Biosynthesis of iridoids lacking C-10 and the chemotaxonomic implications of their distribution

Article abstract of DOI:10.1016/S0031-9422(99)00244-7

Continuing earlier experiments in which deoxyloganic acid proved to be a precursor of the doubly decarboxylated iridoids in Thunbergia alata, we have now shown that deuterium labelled 6-deoxyretzioside was incorporated into stilbericoside in this plant. In the same experiment labelled retzioside was isolated, indicating that this is also a likely intermediate in the biosynthetic pathway. Feedings in two species of Deutzia were also performed with deuterium-labelled iridoids, namely iridodial glucoside, 8-epiiridodial glucoside, 7-hydroxy-iridodial glucoside and 10-hydroxy-iridodial glucoside (= decapetaloside). Of these, only the first and the last were incorporated. Apparently, loss of C-10 in the biosynthesis of scabroside is analogous to that of stilbericoside and the sequence of events after formation of the 7,8- double bond appears in both cases to be the same as that found for the biosynthesis of antirrhinoside. The close relationship between Nuxia, Retzia and the Stilbaceae indicated by chloroplast DNA sequencing results is substantiated by the common occurrence of a special kind of iridoids in these taxa. A similar close relationship is found between Hydrangeaceae and Loasaceae.

Full text of DOI:10.1016/S0031-9422(99)00244-7

Phytochemistry 52 (1999) 1409±1420
Biosynthesis of iridoids lacking C-10 and the chemotaxonomic
implications of their distribution
Lotte Boe Frederiksen, Sùren Damtoft, Sùren Rosendal Jensen*
Department of Organic Chemistry, The Technical University of Denmark, 2800 Lyngby, Denmark
Received 25 March 1999; received in revised form 29 April 1999; accepted 3 May 1999
Abstract
Continuing earlier experiments in which deoxyloganic acid proved to be a precursor of the doubly decarboxylated iridoids in
Thunbergia alata, we have now shown that deuterium labelled 6-deoxyretzioside was incorporated into stilbericoside in this
plant. In the same experiment labelled retzioside was isolated, indicating that this is also a likely intermediate in the biosynthetic
pathway. Feedings in two species of Deutzia were also performed with deuterium-labelled iridoids, namely iridodidal glucoside,
8-epiiridodial glucoside, 7-hydroxy-iridodial glucoside and 10-hydroxy-iridodial glucoside (=decapetaloside). Of these, only the
®rst and the last were incorporated. Apparently, loss of C-10 in the biosynthesis of scabroside is analogous to that of
stilbericoside and the sequence of events after formation of the 7,8-double bond appears in both cases to be the same as that
found for the biosynthesis of antirrhinoside. The close relationship between Nuxia, Retzia and the Stilbaceae indicated by
chloroplast DNA sequencing results is substantiated by the common occurrence of a special kind of iridoids in these taxa. A
similar close relationship is found between Hydrangeaceae and Loasaceae. # 1999 Elsevier Science Ltd. All rights reserved.
Keywords: Thunbergia; Acanthaceae; Deutzia; Hydrangeaceae; Iridoid glucosides; Stilbericoside; Deutzioside; Biosynthesis; Chemotaxonomy;
Loasaceae; Nuxia; Retzia; Stilbaceae
1. Introduction
Buddlejaceae
(Nuxia )
(Jensen,
Ravnkilde
&
Schripsema, 1998), in Acanthaceae (Thunbergia)
(Jensen & Nielsen, 1989; Jensen, Jensen & Nielsen,
1988), Ericaceae (Arbutus and Arctostaphylos )
(Geissman, Knaack & Knight, 1966; Jahodar, Kolb &
Leifertova, 1981), in Hydrangeaceae (Deutzia )
(Bonadies, Esposito & Guiso, 1974; Esposito & Guiso,
1973), and in Loasaceae (Mentzelia) (Danielson &
Hawes, 1973; El-Naggar, Beal & Doskotch, 1982;
Jensen, Mikkelsen & Nielsen, 1981). The ®rst four taxa
all belong to Lamiales and the iridoids from these as
well as those from Ericaceae are characterized by the
lack of both C-10 and C-11, e.g. unedoside (1) and stil-
bericoside (2). In the following, these compounds are
designated C-8 iridoids, as they only have eight carbon
atoms in the aglucone. The iridoids from the genera
Deutzia and Mentzelia are di€erent since they always
have C-11 present as a methyl or an oxymethylene sub-
stituent, examples of compounds found in Deutzia are
deutzioside (3) and scabroside (4).
Iridoid compounds lacking C-11, i.e. aucubin and
congeners, are very common in genera from the families
within the order Lamiales sensu Dahlgren (1989) and
they have been shown to constitute a valuable taxo-
nomic character to delineate these families from those
belonging to Cornales and Gentianales which mainly
contain secoiridoids (Jensen, 1991, 1992). Conversely,
iridoids lacking C-10 have been found in only a limited
number of families of the dicotyledons, namely in
Stilbaceae (Stilbe and Xeroplana ) (Dahlgren, Nielsen,
Goldblatt & Rourke, 1979; Rimpler & Pistor, 1974), in
Retziaceae (Retzia capensis ) (Dahlgren et al., 1979;
Damtoft, Franzyk, Jensen
& Nielsen, 1993c), in
* Corresponding author. Tel.: +45-452-521-03; fax: +45-459-339-
68.
E-mail address: oksrj@pop.dtu.dk (S.R. Jensen).
0031-9422/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved.
PII: S0031-9422(99)00244-7

