Isoferuloyl derivatives of five seed glucosinolates in the crucifer genus Barbarea

Article abstract of DOI:10.1016/j.phytochem.2011.01.034

Five acylated glucosinolates (GSLs) were isolated as desulfated derivatives after enzymatic desulfation of anionic metabolites from seeds of two chemotypes of Barbarea vulgaris, and their structures were elucidated by a combination of spectroscopic methods and HPLC analysis of products of enzymatic de-acylation. The acyl group was in all cases found to be a trans isoferuloyl group at the 6′-position of the thioglucose moiety. The GSL moieties of the native metabolites were found to be one Trp derived; indol-3-ylmethylGSL, as well as four homoPhe derived; phenethylGSL, (S)-2-hydroxy-2-phenylethylGSL, (R)-2-hydroxy-2-phenylethylGSL, and (R)-2-hydroxy-2-(4-hydroxyphenyl)ethylGSL. GSL analysis of B. vulgaris seed extracts by the commonly employed 'desulfoGSL' method (based on binding to anion exchange columns, enzymatic desulfation, elution and HPLC) was optimized for 6′-isoferuloyl derivatives of GSLs. From peak areas before and after de-acylation of the isolated desulfoGSL, the response factor of the 6′-isoferuloyl derivative of (S)-2-hydroxy-2- phenylethylGSL was estimated to be 0.37 (relative to 1.00 for sinigrin), allowing us to estimate the level in B. vulgaris to 3 μmol/g dry wt. in mature seeds and less than 0.1 μmol/g dry wt. in seedlings and floral parts of the insect resistant G-type of B. vulgaris var. arcuata. HPLC analysis of intact GSLs in crude extracts and after group separation did not reveal additional derivatives, but confirmed the existence of the deduced intact GSLs. A taxonomic screen showed that most (14/17) B. vulgaris accessions (with the exception of three accessions of var. vulgaris) contained relatively high levels of 6′-isoferuloyl GSLs. The profiles of 6′-isoferuloylated GSLs matched the profiles of non-acylated GSLs in the same seed accessions, suggesting a low side chain specificity of the isoferuloylation mechanism. A minor peak tentatively identified as a dimethoxycinnamoyl derivative of (S)-2-hydroxy-2-phenylethylGSL was detected by HPLC-MS of one accession, suggesting that GSLs with other acyl groups may occur at low levels. A single analyzed B. plantaginae accession contained relatively high levels of 6′-isoferuloylated phenethylGSL and (S)-2-hydroxy-2-phenylethylGSL. Five other tested Barbarea species (B. australis, B. bracteosa, B. intermedia, B. stricta, B. verna) also contained isoferuloylated GSLs, albeit at lower levels than in B. vulgaris and B. plantaginae, suggesting that seed GSL acylation is a general character of the Barbarea genus and possibly also of related genera including Arabidopsis.

Full text of DOI:10.1016/j.phytochem.2011.01.034

Phytochemistry 72 (2011) 610–623
Contents lists available at ScienceDirect
Phytochemistry
journal homepage: www.elsevier.com/locate/phytochem
Isoferuloyl derivatives of five seed glucosinolates in the crucifer genus Barbarea
⇑
Niels Agerbirk , Carl Erik Olsen
Faculty of Life Sciences, University of Copenhagen, Thorvaldsensvej 40, 1871 Frederiksberg, Denmark
a r t i c l e i n f o
a b s t r a c t
Article history:
Five acylated glucosinolates (GSLs) were isolated as desulfated derivatives after enzymatic desulfation of
anionic metabolites from seeds of two chemotypes of Barbarea vulgaris, and their structures were
elucidated by a combination of spectroscopic methods and HPLC analysis of products of enzymatic de-
Received 22 December 2010
Received in revised form 27 January 2011
Available online 26 February 2011
0
acylation. The acyl group was in all cases found to be a trans isoferuloyl group at the 6 -position of the
thioglucose moiety. The GSL moieties of the native metabolites were found to be one Trp derived;
indol-3-ylmethylGSL, as well as four homoPhe derived; phenethylGSL, (S)-2-hydroxy-2-phenylethylGSL,
Key words:
Acylated glucosinolates
Intact glucosinolate analysis
Desulfoglucosinolate analysis
Taxonomic occurrence
Barbarea
(
R)-2-hydroxy-2-phenylethylGSL, and (R)-2-hydroxy-2-(4-hydroxyphenyl)ethylGSL. GSL analysis of B.
vulgaris seed extracts by the commonly employed ‘desulfoGSL’ method (based on binding to anion
0
exchange columns, enzymatic desulfation, elution and HPLC) was optimized for 6 -isoferuloyl derivatives
of GSLs. From peak areas before and after de-acylation of the isolated desulfoGSL, the response factor of
0
Flowers
the 6 -isoferuloyl derivative of (S)-2-hydroxy-2-phenylethylGSL was estimated to be 0.37 (relative to 1.00
Siliques
Seeds
Germination
Seedlings
Herbivores
Anthocharis cardamines
for sinigrin), allowing us to estimate the level in B. vulgaris to 3
lmol/g dry wt. in mature seeds and less
than 0.1 mol/g dry wt. in seedlings and floral parts of the insect resistant G-type of B. vulgaris var. arcu-
l
ata. HPLC analysis of intact GSLs in crude extracts and after group separation did not reveal additional
derivatives, but confirmed the existence of the deduced intact GSLs. A taxonomic screen showed that
most (14/17) B. vulgaris accessions (with the exception of three accessions of var. vulgaris) contained rel-
0
0
atively high levels of 6 -isoferuloyl GSLs. The profiles of 6 -isoferuloylated GSLs matched the profiles of
non-acylated GSLs in the same seed accessions, suggesting a low side chain specificity of the isoferuloy-
lation mechanism. A minor peak tentatively identified as a dimethoxycinnamoyl derivative of (S)-
2-hydroxy-2-phenylethylGSL was detected by HPLC-MS of one accession, suggesting that GSLs with other
acyl groups may occur at low levels. A single analyzed B. plantaginae accession contained relatively
0
high levels of 6 -isoferuloylated phenethylGSL and (S)-2-hydroxy-2-phenylethylGSL. Five other tested
Barbarea species (B. australis, B. bracteosa, B. intermedia, B. stricta, B. verna) also contained isoferuloylated
GSLs, albeit at lower levels than in B. vulgaris and B. plantaginae, suggesting that seed GSL acylation is a
general character of the Barbarea genus and possibly also of related genera including Arabidopsis.
Ó 2011 Elsevier Ltd. All rights reserved.
1
. Introduction
Glucosinolates are secondary plant metabolites characterized
moiety hydroxyl groups (Iori et al., 1996; Burmeister et al., 1997).
However, a few GSLs with substituents on the thioglucose moiety
have been documented from seeds of some plant species, namely
0
0
by the presence of an amino acid derived sulfated thiohydroximate
and a sulfur-linked b-D-glucopyranosyl moiety referred to as the
thioglucose moiety in the following. The amazingly varied bio-
chemistry of GSLs is intimately related to co-occurrence with the
enzyme myrosinase (thioglucoside glucohydrolase, EC 3.2.1.147),
which may hydrolyse GSLs into an organic isothiocyanate, sulfate,
glucose, and hydrogen ion or many other product combinations
6 -sinapoyl-4-methylsulfinylbut-3-enylGSL (6 -sinapoylglucoraphenin)
0
from Raphanus sativus L. (Linsheid et al., 1980) and 6 -benzoyl
derivatives of 4-methylthiobutylGSL, 4-methylsulfinylbutylGSL,
and 4-benzoyloxybutylGSL from Arabidopsis thaliana (L.) Heynh.
(Reichelt et al., 2002).
Classical methods of GSL analysis were reported to be well sui-
ted for the recognition of GSLs substituted with aromatic acids,
0
(Bones and Rossiter, 2006). The thioglucose moiety of the vast
such as 6 -sinapoylglucoraphenin, as paper chromatograms/elec-
majority of known GSLs is not substituted at any position, and in-
deed the substrate recognition of the well investigated myrosinase
from white mustard involves non-covalent bonds to all thioglucose
tropherograms with the intact compounds could be inspected in
UV light with and without ammonia vapour for visualisation and
preliminary identification of acyl groups (Bjerg and Sørensen,
1
987; Sørensen, 1990; Agerbirk et al., 2001b). The classical GSL
⇑
Corresponding author. Tel.: +45 35 33 24 38; fax: +45 35 33 23 98.
E-mail address: nia@life.ku.dk (N. Agerbirk).
from Barbarea vulgaris R. Br. is (S)-2-hydroxy-2-phenylethylGSL,
which is commonly known as glucobarbarin and abbreviated as
0
031-9422/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.phytochem.2011.01.034

N. Agerbirk, C.E. Olsen / Phytochemistry 72 (2011) 610–623
611
2
S in this paper, while its epimer (R)-2-hydroxy-2-phenylethylGSL
glucosinolates was not presented. In a critical review of known,
natural GSL structures the acylated GSLs in Barbarea seeds were
duly regarded as hypothetical (Fahey et al., 2001). A later review
suggested that acylated GSLs are not substrates of sulfatase and
will not be detected when using analytical methods based on de-
sulfoGSLs (Bellostas et al., 2007). However, other workers had
meanwhile detected several GSLs with benzoyl substituent on
the thioglucose moiety from A. thaliana, using a standard desul-
foGSL protocol (Reichelt et al., 2002).
Species of the genus Barbarea are currently subject to consider-
able attention in the field of chemical ecology (Badenez-Pérez
et al., 2010, 2011; Nielsen et al., 2010a,b; Kuzina et al., 2011) and
experimental taxonomy (Ørgaard and Linde-Laursen, 2008;
Toneatto et al., 2010). We have previously conducted an investiga-
tion of GSL diversity in roots and rosette leaves of several Barbarea
species, and documented a considerable variation in accumu-
lated GSLs within the species B. vulgaris (Agerbirk et al. 2001a,b,
2003a). In contrast to var. vulgaris and the ‘G-type’ of B. vulgaris
var. arcuata (Opiz.) Fries that both accumulated 2S, the ‘P-type’ of
var. arcuata accumulated the epimer 2R, as well as two novel GSLs:
4-hydroxyglucosibarin [(R)-2-hydroxy-2-(4-hydroxyphenyl)ethyl-
is known as glucosibarin and will be abbreviated 2R (Fig. 1). The
existence of a total of six isoferuloyl derivatives of 2S and 2R in
0
Barbarea, with the point of acylation being in the 2 -position, the
0
0
0
6
-position, as well as in both the 2 - and 6 -positions of the thio-
glucose moiety, was suggested in a book chapter and two reviews
(Bjerg and Sørensen, 1987; Sørensen, 1990; Bellostas et al., 2007).
Early reports contained HPLC chromatograms and capillary
electrophoresis electropherograms of both intact and desulfated
suggested isoferuloyl glucosinolates (Bjerg and Sørensen, 1987;
Michaelsen et al., 1992). In a further report, 6 -iF-2S was reported
as a constituent of B. vulgaris seeds (mean level 1.7
with very low levels of 2 -isoferuloylated (0.2
0
l
mol/g), along
0
0
0
l
mol/g) and 2 ,6 -di-
isoferuloylated (0.05
intact GSLs mentioned among several analysis methods used
Andersson et al., 1999), but further evidence for the structures
of the acylated glucosinolates was not presented. Agerbirk et al.
2001) reported that isoferuloylated glucosinolates detectable
lmol/g) 2S, with capillary electrophoresis of
(
(
in B. vulgaris seeds were not significant leaf constituents, based
on paper electrophoresis of crude seed and leaf extracts, but
further evidence for the existence and structures of the acylated
0
Fig. 1. Glucosinolates (GSLs) previously documented to occur in the genus Barbarea (1–7) and glucosinolates with isoferuloyl (iF) groups documented in this paper (6 -iF-1,
0
0
0
0
6
-iF-2R, 6 -iF-2S, 6 -iF-3R, 6 -iF-4). The structure of free isoferulic acid is also shown, and conventional NMR numbering systems for GSLs and cinnamic acid derivatives (with
three primes added as used in this paper) are indicated.

