Metabolic Modifications of Birch Leaf Phenolics by an Herbivorous Insect: Detoxification of Flavonoid Aglycones via Glycosylation

Abstract

The metabolic modifications of birch (Betula pubescens Ehrh.) leaf phenolics in the digestive tract of its major defoliator, larvae of the autumnal moth Epirrita autumnata, were studied. The main phenolic acids of birch, i.e. chlorogenic and p-coumaroylquinic acids, were isomerised in the alkaline digestive tract. Moreover, only 16 to 92% of the ingested amounts of chlorogenic acid were found in the faeces of individual larvae; the missing portion is possibly being used in the formation of reactive o-quinones. Water-soluble flavonoid glycosides were mostly excreted unaltered. In contrast, lipophilic flavonoid aglycones were not excreted as such, but as glycosides after being detoxified by E. autumnata via glycosylation. When the larvae were fed with leaf-painted acacetin and kaempferide, i.e. two naturally occurring birch leaf flavonoid aglycones, acacetin-7-O-glucoside and kaempferide-3-O-glucoside appeared in larval faeces as major metabolites. However, the efficiency of aglycone glycosylation varied-, ranging from 17 to 33%, depending on the aglycone and its dietary level. There was also large variation in the efficiency of glycosylation--from 2 to 57%--among individual larvae. These results demonstrate a compound-specific metabolism of phenolic compounds, leading to different phenolic profiles in the insect gut compared to its leaf diet.

Full text

Metabolic Modifications of Birch Leaf Phenolics by an Herbivorous Insect:
Detoxification of Flavonoid Aglycones via Glycosylation
Juha-Pekka Salminen
a,*
, Maria Lahtinen
a
, Kyösti Lempa
b
, Lauri Kapari
b
,
Erkki Haukioja
b
, and Kalevi Pihlaja
a
a
Laboratory of Environmental Chemistry, Department of Chemistry, University of Turku,
FIN-20014 Turku, Finland. Fax: +3 58-2-3336700. E-mail: j-p.salminen@utu.fi
b
Section of Ecology, Department of Biology, University of Turku, FIN-20014 Turku, Finland
* Author for correspondence and reprint requests
Z. Naturforsch. 59 c, 437Ð444 (2004); January 28/February 27, 2004
The metabolic modifications of birch (Betula pubescens Ehrh.) leaf phenolics in the diges-
tive tract of its major defoliator, larvae of the autumnal moth Epirrita autumnata, were
studied. The main phenolic acids of birch, i.e. chlorogenic and p-coumaroylquinic acids, were
isomerised in the alkaline digestive tract. Moreover, only 16 to 92% of the ingested amounts
of chlorogenic acid were found in the faeces of individual larvae; the missing portion is
possibly being used in the formation of reactive o-quinones. Water-soluble flavonoid glyco-
sides were mostly excreted unaltered. In contrast, lipophilic flavonoid aglycones were not
excreted as such, but as glycosides after being detoxified by E. autumnata via glycosylation.
When the larvae were fed with leaf-painted acacetin and kaempferide, i.e. two naturally
occurring birch leaf flavonoid aglycones, acacetin-7-O-glucoside and kaempferide-3-O-gluco-
side appeared in larval faeces as major metabolites. However, the efficiency of aglycone
glycosylation varied, ranging from 17 to 33%, depending on the aglycone and its dietary
level. There was also large variation in the efficiency of glycosylation Ð from 2 to 57% Ð
among individual larvae. These results demonstrate a compound-specific metabolism of phe-
nolic compounds, leading to different phenolic profiles in the insect gut compared to its
leaf diet.
Key words: Phenolic Metabolism, Flavonoid Aglycones, Glycosylation
Introduction
Phenolic compounds have been considered for
a long time to be important chemical defences of
deciduous trees against insect herbivores. More
than 30 years have now elapsed since Paul Feeny’s
pioneering work (Feeny and Bostock, 1968; Feeny,
1970) showed that the phenolic composition of
oak leaf foliage correlated negatively with the per-
formance of an oak leaf feeding herbivore. How-
ever, the rather crude chemical methods used in
the sixties and seventies Ð e.g. measuring only to-
tal phenolics or total contents of phenolic sub-
groups like hydrolysable or condensed tannins Ð
have only partially been replaced by more accu-
rate measures like HPLC (e.g. Grayer et al., 1994;
Kause et al., 1999; Ossipov et al., 2001; Henriksson
et al., 2003; Tikkanen and Julkunen-Tiitto, 2003).
But even when using more detailed chemical
methods, often correlations between foliar pheno-
lics and insect performance have been weak or in-
consistent between different experiments (Hau-
kioja, 2003). This may depend on a number of
0939Ð5075/2004/0500Ð0437 $ 06.00 ” 2004 Verlag der Zeitschrift für Naturforschung, Tübingen · http://www.znaturforsch.com ·
D
reasons like, for instance, on interactions between
phenolics and other, possibly even non-measured
leaf constituents, or on variable fates of individual
phenolics in digestive tracts of individual insects
included in the correlative studies. Furthermore,
the possibilities for good correlations may have
been destroyed by pooling together individual
compounds, which are processed differently by the
insect (Salminen and Lempa, 2002).
