Analysis of metabolic pathways via quantitative prediction of isotope labeling patterns: A retrobiosynthetic 13C NMR study on the monoterpene loganin

Article abstract of DOI:10.1016/S0031-9422(98)00733-X

The monoterpene loganin serves as a precursor in the biosynthetic pathways of numerous indole alkaloids. In contrast to earlier studies, we present evidence that the biosynthesis of loganin in Rauwolfia serpentina cells proceeds mainly via the deoxyxylulose pathway and not by the mevalonate pathway. This conclusion is based on experiments using a R. serpentina cell culture supplied with ¹³C-labeled samples of glucose, ribose/ribulose, pyruvate or glycerol. Loganin was isolated from biomass, and the hydrolysis of cellular protein afforded amino acids. The isolated metabolites were analyzed by NMR spectroscopy. The ¹³C-labeling patterns of isolated amino acids were then used to reconstruct the labeling patterns of phosphoenol pyruvate, pyruvate and acetyl CoA. These labeling patterns were subsequently used to predict labeling patterns for dimethylallyl pyrophosphate and isopentenyl pyrophosphate via the mevalonate and deoxyxylulose pathway, respectively. The observed labeling patterns of the terpenoid moieties in loganin were in excellent agreement with the deoxyxylulose prediction. The minor incorporation of mevalonate into loganin observed in earlier studies can be attributed to metabolite exchange between the two terpenoid pathways. The possibility of crosstalk between the two pathways in plants and plant cell cultures stresses the need for a quantitative analysis of general carbon metabolism in order to determine the partitioning between the mevalonate and deoxyxylulose pathway. The present study shows that a wide variety of general metabolic precursors can fulfill this task in conjunction with the retrobiosynthetic concept.

Full text of DOI:10.1016/S0031-9422(98)00733-X

Phytochemistry 51 (1999) 223±236
Analysis of metabolic pathways via quantitative prediction of
13
isotope labeling patterns: a retrobiosynthetic C NMR study on
the monoterpene loganin
Dietmar Eichingera, Adelbert Bachera, Meinhart H. Zenkb, Wolfgang Eisenreicha,*
aLehrstuhl fur Organische Chemie und Biochemie, Technische Universitat Munchen, Lichtenbergstraûe 4, D-85747 Garching, Germany
È
È
È
bInstitut fur Pharmazeutische Biologie, Ludwigs-Maximilians-Universitat Munchen, Karlstr. 29, D-80333 Munchen, Germany
È
È
È
È
Received 6 October 1998
Abstract
The monoterpene loganin serves as a precursor in the biosynthetic pathways of numerous indole alkaloids. In contrast to
earlier studies, we present evidence that the biosynthesis of loganin in Rauwol®a serpentina cells proceeds mainly via the
deoxyxylulose pathway and not by the mevalonate pathway. This conclusion is based on experiments using a R. serpentina cell
culture supplied with 13C-labeled samples of glucose, ribose/ribulose, pyruvate or glycerol. Loganin was isolated from biomass,
and the hydrolysis of cellular protein a€orded amino acids. The isolated metabolites were analyzed by NMR spectroscopy. The
13C-labeling patterns of isolated amino acids were then used to reconstruct the labeling patterns of phosphoenol pyruvate,
pyruvate and acetyl CoA. These labeling patterns were subsequently used to predict labeling patterns for dimethylallyl
pyrophosphate and isopentenyl pyrophosphate via the mevalonate and deoxyxylulose pathway, respectively. The observed
labeling patterns of the terpenoid moieties in loganin were in excellent agreement with the deoxyxylulose prediction. The minor
incorporation of mevalonate into loganin observed in earlier studies can be attributed to metabolite exchange between the two
terpenoid pathways. The possibility of crosstalk between the two pathways in plants and plant cell cultures stresses the need for
a quantitative analysis of general carbon metabolism in order to determine the partitioning between the mevalonate and
deoxyxylulose pathway. The present study shows that a wide variety of general metabolic precursors can ful®ll this task in
conjunction with the retrobiosynthetic concept. # 1999 Elsevier Science Ltd. All rights reserved.
Keywords: Rauwol®a serpentina; Apocynaceae; Loganin; Isoprenoid biosynthesis; Deoxyxylulose pathway; Indole alkaloids; NMR; Retrobio-
synthesis
1. Introduction
Francis, 1972; Uesato, Kanomi, Iida, Inouye, & Zenk,
1986).
The ring system of loganin (9, Fig. 1) is assembled
from one molecule each of the universal terpenoid
precursors dimethylallyl pyrophosphate (1) (DMAPP)
and isopentenyl pyrophosphate (2, IPP) via geranyl
pyrophosphate (3). Cyclization of the intermediate
(4) yields the cyclopentane derivative (5) which is
then further elaborated via (8) to the ®nal com-
pound (9) (for review see Banthorpe, Charlwood, &
Pioneering studies by Bloch, Cornforth, Lynen and
their coworkers had shown that IPP (2) can be formed
from three molecules of acetyl CoA (10) via mevalo-
nate (11) in yeast and animal cells (Fig. 2; for review
see Qureshi & Porter, 1981; Bloch, 1992). DMAPP (1)
was shown to be formed from IPP by a reversible iso-
merization step (for review see Ramos-Valdivia, van
der Heijden, & Verpoorte, 1997).
During a period of several decades, the mevalonate
pathway was considered as the unique biosynthetic
route to terpenoids, although numerous experimental
observations obtained with plants and certain bacteria
* Corresponding author. Tel.: +49-89-289-13043; fax +49-89-289-
13363.
E-mail address: wolfgang.eisenreich@ch.sum.de (W. Eisenreich)
0031-9422/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved.
PII: S0031-9422(98)00733-X