6
12
N. Agerbirk, C.E. Olsen / Phytochemistry 72 (2011) 610–623
GSL] (3R) (Fig. 1) and 1,4-dimethoxyindol-3-ylmethylGSL (7). Hy-
brids between the P and G types also accumulated 4-hydroxyg-
lucobarbarin [(S)-2-hydroxy-2-(4-hydroxyphenyl)ethylGSL] (3S),
and this GSL was also found in Barbarea stricta Andrz., but not in
Barbarea verna (Miller) Ascherson or Barbarea intermedia Boreau.
Since our first publications, we have included Barbarea australis
Hook f. to our study material. B. australis is a threatened, endemic
species from Tasmania (Kirkpatrick and Gilfedder, 1998), which we
regard as a potentially distant relative of the otherwise NW Euro-
pean species studied so far, and possibly suitable as an ‘out-group’
in phylogenetic analyses. For the present investigation, Barbarea
bracteosa Guss. and Barbarea plantaginae DC were also available
and were included.
2. Results and discussion
0
2.1. Identification of 6 -isoferuloylglucobarbarin as the desulfated
derivative
We initially analysed seeds of our main experimental plant, the
insect resistant G-type of B. vulgaris var. arcuata, employing our
standard desulfoGSL protocol (Agerbirk et al., 2001a). The seed
chromatograms, using our usual HPLC gradient ending at 60%
MeOH (Agerbirk et al., 2001a,b), showed a late, major peak of a
quite hydrophobic analyte with UV spectrum suggesting a highly
conjugated aromatic substituent (kmax at 328 nm (100%) and
shoulders at 230–240 nm (95%) and 290 nm (80%). After modifying
the gradient to end at 80% MeOH, the peak was well separated
from the part of the chromatogram associated with the MeOH
washing cycle and still exhibited a symmetrical peak shape ex-
pected for a single analyte (Fig. 3A), so we decided to isolate the
compound for structural elucidation.
B. vulgaris has been reported to be resistant to larvae of the
orange tip butterfly that often feed on floral parts of crucifers
(Courtney, 1981; Wiklund and Friberg, 2009; Agerbirk et al.,
2
010c). The glucosinolate detoxification mechanism of orange tip
larvae is believed to depend on myrosinase catalysed hydrolysis of
glucosinolates followed by immediate nitrile formation (Agerbirk
et al., 2006), and we wondered whether inhibition of myrosinase
activity by thioglucose moiety acylation could in some way inter-
fere with this detoxification mechanism. The present investigation
started as part of our investigations (Agerbirk et al., 2010b,c) of the
biochemistry of this plant-insect interaction (Section 2.1. and 2.2.)
but was expanded to include seed glucosinolates of the genus
Barbarea in general (Section 2.3.–2.6.).
The spectroscopic data (Table 1, Table 2) were in accordance
with a desulfoderivative of the previously suggested isoferuloyl
0
derivative of 2S; 6 -isoferuloyl-(S)-2-hydroxy-2-phenylethylGSL
0
(6 -iF-2S). This and other desulfo derivatives are indicated with
0
a prefix ‘d’ in the rest of the text, e.g. d6 -iF-2S for the isolated
desulfo derivative (Fig. 2). HMBC long range interactions were
particularly useful in elucidating the structure, notably an HMBC
0
interaction between the 6 hydrogen in the thioglucoside moiety
The purpose of the present investigation was to apply the de-
sulfoGSL method for a search for acylated GSLs in Barbarea seeds,
seedlings and floral parts, and to compare methods based on intact
and desulfated GSLs for the detection and quantification of such
and the carbonyl carbon confirmed the position of the isoferuloyl
group (Fig. 2). The chemical shift of the thiohydroximate carbon
was as typical for desulfated GSLs but lower than what is typical
for intact GSLs (Agerbirk et al., 2001a), confirming that the isolated
compound was the desulfated derivative. The coupling constant of
the isoferuloyl double bond hydrogens (H7, H8) was 16 Hz, as ex-
pected for the trans isomer (Nishida et al., 2003). The preparation
0
GSLs. We confirm the existence of two suggested 6 -isoferuloylated
0
GSLs in B. vulgaris, and report a further three 6 -isoferuloylated
structures. We suggest conditions for routine identification and
0
13
1
quantification of 6 -isoferuloylated GSLs, and demonstrate that
used for both C and H NMR analysis, which was apparently close
these metabolites mainly occur in seeds, with low levels in devel-
oping floral parts and seedlings, and exhibit a wide taxonomic dis-
tribution in the genus.
to a saturated solution for the D O solvent used, deviated slightly
from an initial, smaller preparation with respect to chemical shifts
of the isoferulic acid H7–H8 protons (Table 1, footnote b), leading
2
Table 1
1
H
H NMR data (d /ppm, multiplicity (J/Hz)) of desulfated glucosinolates.
0
0
0
0
0
d6 -iF-2S
d6 -iF-1
CD OD
d1
CD
d6 -iF-2R
CD OD
d6 -iF-3R
CD OD
d6 -iF-4
CD OD
d4
CD OD
3
NMR solvent:
D
2
O
3
3
OD
3
3
3
Side chain
1
1
2
2
3
4
a
b
2.90 dd (15; 6)
2.95 dd (15; 8)
5.08 dd (8; 6)
7.2–7.3 m
do.
2.86 m
do.
2.97 m
7.2–7.3 m
do.
2.8–3.0 m
do.
do.
7.2–7.3 m
do.
7.16 m
2.87 dd (15;5)
3.06 dd (15;9)
5.06 dd (9;5)
7.40 m
7.28 m
7.20 m
2.82 dd (15; 5)
3.04 dd (15; 8)
4.96 dd (8;5)
6.78 d (8)
7.22 d (8)
–
1a
1b
3.95 d (16)
4.23 d (16)
7.13 s
7.60 d (8)
7.00 t (8)
7.09 t (8)
7.34 d (8)
3.94 d (16)
4.25 d (16)
7.13 s
7.64 br d (8)
7.01 br t (8)
7.10 br t (8)
7.33 br d (8)
0
0
2
4
5
6
7
0
0
0
0
0
0
00
00
00
00
00
+ 6
+ 5
do.
7.16 m
0
0
Thioglucose moietya
0
1
4.95 d (10)
3.48‘t (9)
3.57 t (9)
3.44 t (9)
3.35 m
4.14 dd (12; 8)
4.60 dd (12; 2)
4.92 d (10)
3.25–3.5 m
do.
do.
3.62 m
4.83 d (10)
3.28‘t’ (9)
3.39 t (9)
3.32 t (9)
3.35 m
5.20 d (10)
3.31 ‘t’ (9)
3.49 t (9)
3.35 t (9)
3.68 m
5.17 d (10)
3.31 ‘t’ (9)
3.47 t (9)
3.35 t (9)
3.68 m
n.d.
3.1–3.4 m
do.
do.
do.
4.70 d (10)
3.14 ‘t’ (9)
3.05 t (9)
3.25 t (9)
3.10 m
0
0
2
0
3
0
4
0
5
0
0
6
6
a
b
4.22 dd (12;8)
4.57 dd (12;2)
3.62 dd (12;5)
3.81 dd (12;2)
4.21 dd (12;8)
4.54 dd (12:2)
4.22 dd (12;8)
4.53 dd (12;1)
4.22 dd (12;8)
4.52 dd (12;1)
3.60 dd (12:6)
3.80 dd (12;2)
Acyl moiety
0
0
0
0
0
00
00
00
00
00
2
5
6
7
8
6.81 br s
6.90 d (8)
6.93 d (2)
6.91 d (8)
6.86 dd (8;2)
7.46 d (16)
6.07 d (16)
3.91 s
–
–
–
–
–
–
6.91 d (2)
6.88 d (8)
6.85 dd (8;2)
7.46 d (16)
6.02 d (16)
3.88 s
6.95 d (2)
6.90 d (8)
6.86 dd (8;2)
7.48 d (16)
6.09 d (16)
3.89 s
7.03 d (2)
6.89 d (8)
6.95 dd (8;2)
7.65 d (16)
6.34 d (16)
3.87 s
–
–
–
–
–
–
6.79 dd (8; 1)
bc
7.25–7.4 d (16)
b
5.85–6.0 d (16)
OMe
3.84 s
a
Individual thioglucose moiety proton signals of d6 -iF-2S, d1, d6 -iF-2R, d6 -iF-3R and d4 assigned by a combination of 2D spectra and resolved by a J-resolved 1H NMR
0
0
0
0
0
0
spectrum (JRES). Signal of H1 in d6 -iF-4 presumably hidden under solvent peak. Chemical shifts of d2R in CD OD (not shown) included 6 a: 3.64 ppm and 6 b: 3.82 ppm.
3
b
Chemical shift concentration dependent in D
2
O, with higher chemical shift at lower concentration. The range of observed chemical shifts is indicated.
Peak overlap with side chain signals at high concentration but the signal and its coupling to H 8 confirmed by COSY.
c
000

N. Agerbirk, C.E. Olsen / Phytochemistry 72 (2011) 610–623
613
Table 2
ucts: prolonged incubation with porcine liver esterase (Reichelt
et al., 2002) resulted in a significant decrease in the peak area of
d6 -iF-2S and appearance of two product peaks, identified as d2S
and trans isoferulic acid (base line separated from a standard of
authentic ferulic acid) by comparison of their migration time and
UV spectra with those of authentic references. The observed UV
spectrum of isoferulic acid from the hydrolysis experiment was
also in accordance with the UV spectrum of trans isoferulic acid re-
ported by Borges et al. (1991).
1
3C NMR data (d/ppm) of intact and desulfated glucosinolates.
0
0
0
0
0
0
d6 -iF-
6 -iF-
6 -iF-
d2R
CD OD
d6 -iF-
d6 -iF-
2
S
2R
2S
2R
3R
CD OD
3
NMR-
solvent
D
2
O
D
2
O
D
2
O
3
CD
3
OD
d
C
d
C
d
C
d
C
d
C
d
C
Side chain
0
1
2
152.4
42.9
161.6
41.7
160.4
43.0
152.0
42.9
152.4
42.8
73.7
145.4
127.2
129.5
128.6
152.6
42.8
73.5
136.3
116.3
128.5
158.0
71.9
72.4
71.6
73.1
The corresponding intact GSL was isolated but not published
after extended elution of the same anion exchange column used
for isolation of 2S (Agerbirk et al., 2001b; J.K. Nielsen, unpublished
0
0
0
0
0
0
0
0
1
2
3
4
143.2
126.3
129.6
128.9
143.1
126.5
129.6
129.0
143.1
126.3
129.6
129.0
145.2
127.3
129.3
128.5
00
00
+ 6
+ 5
1
result), and was identified by one-dimensional H NMR (at
1
3
2
50 MHz) and C NMR (63 MHz) and comparison with the as-
Thioglucose moiety
0
signed spectrum of the desulfated derivative (Table 2). The chem-
ical shift of C0 of this intact GSL was 160.4 ppm, comparable to C0
chemical shifts for intact GSLs (Agerbirk et al., 2001b), and the m/z
1
2
3
4
5
6
82.1
73.1
77.8
70.7
78.9
64.6
82.7
72.7
77.7
70.7
78.7
64.7
82.3
72.9
77.7
70.6
79.0
64.5
83.5
74.6
79.7
71.3
82.1
62.7
83.4
74.4
79.5
71.8
79.4
65.2
83.5
74.5
79.7
71.9
79.4
65.3
0
0
0
0
0
ꢀ
value of the GSL ion [M–H] was 614, confirming the existence of
0
0
intact 6 -iF-2S and justifying that analytical identification of d6 -iF-
2
S after sulfatase treatment is indeed due to the presence of the
Acyl moiety
0
corresponding intact GSL in the seeds. Intact 6 -iF-2R was likewise
isolated as a minor constituent of seeds of the P-type of B. vulgaris
var. arcuata (Nielsen, unpublished result) previously reported to be
dominated by 2R. Minor differences in chemical shifts between the
two epimers were observed in position 0, 1 and 2 of the side chain,
0
0
0
0
0
0
0
0
0
00
00
00
00
00
00
00
00
00
1
2
3
4
5
6
7
8
9
127.8
127.7
114.7
145.8
150.7
112.5
123.4
146.8
115.2
169.4
56.6
127.8
114.8
145.7
150.7
112.7
123.7
146.8
115.4
169.3
56.6
–
–
–
–
–
–
–
–
–
–
128.8
115.1
147.9
151.5
112.4
122.9
146.8
115.8
168.8
56.5
128.8
115.2
148.1
151.6
112.6
122.7
146.9
115.8
168.9
56.5
114.9a
a
a
145.8
150.8
112.7
123.5
146.8
115.1a
169.3
56.6
0
0
position 1 and 5 of the thioglucose moiety, as well as in position
a
a
000
6
of the isoferuloyl moiety, suggesting that these positions of the
thioglucose and isoferuloyl moieties were spatially close to the
OMe
1
side chain OH group in solution (Table 2). The 250 MHz H NMR
0
0
a
spectra of intact 6 -iF-2R and 6 -iF-2S were similar to the reported
4
Signals may be interchanged.
0
00 MHz spectrum of d6 -iF-2S (results not shown) except for the
following differences between the R and S epimers: The anomeric
proton of the R-epimer had d = 5.26 in contrast to d = 4.94
for the S-epimer. And d of the two H1 side chain protons of the
H
H
H
S-epimer were identical, resulting in a single doublet at 3.05 ppm
in contrast to two dd signals at 2.98 ppm and 3.11 ppm in the case
of the R-epimer.
0
0
2
.2. Quantification of 6 - isoferuloylglucobarbarin (6 -iF-2S) in seeds,
floral parts and seedlings by the desulfoglucosinolate method
0
The isoferuloylated desulfo derivative d6 -iF-2S eluted later
than d2S from the anion exchange column used for desulfation; se-
ven column volumes (5 ꢁ 1 mL) of water were needed for quanti-
tative elution of the desulfated analyte (Fig. 4A), compatible with
standard desulfoGSL protocols optimized to include the indole
GSL 4-hydroxyglucobrassicin (Agerbirk et al., 2001a).
Fig. 2. Selected HMBC long-range interactions (arrows) that confirm the connec-
tivity of isolated desulfated 6 -iF-2S (abbreviated d6 -iF-2S).
0
0
The response factor sensu Wathelet et al. (2004) (a factor to be
multiplied with the peak area before comparison with the peak
area of the desulfo derivative of the standard, sinigrin) of the iso-
feruloylated GSL was estimated from the peak areas observed in the
hydrolysis experiment (section 2.1.). With the usual HPLC condi-
tions for desulfoGSLs (solvent not acidified), the peak area increase
of desulfoglucobarbarin was 39% of the peak area decrease of the
isoferuloyl derivative when measured at 229 nm. The response fac-
to the temporary hypothesis that the initial preparation was a dif-
ferent compound. However, this hypothesis could not be con-
firmed: all attempts to distinguish the two preparations by HPLC
failed (results not shown), leading to the alternative hypothesis
that the same compound exhibited slightly different NMR spectra
due to experimental conditions. Indeed, repeated 1H NMR of the
major preparation at two serial dilutions demonstrated a concen-
tration-effect for the relevant 1H NMR signals (Table 1 footnote
b), and we hence concluded that both our preparations were of
the same compound and the slightly different 1H NMR spectra of
two preparations were due to a concentration effect on selected
NMR chemical shifts.
0
tor of 6 -iF-2S would then, all other things being equal (i.e., assum-
ing that on-column desulfation and elution yield were comparable
0
for 2S and 6 -iF-2S, the former assumption not tested but the
second assumption confirmed, Fig. 4A), be 39% of the response
factor of 2S. Assuming the same response factor of 2S as the
known response factor of the structurally similar benzylGSL
0
(
0.95) (Wathelet et al., 2004), the response factor of 6 -iF-2S would
High resolution MS of the sodium adduct [M + Na]+ was in
then be 0.37, calculated as 39% of the assumed response factor
of 2S. (With acidic HPLC eluent (0.1% TFA) isoferulic acid was
detectable as a sharp peak accounting for the remaining lost peak
area of the isoferuloylated desulfoGSL).
+
accordance with the suggested structure (Found for [M + Na] ):
+
5
58.1396, calculated for [C25
H
29
O
10NSNa] : 558.1404). The in-
ferred structure was also confirmed by analysis of hydrolysis prod-

