role of the lac-
tone moiety remains to be established.
Discussion
Two fundamental factors emerge from our chemical
experimental results: the acid catalyzed rearrangements
regioselectively furnish products derived from a 3ary
carbocation as final intermediate and the electrophile
induced transannular cyclization occurs only when the
distance between the bonding carbon atoms is shorter
than the rough averaged value of 310 pm.
Conclusions
Following our research protocol, at point (i) we can
confidentially state that, within the four tested germa-
cranolides, the exocyclic a-methylene-g-lactone moiety
possesses no or negligible role in chemotaxis indexes.
However slight structural changes among 1 or 2 and 3
or 4 exert major influences on their biological proper-
ties, because modifications in the molecular spatial
arrangement, point (ii), affect largely their chemical
reactivity, point (iii). The (E)/(Z) geometric isomerism
drastically modifies the biological activity so it should
not originate from a particular set of functional groups
(electrophilic, nucleophilic, acidic, basic, etc.). More
plausible causes are the tendency to give the electro-
philic transannular cyclization or the structure of the
reaction products.
The present study has corrected a previous literature
report and achieved the photochemical conversion of a
germacrolide into a melampolide SL. The investigation
on the chemical properties of compounds 1, 2 and 3, in
both basic and acid solutions, explained the regiochem-
ical selectivity and appraised the minimum interatomic
distance needed for transannular bonding.
The evaluation of the chemotactic response sheds a new
light on the action of 1 as an antiinflammatory natural
product. While not exhaustive, this information pro-
vides additional hints for drug designers to decide what
compounds to make or/and screen for efficacy. The
chemical functionalities and the way of biological inter-
action of a possible major agent, are the subject of on
going research in our laboratories.
In actual fact, the bicyclic compounds 6, 7 and 9,
derived from acidic treatment of parthenolide, inhibited
chemotaxis more than the SL substrate. This lessens the
route of alkylation at C(10) by endogenous nucleo-
philes, as proposed by Fisher et al., and opens the pos-
sibility that 1 undergoes metabolic transformations to
produce more active binders. We therefore hypothesize
that, in the damaged tissue or at the active site of the
enzyme, 1 turns into a series of products that are the
effective anti-inflammatory agents. The low pH value at
the injured tissue coupled to the peculiar tendency of 1
to change into a guaiane skeleton, may constitute a sort
of clever trick for delivering the drug to the targeted
cells.
Experimental
Chemistry. Flash chromatography (FC): Merck RP-18
LiChroprep (40–65 mm). TLC: Merck-Kieselgel 60
PF254. HPLC: Reinin Dynamax 60A CN (8mm); Mach-
ery-Nagel Nucleosil 100-5 C18; 25 Â 1 cm column, solvent
flux 3mL/min. Optical rotation: JASCO-DIP-181 polari-
meter, [a]D in deg mL dmÀ1 g NMR: Varian XL-300
.
(1H at 299.94 MHz, 13C at 75.4 MHz), d in ppm using
residual solvent signals as internal standard
(CDCl3=77.0 ppm, CHCl3=7.25 ppm), J values in Hz,

1508
H. Neukirch et al. / Bioorg. Med. Chem. 11 (2003) 1503–1510
multiplicities and peak assignments from DEPT, 1H,
(d, C-7), 42.6 (d, C-11), 38.6 (t, C-3), 28.5* (t, C-9),
28.2* (t, C-8), 25.0 (t, C-2), 22.7 (q, C-14), 19.1 (q, C-15)
14.3 (q, C-13). MS: m/z 250 (M+, 4.5%), 43 (100%).
HRMS calcd for C15H22O3 (M+), 250.15689; found,
250.15698.
1H-COSY, JCH- and JCH-COSY. MS: Kratos MS80
with home-built acquisition system. Molecular mechan-
ics (MM) calculations were carried out with the
PCMODEL V7.5 computer program, searching for the
global energy minima.
