Báo cáo khoa học: Comparative biochemical characterization of nitrile-forming proteins from plants and insects that alter myrosinase-catalysed hydrolysis of glucosinolatesk docx

Comparative biochemical characterization of
nitrile-forming proteins from plants and insects that alter
myrosinase-catalysed hydrolysis of glucosinolates
Meike Burow, Jana Markert, Jonathan Gershenzon and Ute Wittstock
Max Planck Institute for Chemical Ecology, Department of Biochemistry, Jena, Germany
Plants defend themselves against herbivore and patho-
gen attack using a diverse array of repellent or toxic
secondary metabolites [1,2]. Among these chemical
defences, the glucosinolates found in plants of the
order Capparales have been studied intensively as they
have significant effects on the taste, flavour, nutritional
value and pest resistance of crops belonging to the
Brassicaceae family, such as oilseed rape, cabbage and
broccoli [3,4]. Glucosinolates are amino acid-derived
thioglycosides with aliphatic, aromatic or indole side
chains (Fig. 1). The biological activities of glucosino-
late-containing plants are usually attributed to the
hydrolysis products formed from glucosinolates upon
tissue disruption by endogenous thioglucosidases
(known as myrosinases EC 3.2.3.1., Fig. 2) rather than
to the parent glucosinolates, which are spatially separ-
ated from myrosinases in the intact plant [5,6]. The
most common type of glucosinolate hydrolysis type of
Keywords
epithionitrile; epithiospecifier protein;
glucosinolate; nitrile; nitrile-specifier protein
Correspondence
U. Wittstock, Institut fu
¨
r Pharmazeutische
Biologie, Technische Universita

¨
t
Braunschweig, Mendelssohnstr. 1,
D-38106 Baunschweig, Germany
Fax: +49 531 391 8104
Tel: +49 531 391 5681
E-mail:
(Received 1 March 2006, accepted
30 March 2006)
doi:10.1111/j.1742-4658.2006.05252.x
The defensive function of the glucosinolate–myrosinase system in plants of
the order Capparales results from the formation of isothiocyanates when
glucosinolates are hydrolysed by myrosinases upon tissue damage. In some
glucosinolate-containing plant species, as well as in the insect herbivore
Pieris rapae, protein factors alter the outcome of myrosinase-catalysed glu-
cosinolate hydrolysis, leading to the formation of products other than
isothiocyanates. To date, two such proteins have been identified at the
molecular level, the epithiospecifier protein (ESP) from Arabidopsis thaliana
and the nitrile-specifier protein (NSP) from P. rapae. These proteins share
no sequence similarity although they both promote the formation of nit-
riles. To understand the biochemical bases of nitrile formation, we com-
pared some of the properties of these proteins using purified preparations.
We show that both proteins appear to be true enzymes rather than alloster-
ic cofactors of myrosinases, based on their substrate and product specif-
icities and the fact that the proportion of glucosinolates hydrolysed to
nitriles does not remain constant when myrosinase activity varies. No sta-
ble association between ESP and myrosinase could be demonstrated during
affinity chromatography, nevertheless some proximity of ESP to myrosin-
ase is required for epithionitrile formation to occur, as evidenced by the
lack of ESP activity when it was spatially separated from myrosinase in a

dialysis chamber. The significant difference in substrate- and product spe-
cificities between A. thaliana ESP and P. rapae NSP is consonant with their
different ecological functions. Furthermore, ESP and NSP differ remark-
ably in their requirements for metal ion cofactors. We found no indications
of the involvement of a free radical mechanism in epithionitrile formation
by ESP as suggested in earlier reports.
Abbreviations
BPDS, bathophenanthroline disulfonic acid; ESP, epithiospecifier protein; FID, flame ionization detection; NSP, nitrile-specifier protein.
2432 FEBS Journal 273 (2006) 2432–2446 ª 2006 The Authors Journal compilation ª 2006 FEBS
products, the isothiocyanates, has been shown to pos-
sess antimicrobial and insecticidal activities [7], and
have stimulated much interest as cancer-preventing
agents [8,9]. In addition to isothiocyanates, other
hydrolysis products such as epithionitriles and thiocya-
nates are formed in other species of the Brassicaceae
under the influence of certain protein factors [10–12].
For example, epithiospecifier proteins (ESPs) have
been identified in several species of Brassicaceae that
alter the outcome of glucosinolate hydrolysis without
having hydrolytic activity on glucosinolates themselves
[13–15].
Since the first description of ESP activity in plants
in 1973 [13], only a few studies have investigated its
biochemical properties, probably because of the diffi-
culty in isolating the active protein from plant mater-
ial. ESPs were originally described as 35–40 kDa
proteins that promote the formation of epithionitriles,
rather than isothiocyanates, from alkenylglucosinolates
upon myrosinase-catalysed glucosinolate hydrolysis
[12–14]. During nitrile formation from alkenylglucosin-