1410
L.B. Frederiksen et al. / Phytochemistry 52 (1999) 1409±1420

L.B. Frederiksen et al. / Phytochemistry 52 (1999) 1409±1420
1411
The biosynthesis of the aucubin type of iridoid glu-
cosides is well known (cf. Jensen, 1991), proceeding
through a pathway including the key compounds 8-
epiiridodial (5) and epideoxyloganic acid (6), with C-
11 being lost by decarboxylation of an a,b-unsaturated
acid (Damtoft, Jensen, Jessen & Knudsen, 1993b).
Compounds derived from such a pathway including 5
and 6 are widespread in Lamiales (Jensen, 1991, 1992).
Surprisingly, our initial results with Thunbergia
(Damtoft, Frederiksen & Jensen, 1994), and more
recently with Nuxia (Jensen et al., 1998) on the biosyn-
thesis of unedoside (1) type iridoids which lack both
C-10 and C-11, have shown that these compounds are
biosynthesized through deoxyloganic acid (7), and not
from epideoxyloganic acid (6). The latter had earlier
been the surmised intermediate (Jensen, 1992) due to
lack of speci®c evidence. Furthermore, administration
of deuterium labelled 7 to T. alata indicated that the
loss of C-10 does not involve decarboxylation of an
a,b-unsaturated acid since labels at both H-7 and H-8
were retained in the isolated stilbericoside (2) (Damtoft
et al., 1994). To further investigate the biosynthetic
sequence after the decarboxylation of C-10 in stilberi-
coside (2), deuterium labelled 6-deoxy-retzioside (8)
has been synthesized and tested as a precursor in T.
alata.
Regarding the compounds found in Deutzia, some
experiments have been performed with D. crenata.
Thus, Inouye, Ueda and Uesato (1977) showed that
[2-14C]-MVA was incorporated into deutzioside (3)
and scabroside (4) without scrambling. Also, tritium
labelled 10-hydroxygeraniol, iridodial and iridodial
glucoside (9) have been shown (Inouye, Ueda, Uesato
& Kobayashi, 1978) to be precursors for 3. Finally,
parallel administrations (Uesato, Miyauchi, Itoh &
Inouye, 1986) of iridodial glucoside (9) and 8-epiirido-
dial glucoside (10) indicated a lack of stereospeci®city
since both were incorporated into 3 although 9 gave a
much better incorporation than 10. However, both
precursors were probably contaminated with their 8-
epimer, but in di€erent ratios, which somewhat
obscured the results (Jensen, 1991). In order to fully
clarify this and to gain further knowledge of the bio-
synthesis of the Deutzia iridoids, we have tested some
possible precursors.
was accomplished starting from gardenoside (11)
obtained from Gardenia jasminoides. Selective oxi-
dation with sodium periodate cleaved only the C8±C10
bond (cf. Damtoft, Jensen & Nielsen, 1993a) of the
aglucone to give randioside (12) which was protected
as the trimethylsilyl ether (12s)1 and then reduced with
sodium borohydride to give, selectively, the 8a-
hydroxy compound (13s). In order to avoid an E2
elimination which would presumably produce an 8,9-
double bond, a Chugaev reaction (Depuy & King,
1960) was attempted. Conversion of the free hydroxy
group to the methylxanthate was performed using the
phase transfer procedure by Lee, Chan, Wong and
Wong (1989). Surprisingly, the primary TMS group at
C-6' was unstable under these conditions since an ad-
ditional methyl xanthate group was introduced also at
this position in the product (13sx) as seen by NMR,
mainly by the low ®eld position of C-6' (d 72.5).
Pyrolysis in quinoline using the procedure of Bordwell
and Landis (1958) gave the 10-nor-geniposide deriva-
tive 14sx methyl ester as a crude product. Desilylation
and saponi®cation provided 10-nor-geniposidic acid
(14) in only 12% yield from 13sx. Acetylation to 14a
followed by a copper-catalyzed decarboxylation in qui-
noline (Murai & Tagawa, 1980) yielded the tetraace-
tate 8a in 21% yield. Finally, deacetylation provided 8
in an overall yield of 0.8% from 11. Labelling was
introduced selectively into the 6b- and 8b-positions by
performing the reduction of randioside tetra-O-TMS
ether (12s) with sodium borodeuteride. The resulting
labelled 6,8-[2H2]-retzioside (d-8) contained more than
2
1
0.9 H at both C-6 and C-8 as measured by H NMR.
The yields in this sequence were somewhat better than
those above, namely 1.8% from 11.
Our previous syntheses of iridodial glucoside (9)
(Jensen, Kirk & Nielsen, 1989) and of 8-epiiridodial
glucoside (10) (Breinholt, Damtoft, Demuth, Jensen &
Nielsen, 1992) have shown that both these compounds
are more or less contaminated (ca. 15 and 2±5%, re-
spectively) with the 8-epimer, when synthesized from
11 or from geniposide (15). In our hands, these com-
pounds co-crystallize and no signi®cant puri®cation is
obtained by repeated recrystallisation. In order to
secure an epimer-free preparation of deuterium
labelled 9, we in this case treated 5-deoxystansioside
(iridotrial
glucoside),
readily
obtained
from
Chaenostomum foetidum (Jensen et al., 1989), with
sodium borodeuteride. Transformation of the resulting
labelled 11-OH-iridodial glucoside to selectively
labelled d-9 as earlier described (Jensen et al., 1989)
provided the desired epimer-free preparation.
Conversely, we were able to purify the above pre-
viously prepared multiply labelled d-10 by chromatog-
raphy on an ecient RP-18 column (see
Experimental).
2. Results
2.1. Synthesis of labelled precursors
The synthesis (Scheme 1) of 6-deoxyretzioside (8)
1 The following suxes are used: a: acetate; s: trimethylsilylether;
x: xanthate. Deuterated compounds have the pre®x: d-.
Labelled 7-hydroxy-iridodial glucoside (d-16) was

1412
L.B. Frederiksen et al. / Phytochemistry 52 (1999) 1409±1420
Table 1
Results of administrations to T. alata
acid gave the desired multiply labelled decapetaloside
(d-23).
Deoxyloganic acid
(d-7)
6-Deoxyretzioside
(d-8)
2.2. Feeding experiments
amount
(mg)
incorp.
(%)
amount
(mg)
incorp.
(%)
In the biosynthetic pathway to the compounds
found in Thunbergia earlier proposed (Damtoft et al.,
1994), we suggested 6-deoxy-retzioside (8) as a likely
intermediate. The labelled analogue d-8 was therefore
fed to leaves of young T. alata plants. In addition,
deuterium labelled deoxyloganic acid (d-7) was admi-
nistered in a control experiment and the results are
presented in Table 1. Consistent with the proposed
pathway, both compounds were incorporated into stil-
bericoside (2) and the higher incorporation of d-8
(4%) than of d-7 (3%) is consistent with 8 being a
probable later intermediate. A trace of unlabelled thu-
naloside (24) as well as labelled retzioside (25) was
also isolated from the feeding experiment with 6-deox-
yretzioside (d-8); in the 1H NMR spectrum of 25, a
signal from H-8 (d 6.0) was absent and it was obvious
that no signi®cant dilution with unlabelled material
had taken place. This iridoid glucoside has so far not
been isolated from Thunbergia species, but it has been
found in Retzia capensis (Damtoft et al., 1993c) which
also contains the 10,11-decarboxylated iridoids includ-
ing 2. No incorporation into 6-epi-stilbericoside was
seen in either case, but when feeding with d-7 a small
fraction containing labelled thunaloside (24) in admix-
ture with tryptophan (1:2) was isolated. The label (at
C-10) in 24 was not diluted to any visible degree, but
due to the small amount and to the presence of trypto-
phan in the fraction, it was not possible to determine
this with certainty.
Precursor fed
Crude extract
Sugars
43
180
106
10
10
2
20
190
104
8
0
0
0
0
4
0
3
6
Epistilbericoside
Stilbericoside (2)
Thunaloside (24)a
Retzioside (25)a
MeOH-eluateb
3
9
1
trace
3.5
ca. 10
±
±
14
9
a The fraction was contaminated with tryptophan.
b Fraction containing recovered precursor.
prepared (Scheme 1) starting from loganin (17).
Protection to give the trimethylsilyl ether, followed by
reduction of the carboxyl ester with diisobutyl alu-
minium hydride and deprotection, gave 7,11-dihy-
droxy-iridodial glucoside (18). Deuterium label was
introduced at C-11 by transfer hydrogenation of the
hexaacetate (18a) using Pd/C and dideutero formic
acid. Deacetylation gave selectively labelled d-16.
Palladium catalyzed reduction of geniposide (15)
with deuterium gas and added triethylamine gave rise
to seemingly unpredictable mixtures (5:1 to 1:3) of
multiply labelled adoxoside (d-19) and the 8-epimer
(d-20), which could be separated only with diculty
(Damtoft et al., 1994). However, since separation of
the epimers proved possible at a later stage, such a
mixture (2:3) was acetylated and reduced with diisobu-
tyl aluminium hydride. From this experiment, both the
labelled 11-hydroxy-decapetaloside (d-21) and its epi-
mer d-22 could be isolated. Finally, transfer hydrogen-
ation of the hexa-acetate d-21a using Pd/C and formic
The earlier reported apparent lack of stereospeci®-
city in the biosynthesis of the Deutzia iridoids (Uesato
et al., 1986) was investigated by feeding labelled, epi-
merically pure iridodial glucoside (d-9) and 8-epiirido-
dial glucoside (d-10) to young shoots of D.
Table 2
Results of administrations to Deutzia species
Plant
D. schneideriana
D. scabra
d-9
d-10
d-16
d-23
mg
incorp. (%)
mg
incorp. (%)
mg
incorp. (%)
mg
incorp. (%)
Precursor fed
Crude extract
Sugars
19
18
1600
800
30
1030
550
31
28
1070
450
33
1200
640
300
80
Scabroside (4)
Deutzioside (3)
MeOH-eluateb
3
5
320
100
0
0
0
0
1a
4a
6
55
160
58
165
100
33
150
16
10
a Incorporations calculated are minimum values assuming that all 2H-7 and 2H-8 from precursor are retained.
b Fraction containing recovered precursor.