224
D. Eichinger et al. / Phytochemistry 51 (1999) 223±236
Fig. 1. Pathway which has been proposed in the literature for the biosynthesis of loganin (for review, see Banthorpe et al., 1972; Stockigt, 1979).
The isoprene dissection of the monoterpene is indicated by bold bars.
were not easily explained. An alternative terpenoid
pathway operative in plants and certain microorgan-
isms was detected relatively recently by independent
contributions from Rohmer, Arigoni, Sahm and their
research groups (Fig. 2; for review see Eisenreich et
al., 1998; Rohmer, 1998). More speci®cally, Rohmer,
Sahm and their coworkers showed that the transform-
ation of 13C-labeled glucose and acetate into bacterial
hopanoids could not be explained by the classical
moieties are involved in the mevalonate pathway.
Using various 13C labeled glucose samples, Arigoni
and his coworkers were also able to show that the for-
mation of ubiquinone in Escherichia coli proceeds via
the non-mevalonate pathway (Broers, 1994).
The enzyme catalyzing the formation of 1-deoxy-D-
xylulose 5-phosphate (14) from glyceraldehyde 3-phos-
phate (13) and pyruvate (12) in the alternative pathway
has been described from E. coli (Sprenger et al., 1997;
Campos, Lois, & Boronat, 1997; Lois et al., 1998) and
from Mentha piperita (Lange, Wildung, McCaskill, &
Croteau, 1998). The further conversion of 1-deoxy-D-
xylulose 5-phosphate (14) to 2-C-methyl-D-erythritol 4-
phosphate (15) was recently demonstrated to proceed
in a single step by intramolecular rearrangement and
reduction via an enzyme which was cloned and
mevalonate hypothesis (Flesch
& Rohmer, 1988;
Rohmer, Knani, Simonin, Sutter, & Sahm, 1993).
Independently, Arigoni and his coworkers (Schwarz,
1994) using uniformly 13C-labeled glucose as precursor
had shown that the biosynthesis of ginkgolides by
seedlings of Ginkgo biloba involves the incorporation
of a three-carbon fragment, whereas only two-carbon

D. Eichinger et al. / Phytochemistry 51 (1999) 223±236
225
Fig. 2. Diversion of carbon atoms from precursors to isoprenoid monomers by the mevalonate pathway (precursor, acetyl CoA) and the deoxyx-
ylulose pathway (precursors pyruvate and glyceraldehyde 3-phosphate). The deoxyxylulose pathway involves a rearrangement of 1-deoxy-D-xylu-
lose 5-phosphate which interrupts the contiguity of the carbon atoms derived from the triose phosphate precursor.
expressed from E. coli (Takahashi, Kuzuyama,
Watanabe, & Seto, 1998).
other hand, no incorporation from radiolabeled meva-
lonate into loganin was observed using other Vinca
species (D. Arigoni, personal communication).
Similarly, 14C-mevalonate was not signi®cantly incor-
porated into loganin from Menyanthes trifoliata, but
was eciently used as a precursor for triterpenes and
sterols in the same biological system (Brechbuhler-
Bader, Coscia, Loew, Sczepanski, & Arigoni, 1968).
Despite those con¯icting results, it was widely assumed
until recently that loganin is derived from mevalonate
Both terpenoid pathways have been shown to occur
in cells of higher plants (for review see Eisenreich et
al., 1998; Rohmer, 1998). Speci®cally, the mevalonate
pathway appears to be operative in the cytoplasmic
compartment where it is supplying the building blocks
for the biosynthesis of sterols. The deoxyxylulose path-
way (Fig. 2) appears to be compartmentalized in plas-
tids and has been shown to yield phytol as well as a
variety of hemiterpenes, monoterpenes, diterpenes and
carotenoids (Arigoni et al., 1997; Schwender et al.,
1997).
The biosynthetic origin of DMAPP and IPP used
for the biosynthesis of loganin had been investigated
repeatedly using 14C- and 3H-labeled mevalonate as
precursors. Incorporation of radioactivity from meva-
lonate into the terpenoid moiety of loganin from Vinca
rosea was documented unequivocally, however in low
rates (Battersby et al., 1968; Escher, Loew, & Arigoni,
1970; Guarnaccia, Botta, & Coscia, 1974). On the
(Verpoorte, van der Heijden,
& Moreno, 1997;
Whitmer, Canel, Hallard, Goncalves, & Verpoorte,
1998; Shanks, Bhadra, Morgan, & Rijhwani, 1998).
In light of the important role of loganin as a precur-
sor of over 2000 indole alkaloids known to date, we
decided to investigate its biosynthetic origin by retro-
biosynthetic analysis which permits the quantitative
analysis of metabolite ¯ux patterns. This method uses
information gleaned from isotope distribution patterns
in primary metabolites in order to interpret the label-
ing patterns of secondary metabolites.