1
n
5: (À)(1aR,4Z,7aS,10aS,10bS)-2,3,6,7,7a,8,10a,10b-Octa-
hydro-8-(methoxymethyl)-1a,5-dimethyloxireno[9,10]cyclo-
deca[1,2-b]furan-9(1aH)-one. To a solution of 1 (4.0 mg,
0.016 mmol) in 1 mL of MeOH was added an excess of
tetramethylguanidine (100 mL) at room temperature.
After 15 min the reaction mixture was purified by TLC
(Si60, hexane/ether 3:7) to yield 5 as a white solid
(3.6 mg, 0.0128 mmol, 80%). [a]2D5 À30 (c=1.5, EtOH).
1H NMR (CDCl3): d 5.20 (bdd, J=12, 2 Hz, 1H, H-1),
3.82 (t, J=9.0 Hz, 1H, H-6), 3.74 (dd, J=10.0, 3.0 Hz,
1H, H-13), 3.65 (dd, J=10.0, 3.0 Hz, 1H, H-13), 3.38 (s,
3H, MeO), 2.75 (d, J=9.0 Hz, 1H, H-5), 2.34.46 (m,
2H, H-7, H-11), 2.30.40 (m, 1H, H-2b), 2.20.30 (m, 1H,
H-9b), 2.05.20 (m, 3H, H-2a, H-3b, H-9a), 1.9.0 (m,
1H, H-8a), 1.69 (bs, 3H, H-14), 1.62 (m, 1H, H-8b), 1.28
(s, 3H, H-15), 1.15.30 (m, 1H, H-3a). 13C NMR
(CDCl3): d 175.0 (s, C-12), 134.5 (s, C-10), 125.2 (d,
C-1), 82.2 (d, C-6), 68.2 (t, C-13), 66.3 (d, C-5), 61.5 (s,
C-4), 59.4 (q, MeO), 48.6 (d, C-7), 45.5 (d, C-11), 41.0
(t, C-3), 36.6 (t, C-9), 30.0 (t, C-8), 24.1 (t, C-2), 17.2 (q,
C-15), 16.9 (q, C-14). MS: m/z 280 (M+, 2.0%), 45
(100%). HRMS calcd for C16H24O4 (M+), 280.16746;
found, 280.16710.
1: Parthenolide. Tanacetum parthenium was generously
given by ISAFA (Villazzano, Trento). The AcOEt
extract (3.4 g) of fresh plant material was subjected to
FC on silica and eluted with gradient of AcOEt in hex-
ane to afford 10 fractions. Central fractions were eva-
porated and the residues were submitted to HPLC
purification (CN, l=225 nm, hexane/ethanol 80:20),
thus furnishing 1 (75 mg).
2: 11,13-Dihydroparthenolide. 2 Was obtained from 1
according to literature procedure.9
3: (À)(1aR,4Z,7aS,10aS,10bS)-2,3,6,7,7a,8,10a,10b-Octa-
hydro-1a,5-dimethyl-8-methyleneoxireno[9,10]cyclodeca
A solution of 1 (6.5 mg,
[1,2-b]furan-9(1aH)-one.