Recently, we reported different fates of indivi-
dual birch (Betula pubescens) leaf hydrolysable
tannins in the guts of larvae of the geometrid moth
Epirrita autumnata, the main defoliator of birch
(Salminen and Lempa, 2002). The aim of this
study was to investigate in detail how the rest of
the main HPLC-detectable birch leaf phenolics Ð
i.e. chlorogenic and p-coumaroylquinic acid, flavo-
noid glycosides and aglycones Ð are modified by
E. autumnata larvae. To study the metabolism of
phenolics would be relatively easy using artificial
diets and commercially available standards (e.g.
Martin et al., 1987; Barbehenn and Martin, 1992,
1994; Barbehenn et al., 1996; Zimmer, 1999), but

438 J.-P. Salminen et al. · Metabolic Modifications of Phenolics by an Herbivorous Insect
since these diets lack e.g. the foliar oxidants, sur-
factants and the other naturally occurring com-
pounds, it would be difficult to estimate how well
these kinds of “artificial” studies are able to reveal
the “natural” metabolic modifications of phenolic
compounds within insect guts. Therefore meta-
bolic studies herein were done with natural leaf
diets only. A particular interest was paid to the
metabolism of flavonoid aglycones, a group of
phenolics recently found on the lipophilic surface
of birch leaves (Keinänen and Julkunen-Tiitto,
1998; Valkama et al., 2003). By feeding E. autum-
nata with two leaf-painted flavonoid aglycones of
birch, i.e. acacetin (5,7-dihydroxy-4⬘-methoxyfla-
vone) and kaempferide (3,5,7-trihydroxy-4⬘-me-
thoxyflavonol), we wanted to find out if even the
difference of a single hydroxyl group between the
two phenolics makes a difference to their metabo-
lism.
Materials and Methods
Monitoring metabolic modifications of birch
phenolics by Epirrita autumnata larvae
10 larvae of the autumnal moth, E. autumnata,
from three different broods of southern Finnish
origin were randomly allocated to control leaves
originated from a single mature white birch (Be-
tula pubescens Ehrh.) growing in the botanical gar-
den of the University of Turku (the same tree as
used by Salminen and Lempa, 2002). The larvae
were reared under a regime of ambient temper-
ature (approx. 0 ∞Cto23∞C) and light (approx.
16 h light/8 h dark), and the faeces of individual
larvae were collected on a daily basis during the
4
th
and 5
th
larval instars (21
st
MayÐ4
th
June 2000).
The combined faeces produced by each individual
larva was freeze-dried and extracted with 70%
acetone (containing 0.1% ascorbic acid). The phe-
nolic composition of the faeces extracts was com-
pared to that extracted from the foliage of the ex-
perimental tree. For further details, see Salminen
and Lempa (2002).
The 72-h bioassay with leaf-painted flavonoid
aglycones and Epirrita autumnata larvae
To study in more detail the fates of individual
flavonoid aglycones of birch leaf surface in the di-
gestive tract of E. autumnata, we conducted an ad-
ditional 72-h bioassay with the most voracious 5
th
instar E. autumnata larvae (27
th
Ð30
th
May 2002).
The larvae were fed with commercially available
acacetin and kaempferide at two levels, 5 and
10 mg/g leaf dry weight. These levels were chosen
since they are close to the average contents of aca-
cetin and kaempferide on the surface of young
white birch leaves (Valkama et al., 2003; kaempfer-
ide preliminarily identified as flavonol methyl
ether). The compounds were dissolved into 90%
acetone and known volumes were painted onto
the surface of fresh birch leaves (originated from
two mature Betula pubescens trees growing in the
botanical garden of the University of Turku) to
approximate the desired content. Details of the
painting procedure have been described earlier by
Salminen and Lempa (2002). The only exception
to their method was that before painting the
leaves were washed to remove the flavonoids nat-
urally occurring on the birch leaf surface. Washing
was conducted by immersing the leaves in 50 ml
of 95% ethanol for 10 s and subsequently in 50 ml
of water for 5 s to remove the traces of ethanol.
The excess water was gently removed by keeping
the leaf between two pieces of filter paper. This
procedure removed the lipophilic substances Ð
like flavonoid aglycones Ð dissolved in the sticky
surface of leaves, as noticed by the lack of sticki-
ness in the washed leaves compared to unwashed
ones. Three kinds of leaves were used as controls:
(1) non-washed and non-painted, (2) washed but
non-painted, and (3) washed and painted with
90% acetone. The bioassay included 84 larvae of
E. autumnata, from four different broods of south-
ern Finnish origin. 12 larvae were randomly allo-
cated to each level of painted compounds and each
type of controls. Larvae and leaves were placed in
20-ml vials at + 12 ∞C and approx. 16 h light/8 h
dark. The bioassay was a no-choice experiment:
the larvae were placed individually in vials and re-
ceived only one type of food. Leaves were re-
placed with fresh ones after 24 and 48 h. At the
same time, and after 72 h, the faeces of individual
larvae was collected, freeze-dried and weighed.