226
D. Eichinger et al. / Phytochemistry 51 (1999) 223±236
The data show that the bulk of loganin (>95%) in
quantitatively using the experimentally determined
labeling patterns of central, primary intermediates.
These predicted labeling patterns can then be com-
pared with the observed labeling pattern of the second-
ary metabolite under study. Metabolic pathways which
predict labeling patterns at odds with the observed pat-
tern can be ruled out with a high degree of certainty.
In cases where a given metabolite can be formed sim-
ultaneously via more than one metabolic pathway, the
partitioning of metabolite ¯ux through the di€erent
pathways can be estimated (for review see Bacher et
al., 1998). The application of this general method to
the biosynthesis of loganin in R. serpentina cell cul-
tures is described in detail below.
R. serpentina is formed via the deoxyxylulose pathway.
The earlier studies with radioisotopes have apparently
detected a minor contribution from the mevalonate
pathway to the two C5 building blocks from which
loganin is assembled.
2. Results
2.1. Experimental strategy
The products of secondary metabolism are derived
from amino acids or from central metabolic intermedi-
ates (e.g. of the Embden±Meyerhof pathway, the pen-
tose phosphate cycle and the citrate cycle). In isotope
labeling experiments, the labeling pattern of each
metabolite studied must therefore re¯ect the labeling
patterns of the precursor molecules from which it has
been assembled in the biosynthetic process.
The retrobiosynthetic method described in the fol-
lowing section utilizes general metabolic precursors
such as carbohydrates, polyols and carboxylic acids as
isotope-labeled tracers. These general metabolites can
divert isotopic label to virtually all branches of cellular
metabolism. The metabolites generated in these exper-
iments are complex isotopomer mixtures, since the
label of the pro€ered precursor can be reshu‚ed
extensively during its passage through linear and/or
cyclic pathways of intermediary metabolism (Werner,
Bacher, & Eisenreich, 1997).
2.2. Isotope incorporation studies
Cell cultures of R. serpentina were grown with D-glu-
cose as carbon source. The time for doubling of cell
mass was 18 days. Isotope labeled precursors were
added as a bolus at the beginning of the growth phase
(Table 1). After two weeks of growth, the cells were
harvested, and loganin was isolated as described in
Section 4. Typically, 45 mg of the monoterpene were
obtained from 100 g of wet cell mass. The remaining
cell mass was subsequently hydrolyzed to yield amino
acids. The 13C labeling patterns of biosynthetic loganin
and selected amino acids were assessed quantitatively
by one- and two-dimensional NMR analysis (Table 2).
Amino acids can be easily isolated after hydrolytic
treatment of isotope-labeled biomass. Their isotopomer
compositions can be determined quantitatively by one-
2.3. Labeling patterns of central intermediates
In order to predict the labeling pattern of DMAPP/
IPP via the mevalonate and the deoxyxylulose
pathway Fig. 2, the labeling patterns of the speci®c
precursors for each of these pathways are required.
Speci®cally, acetyl CoA is the unique precursor of the
mevalonate pathway; glyceraldehyde 3-phosphate and
pyruvate act as precursors of 1-deoxy-D-xylulose 5-
phosphate in the alternative pathway.
and two-dimensional NMR analysis when 13C, 15N, H
2
or any combination of these isotopes are used as labels
in the pro€ered precursor. On basis of known biosyn-
thetic pathways, these data can then be used to recon-
struct the labeling patterns of central intermediary
metabolites.
If a given metabolite can be biosynthesized alterna-
tively via di€erent biosynthetic pathways, the labeling
pattern for each respective pathway can be predicted
The carboxylic group and the a carbon of leucine
(16) are derived from acetyl CoA (3) via condensation
Table 1
Isotope incorporation experiments with cell cultures of Rauwol®a serpentina. The precursors were added as a bolus at the beginning of the
growth phase
Precursor
Concentration (mM)
Unlabeled glucose (mM)
Culture volume (ml)
[U±13C6]glucose
[1±13C]glucose
6.4
165.7
82.9
160.0
0
750
100
30
[2±13C]glucose
83.3
0
[3±13C]glucose
165.7
2.1/1.0
29.7
17
[U±13C5]ribulose/ribose
[3±13C]sodium pyruvate
[2±13C]glycerol
166.7
83.3
72.2
600
300
250
43.0

D. Eichinger et al. / Phytochemistry 51 (1999) 223±236
227
Table 2
13C labeling patterns of loganin from cell cultures of Rauwol®a serpentina grown with 13C-labeled glucoses, pyruvate, glycerol or ribulose/ribose
Position
13C-Abundances and 13C13C couplings
[U±13C6]glucose
[1-13C]-
glucose
[2-13C]-
glucose
[3-13C]-
glucose
[U±13C5]ribulose/ribose
[3-13C]-
pyruvate
[2-13C]-
glycerol
%
13Ca
%
13Ca
%
13C13Cb
%
13Ca
%
13Ca
%
13Ca
%
13Ca
%
13C13Cb
%
13Ca
1
3.5
3.4
3.5
3.2
3.4
3.3
3.1
3.5
3.4
3.3
3.4
3.7
3.8
3.8
3.4
3.5
3.6
60.9(9), 36.1(7)
61.6(1)
28.4
2.3
4.1
4.5
2.3
2.3
2.2
2.1
2.3
2.3
2.3
2.1
2.2
2.1
1.9
1.8
2.1
2.0
1.8
1.8
2.2
39.2(9), 21.1(7)
39.7(1)
1.9
1.8
2.8
6.0
1.7
1.9
1.8
2.7
3.7
4.0
1.9
1.8
1.6
1.6
1.8
1.8
2.8
3.0
9
12.1
11.7
4.1
10.8
9.3
16.0
17.3
3.1
8
56.2(10)
56.6(8)
7.0
30.0(10), 5.0(7)
31.5(8)
5.9(8)
10
7
30.5
4.8
4.7
5.6(8)
58.7(5)
63.0(6)
4.0
4.2
32.8
5.4
2.9
3.5
6
28.5
2.6
38.9(5)
39.9(6)
5
11.9
12.5
3.8
10.5
7.3
16.2
17.4
3.0
4
25.4(3), 28.0(11)
29.0(4)
30.2(4)
6.3
15.4(3), 15.2(11)
15.4(4)
16.0(4)
11
3
17.9
18.4
52.6
2.2
17.6
18.4
4.4
3.8
3.6
3.1
2.8
1'
2'
3'
4'
5'
6'
12
65.3(2')
20.3(2'), 6.3(3')
21.7(1'), 8.5(1' +3')
7.5(2'), 10.4(4')
10.0(3'), 15.7(5')
16.2(6'), 15.2(4' +6')
30.9(5')
45.8(1' and 3'), 21.5(1')
32.7(2' and 4'), 27.9(2' or 4')
31.3(3' and 5'), 34.0(3' or 5')
49.6(4' and 6'), 19.0(6')
66.0(5')
21.6
4.1
9.5
12.5
2.2
3.9
3.2
68.9
26.3
7.2
2.8
9.8
2.3
2.0
12.8
2.5
24.1
21.4
3.1
5.0
3.6
4.9
6.6
a Absolute 13C abundance.
b Percent coupling calculated as percentage of total signal for a given carbon atom involved in 13C13C coupling. Carbons coupled to the re-
spective index carbon are in parentheses.
with a-ketoisovalerate. The labeling pattern of C-1 and
C-2 of the leucine therefore reveals the labeling pattern
of acetyl CoA (Fig. 3). Alanine (17) is biosynthetically
obtained by reductive amination of pyruvate (12).
Hence, the labeling pattern of pyruvate in each re-
spective experiment can be gleaned from that of ala-
nine isolated from cell protein Fig. 3. The side chains
of phenylalanine (18) and tyrosine (19) are derived
from phosphoenol pyruvate (20) via chorismic acid.
Hence, the labeling patterns of the aromatic amino
Fig. 3. Retrobiosynthetic analysis of amino acids, synthesized from central metabolic intermediates. Reconstruction of central labeling patterns is
indicated by the symbol c.