L.B. Frederiksen et al. / Phytochemistry 52 (1999) 1409±1420
1413
schneideriana in June. The results (Table 2) clearly
show that only iridodial glucoside (9) is a precursor
for deutzioside (3) and scabroside (4).
scabroside (4), assuming it to be analogous to the
pathways discussed above (Scheme 1) after the for-
mation of the 7,8-double bond. Support for such a
pathway in Deutzia and in Mentzelia (Loasaceae) is
found in the actual isolation of the hypothetical inter-
mediates decapetaloside (23) and decaloside (28) from
Mentzelia decapetala (El-Naggar et al., 1982).
The next experiments were performed with growing
shoots of D. scabra in late August when the other
species seemed too mature to exhibit full biosynthetic
activity. When feeding 7-hydroxy-iridodial glucoside
and decapetaloside (d-16 and d-23, respectively), only
the latter was incorporated into deutzioside (3) and
scabroside (4) (4 and 1%, respectively; see Table 2).
The incorporations are minimum values since they
3.2. Chemotaxonomy
The common presence of C8 iridoids in Nuxia,
Retzia, Stilbaceae and Thunbergia, suggests a common
biosynthetic pathway and thus indicates the possibility
of a phylogenetic relationship which will be discussed
in the following. Nuxia and Retzia have traditionally
been included in Loganiaceae in the tribe Buddlejeae
(Leeuwenberg & Leenhouts, 1980) while Stilbaceae
was thought to be closely related to Verbenaceae
(Cronquist, 1988). The presence of C8 iridoids in both
Retzia and Stilbaceae, however, led Dahlgren et al.
(1979) to propose a close anity between Retzia and
Stilbaceae, not only because of the iridoids present,
but also for morphological and other reasons. This
view was later corroborated by the embryological in-
vestigation by Engell (1987). Most contemporary taxo-
nomists have raised the tribe Buddlejeae (including
Nuxia) to family rank and placed Buddlejaceae close
to Scrophulariaceae (Cronquist, 1988; Dahlgren, 1989;
Thorne, 1992). Very recent results based on chloro-
plast DNA sequencing (Oxelman, Backlund & Bremer,
1999) implicates that Nuxia, Retzia and Stilbaceae are
more related to each other than to any other taxa. The
sequencing results also indicate a much closer relation-
ship between this assembly and Scrophulariaceae/
Buddlejaceae than with Loganiaceae. This is comple-
tely in concert with the chemical pro®les, where
Loganiaceae are characterized by loganin and secoiri-
doids combined with lack of verbascoside and other
ca€eoyl phenylethanoid glucosides (CPG's), while
Scrophulariaceae and Buddlejaceae usually contain
decarboxylated iridoids and CPG's, a combination
also found in Nuxia, Retzia and Stilbaceae (Jensen,
1992). The relationship between these three taxa and
Thunbergia is less striking since the latter appears to
be a true member of Acanthaceae (Oxelman et al.,
1999; Scotland, Sweere, Reeves & Olmstead, 1995), a
family placed in the same order. The signi®cance of
the C8 iridoids reported from members of the
Ericaceae is dicult to assess, since this family is taxo-
nomically unrelated to the taxa discussed above.
2
2
were calculated assuming that both H-7 and H-8 are
fully retained.
3. Discussion
3.1. Biosynthesis
The good incorporation of labelled 6-deoxyretzioside
(d-8) apparently supports our earlier (Damtoft et al.,
1994) proposed route for the biosynthesis of the C8 iri-
doid glucosides in Thunbergia (Scheme 2). The fact
that when feeding with d-8, the produced retzioside
(25) and the recovered precursor were fully labelled,
might indicate that these two compounds are not natu-
ral intermediates on the biosynthetic pathway.
However, also in the experiment where d-7 was used
as precursor, the produced thunaloside (24) was only
little diluted, if at all. The explanation could be that
the experiments were performed with a large amount
of precursor compared to the total amount of iridoids
present in the plant. Thus, de novo synthesis of iri-
doids may have been suppressed more or less comple-
tely and thus depleting the pool of natural
intermediates by the massive ¯ux of labelled com-
pounds. As mentioned above, the unedoside (1) deriva-
tives in Nuxia has also been shown to be derived from
deoxyloganic acid (7) and not from 6. This, together
with the fact that Retzia capensis and members of the
Stilbaceae have the same iridoid garniture as that
found in Thunbergia, indicates that the same biosyn-
thetic route to the C8 iridoids may be present in all
these taxa. It is notable that in the proposed biosyn-
thetic pathway to stilbericoside (2), the last steps after
formation of the 7,8-double bond are completely ana-
logous to the last steps in the biosynthesis of antirrhi-
noside (27; Scheme 2) (Damtoft, Jensen & Schacht,
1995), the only di€erence being the presence or
absence of C-10.
As might have been expected, only iridodial gluco-
side (9) was incorporated into the iridoid glucosides of
Deutzia when epimer-free, labelled precursors were
tested. The fact that 10-hydroxyiridodial glucoside (23)
was incorporated while 16 was not, allows us to sketch
a probable biosynthetic pathway to deutzioside (3) and
Deutzia is the only genus in Hydrangeaceae reported
to have iridoids lacking C-10, and likewise, Mentzelia
is unique in Loasaceae with this feature. Until recently,
most taxonomists considered the two taxa not to be
particularly closely related. Thus, Cronquist (1988)
places these two families in two di€erent superorders

