Biosynthesis of angular furanocoumarins: Mechanism and stereochemistry of the oxidative dealkylation of columbianetin to angelicin in Heracleum mantegazzianum (Apiaceae)

Article abstract of DOI:10.1002/(SICI)1522-2675(19980909)81:9<1596::AID-HLCA1596>3.0.CO;2-F

Deuterium-labelled 5-fluorocolumbianetin 13 was synthesized as a metabolic probe to examine the stereochemical course of the bioconversion of (+)-columbianetin (12) into the angular furocoumarin angelicin (5). In leaves of Heracleum mantegazzianum, oxidative dealkylation of the specifically deuterated fluorocolumbianetin 13 proceeded by syn-elimination of a D-atom, from C(9), and the vicinal 1-hydroxy-1-methylethyl substituent, yielding 5- fluoroangelicin (23). This matches the stereochemical course of the related reaction converting (+)-marmesin (10) into the linear furocoumarin psoralen (1). Key steps in the synthesis of 5-fluorocolumbianetin (13) were the copper-catalysed alkynylation/cyclization of 5-fluoro-8-iodoumbelliferone (15) followed by a transfer hydrogenation, which established the cis- orientation of the D-Atom and the 1-hydroxy-1-methylethyl substituent.

Full text of DOI:10.1002/(SICI)1522-2675(19980909)81:9<1596::AID-HLCA1596>3.0.CO;2-F

1596
Helvetica Chimica Acta ± Vol. 81 (1998)
Biosynthesis of Angular Furanocoumarins: Mechanism and Stereochemistry
of the Oxidative Dealkylation of Columbianetin to Angelicin
in Heracleum mantegazzianum (Apiaceae)
by Volker Stanjek and Wilhelm Boland*
Max-Planck-Institut für Chemische Ökologie, Tatzendpromenade 1a, D-07745 Jena
Deuterium-labelled 5-fluorocolumbianetin 13 was synthesized as a metabolic probe to examine the
stereochemical course of the bioconversion of (‡)-columbianetin (12) into the angular furocoumarin angelicin
(5). In leaves of Heracleum mantegazzianum, oxidative dealkylation of the specifically deuterated
fluorocolumbianetin 13 proceeded by syn-elimination of a D-atom, from C(9), and the vicinal 1-hydroxy-1-
methylethyl substituent, yielding 5-fluoroangelicin (23). This matches the stereochemical course of the related
reaction converting (‡)-marmesin (10) into the linear furocoumarin psoralen (1). Key steps in the synthesis of
5-fluorocolumbianetin (13) were the copper-catalysed alkynylation/cyclization of 5-fluoro-8-iodoumbelliferone
(15) followed by a transfer hydrogenation, which established the cis-orientation of the D-Atom and the 1-
hydroxy-1-methylethyl substituent.
1. Introduction. ± Plants have evolved many traits, both physical and chemical, that
protect them from damage by pathogens and herbivores. Unlike physical traits like
spines, chemical defences may be either constitutively present or inducible. Induced
responses to herbivory may be localized, restricted to a small area around the damage
site, or systemic, causing effects far distant from the original injury [1]. One of the most
intensively studied constitutive and/or inducible chemical defenses is the production of
furocoumarins, e.g., 1 ± 7. Furocoumarins (ˆ furanocoumarins) represent a large class
of compounds with widespread occurrence in the Apiaceae and Rutaceae [2]. They are
typical phytoalexins, being (photo)toxic to a wide variety of organisms, including
bacteria, fungi, insects, and mammals [3]. Furocoumarin biosynthesis can be induced
by insect herbivory [4], infection with pathogens [5], light stress [6], and even by
airborne methyl jasmonate [7]. In some plants such stimulation not only increases the
amount of furocoumarins within the leaf tissue, but also leads to higher titres on the leaf
surface [8]. Contact with induced leaves causes blistering and erythrema, accompanied
by allergic reactions, well-known occupational diseases during the harvesting of celery,
parsley, or parsnip [9]. All furocoumarins are able to intercalate with DNA, but only
linear furocoumarins of type 1 ± 4 will cross-link the single-strands, by forming [2 ‡ 2]-
photoadducts between appropriately located nucleobases [10].
Unlike the widely distributed linear furocoumarins, in which the furan moiety is
attached to the atoms C(6) and C(7) of the 1-benzopyran-2-one moiety, angular
furocoumarins like, e.g., angelicin (5) and sphondin (6) are much less common. In many
plants, they are found as minor components in mixtures dominated by linear
furocoumarins. Since no plants are known which exclusively produce angular
furocoumarins, their biosynthesis may be considered as a more recently evolved trait.
The naturally occurring mixtures of both linear and angular furocoumarins are more

Helvetica Chimica Acta ± Vol. 81 (1998)
1597
toxic to herbivorous insects, especially in the absence of light, than pure psoralen (1) or
xanthotoxin (3) [3] [11]. As shown recently, this synergistic effect is apparently due to a
strong inhibition of the detoxifying P450-type enzymes of the insects by the angular
furocoumarins [3] [12]. Thus, plants which produce mixtures of both types of
furocoumarins are generally better protected against even those herbivores which
are adapted to plants containing large amounts of linear furocoumarins. For example,
angelicin seriously interferes with the growth and reproduction of the black swallowtail
(Papilio polyxenes), a specialist that is adapted to feed on plants containing the linear
furocoumarin xanthotoxin (3) [13].
These findings suggest that the biosynthesis of angular furocoumarins may have
evolved from the pathway leading to linear furocoumarins in response to selection
pressure by herbivores [3] [14]. The required diversification of the primordial pathway
could easily have been achieved by selection of an enzyme catalysing the prenylation of
umbelliferone (8) at C(8), instead of C(6) as is the case in the biosynthesis of the linear
furocoumarins (Scheme 1).
By subsequent evolutionary optimization of the other enzymes of the ancient
pathway, a new set of biocatalysts, adapted to the angular regioisomers, may have
developed. Accordingly, both pathways share the same type of intermediates like 9 and
11, or 10 and 12, and in both pathways, the unfunctionalized parent systems 1 and 5 are
further modified by oxidations and methylations [15]. Although several enzymes from
the linear furocoumarin pathway have been isolated and studied in detail [16], almost
nothing is known about the corresponding enzymes from the angular furocoumarin
pathway. A most remarkable transformation in the biosynthesis of both types of
furocoumarins is the oxidative dealkylation of (‡)-marmesin (10) or (‡)-columbia-
netin (12). In the case of the psoralen (1) biosynthesis, the reaction is catalysed by a
cytochrome P450 and requires molecular oxygen and NADPH as cofactors [17].
Recently, the mechanism of this bioconversion was established as a single-step
transformation cleaving (‡)-marmesin (10) into psoralen (1) and acetone [18] [19].
As outlined in Scheme 2, the HSi atom at C(3) of (‡)-marmesin (10) is attacked by
.
‡
the electrophilic [FeIV(porphyrinato)(ˆO)] species, and a benzylic radical is formed.
The reactive intermediate is stabilized by b-cleavage generating psoralen (1) and a 1-
hydroxy-1-methylethyl radical. Subsequent reaction with neighbouring [FeIV(porphyr-