olates (Fig. 1), the sulfur released from the thioglycosi-
dic bond is captured by the terminal double bond in
the glucosinolate side chain to form a thiirane (episul-
fide) ring and an epithionitrile is formed [16] (Fig. 2).
However, the mechanism by which ESPs catalyse this
intramolecular sulfur transfer is not known. It has
been suggested that the mechanism is analogous to the
formation of epoxides catalysed by cytochrome P450-
dependent monooxygenases [17]. The first ESP gene
was isolated several years ago from the model plant
Arabidopsis thaliana ecotype Landsberg erecta (Ler)
(Brassicaceae) [15]. It encodes a 37-kDa protein (341
amino acids) with 45–55% amino acid sequence iden-
tity to several A. thaliana myrosinase-binding proteins.
In the presence of myrosinase, crude extracts of
Escherichia coli expressing recombinant A. thaliana
ESP catalysed the conversion of alkenylglucosinolates
to epithionitriles, as well as the conversion of nonalke-
nylglucosinolates to simple nitriles lacking the thiirane
ring [15] (Fig. 2).
More recently, we identified a functionally related
protein factor, designated a nitrile-specifier protein
(NSP), in the midgut of larvae of the cabbage white
butterfly, Pieris rapae [18]. P. rapae NSP cDNA
encodes a polypeptide of 632 amino acids with a
molecular mass of 73 kDa. Like plant ESPs, P. rapae
NSP does not directly hydrolyse glucosinolates, but
promotes the formation of nitriles rather than toxic
1
n

NC
S
SC
R
N
-
RCN
R
3
-
OSO
N
S
R
aglycone
2
3
-
SOO
clG
N
S
3
Fig. 2. Glucosinolate hydrolysis. Upon activation of the glucosinolate–myrosinase system, glucosinolates are hydrolysed by myrosinases
yielding unstable aglycones. These aglycones can then spontaneously undergo a Lossen rearrangement to form the corresponding isothio-
cyanates (1). In several species of the Brassicaceae, ESPs promote the formation of epithionitriles (2) from aliphatic glucosinolates with a
terminal double bond such as allylglucosinolate. ESP from A. thaliana, however, is capable of producing epithionitriles from alkenylglucosino-
lates as well as simple nitriles (3) from nonalkenyl substrates. Simple nitrile formation is also promoted by P. rapae NSP or the presence of
Fe
2+

. Depending on the nature of the glucosinolate side chain, other hydrolysis products such as thiocyanates and oxazolidine-2-thiones can
also be formed. R indicates the variable side chain (aliphatic, aromatic or indole) of the parent glucosinolate.
S
N
Glc
OSO
3
-
SS
N
Glc
OSO
3
-
S
N
Glc
OSO
3
-
S
N
Glc
OSO
3
-
S
O
21
43

Fig. 1. Chemical structures of glucosinolates used in this study.
1, Allylglucosinolate; 2, 4-methylthiobutylglucosinolate; 3, benzylglu-
cosinolate; 4, 4-methylsulfinylbutylglucosinolate).
M. Burow et al. Nitrile-forming proteins from plants and insects
FEBS Journal 273 (2006) 2432–2446 ª 2006 The Authors Journal compilation ª 2006 FEBS 2433
isothiocyanates upon glucosinolate hydrolysis catalysed
by the myrosinases ingested with the plant tissue. The
nitriles are excreted with the faeces [18]. Although nit-
rile formation in both plants and their insect herbiv-
ores is accomplished through the action of protein
factors, the plant ESP and insect NSP do not show
any significant amino acid sequence similarity [18].
Thus, it is an open question if ESP and NSP share any
biochemical properties and whether nitrile formation
mediated by these two proteins occurs via the same
mechanism.
Because ESP and NSP activities have been detected
only in association with myrosinase, it is not yet
known if these proteins act as cofactors of myrosin-
ase or possess catalytic activity of their own. In the
case of ESP, it has been suggested that this protein
interacts with myrosinases in an allosteric manner
leading to conformational changes in the myrosinase
active site that modify the proportions of hydrolysis
products formed [14]. Alternatively, both ESP and
NSP may possess catalytic activity and control the
outcome of glucosinolate hydrolysis by converting the
unstable aglycone intermediate released by myrosin-
ase, which typically rearranges to an isothiocyanate,
to a nitrile product instead. In either case, it can be