Larval leaf consumption during the experiment
was measured as described by Salminen and
Lempa (2002).
Sample preparation and HPLC-DAD/
HPLC-ESI-MS analysis
Leaf samples were collected in the beginning of
each bioassay to monitor the foliar phenolic com-
position of the experimental trees. Leaf sample
collection, sample preparation and extraction fol-

J.-P. Salminen et al. · Metabolic Modifications of Phenolics by an Herbivorous Insect 439
lowed Salminen et al. (2001). In short, freeze-dried
leaves were extracted with 70% acetone (contain-
ing 0.1% ascorbic acid). After freeze-drying of the
extracts, water-soluble phenolics were dissolved
into 3 ¥ 2 ml water; the lipophilic residues of the
extracts were then dissolved into 3 ¥ 1 ml ethanol.
The combined faeces produced by each individual
larva were freeze-dried and extracted in the same
way as the birch leaves. Leaf and faeces extracts
were filtered through 0.45-µm PTFE filters and
analysed with HPLC-DAD at 280 nm and 349 nm.
The HPLC system (Merck-Hitachi, Tokyo, Japan)
consisted of a pump L-7100, a diode array detector
L-7455, a programmable autosampler L-7250, and
an interface D-7000. Column and chromato-
graphic conditions were as described earlier (Sal-
minen et al., 1999), except that 0.1 m H
3
PO
4
was
replaced with 0.05 m H
3
PO
4
. A selected set of
samples was analysed also with HPLC-ESI-MS as
described by Salminen et al. (1999). Phenolic com-
pounds were identified on the basis of their UV
and mass spectra and retention times reported in
the literature (Ossipov et al., 1995, 1996; Salminen
et al., 1999, 2001; Valkama et al., 2003). Phenolic
acids were quantified as gallic, chlorogenic and p-
coumaric acid, myricetin glycosides as myricetins,
quercetin glycosides as quercetins, kaempferol gly-
cosides as kaempferols and flavonoid aglycones
as acacetins.
Isolation and identification of the metabolic
products of acacetin and kaempferide
To produce enough larval metabolites of aca-
cetin and kaempferide for their isolation and
structure elucidation, two separate sets of 5
th
in-
star E. autumnata were fed with washed birch
leaves artificially enriched with high levels
(> 20 mg/g dry weight) of acacetin or kaempferide
only. After all leaf-painted compounds had been
consumed, faeces of the two sets of larvae were
freeze-dried and extracted, and one major metab-
olite detected in the extract of both sets. The two
metabolites were isolated from the water-soluble
fractions of the faeces extracts with Merck Li-
Chroprep RP-18 column (44 ¥ 3.7 cm in diameter,
40Ð63 µm; Darmstadt, Germany) using different
gradients of H
2
OinCH
3
CN. The metabolites were
analysed with HPLC-ESI-MS, and
1
H NMR
spectra were recorded on a Jeol Lambda 400 MHz
series spectrometer (Tokyo, Japan) and a Bruker
routine AM-200 NMR spectrometer (Karlsruhe,
Germany) using DMSO-d
6
as a solvent. To con-
firm the identities of the sugar moieties of the me-
tabolites, they were hydrolysed according to Chin
et al. (2000) and analysed by GC-MS (Perkin-
Elmer Autosystem XL/TurboMass Gold, Norwalk,
USA) after trimethylsilylation.
Results and Discussion
Comparing the composition and content of phe-
nolic compounds in the leaf diet and in the faeces
of each individual larva, we were able to elucidate
the metabolic modifications of different com-
pounds in the larval digestive tract. The HPLC-
DAD chromatograms of the water-soluble and li-
pophilic fractions of the birch leaf diet and the cor-
responding fractions of larval faeces revealed sig-
nificant differences in the composition of phenolic
compounds (quantitative results shown in Table I).
The appearance of uric acid in larval faeces
The water-soluble fraction of larval faeces con-
tained a non-dietary compound as one of the ma-
jor metabolites (compound 23 in Table I). On the
basis of its UV spectrum (λ
max
230 and 283 nm),
molar mass (168 g/mol) and mass spectral frag-
mentation (m/z of the main fragment was 124) in
ESI-MS and co-elution with a commercially avail-
able standard, this nitrogen-containing metabolite
was identified as uric acid. It was surprising to find
such a major nitrogenous waste product in the lar-
val faeces of E. autumnata, an herbivore whose
growth is supposed to be limited by the nitrogen
intake from the diet (e.g. Ruohomäki et al., 1996;
Kaitaniemi et al., 1998; Kause et al., 1999). How-
ever, due to practical reasons we had sampled the
faeces once a day only; this might have enabled
post-intestinal modification of the larval metabo-
lites and the formation of artefacts in the faeces.