1414
L.B. Frederiksen et al. / Phytochemistry 52 (1999) 1409±1420

L.B. Frederiksen et al. / Phytochemistry 52 (1999) 1409±1420
1415
with
Hydrangeaceae
in Rosidae/Rosales and
H-3), 3.96 (br d, J = 6.9 Hz, H-5), 8.00 (dd, J = 6.9
and 2.8 Hz, H-6), 6.18 (dd, J = 6.9 and 1.4 Hz, H-7),
3.10 (dd, J = 6.9 and 4.2 Hz, H-9), 3.69 (s, 11-OMe),
4.68 (d, J = 8.3 Hz, H-1'), 3.22 (t, J = 8.3 Hz, H-2'),
3.43 (t, J = 8.3 Hz, H-3'), 3.34 (t, J = 8.3 Hz, H-4'),
3.40 (m, H-5'), 3.83 (dd, J = 12.5 and 1.4 Hz, H-6a'),
3.65 (dd, J = 12.5 and 5.6 Hz, H-6b'), essentially as
reported by Uesato, Ali, Nishimura, Kawamura and
Inouye (1989). 13C NMR (62.5 MHz, D2O): d 95.1 (C-
1), 153.2 (C-3), 110.1 (C-4), 36.8 (C-5), 132.3 (C-6),
169.0 (C-7), 209.6 (C-8), 50.4 (C-9), 169.2 (C-11), 53.0
(11-OMe), 99.3 (C-1'), 73.4 (C-2'), 76.4 (C-3'), 70.3
(C-4'), 77.1 (C-5'), 61.5 (C-6').
Loasaceae in Dilleniidae/Violales. In Dahlgren's sys-
tem (1989), Hydrangeaceae are a member of the
Cornales and Loasaceae have been assigned an order
of its own. Thorne (1992) prefers a similar arrange-
ment. However, a very close relationship between the
two families was found in a cladistic analysis on mor-
phological and chemical grounds by Hu€ord (1992).
More recently, this result has been corroborated by
chloroplast DNA sequencing analyses (Hempel,
Reeves, Olmstead & Jansen, 1995; Xiang, Soltis &
Soltis, 1998), indicating that Hydrangeaceae and
Loasaceae make up a clade which should be regarded
as a sister group to the remainder of a restricted
Cornales. These ®ndings are fully corroborated by the
chemical data (Jensen, Nielsen & Dahlgren, 1975;
Jensen et al., 1981) where most genera of the two
families are characterized by the presence of loganin
and secoiridoids, but each with a single genus produ-
cing the structurally unique iridoid glucosides lacking
C-10 (but keeping C-11), compounds so far not
reported from other plant taxa.
4.2.2. Randioside tetra-O-trimethylsilylether (12s)
Randioside (12, 4.40 g) was dissolved in Py (30 ml)
followed by HMDS (hexamethyldisilazane; 10 ml) and
TMCS (trimethylchlorosilane; 10 ml), under stirring.
After 15 min Et2O (100 ml) was added followed by
5% AcOH (100 ml) and the phases were separated.
The Et2O layer was washed successively with satd aq.
NaHCO3 and H2O and dried (Na2SO4). Evapn of the
solvent gave crude 12s (6.74 g, 88%). This preparation
was used in the next step without further characteriz-
ation.
4. Experimental
4.1. General
4.2.3. Tetrahydrorandioside tetra-O-trimethylsilylether
(13s)
Microanalyses
were
performed
by
Leo
The above TMS-ether (12s, 3.50 g) was dissolved in
MeOH (30 ml) and NaBH4 (217 mg) was added under
stirring. After 5 min AcOH (0.5 ml) was added and
the solvent removed. The residue was dissolved in
PhMe, ®ltered and taken to dryness, leaving crude 13s
as a yellow syrup (3.27 g, 94%). The product was
Micronalytical Laboratory, Ballerup, Denmark.
M.p.'s: uncorr. 1H and 13C NMR spectra were
recorded in D2O, using the solvent peak (d 4.75) and
the C-6' signal (d 61.5) (Damtoft, Jensen & Nielsen,
1981), respectively, as standards or in CDCl3 (stan-
dards d 7.27 and 77.0). 2H NMR (77 MHz) spectra
were recorded in H2O with 0.0156% 2H of natural
abundance. Preparative chromatography was per-
formed at medium pressure (MPLC) on reverse phase
SiGel (Polygoprep C18 50±60 mm; 550 g) or on Merck
Lobar C18-columns (size B or C), eluting with H2O±
MeOH mixtures as speci®ed and monitoring simul-
taneously at 206 and 254 nm with a UV detector.
Plant material was either grown in a greenhouse (T.
alata) or in the free (D. species ). Both were provided
by the Botanical Garden of Copenhagen.
1
characterized solely by H NMR (250 MHz, CDCl3): d
5.28 (d, J = 8.3 Hz, H-1), 7.44 (s, H-3), 2.72 (br q,
J = 8.3 Hz, H-5), 1.74 (m, H-6a), 1.49 (m, H-6b), 2.20
(m, H-7a), 1.74 (m, H-7b), 4.52 (m, H-8), 2.31 (t,
J = 8.3 Hz, H-9), 3.70 (s, 11-OMe), 4.72 (d, J = 8.3
Hz, H-1'), 3.45-3.21 (m, 4H, H-2', H-3', H-4', H-5'),
3.79 (dd, J = 12.5 and 1.4 Hz, H-6a'), 3.58 (dd,
J = 12.5 and 5.6 Hz, H-6b'), 0.20-0.08 (4 Â s,
4 Â Tms). 13C NMR (62.5 MHz, CDCl3): d 95.4 (C-1),
151.9 (C-3), 111.5 (C-4), 33.7 (C-5), 30.3 (C-6), 34.5
(C-7), 74.7 (C-8), 43.5 (C-9), 167.5 (C-11), 51.1 (11-
OMe), 99.8 (C-1'), 73.5 (C-2'), 77.2 (C-3'), 71.3 (C-4'),
77.6 (C-5'), 61.5 (C-6'), 1.3±0.4 (4 Â Tms).
4.2. Synthesis of 6-deoxyretzioside
4.2.1. Randioside (12)
4.2.4. Tetrahydrorandioside tri-O-trimethylsilylether
6',8-dixanthate (13sx)
Gardenoside (11, 6.0 g) in phosphate bu€er (pH 6.0,
40 ml) was stirred at RT with NaIO4 (4.31 g) for 15
min when the mixt. was ®ltered through act. C on
Celite. The ®ltrate was evapd to a foam which was
fractd by MPLC (C-column). Elution with H2O±
MeOH (3:1) gave pure 12 (4.40 g, 73%). 1H NMR
(250 MHz, D2O): d 5.60 (d, J = 4.2 Hz, H-1), 7.35 (s,
To a solution of 13s (3.27 g) and MeI (340 ml) in
PhMe (25 ml), CS2 (25 ml) and 50% aq. NaOH (25
ml) were added with (n-Bu)4NHSO4 (171 mg). After
vigorous stirring for 3 h at RT the phases were separ-
ated. The aq. phase was extracted with PhMe (2 Â 25
ml), the combined org. layers were dried (Na2SO4) and