1598
Helvetica Chimica Acta ± Vol. 81 (1998)
Scheme 1. Biosynthesis of the Linear Furocoumarin Psoralen (1) and the Angular Furocoumarin Angelicin (5)
from the Common Precursor Umbelliferone (8)
inato)(OH)] yields acetone, presumably via a geminal diol, together with the reduced
cytochrome catalyst. Although the dealkylation of (‡)-columbianetin 12) is formally
related, no mechanistic or stereochemical information is available yet. We now report i)
the synthesis of D-labelled 5-fluorocolumbianetin 13, a valuable metabolic probe for
the mechanistic analysis of angular furocoumarin biosynthesis, and ii) the first data on
the stereochemical course of oxidative dealkylation of columbianetin to angelicin in
leaves of Heracleum mantegazzianum.
Scheme 2. Mechanistic Rationale for the P450-Catalysed Dealkylation of (‡)-Marmesin (10) to Psoralen (1) and
Acetone
Results and Discussion. ± 1. Synthesis of cis-5-Fluoro(9-D)columbianetin (13) as a
Metabolic Probe. While mechanistic studies towards linear furocoumarins can rely on
microsomal preparations from well-established cell cultures of several Apiaceae, in
particular Ammi majus [17] [18], no such system is available for the production of
angular furocoumarins. Therefore, the experiments and the metabolic probe(s) were
designed for the use with intact plants. To secure identification of the expected
metabolites, undiluted by the large amounts of natural furocoumarins, we decided to
use a C(5)-fluorinated columbianetin 13 as a metabolic probe. Due to the similar size of
F- and H-atoms, space filling of the probe and the natural precursor will be almost
identical; negative effects from the electron-withdrawing substituent are not expected
to be strong enough to prevent the oxidative dealkylation. To gain insight into the

Helvetica Chimica Acta ± Vol. 81 (1998)
1599
stereochemistry of the transformation, the precursor has to be additionally labelled by
a D-atom at C(9) in a configurationally defined manner, either cis or trans to the
hydroxy(methyl)ethyl substituent. These structural demands are easily fulfilled
following a synthetic route developed previously for the preparation of configuration-
ally defined, deuterium-labelled (Æ)-marmesins [19]. As outlined in Scheme 3, exclusive
cis-orientation of the D-atom at C(9) and the side chain, as in 13, is feasible via transfer
hydrogenation of 14 in the presence of Pd/C using formate as a H2 source [20].
Scheme 3. retro-Synthetic Route to 5-Fluoro-8-iodoumbelliferone (15)
The substituted furocoumarin 14 should be accessible from dihalide 15 by a copper-
catalysed alkynylation with 2-methylbut-3-yn-2-ol [20]. Thus, only the synthesis of the
dihalide remained to be developed.
Although direct fluorination of the readily available 8-iodoumbelliferon [21] should
be possible, in principle, the dihalide 15 was synthesized utilizing the protocol shown in
Scheme 4. The symmetric precursor 16 already containing the F-atom was converted to
13 in only a few steps and high overall yield. Dangerous manipulations with elemental
F2 or unstable fluorinating reagents were, thus, not required.
Dissymmetrization of commercial 16 was achieved by a Vilsmeier-Haak formyla-
tion, yielding the two regioisomeric aldehydes 17a and 17b in almost quantitative yield.
Due to the lower steric hindrance next to the F-atom, the desired regioisomer 17a was
formed in excess (17a/17b 5.5 :1). Since it proved difficult to separate the two
regioisomers by column chromatography, the mixture was directly subjected to a
Wittig-Horner olefination. The reaction proceeded virtually quantitatively, and after
chromatographic separation (SiO2), the configurationally pure (E)-ester 18 was
obtained in 73% overall yield. Subsequent ether cleavage with BBr3 proceeded
instantaneously and was accompanied by an (E) !(Z) isomerization of the acrylic side
chain with the (Z)-isomer being trapped by cyclization to the 5-fluoroumbelliferone
(19) in a single operation. Introduction of the I-atom at C(8) was achieved with
exceptionally high positional preference using I2 and aqueous ammonia in dioxan.
Following the sequence of Scheme 4, the required 5-fluoro-8-iodoumbelliferone (15)
was obtained in ca. 40% overall yield from 16.
The C-skeleton of 5-fluorocolumbianetin 13 was completed by copper-catalysed
alkynylation of 15 with deuterium-labelled 2-methylbut-3-yn-2-ol as depicted in
Scheme 5. The introduction of a D-atom at C(9) of 14 was achieved by employing an
alkynol possessing a D-atom at the terminal C-atom and at the hydroxy group.
Moreover, a high degree of D-labelling (> 98%) at C(9) of 14 required an absolute
solvent (pyridine), a rigorous exclusion of moisture, and a previous exchange of the
phenolic proton of 15 by a D-atom (by stirring with MeOD in the presence of a catalytic
amount of DCl) [19]. The final transfer hydrogenation placed the two H-atoms in a cis-

1600
Helvetica Chimica Acta ± Vol. 81 (1998)
Scheme 4. Synthesis of 5-Fluoro-8-iodoumbelliferone (15)
fashion across the CˆC bond of the furan moiety [19] [20]. However, the yields were
low (ca. 5 ± 8%) and could not be optimized, since the reduction of the unsaturated
lactone proceeded faster. To overcome the low yield, both CˆC bonds were reduced
yielding 20. Then, the unsaturated lactone substructure was regenerated by successive
deprotonation and selenenylation of 20 with lithium diisopropylamide and phenyl-
selenenyl chloride [22]. Oxidation of the selenide with H2O2 and final elimination of
phenylselenenic acid provided 13 in 37% overall yield from 14.
Scheme 5. Synthesis of Stereospecifically Deuterated Fluorocolumbianetin 13 as a Metabolic Probe for Analysis
of the Stereochemical Course of Angular Furocoumarin Biosynthesis
Fluorinated angelicin 23, required as an authentic reference for the biotransfor-
mation of the cis-5-fluoro(9-D)columbianetin (13) was synthesized along a similar
protocol (Scheme 6). Key step of the sequence was the copper-catalysed decarbox-
ylation of the readily available angelicin-type ester 22 [21]. Thus, copper-catalysed
alkynylation of 15 with ethyl propiolate furnished ester 22. Hydrolysis and thermal
decarboxylation of the resulting acid for 1 h in quinoline/copper at 2108 afforded the
required 5-fluoroangelicin (23) in 43% yield.
2. Incubation Experiments with Deuterium-Labelled 5-Fluorocolumbianetin 13. The
labelled cis-5-fluoro(9-D)columbianetin (13), Triton X 100, and jasmonic acid