6
14
N. Agerbirk, C.E. Olsen / Phytochemistry 72 (2011) 610–623
The tentatively determined response factor allowed a quantita-
0
tive estimate of 6 -iF-2S in seeds at a level of 3
lmol/g dry wt.
(Table 3). This level was comparable to levels reported by previous
authors (1.7 lmol/g dry wt. in seeds of the same species, Andersson
et al., 1999), leading us to tentatively conclude that the analytical
method was reliable. The peak was nearly absent from the chro-
0
matograms of green silique samples: levels of 6 -iF-2S in green
siliques and flowers of B. vulgaris were less than 0.1 lmol/g dry
wt., i.e., many fold less than levels in mature seeds (Table 3).
Upon germination in Petri-dishes for 5–7 days, the content of
-iF-2S had dropped significantly (Table 4). The decline in the
0
6
isoferuloylated GSL was due to breakdown rather than leaching,
as the missing GSL was not recovered from the liquid medium
used for the germination. As the amount of un-substituted 2S
0
was much higher than the amount of 6 -iF-2S, a tendency for a
slight increase in the level of 2S after germination could be due
to both de novo synthesis, statistical uncertainty or possibly
0
hydrolysis of 6 -iF-2S. Levels of 1 and the indole GSLs typically
found in leaves (4) (Agerbirk et al., 2001b) or roots (4, 5, 6)
(Agerbirk et al., 2001a) increased markedly during germination,
demonstrating that GSL biosynthesis took place in seedlings
within the first days after germination. The significant, yet
0
incomplete decrease of 6 -iF-2S during germination for 5–7 days
was less dramatic than the complete disappearance of sinapoyl
choline in Brassica napus during just two days of germination
(Baumert et al., 2005).
The samples of flowers and siliques were harvested in the larval
feeding period of the orange tip butterfly (Agerbirk et al., 2010c).
The levels in tissues experienced by orange tip larvae were much
less than reported leaf concentrations of feeding inhibiting
saponins in B. vulgaris (hederagenin cellobioside 15 lmol/g dry
wt., Shinoda et al., 2002; oleanolic acid cellobioside ca. 1
dry wt., Agerbirk et al., 2003b; sum of both saponins 4
l
l
mol/g
mol/g
dry wt., Badenes-Pérez et al., 2011). Likewise, the isoferuloylated
0
GSL constituted only a tiny fraction of GSLs in seedlings. 6 -iF-2S
would therefore only be expected to play a role in resistance
against orange tip larvae or other flower or seedling herbivores
(Smallegange et al., 2007; Ahuja et al., 2010; Bandeili and Müller,
2
010) if the compound is extremely bioactive. If the previously
described saponins from leaves of the species are present in floral
parts, saponins rather than acylated glucosinolates may be respon-
sible for reduced orange tip larval feeding.
2.3. Identification of additional seed glucosinolates from the genus
Barbarea
As our past taxonomic screen of Barbarea glucosinolates only fo-
cused on leaves and roots (Agerbirk et al., 2003) and obviously had
missed a part of the glucosinolate diversity of the genus, we
decided to extend our past taxonomic screen of the genus to in-
clude seed GSLs. During this screen, based on HPLC-MS and
HPLC-DAD of desulfated GSL derivatives, four additional putative
isoferuloylated GSLs were discovered by inspection of HPLC-MS
data and UV328nm chromatograms (Fig. 3, Fig. 5). The P-type of
B. vulgaris var. arcuata showed signals of all four compounds, and
an HPLC-NMR approach (supplemented by analysis of products
of esterase-catalysed hydrolysis) allowed isolation and structural
elucidation of all four from a single HPLC gradient (Fig. 6). The elu-
Fig. 3. Chromatograms illustrating the diversity of glucosinolates (GSLs) in two
chemotypes of Barbarea vulgaris and the output of different analytical methods.
Chemotype and detection wavelength as indicated on each chromatogram. A–C:
Desulfoderivatives analysed by HPLC, peaks of desulfated GSLs indicated with the
symbols of the corresponding intact GSLs for simplicity. Insert in A shows UV
0
0
cidated structures were 6 -isoferuloyl2R (6 -iF-2R), previously pro-
posed by Sørensen (1990), and three additional acylated GSLs that
0
0
0
had not previously been proposed: 6 -isoferuloyl3R (6 -iF-3R), 6 -
0
0
0
spectrum of desulfo 6 -iF-2S. D: Intact GSLs isolated by ion exchange chromatog-
isoferuloyl-phenethylGSL (6 -iF-1), and 6 -isoferuloyl-indol-3-yl-
methylGSL (6 -iF-4) (Fig. 1). Due to our experiences with desulfo
-iF-2S and an observed very low water solubility of desulfo 6 -
iF-1, CD OD was selected as NMR solvent for all four compounds
(Table 1, Table 2).
raphy and analysed by HPLC using acidic eluent, selected peaks identified by
isolation and subsequent MS and conversion to desulfoderivatives (see text). E, F:
Crude extracts (after solvent change to water) analysed at the same conditions as
used in D, with major GSLs visible as well separated peaks at both 229 nm (E) and
0
0
0
6
3
3
28 nm (F). Insert in F shows UV spectrum of dominating peak at 31 min.

N. Agerbirk, C.E. Olsen / Phytochemistry 72 (2011) 610–623
615
0
Fig. 4. Delayed elution of isoferuloylated glucosinolates (GSLs) and desulfoGSLs from carbohydrate based ion exchange columns (DEAE-Sephadex A-25, 100 mg dry wt. in
0
both cases). A: Elution of desulfoGSLs by H
for 2R and 6 -iF-2R. B: Elution of intact GSLs by K
2
O upon over night incubation with sulfatase. Results obtained for 2S and 6 -iF-2S (not shown) were similar to the results obtained
0
2
SO
4
(sat) in H
2
O–EtOH (95:5). The three last 2 mL fractions (24–28 mL) were eluted after an overnight pause.
Table 3
Glucosinolate contents (
of the G-type of Barbarea vulgaris var. arcuata, either cultivated or collected in nature.
from a C-H correlated spectrum (HSQC). The chemical shifts of
both C6 and H6 deviated significantly from the corresponding val-
0
0
l
mol/g dry wt., mean) of various floral and vegetative parts
ues for the non-acylated 2R (Table 1, Table 2), suggesting that the
0
isoferuloyl group was positioned at C6 (Nishida et al., 2003). This
0
Plant part
1
2R
2S
6 -iF-2S
4
6
N
0
suggestion was confirmed by an HMBC interaction between the 6
Cultivated, accession B44
Upper leaves (May)
Flowers (May)
Siliques (May)
Siliques (July)
0.2
0.05
0.1
0.05
0.7
1.6
1.1
1.1
0.3
1.1
92
90
90
17
57
(0.05)
(0.16)
(0.14)
(0.14)
3.0
0.9
2.1
2.0
0.1
0.4
0.00
0.00
0.01
0.00
0.00
1
1
1
1
5
hydrogen in the thioglucoside moiety and the isoferuloyl carbonyl
0
carbon as also found for d6 -iF-2S.
0
0
We found d6 -iF-2R to be indistinguishable from d6 -iF-2S by
0
our routine HPLC-DAD analysis, indicating that the epimers d6 -
Mature seeds
0
Collected in naturea
iF-2S and d6 -iF-2R had the same UV spectra and HPLC retention
Upper leaves
Flowers
Siliques (June)
0.1
0.2
0.03
1.0
1.0
0.4
77
81
23
(0.06)
(0.06)
n.d.
0.6
1.9
0.2
0.09
0.03
0.03
3
3
2
times at our conditions. Two other column types (C18 and phenyl-
hexyl stationary phases) were tested at a range of temperatures
(
25–55 °C), resulting in a weak partial separation (0.2 min, S-
n.d., not detected. Values in parentheses are maximum estimates based on trace
peaks.
epimer eluting first) of the epimers which was rather independent
of temperature. Chromatographic distinction of the epimers was
a
Sampled in May–June at the orange tip butterfly locality DK1 (Agerbirk et al.,
later acomplished in other ways as described in sections 2.4. and
2
010c).
0
2
.5. Upon esterase-catalysed hydrolysis of d6 -iF-2R, the increase
of the peak area at 229 nm of d2R (and a trace of d2S) was 34%
0
of the decrease of the peak area of d6 -iF-2R, in slight contrast to
0
Table 4
the value of 39% for d2S relative to d6 -iF-2S (Section 2.2.). How-
Glucosinolate contents of B. vulgaris (G-type) seeds, seedlings, and liquid
medium + filter paper after germination for 5 or 7 days.
ever, we did not consider this difference significant, and decided
0
to use the same HPLC response factor for quantification of 6 -iF-
0
2R and 6 -iF-2S.
0
N
1
2R
2S
4
5
6
6 -iF-2S
A less prominent desulfoGSL eluting ca. 4 min earlier was iden-
tified as d6 -iF-3R by a similar combination of methods. The exact
nmol per seed(ling)
0
Dry seeds
a
m/z value from high resolution MS was compatible with the molec-
1
0.09
0.62
32.2
0.29
n.d.
n.d.
1.32
ular formula of an isoferuloyl derivative of d3R (Found 574.1358,
Seedlings
+
calc. for [C25
H
29
O
11NSNa] : 574.1354). The UV spectrum showed
5
7
d
d
1
3
1.04
0.44
0.83
0.74
35.2
33.9
0.78
1.35
0.15
0.27
2.8
2.3
0.22
0.02
b
the same isoferuloyl band at 328 nm as that of d6 -iF-2S as well
0
as a maximum at 219 nm attributable to the side chain phenol
group. The NMR spectra were similar to the spectra of d6 -iF-2R
Filter paper + liquid medium
c
0
7
d
2
n.d.
n.d.
0.4
0.02
n.d.
0.02
n.d.
except for the signals of the side chain phenol group as expected
from the presence of a 4-hydroxy group (Table 1, Table 2), and
a
On a weight basis, this result was comparable to other analyses of the same
accession.
0
b
0
the position of the acylation at the thioglucose moiety 6 position
The mean of three replicates is reported. RSD was 95% for 6 -iF-2S and ranged
0
0
from 7% to 13% for the remaining GSLs.
was confirmed from the chemical shifts of C6 and H6 and by
c
The mean of two replicates is reported. RSD ranged from 3% to 69%.
0
HMBC as explained for d6 -iF-2R. The identity of the acyl group
as isoferulic acid was concluded from the close similarity of the rel-
0
0
evant parts of the NMR spectra of d6 -iF-3R, d6 -iF-2R, and litera-
The dominating desulfated acylGSL from P-type seeds was iden-
tified as d6 -iF-2R as follows: The nominal m/z value, approximate
ture data (Nishida et al., 2003) and from HMBC interactions as
follows: The signal of C4 was identified with certainty from
0
000
0
00
000
000
HPLC retention time and UV spectrum were the same as those of
HMBC interactions of H2 , H5 and H6 . The main HMBC interac-
0
000
the epimer d6 -iF-2S, but incubation with esterase resulted in for-
tion of the OMe hydrogens at 3.89 ppm was also to C4 , proving
that the acyl group had a 4 -methoxy substituent as in isoferulic
0
00
mation of d2R (and trans isoferulic acid). NMR of the desulfated
0
00
derivative confirmed the expected signals of the various moieties
acid, not a 3 -methoxy substituent as in ferulic acid. Preliminary
experiments with esterase treatment showed a low hydrolysis
0
1
(
Table 1, Table 2). The signals of H6 in the thioglucose moiety were
0
identified with certainty from the H NMR including coupling ana-
lysis by COSY, because these protons constitute the only (pseudo)
rate, suggesting that d6 -iF-3R was a poor substrate of the used
porcine enzyme. For this reason, a response factor for quantifica-
tion was assumed (Section 3.3.) rather than measured experimen-
tally. But after incubation for four days, d3R and isoferulic acid
0
ABX system (with X = H5 coupled to further signals) in the thio-
0
glucose moiety. The corresponding carbon, C6 , was also identified