228
D. Eichinger et al. / Phytochemistry 51 (1999) 223±236
Table 3
13C isotopomer composition of acetyl CoA, pyruvate and phosphoenol pyruvate in Rauwol®a serpentina after labeling with 13C glucoses, pyru-
vate, glycerol and ribulose/ribose
Isotopomer
Abundance (mol %)
[U±13C6]glucose
[1-13C]glucose
[2-13C]glucose
[3-13C]glucose
[3-13C]pyruvate
[2-13C]glycerol
[U-13C5]ribulose/ribose
Acetyl CoA
[1-13C]
1.1
1.5
1.2
3.9
25.5
12.0
3.1
9.8
5.8
2.2
15.5
17.2
3.1
1.5
1.3
0.5
[2-13C]
[1,2-13C2]
Pyruvate
[1-13C]
1.9
1.2
1.3
0.1
0.7
1.1
4.5
5.2
6.9
11.5
4.1
31.5
5.6
2.0
1.3
6.8
4.8
17.7
4.0
1.8
1.5
1.6
0.1
0.3
0.3
[2-13C]
[3-13C]
31.1
3.5
[1,2-13C2]
[2,3-13C2]
[1,2,3-13C3]
Phosphoenol pyruvate
[1-13C]
2.6
1.3
1.4
0.1
0.8
1.4
3.6
2.6
3.1
12.8
3.0
NDa
NDa
NDa
1.3
2.2
2.2
2.5
15.9
3.2
1.4
1.5
1.7
0.1
0.4
0.3
[2-13C]
[3-13C]
28.9
[1,2-13C2]
[2,3-13C2]
[1,2,3-13C3]
a ND means not determined.
acid side chains must re¯ect that of phosphoenol pyru-
vate (Fig. 3).
pyruvate and phosphoenol pyruvate from totally
labeled precursors (Table 3).
Reconstructed isotopomer compositions for acetyl
CoA, pyruvate and phosphoenol pyruvate from exper-
iments using 13C labeled specimens of glucose, ribu-
lose/ribose, pyruvate and glycerol are summarized in
Table 3. Even in the experiments with single labeled
precursors, the acetyl CoA, pyruvate and phosphoenol
pyruvate pool consisted of a complex mixtures of iso-
topomers. This is easily explained because the general
precursors used in this study can be recycled in the
central compartments of intermediary metabolism.
Similarly, the multiply labeled precursors [U±
13C6]glucose and [U±13C5]ribulose/-ribose a€orded
complex isotopomer mixtures including various multi-
ply labeled molecular species. Universally 13C labeled
C2- and C3-fragments can be generated metabolically
from the breakdown of totally labeled precursors and
can be diverted to downstream metabolites. On the
other hand, the recycling of universally labeled frag-
ments through central metabolic pathways can result
in breaking of CC bonds. Subsequently, the resulting
fragments can be used for the regeneration of metab-
olites such as pyruvate and phosphoenol pyruvate.
This will usually involve the recombination of frag-
ments from both labeled and unlabeled precursor mol-
ecules, and this explains the formation of single-
labeled and double-labeled as well as triple-labeled
2.4. Labeling patterns of loganin
The mevalonate pathway implies the formation of
IPP from three acetate moieties (Fig. 2). On basis of
the reconstructed acetyl CoA labeling patterns (Table
3), the hypothetical labeling pattern of IPP via the
mevalonate pathway can be predicted quantitatively,
as shown in Fig. 4c for the experiment with the mix-
ture of [U±13C6]glucose and unlabeled glucose.
Speci®cally, the reconstructed pattern of acetyl CoA is
characterized by a signi®cant contribution from a dou-
bly 13C labeled isotopomer. Following the mevalonate
pathway, contiguously 13C labeled atoms would then
be present in C1/C2 and C3/C5 of IPP and DMAPP
(Fig. 4c).
The formation of the loganin ring system has been
studied in considerable detail (for review see
Banthorpe et al., 1972; Stockigt, 1979). The committed
precursor, geranyl pyrophosphate, biosynthesized from
one molecule each of IPP and DMAPP, is converted
via cyclization and oxidation of the intermediate (4)
(Fig. 1) into the cyclopentane intermediate (5) which is
in rapid equilibrium with its tautomeric form (7) via
(6). As a consequence, the terminal enolic carbon of
Ã
(5) as well as that of (7) (labeled with . and in Fig.
1) participate in the bond making process yielding the