Helvetica Chimica Acta ± Vol. 81 (1998)
Scheme 6. Synthesis of 5-Fluoroangelicin (23)
1601
(required for elicitation of the biosynthesis of the furocoumarins [8]) were dispersed in
tap water and sonicated until a stable emulsion resulted. Freshly detached leaves of
two- to three-month-old plants of Heracleum mantegazzianum were immediately
placed into the emulsion. After five days, the nonpolar surface compounds including
the furocoumarins and the metabolites of 13 were collected from the leaf surface by
briefly dipping the leaves in CH2Cl2. The constituents were analysed by GLC/MS and
identified by comparison with reference compounds [8]. Analysis of the surface lipids
was preferred, because of the ease of extraction and the simplicity of the compound
spectrum, compared to that of corresponding leaf-tissue extracts.
Alternatively, the leaf surfaces of young plants (two- to three-week-old) of
H. mantegazzianum were repeatedly sprayed with the above emulsion at four-day
intervals. After 3 ± 4 weeks, the surface lipids were collected as described above, and the
constituents were analysed by GLC/MS. Unlike the direct administration experiment
which resulted in ca. 50% mortality of the treated plants, surface application had no
visible negative impact on the plants.
As a matter of fact, both application modes resulted in the formation of 5-
fluoroangelicin 23 from the administered 5-fluorocolumbianetin, albeit in very low
yield. About 0.5 ± 1% of the total angelicin isolated from the surface of the leaves
proved to be 5-fluoroangelicin (23). However, as shown in Fig. 1, the mass spectra of
the metabolite and the synthetic reference were fully congruent, as were their retention
times.
The molecular ion of the metabolite at m/z 204 Da indicated complete loss of D-
labelling and, hence, the oxidative dealkylation of cis-5-fluoro(9-D)columbianetin (13)
had proceeded exclusively by syn-elimination. This observation was of great
importance, since exclusive removal of the D-atom, despite a potential isotope effect
(vide infra), characterized the metabolite as a product of an enzymatic transformation.
Moreover, syn-elimination of D C(9) and the 1-hydroxy-1-methylethyl substituent is
in perfect agreement with the stereochemical course of the corresponding oxidative
dealkylation of (‡)-marmesin 10 to the linear furocoumarin psoralen (1) [18] [20].
Thus, it is reasonable to assume that the oxidative degradation of (‡)-columbianetin
(12) to angelicin (5) is also catalysed by a cytochrome P450, similar to that involved in
the biosynthesis of psoralen (1).
3. Biomimetic Dealkylation of 5-Fluoro(9-D)columbianetin. Recently, we described
the first chemical model system for the oxidative dealkylation of marmesin to psoralen
using (tetraphenylporphyrinato)manganese and iodosylbenzene [18]. This reagent
combination has been frequently used to simulate cytochrome P450 catalysed

1602
Helvetica Chimica Acta ± Vol. 81 (1998)
Fig. 1. Mass spectrum of fluoroangelicin 23 resulting from the enzymatic transformation of deuterated
fluorocolumbianetin 13. Insert: molecular ion and first fragments of the synthetic reference. Both spectra are
congruent, i.e., the vicinal cis-arranged D-atom removed from C(9) of 13 during the oxidative dealkylation.
oxygenation reactions [23]. Although only minor amounts (< 1%) of the D-labelled 5-
fluorocolumbianetin 13 were dealkylated to 5-fluoroangelicin 23, the mass spectrum of
the product clearly indicated that in this case, unlike in the enzymatic conversion (see
above), the trans-arranged H-atom was preferentially removed, leading to a product
carrying a D-atom (70% D C(9); see Fig. 2). The rather low conversion rate of 13 was
largely due to the electronic effect of the F-substituent, since the corresponding non-
fluorinated columbianetin was converted to angelicin in much higher yield (ca. 10%).
Fig. 2. Mass spectrum of fluoroangelicin 23 resulting from the biomimetic dealkylation of fluorocolumbianetin
13 with (tetraphenylporphyrinato)manganese(III) and iodosobenzene. The isotopomer of the deuterated
fluoroangelicin prevails, indicating a predominant loss of the trans-arranged H-atom at C(9) of 13.