6
16
N. Agerbirk, C.E. Olsen / Phytochemistry 72 (2011) 610–623
0
0
0
0
0
0
Fig. 5. Virtually identical MS2 spectra of four desulfated 6 -isoferuloylated glucosinolates (d6 -iF-1, d6 -iF-2S, d6 -iF-3R, d6 -iF-4). The MS2 spectrum of d6 -iF-2R (not
0
shown) was indistinguishable from the spectrum of d6 -iF-2S. In each case, the major fragmentation pattern can be interpreted as loss of a neutral fragment corresponding to
+
+
the aglycone + H or the aglycone ꢀ SH, resulting in identical fragment ions representing the Na adduct of the isoferuloylated glucose moiety –H, [C16
18 8
H O NA] , m/z = 361, or
+
+
the isoferuloylated glucose moiety + SH, [C16
H
19
O
8
SNa] , m/z = 395. In contrast, an unidentified peak (m/z 572 for [M + Na] , lowermost panel) from accession M0097–68 gave
0
MS2 fragments with m/z values 14 higher than the above, which would be compatible with the presence of methyl 6 -iF-2S in this accession, but this structure was not
confirmed.
were detected as (the only) significant products, confirming the
suggested structure.
sponding to the isoferuloyl group should otherwise be lost along
with the aglucone fragment.
0
0
0
0
Two additional desulfoGSLs, d6 -iF-1 and d6 -iF-4, were found
at low levels, and were first suggested based on routine screen-
ing for potential isoferuloyl derivatives of the major desulfoGSLs
in each HPLC-MS chromatogram. As peaks with reasonable
retention times and m/z values for hypothetical isoferuloylated
d1 and d4 were detected, the daugther ions of the molecular
ions were also examined. The daugther ion (MS2) mass spectrum
was in each case highly similar to those of the isoferuloylated
desulfoGSLs elucidated above (Fig. 5), and this invariable daugh-
ter ion spectrum could be rationalized by assuming loss of neu-
tral fragments corresponding to the aglucone plus H and the
aglucone minus SH. The similarity of the MS2 spectrum of all
the putative isoferuloylated desulfoGSLs suggested that they
were all substituted on the glucose moiety, as the mass corre-
The minor amounts of d6 -iF-1 and d6 -iF-4 obtained by pre-
1
parative HPLC were only sufficient for conventional H NMR, but
assignment of individual proton signals except most of the thiog-
0
0
lucoside signals H1 -H5 (overlapping each other and/or strong sol-
vent signals at 4.7–4.9 ppm and 3.3 ppm) and those of the phenyl
0
group of d6 -iF-1 was possible from comparison with spectra of d1,
0
0
d4, d6 -iF-2R and d6 -iF-3R. In particular, the chemical shifts of the
0
H6 protons were similar to the chemical shifts of the correspond-
ing protons in other 6 -isoferuloylated GSLs but different from
0
chemical shifts of the corresponding protons in the non-acylated
desulfoGSLs d1 and d4 (Table 1), allowing us to conclude that the
0
isoferuloyl group was positioned at C6 in both compounds. For lo-
gistic reasons, these dGSLs were not analysed by high resolution
MS.

N. Agerbirk, C.E. Olsen / Phytochemistry 72 (2011) 610–623
617
0
that d6 -iF-1 was an isoferuloyl derivative of d1. Detection of d4
0
and isoferulic acid from d6 -iF-4 by the same procedure likewise
0
demonstrated that d6 -iF-4 was an isoferuloyl derivative of d4.
The peak area increases at 229 nm of d1 and d4 when these sam-
ples were analysed with the standard desulfoGSL HPLC method
were 40% and 72%, respectively, of the decreases of the isoferuloy-
lated peaks. Assuming the response factor of 1 to be 0.95 and that
of 4 to be 0.29, the response factors of the isoferuloyl derivatives
could hence be assumed to be 0.38 and 0.21, respectively. We used
0
the latter factor in subsequent analyses, but assumed 6 -iF-1 to
0
have the same response factor as 6 -iF-2S like we did in the case
0
of 6 -iF-2R. The HPLC-DAD UV spectra of these minor GSLs were
0
in accordance with their structures: the spectrum of d6 -iF-1 was
0
0
identical to that of d6 -iF-2R/S, while that of d6 -iF-4 showed
bands at 218 nm and 289 nm attributable to the indolyl group
and a 328 nm band from the isoferuloyl group.
Identification of these novel aromatic GSLs adds new members
to the steadily growing list of documented indole GSLs (reviewed
by Agerbirk et al., 2009), acyl conjugates of GSLs (Frechard et al.,
2
001; Reichelt et al., 2002; Kliebenstein et al., 2007) and GSLs in
general (Clark, 2010; Agerbirk et al., 2010b).
0
A small peak immediately before d6 -iF-2S in the desulfogluco-
sinolate chromatogram of G-type seeds (Fig. 3A) did not show
absorbance above 300 nm, and was therefore not a potential isof-
eruloylated desulfoglucosinolate. This peak was not investigated
further. During the preparative HPLC for NMR of P-type desulfog-
lucosinolates, a number of very low intensity UV peaks were de-
tected (Fig. 6) in addition to the four peaks identified above. A
tiny peak at 35 min had kmax = 322 nm, two tiny peaks at 36.9
and 37.4 min both had kmax = 328 nm, a tiny peak at 40.9 min
had kmax = 231 and 324 nm, and a tiny peak at 41.3 min had
kmax = 302 and 327 nm. All these five peaks from P-type seeds
could hence represent acylated glucosinolates at trace levels, some
of them perhaps with different acyl groups, but their identification
was not considered to be realistic at the time of the experiment.
Unfortunately, the peaks were not even collected for MS investiga-
tion. The UPM accession M0097–68 gave an additional small
+
[
M + Na] peak in HPLC-MS with m/z = 572 (14 higher than the
0
m/z value corresponding to d6 -iF-2S) and with retention time
.6 min higher than that of d6 -iF-2S in HPLC-MS. The MS2 spec-
0
0
trum suggested that the additional 14 mass units were associated
with the acylated thioglucose moiety (Fig. 5 lowermost panel), as
would for example be the case for a conjugate of d2S with O-
methyl isoferulic acid (or another dimethoxy cinnamic acid). As
0
the peak height was only 3% of the peak height of d6 -iF-2S and less
than 50 mg seeds were available, this possible additional minor
GSL was not further investigated, but its detection suggests that
acylation with additional aromatic acids may occur in the genus,
and illustrates the utility of MS2 spectra for distinction of thioglu-
cose moiety-acylated GSLs (Reichelt et al., 2002).
2.4. Intact GSL analysis compared to desulfoglucosinolate analysis
A standard HPLC gradient (modified from Mellon et al., 2002)
was apparently well suited for separation of intact Barbarea GSLs,
isolated as a group by a standard ion exchange chromatography
procedure (modified from Thies, 1988) employing salt elution
(Fig. 3D). But elution of the isoferuloylated GSLs from the DEAE
ion exchange column was obviously not quantitative when using
the normal elution volume of ca. ten column volumes (peaks of
Fig. 6. Identification of trace level and medium level isoferuloyl glucosinolate peaks
by HPLC-NMR of P-type B. vulgaris var. arcuata (Fig. 3B + C). Upper panel: UV trace
1
(
328 nm). Lower panels: H NMR of four peaks. Assignment of selected signals is
0
00
000
000
indicated. For condensation, ‘‘iF’’ signifies protons 2 , 5 and 6 of the isoferuloyl
group. S = solvent-associated signals.
0
0
6
-iF-2R and 6 -iF-3R were much lower than expected, results
not shown). Extended elution of an analytical ion exchange column
was performed with the hope of establishing satisfactory elution
conditions.
Unfortunately, complete elution was not accomplished
(Fig. 4B): While Barbarea GSLs without substituents on the thioglu-
0
A further test of the identity of the putative isoferuloylated d6 -
0
iF-1 and d6 -iF-4 was carried out by analysis of products of ester-
ase treatment: isoferulic acid and d1 were produced from d6 -iF-1
demonstrated by HPLC-DAD with acidified eluent), demonstrating
0
(