assumed that ESP and NSP have to be closely associ-
ated with myrosinases, either to bind them as cofac-
tors or to bind the unstable aglycone before it
rearranges.
The isolation of two different types of nitrile-form-
ing proteins, ESP from A. thaliana and NSP from
P. rapae, presented an opportunity to learn more
about the biochemical requirements for nitrile forma-
tion. Here, we compare some characteristics of the
purified proteins to investigate their role in nitrile for-
mation. We found striking differences between ESP
and NSP which are in agreement with their proposed
biological functions, but both appear to act as enzymes
rather than myrosinase cofactors. Our data do not
support the mechanism for epithionitrile formation
suggested in an earlier report.
Results
Development of an enzyme assay to measure
ESP-dependent nitrile formation
ESPs have been shown to redirect the hydrolysis of
glucosinolates catalysed by myrosinases from isothio-
cyanate formation towards formation of the corres-
ponding nitriles [10,13,15,17,19]. The ability of ESPs
to promote nitrile formation in the presence of myros-
inase has been referred to as ESP activity even though
it is not yet certain if ESPs are cofactors rather than
true enzymes. Their activity has been detected only in
conjunction with myrosinase. If ESPs are catalysts,
they can be assumed to convert the aglycones pro-
duced by myrosinase to nitrile products. The instability

of these aglycones precludes rigorous kinetic studies,
and no other natural or synthetic compound has been
reported to serve as a substrate for ESP.
With these limitations in mind, optimal assay con-
ditions for the biochemical characterization of the
purified, recombinant A. thaliana ESP were sought. A
purified preparation from Sinapis alba was used as
the myrosinase source, and a number of buffers and
pH conditions were surveyed. Surprisingly, ESP was
completely inactive in citrate and phosphate buffers,
but promoted nitrile formation in many biological
buffers as measured by the production of epithiopro-
pyl cyanide [2-(thiirane-2-yl)acetonitrile] from allyl
glucosinolate (Figs 1,2). The highest activity was
found in Mes buffer (50 mm) at pH 6.0. Fe
2+
has
been previously reported to be an essential cofactor
for ESP from Brassica napus [17] and Crambe abyssi-
nica [13]. This was also true for our recombinant
A. thaliana protein, with the optimal concentration
found to be 0.5 mm. The temperature for standard
assays was chosen to be 20 °C. Higher temperatures
accelerated the hydrolysis of allyl glucosinolate, but
the proportion of epithiopropyl cyanide declined rel-
ative to the amounts of other hydrolysis products
(Table 1).
Irrespective of whether ESP is a protein cofactor or
a true enzyme, the molar ratio of ESP to myrosinase
can be assumed to be crucial for the proportion of nit-

riles produced upon glucosinolate hydrolysis. Under
the standard conditions used for ESP assays, a 260-
fold molar excess of ESP was employed relative to the
quantity of myrosinase protein, which is a dimer.
Lower molar ratios of ESP:myrosinase resulted in
decreased absolute and relative nitrile formation, as
measured by the production of epithionitrile from
allylglucosinolate. When higher molar ratios of ESP to
myrosinase were used, myrosinase activity was found
to be strongly reduced which reduced net nitrile forma-
tion. With a 260-fold molar excess of ESP:myrosinase,
nitrile products and isothiocyanates each accounted for
 50% of the hydrolysis products formed. The produc-
tion of considerable amounts of isothiocyanate indi-
cates that ESP (assuming it acts as an enzyme) is
saturated with its substrate. The concentration of
glucosinolates used in the assays (1–3 mm) was satur-
ating with respect to myrosinase activity. The same
considerations were used in developing an assay for
NSP activity.
Nitrile-forming proteins from plants and insects M. Burow et al.
2434 FEBS Journal 273 (2006) 2432–2446 ª 2006 The Authors Journal compilation ª 2006 FEBS
Physical proximity of myrosinase and ESP is
essential for epithionitrile formation, but a stable
interaction could not be detected
Examination of the predicted structure of the A. thali-
ana ESP showed that most of the protein was made up
of a series of b sheets known as Kelch motifs [20]
(Fig. 9), as predicted by the program interproscan
( European Bioin-