Helvetica Chimica Acta ± Vol. 81 (1998)
1603
On the other hand, the amount of syn-elimination (ca. 30%) in the biomimetic
dealkylation of 13 was remarkably high and may be due to a preorientation of the
porphyrinato system and the 5-fluorocolumbianetin via a H-bond between the
substrate and the [MnV(porphyrinato)(ˆO)] complex as outlined in Scheme 2. In
the enzymatic transformation, the facial arrangement of the reactive [FeIV(porphyri-
‡
nato)(ˆO)] complex is responsible for the observed complete syn-elimination of the
HSi atom at C(9) and the hydroxyalkyl substituent. This peculiar arrangement at the
same time prevents the escape of long-lived reactive radicals from the active centre,
since the close, cage-type arrangement of all interacting species secures almost
immediate trapping of the ensuing radicals.
Conclusions. ± The stereochemical course of the bioconversion of (‡)-columbia-
netin (12) into angelicin (5), as shown above, is identical to the stereochemical course
of the related conversion of (‡)-marmesin (10) into the linear furocoumarin psoralen
(1) [18]. In both cases, the oxidative dealkylation proceeds by syn-elimination of an
H-atom and a vicinal 1-hydroxy-1-methylethyl substituent (cf. Scheme 1). Although the
present study was performed with intact plants, which did not allow for a direct
analytical proof of the acetone fragment, its formation is reasonable because of the
perfect stereochemical analogy with psoralen synthase [18]. As with psoralen synthase,
angelicin synthase catalyses the oxidative dealkylation of stereospecifically deuterated
columbianetin 13 against a potential kinetic isotope effect with a quantitative loss of the
vicinal cis-located D-atom. This is especially remarkable, since the biomimetic
dealkylations with D-labelled marmesin and (tetraphenylporphyrinato)-manganese,
activated by iodosylbenzene, exhibit a kinetic isotope effect of kH/kD ꢀ 4 (Tˆ 208) [18]
suggesting that the removal of the H radical from C(3) is rate-contributing (cf.
Scheme 2).
The perfect mechanistic and stereochemical analogies between the two pathways
leading to linear and angular furocoumarins strongly support the idea that the pathway
to angular furocoumarins indeed originated from a primordial route to linear
furocoumarins [24]. Considering the negative impact of angelicin on the detoxification
of linear furocoumarins [13] in insects as an important evolutionary driving force,
accidental appearance of angular furocoumarins may have been selected immediately
as a trait for chemical protection and may explain its subsequent phylogenetic inertia.
Although todayꢀs prenyl transferases are enzymes of high regiospecificity [24], it is
tempting to speculate that a simple switch of the selectivity of the prenylation of
umbelliferone from C(6) to C(8) has initiated the evolutionary shift from linear to
angular furocoumarins. Probably due to a certain structural tolerance of the ancient
enzymes processing the compounds downstream of demethylsuberosin (9; Scheme 1),
also osthenol (11) may have been further processed to give angelicin (5), albeit with
low efficiency. However, subsequent evolutionary adaptation of the enzymes may have
created a new set of biocatalysts optimized to the molecular architecture of the angular
compounds. The current work supports such an evolutionarily driven development in as
much as the stereochemical and mechanistic aspects of the crucial oxidative deal-
kylation are identical in both pathways. However, further comparative studies on the
enzymes of these two pathways are urgently needed to establish more parallels between
the transformations and their biocatalysts.

1604
Helvetica Chimica Acta ± Vol. 81 (1998)
Experimental Part
General. Angelicin (5) was purchased from Roth, Karlsruhe, Germany. Reactions were performed under
Ar; solvents were dried according to standard procedures. Column chromatography (CC): silica gel Si60
(0.200 ± 0.063 mm; E. Merck, Darmstadt, Germany). TLC: Silica-gel plates Polygram Sil GF254, from E. Merck.
GLC: Carlo Erba, series 4100, equipped with fused silica capillaries coated with SE30 (15 m  0.31 mm) from
Macherey & Nagel (Düren, Germany). IR: Perkin-Elmer-1600-FTIR spectrophotometer, n in cm 1. 1H- and 13C-
NMR (1H-decoupled): Bruker-AC-250 or Bruker-AC-400 spectrometer; CDCl3 or (D6)acetone as solvent,
chemical shifts d in ppm downfield from SiMe4 (ˆ 0 ppm), J in Hz. GLC/MS (70 eV): Finnigan ITD 800 coupled
with a Carlo Erba GC 6000, model Vega or Fisons MD 800 GLC-MS system. HR-MS: Kratos MS 50.
2-Fluoro-4,6-dimethoxybenzaldehyde (17a) and 4-Fluoro-2,6-dimethoxybenzaldehyde (17b). POCl3