6
18
N. Agerbirk, C.E. Olsen / Phytochemistry 72 (2011) 610–623
cose moiety eluted relatively early with saturated K
2
SO
4
from the
(Fig. 3A–D). Analysis of desulfated derivatives had a number of
advantages compared to analysis of group separated intact GSLs,
notably a better recovery of an acylated phenolic GSL with strong
ion exchange column, the elution of isoferuloylated GSLs was
markedly delayed: Even after stepwise elution with 39 column vol-
umes from a 100 mg dry wt. (ca. 0.7 mL) column, the elution curve
0
affinity for the used ion exchange column material (6 -iF-3R).
0
0
suggested that complete elution of 6 -iF-3R had not been achieved.
However, chromatographic distinction of the epimers 6 -iF-2R
0
0
0
Elution of 6 -iF-2R and 6 -iF-3R peaked around 17 and 22 column
volumes, respectively, in contrast to non-substituted GSLs (2R, 3R)
which peaked around four column volumes. Comparison with the
elution profile of desulfoGSLs (Fig. 4A) suggested that the affinities
of isoferuloyl groups and phenol groups for the column material
had a much greater effect in the presence of a sulfate group (an
expectable effect (‘‘avidity’’) of the combination of several indepen-
dent binding sites), even at the high ionic strength employed for
elution.
and 6 -iF-2S was better accomplished by analysis of intact GSLs
and was possible even with crude extracts. In this way, the desul-
foglucosinolate method and direct analysis of crude extracts com-
plemented each other well, while the tested method for group
separation of intact glucosinolates would need considerable
improvement in order to be useful in an analytical context. Elution
0
of the indole glucosinolate 6 -iF-4 after desulfation was not specif-
ically optimized (Fig. 4A), and given the known affinity of indolyl
groups for DEAE column material (Agerbirk et al., 2001a), the
5 mL elution volume used here may need to be moderately in-
creased for full recovery.
Using fractions from this experiment, identification of each
individual peak was accomplished by collection of that peak fol-
lowed by mass spectrometry and sulfatase catalysed conversion
to the corresponding desulfoGSL, identified by HPLC retention
time. In order to avoid the majority of the dominating peak 2R,
the first 4 mL fractions from the ion exchange column were
HPLC-MS and HPLC-NMR (with manual peak collection) were
employed for identification of both major and minor peaks in
one sample. In contrast to the MS detection that allowed a partial
structure elucidation (from MS2 spectra demonstrating acylation
of the thioglucose moiety, Fig. 5), the NMR detection allowed struc-
avoided for preparative HPLC. Molecular masses (as m/z-values of
-
0
[
M-H] intact GSL ions) and HPLC-DAD retention times (min) of
tural elucidation to a specific positional isomer (6 -acylation,
desulfated derivatives were: 2R (438, 17.2); 2S (438, 18.5); 4
Fig. 6). Insufficient distinction of isomers is a frequently experi-
enced problem in HPLC-MS analysis of GSLs that can be solved
by NMR (Agerbirk et al., 2010b), and the results presented here
confirms the potential of HPLC with NMR detection. In order to
minimize the work-load involved in HPLC-NMR, use of an auto-
mated interphase for systematic HPLC-NMR (Sprogø et al., 2007)
is currently considered.
0
0
0
(
(
447, 23.9); 6 -iF-1 (598, 42.8); 6 -iF-2R (614, 39.0); 6 -iF-3R
630, 35.4). Isolated peak 3R was lost, but the m/z value was 454
as expected in the combined sample. A moderate size peak labeled
‘b’’ in Fig. 3D gave an m/z-value of 601 in MS and was not a sub-
‘
strate of sulfatase. As the UV spectrum showed absorption bands
at 227 and 277 nm but not at 328 nm expected for an isoferuloylat-
ed GSL, the identity of this peak was not further investigated. Two
other peaks labelled ‘‘?’’ in Fig. 3D were not identified due to small
size (27 min) or vicinity to a major peak (45 min).
0
The previously suggested existence of 2 -isoferuloylated and
0
0
2 ,6 -diisoferuloylated GSLs in B. vulgaris seeds (Sørensen, 1990;
Bellostas et al., 2007) could neither be confirmed nor rejected de-
spite investigation of both sulfatase-treated and intact GSLs. At
present, it cannot be concluded whether these hypothetical com-
pounds exist (Fahey et al., 2001), and whether they would be
detectable at any of the conditions tested here. A considerable
number of trace level peaks with absorption at 328 nm was de-
tected during the HPLC separation for NMR (Section 2.3.), but the
mere detection of these UV peaks is not sufficient evidence for con-
firmation of the hypothetic structures – indeed, many other struc-
tures could be responsible for these trace peaks, including acyl
derivatives at the side chain OH of 2R, 2S and 3R, double bond
cis-isomers of isoferuloylated GSLs, and additional kinds of acyl
groups (Linsheid et al., 1980; Reichelt et al., 2002; Section 2.3.).
Rather than searching for more efficient elution regimes for in-
tact GSLs, we tested direct HPLC analysis of crude extracts as an
analysis method that could potentially reveal all major GSLs pres-
ent, unbiased from differences in recovery during group separa-
tion. Both direct injection of crude extracts (in ca. 50–70% MeOH)
and injection after solvent change to pure water (Fig. 3E) resulted
in a less than satisfactory peak shape of 3R, while the appearance
of the late two thirds of the chromatogram appeared identical with
both solvents. The major GSLs could be clearly identified in chro-
matograms of crude extracts, but many apparently non-GSL peaks
were present which might interfere with minor GSL peaks.
Inspection of crude extract chromatograms at 328 nm was an
0
0
attractive way of detecting isoferuloylated GSLs such as 6 -iF-2R.
We conclude that 6 -isoferuloyl derivatives of at least five dif-
0
0
0
0
This simple screen also revealed a major and a number of minor
potentially acylated metabolites that might be of interest in future
investigations of non-GSL aromatic acid derivatives (Fig. 3E). Possi-
ble identities include choline esters of isoferulic and coumaric
acids (Andersson et al., 1999) or aromatic ester conjugates of ma-
late, flavonoids or carbohydrates (Baumert et al., 2005). Partial sep-
ferent B. vulgaris seed GSLs exist (6 -iF-1, 6 -iF-2R, 6 -iF-2S, 16 -
0
iF-3R, 6 -iF-4) and can be quantified using the desulfoGSL method,
and that no further acylated GSLs have so far been documented in
the genus Barbarea.
2.6. Seed glucosinolate variation in the genus Barbarea
0
0
aration of intact 6 -iF-2R and 6 -iF-2S was also possible in crude
extracts, but constant temperature was needed in order to obtain
sufficiently stable retention times. At a column temperature of
The profiles of non-acylated GSLs (Table 5) in seeds of the inves-
tigated taxa had in common that the dominating GSL was appar-
ently derived from homoPhe: most taxa were dominated by 1, 2S
or 2R. For the species with previously reported leaf GSL profile
(Agerbirk et al., 2003), the seed and leaf GSL profiles were similar
(in contrast to root GSL profiles which in general contain higher
levels of 1 and the substituted indole GSLs 5–8). Presence of 3R
in P-type B. vulgaris seeds and of 3S in B. stricta seeds was also in
accordance with leaf GSL profiles (Agerbirk et al., 2003a). A sugges-
tion of B. intermedia seeds being dominated by 2R (Andersson et al.,
0
0
2
(
5 ꢀC , intact 6 -iF-2S eluted earlier (41.8 min) than 6 -iF-2R
42.2 min). This order of elution was in contrast to the order of elu-
tion of the corresponding free intact GSLs, with 2S eluting at
9.9 min and 2R eluting at 18.4 min (at 25 ꢀC ). Retention times at
1
ambient temperature (ca. 20 ꢀC , Fig. 3D–F) were slightly higher,
but with the same elution order.
2.5. Analytical and structural conclusion
1
999, without experimental details) could not be confirmed, we
The specificities of the group separation procedures, whether
intact or desulfated GSLs were prepared, resulted in simple chro-
matograms in which all or nearly all peaks corresponded to GSLs
have previously argued that results attributed to B. intermedia from
that period and laboratory may in reality have been obtained with
mis-identified specimens of the P-type of B. vulgaris var. arcuata

N. Agerbirk, C.E. Olsen / Phytochemistry 72 (2011) 610–623
619
0
(
Agerbirk et al., 2001b). Dominance of 1 was frequent in the genus:
The majority of the tested species exhibited considerable levels of
in seeds (B. intermedia, B. verna, B. bracteosa, B. plantaginae, and
of 6 -iF-1 were only seen in those accessions were 1 was a
0
dominating seed GSL (Table 5), the dominating epimer of 6 -iso-
1
feruloylated 2R/2S corresponded to the dominating epimer of
non-acylated 2R/2S (this conclusion was only based on epimer-
identification of a subset of the accessions, see footnote to Table 5),
and 6 -iF-3R was only seen in P-type accessions containing 3R
(Table 5, Fig. 3). In the case of 6 -iF-1 in the Nordgen accessions,
B. australis). A further species, Barbarea rupicola was also recently
reported to be dominated by 1 in leaves (Badenez-Pérez et al.,
0
2
011) as was an interesting Dutch ‘‘NAS’’ chemotype of B. vulgaris
with very high levels of 1 and only traces of 2R/2S (Van Leur et al.,
006). High levels of both 1 and 2S is hence typical of roots, seeds
0
2
a correlation analysis seemed relevant because of the considerable
and sometimes of leaves of many Barbarea species and may be
more common than previously thought. In this respect, the GSL
profile of B. australis, not previously investigated, was typical of
the genus, with seeds and roots dominated by both 1 and 2S but
leaves dominated by 2S only (Table 5 and results not shown).
The low frequency in the genus of 2R as dominating GSL was
confirmed, but B. plantaginae would seem to share this character
with the P-type of B. vulgaris (Table 5). In this connection, it is
interesting that a major gene for 2R in the P-type was mapped
to another genetic locus than a major gene for 2S in the G-type
variation in levels of 1 and the common origin of the accessions.
0
2
Levels of 1 (x) and 6 -iF-1 (y) were highly correlated (R = 0.9966),
and linear regression resulted in the equation y = 0.0553 x + 0.0004,
confirming that levels of the derivative depended in a simple
manner on levels of the free GSL. We interpret these correlations
0
as the outcome of an un-specific mechanism for 6 -isoferuloylation
of all B. vulgaris seed glucosinolates. According to this interpreta-
0
tion, the repeatedly observed low level of 6 -iF-4 (Table 3, Fig. 6)
may simply be due to the low levels of non-acylated 4, rather than
any lack of potential for acylation of this GSL. Extending genetic
modification of GSL profiles (e.g. Bak et al., 1999; Wittstock and
Halkier, 2000) to B. vulgaris would be an elegant way of testing this
(Kuzina et al., 2011). It seems likely that a gene for 2R (with low
expression in most species) is present in the entire genus (Table 5).
Possibly, the reversed 2R/2S ratio in the P-type is due to a gene
expression mutation rather than a structural mutation.
0
hypothesis and possibly creating plants with novel 6 -isoferuloy-
0
lated GSLs. If we accept the hypothesis of lack of specificity in 6 -
In B. verna, we and others had previously not reported 2R and 2S
from HPLC analyses with UV detection (Andersson et al., 1999;
Agerbirk et al., 2003a). But with the more sensitive HPLC-MS, we
detected levels of these GSLs well above the detection limit. Corre-
sponding trace peaks in HPLC-DAD were also observed (Table 5).
To test whether the apparent presence of 2R and 2S in B. verna
could be an artifact due to contamination from other species ex-
tracted in parallel, we repeated the extraction on a separate day,
with extraction of Brassica nigra (L.) Koch seeds in parallel, and
HPLC-MS of the resulting desulfated GSLs in a separate analytical
sequence. In the control sample from B. nigra seeds, 2R and 2S were
not detected, while the samples from B. verna contained similar
levels of 2R and 2S as observed in the first block of experiments
isoferuloylation, specific determination of the dominating epimer
of 6 -iF-2R/S by chromatography of crude extracts may not be
0
needed for routine GSL analysis of Barbarea seeds by the desul-
foGSL analysis method, as the ratio can probably be inferred from
the ratio of the non-acylated epimers that are well resolved with
this method. The hypothesis of a rather un-specific isoferuloylation
mechanism also accentuates the relevance of identifying non-GSL
peaks in crude extracts (Fig. 3E). However, this task was not within
the scope of the present investigation.
0
The second variable character was the extent of 6 -isoferuloylation.
According to the discussion above, the logical character to analyse
in a taxonomical screen is the presence and intensity of isofer-
uloylation as such, irrespective of the actual GSL carrying the
isoferuloyl group. Within the species B. vulgaris levels spanned
two orders of magnitude, with high levels in accessions of the P-
type and the G-type of var. arcuata and two accessions of var. vul-
garis, an intermediate level in the North American accession B15,
and low levels in three accessions of var. vulgaris. Among the addi-
tional Barbarea species investigated, only B. plantaginae accumu-
(Table 5). The two previously unnoticed B. verna GSLs were in an
approximate 1:1 ratio, in contrast to the highly skewed ratios in
all other Barbarea species, suggesting that their quantitatively
reproducible 1:1 presence in the analysed seeds was not due to
contamination of the B. verna seed batches with low levels of an-
other Barbarea species. We concluded that 2R and 2S are minor
constituents of B. verna seeds, often in approximately 1:1 ratio.
In contrast to an earlier interpretation of ours (Agerbirk et al.,
0
0
lated relatively high levels of 6 -isoferuloylated GSLs, with 6 -iF-1
being dominant in agreement with 1 being the dominant non-
acylated GSL in the species. This accession, as well as various acces-
sions from Nordgen and accession B4 (deposited as UPM 9971–05),
2
003a), B. verna is hence not qualitatively different from the other
species with respect to occurrence of 2-hydroxylation of homoPhe
derived GSLs, but rather an extreme of a continuum from very high
over medium to very low extent of such hydroxylation.
As we learned that a huge number (569) of B. vulgaris accessions
were kept at the Nordic gene bank ‘Nordgen’, all deposited by the
same collector not associated with our laboratory, we decided to
analyze five of those, kindly selected by the Nordgen staff with
the purpose of maximizing diversity. These accessions were quite
diverse chemically, with relatively high levels of 2R in all of them
0
may serve as convenient reference materials for the 6 -isoferuloy-
lated GSLs reported here. We plan to deposit additional P-type and
G-type seeds at Nordgen.
2.7. Perspectives
Acylation of the thioglucose moiety has rarely been considered
in structural investigations (Linsheid et al., 1980; Reichelt et al.,
2002). However, this group of GSLs may often have been over-
looked simply due to routine use of relatively short HPLC gradients
or low elution volumes after desulfation or ion exchange. As the
rather closely related species A. thaliana (Windsor et al., 2005) also
contains acylated seed GSLs (Reichelt et al., 2002), albeit with a dif-
ferent acyl group, a search for acylated GSLs in other Arabidopsis
and Barbarea relatives may be worthwhile. Biosynthesis of thioglu-
cose moiety acylated glucosinolates have not so far been investi-
gated (Halkier and Gershenzon, 2006; Sønderby et al., 2010) in
contrast to the mechanism of side chain acylation in A. thaliana
(Graser et al., 2001; Kliebenstein et al., 2007; Ibdah and Pichersky,
2009). Catabolism in vivo of thioglucose moiety acylated gluco-
sinolates has also not been investigated. Given the substrate
(
but still dominance of 2S) and high levels of 1 in three of them.
This variation may possibly be a result of P-type ꢁ G-type hybrid-
ization and subsequent genetic recombination (Agerbirk et al.,
2
001b; Kuzina et al., 2011), and/or involve hybridization with the
NAS genotype of B. vulgaris (van Leur et al., 2006).
0
The 6 -isoferuloylated GSLs focused on in this paper were pres-
ent in all analysed members of the genus suggesting that this type
of seed GSL is a general trait in the genus. However, the kinds and
levels of these GSLs varied in two interesting ways. The first vari-
able character was the kind of isoferuloylated GSLs seen. Within
the B. vulgaris accessions, it was obvious that the relative levels
of isoferuloylated GSLs reflected the relative levels of correspond-
ing non-acylated GSLs (Table 5, Fig. 3). In particular, high levels