To rule out this possibility, we did an additional
experiment with fifteen 5
th
instar E. autumnata
larvae. Larvae were fed with birch (Betula pu-
bescens) leaves for 9 h and during that time every
second drop of faeces produced by individual lar-
vae was collected directly into 70% acetone, i.e.
the extraction solvent. The other drops of faeces
were collected in a separate vial and left to stand
at room temperature for 24 h; after that time both
sets of faeces were extracted with 70% acetone.
Interestingly, the only difference between the
HPLC traces of the two differently collected fae-
ces samples was the absence of uric acid in the

440 J.-P. Salminen et al. · Metabolic Modifications of Phenolics by an Herbivorous Insect
Table I. Distribution of the water-soluble and lipophilic phenolic compounds (mg/g dry weight) in the leaf diet
(Betula pubescens) and faeces of the 5
th
instar Epirrita autumnata larvae.
Leaf diet Faeces
Class No Compound Water-soluble Lipophilic Water-soluble Lipophilic
fraction fraction fraction
a
fraction
a
Vacuolar birch leaf 1 Chlorogenic acid 6.85 Ð 4.42
b
Ð
phenolics 2 p-Coumaroylquinic acid 0.52 Ð 1.67
b
Ð
3 Myricetin-3-O-galactopyranoside 1.78 Ð 0.68 Ð
4 Quercetin-3-O-galactopyranoside 2.34 Ð 2.05 Ð
5 Quercetin-3-O-glucuronopyranoside 8.80 Ð 8.46 Ð
6 Quercetin-3-O-arabinofuranoside 1.10 Ð n.q. Ð
7 Kaempferol-3-O-glucopyranoside 0.99 Ð n.q. Ð
8 Kaempferol glycoside 2.26 Ð 1.98 Ð
9 Kaempferol glycoside 1.06 Ð 0.82 Ð
10 Kaempferol-3-O-rhamnopyranoside 0.65 Ð n.q. Ð
Cuticular birch leaf 11 Naringenin Ð 0.47 ÐÐ
phenolics 12 Apigenin Ð 0.92 ÐÐ
13 Kaempferol Ð 1.86 ÐÐ
14 Flavonol methyl ether Ð 2.02 ÐÐ
15 Flavanone methyl ether Ð 7.09 ÐÐ
16 Acacetin Ð 2.72 ÐÐ
17 Kaempferide Ð 7.73 Ð traces
18 Flavonol dimethyl ether Ð 3.84 ÐÐ
19 Pentahydroxyflavone trimethyl ether Ð 1.31 ÐÐ
20 Flavanone Ð 2.01 Ð traces
21 Kaempferol derivative Ð 1.97 Ð traces
22 Apigenin derivative Ð 1.53 ÐÐ
Additional faecal 23 Uric acid ÐÐ 7.41
Ð
metabolites
24
Flavonoid glycosides (aglycones ÐÐ 18.83 Ð
glycosylated by E. autumnata
larvae)
c
a
Mean value of ten larvae.
b
A sum of four isomers.
c
A sum of eleven flavonoid glycosides.
n.q. = not quantified (peaks overlapped by compounds 24).
extract of 70% acetone collection (data not
shown). This clearly suggested that rather than a
larval waste product, uric acid actually was an ar-
tefact caused by post-intestinal modification of lar-
val metabolites. This finding was further supported
when the uric acid free extract was left to stand at
room temperature; the composition of the pheno-
lic metabolites was not altered but increasing
amounts of uric acid appeared in the extract as a
function of time. Thus it is presumable that uric
acid is a catabolic product of some water-soluble,
non-phenolic and nitrogenous metabolites that as
such could not be detected by UV spectroscopy.
Nevertheless, this result questions whether the
previous studies reporting the presence of uric
acid in insect faeces have been dealing just with
an artefact, and not with a real product of insect
metabolism as such (e.g. Bursell, 1965; Ritter,
1996).