1416
L.B. Frederiksen et al. / Phytochemistry 52 (1999) 1409±1420
evapd to give a yellow syrup (2.17 g, 57%) which in
1
NMR appeared suciently pure for the next step. H
15 min, Ac2O (0.4 ml) was added and stirring contin-
ued for 1 h when EtOH (5 ml) was added. The dark
mixt. was ®ltered and the solvents evapd and the oily
residue partitioned between 2 M H2SO4 (50 ml) and
CHCl3 (50 ml). The org. phase was washed with H2O
(50 ml) and satd aq. NaHCO3 (2 Â 50 ml) and dried
(MgSO4). Evapn of the solvent gave a brownish syrup
(150 mg) which was deacetylated at RT with 0.1 M
NaOMe in MeOH (4 ml, 2 h). Chromatography (B-
NMR (250 MHz, CDCl3): d 5.60 (d, J = 4.2 Hz; H-1),
7.52 (s, H-3), 3.09 (q, J = 6.9 Hz, H-5), 1.52 (m, H-6),
1.77 (2H, H-6 and H-7), 2.06 (m, H-7), 6.24 (q,
J = 5.6 Hz, H-8), 2.38 (m, H-9), 3.78 (s, OMe), 4.69
(d, J = 6.9 Hz, H-1'), 3.4±3.7 (3H, H-2', H-3', H-4'),
2.79 (m, H-5'), 4.57 (dd, J = 5.6 and 12.5 Hz, H-6a'),
4.97 (br. d, J = 12.5 Hz, H-6b'), 2.60 and 2.65
(2 Â 3H, s's, 2 Â SMe). 13C NMR (62.5 MHz, CDCl3):
d 92.5 (C-1), 151.0 (C-3), 111.5 (C-4), 30.5 (C-5), 29.8
(C-6), 30.5 (C-7), 84.3 (C-8), 43.5 (C-9), 166.5 (C-11),
50.9 (11-OMe), 98.3 (C-1'), 74.2 (C-2'), 77.5 (C-3'),
71.6 (C-4'), 76.4 (C-5'), 72.5 (C-6'), 18.2 and 18.8
(2 Â SMe), 214.4 and 215.3 (2 Â C=S).
1
column; 3:1) gave pure 8 as a syrup (13 mg, 21%). H
NMR (250 MHz, D2O): d 5.32 (d, J = 4.2 Hz, H-1),
6.19 (br d, J = 5.6 Hz, H-3), 4.88 (dd, J = 5.6 and 2.8
Hz, H-4), 2.87 (m, H-5), 2.63 (br dd, J = 13.9 and 6.9
Hz, H-6b), 2.11 (br d, J = 13.9 Hz, H-6a), 5.90 (m,
H-7), 5.68 (m, H-8), 2.96 (m, H-9), 4.73 (d, obsc. by
the HDO-signal, H-1'), 3.25 (t, J = 8.3 Hz, H-2'), 3.5-
3.3 (3H, H-3', H-4' and H-5'), 3.87 (dd, J = 12.5 and
1.4 Hz, H-6a'), 3.66 (dd, J = 12.5 and 5.6 Hz, H-6b');
13C NMR (62.5 MHz, D2O): d 96.2 (C-1), 139.3 (C-3),
110.0 (C-4), 40.0 (C-5), 31.6 (C-6), 134.4 (C-7), 130.0
(C-8), 48.9 (C-9), 99.2 (C-1'), 73.6 (C-2'), 76.5 (C-3'),
70.4 (C-4'), 77.0 (C-5'), 61.5 (C-6').
4.2.5. 10-Nor-geniposidic acid tetra-acetate (14a)
The above crude dixanthate (13sx, 2.17 g) was dis-
solved in quinoline (20 ml) in a 100 ml ¯ask and ther-
molyzed in an oil bath under a stream of N2 at 2008C
for 90 min. The dark residue was partitioned between
CH2Cl2 (100 ml) and 2 M H2SO4 (50 ml) and the or-
ganic phase washed with more 2 M H2SO4 (50 ml) and
satd NaHCO3 (50 ml), dried and taken to dryness.
Desilylation was performed in MeOH (60 ml) with
AcOH (20 ml 50%). After 17 h the solvents were
removed and the crude product hydrolyzed with 1 M
NaOH (10 ml, 3.5 hrs). After neutralisation with
AcOH, fractionation by MPLC (C-column) with H2O±
MeOH (2:1) a€orded impure 10-nor-geniposidic acid
(14, 120 mg, 12%). Acetylation with Ac2O/Py 1:2 (2
ml, 2 h) gave after work-up the tetra-acetate (14a, 160
mg, 92%) which crystallized. Recryst. from CHCl3
gave the analytical sample, m.p. 181.5±182.58C (found:
C, 53.0; H, 5.7. C23H28O13Á1/2H2O requires: C 53.0;
4.3. Synthesis of deuterium labelled 6-deoxyretzioside
(d-8)
4.3.1. [6b,8b-2H2]-Tetrahydrorandioside tetra-O-
trimethylsilylether (d-13s)
Reduction of 12s (3.09 g) with NaBD4 (240 mg) as
2
above gave d-13s (3.12 g, quant), containing 0.95 H
1
in both the 6b- and 8b-positions according to the H
NMR spectrum.
4.3.2. [6b,8b-2H2]-Tetrahydrorandioside tri-O-
trimethylsilylether 6',8-dixanthate (d-13sx)
Treatment of (d-13s, 3.1 g) in PhMe (25 ml) with
MeI (0.32 ml) as in Section 4.2.4 gave d-13sx (1.7 g,
47%).
1
H, 5.6%); [a ]2D0=+298 (CHCl3; c 0.5); H NMR (250
MHz, CDCl3): d 5.14 (d, J = 4.5 Hz, H-1), 7.54 (s, H-
3), 3.11 (m, H-5), 2.84 (br dd, J = 16.5 and 8.0 Hz, H-
6b), 2.25 (br d, J = 16.5 Hz, H-6a), 5.91 (br d,
J = 8.0 Hz, H-7), 5.66 (br d, J = 8.0 Hz, H-8), 2.97
(m, H-9), 4.88 (d, J = 8.0 Hz, H-1'), 5.10 (dd, J = 9.3
and 8.0 Hz, H-2'), 5.25 (t, J = 9.3 Hz, H-3'), 5.02 (t,
J = 9,3 Hz, H-4'), 3.75 (m, H-5'), 4.29 (dd, J = 12.0
and 4.5 Hz, H-6a'), 4.12 (dd, J = 12.0 and 2.1 Hz, H-
6b'); 13C NMR (62.5 MHz, CDCl3): d 96.1 (C-1),
152.6 (C-3), 111.5 (C-4), 39.0 (C-5), 31.7 (C-6), 133.5
(C-7), 128.1 (C-8), 46.7 (C-9), 172.3 (C-11), 96.3 (C-
1'), 70.4 (C-2'), 71.8 (C-3'), 68.0 (C-4'), 72.3 (C-5'),
61.5 (C-6'), 170.5±169.1 (4 Â OAc), 20.5±20.1
(4 Â Ac±CH3).
4.3.3. [6b,8b-2H2]-10-Nor-geniposidic acid tetraacetate
(d-14a)
Crude d-13sx (1.7 g) was thermolyzed as in Section
4.2.5 for 75 min, desilylated with AcOH and ®nally
hydrolyzed with 1 M NaOH to give, after chromatog-
raphy d-14 (160 mg, 21%). Acetylation gave d-14a
(200 mg, 92%) as a cryst. residue.
4.3.4. [6b,8b-2H2]-6-Deoxyretzioside (d-8)
Decarboxylation of d-14a (200 mg) was performed
in quinoline (5 ml) with CuCOÁ3Cu(OH)2 (34 mg) as in
Section 4.2.6. The crude product (163 mg) was deace-
tylated and gave, after chromatography d-8 (37 mg,
26%).The 1H NMR spectrum was identical to the
spectrum of the unlabelled compound, except that the
compound contained 0.95 2H at the 6b-position (d
4.2.6. 6-Deoxyretzioside (8)
The above tetra-acetate (14a, 104 mg) was dissolved
in dry quinoline (5 ml) and CuCO.3Cu(OH)2 (13 mg)
was added. Under stirring, the mixt. was heated in an
oil bath at 2008C for 2 h. After cooling in the air for
2
2.63) and 0.92 H at the 8-position (d 5.68).