0.026 mmol) in 3 mL of benzene was degassed for 30 min
by bubbling in N2, irradiated with UV-light
(l=254 nm) for 20 h and the solvent removed. The
crude reaction product was purified by means of HPLC
(CN, l=225 nm, hexane/ethanol 80:20) to provide
unreacted 1 (3.4 mg, 0.014 mmol) and 3 (2.3 mg,
0.009 mmol, 76% yield). [a]2D5 À31 (c=0.87, EtOH). H
1
NMR (CDCl3): d 6.23 (d, J=3.6 Hz, 1H, H-13), 5.53 (d,
J=3.2 Hz, 1H, H-13), 5.34 (bt, 7 Hz, 1H, H-1), 3.83 (t,
J=9.4 Hz, 1H, H-6), 2.90 (d, J=9.4 Hz, 1H, H-5), 2.76
(ddddd, J=12.0, 9.4, 3.6, 3.2, 3.0 Hz, 1H, H-7), 2.46 (td,
J=13.0, 5.0 Hz, 1H, H-9b), 2.36 (tdd, J=13.0, 5.0,
3.0 Hz, 1H, H-8a), 2.24 (m, 1H, H-2), 2.15.05 (m, 3H,
H-2, H-3b, H-9a), 1.70 (bs, 3H, H-14), 1.63 (m, 1H,
H-8b), 1.52 (s, 3H, H-15), 1.07 (m, H-3a). 13C NMR
(CDCl3): d 169.5 (s, C-12), 139.0 (s, C-11), 135.7 (s,
C-10), 125.8 (d, C-1), 119.9 (t, C-13), 81.2 (d, C-6), 63.5
(d, C-5), 60.1 (s, C-4), 42.4 (d, C-7), 37.2 (t, C-3), 26.9
(t, C-9), 25.8 (t, C-8), 23.9 (t, C-2), 21.5 (q, C-14), 18.0
(q, C-15). MS: m/z 248 (M+, 3%), 43 (100%). HRMS
calcd for C15H20O3 (M+), 248.14124; found,
248.14106.
Acid treatment of 1. To a solution of 1 (40.0 mg,
0.161 mmol) in 6 mL of MeOH was added pTsOH
(55 mg, 0.32 mmol, 2 equiv) at room temperature. After
5 h the reaction mixture was neutralized with Na2HPO4
(2 equivin 1 mL H O), added of RP18 stationary phase
2
and dried. The adsorbed material was washed with
water and eluted with acetone. This portion (45 mg) was
subjected to HPLC (Nucleosil, CH3CN/H2O 75: 25,
l=213 nm) to give 8 (10.3 mg, 0.042 mmol, 26%), 7
(7.0 mg, 0.028 mmol, 18%) and 22.4 mg of a mixture
which by HPLC (CN, hexane/EtOH 80 : 20, l=213 nm)
furnished pure 6 (21.0 mg, 0.075 mmol, 47%) and 9
(4.0 mg, 0.014 mmol, 9%).
6: (À)(3aS,6R,6aR,9R,9aS,9bS)-Decahydro-9-hydroxy-
6-methoxy-6,9-dimethylazuleno[4,5-b]furan-2(3H)-one.
[a]2D5 À31 (c=4.2, EtOH). 1H NMR (CDCl3): d 6.20
(d, J=3.5 Hz, 1H, H-13), 5.49 (d, J=3.1 Hz, 1H,
H-13), 4.18 (dd, J=11.7, 9.7 Hz, 1H, H-6), 3.17 (s,
3H, MeO), 2.80 (dd, J=11.5, 8.0 Hz, 1H, H-1),
2.70.88 (m, 1H, H-7), 2.32 (dd, J=11.7, 11.5 Hz, 1H,
H-5), 2.15 (dddd, 13.0, 5.5, 4.7, 4.0 Hz, 1H, H-8a),
1.95 (ddd, 14.0, 5.5, 4.7 Hz, 1H, H-9b), 1.75.90 (m,
3H, H-2, 2H-3), 1.68 (m, 1H, H-9a), 1.44.59 (m, 1H,
H-2), 1.22.44 (m, 1H, H-8b), 1.36 (s, 3H, H-15), 1.14
(s, 3H, H-14). 13C NMR (CDCl3): d 169.6 (s, C-12),
138.9 (s, C-11), 120.1 (t, C-13), 82.7 (d, C-6), 80.4 (s,
C-4), 78.1 (s, C-10), 55.4 (d, C-5), 48.3 (q, MeO), 46.8
(d, C-7), 45.9 (d, C-1), 39.0 (t, C-3), 36.0 (t, C-9),
25.5 (t, C-2), 24.4 (t, C-8), 24.4 (q, C-15), 22.5 (q,
C-14). MS: m/z 280 (M+, 2.4%), 43 (100%). HRMS
calcd for C16H24O4 (M+), 280.16746; found,
280.16734.