(1.8 ml, 19.3 mmol) was added slowly to a well-stirred mixture of 1-fluoro-3,5-dimethoxybenzene (2.6 ml,
19.25 mmol) and N-methylformanilid (2.5 ml, 20 mmol) while the temp. was kept below
58. Stirring was
continued at r.t. for 1.5 h and at 608 for another 2 h. Then, the chilled mixture was hydrolysed with ice-water
(60.0 ml), and the resulting suspension neutralized by addition of 5n aq. NaOH. Following extraction with
AcOEt (2 Â 30 ml), the aq. phase was adjusted to pH ca. 10 by 5n NaOH and reextracted with AcOEt (2 Â
30 ml). The combined org. phase was washed with sat. aq. NaHCO3 soln. (30 ml) and brine (10 ml), dried
(MgSO4), and evaporated. CC (silica gel, hexane/AcOEt 1:1) afforded a colourless solid consisting of 17a/b
(5.5:1) which were separated only for anal. purposes. Yield of 17a/b: 3.47 g, 98%.
Data of 17a. M.p. 86 ± 878. IR (KBr): 2861, 2784, 1679, 1628, 1572, 1478, 1457, 1410, 1344, 1311, 1208, 1157,
1095, 824, 802. 1H-NMR (CDCl3, 400 MHz): 10.26 (d, Jˆ 1.2, CHO); 6.25 (dd, Jˆ 12.8, 2.2, H C(3)); 6.25 (Jꢀ 2.0,
H
C(5)), 3.89 (s, 1 MeO); 3.85 (s, 1MeO). 13C-NMR (CDCl3, 100 MHz): 185.8 (d, J ˆ 3.1, CHO); 166.0 (d, J ˆ
15.2, C(4)); 165.2 (d, J ˆ 260.6, C(2)); 163.3 (d, J ˆ 8.2, C(6)); 108.1 (d, J ˆ 9.2, C(1)); 94.3 (d, J ˆ 3.6, C(5));
.
‡
94.0 (d, J ˆ 25.3, C(3)); 56.2 (s, MeO); 55.9 (s, MeO). MS (70 eV): 184(100, M ), 183(71), 169(10), 168(22),
167(56), 166(37), 155(13), 153(40), 140(20), 139(37), 138(34), 127(12), 124(25), 112(10), 110(13),
‡
97(13), 95(10), 82(12), 81(13), 63(20). HR-MS: 184.0535 (C9H9FO3 ; calc. 184.0535).
Methyl (E)-3-(2'-Fluoro-4',6'-dimethoxyphenyl)prop-2-enoate (18).
A stirred soln. of 17a/b (3.4 g,
18.5 mmol) in MeCN (200.0 ml) was treated with methyl(triphenylphosphoranylidene)acetate (8.0 g,
23.9 mmol) at r.t. The resulting suspension was kept at 55 ± 608 for 12 h. After cooling, AcOEt (200 ml) was
added and the org. layer washed with a sat. NaHCO3 soln. (100 ml). The aq. phase was reextracted with AcOEt
(2 Â 50 ml) and the combined org. extract dried (MgSO4) and evaporated. The regioisomer mixture was
submitted to CC (silica gel, hexane/AcOEt 2.5:1): 3.25 g (73%) of 18 as a colourless solid. M.p. 768. IR (KBr):
2951, 1711, 1613, 1431, 1314, 1293, 1194, 1170, 1090, 992, 858, 809. 1H-NMR (CDCl3, 400 MHz): 7.79 (d, J ˆ
16.2, H C(3)); 6.59 (d, J ˆ 16.3, H C(2)); 6.16 ± 6.20 (m, H C(3'), H C(5')); 3.79 (s, 1 MeO); 3.73 (s, 1 MeO);
3.71 (s, 1 MeO). 13C-NMR (CDCl3, 100 MHz): 168.6 (s, C(1)); 163.6 (d, J ˆ 250.0, C(2')); 162.4 (d, J ˆ 17.2,
C(4')); 160.6 (d, J ˆ 8.9, C(6')); 133.4 (d, J ˆ 3.3, C(3)); 119.0 (d, J ˆ 10.2, C(2)); 105.4 (d, J ˆ 13.6, C(1'));
94.5 (d, J ˆ 3.0, C(5')); 93.6 (d, J ˆ 27.4, C(3')); 55.9 (s, ArOMe); 55.6 (s, ArOMe); 51.5 (s, CO2Me). MS
.
‡
(70 eV): 240(63, M ), 209(100), 194(21), 180(9), 179(9), 167(29), 166(27), 151(16), 139(14). HR-MS:
‡
240.0800 (C12H13FO4 ; calc. 240.0798).
5-Fluoroumbelliferone (ˆ 5-Fluoro-7-hydroxy-2H-1-benzopyran-2-one; 19). A stirred soln. of 18 (3.1 g,
12.9 mmol) in CH2Cl2 (250 ml) at 228 was gradually treated with 1m BBr3 in hexane (110 ml). After 2 h, the
mixture was slowly allowed to come to r.t., and after 36 h at r.t., the red soln. was hydrolysed with ice-water
(100 ml). The aq. layer was extracted with AcOEt (3 Â 70 ml), the combined org. extract washed with sat. aq.
NaHCO3 soln. (50 ml) and brine, dried (MgSO4), and evaporated, and the residue submitted to CC (silica gel,
hexane/AcOEt 1:1): 1.88 g (81%) of 19. Colourless solid. M.p. 245 ± 2468. IR (KBr): 3191(br.), 1700, 1629,
1569, 1466, 1394, 1334, 1286, 1246, 1157, 1113, 1059, 899, 837, 705, 657. 1H-NMR ((D6)acetone, 400 MHz):
9.74 (br. s, OH); 7.81 (d, J ˆ 9.7, H C(4)); 6.47 ± 6.52 (m, H C(6), H C(8)); 6.10 (d, J ˆ 9.7, H C(3)). 13C-
NMR ((D6)acetone, 100 MHz): 162.5 (d, J ˆ 13.9, C(7)); 160.4 (d, J ˆ 251.0, C(5)); 160.4 (s, C(2));
157.2 (d, J ˆ 7.4, C(8a)); 137.2 (d, J ˆ 3.9, C(4)); 113.3 (s, C(3)); 102.8 (d, J ˆ 19.8, C(4a)); 100.1 (d, J ˆ 23.16,
.
‡
C(6)); 99.9 (d, J ˆ 3.0, C(8)). MS (70 eV): 180(82, M ), 152(100), 124(10), 123(17), 96(43), 95(13), 83(8),
‡
75(16), 69(11). HR-MS: 180.0231 (C9H5FO3 ; calc. 180.0223).
5-Fluoro-8-iodoumbelliferone (ˆ 5-Fluoro-7-hydroxy-8-iodo-2H-1-benzopyran-2-one; 15). A soln. of 19
(1.00 g, 5.56 mmol) in dioxan (9.0 ml) was treated with 20% aq. NH3 soln. (25.0 ml). Then, a soln. of I2 (1.6 g,
6.10 mmol) and KI (2.5 g, 15 mmol) in H2O (38.0 ml) was slowly added with stirring at 08. After 1 h, the chilled
mixture was slightly acidified with 2.5n H2SO4 and extracted with AcOEt (1 Â 80 ml, 2 Â 30 ml). The org. layer
was washed with sat. aq. NaHCO3 soln. (50 ml) and brine (20 ml), dried (MgSO4), and evaporated and the