The metabolic modifications of chlorogenic and
p-coumaroylquinic acid
The HPLC-DAD and HPLC-ESI-MS analysis of
the faeces showed that the main phenolic acids of
the leaf diet Ð having a difference of only one hy-
droxyl group i.e. chlorogenic and p-coumaroyl-
quinic acid (compounds 1 and 2 in Table I) Ð were
isomerised in the larval digestive tract. Such iso-
merisation has been observed to happen to chlo-
rogenic acid in vitro under basic conditions (Na-
gels et al., 1980), suggesting that basic conditions
would prevail also in the gut of E. autumnata. To
ensure this the gut pH of some 5
th
instar larvae
was measured and found to be close to 9, i.e.
strongly basic just like in most lepidopteran spe-
cies (Martin and Martin, 1983; Appel, 1993). Inter-
estingly, these kinds of basic conditions may pro-
mote the auto-oxidation of phenolic compounds

J.-P. Salminen et al. · Metabolic Modifications of Phenolics by an Herbivorous Insect 441
even in the absence of polyphenol oxidases (Ap-
pel, 1993). Moreover, the oxidation of phenolics
into quinones, especially o-quinones, is thought to
be one of the main mechanisms in turning other-
wise harmless compounds into reactive and harm-
ful ones, at least from the insect’s point of view
(Felton et al., 1989; Appel, 1993). Chlorogenic
acid, one of the main phenolics of birch leaves (see
Table I), would be an ideal substrate for the pro-
duction of o-quinones due to its benzene ring
located o-dihydroxyl group. Interestingly, on
average 63% of the ingested chlorogenic acid was
missing from the faeces of the larvae (when calcu-
lated as a sum of all isomers), possibly as a result
of the transformation of chlorogenic acid (or its
isomers) into o-quinones in the digestive tract of
E. autumnata.
The metabolic modifications of flavonoid
glycosides and aglycones
Flavonoid glycosides of the birch leaf diet were
the only group of compounds that were mostly
excreted unchanged (e.g. compounds 4, 5, 8 and 9
in Table I), although there were compound-spe-
cific differences in their metabolism as well. To
pinpoint the differences in larval metabolism even
within one phenolic subgroup, i.e. the flavonoids,
the most easily observed changes in the composi-
tion of phenolics between the leaf diet and faeces
were related to flavonoid aglycones (compounds
11Ð22 in Table I). The lipophilic aglycones were
almost totally absent in the lipophilic fraction of
the faeces, while several metabolites having UV
spectra similar to the flavonoid aglycones ap-
peared into the water-soluble fraction (compounds
24 in Table I). However, the retention times or
molar masses of these metabolites did not match
with the flavonoid aglycones or glycosides present
in the larval diet (Ossipov et al., 1995, 1996; Val-
kama et al., 2003). In contrast, the mass spectral
characteristics of the metabolites suggested them
to be flavonoid glycosides, of which the aglycone
parts had similar molar masses as the aglycones
found on birch leaf surface (Valkama et al., 2003).
Moreover, the level of these glycosides was highly
suppressed when E. autumnata larvae were fed
with leaves from which most of the leaf surface
aglycones were removed by washing with 95%
ethanol. Therefore the new faecal flavonoid glyco-
sides were preliminarily classified as glycosylation
products of flavonoid aglycones by E. autumnata
larvae. The aglycones as such are lipophilic but be-
come water-soluble once glycosylated; this ex-
plains the almost total absence of flavonoid agly-
cones in the lipophilic fraction of faeces and the
presence of the additional flavonoid glycosides in
the water-soluble fraction (see Table I).
Ensuring the glycosylation of flavonoid aglycones
by Epirrita autumnata larvae
Since flavonoid aglycones were the group of
phenolics that were most modified by E. autum-
nata, more attention was paid to their metabolic
fate in the larval digestive tract. To prove the
above mechanism of aglycone glycosylation, two
separate sets of 5
th
instar E. autumnata were fed
with washed birch leaves that were artificially en-
riched with high levels (> 20 mg/g dry weight) of
acacetin (compound 16 in Table I) or kaempferide
(compound 17 in Table I) only. When all leaf-
painted compounds had been consumed by larvae,
the faeces of the two sets of larvae were freeze-
dried and extracted, and one major metabolite was
detected in the extract of both sets. The m/z values
from negative ion ESI-MS for the metabolite of
acacetin were 445, 891 and 283, and for the metab-
olite of kaempferide 461, 923 and 299. These val-
ues corresponded to [M-H]
Ð
, [2M-H]
Ð
and [agly-
cone-H]
Ð
, respectively. As the [M-H]
Ð
value for
acacetin is 283 and for kaempferide 299, the me-
tabolites were tentatively identified as acacetin
and kaempferide glycosides. After GC-MS analy-
sis of the products of hydrolysis of these com-
pounds, the sugar moiety of both of the flavonoids
was confirmed to be glucose.