4: (À)(1aR,4Z,7aS,8S,10aS,10bS)-2,3,6,7,7a,8,10a,10b-
Octahydro-1a,5,8-trimethyloxireno[9,10]cyclodeca[1,2-b]
furan-9(1aH)-one. A solution of 2 (9.0 mg, 0.036 mmol)
in 3 mL of benzene was degassed for 30 min by bubbling
in N2, irradiated with UV-light (l=254 nm) for 20 h
and the solvent removed. The crude reaction product was
purified by means of HPLC (CN, l=213 nm, hexane/
ethanol 80:20) to provide unreacted
2 (4.8 mg,
0.019 mmol) and 4 (4.1 mg, 0.017 mg, 100%). [a]2D5 À74
(c=3.4, EtOH). 1H NMR (CDCl3): d 5.29 (bt, 7 Hz, 1H,
H-1), 3.81 (t, J=9.5 Hz, 1H, H-6), 2.80 (d, J=9.5 Hz,
1H, H-5), 2.45 (td, J=13.0, 5.0 Hz, 1H, H-9b), 2.27 (dq,
J=12.0, 7.0 Hz, 1H, H-11), 1.98.20 (m, 5H, 2H-2, H-3b,
H-8a, H-9a), 1.85 (tdd, J=12.0, 9.5, 3.0 Hz, 1H, H-7),
1.67 (bs, 3H, H-14), 1.53 (m, 1H, H-8b), 1.52 (s, 3H,
H-15), 1.26 (d, 3H, H-13), 1.04 (m, 1H, H-3a). 13C
NMR (CDCl3): d 179.2 (s, C-12), 137.2 (s, C-10), 126.8
(d, C-1), 82.6 (d, C-6), 64.9 (d, C-5), 61.2 (s, C-4), 47.3

H. Neukirch et al. / Bioorg. Med. Chem. 11 (2003) 1503–1510
1509
9: (À)(3aS,6R,6aR,8aR,9S,9aS)-Decahydro-9-hydroxy-
6-methoxy-6,8a-dimethyl-3-methylenecyclobuta[6,7]cyclo-
octa[1,2-b]furan-2(3H)-one. [a]2D5 À45 (c=1.0, EtOH).
1H NMR (CDCl3): d 6.25 (d, J=3.5 Hz, 1H, H-13), 5.58
(d, J=3.1 Hz, 1H, H-13), 4.18 (dd, J=9.8, 7.2 Hz, 1H,
H-6), 3.58 (d, 9.8 Hz, 1H, H-5), 3.16 (s, 3H, MeO),
3.08.18 (m, 1H, H-7), 2.52 (dd, J=10.9, 8.0 Hz, 1H, H-
1), 2.23 (tdd, 13.5, 5.5, 4.5 Hz, 1H, H-8a), 2.14 (dddd,
J=10.9, 10.7, 10.5, 9.9 Hz, 1H, H-2b), 1.90 (m, 1H, H-
2a), 1.60.84 (m, 3H, 2H-3, H-9), 1.40 (ddt, J=13.5,
11.5, 4.0 Hz, 1H, H-8b), 1.20.28 (m, 1H, H-9), 1.28 (s,
3H, H-15), 1.23 (s, 3H, H-14). 13C NMR (CDCl3): d
169.2 (s, C-12), 140.6 (s, C-11), 122.2 (t, C-13), 84.5 (d,
C-5), 82.1 (d, C-6), 76.6 (s, C-10), 49.3 (q, MeO), 47.6
(d, C-1), 43.7 (d, C-4), 40.3 (t, C-7), 34.2 (t, C-3), 32.5 (t,
C-9), 27.0 (t, C-8), 23.6 (q, C-14), 22.3 (t, C-2), 14.0 (s,
C-15). MS: m/z 280 (M+, 0.2%), 85.1 (100%). HRMS
calcd for C16H24O4 (M+), 280.16746; found, 280.16702.