D. Eichinger et al. / Phytochemistry 51 (1999) 223±236
229

230
D. Eichinger et al. / Phytochemistry 51 (1999) 223±236
hemiacetal ring system in (8) (Escher et al., 1970).
Thus, the original methyl groups of the DMAPP unit
become randomized with respect to their label distri-
bution. In other words, the carbon atoms 3 and 11 of
loganin (Fig. 1) become indistinguishable with respect
to their 13C-abundance. On basis of these established
mechanisms, IPP contributes the carbon atoms 1, 7, 8,
9 and 10 of loganin, and DMAPP contributes the
loganin carbons 3, 4, 5, 6 and 11 of loganin (9, Fig. 1).
Following this dissection the labeling patterns of loga-
nin are now predicted from the labeling patterns of the
C5 building blocks as shown for the experiment with
[U±13C6]glucose in Fig. 4c. The hypothetical labeling
pattern of loganin biosynthesized via the mevalonate
pathway is then characterized by blocks of 13C atoms
including the carbon pairs C1/C9, C8/C10, C5/C6, C3/
C4, and C4/C11 (shown by bold lines in Fig. 4c).
Using this isotopomer pattern, hypothetical 13C NMR
signals of loganin can now be simulated. For an
example, the 13C NMR signal of C-1 of loganin
(indicated by an arrow in Fig. 4) biosynthesized via
mevalonate would re¯ect a [1-13C1]- and a [1,9-13C2]-
isotopomer resulting in a singlet and a doublet
substructure of the 13C NMR signal, respectively
(Fig. 4c).
As shown in Fig. 4, the observed labeling patterns
of loganin and its IPP building block, as well as the
observed 13C NMR signal of C-1 is in almost perfect
agreement to the predictions via the deoxyxylulose
pathway.
Starting from various 13C-labeled samples of glu-
cose, ribose/ribulose, pyruvate or glycerol and using
the retrobiosynthetic concept described above in detail
the predicted labeling patterns of IPP formed via the
deoxyxylulose pathway are in close agreement with the
experimentally determined IPP labeling patterns in
loganin (Fig. 5). On the other hand, the predictions
via the mevalonate pathway are clearly at odds with
the observed labeling patterns. It follows that, if not
completely, at least the major fraction of loganin
(>95%) has been obtained via the deoxyxylulose path-
way in the experimental system used.
Since loganin is undoubtedly the precursor of seco-
loganin (Battersby, Burnett,
&
Parsons, 1969;
Tanahashi, Nagakura, Inouye, & Zenk, 1984) one can
assume that all monoterpenoid indole alkaloids as well
as the dopamine derived isoquinoline alkaloids of the
emetine type are biosynthesized via condensation of
pyruvate and glyceraldehyde-3-phosphate.
This experimental approach is now used for a quan-
titative prediction of the isotopomer pattern of loganin
biosynthesized via the deoxyxylulose pathway. A cen-
tral intermediate in the novel terpenoid pathway is 1-
deoxy-D-xylulose 5-phosphate (14) which is assembled
from glyceraldehyde 3-phosphate (13) and pyruvate
(12) (Fig. 2). The labeling pattern of glyceraldehyde 3-
phosphate is not directly represented in the proteolytic
amino acids. However, since phosphoenol pyruvate is
obtained from glyceraldehyde 3-phosphate (13) by
dehydrogenation, isomerization and dehydration, its
labeling pattern can serve as a substitute for the elusive
glyceraldehyde pattern (Fig. 3). The hypothetical label-
ing pattern of IPP formed via the deoxyxylulose path-
way is therefore reconstructed from the labeling
patterns of pyruvate and glyceraldehyde 3-phosphate.
Speci®cally, a signi®cant contribution from [1,2,3-
13C3]glyceraldehyde 3-phosphate is expected in the iso-
topomer pattern of compound 14 yielding [3,4,5-
13C3]1-deoxy-D-xylulose 5-phosphate. Due to the re-
arrangement process yielding 2-C-methyl-D-erythritol
4-phosphate (15) (Fig. 2) the formation of [1,2,4-
13C3]IPP would result, which is subsequently converted
into [1,7,9-13C3]loganin.
3. Discussion
Recent studies show that the universal terpenoid
precursors, IPP and DMAPP, can be biosynthesized
via two di€erent pathways in higher plants. There is
indirect evidence that the mevalonate pathway is oper-
ative in the cytoplasm, and the deoxyxylulose pathway
is operative in plastids. However, the compartmental
separation of the two pathways is not absolute.
Exchange of metabolites between the two pathways
occurs to a variable degree in di€erent experimental
systems (Arigoni et al., 1997). The crosstalk between
the two pathways is not understood in detail. The
speci®c metabolite or metabolites involved in the sup-
posed intercompartmental exchange process are
unknown.
In light of these recent ®ndings, it is easily under-
stood that earlier studies documented the incorpor-
ation of mevalonate into virtually all plant terpenoids
studied. The low incorporation rates of mevalonate
observed in many experiments were explained by poor
uptake of the isotope-labeled precursor and compart-
mentalization of metabolic pathways. Direct exper-
iments to determine the uptake rate were not
performed and, indeed, would have been dicult to
perform with sucient accuracy. In retrospect, it
appears that in many cases the observed incorporation
of mevalonate were due to the crosstalk between the
pathways and did not re¯ect the mainstream metab-
olite ¯ux.
As described above, hypothetical 13C NMR signals
are now simulated from the predicted labeling patterns
(Fig. 4b). Most notably, the [1,7,9-13C3]loganin isoto-
pomer in loganin causes the appearance of a double
doublet substructure in the 13C NMR signal of C-1
re¯ecting simultaneous 13C coupling to C-9
(1JCC = 43.5 Hz) and C-7 (3JCC =4.15 Hz).

D. Eichinger et al. / Phytochemistry 51 (1999) 223±236
231
Fig. 5. Observed and predicted isotopomer patterns of the IPP moiety in loganin after growing cell cultures of Rauwol®a serpentina with various
stable isotope labeled precursors. Predictions of the IPP labeling pattern were made on basis of the labeling data of pyruvate and phosphoenol
pyruvate (deoxyxylulose pathway) or acetate (mevalonate pathway), as reconstructed from amino acids.
In recent studies, isotope-labeled 1-deoxy-D-xylulose
has been used as a precursor to study the biosynthetic
origin of terpenoids. These studies are faced with
similar problems as the older work with mevalonate.
The uptake of deoxyxylulose into the cell compart-
ments involved in the biosynthesis of a speci®c
terpenoid is unpredictable. For example, the carbo-
hydrate is eciently utilized for biosynthesis of ubiqui-
nones in E. coli (Broers, 1994), of 2-C-methyl-D-
erythritol in leaves of Liriodendron tulipifera (Sagner et
al., 1998), and of carotenoids in cell cultures of
Catharanthus roseus (Arigoni et al., 1997). On the
other hand, 1-deoxy-D-xylulose was not signi®cantly
incorporated into taxoids in cells of Taxus chinensis
(Menhard, 1998) and in cultures of Methylobacterium
organophilum, Corynebacterium ammoniagenes (T.
Duvold, P. Cali and M. Rohmer, unpublished work,
quoted in Disch, Schwender, Muller, Lichtenhaler, &