L.B. Frederiksen et al. / Phytochemistry 52 (1999) 1409±1420
1417
4.4. Synthesis of [11-2H]-iridodial glucoside (d-9)
4'), 3.53 (m, H-5'), 3.75 (br s, 2H, H-6'), 0.27-0.03
(5 Â s, 5 Â Tms). 13C NMR (62.5 MHz, CDCl3): d
96.7 (C-1), 151.2 (C-3), 112.2 (C-4), 41.5 (C-5), 31.6
(C-6), 74.7 (C-7), 42.7 (C-8), 44.5 (C-9), 13.7 (C-10),
167.4 (C-11), 50.8 (11-OMe), 98.6 (C-1'), 74.3 (C-2'),
76.6 (C-3'), 70.9 (C-4'), 77.9 (C-5'), 61.5 (C-6').
Iridotrial glucoside (800 mg), isolated from
Chaenostomum foetidum (cf. Jensen et al., 1989), in
H2O (15 ml) was reduced with NaBD4 (125 mg) under
stirring for 1 h, when AcOH was added until neutral.
Evapn and chromatography (C-column; 2:1) gave 11-
OH-(11-2H(-iridodial glucoside (721 mg, 90%). An ali-
quot (480 mg) was acetylated with Py (6 ml) and Ac2O
(3 ml) for 2 h to give the penta-acetate (770 mg). This
was dissolved in dioxane (10 ml) and Pd/C (5%, 120
mg) and HCOOH (72 mg) was added under stirring
and heating to re¯ux. After 1 h, TLC showed only
10% conversion and more Pd/C (140 mg) was added.
The reaction still commenced sluggishly and after 10 h
the reaction was stopped and the mixt. ®ltered, taken
to dryness and evapd twice with EtOH (2 Â 50 ml).
Deacetylation was performed in MeOH (20 ml) and
NaOMe was added to pH 14. After 10 min, the mixt.
was neutralized with AcOH. Evapn and chromatog-
raphy (C-column; 1.5:1 and 1:1) gave ®rst unconverted
11-OH-iridotrial glucoside (140 mg, 29%), then d-9
4.6.2. 7,11-Dihydroxyiridodial glucoside (18)
The silylether 17s (2.76 g) was dissolved in dry
PhMe (25 ml) in a three-necked ¯ask, ®tted with a
rubber septum, a stopper and a T-tube. The solution
was stirred under dry N2 while cooling to 608C.
DIBAL (20% in PhMe, 8.0 ml) was slowly added by
syringe. Stirring was continued for an additional 10
min at 608C and then AcOH (10 ml) was added.
When the reaction mixt. had reached RT, more PhMe
(50 ml) was added and it was washed with H2O (25
ml). The PhMe soln was ®ltered through Celite and
taken to dryness. Desilylation of the residue (2.51 g)
was performed with 10% aq. AcOH (20 ml) in MeOH
(100 ml) in 90 min when the solvents were removed to
give a crude product (1.5 g). Chromatography on a C-
column (2 runs), eluting with 4:1 gave slightly impure
1
(260 mg, 56%); H NMR (250 MHz, D2O) was essen-
1
tially as reported (Jensen et al., 1989) but the prep-
aration was free of the 8-epimer as seen by the absence
of the peak at d 5.39 present in the earlier preparation.
The deuterium content was determined to be 1.0 H at
C-11 by the integral of the peak at d 1.48.
18 (430 mg, 32%), characterized solely by NMR: H
NMR (250 MHz, D2O): d 5.20 (d, J = 2.8 Hz, H-1),
6.20 (s, H-3), 2.83 (br q, J = 6.9 Hz, H-5), 2.13-1.91
(4H, 2 Â H-6, H-8 and H-9), 4.10 (m, H-7), 0.97 (d,
J = 6.9 Hz, 3H, H-10), 4.04 (br d, J = 12.5 Hz, H-
11a), 3.82 (d, J = 12.5 Hz, H-11b), 4.75 (obsc. by the
HDO-signal, H-1'), 3.56±3.20 (4H, m, H-2', H-3', H-
4' and H-5'), 3.68 (dd, J = 12.5 and 5.6 Hz, H-6a'),
3.86 (partly obsc. by H-11; H-6b'); 13C NMR (62.5
MHz, D2O): d 96.7 (C-1), 137.0 (C-3), 119.6 (C-4),
38.6 (C-5), 30.9 (C-6), 75.3 (C-7), 40.7 (C-8), 46.3 (C-
9), 12.9 (C-10), 61.7 (C-11), 99.1 (C-1'), 73.5 (C-2'),
76.4 (C-3'), 70.4 (C-4'), 77.1 (C-5'), 61.5 (C-6').
2
4.5. [6,7,8,10-2H4]-8-Epiiridodial glucoside (d-10)
The earlier prepared compound (Breinholt et al.,
2
1992), had a H content of ca. 0.8 at C-6, 0.8 at C-7,
0.9 at C-8 and 1.9 at C-10. Prep. HPLC using a Merck
HiBar coloumn (250 Â 25 mm) packed with
LiChrosorb RP-18 (7 mm) and eluting with 1:1 gave
virtually baseline separation of d-10 from the small
contamination of d-9.
4.6.3. 7-Hydroxyiridodial glucoside (16)
The glucoside 18 (430 mg) was acetylated with
Ac2O/Py 1:1 (2 h) to give the hexaacetate (18a, 550
mg). This was dissolved in EtOH (25 ml) and Pd/C
(5%, 213 mg) was added and the soln was stirred
under gentle re¯ux. A mixture of NEt3 (42 mg) and
HCOOH (46 mg, 1.1 eq) in EtOH (10 ml) was added
dropwise (5 min) and the reaction was followed by
TLC (Et2O). After 65 min the reaction mixt. was ®l-
tered through Celite and evapn of the ®ltrate gave a
residue (498 mg). Deacetylation was performed in 0.1
M NaOMe (10 ml, 2.5 h) and chromatography (B-col-
umn, 3:1) gave recovered 18 (50 mg, 12%) and 16 (190
mg, 61%). An analytical sample was cryst. from
EtOH, m.p. 214±2158C (found: C, 55.4; H, 7.6.
C16H26O8 requires: C, 55.5; H, 7.6%); [a]D22
96.28(EtOH; c 0.2,); 1H NMR (250 MHz, D2O): d
5.14 (d, J = 3.0 Hz, H-1), 5.92 (br s, H-3), 2.62 (q,
J = 6.9 Hz, H-5), 1.95 (m, H-6a), 1.69 (m, H-6b), 4.08
4.6. Synthesis of 7-hydroxyiridodial glucoside (16)
4.6.1. Loganin penta-O-trimethylsilylether (17s)
Loganin (17, 1.60 g) in Py (15 ml) was treated with
HMDS (3.5 ml) and TMCS (3.5 ml) under stirring for
20 min. Et2O (75 ml) was added, followed by 5%
AcOH (60 ml). The org. phase was washed successively
with satd NaHCO3 and H2O (50 ml each) and dried
(MgSO4). Evapn of the solvent gave a residue which
was evapd from PhMe (25 ml), to give 17s as a yellow-
1
ish syrup (2.76 g, 90%). H NMR (250 MHz, CDCl3):
d 5.06 (d, J = 5.6 Hz, H-1), 7.43 (s, H-3), 3.20 (m, H-
5), 2.26 (dd, J = 13.9 and 6.9 Hz, H-6a), 1.41 (m, H-
6b), 4.01 (t, J = 1.4 Hz, H-7), 1.84 (m, H-8), 2.00 (q,
J = 8.3 Hz, H-9), 1.04 (d, J = 6.5 Hz, 3H, H-10), 3.72
(s, 11-OMe), 4.62 (d, J = 8.3 Hz, H-1'), 3.20 (m, H-
2'), 3.47 (t, J = 8.3 Hz, H-3'), 3.31 (t, J = 8,3 Hz, H-