By comparing the
1
H NMR spectra of acacetin
and kaempferide glucosides to those measured
separately for acacetin and kaempferide, and to
the matching spectra found in the literature
(Sharaf et al., 1997; Nielsen et al., 1998; Curir et al.,
2001), the two metabolites were finally identified
as acacetin-7-O-glucoside and kaempferide-3-O-
glucoside, respectively (see Fig. 1). The
1
H NMR
data for the acacetin-7-O-glucoside are as follows:
8.00 (2H, d, J = 8.8 Hz, H-2⬘and H-6⬘), 7.10 (2H,
d, J = 8.8 Hz, H-3⬘and H-5⬘), 6.75 (1H, d, J =
4.8 Hz, H-8), 6.73 (1H, s, H-3), 6.70 (1H, d, J =
4.8 Hz, H-6), 5.69 (1H, d, J = 4.0 Hz, H-1 of glu-
cose), 3.85 (3H, s, H-4⬘, OMe), 3.15Ð3.50 (sugar
protons). The
1
H NMR data for the kaempferide-
3-O-glucoside are as follows: 12.45 (1H, s, H-5,
OH), 8.09 (2H, d, J = 8.96 Hz, H-2⬘and H-6⬘), 7.05

442 J.-P. Salminen et al. · Metabolic Modifications of Phenolics by an Herbivorous Insect
R
1
R
2
Acacetin OH H
Kaempferide OH OH
Acacetin-7-O -glucoside O -glucose
H
Kaempferide-3-O -glucoside
OH
O -glucose
OCH
3
R
2
R
1
OH
O
O
A
B
6´
1
2
3
4
5
6
7
8
1´
2´
3´
4´
5´
Fig. 1. Structures of the flavonoid aglycones (acacetin
and kaempferide) fed to the 5
th
instar Epirrita autum-
nata larvae, and the corresponding glucosides (acacetin-
7-O-glucoside and kaempferide-3-O-glucoside) found in
larval faeces after detoxification of the aglycones by
Epirrita autumnata via glycosylation.
(2H, d, J = 8.96 Hz, H-3⬘and H-5⬘), 6.18 (1H, s,
H-8), 5.98 (1H, s, H-6), 5.40 (1H, d, J = 7.90 Hz,
H-1 of glucose), 3.83 (3H, s, H-4⬘, OMe), 3.10Ð
3.60 (sugar protons).
The levels of glycosylation of flavonoid aglycones
by 5
th
instar Epirrita autumnata larvae
To study the levels of glycosylation of acacetin
and kaempferide by 5
th
instar Epirrita autumnata,
we analysed the faeces produced by individual lar-
Fig. 2. The amounts of ingested acacetin
(A: dietary level 5 mg/g; B: dietary level
10 mg/g) and kaempferide (C: dietary level
5 mg/g; D: dietary level 10 mg/g) by the 5
th
instar Epirrita autumnata larvae vs. the ex-
creted amounts of these aglycones together
with their glycosidic metabolites. In all, n =
r = 0.874
r = 0.810
0
20
40
60
80
100
120
140
160
0 100 200 300 400
r = 0.954
r = 0.966
0
20
40
60
80
100
120
140
160
0 100 200 300 400
r = 0.901
r = 0.934
0
20
40
60
80
100
120
140
160
0 200 400 600 800
r = 0.931
r = 0.771
0
20
40
60
80
100
120
140
160
0 200 400 600 800
Acacetin intake [
µ
g]
A
C
D
B
Kaempferide intake [
µ
g]
Acacetin intake [
µ
g]
Kaempferide intake [
µ
g]
Excretion of the aglycon
e
and its glycoside [
µ
g]
Excretion of the aglycon
e
and its glycoside [
µ
g]
excretion of the aglycone as such
excretion of the glycosidic
form of the aglycone
12 larvae.
vae during the 72-h bioassay. As a further proof
for the observed detoxification mechanism of fla-
vonoid aglycones, the amounts of faecal acace-
tin-7-O-glucoside and kaempferide-3-O-glucoside
correlated positively with the ingested amounts of
acacetin and kaempferide, regardless of the di-
etary levels of the painted aglycones (see Fig. 2).
However, the efficiency of aglycone glycosylation
was not even close to complete; its average level
varied from 17 ð 2% to 33 ð 4% (mean ð SE)
depending on the aglycone and its dietary level.
Interestingly, as the overall glycosylation efficiency
by the larvae was 22 ð 2% (mean ð SE), within
individual larvae the efficiency varied even from 2
to 57%. Moreover, the capacity of E. autumnata
to glycosylate acacetin seemed to be overloaded
already at the 5 mg/g level; contents of acacetin-7-
O-glucoside did not any more increase in the fae-
ces when the absolute intake of acacetin was ex-
perimentally doubled (compare Fig. 2A, B). On
the contrary, although kaempferide was glycosy-
lated less efficiently than acacetin at the 5 mg/g
dose (compare Fig. 2A, C), the yield of faecal
kaempferide-3-O-glucoside was almost doubled
with the doubled dietary intake at the 10 mg/g
level (compare Fig. 2C, D). These findings suggest
that even a difference of one hydroxyl group be-
tween the flavonoids makes a clear difference to
their metabolism. Moreover, also traces of non-
transformed acacetin and kaempferide were found
in the faeces (0.5Ð1.3% of the ingested amounts;

J.-P. Salminen et al. · Metabolic Modifications of Phenolics by an Herbivorous Insect 443
see Fig. 2). While 65Ð83% of the ingested agly-
cones remained undetected in the faeces (as agly-
cones or glycosides), their ultimate fate represents
an open question; presumably the undetected por-
tion was either oxidised in the larval digestive
tract, or due to the lipophilic nature bound to the
fat body or peritrophic membranes of the insects
(Barbehenn, 2001).