Helvetica Chimica Acta ± Vol. 81 (1998)
1605
beige solid purified by CC (silica gel, hexane/AcOEt 1:1): 1.21 g (72%) of 15. M.p. 2208 (dec.). IR (KBr):
3200(br.), 1705, 1618, 1561, 1503, 1436, 1387, 1326, 1240, 1172, 1115, 1065, 929, 839, 823, 751, 712, 686, 656.
1H-NMR ((D6)acetone, 400 MHz): 10.55 (br. s, OH); 7.94 (d, J ˆ 9.5, H C(4)); 6.83 (d, J ˆ 10.6, H C(6));
6.29 (d, J ˆ 9.4, H C(3)). 13C-NMR ((D6)acetone, 100 MHz): 161.8 (d, J ˆ 13.4, C(7)); 160.7 (d, J ˆ 252.1,
C(5)); 160.1 (s, C(2)); 156.6 (d, J ˆ 8.3, C(8a)); 137.2 (d, J ˆ 3.7 C(4)); 113.8 (d, J ˆ 1.3, C(3)); 103.6 (d, J ˆ
.
‡
19.6, C(4a)); 99.1 (d, J ˆ 23.5, C(6)); 68.7 (d, J ˆ 3.9, C(8)). MS (70 eV): 306(100, M ), 278(57), 151(16),
‡
123(23), 95(12), 94(13), 75(14). HR-MS: 305.9195 (C9H4FIO3 ; calc. 305.9189).
2-Methyl(4-2H)but-3-yn-2-(2H)ol. A soln. of 2-methylbut-3-yn-2-ol (0.75 ml, 7.73 mmol) in Et2O (25.0 ml)
was slowly added with stirring to EtMgBr (22.5 mmol) in Et2O (22.5 ml). The mixture was refluxed for 1 h,
chilled, and hydrolysed with 2H2O (2.5 ml) and conc. 2HCl (1.0 ml). Extractive workup with Et2O and
evaporation afforded the crude 2-methyl(4-2H2)but-3-yn-2-(2H)ol which was used for the copper-catalysed
alkynylation/cyclization without further purification to avoid loss of the isotopes.
5-Fluoro-8-(1-hydroxy-1-methylethyl)(9-2H)angelicin (ˆ 5-Fluoro-8-(1-hydroxy-1-methylethyl)(9-2H)-2H-
furo[2,3-h]-1-benzopyran-2-one; 14). A suspension of 15 (0.75 g, 2.45 mmol) in MeO2H (15 ml) was treated
with one drop of conc. 2HCl soln. and refluxed until a clear soln. was obtained. Then, the solvent was evaporated
and the remaining solid dissolved in absolutely dry pyridine (12 ml). Then, 2-methyl(4-2H)but-3-yn-2-(2H)ol
2
(7.73 mmol), copper(I) oxide (0.30 g, 2.1 mmol), and H2O (0.10 ml) were added and refluxed for 2 h. After
cooling to r.t., the mixture was filtered through a small pad of silica gel, diluted with AcOEt (50 ml), and
acidified with 2n HCl, the org. phase washed with sat. aq. NaHCO3 soln. (50 ml) and brine (10 ml), dried
(MgSO4), and evaporated, and the residue submitted to CC (silica gel, pentane/AcOEt 1:1): 0.54 g (84%) of 14.
Faintly yellow solid. Deuterium incorporation: > 98%. M.p. 148 ± 1498. IR (KBr): 3488, 3056, 2984, 1711, 1628,
1570, 1465, 1371, 1330, 1290, 1259, 1196, 1121, 1018, 976, 934, 882, 835, 771. 1H-NMR (CDCl3, 400 MHz):
7.93 (d, J ˆ 9.8, H C(4)); 7.07 (d, J ˆ 8.9, H C(6)); 6.33 (d, J ˆ 9.8, H C(3)); 2.26 (br. s, OH); 1.63 (s, 2 Me).
13C-NMR (CDCl3, 100 MHz): 164.4 (d, J ˆ 3.5, C(8)); 160.1 (s, C(2)); 156.1 (d, J ˆ 250.1, C(5)); 156.0 (d, J ˆ
15.5, C(6 a)); 147.6 (d, J ˆ 7.1, C(9 b)); 137.7 (d, J ˆ 4.5, C(4)); 114.0 (d, J ˆ 1.3, C(3)); 113.7 (d, J ˆ 2.9, C(9 a));
.
‡
104.5 (d, J ˆ 21.0, C(4a)); 95.7 (d, J ˆ 26.7, C(6)); 69.1 (s, Me2C); 28.6 (s, Me2C). MS (70 eV): 263(26, M ),
248(100), 246(6), 220(9), 206(17), 150(6). HR-MS: 263.0711 (C14H102HFO4 ; calc. 263.0704).
‡
(Æ)-cis-5-Fluoro-3,4,8,9-tetrahydro-8-(1-hydroxy-1-methylethyl)(9-2H)angelicin (ˆ (Æ)-cis-5-Fluoro-3,4,8,9-
tetrahydro-8-(1-hydroxy-1-methylethyl)(9-2H)-2H-furo[2,3-h]-1-benzopyran-2-one; 20). A suspension of 14
(0.50 g, 1.90 mmol) and 10% Pd/C (500.0 mg) in acetone (15 ml) was stirred for 4 h at r.t. under H2. Evaporation
and chromatography (silica gel, pentane/AcOEt 1:1) yielded pure 20 (0.38 g, 72%). Colourless solid. M.p. 1318.
IR (KBr): 3515, 3010, 2926, 1754, 1645, 1616, 1488, 1366, 1256, 1237, 1177, 1145, 1092, 1057, 1008, 974, 907, 834,
755, 635. 1H-NMR (CDCl3, 400 MHz): 6.28 (d, J ˆ 9.1,
H
C(6)); 4.62 (d, J ˆ 9.5, H C(8)); 3.08 (d, J ˆ
9.5, H C(9)); 2.87 (ꢁtꢀ, Jˆ 7.0, 2 H C(4)); 2.70 (ꢁtꢀ, Jˆ 7.1, 2H C(3)); 1.78 (br. s, OH); 1.27 (s, 1 Me);
1.14 (s, 1 Me). 13C-NMR (CDCl3, 100 MHz): 167.6 (s, C(2)); 160.2 (d, J ˆ 14.2, C(6a)); 159.4 (d, J ˆ
243.4, C(5)); 148.5 (d, J ˆ 10.2, C(9b)); 110.1 (d, J ˆ 3.4, C(9a)); 102.1 (d, J ˆ 24.5, C(4a)); 93.8 (d, J ˆ 27.8,
C(6)); 91.0 (s, C(8)); 71.8 (s, Me2C); 28.6 (s, C(3)); 27.3 (t, J ˆ 20.7, C(9)); 26.1 (s, Me); 23.9 (s, Me);
.
‡
17.0 (d, J ˆ 3.2, C(4)). MS (70 eV): 267(29, M ), 233(12), 209(70), 208(29), 194(7), 191(7), 181(10),
167(31), 166(23), 154(11), 138(8), 110(10), 59(100), 55(9). HR-MS: 267.1015 (C14H142HFO4 ; calc. 267.1017.
‡
(Æ)-cis-5-Fluoro(9-2H)columbianetin (ˆ(Æ)-cis-5-Fluoro-8,9-dihydro-8-(1-hydroxy-1-methylethyl)(9-2H)-
2H-furo[2,3-h]-1-benzofuran-2-one; 13): A soln. of 20 (0.38 g, 1.41 mmol) in THF (10 ml) was added to a well-
stirred, cold ( 788) LDA soln. in THF (20 ml, 4.20 mmol), followed by an immediate and rapid addition of a
phenylselenenyl chloride (0.804 g, 4.20 mmol) soln. in THF (10 ml). After stirring at 788 for 1 h, the mixture
was allowed to come to 08, and an AcOH (1.5 ml) soln. in H2O (5 ml) was added, followed by slow addition of a
30% H2O2 soln. (5.0 ml). Stirring was continued for 30 min at 08 and 30 min at r.t., followed by extraction with
AcOEt (3 Â 30 ml). Hexane (50.0 ml) was added to the combined org. extract, the resulting soln. washed with
sat. aq. NaHCO3 (30 ml) and NaCl soln. (20 ml), dried (MgSO4), and evaporated, and the crude solid purified
by CC (silica gel, hexane/AcOEt 1:1): 0.194 g (52%) of 13. M.p. 1718. IR (KBr): 3448 (br.), 3072, 2970, 1718,
1627, 1467, 1398, 1357, 1292, 1267, 1171, 1120, 1074, 1034, 1008, 955, 906, 851, 827, 758. 1H-NMR (CDCl3,
400 MHz): 7.75 (d, J ˆ 9.8, H C(4)); 6.43 (d, J ˆ 9.7, H C(6)); 6.14 (d, J ˆ 9.7, H C(3)); 4.74 (d, J ˆ 9.6,
H
C(8)); 3.21 (d, J ˆ 9.7, H C(9)); 1.80 (br. s, 1 OH); 1.29 (s, 1 Me); 1.17 (s, 1 Me). 13C-NMR (CDCl3,
100 MHz): 163.7 (d, J ˆ 14.6, C(6a)); 160.4 (s, C(2)); 159.4 (d, J ˆ 253.1, C(5)); 150.8 (d, J ˆ 7.6, C(9b));
137.2 (d, J ˆ 4.0, C(4)); 112.1 (d, J ˆ 1.6, C(3)); 109.6 (d, J ˆ 3.0, C(9a)); 102.9 (d, J ˆ 20.2, C(4a));
94.4 (d, J ˆ 25.3, C(6)); 92.1 (s, C(8)); 71.8 (s, Me2C); 26.9 (t, J ˆ 20.5, C(9)); 26.0 (s, Me); 24.1 (s, Me). MS
.
‡
(70 eV): 265(41, M ), 231(13), 209(14), 208(17), 207(76), 206(66), 205(18), 195(10), 194(15), 179(32),
178(14), 150(20), 102(9), 59(100), 43(17). HR-MS: 265.0855 (C14H122HFO4 ; calc. 265.0861).
‡