1418
L.B. Frederiksen et al. / Phytochemistry 52 (1999) 1409±1420
(br s, H-7), 1.81 (m, H-8), 1.95 (m, H-9), 1.00 (d,
J = 13.9 Hz, 3H, H-10), 1.48 (s, 3H, H-11), 4.68 (d,
J = 8.3 Hz, H-1'), 3.23 (t, J = 8.3 Hz, H-2'), 3.44 (t,
J = 8.3 Hz, H-3'), 3.35 (t, J = 8.3 Hz, H-4'), 3.40 (m,
H-5'), 3.84 (dd, J = 12.5 and 2.8 Hz, H-6a'), 3.65 (dd,
J = 12.5 and 5.6 Hz, H-6b'); 13C NMR (62.5 MHz,
D2O): d 96.6 (C-1), 132.4 (C-3), 117.8 (C-4), 39.2 (C-
5), 34.9 (C-6), 75.4 (C-7), 41.1 (C-8), 46.8 (C-9), 12.0
(C-10), 16.1 (C-11), 99.0 (C-1'), 73.6 (C-2'), 76.5 (C-
3'), 70.4 (C-4'), 77.0 (C-5'), 61.5 (C-6').
4), 34.8 (C-5), 27.4Ã (C-6), 29.5Ã (C-7), 42.5Ã (C-8),
44.6 (C-9), 66.0 (C-10), 61.6 (C-11), 99.5 (C-1'), 73.5
(C-2'), 76.4 (C-3'), 70.4 (C-4'), 77.0 (C-5'), 61.5 (C-6'),
close to those reported (Damtoft et al., 1981).
4.8.3. [6,7,8,10-2H]-Decapetaloside (d-23)
Impure d-21 (145 mg) was acetylated to give the
hexaacetate (d-21a, 143 mg) as a colourless syrup. It
was hydrogenated as above in EtOH (20 ml) with 5%
Pd/C (220 mg) and of NEt3 (14 mg) and HCOOH (17
mg, 1.1 eq). The reaction was followed by TLC
(Et2O). After 75 min more HCOOH (17 mg, 1.1 eq)
was added. The catalyst was ®ltered o€ after a total of
6 h and the ®ltrate was taken to dryness (133 mg) and
deacetylated with NaOMe/MeOH. Chromatography
(B-column, 4:1 to 2:1) gave ®rst d-21 (21 mg, 25%),
and then d-23 (28 mg, 34%) as a syrup; 1H NMR
4.7. Synthesis of [11-2H]-7-Hydroxyiridodial glucoside
(d-16)
The hexa-acetate 18a (130 mg) in EtOD (5 ml) was
reduced with Pd/C (5%, 63 mg), DCOOD (2 Â 10 ml)
and NEt3 (10 ml) during 4 h. Deacetylation and chro-
matography gave 18 (10 mg) and d-16 (40 mg, 47%).
1H NMR showed that the signal at d 1.48 (H-11) had
an integral corresponding to 2.3 H.
2
2
showed a H content of 0.6 H in the 6a-position, 0.4
2H in the 6b-position, 0.8 2H in the 6a-position, 0.6
2
2
2H in the 7b-position, 0.8 H at C-8 and 0.4 H at C-
1
10. H NMR (500 MHz, D2O): d 5.19 (d, J = 5.6 Hz,
4.8. Synthesis of [6,7,8,10-2H]-decapetaloside (d-23)
H-1), 6.10 (s, H-3), 2.53 (t-like, J = 7.0 Hz, H-5), 1.88
(0.4H, m, H-6a), 1.59 (obsc. by the 11-Me-signal, ca.
0.5H, H-6b), 1.67 (0.2H, m, H-7a), 1.35 (0.4H, m, H-
7b), 2.08 (0.2H, q-like, J = 7.0 Hz, H-8), 1.96 (m, H-
9), 3.62 (d, J = 11.1 Hz, H-10a), 3.56 (d, J = 11.1 Hz,
H-10b), 1.57 (s, 3H, H-11), 4.76 (obsc. by the HDO-
signal, H-1'), 3.31 (t, J = 9.7 Hz, H-2'), 3.53 (t,
J = 9.7 Hz, H-3'), 3.42 (t, J = 9.7 Hz, H-4'), 3.49 (m,
H-5'), 3.94 (dd, J = 12.5 and 2.8 Hz, H-6a'), 3.74 (dd,
J = 12.5 and 5.6 Hz, H-6b'), some assignments are
uncertain (similar to that reported by Jensen et al.
(1985)). 13C NMR (125 MHz, D2O; Ãlow intensity
peaks): d 97.3 (C-1), 134.1 (C-3), 116.2 (C-4), 38.6 (C-
5), 29.6Ã (C-6), 27.4Ã (C-7), 43.0Ã (C-8), 45.1 (C-9),
66.1Ã (C-10), 15.8 (C-11), 99.4 (C-1'), 73.6 (C-2'), 76.6
(C-3'), 70.4 (C-4'), 77.0 (C-5'), 61.5 (C-6'), almost
identical to that reported by Jensen et al. (1981).
4.8.1. [6,7,8,10-2H]-Adoxoside/epiadoxoside (d-19/d-
20)
Reduction of geniposide (15, 1.20 g) over Pd/C with
D2 and added Et3N (0.25 ml) was performed as
described (Damtoft et al., 1994), to give, after chroma-
tography (C-column, 2 runs, 2:1), a 2:3 mixt. (880 mg,
73%) of labelled adoxoside (d-19) and 8-epiadoxoside
(d-20).
4.8.2. [6,7,8,10-2H]-11-Hydroxydecapetaloside (d-22)
The above mixt. (d-19/d-20; 410 mg) was acetylated
with Ac2O/Py 1:1 (7 ml, 2 h) to give the mixt. of pen-
taacetates (d-19a/d-20a, 560 mg, 83%). An aliquot
(280 mg) in dry PhMe at 408C (50 ml) was reduced
with DIBAL (20% in PhMe,14.0 ml) for 10 min, then
AcOH (10 ml) was added. The reaction mixt. was
extracted with H2O (2 Â 25 ml) and the combined aq.
extracts were ®ltered through celite and evapd to a
white foam. Chromatography (C-column, 4:1) gave 2
frs with mainly d-21 (96 mg) and mainly d-22 (58 mg).
4.9. Feeding experiments
4.9.1. Thunbergia alata
The labelled precursors (d-7 and d-8) were dissolved
in H2O (2 ml) in 10 ml beakers. Leaves (ca. 14 g) from
young T. alata plants (with 8±10 leaves), were
immersed into the above solns to absorb the precursor.
In order to ensure complete transfer to the plant, more
H2O (2 Â 1 ml) was added when almost all had been
taken up. After 3±5 h the leaves were placed in a lar-
ger beaker to metabolize the precursor. Due to begin-
ning withering, the plants were worked up after 3
days. The leaves were homogenized with EtOH (250
ml), ®ltered and concentrated. The residue was parti-
tioned between H2O and Et2O. The aq. phase was
evapd to a greenish syrup which was dissolved in
MeOH (20 ml) and ®ltered through a layer of act. C
1
Spectral data for d-21: H NMR (250 MHz, D2O): d
5.22 (d, J = 4.2 Hz, H-1), 6.37 (br s, H-3), 2.74 (br t,
2
J = 5.6 Hz, H-5), 1.74 (m, 0.4 H, 0.6 H, H-6a), 1.32
2
2
(m, 0.7 H, 0.3 H, H-6b), 1.87 (m, 0.2 H, 0.8 H, H-
7a), 1.59Ã (m, 0.5 H, 0.5 H, H-7b), 2.02 (m, 0.3 H, 0.7
2
2H, H-8), 1.99 (m, H-9), 3.60-3.37 (m, H-10a, H-10b),
4.12 (br d, J = 13.9 Hz, H-11a), 3.90 (br d, J = 13.9
Hz, H-11b), 4.78 (obsc. by the HDO-signal, H-1'),
3.29 (t, J = 8.3 Hz, H-2'), 3.60±3.37 (m, H-3', H-4'
and H-5'), 3.90 (br d, J = 12.5 Hz, H-6a'), 3.69 (dd,
J = 12.5 and 5.6 Hz, H-6b'), similar to those reported
(Jensen et al., 1985); 13C NMR (62.5 MHz, D2O; low
Ã
intensity signals): d 97.3 (C-1), 138.6 (C-3), 117.7 (C-