formatics Institute, Hinxton, Cambridge, UK). These
features are known to mediate protein–protein interac-
tions providing some support for the hypothesis that
ESP functions as an allosteric cofactor of myrosinase.
In this case, a very close physical interaction between
the two proteins would be expected. To assess if an
association of ESP and myrosinase is essential for
epithionitrile formation, we spatially separated the two
proteins by placing ESP into a dialysis cassette with
myrosinase in the external buffer. Under these condi-
tions, epithionitrile formation from allylglucosinolate
was completely abolished, with allyl isothiocyanate
being the only hydrolysis product formed (data not
shown). These results suggested that ESP and myrosin-
ase must have some proximity for nitrile formation to
occur. To test for a stable association of ESP and my-
rosinase, crude extracts of A. thaliana (Col-0) leaves
were added to an amino-link agarose resin coupled
with an antibody to ESP that had been presaturated
with ESP. However, myrosinase activity was found
only in the wash fractions containing unbound compo-
nents of the plant extract. When ESP was eluted from
the resin, no myrosinase activity was detectable in the
eluates (Fig. 10). Furthermore, the recombinant ESP
did not comigrate with the S. alba myrosinase in
native PAGE gels (data not shown).
ESP and NSP: enzymes or myrosinase cofactors?
Another approach employed to investigate the role of
ESP and NSP in nitrile formation was to manipulate
the activity of myrosinase and observe its effect on nit-

rile formation. If ESP interacts with myrosinase as an
allosteric cofactor, manipulation of myrosinase in the
presence of ESP should only affect the total amount of
hydrolysis products formed without altering the ratio
between nitrile and isothiocyanate. We stimulated
myrosinase by addition of low concentrations of
l-ascorbate [21–23]. l-Ascorbate facilitates release of
the glucose moiety from the active site of myrosinases,
the rate-limiting step in glucosinolate hydrolysis
[21]. The addition of 0.05–2 mml-ascorbate to ESP
assays increased the formation of total hydrolysis prod-
ucts, as described previously, but decreased the amount
of epithionitrile formed from allylglucosinolate relative
to the amount of isothiocyanate (Fig. 3A). A parallel
series of assays carried out with the P. rapae NSP gave
similar results. The increase in total hydrolysis products
formed from benzylglucosinolate with l-ascorbate addi-
tion was accompanied by a substantial decrease in the
ratio of nitrile to isothiocyanate (Fig. 3B).
ESP and NSP have different substrate and
product specificities
To compare the properties of A. thaliana ESP, purified
after heterologous expression in E. coli, and P. rapae
NSP, purified from larval midguts, we examined their
effects on the myrosinase-catalysed hydrolysis of differ-
ent aliphatic glucosinolates and the aromatic benzyl-
glucosinolate (Fig. 4, Table 2). When allylglucosinolate
was incubated with myrosinase only, allyl isothiocya-
nate was the major hydrolysis product. ESP redirected
the hydrolysis of allylglucosinolate towards the forma-

tion of the corresponding epithionitrile, epithiopropyl
cyanide [2-(thiirane-2-yl)acetonitrile] with minor
amounts of the simple nitrile, allyl cyanide (but-3-ene
nitrile). Both diastereomers of epithiopropyl cyanide
Table 1. Biochemical characteristics of ESP. The effects of variable
temperature, phosphate ions, radical scavengers, and reducing
agents were tested on ESP-catalysed formation of the epithionitrile
from allylglucosinolate, epithiopropyl cyanide, in ESP assays per-
formed under standard conditions as described in Experimental pro-
cedures.
Test parameter Effect on ESP
Temperature Higher absolute and relative nitrile formation
at low temperatures (0–20 °C) despite
reduced myrosinase activity
Phosphate Complete inhibition with 5 m
M phosphate
(K
i
¼ 1mM); activity was restored by
ferrous ions
Sulfate Acts as myrosinase inhibitor; addition of
sulfate (0–100 m
M concentration tested)
resulted in reduced absolute but constant
relative activity
Sorbitol No effect (0–100 m
M tested)
L-Cysteine No effect, but 5 mML-cysteine led to a
decrease in myrosinase activity
Superoxide