1606
Helvetica Chimica Acta ± Vol. 81 (1998)
8-(Ethoxycarbonyl)-5-fluoroangelicin (ˆ Ethyl 5-Fluoro-2-oxo-2H-furo[2,3-h]-1-benzofuran-8-carboxy-
late; 22). Ethyl propiolate (0.4 ml, 3.94 mmol) and a soln. of 15 (0.27 g, 0.88 mmol) in dry DMF (4.0 ml)
were added to copper(I) oxide (0.15 g, 1.05 mmol) suspended in dry DMF (4.0 ml). The mixture was heated to
1108 for 24 h, then filtered through a small pad of silica gel, and diluted with AcOEt (20 ml). The org. soln. was
washed with 1n HCl, sat. aq. NaHCO3 soln. (50 ml) and brine, dried (MgSO4), and evaporated. The residue was
submitted to CC (silica gel, hexane/AcOEt 1:1): pure 22 (0.17 g, 70%). Faintly yellow solid. IR (KBr): 3054,
2985, 1729, 1621, 1563, 1473, 1368, 1290, 1207, 1143, 1014, 864, 847, 760, 642. 1H-NMR (CDCl3, 400 MHz):
7.95 (d, J ˆ 9.8, H C(4)); 7.71 (d, J ˆ 0.7, H C(9)); 7.18 (dd, J1 ˆ 8.9, 0.8, H C(6)); 6.40 (d, J ˆ 9.8, H C(3));
4.39 (q, J ˆ 7.1, CH2); 1.37 (t, J ˆ 7.1, Me). 13C-NMR (CDCl3, 100 MHz): 159.2, 158.5 (2s, C(2), CO2Et);
158.2 (d, J ˆ 253.8, C(5)); 156.6 (d, J ˆ 15.3, C(6a)); 149.1 (d, J ˆ 7.7, C(9b)); 147.0 (d, J ˆ 4.4, C(8));
137.3 (d, J ˆ 5.7, C(4)); 115.0 (s, C(3)); 113.3 (d, J ˆ 2.4, C(9a)); 110.5 (d, J ˆ 1.5, C(9)); 105.5 (d, J ˆ 21.8,
.
‡
C(4a)); 96.2 (d, J ˆ 25.8, C(6)); 62.0 (s, MeCH2); 14.3 (s, MeCH2). MS (70 eV): 276(100, M ), 248(23),
‡
231(30), 220(48), 204(18), 176(12), 175(10), 147(15), 119(11), 99(10). HR-MS: 276.0435 (C14H9FO5 , calc.
276.0434).
5-Fluoroangelicin (ˆ 5-Fluoro-2H-furo[2,3-h]-1-benzofuran-2-one; 23).
A suspension of 22 (0.13 g,
0.47 mmol) in 20% aq. NaOH soln. (15.0 ml) was refluxed for 2 h. After cooling and acidification with conc.
HCl soln., the mixture was extracted with AcOEt (1 Â 100 ml, 2 Â 50 ml). The combined org. layer washed with
sat. aq. NaHCO3 soln. (40 ml), dried, and evaporated and the remaining solid taken up in quinoline (6.0 ml).
Copper powder (20.0 mg, 0.31 mmol) was added and the suspension heated to 2108 for 1 h. After cooling, cold
1n HCl (50.0 ml) was added and the aq. layer extracted with AcOEt. The org. layer was washed with sat. aq.
NaHCO3 soln. (20 ml) and brine (10 ml), dried (MgSO4), and evaporated. CC (silica gel, hexane/AcOEt 3:2)
yielded 23 (40.0 mg, 43%). Colourless solid. M.p. 1558. IR (KBr): 3080, 3066, 1737, 1631, 1533, 1470, 1370, 1309,
1255, 1189, 1167, 1141, 1113, 1085, 1074, 990, 829, 749, 630. 1H-NMR (CDCl3, 400 MHz): 7.95 (d, J ˆ 9.9,
H
C(4)); 7.60 (d, J ˆ 2.2, H C(8)); 7.11 (dd, J1 ˆ 9.1, 1.0, H C(6)); 7.01 (dd, J ˆ 2.2, 0.8, H C(9)); 6.35 (d, J ˆ
9.7, H C(3)). 13C-NMR (CDCl3, 100 MHz): 160.0 (s, C(2)); 156.4 (d, J ˆ 249.9, C(5)); 156.3 (d, J ˆ 15.9,
C(6 a)); 147.9 (d, J ˆ 7.3, C(9b)); 146.0 (d, J ˆ 3.8, C(8)); 137.7 (d, J ˆ 4.5, C(4)); 114.1 (d, J ˆ 1.4, C(3));
113.3 (d, J ˆ 2.0, C(9a)); 104.6 (d, J ˆ 20.7, C(4a)); 104.0 (d, J ˆ 1.72, C(9)); 95.8 (d, J ˆ 26.3, C(6)). MS
.
‡
‡
(70 eV): 204(100, M ), 176(71), 148(20), 120(27). HR-MS: 204. 0221 (C11H5FO3 ; calc. 204.0223).
Induction Experiments Plants: Young plants of Heracleum mantegazzianum were bought from a garden
centre (Hof Berg-Garten, Lindenweg 17, D-79737 Herrischried) and grown in 49-cm2 diameter pots filled with
potting compost. The plants were kept at r.t. and illuminated with daylight fluorescent tubes (ca. 270 mEs 1 m
,
2
using a regime of 14 h light and 10 h dark period). Plants used for induction experiments were about two- or
three-month-old.
Application of the Test Compounds: A soln. of cis-5-fluoro(9-2H)columbianetin (13; 0.8 mg/ml), Triton X
100 (ca. 4 mg per ml), and jasmonic acid (0.2 mg per ml) were sonicated (ca. 5 ± 10 min) until a stable emulsion
resulted. The emulsified test soln. was stable for the time of the induction experiment (5 days) without noticeable
decomposition. Freshly detached leaves were immediately placed into the emulsions. About 50% of the directly
treated plants were lost during the first 3 days due to shrinking of the stem and subsequent necrotization of the
leaf. No 5-fluoroangelicin 23 was found among the surface lipids of the necrotized plant leaves.
Surface Treatment of Leaves: The above emulsion containing cis-5-fluoro(9-2H)columbianetin (13; 1 mg
per ml), Triton X 100 (ca. 4 mg per ml), and jasmonic acid (0.2 mg per ml) was repeatedly sprayed onto the
leaves of the intact plants (four-days intervals). This treatment had no visible impact onto the plants. About
three weeks after the beginning of the treatment, the leaves were detached and their surface lipids were analysed
for metabolites.
Collection of the Surface Lipids: The nonpolar surface compounds were obtained by brief dipping (2 Â 30 s)
of the pretreated, but undamaged leaves into CH2Cl2 (ca. 10 ml) [25]. After filtration and controlled
evaporation at ca. 300 Torr; the residue was dissolved in AcOEt (100 ± 200 ml). The resulting soln. was used for
GLC/MS without further purification.
Mass Spectroscopic Analysis of Surface Compounds: The solns. containing the surface lipids of the
pretreated plants or from the leaves of untreated control plants were directly analysed by GLC/MS by injecting
(1 ml) onto a fused silica column coated with DB 1 (15 m  0.25 mm, 0.25 mm). Compounds were separated
under programmed conditions (508 for 2 min, then to 2808 at 208/min). Transfer line: 2508. Scan range: 35 ±
350 Da/s. Authentic references were used for the unequivocal identification of all furocoumarins and the
fluorinated metabolites.
Biomimetic Dealkylation of cis-5-Fluoro(9-2H)columbianetin (13): A soln. of (tetraphenylporphyrinato)-
manganese(III) (1.4 mmol), 5-fluoro(9-2H)columbianetin (13; 8.1 mmol) and iodosylbenzene (22.0 mmol) in