13: (À)(3aS,6Z,11S,11aS)-3a,4,5,8,9,10,11,11a-Octahydro-
11-hydroxy-6-methyl-3,10-bis(methylene)cyclodeca[b]furan-
2(3H)-one. [a]D25 À94 (c=0.80, EtOH). 1H NMR
(CDCl3): d 6.21 (d, J=1.7 Hz, 1H, H-13), 5.58 (d,
J=1.5 Hz, 1H, H-13), 5.12 (bdd, J=12, 4 Hz, 1H, H-1),
4.96 (bs, 1H, H-15), 4.92 (d, J=1.7 Hz, 1H, H-15), 4.55
(d, J=9.4 Hz, 1H, H-5), 3.88 (d, J=9.4 Hz, 1H, H-6),
2.68.82 (m, 2H, H-3a, H-7), 2.60 (bddd, J=13, 11, 6 Hz,
1H, H-9b), 2.26.42 (m, 2H, H-2b, H-3b), 2.02.20 (m,
2H, H-2a, H-8a), 1.82.92 (m, 1H, H-9a), 1.69 (d,
1.4 Hz, 3H, H-14), 1.65 (m, 1H, H-8b). 13C NMR
(CDCl3): d 169.8 (s, C-12), 144.9 (s, C-4), 139.4 (s, C-
11), 132.3 (s, C-10), 126.7 (d, C-1), 122.4 (t, C-13),
120.2 (t, C-15), 82.6* (d, C-5), 81.0* (d, C-6), 38.3 (d,
C-7), 31.4** (t, C-3), 30.6** (t, C-2), 27.4 (t, C-9), 26.2
(t, C-8), 22.3 (q, C-14). MS: m/z 248 (M+, 8.1%), 41
(100%). HRMS calcd for C15H20O3 (M+), 248.14124;
found, 248.14037.
Kinetic investigations of the acid-catalysed reactions of 1
and 2 with MeOH. To each solution of 1 (6.5 mg,
0.026 mmol) and 2 (6.5 mg, 0.026 mmol) in 6.5 mL of
MeOH-d4 was added pTsOH (8.9 mg, 0.052 mmol,
2 equiv) and the reactions at 20 ꢀC were monitored by
14: (À)(3aS,6Z,10R,11R,11aS)-3a,4,5,8,9,10,11,11a-Octa-
hydro-11-hydroxy-10-methoxy-6,10-dimethyl-3-methylene-
cyclodeca[b]furan-2(3H)-one. [a]2D5 À83 (c=0.53, EtOH).
1H NMR (CDCl3): d 6.19 (d, J=2.1 Hz, 1H, H-13), 5.58
(d, J=1.9 Hz, 1H, H-13), 5.26 (bt, J=8 Hz, 1H, H-1),
4.82 (dd, J=5.4, 2.5 Hz, 1H, H-6), 3.58 (bd, J=5 Hz,
1H, H-5), 3.47 (ddddd, J=13.0, 5.1, 2.5, 2.1, 1.9 Hz,
1H, H-7), 3.20 (s, 3H, MeO), 2.20.35 (m, 1H, H-2),
1.9.2 (m, 3H, H-2, 2H-9), 1.5.8 (m, 4H, 2H-3, 2H-8),
1.71 (s, 3H, H-14), 1.22 (s, 3H, H-15). 13C NMR
(CDCl3): d # (s, C-12), 140.1 (s, C-11), # (s, C-10),
127.4 (d, C-1), 121.6 (t, C-13), 81.6 (d, C-6), 79.9 (s,
C-4), 77.6 (d, C-5), 50.8 (q, MeO), 49.0 (d, C-7), 31.9*
(t, C-3), 31.5* (t, C-8), 28.6* (t, C-9), 27.1* (t, C-2),
23.1 (q, C-15), 21.8 (q, C-14). MS: m/z 280 (M+,
4.8%), 72 (100%). HRMS calcd for C16H24O4 (M+),
280.16746; found, 280.16750.