232
D. Eichinger et al. / Phytochemistry 51 (1999) 223±236
Rohmer, 1998) and Ochromonas danica (C. Muller
and H.K. Lichtenthaler, unpublished work, quoted in
Disch et al., 1998) although it has been shown by
other experiments that the terpenoids under study
were assembled predominantly from isoprenoid precur-
sors derived from the deoxyxylulose pathway. Thus, 1-
deoxy-D-xylulose and mevalonate may not be suitable
as exogenous precursor for labeling experiments in
order to examine the partitioning between the two ter-
penoid pathways in plants and plant cell cultures.
The retrobiosynthetic concept described in this
article is a universal method for analysis of metabolite
¯ux. Any labeled compounds which can be diverted to
all major areas of cellular metabolism can be used as
precursor for this method. The labeling patterns of pri-
mary metabolites (amino acids, nucleosides) are deter-
mined and are used to reconstruct the isotopomer
patterns of central metabolites on basis of known bio-
synthetic pathways. These procedures have been
reviewed recently (Bacher et al., 1998) and are there-
fore not described in detail.
In order to assess the ¯ux contributions of di€erent
pathways for the formation of given compounds which
can be obtained via more than one route, the isotopo-
mer pattern is predicted quantitatively for each of the
possible pathways from the labeling patterns of the
central metabolites (Fig. 5). The predicted isotopomer
patterns are then compared to the experimentally
determined label distribution in the metabolite under
study. In the present case, the labeling patterns of
loganin are in excellent agreement with the deoxyxylu-
lose prediction (Fig. 5). This shows that the bulk
amount of loganin is biosynthesized via the deoxyxylu-
lose pathway.
Earlier reports suggesting the incorporation of
mevalonate into loganin in low yields (Battersby et al.,
1968; Escher et al., 1970; Guarnaccia et al., 1974)
remain valid, even in light of the data reported in this
paper, but they need to be reinterpreted. The low in-
corporation rate found in the old experiments were
not (or at least not exclusively) a consequence of poor
metabolite uptake. Instead, they represent a small
amount of loganin formed from precursors contributed
via the mevalonate pathway via crosstalk between the
pathways.
In principle, each of the seven incorporation exper-
iments in the present study would have been sucient
to prove that the formation of loganin in R. serpentina
cells occurs predominantly via the deoxyxylulose
branch of terpenoid metabolism. The parallel exper-
iments with di€erent metabolites and/or label positions
in a given precursor were performed to check the gen-
eral validity of the retrobiosynthetic concept. Indeed,
the data show that each of the precursor species used
yields the same conclusion (Fig. 5). The close agreement
between the experiments actually suggests that every
labeling strategy conducive to the diversion of label to
all metabolic branches, but not to a rigorously stochas-
tic label distribution (equivalent to a maximum of the
isotopomer entropy term) can be used for the analysis
of metabolic partitioning based on analysis of isotopo-
mer patterns. Thus, precursors can be selected on basis
of price, availability and eciency of metabolic utiliz-
ation rather than of their speci®c metabolic proximity
to the intermediate of the respective pathway.
4. Experimental
4.1. Chemicals
[U±13C6]-, [1-13C]-, [2-13C]-, [3-13C]glucose, [3-
13C]sodium pyruvate and [2-13C]glycerol were pur-
chased from Omicron Biochemicals, Inc. (South Bend,
IN). Other chemicals were of the highest purity avail-
able.
4.2. Preparation of [U±13C5]ribulose/ribose
A reaction mixture containing 100 mM Tris hydro-
chloride (pH 7.8), 6 mM dithiothreitol, 60 mM MgCl2,
80 mM ATP, 200 mM ammonium acetate, 200 mM a-
ketoglutarate, 5 mM NADP+, 1 g of [U±13C6]glucose,
250 U of hexokinase, 250 U of glucose 6-phosphate
dehydrogenase and 250 U of glutamate dehydrogenase
in a total volume of 105 ml was incubated at 378C.
The pH of the solution was kept at pH 7.8 by adding
of 3 M potassium hydroxide at intervals. The reaction
was monitored by 13C NMR spectroscopy. When the
NMR signals of the intermediate, 6-phosphogluconate
(C-1, 178.7 ppm; C-2, 74.1 ppm), had disappeared, 200
U of pentose isomerase were added, and the mixture
was incubated for 2 h. To the resulting equilibrium
mixture of [U±13C5]ribulose- and [U±13C5]ribose phos-
phate (2:1), 15 ml of 1 M barium chloride were added,
and the suspension was centrifuged. The supernatant
was diluted with 420 ml of ethanol and was kept at
48C for 24 h. The precipitate of carbohydrate phos-
phate barium salts was harvested by centrifugation
and redissolved in 100 ml of water. Six ml of 1 M
(NH4)2SO4 and 4 ml of 1 M K2SO4 were added and
the precipitate of BaSO4 was removed by centrifu-
gation. The supernatant was concentrated at 358C
under reduced pressure to a small volume (80 ml).
After addition of 2.5 ml of 100 mM MgCl2 and 2.5 ml
of 100 mM ZnCl2, the pH value was adjusted to 7.9
with 1 M NaOH. Alkaline phosphatase (50 U) was
added. The mixture was incubated at 378C for 4 h and
lyophilized.
The dry powder was dissolved in 40 ml of water,
and the solution was applied on a column (2.2 Â 43
cm) containing a mixture of 31.2 g of DOWEX