Ecological implications of the observed patterns of
phenolic metabolism in Epirrita autumnata
Previously we showed that birch leaf hydrolysa-
ble tannins are partially hydrolysed by E. autum-
nata larvae (Salminen and Lempa, 2002). The pre-
sent study further highlights the fact that
individual phenolics face highly differential fates
in the digestive tract of a lepidopteran herbivore.
Chlorogenic and p-coumaroylquinic acid were iso-
merised, flavonoid glycosides excreted without
visible metabolic modifications, whereas flavonoid
aglycones were partially detoxified via glycosyla-
tion. These patterns strongly indicate that the phe-
nolic profiles in the insect gut are very much dif-
ferent from those in the leaf diet. Further, the
efficiency of processing of even a single phenolic
compound was highly variable among individual
insects (e.g. 8Ð84% of the ingested chlorogenic
acid was recovered from the faeces of larvae eat-
ing foliage of the same tree). The reasons behind
Appel H. M. (1993), Phenolics in ecological interactions: Ginko biloba extract and its products. J. Food Drug
the importance of oxidation. J. Chem. Ecol. 19, Anal. 8, 289Ð296.
1521Ð1552. Curir P., Dolci M., Lanzotti V., and Tagliatela-Scafati O.
Barbehenn R. V. (2001), Roles of peritrophic mem- (2001), Kaempferide triglycoside: a possible factor of
branes in protecting herbivorous insects from ingested resistance of carnation (Dianthus caryophyllus)toFu-
plant allelochemicals. Arch. Insect Biochem. Physiol. sarium oxysporum f. sp. Dianthi. Phytochemistry 56,
47,86Ð99. 717Ð721.
Barbehenn R. V. and Martin M. M. (1992), The protec- Feeny P. P. (1970), Seasonal changes in oak leaf tannins
tive role of the peritrophic membrane in the tannin- and nutrients as a cause of spring feeding by winter
tolerant larvae of Orgyia leucostigma (Lepidoptera). moth caterpillars. Ecology 51, 565Ð581.
J. Insect. Physiol. 38, 973Ð980. Feeny P. P. and Bostock H. (1968), Seasonal changes in
Barbehenn R. V. and Martin M. M. (1994), Tannin sensi- the tannin content of oak leaves. Phytochemistry 7,
tivity in larvae of Malacosoma disstria (Lepidoptera): 871Ð880.
roles of the peritrophic envelope and midgut oxida- Felton G. W., Donato K., Del Vecchio R. J., and Duffey
tion. J. Chem. Ecol. 20, 1985Ð2001. S. S. (1989), Activation of plant foliar oxidases by in-
Barbehenn R. V., Martin M. M., and Hagerman A. E. sect feeding reduces the nutritive quality of foliage for
(1996), Reassessment of the roles of the peritrophic noctuid herbivores. J. Chem. Ecol. 15, 2667Ð2694.
envelope and hydrolysis in protecting polyphagous Grayer R. J., Harborne J. B., Kimmins F. M., Stevenson
grasshoppers from ingested hydrolyzable tannins. J. P. C., and Wijayagunasekera H. N. P. (1994), Phenolics
Chem. Ecol. 22, 1901Ð1919. in rice phloem sap as sucking deterrents to the brown
Bursell E. (1965), The excretion of nitrogen in the in- planthopper, Nilaparvata lugens. Acta Horticult. 381,
sects. Adv. Insect Physiol. 4,33Ð67. 691Ð694.
Chin L., Lin Y., Huang C., and Wen K. (2000), Evalua- Haukioja E. (2003), Putting the insect into the birch-
tion of quantitative analysis of flavonoid aglycons in insect interactions. Oecologia 136, 161Ð168.
such a variation are largely unknown, but at any
case it is hardly surprising that the results of eco-
logical studies trying to correlate insect perfor-
mance directly with the phenolic composition of
leaves have been variable and hard to repeat
(Haukioja, 2003). We propose that finding the true
effects of particular phenolic compounds on insect
performance demands knowledge of the ways and
levels of phenolic metabolism in the digestive tract
of individual insects. For instance, instead of using
the foliar content of chlorogenic acid in the corre-
lations it could be more useful to correlate insect
performance with the level of chlorogenic acid not
recovered from the faeces; this level presumably
being closer to the portion of chlorogenic acid bio-
logically active against that particular insect. To
conclude, in the absence of detailed studies on the
metabolism of plant phenolics in the digestive
tracts of different species of herbivorous insects, it
is likely that many of the mechanisms of action Ð
or even the lacks of long-proposed actions Ð of
these compounds are still to be discovered.