1
means of H NMR spectroscopy. The relative amounts
of the main products 6 and 10 were calculated from the
integral ratio of the protons H-13 of 6/(1+6) and of the
proton H-6 of 10/(2+10).
Acid treatment of 3. To a solution of 3 (15.0 mg,
0.0605 mmol) in 3 mL of MeOH was added pTsOH
(21 mg, 0.12 mmol, 2 equiv) at room temperature. After
22 days the reaction mixture was neutralized with
Na2HPO4 (2 equivin 1 mL H O), added of RP18 sta-
2
tionary phase and dried. The adsorbed material was
washed with water and eluted with acetone (14 mg). The
1H NMR spectrum of the crude eluate suggested the
following composition: 12 (50%), 13 (25%), 14 (12%).
The mixture was then purified by HPLC (Nucleosil,
CH3CN/H2O 38:62, l=213 nm) to give 12 (1.7 mg,
0.007 mmol, 11%), 13 (1.2 mg, 0.0043 mmol, 8.0%) and
14 (0.8 mg, 0.0029 mmol, 5%).
#
*, ** May be interchanged; missing.
Biology. RPMI 1640 with phenol red was purchased
from Biological Industries (Kibbutz Beit Haemek,
Israel). Bovine serum albumin (BSA) was from Dade
Behring (Marburg, Germany), IL-8 from Roche Diag-
nostics (Mannheim, Germany). Lymphoprep1 was from
Nycomed Pharma AS (Oslo, Norway), 48-blindwell
microchemotaxis chamber from Neuroprobe (Gaithers-
burg, MD, USA), 5 mm pore-sized nitrocellulose filters
from Sartorius (Goettingen, Germany). All experiments
were obtained by a single technician who was blinded for
test reagents in order to avoid systematic error in
answering this sensitive question.
12: (À)(3aS,6Z,9Z,11S,11aS)-3a,4,5,8,11,11a-Hexahydro-
11-hydroxy-6,10-dimethyl-3-methylenecyclodeca[b]furan-
2(3H)-one. [a]D25 À53 (c=0.87, EtOH). 1H NMR
(CDCl3): d 6.26 (d, J=3.0 Hz, 1H, H-13), 5.56 (d,
J=2.6 Hz, 1H, H-13), 5.51 (bt, J=8.0 Hz, 1H, H-3), 5.26
(bt, J=7.5 Hz, 1H, H-1), 4.51 (d, J=5.1 Hz, 1H, H-5),
4.43 (t, J=5.1 Hz, 1H, H-6), 2.86–2.96 (m, 1H, H-7),
2.8.9 (overlapped, 1H, H-2), 2.72 (dt, J=13.0, 7.0 Hz,
1H, H-2), 2.57 (bddd, J=13.0, 11.0, 2.5 Hz, 1H, H-9b),
2.14 (ddd, J=13.0, 6.0, 2.5 Hz, 1H, H-9a), 1.90 (dddd,
J=13.0, 11.0, 3.5, 2.5 Hz, 1H, H-8a), 1.78 (d, 1.4 Hz, 3H,
H-15), 1.71 (d, J=1.2 Hz, 3H, H-14), 1.60 (dddd,
J=13.0, 10.5, 6.0, 2.5 Hz, 1H, H-8b). 13C NMR
(CDCl3): d 169.8 (s, C-12), 139.6 (s, C-11), 134.9 (s,
C-4), 133.6 (s, C-10), 125.8 (d, C-3), 123.2 (d, C-1),
122.4 (t, C-13), 84.4 (d, C-6), 74.1 (d, C-5), 41.0 (d,
C-7), 30.8* (t, C-2), 28.3* (t, C-8), 27.8* (t, C-9), 24.2
(q, C-14), 20.1 (q, C-15). MS: m/z 248 (M+, 23.8%),
41 (100%). HRMS calcd for C15H20O3 (M+);
248.14124, found, 248.14097.