D. Eichinger et al. / Phytochemistry 51 (1999) 223±236
233
50W Â 8 (H+-form) and 42.9 g of DOWEX 1 Â8
(OH -form). The column was developed with water.
The e‚uent (450 ml) was collected, concentrated and
lyophilized yielding 300 mg of a 2:1 mixture of [U±
13C5]ribulose/ribose (total yield 37%).
supernatant was applied on a column of Amberlite
XAD-2 (2 Â 10 cm). The column was washed with 250
ml of water. Loganin was eluted with 150 ml of metha-
nol. The e‚uent was concentrated under reduced
pressure to a volume of 10 ml.
Crude loganin was puri®ed by preparative reversed-
phase HPLC using a column of Nucleosil RP18
(16 Â 250 mm). The column was developed with 40
vol% methanol (¯ow rate, 8 ml/min). The e‚uent was
monitored photometrically (254 nm). The retention
volume of loganin was 55 ml. Fractions containing
loganin were combined and concentrated to dryness
under reduced pressure.
4.3. Culture medium
Suspension cultures of R. serpentina were grown in
LS medium (Linsmaier & Skoog, 1965) containing 1
mM 2,4-dichlorophenoxyacetic acid and
1
mM
naphthylacetic acid. Alternatively, the cells were grown
in AP medium (Zenk et al., 1977) containing 1 mM 3-
indolylacetic acid, 5 mM N6-benzyladenine and 20 mM
MgSO4. Sucrose was replaced by D-glucose in both
culture media used (LS medium, 15 g D-glucose per
liter; AP medium, 30 g D-glucose per liter, unless
otherwise stated). The pH value of the media was
adjusted to 6.0 with 0.1 M NaOH prior to steriliza-
tion.
4.7. Isolation of amino acids
Methanol-extracted cells were treated for 24 h at
1208C with 150 ml of 6 M HCl containing 0.5 M thio-
glycolic acid. The mixture was ®ltered and the super-
natant was evaporated to dryness. Amino acids were
isolated from the residue as described earlier
(Eisenreich, Schwarzkopf, & Bacher, 1991).
4.4. Cell culture
Cells of R. serpentina were grown in 250-ml ¯asks
containing 130 ml of LS medium. After 7 days of
growth, 60 ml of cell suspension were added to 70 ml
of LS medium. The cultures were incubated on a
rotary shaker (130 rpm) at 218C under a ¯uorescent
tube (40 W, Duro Lite Inc., Midland Park, NJ).
For isotope incorporation experiments, 20 g of wet
cell mass grown on LS medium for 7 days were sus-
pended in 100 ml of modi®ed AP-medium. 13C-labeled
nutrients were added as sterile solutions (Table 1).
After a growth period of 18 days, the cells were har-
vested, frozen in liquid nitrogen and stored at 208C.
4.8. NMR-spectroscopy
1H and 13C NMR spectra were recorded at 500 and
125 MHz, respectively, using a Bruker DRX 500 spec-
trometer. One- and two-dimensional experiments
(DQF-COSY, NOESY, HMBC and INADEQUATE)
were performed at 278C using standard Bruker soft-
ware (XWINNMR 1.3). Loganin was measured in
D2O as solvent. Amino acids were measured in D2O at
pH 1 or 13.
4.9. NMR signal assignment
4.5. High performance liquid chromatography
The correct interpretation of 13C labeling exper-
iments by quantitative NMR spectroscopy requires
unequivocal assignments of H and 13C NMR signals.
1
HPLC analysis of loganin was performed with a
column of Nucleosil RP18 (4.55 Â 250 mm) and an
eluent containing 40 vol% methanol in water (¯ow
rate, 1 ml/min). The e‚uent was monitored photo-
metrically (254 nm). The retention volume of loganin
was 9.5 ml.
NMR signals of loganin were assigned earlier on basis
of chemical shift arguments (Heckendorf, Mattes,
Hutchinson, Hagaman, & Wenkert, 1976; Jensen,
Lyse-Petersen, & Nielsen, 1979; Calis, Lahloub, &
Sticher, 1984). In order to obtain unequivocal NMR
signals assignments, we performed a comprehensive
NMR analysis by two-dimensional homo- and hetero-
correlation experiments (Table 4). The 13C13C coupling
patterns of loganin from the incorporation experiment
with [U±13C6]glucose and [U±13C5]ribulose/ribose gave
additional con®rmation of the 13C NMR assignments
via INADEQUATE experiments.
4.6. Isolation of loganin
R. serpentina wet cell mass (100 g) was suspended in
50 ml of water and homogenized with an Ultra Turrax
(10000 U/min, 5 min). The mixture was freeze-dried
(3.5 g). Lyophilized plant material was extracted twice
with 300 ml of methanol at room temperature for 24
h. The extracts were combined and concentrated under
reduced pressure to a volume of 20 ml; 180 ml of
water were added and insoluble material was removed
by centrifugation (15 min, 7500 U/min, 48C). The
4.10. Quantitative analysis of 13C labeling patterns
Absolute 13C abundance was determined for each
carbon atom of loganin from the feeding experiments