6
20
N. Agerbirk, C.E. Olsen / Phytochemistry 72 (2011) 610–623
Table 5
Seed glucosinolate profiles of Barbarea accessions (lmol/g dry wt.).
0
0
a
0
0
1
2R
2S
3R
3S
4
6 -iF-1
6 -iF-2
6 -iF-3R
6 -iF-4
B. vulgaris var. vulgaris
B5
B6
B51
B52
B53
0.9
1.0
1.5
1.2
0.6
0.9
41
48
43
33
34
n.d.
n.d.
n.d.
n.d.
n.d.
tr.
tr.
n.d.
n.d.
n.d.
0.2
0.5
0.4
0.2
0.4
tr.
0.07
0.13
0.07
2.3
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
tr.
0.02
0.02
1.4
1.0
0.9
0.7
0.01
n.d.
0.08
0.09
2.4
B. vulgaris var. arcuata ‘G-type’
S
B1
0.3
0.2
0.3
0.2
1.1
0.9
0.7
1.0
56
47
31
54
n.d.
n.d.
n.d.
n.d.
tr.
tr.
tr.
tr.
0.3
0.06
0.2
0.08
0.01
0.01
0.01
3.1
2.5
0.9
2.8
n.d.
n.d.
n.d.
n.d.
0.01
0.01
0.01
tr.
S
B2
S
B15
B44
S
0.3
B. vulgaris var. arcuata ‘P-type’
R
B3
B4
0.4
0.3
29
48
0.9
0.7
1.3
1.8
0.02
0.02
0.3
0.2
0.01
0.07
4.2
5.0
0.3
0.3
0.01
0.04
R
B. vulgaris from seed banks, var. and type not known
M0097-68
N10001
N10106
N10239
N10365
N10508
0.2
0.3
19
6.0
13
0.4
0.8
12
18
13
16
37
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
0.5
0.6
0.6
0.7
0.5
0.6
0.02
0.01
1.0
0.3
0.7
2.6
2.4
2.9
2.2
2.5
2.6
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
0.07
0.05
0.06
n.d.
0.05
n.d.
54
52
33
35
31
8.5
0.03
B. intermedia
B7
B39
14
9.9
19
0.4
0.3
0.4
43
35
51
n.d.
n.d.
n.d.
tr.
tr.
n.d.
0.8
0.5
0.9
0.01
0.01
0.01
0.03
0.03
0.04
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
M1261-67
B. verna
B16
B36
61
58
88
0.03
0.01
0.04
0.02
0.01
0.1
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
0.06
0.08
0.06
0.09
0.06
0.19
tr.
tr.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
n.d.
M0192-68
B. stricta
B40
0.1
0.4
2.1
49
32
4.6
0.02
n.d.
n.d
1.6
n.d.
n.d
0.7
0.5
0.1
n.d.
0.04
3.7
0.03
0.04
1.1
n.d.b
n.d.
n.d.
n.d.
n.d.
0.02
B. bracteosa
M6203-84
41
59
B. plantaginae
M3773-75
17
B. australis
B45
B48 (N23039)
45
33
0.3
0.4
29
40
n.d.
n.d.
tr.
tr.
0.6
1.2
0.09
0.04
0.05
0.04
nd.
nd.
n.d.
n.d.
n.d.; not detected. tr.: trace.
a
0
0
0
Measured by HPLC of desulfoderivatives as the sum (‘‘6 -iF-2’’) of the epimers 6 -iF-2R and 6 -iF-2S. The predominant epimer, determined for selected accessions by HPLC
of intact GSLs in crude extracts (and in the cases of B1 and B4, also by esterase-catalysed hydrolysis of the desulfoGSL and subsequent identification of the corresponding non-
acylated desulfoGSL by HPLC), is indicated by a superscript R or S.
b
0
0
No GSL with the proper m/z ratio of neither 6 -iF-3R nor the hypothetical 6 -iF-3S detected by HPLC-MS.
recognition of all thioglucose moiety hydroxy groups of the few
well investigated plant myrosinases (Burmeister et al., 1997), it is
possible that acylated glucosinolates require de-acylation before
catabolism. However, more recently discovered expansion of
myrosinase diversity (Huseby et al., 2005; Bednarek et al., 2009)
should induce us to be extremely cautious in predictions of
myrosinase capabilities. The recent genetic ablation of myrosin
cells from crucifer seeds (Borgen et al., 2010) may lead to a future
increase of our understanding of crucifer germination including the
possibility of yet unknown myrosinase-like enzymes.
It is noticeable that brassicaceous plants also contain enzymes
and protein factors that can convert GSL aglucones to aromatic car-
boxylic acids by the combined action of plant nitrile specifier pro-
tein (Kissen and Bones 2009; Burow et al., 2009) and plant
nitrilases (Agerbirk et al., 2008, 2010; Janowitz et al., 2009). An
example is enzymatic formation of a coumaric acid precursor from
3R in Arabis hirsuta (L.) Scop ingested by larvae of the butterfly Pie-
ris rapae (L.) (Agerbirk et al., 2010a). The precursor, 2-hydroxy-3-
(4-hydroxyphenyl)propanoic acid, spontaneously dehydrated to
form coumaric acid in acidic solution. This means that free aro-
matic glucosinolates also represent potential sources of aromatic
carboxylic acids during germination or other developmental
phases involving glucosinolate turn-over.
The apparent restriction of thioglucose moiety-acylated GSLs to
seeds (with rapid degradation during germination) may suggest
that they play a role as storage forms of aromatic carboxylic acids
(
and free GSLs) for use during germination, thereby supplementing
other sources (Linsheid et al., 1980; Baumert et al., 2005). The iso-
feruloyl GSLs in Barbarea are a puzzling contrast to an isoferuloyl
containing spermidin conjugate in Brassica napus L. seeds (Baumert
et al., 2005) and various sinapic acid conjugates in other brassica-
ceous plants (Milkowski and Strack, 2010), and further investiga-
tion may lead to the identification of anabolic and catabolic
enzymes with specificities matching this apparent difference in
aromatic acid metabolism.
3. Experimental
3.1. Chemicals and general experimental procedures
Ferulic acid (4-hydroxy-3-methoxycinnamic acid, No. F-3500),
sulfatase from Helix pomatia (No. S9626) and esterase from porcine
liver (No. E2884) were from Sigma (St. Louis, USA). Isoferulic acid
(3-hydroxy-4-methoxycinnamic acid, No. 103012, ‘‘predominantly