Acknowledgements
We thank Marjo Lukkarinen for assistance dur-
ing this work that was supported in part by grants
from the Emil Aaltonen Foundation (to J.-P. S.,
and M. L.) and from the Academy of Finland
(grant no 204209).

444 J.-P. Salminen et al. · Metabolic Modifications of Phenolics by an Herbivorous Insect
Henriksson J., Haukioja E., Ossipov V., Ossipova S., Sil- pounds from leaves of Betula pubescens and Betula
lanpää S., Kapari L., and Pihlaja K. (2003), Effects of pendula. J. Chromatogr. A 721,59Ð68.
host shading on consumption and growth of the geome- Ossipov V., Haukioja E., Ossipova S., Hanhimäki S., and
trid Epirrita autumnata: interactive roles of water, pri- Pihlaja K. (2001), Phenolic and phenolic-related
mary and secondary compounds. Oikos 103,3Ð16. factors as determinants of suitability of mountain
Kaitaniemi P., Ruohomäki K., Ossipov V., Haukioja E., birch leaves to an herbivorous insect. Biochem. Sys-
and Pihlaja K. (1998), Delayed induced changes in the tem. Ecol. 29, 223Ð240.
biochemical composition of host plant leaves during Ritter P. (1996), Biochemistry, A Foundation. Brooks/
an insect outbreak. Oecologia 116, 182Ð190. Cole Publishing Company, California, USA, p. 521.
Kause A., Ossipov V., Haukioja E., Lempa K., Han- Ruohomäki K., Chapin F. S. III, Haukioja E., Neuvonen
himäki S., and Ossipova S. (1999), Multiplicity of bio- S., and Suomela J. (1996), Delayed inducible resis-
chemical factors determining the quality of growing tance in mountain birch in response to fertilization
birch leaves. Oecologia 120, 102Ð112. and shade. Ecology 77, 2302Ð2311.
Keinänen M. and Julkunen-Tiitto R. (1998), High-per- Salminen J.-P. and Lempa K. (2002), Effects of hydrolys-
formance liquid chromatographic determination of able tannins on an herbivorous insect: fate of indivi-
flavonoids in Betula pendula and Betula pubescens dual tannins in insect digestive tract. Chemoecology
leaves. J. Chromatogr. A 793, 370Ð377. 12, 203Ð211.
Martin J. S. and Martin M. M. (1983), Precipitation of Salminen J.-P., Ossipov V., Loponen J., Haukioja E., and
ribulose-1,6-bis-phosphate carboxylase/oxygenase by Pihlaja K. (1999), Characterisation of hydrolysable
tannic acid, quebracho, and oak foliage extracts. J. tannins from leaves of Betula pubescens by high-per-
Chem. Ecol. 9, 285Ð294. formance liquid chromatography-mass spectrometry.
Martin J. S., Martin M. M., and Bernays E. A. (1987), J. Chromatogr. A 864, 283Ð291.
Failure of tannic acid to inhibit digestion or reduce Salminen J.-P., Ossipov V., Haukioja E., and Pihlaja K.
digestibility of plant protein in gut fluids of insect her- (2001), Seasonal variation in the content of hydrolysa-
bivores: Implications for theories of plant defence. J. ble tannins in the leaves of Betula pubescens. Phyto-
Chem. Ecol. 13, 605Ð621. chemistry 57,15Ð22.
Nagels L., van Dongen W., de Brucker J., and de Potter Sharaf M., El-Ansari M., Matlin S., and Saleh N. (1997),
J. (1980), High-performance liquid chromatographic Four flavonoid glycosides from Peganum harmala.
separations of naturally occurring esters of phenolic Phytochemistry 44, 533Ð536.
acids. J. Chromatogr. 187, 181Ð187. Tikkanen O.-P. and Julkunen-Tiitto R. (2003), Phenolog-
Nielsen J., Norbaek R., and Olsen R. (1998), Kaemp- ical variation as protection against defoliating insects:
ferol tetraglucosides from cabbage leaves. Phyto- the case of Quercus robur and Operophtera brumata.
chemistry 49, 2171Ð2176. Oecologia 136, 244Ð251.
Ossipov V., Nurmi K., Loponen J., Prokopiev N., Hau- Valkama E., Salminen J.-P., Koricheva J., and Pihlaja K.
kioja E., and Pihlaja K. (1995), HPLC isolation and (2003), Comparative analysis of leaf trichome struc-
identification of flavonoids from white birch Betula ture and composition of epicuticular flavonoids in Fin-
pubescens leaves. Biochem. Syst. Ecol. 23, 213Ð222. nish birch species. Ann. Bot. 91, 643Ð655.
Ossipov V., Nurmi K., Loponen J., Haukioja E., and Pih- Zimmer M. (1999), The fate and effects of ingested hy-
laja K. (1996), High-performance liquid chromato- drolyzable tannins in Porcellio scaber. J. Chem. Ecol.
graphic separation and identification of phenolic com- 25, 611Ð628.