234
D. Eichinger et al. / Phytochemistry 51 (1999) 223±236

D. Eichinger et al. / Phytochemistry 51 (1999) 223±236
235
Battersby, A. R., Byrne, J. C., Kapil, R. S., Martin, J. A., Payne, T.
G., Arigoni, D., Loew, P. (1968). J. Chem. Soc. Chem. Commun.,
951±953.
by
quantitative
NMR
&
spectroscopy
(Strauû,
Eisenreich, Bacher,
Fuchs, 1992). Brie¯y, 1H
decoupled 13C NMR spectra of the biosynthetic
samples and of samples with natural 13C abundance
(1.1% 13C) were measured under identical conditions.
This standardization considers di€erent relaxation
behavior of carbon atoms and prevents false quanti®-
cation of 13C abundances due to di€erences in relax-
ations times. Relative 13C abundance of individual
carbon atoms was then calculated from the integrals of
biosynthetic samples by comparison with the natural
abundance samples. In order to obtain absolute 13C
Bloch, K. (1992). Steroids, 57, 378±382.
Brechbuhler-Bader, S., Coscia, C. J., Loew, P., v. Sczepanski, C.,
Arigoni, D. (1968). J. Chem. Soc. Chem. Commun., 136±137.
Broers, S. T. J. (1994). Thesis No. 10978, ETH Zurich, Switzerland.
Calis, I., Lahloub, M. F., & Sticher, O. (1984). Helv. Chim. Acta, 67,
160±165.
Campos, N., Lois, L. M., & Boronat, A. (1997). Plant Physiol., 115,
1289.
Disch, A., Schwender, J., Muller, C., Lichtenthaler, H. K.,
Rohmer, M. (1998). Biochem. J., 333, 381±388.
&
Eisenreich, W., Sagner, S., Zenk, M. H., & Bacher, A. (1997).
Tetrahedron Lett., 38, 3889±3892.
abundances, the 13C satellites in the H NMR of each
1
Eisenreich, W., Schwarz, M., Cartayrade, A., Arigoni, D., Zenk,
M. H., & Bacher, A. (1998). Chem. Biol., 5, 221±233.
Eisenreich, W., Schwarzkopf, B., & Bacher, A. (1991). J. Biol.
Chem., 266, 9622±9631.
metabolite were analyzed yielding absolute 13C abun-
dance values for selected carbon atoms. The relative
abundances were subsequently referenced to these re-
spective carbon atoms (Eisenreich, Sagner, Zenk, &
Bacher, 1997).
Escher, S., Loew, P., Arigoni, D. (1970). J. Chem. Soc. Chem.
Commun., 823±825.
Flesch, G., & Rohmer, M. (1988). Eur. J. Biochem., 175, 405±411.
Guarnaccia, R., Botta, L., & Coscia, C. J. (1974). J. Am. Chem.
Soc., 96, 7079±7084.
4.11. Analysis of isotopomer composition
Heckendorf, A. H., Mattes, K. C., Hutchinson, C. R., Hagaman,
E. W., & Wenkert, E. (1976). J. Org. Chem., 41, 2045±2047.
The fraction of multiply labeled isotopomers was
calculated from H-decoupled 13C NMR spectra as the
1
Jensen, S. R., Lyse-Petersen, S. E.,
Phytochemistry, 88, 273±277.
& Nielsen, B. J. (1979).
fraction of 13C13C coupled satellites in the global 13C
NMR integral of the respective carbon atom.
Multiplication with the absolute 13C abundance value
of the corresponding position yielded the fraction of
multiply labeled isotopomers in mol%.
Lange, B. M., Wildung, M. R., McCaskill, M. R., & Croteau, D.
(1998). Proc. Natl. Acad. Sci. USA, 95, 2100±2104.
Linsmaier, E. M., & Skoog, F. (1965). Physiol. Plant., 18, 100±127.
Lois, L. M., Campos, N., Putra, S. R., Danielsen, K., Rohmer, M.,
& Boronat, A. (1998). Proc. Natl. Acad. Sci. USA, 95, 2105±
2110.
Menhard, B. (1998). Biosynthesis of taxans and their elicitation.
International Symposium Antitumor Products from Higher
Plants, Paris, January 8±10, 1998.
4.12. NMR spectra simulation
1H- and 13C-coupling patterns were simulated with
the program package NMRSIM (Bruker).
Qureshi, N., & Porter, J. W. (1981). In: J. W. Porter, & S. L.
Spurgeon. Biosynthesis of isoprenoid compounds, (Vol. 1, pp. 47±
94). New York: John Wiley.
Ramos-Valdivia, A., van der Heijden, R., & Verpoorte, R. (1997).
Nat. Prod. Rep., 14, 591±603.
Acknowledgements
Rohmer, M. (1998). Prog. Drug. Res., 50, 135±154.
Rohmer, M., Knani, M., Simonin, P., Sutter, B., & Sahm, H. (1993).
Biochem. J., 295, 517±524.
This work was supported by the Deutsche
Forschungsgemeinschaft (SFB 369) and the Fonds der
chemischen Industrie. We thank Professor Duilio
Arigoni for many helpful discussions and Angelika
Werner and Fritz Wendling for help with the prep-
aration of the manuscript.
Sagner, S., Eisenreich, W., Fellermeier, M., Latzel, C., Bacher, A., &
Zenk, M. H. (1998). Tetrahedron Lett., 39, 2091±2094.
Schwarz, M. K. (1994). Thesis No. 10951, ETH Zurich, Switzerland.
Schwender, J., Lichtenthaler, H. K., et al. (1997). FEBS Lett., 414,
129±134.
Shanks, J. V., Bhadra, R., Morgan, J., & Rijhwani, S. (1998).
Biotechnol. Bioeng., 58, 333±338.
Sprenger, G. A., Sahm, H., et al. (1997). Proc. Natl. Acad. Sci. USA,
94, 12857±12862.
Stockigt, J. (1979). Habilitation. Germany: Ruhruniversitat Bochum.
Strauû, G., Eisenreich, W., Bacher, A., & Fuchs, G. (1992). Eur. J.
Biochem., 205, 853±866.
References
Takahashi, S., Kuzuyama, T., Watanabe, H., & Seto, H. (1998).
Proc. Natl. Acad. Sci. USA, 95, 9879±9884.
Arigoni, D., Sagner, S., Latzel, C., Eisenreich, W., Bacher, A., &
Zenk, M. H. (1997). Proc. Natl. Acad. Sci. USA, 94, 10600±
10605.
Tanahashi, T., Nagakura, N., Inouye, H., & Zenk, M. H. (1984).
Phytochemistry, 23, 1917±1922.
Bacher, A., Rieder, C., Eichinger, D., Arigoni, D., Fuchs, G., &
Eisenreich, W. (1998). FEMS Microbiol. Rev., in press.
Banthorpe, D. V., Charlwood, B. V., & Francis, M. J. O. (1972).
Chem. Rev., 72, 115±155.
Uesato, S., Kanomi, S., Iida, A., Inouye, H., & Zenk, M. H. (1986).
Phytochemistry, 25, 839±842.
Verpoorte, R., van der Heijden, R., & Moreno, P. R. H. (1997). In:
G. A. Cordell. The alkaloids (Vol. 49, pp. 221±299). New York:
Academic Press.
Battersby, A. R., Burnett, A. R., Parsons, P. G. (1969). J. Chem.
Soc. (C), 1187±1192.

236
D. Eichinger et al. / Phytochemistry 51 (1999) 223±236
Werner, I., Bacher, A., & Eisenreich, W. (1997). J. Biol. Chem., 272,
25474±25482.
Zenk, M. H., El-Shag,i Arens, H., Stockigt, J., Weiler, E. W., &
Deus, B. (1977). In: W. Barz, E. Reinhard, & M. H. Zenk. Plant
tissue culture and its biotechnological application (pp. 27). Berlin:
Springer.
Whitmer, S., Canel, C., Hallard, D., Goncalves, C., & Verpoorte, R.
(1998). Plant Physiol., 116, 853±857.