N. Agerbirk, C.E. Olsen / Phytochemistry 72 (2011) 610–623
621
trans’’), was from Aldrich (Steinheim, Germany). High resolution
MS was carried out using a Bruker micrOTOF-Q instrument (Bruker
Daltonics, Bremen, Germany), HPLC-MS was carried out as previ-
ously reported (Agerbirk et al., 2008), and NMR was carried out
same way. The combined crude extract was applied to a 0.1 g dry
wt. column of DEAE Sephadex A 25 (Pharmacia, Uppsala, Sweden)
on AcO form (achieved by incubation ON with 3 mL of aq. 0.5 M
-
AcOH-NaAcO pH 5.0, followed by wash with water). The column
1
with a Bruker Avance 400 instrument at 400.1 MHz ( H) and
was washed with 5 ꢁ 2 mL of H
2
O followed by 2 mL aq. 0.02 M
1
3
1
2
6
00.6 MHz ( C) or, in the case of intact GSLs isolated before
AcOH-NaAcO pH 5.0. Then, 0.5 mL of Helix pomatia sulfatase
purified by EtOH precipitation according Graser et al. (2001) (kept
frozen, immediately before use diluted 10ꢁ with aq. 0.02 M
AcOH-NaAcO pH 5.0) was applied to the column, which was left
ON. The resulting desulfoGSLs were eluted with 5 ꢁ 1 mL of water
1
000, with a Bruker AC250P instrument at 250.1 MHz ( H) and
1
3
2.9 MHz ( C).
3.2. Seeds and plant samples
and subjected (100 lL injection) to HPLC analysis. The HPLC appa-
Seeds of the Barbarea accessions B1, B2, B3, B4, B5, B6, B7, B15,
ratus was a LC-10AT with a SPD-M10AVP diode array detector,
B16, B36, B39, and B40 were as reported in a previous taxonomic
investigation which was documented by voucher specimens
both from Shimadzu, Kyoto, Japan, equipped with a Supelcosil
LC-ABZ column, 25 cm ꢁ 4.6 mm, 5
PA. The flow rate was 1 mL/min, and the dGSLs were eluted with
O for 2 min followed by a 48 min linear gradient of MeOH from
0 to 80%, followed by a brief wash and equilibration with H O. Peak
lm, from Supelco, Bellefonte,
(Agerbirk et al., 2003a). Some additional Barbarea accessions in-
cluded were not documented by voucher specimens for various
reasons. B. australis was a kind gift from Mick Ilowski, Threatened
Species Unit, Nature Conservation Branch, Dept. of Primary Indus-
tries, Water and Environment, Hobart, Tasmania, Australia. Seeds
of B. australis from Ouse River (March 1999) were numbered B45,
seeds of B. australis from Nive River (Jan. 2001) were numbered
B47, the localities were described by Kirkpatrick and Gilfedder
H
2
2
identification was based on comparison of retention times and UV
spectra with those of authentic standards. The order of elution
(with typical retention times/min) at these conditions were: d3R
0
(13.2), d3S (13.9), d2R (17.6), d2S (18.8), d4 (24.4), d1 (25.8), d6 -
0
0
0
iF-3R (35.8), d6 -iF-2R and d6 -iF-2S (coeluting, 39.3), d6 -iF-4
0
(
1998). Seeds propagated from B47 in a Copenhagen garden were
(41.8), d6 -iF-1 (42.7). For quantification, a solution containg
numbered B48 (unfortunately, voucher specimens of flowering
plants were not collected). Accession B48 has been deposited
0.6
lmol sinigrin was analysed in parallel (Rt = 8.4 min), and levels
(lmol/g dry wt.) in samples calculated from sample weights and
(
acc. 23039) at Nordgen (www.nordgen.org). Three additional
peak areas relative to that of desulfosinigrin, using the known rel-
ative response factors for sinigrin (1.00), 1 (0.95), 4 (0.29) and
assuming the factors of 2R and 2S to be 0.95, the factor of 3R
accessions of B. vulgaris var. vulgaris, numbered B51, B52 and B53
and all collected in Denmark, were a kind gift of J.K. Nielsen who
in each case identified the variety in the field at the time of collec-
tion. B51 was from Suserup ‘‘at the golf court’’ 19. Aug. 2004, B52
was from Tarup 9. Aug. 2008, and B53 was from Lohals 18. Aug.
0
and 3S to be 0.5 (as for 4-hydroxybenzylGSL), the factor of 6 -iF-
0
0
2R/2S and 6 -iF-1 to be 0.39 (Section 2.3.), the factor of 6 -iF-3R
0
to be 0.30, and the factor of 6 -iF-4 to be 0.21 (Section 2.3.).
2
009. Additional accessions were a kind gift from Prof. C. Gómez-
Most samples were additionally analyzed by HPLC-MS (Agerb-
irk et al., 2008) for control of peak identity and for detection of
trace analytes. Retention times and m/z values (Rt/min, m/z of
Campo of the UPM crucifer seed bank at Dept. Biología Vegetal,
Universidad Politécnica de Madrid, Spain; B. vulgaris, B. verna,
B. intermedia, B. bracteosa, and B. plantaginae, or from Nordgen;
B. vulgaris. Collection data of all five Nordgen accessions were:
Roskilde (= a Danish town) 1990. Accession numbers indicated in
Table 5 are the original numbers, with a prefix‘M’ or ‘N’ added
for ‘Madrid’ or ‘Nordgen’, respectively.
+
[M + Na] ) in HPLC-MS were d3R (2.1, 398), d3S (2.2, 398), d2R
0
(5.7, 382), d2S (5.9, 382), d4 (6.8, 391), d1 (7.3, 366), d6 -iF-3R
0
0
0
(8.1, 574), d6 -iF-2R and d6 -iF-2S (9.1, 558), d6 -iF-4 (9.6, 567),
0
d6 -iF-1 (9.9, 542).
G-type seeds of B. vulgaris var. arcuata for initial small scale iso-
1
0
lation and H NMR of desulfo 6 -isoferuloylglucobarbarin were of
accession B44 (Agerbirk et al., 2003b). Additional seeds for large
scale isolation were garden-propagated seeds from 1994 of the
G-type originating from the same locality as B44 and were a kind
gift from J.K. Nielsen. P-type seeds of B. vulgaris var. arcuata for iso-
3.4. Isolation and spectroscopic identification of desulfo derivatives of
acylated glucosinolates
After some small scale trials, extract of 19.5 gram G-type seeds
was partitioned between 16 DEAE Sephadex A25 columns (0.1 g
dry wt.) and another 16 columns of 0.2 g dry wt. Wash and sulfa-
tase addition was as specified in section 3.3, elution was with 5 mL
0
lation of desulfo 6 -isoferuloyl derivatives of 1, 2R, 3R and 4 were
garden-propagated seeds from 1993 from the same locality as B3
and were also a kind gift from J.K. Nielsen. Leaves and floral parts
of flowering plants were from B44 plants cultivated in NA’s garden
and from wild plants collected at the orange tip butterfly locality
DK1 (Agerbirk et al., 2010c). Seed germination took place in Petri
dishes with filter paper moistened with 1 mg/mL gibberelic acid
H
2
O (small columns) or 10 mL H
pooled, freeze dried, re-dissolved in H
to semi-preparative HPLC on the analytical column specified in
section 3.3., injecting 400 L per run.
2
O (large columns). Eluates were
2
0 (3.6 mL) and subjected
l
Acylated GSLs from P-type seeds were isolated and identified in
a similar way: Extract of 40 gram seeds was distributed on 40
DEAE columns (each 0.1 g dry wt.). A solution of crude sulfatase
(9 mg powder in 0.5 mL buffer) was added to each column in the
desulfation step. Of 6 ꢁ 1 mL eluate, the first mL was discarded
to obtain an eluate (2–6 mL) enriched in isoferuloylated GSLs that
was freeze dried, re-dissolved in ca. 14 mL of water + ca 4 mL 50%
(
3
GA , Sigma) incubated in a dark drawer at room temperature,
and 65 seedlings (incl. roots) were extracted directly for each anal-
ysis. For comparison, 65 dry seeds (ca. 50 mg) were lightly crushed
in a mortar and extracted in parallel.
3
.3. Determination of glucosinolates by the desulfoglucosinolate
method
MeOH, and separated by HPLC (44 injections, each of 400 lL) with
manual peak collection.
0
Seeds were lightly crushed in a porcelain mortar, the resulting
material was weighed (50.0 mg) and extracted in 3 mL boiling
0% aq. MeOH in a test tube periodically immersed in a boiling
water bath with visual inspection of the extent of boiling. After a
brief centrifugation, if needed, the extract was decanted from the
seed residue, which was extracted an additional two times in the
Fractions with isolated desulfo 6 -iF-2S were freed from MeOH
by partial evaporation under an air stream, frozen, freeze dried,
1
7
dissolved in
D
2
O
and subjected to analysis by
H NMR
1
3
(400.1 MHz) and C NMR (100.6 MHz), including DEPT, J-resolved
1
H NMR and the 2D experiments COSY, HSQC and HMBC, with
H C
dioxane as internal reference (d = 3.75, d = 67.4). Fractions with

6
22
N. Agerbirk, C.E. Olsen / Phytochemistry 72 (2011) 610–623
P-type desulfoGSLs were treated and analysed in the same way, ex-
cept that the NMR solvent was CD OD, with TMS as int. ref.
Acknowledgements
3
We are immensely grateful to the late Prof. César Gómez-
Campo (1933–2009) for his unselfish efforts to provide us and
others with crucifer seeds. We thank Jens Kvist Nielsen for
generously permitting us to analyze intact 6 -iF-2R and 6 -iF-2S
isolated by him. We also thank Birgitte B. Rasmussen for skilful
GSL isolation and analysis, J.K. Nielsen, Fredrik Ottosen, Gert
Poulsen and Mick Ilowski for their helpfulness in connection
with seed requests and seed donations, and two anonymous
reviewers for helpful comments to an earlier version of this
paper.
3
.5. Esterase-catalysed hydrolysis of acylated glucosinolates
0
0
0
Desulfo 6 -iF-2S from NMR analysis (0.1
with 60 U (20 L) of porcine liver esterase (Sigma E2884) in
00 L of 50 mM sodium phosphate buffer pH 7 for several days
at room temp in an HPLC vial. Aliquots (10 L) were subjected to
lmol) was incubated
l
2
l
l
HPLC analysis each day, and compared with a control mixture pre-
pared in the same way, except that the enzyme volume was substi-
tuted with water. HPLC at the conditions for B. vulgaris desulfoGSLs
served to determine relative changes in the substrate peak and
produced d2S. HPLC at modified conditions (Luna phenylhexyl col-
References
umn (Phenomenex, 250 ꢁ 4.6 mm, 5
lm, with aqueous eluent
Agerbirk, N., Petersen, B.L., Olsen, C.E., Halkier, B.A., Nielsen, J.K., 2001a. 1,4-
Dimethoxyglucobrassicin in Barbarea and 4-hydroxyglucobrassicin in Arabidopsis
and Brassica. J. Agric. Food. Chem. 49, 1502–1507.
Agerbirk, N., Olsen, C.E., Nielsen, J.K., 2001b. Seasonal variation in leaf gluco-
sinolates and insect resistance in two types of Barbarea vulgaris ssp. arcuata.
Phytochemistry 58, 91–100.
Agerbirk, N., Ørgaard, M., Nielsen, J.K., 2003a. Glucosinolates, flea beetle resistance,
and leaf pubescence as taxonomic markers in the genus Barbarea (Brassicaceae).
Phytochemistry 63, 69–80.
Agerbirk, N., Olsen, C.E., Bibby, B.M., Frandsen, H.O., Brown, L.D., Nielsen, J.K.,
Renwick, J.A.A., 2003b. A saponin correlated with variable resistance of Barbarea
vulgaris to the diamondback moth Plutella xylostella. J. Chem. Ecol. 29, 1417–
1433.
acidified by addition of trifluoroacetic acid to 0.1%, same linear gra-
dient ending at 80% MeOH as in section 3.3.) served to identify the
remaining product as isoferulic acid (Rt 36.5 min, kmax 323 nm and
2
96 nm), not ferulic acid (main component of authentic standard
Rt 35.5 min, kmax 323 nm, shoulder at 296 nm). Additional reten-
0
tion times: d2S 23.8 min, d6 -iF-2S 42.4 min. Esterase treatment
0
0
0
of four acylated GSLs from P-type seeds (6 -iF-1, 6 -iF-2R, 6 -iF-
0
3
R, 6 -iF-4) was carried out similarly, except that the reaction
mix contained 5–10% MeOH from the NMR solvent, with identifica-
tion of reaction products by the latter method with acidified
eluent.
Agerbirk, N., Müller, C., Olsen, C.E., Chew, F.S., 2006. A common pathway for
metabolism of 4-hydroxybenzylglucosinolate in Pieris and Anthocaris
(
Lepidoptera: Pieridae). Biochem. Syst. Ecol. 34, 189–198.
3
.6. Identification of glucosinolates in their intact form
Agerbirk, N., Warwick, S., Hansen, P.R., Olsen, C.E., 2008. Sinapis phylogeny and
evolution of glucosinolates and specific nitrile degrading enzymes.
Phytochemistry 69, 2937–2949.
Agerbirk, N., De Vos, M., Kim, J.H., Jander, G., 2009. Indole glucosinolate breakdown
and its biological effects. Phytochem. Rev. 8, 101–120.
Agerbirk, N., Olsen, C.E., Poulsen, E., Jacobsen, N., Hansen, P.R., 2010a. Complex
metabolism of aromatic glucosinolates in Pieris rapae caterpillars involving
nitrile formation, hydroxylation, demethylation, sulfation, and host plant
dependent carboxylic acid formation. Insect Biochem. Mol. Biol. 40, 126–
For comparison with the desulfoGSL method, intact GSLs were
determined by group separation on ion exchange columns and
analysis by HPLC with weak acid ion pairing and UV detection
(Mellon et al., 2002; Agerbirk et al., 2010a). Only seeds of the P-
type were analysed in this way. Crude extract (eq. to 300 mg seeds)
as above, in 70% aq. MeOH, was applied to a 0.1 g dry wt. column of
DEAE Sephadex exactly as section 3.3. The column was washed
137.
Agerbirk, N., Olsen, C.E., Chew, F.S., Ørgaard, M., 2010b. Variable glucosinolate
profiles of Cardamine pratensis (Brassicaceae) with equal chromosome numbers.
J. Agric. Food. Chem. 58, 4693–4700.
Agerbirk, N., Chew, F.S., Olsen, C.E., Jørgensen, K., 2010c. Leaf and floral parts feeding
by orange tip butterfly larvae depends on larval position but not on
glucosinolate profile or nitrogen level. J. Chem. Ecol. 36, 1335–1345.
Ahuja, I., Rohloff, J., Bones, A.M., 2010. Defence mechanisms of Brassicaceae:
implications for plant-insect interactions and potential for integrated pest
management. A review. Agron. Sustain. Dev. 30, 311–348.
Andersson, A.A.M., Merker, A., Nilsson, P., Sørensen, H., Åman, P., 1999. Chemical
composition of the potential new oilseed crops Barbarea vulgaris, Barbarea verna
and Lepidium campestre. J. Sci. Food Agric. 79, 179–186.
Badenes-Pérez, F.R., Reichelt, M., Heckel, D.G., 2010. Can sulfur fertilisation improve
the effectiveness of trap crops for the diamondback moth, Plutella xylostella (L.)
with MeOH-H
1 mL). Stepwise elution with 28 ꢁ 1 mL (= 40 column volumes)
SO (sat., almost 0.5 M) in EtOH- H O (1:19) was performed with
2 2 2
O (7:3) (4 mL), AcOH-EtOH- H O (1:1:3), and H O
(
K
2
4
2
irregular intervals (5–30 min) between addition of each mL, except
one over night interval after 22 mL eluent. Fractions of 2 mL were
collected and each desalted by addition of 96% aq. EtOH (3 mL),
freezing and a brief centrifugation (4000 g, 2 min) to remove pre-
cipitated K
of atmospheric air ON, re-dissolved in H
by HPLC using the phenylhexyl column with acidified eluent spec-
ified in section 3.5. A representative mixture of all fractions (100
2
SO
4
. Supernatants were dried under a gentle stream
2
O (1 mL) and analysed
(
Lepidoptera: Plutellidae)? Pest Manag. Sci. 66, 832–838.
lL
Badenes-Pérez, F.R., Reichelt, M., Gershenzon, J., Heckel, D.G., 2011. Phylloplane
location of glucosinolates in Barbarea spp. (Brassicaceae) and misleading
assessment of host suitability by a specialist herbivore. New Phytologist 189,
of each 2 mL fraction) was reconstituted and analysed at the same
HPLC conditions to represent the GSL profile of the seeds (Fig. 3D).
A mixture of the remaining eluate between 6 mL and 28 mL was
549–556.
Bak, S., Olsen, C.E., Petersen, B.L., Møller, B.L., Halkier, B., 1999. Metabolic
engineering of p-hydroxybenzylglucosinolate in Arabidopsis by expression of
the cyanogenic CYP79A1 from Sorghum bicolor. Plant J. 20, 663–671.
Bandeili, B., Müller, C., 2010. Folivory versus florivory–adaptiveness of flower
feeding. Naturwissenschaften 97, 79–88.
Baumert, A., Milkowski, C., Schmidt, J., Nimtz, M., Wray, V., Strack, D., 2005.
Formation of a complex pattern of sinapate esters in Brassica napus seeds,
catalyzed by enzymes of a serine carboxypeptidase-like acyltransferase family?
Phytochemistry 66, 1334–1345.
Bednarek, P., Pi s´ lewska-Bednarek, M., Svatoš, A., Schneider, B., Doubsk y´ , J.,
Mansurova, M., Humphry, M., Consonni, C., Panstruga, R., Sanchez-Vallet, A.,
Molina, A., Schulze-Lefert, P., 2009. A glucosinolate metabolism pathway in
living plant cells mediates broad-spectrum antifungal defense. Science 323,
101–106.
Bellostas, N., Sørensen, A.D., Sørensen, J.C., Sørensen, H., 2007. Genetic variation and
metabolism of glucosinolates. Adv. Bot. Res. 45, 369–415.
Bjerg, B., Sørensen, H., 1987. Isolation of intact glucosinolates by column
chromatography and determination of their purity. In: Wathelet, J.-P. (Ed.),
Glucosinolates in Rape Seeds: Analytical Aspects. Martinus Nijhoff, New York,
pp. 59–75.
frozen, freeze dried, re-dissolved in 1.5 mL H
preparative HPLC (using the analytical column but injecting
00 L in each of 3 runs), with collection of 9 peaks = all major
2
O, and subjected to
4
l
peaks and some minor peaks suspected to be GSLs. Fractions with
individual collected peaks were neutralised by 1% (v/v) 3 M KAcO
to avoid TFA-catalysed hydrolysis, peak purity was checked by
HPLC-DAD, MeOH was evaporated under an air stream, water re-
2
moved by freeze drying, residue re-dissolved in 100 lL H O, and
the peak subjected to LC-MS (negative mode, conditions as in
Agerbirk et al., 2010a) for determination of molecular weights of
the isolated intact GSLs. Subsequent to LC-MS, GSL identities were
confirmed by addition of 2
2
the sulfate ester bond to occur (with 0.5 mM benzyl GSL as con-
trol), and subsequent HPLC-DAD with authentic desulfoGSLs as
references.
lL purified sulfatase (Graser et al.,
001) to each fraction, incubation ON for complete hydrolysis of
Bones, A.M., Rossiter, J.T., 2006. The enzymic and chemically induced
decomposition of glucosinolates. Phytochemistry 67, 1053–1067.