Helvetica Chimica Acta ± Vol. 81 (1998)
1607
CH2Cl2 (3.0 ml) was stirred at 48 for 8 h. The resulting products were analysed by GLC/MS as above. The yield
of 5-fluoroangelicin 23 was ca. 1 ± 3%. Due to the low conversion, no attempts were made to trap acetone as the
second fragment, as has been done previously for the comparable transformation of (‡)-marmesin (10) to
psoralen (1).
Financial support by the Deutsche Forschungsgemeinschaft, Bonn, and the Fonds der Chemischen Industrie,
Frankfurt, is gratefully acknowledged. We also thank the BASF AG, Ludwigshafen, and the Bayer AG,
Leverkusen, for generous supply of chemicals and solvents. The authors thank Mrs. Katja Bandemer for her
assistance.
REFERENCES
[1] R. A. Creelman, J. E. Mullet, Annu. Rev. Plant. Mol. Biol. 1997, 48, 355.
Â
[2] R. D. H. Murray, J. Mendes, S. A. Brown, in ꢁThe Natural Coumarins: Occurrence, Chemistry and
Biochemistryꢀ, Wiley and Sons, Chichester ± New York, 1982.
[3] M. R. Berenbaum, A. R. Zangerl, Recent Adv. Phytochem. 1996, 30, 1.
[4] A. R. Zangerl, M. R. Berenbaum, Ecology 1990, 71, 1933.
[5] K. D. Hauffe, K. Hahlbrock, D. Scheel, Z. Naturforsch., C 1986, 41, 228.
[6] R. C. Beier, E. H. Oertli, Phytochemistry 1983, 22, 2595.
[7] M. Miksch, W. Boland, Experientia 1996, 52, 739.
[8] V. Stanjek, C. Herhaus, U. Ritgen, W. Boland, E. Städler, Helv. Chim. Acta 1990, 80, 1408.
[9] L. D. Scheel, V. B. Perone, R. L. Larkin, R. E. Kupel, Biochemistry 1963, 2, 1127.
[10] L. Musajo, G. Rodighiero, G. Columbo, V. Torlone, F. DellꢀAcqua, Experientia 1967, 23, 335.
[11] M. M. Diawara, J. T. Trumble, K. K. White, W. G. Carson, L. A. Martinez, J. Chem. Ecol. 1993, 19, 2473.
[12] M. R. Berenbaum, P. Feeney, Science (Washington, D.C.) 1981, 212, 927.
[13] M. R. Berenbaum, A. R. Zangerl, Oecologia 1993, 95, 370.
[14] M. R. Berenbaum, A. R. Zangerl, Evolution 1992, 46, 1373.
[15] U. Matern, P. Lüer, D. Kreusch, in ꢁComprehensive Natural Products Chemistryꢀ, Ed. U. Sankawa, Elsevier,
Oxford, in press.
Â
[16] A. Estevez-Braun, A. G. Gonzales, Nat. Prod. Rep. 1997, 14, 465.
[17] H. Wendorff, U. Matern, Eur. J. Biochem. 1986, 161, 391.
[18] V. Stanjek, M. Miksch, U. Matern, P. Lüer, W. Boland, Angew. Chem., 1998, in press.
[19] R. A. W. Johnstone, A. H. Wilby, I. D. Entwistle, Chem. Rev. 1985, 85, 129.
[20] V. Stanjek, M. Miksch, W. Boland, Tetrahedron 1997, 53, 17699.
[21] E. Zubía, F. R. Luis, G. M. Massanet, I. G. Collado, Tetrahedron 1992, 48, 4239.
[22] H. J. Reich, Acc. Chem. Res. 1979, 12, 22.
[23] T. J. McMurry, J. T. Groves, in ꢁCytochrome P450: Structure, Mechanism and Biochemistryꢀ, Ed. P. R. Ortiz
de Montellano, Plenum Press, New York, 1986.
[24] D. S. Dillon, S. A. Brown, Arch. Biochem. Biophys. 1976, 177, 74.
[25] E. Städler, H. R. Buser, Experientia 1984, 40, 1157.
Received May 6, 1998