Chemotaxis assay. Neutrophils were obtained from
peripheral EDTA anticoagulated blood of healthy
volunteers after discontinuous density gradient cen-
trifugation on Lymphoprep1 by dextran sedimentation
and centrifugation through a layer of Ficoll–Hypaque,
followed by hypotonic lysis of contaminating ery-
throcytes using sodium chloride solution.14 Cell pre-
parations yielded more than 95% neutrophils (by
morphology in GIEMSA stains) and more than 99%
viability (by trypan dye exclusion). Experiments were
performed in RPMI 1640 with 0.5% BSA.

1510
H. Neukirch et al. / Bioorg. Med. Chem. 11 (2003) 1503–1510
For deactivation, cells were incubated for 20 min with
different substrates at various concentrations or
remained untreated and were washed twice in phos-
phate buffered saline (PBS) before testing for chemo-
taxis. 50 mL of the cell suspension (1 Â 106 cells/mL) was
put into the upper compartment of the chemotaxis
chamber and cells were allowed to migrate for 30 min at
37 ꢀC in a humidified atmosphere (5% CO2) toward
IL-8 (1 nM), which was put in the lower wells as chemo-
attractant. After this migration period the nitrocellulose
filters were dehydrated, fixed and stained with haema-
toxylin-eosin. Migration depth was quantified micro-
scopically, measuring the distance (mm) from the surface
of the filter to the leading front of three cells.15
References and Notes
1. Knight, D. W. Nat. Prod. Rep. 1995, 12, 271.
2. Rodriguez, E.; Towers, G. H. N.; Mitchell, J. C. Phyto-
chemistry 1976, 15, 1573.
3. Fischer, N. H.; Weidenhamer, J. D.; Riopel, J. L.;
Quijano, L.; Menelaou, M. A. Phytochemistry 1990, 29,
2479.
4. Macias, F. A.; Galindo, J. C. G.; Massanet, G. M. Phyto-
chemistry 1992, 31, 1969.
ˇ
5. Fischer, N. H.; Lu, T.; Cantrell, C. L.; Castaneda-Acosta,
J.; Quijano, L.; Franzblau, S. G. Phytochemistry 1998, 49,
559.
6. Rungeler, P.; Castro, V.; Mora, G.; Goren, N.; Vichnewski,
¨
¨
W.; Pahl, H. L.; Merfort, I.; Schmidt, T. J. Bioorg. Med.
Chem. 1999, 7, 2343.
7. Smith, W. B.; Gamble, J. R.; Clark-Lewis, I.; Vadas, M. A.
Immunology 1991, 72, 65.
8. Jonas, E.; Dwenger, A.; Lueken, B.; Boehme, U. J. Biolumin.
Chemilumin. 1991, 6, 19.
9. Govindachari, T. R.; Joshi, B. S.; Kamat, V. N. Tetra-
hedron 1965, 21, 1509.
10. Castaneda-Acosta, J.; Fischer, N. H.; Vargas, D. J. Nat.
ˇ
Data are expressed as mean and SEM. Means were com-
pared by Mann–Whitney U-test after Kruskal–Wallis
analysis of variance. A P-value <0.05 was considered
significant. Analyses were performed using the StatView
software package (Abacus Concepts, Berkeley, CA).
Prod. 1993, 56, 90.
11. Wilton, J. H.; Doskotch, R. W. J. Org. Chem. 1983, 48,
4251.
12. Calabrese, E. J. Crit. Rev. Toxicol. 2001, 31, 615.
13. Wiedermann, C. J. unpublished observation.
14. Wiedermann, C. J.; Niedermuhlbichler, M.; Braunsteiner,
H.; Widermann, C. J. J. Clin. Invest. 1992, 89, 1580.
15. Haston, W. S.; Wilkinson, P. C. Methods Enzymol. 1988,
162, 17.
Acknowledgements
We thank the University of Bolzano for financial support
of H.N. We are grateful to Dr. C. Vender and the ISAFA
(Istituto Sperimentale per l’Assestamento Forestale e per
l’Alpicoltura, Villazzano - Trento) for providing the plant
material. This work was supported by M.U.R.S.T.
(Cofin2000).