N. Agerbirk, C.E. Olsen / Phytochemistry 72 (2011) 610–623
623
Borgen, B.H., Thangstad, O.P., Ahuja, I., Rossiter, J.T., Bones, A.M., 2010. Removing
the mustard oil bomb from seeds: transgenic ablation of myrosin cells in
oilseed rape (Brassica napus) produces MINELESS seeds. J Exp Bot 61, 1683–
Michaelsen, S., Møller, P., Sørensen, H., 1992. Factors influencing the separation and
quantification of intact glucosinolates and desulphoglucosinolates by micellar
electrokinetic capillary chromatography. J. Chromatogr. 608, 363–374.
Milkowski, C., Strack, D., 2010. Sinapate esters in brassicaceous plants:
biochemistry, molecular biology, evolution and metabolic engineering. Planta
232, 19–35.
Nielsen, J.K., Nagao, T., Okabe, H., Shinoda, T., 2010a. Resistance in the plant,
Barbarea vulgaris, and counter-adaptations in flea beetles mediated by saponins.
J. Chem. Ecol. 36, 277–285.
1
697.
Borges, M.F.M., Roleira, F.M.F., Pinto, M.M.M., 1991. Isocratic HPLC separation
of scopoletin and cis/trans isomers of ferulic acid as well as isoscopoletin
and cis/trans isomers of isoferulic acid. J. Liquid Chromatography 14, 2307–
2
316.
Burmeister, W.P., Cottaz, S., Driguez, H., Iori, R., Palmieri, S., Henrissat, B., 1997. The
crystal structure of Sinapis alba myrosinase and a covalent glycosyl-enzyme
intermediate provide insights into the substrate recognition and active-site
machinery of an S-glycosidase. Structure 5, 663–675.
Nielsen, N.J., Nielsen, J., Staerk, D., 2010b. New resistance-correlated saponins from
the insect-resistant crucifer Barbarea vulgaris. J. Agric. Food. Chem. 58, 5509–
5514.
Burow, M., Losansky, A., Müller, R., Plock, A., Kliebenstein, D.J., Wittstock, U., 2009.
The genetic basis of constitutive and herbivore-induced ESP-independent nitrile
formation in Arabidopsis. Plant Physiol. 149, 561–574.
Nishida, M., Yoshimitsu, H., Nohara, T., 2003. Three cycloartane glycosides from
Cimicifuga rhizome and their immunosuppressive activities in mouse allogeneic
mixed lymphocyte reaction. Chem. Pharm. Bull. 51, 354–356.
Reichelt, M., Brown, P.D., Schneider, B., Oldham, N.J., Stauber, E., Tokuhisa, J.,
Kliebenstein, D.J., Mitchell-Olds, T., Gershenzon, J., 2002. Benzoic acid
glucosinolate esters and other glucosinolates from Arabidopsis thaliana.
Phytochemistry 59, 663–671.
Shinoda, T., Nagao, T., Nakayama, M., Serizawa, H., Koshioka, M., Okabe, H., Kawai,
A., 2002. Identification of a triterpenoid saponin from a crucifer, Barbarea
vulgaris, as a feeding deterrent to the diamondback moth, Plutella xylostella. J.
Chem. Ecol. 28, 587–599.
Clark, D.B., 2010. Glucosinolates, structures and analysis in food. Anal. Methods 2,
3
10–325.
Courtney, S.P., 1981. Coevolution of Pierid Butterflies and their cruciferous host
plants III. Anthocharis cardamines (L.) survival, development and oviposition on
different host plants. Oecologia 51, 91–96.
Fahey, J.W., Zalcmann, A.T., Talalay, P., 2001. The chemical diversity and distribution
of glucosinolates and isothiocyanates among plants. Phytochemistry 56, 5–51.
Frechard, A., Fabre, N., Pean, C., Montaut, S., Fauvel., M.-T., Rollin, P. Fourasté, I.,
2
001. Novel indole-type glucosinolates from woad (Isatis tinctoria L.).
Smallegange, R.C., van Loon, J.J.A., Blatt, S.E., Harvey, J.A., Agerbirk, N., Dicke, M.,
2007. Flower vs. leaf feeding by Pieris brassicae: glucosinolate-rich flower
tissues are preferred and sustain higher growth rate. J. Chem. Ecol. 33, 1831–
1844.
Sprogø, K., Stærk, D., Jäger, A.K., ADsersen, A., Hansen, S.H., Witt, M., Landbo, A.-K.R.,
Meyer, A.S., Jaroszewski, J.W., 2007. Targeted natural product isolation guided
by HPLC-SPE-NMR: constituents of Hubertia species. J. Nat. Prod. 70, 1472–
1477.
Tetrahedron Lett. 42, 9015–9017.
Graser, G., Oldham, N.J., Brown, P.D., Temp, U., Gershenzon, J., 2001. The
biosynthesis of benzoic acid glucosinolate esters in Arabidopsis thaliana.
Phytochemistry 57, 23–32.
Halkier, B.A., Gershenzon, J., 2006. Biology and Biochemistry of glucosinolates.
Annu. Rev. Plant Biol. 57, 303–333.
Huseby, H., Artz, S., Burmeister, W.P., Härtel, F.V., Brandt, A., Rossiter, J.T., Bones,
A.M., 2005. Crystal structure at 1.1 Å resolution of an insect myrosinase from
Brevicoryne brassicae shows its close relationship to b-glucosidases. Insect
Biochem. Mol. Biol 35, 1311–1320.
Ibdah, M., Pichersky, E., 2009. Arabidopsis Chy1 null mutants are deficient in
benzoic acid-containing glucosinolates in the seeds. Plant Biol. 11, 574–581.
Iori, R., Rollin, P., Streicher, H., Thiem, J., Palmieri, S., 1996. The myrosinase-
glucosinolate interaction mechanism studied using some synthetic competitive
inhibitors. FEBS Lett. 385, 87–90.
Janowitz, T., Trompetter, I., Piotrowski, M., 2009. Evolution of nitrilases in
glucosinolate-containing plants. Phytochemistry 70, 1680–1686.
Kirkpatrick, J.B., Gilfedder, L., 1998. Conserving weedy natives: two Tasmanian
endangered herbs in the Brassicaceae. Aust. J. Ecol. 23, 466–473.
Kissen, R., Bones, A.M., 2009. Nitrile-specifier proteins involved in glucosinolate
hydrolysis in Arabidopsis thaliana. J. Biol. Chem. 284, 12057–12070.
Kliebenstein, D.J., D’Auria, J.C., Behere, A.S., Kim, J.H., Gunderson, K.L., Breen, J.N.,
Lee, G., Gershenzon, J., Last, R.L., Jander, G., 2007. Characterization of seed-
specific benzoyloxyglucosinolate mutations in Arabidopsis thaliana. Plant J. 51,
Sønderby, I.E., Geu-Flores, F., Halkier, B.A., 2010. Biosynthesis of glucosinolates–
gene discovery and beyond. Trends Plant Sci. 15, 283–290.
Sørensen, H., 1990. Glucosinolates: structure, properties, function. In: Shahidi, F.
(Ed.), Canola and rapeseed. Production, chemistry, nutrition and processing
technology. Van Nostrand Reinhold, New York, pp. 149–172.
Thies, W., 1988. Isolation of sinigrin and glucotropaeolin from cruciferous seeds. Fat
Sci. Technol. 90, 311–314.
Toneatto, F., Nielsen, J.K., Ørgaard, M., Hauser, T.P., 2010. Genetic and sexual
separation between insect resistant and susceptible Barbarea vulgaris plants in
Denmark. Mol. Ecol. 19, 3456–3465.
Van Leur, H., Aaijmakers, C.E., Van Dam, N.M., 2006. A heritable glucosinolate
polymorphism within natural populations of Barbarea vulgaris. Phytochemistry
67, 1214–1223.
Wathelet, J.-P., Iori, R., Leoni, O., Rollin, P., Quinsac, A., Palmieri, S., 2004. Guidelines
for glucosinolate analysis in green tissues used for biofumigation. Agroindustria
3, 257–266.
Wiklund, C., Friberg, M., 2009. The evolutionary ecology of generalization: among-
year variation in host plant use and offspring survival in a butterfly. Ecology 90,
3406–3417.
Windsor, A.J., Reichelt, M., Figuth, A., Svatoš, A., Kroymann, J., Kliebenstein, D.,
Gershenzon, J., Mitchell-Olds, T., 2005. Geographic and evolutionary
diversification of glucosinolates among near relatives of Arabidopsis thaliana
(Brassicaceae). Phytochemistry 66, 1321–1333.
Wittstock, U., Halkier, B.A., 2000. Cytochrome P450 CYP79A2 from Arabidopsis
thaliana L. catalyzes the conversion of L-phenylalanine to phenylacetaldoxime
in the biosynthesis of benzylglucosinolate. J. Biol. Chem. 275, 14659–14666.
Ørgaard, M., Linde-Laursen, I., 2008. Meiotic analysis of Danish species of Barbarea
(Brassicaceae) using FISH: chromosome numbers and rDNA sites. Hereditas
145, 215–219.
1
062–1076.
Kuzina, V., Nielsen, J.K., Augustin, J.M., Torp, A.M., Bak, S., Andersen, S.B., 2011.
Barbarea vulgaris linkage map and quantitative trait loci for saponins,
glucosinolates, hairiness and resistance to the herbivore Phyllotreta nemorum.
Phytochemistry 72, 188–198.
Linsheid, M., Wendisch, D., Strack, D., 1980. The structures of sinapic acid esters and
their metabolism in cotyledons of Raphanus sativus. Z. Naturforsch. 35c, 907–
9
14.
Mellon, F.A., Bennett, R.N., Holst, B., Williamson, G., 2002. Intact glucosinolate
analysis in plant extracts by programmed cone voltage electrospray LC/MS:
performance and comparison with LC/MS/Ms methods. Anal. Biochem. 306,
8
3–91.