L.B. Frederiksen et al. / Phytochemistry 52 (1999) 1409±1420
1419
NMR spectroscopy as a tool in the con®gurational analysis of iri-
doid glucosides. Phytochemistry, 20, 2717±2732.
(400 mg) on Celite. The ®ltrate was taken to dryness
and chromatographed (B-column; 1:0 to 1:1). The frac-
Damtoft, S., Jensen, S. R., & Nielsen, B. J. (1993a). Schismoside, an
iridoid glucoside from Schismocarpus matudai. Phytochemistry,
32, 885±889.
2
tions thus obtained (Table 1) were analyzed by H and
1H NMR for incorporation and for identity, respect-
ively. The compounds containing measurable amounts
Damtoft, S., Jensen, S. R., Jessen, C. U., & Knudsen, T. B. (1993b).
Late stages in the biosynthesis of aucubin in Scrophularia.
Phytochemistry, 33, 1089±1093.
2
of H incorporation were: 2 (d 3.75, D-8), 24 (d 3.7,
D-10), 25 (d 6.0, D-8) and 8 (d 5.7, D-8). The 1H
NMR spectrum of the fraction with 25 showed that
this compound was contaminated with an equimolar
amount of tryptophan but also that it was almost
Damtoft, S., Franzyk, H., Jensen, S. R., & Nielsen, B. J. (1993c).
Iridoids and verbascosides in Retzia. Phytochemistry, 34, 239±
243.
Damtoft, S., Frederiksen, L., & Jensen, S. R. (1994). Biosynthesis of
iridoid glucosides in Thunbergia alata. Phytochemistry, 37, 1599±
1603.
1
devoid of H at C-8 (250 MHz, D2O): d 5.26 (br d,
J = 5.6 Hz, H-1), 6.18 (br d, J = 6.9 Hz, H-3), 5.0
(m, H-4), 2.64 (m, H-5), 4.50 (br. s, H-6), 5.94 (br s,
H-7), 6.0 (<0.05H, m, H-8), 3.2 (m, H-9), 4.7 (obsc.
by the HDO-signal, H-1'), almost identical to that
reported (Damtoft et al., 1993c).
Damtoft, S., Jensen, S. R., & Schacht, M. (1995). Last stages in the
biosynthesis of antirrhinoside. Phytochemistry, 39, 549±551.
Danielson, T. J., & Hawes, E. M. (1973). Iridoids of Mentzelia deca-
petala (Purch). Mentzelioside. Canadian Journal of Chemistry, 51,
760±766.
DePuy, C. H., & King, R. W. (1960). Pyrolytic cis eliminations.
Chemical Reviews, 60, 444±457.
4.9.2. Deutzia species
El-Naggar, L. J., Beal, J. L., & Doskotch, R. W. (1982). Iridoids glu-
cosides from Mentzelia decapetala. Journal of Natural Products,
45, 539±550.
Precursors were dissolved in H2O (3 ml) in 10 ml
beakers. Young shoots (ca. 50 g) from D.
Schneideriana or D. scabra were immersed into the
solns and after a metabolic period of 4 days the plants
were worked up as in Section 4.9.1. The aq. extracts
were treated with Al2O3 (neutral, 50 g).
Chromatography was performed as above and the
Engell, K. (1987). Embryology and taxonomical position of Retzia
capensis (Retziaceae). Nordic Journal of Botany, 7, 117±124.
Esposito, P., & Guiso, M. (1973). Scabroside: structure and con®gur-
ation. Gazetta Chimica Italiana, 103, 517±523.
Geissman, T. A., Knaack Jr, W. F., & Knight, J. O. (1996).
Unedoside, a novel iridoid compound. Tetrahedron Letters, %
1245±1249.
2
fractions were analyzed by H NMR. The results are
listed in Table 2.
Hempel, A. L., Reeves, P. A., Olmstead, R. G., & Jansen, R. K.
(1995). Implications of rbcL sequence data for higher order re-
lationships of the Loasaceae and the anomalous aquatic plant
Hydrostachys (Hydrostachyaceae). Plant Systematics and
Evolution, 194, 25±37.
Acknowledgements
Hu€ord, L. (1992). Rosidae and their relationships to other nonmag-
noliid dicotyledons: a phylogenetic analysis using morphological
and chemical data. Annals of the Missouri Botanical Garden, 79,
218±248.
We thank the sta€ at The Botanical Garden of
Copenhagen for growing and providing the plant ma-
terial and LEO Pharmaceuticals A/S, Ballerup,
Denmark for performing the elementary analyses.
Inouye, H., Ueda, S., & Uesato, S. (1977). Intermediacy of iridodial
in the biosynthesis of some iridoid glucosides. Phytochemistry, 16,
1669±1675.
Inouye, H., Ueda, S., Uesato, S., & Kobayashi, K. (1978). Studies
on monoterpene glucosides and related products XXXVII.
Biosynthesis of the iridoid glucosides in Lamium amplexicaule,
Deutzia crenata and Galium spurium var. echinosperma. Chemical
and Pharmaceutical Bulletin, 26, 3384±3394.
References
Jahodar, L., Kolb, I.,
& Leifertova, I. (1981). Unedoside in
Bonadies, F., Esposito, P., & Guiso, M. (1974). Iridoids: XII.
Deutzioside: structure and con®guration. Gazetta Chimica
Italiana, 104, 17±24.
Arctostaphylos uva-ursi roots. Pharmazie, 36, 294±296.
Jensen, S. R., & Nielsen, B. J. (1989). Iridoids in Thunbergia species.
Phytochemistry, 28, 3059±3061.
Bordwell, F. G., & Landis, P. S. (1958). Elimination reactions. IX.
Mechanisms for the pyrolysis of methyl cyclohexyl sul®tes.
Journal of the American Chemical Society, 80, 6379±6382.
Breinholt, J., Damtoft, S., Demuth, H., Jensen, S. R., & Nielsen, B.
J. (1992). Biosynthesis of antirrhinoside in Antirrhinum majus.
Phytochemistry, 31, 795±797.
Jensen, S. R., Nielsen, B. J., & Dahlgren, R. (1975). Iridoid com-
pounds, their occurrence and systematic importance in the
angiosperms. Botaniska Notiser, 128, 148±180.
Jensen, S. R., Mikkelsen, C. B., & Nielsen, B. J. (1981). Iridoid
mono- and di-glycosides in Mentzelia. Phytochemistry, 20, 71±83.
Jensen, H. F. W., Jensen, S. R.,
& Nielsen, B. J. (1988).
Cronquist, A. (1988). The evolution and classi®cation of ¯owering
plants (2nd ed.). New York: The New York Botanical Garden.
Dahlgren, G. (1989). The last Dahlgrenogram. System of classi®-
cation of the dicotyledons. In K. Tan, Plant taxonomy, phytogeo-
graphy and related subjects (pp. 249±260). Edinburgh: The Davis
and Hedge Festschrift.
Chemotaxonomy of the Acanthaceae. Iridoid and quaternary
amines. Phytochemistry, 27, 2581±2589.
Jensen, S. R., Kirk, O., & Nielsen, B. J. (1989). Biosynthesis of the
iridoid glucoside cornin in Verbena ocinalis. Phytochemistry, 28,
97±105.
Dahlgren, R., Nielsen, B. J., Goldblatt, P., & Rourke, J. P. (1979).
Further notes on Retziaceae: its chemical contents and anities.
Annals of the Missouri Botanical Garden, 66, 545±556.
Damtoft, S., Jensen, S. R., & Nielsen, B. J. (1981). 13C and 1H
Jensen, S. R. (1991). Plant iridoids, their biosynthesis and distri-
bution in angiosperms. In J. B. Harborne, & F. A. Tomas-
Barbaran, Ecological Chemistry and Biochemistry of Plant
Terpenoids (pp. 133±158). Oxford: Clarendon Press.

1420
L.B. Frederiksen et al. / Phytochemistry 52 (1999) 1409±1420
Jensen, S. R., Ravnkilde, L., & Schripsema, J. (1998). Unedoside de-
rivatives in Nuxia and their biosynthesis. Phytochemistry, 47,
1007±1011.
Rimpler, H., & Pistor, H. (1974). Stilbericosid, ein neues C8-iridoid
aus Stilbe ericoides L. Zeitschift fur Naturforschung, 29c, 368±373.
È
Scotland, R. W., Sweere, J. A., Reeves, P. A., & Olmstead, R. G.
(1995). Higher-level systematics of Acanthaceae determined by
chloroplast DNA squences. American Journal of Botany, 82, 266±
272.
Jensen, S. R. (1992). Systematic implications of the distribution of
iridoids and other chemical compounds in the Asteridae. Annals
of the Missouri Botanical Garden, 79, 284±302.
Lee, A. W. M., Chan, W. H., Wong, H. C., & Wong, M. S. (1989).
One pot transfer synthesis of O-alkyl,S-methyl dithiocarbonates
(xanthates). Synthetic Communications, 19, 547±552.
Thorne, R. F. (1992). An updated phylogenetic classi®cation of the
¯owering plants. Aliso, 13, 365±389.
Uesato, S., Miyauchi, M., Itoh, H.,
& Inouye, H. (1986).
Leeuwenberg, A. J. M., & Leenhouts, P. W. (1980). In A. J. M.
Leeuwenberg, Die Naturlischen P¯anzenfamilien. I. Angiospermae:
Biosynthesis of iridoid glucosides in Galium mollugo, G. spurium
var. echinospermon and Deutzia crenata. Intermediacy of deoxylo-
ganic acid, loganin and iridodial glucoside. Phytochemistry, 25,
2515±2521.
È
Ordnung Gentianales, Fam. Loganiaceae (pp. 8±96). In Taxonomy,
Vol. 28b. Berlin: Duncker und Humblot.
Murai, F., & Tagawa, M. (1980). The absolute con®guration of
boschnaloside and the chemical conversion of genipin into bosch-
naloside. Chemical and Pharmaceutical Bulletin, 28, 1730±1735.
Oxelman, B., Backlund, M. & Bremer, B. (1999). Relationships of
the Buddlejaceae s.l. inferred from chloroplast ndhF and rbcL
sequences. Systematic Botany, 24, in press.
Uesato, S., Ali, E., Nishimura, H., Kawamura, I., & Inouye, H.
(1989). Four iridoids from Randia canthioides. Phytochemistry,
21, 353±357.
Xiang, Q.-Y., Soltis, D. E., & Soltis, P. S. (1998). Phylogenetic re-
lationships of Cornaceae and close relatives inferred from matK
and rbcL sequences. American Journal of Botany, 85, 285±297.