dismutase
No effect (5 lgÆassay
)1
)
Argon Flushing of assay mixtures with argon
had no effect
Dithiothreitol Addition 0.1 m
M dithiothreitol resulted in
a threefold increase in activity
(only 0.5-fold increase in presence of
0.5 m
M ferrous ions)
b-Mercaptoethanol 7 m
M b-mercaptoethanol resulted in
a 20% increase in activity
M. Burow et al. Nitrile-forming proteins from plants and insects
FEBS Journal 273 (2006) 2432–2446 ª 2006 The Authors Journal compilation ª 2006 FEBS 2435
were detected by GC-MS using a chiral column for
separation (data not shown). The proportion of these
diastereomers was  1:1, suggesting that sulfur capture
by the terminal double bond is equally probable from
above or below. When NSP was added to the assay,
hydrolysis was redirected only to the simple nitrile,
and no epithionitrile was formed (Fig. 4).
The percentage of nitriles (relative to the total
amount of hydrolysis products) formed from different
glucosinolates in the presence of ESP varied depending
on the glucosinolate side chain. ESP catalysed the
formation of substantial amounts of nitriles from allyl-
glucosinolate, 4-methylthiobutylglucosinolate and,

to a lesser extent, 4-methylsulfinylbutylglucosinolate.
However, almost no nitrile was produced from benzyl-
glucosinolate under the same conditions. The epithio-
nitrile was the predominant nitrile formed from
allylglucosinolate with small amounts of the simple nit-
rile detected. Only simple nitriles were formed from
the remaining glucosinolates. In contrast, NSP
promoted the formation of simple nitriles from allyl-,
4-methylthiobutyl-, 4-methylsulfinylbutyl- and benzyl-
glucosinolate with comparable efficiencies. No epithio-
nitrile was seen to be formed from allylglucosinolate.
In the absence of ESP or NSP, all four glucosinolates
were hydrolysed almost exclusively to the correspond-
ing isothiocyanates (Table 2, third row).
Fe
2+
promotes nitrile formation from different
glucosinolates in the presence of ESP and NSP
Fe
2+
has previously been reported to be essential for
the catalytic activity of ESP from Brassica napus [17]
and Crambe abyssinica [13]. Thus the effects of Fe
2+
on A. thaliana ESP and P. rapae NSP were investigated
using different glucosinolates as substrates (Table 2).
The addition of Fe
2+
to ESP assays increased the pro-
portion of epithionitrile formed from allylglucosinolate

from 27.0% (without addition of Fe
2+
) to 87.2%
(0.01 mm Fe
2+
) and 93.9% (0.5 mm Fe
2+
) of the total
amount of hydrolysis products. No epithionitrile was
formed by myrosinase in the presence of Fe
2+
without
addition of ESP, indicating that ferrous ions and
myrosinase by themselves are not sufficient for the for-
mation of the thiirane ring. In contrast, Fe
2+
promoted
the formation of simple nitriles from the tested gluco-
Fig. 3. Predicted protein structure of
A. thaliana Ler ESP. Amino acid sequence
of A. thaliana ESP (GenBank accession num-
ber AAL14623) and Kelch motif repeats as
predicted by
INTERPROSCAN. Amino acids of
the Kelch repeats are printed in bold, and
the repeats are indicated by arrows. Each
Kelch repeat forms a b sheet (b
1)5
).
A

kDa
W1 W2L1 L2 E1 E2 E3
52
35
0
20
40
60
80
100
myrosinase activity
[% recovered]
B
W2L2 E1 E2 E3
Fig. 4. Test for physical association between ESP and myrosinase.
A chromatography resin coupled with an anti-ESP serum was satur-
ated with ESP before it was loaded with a crude protein extract
from rosette leaves of A. thaliana Col-0 containing myrosinase.
After a washing step, ESP and bound proteins were eluted. Load-
ing, washes and elution were monitored by western blot analysis
with an anti-ESP serum (A) and by myrosinase assays (B). (L1,
flow-through after binding of ESP; W1, wash step after binding of
ESP; L2, flow-through after loading of plant extract containing
myrosinase; W2, wash step after loading of plant extract; E1–E3
eluted fractions of ESP and bound proteins)
Nitrile-forming proteins from plants and insects M. Burow et al.
2436 FEBS Journal 273 (2006) 2432–2446 ª 2006 The Authors Journal compilation ª 2006 FEBS
sinolates in the presence of myrosinase even without
ESP. The addition of ESP further increased the propor-
tion of simple nitriles formed from 4-methylthiobutyl-,

4-methylsulfinylbutyl- and benzylglucosinolate. In the
presence of 0.01 mm Fe
2+
, ESP caused an increase
in the formation of simple nitriles from 4-methyl-
sulfinylbutyl- and to a lesser extent from benzylglucosi-
nolate compared with assays without ESP. In the
presence of 0.5 mm Fe
2+
, ESP addition led to nearly
100% simple nitrile formation for all tested nonalke-
nylglucosinolates. For NSP, addition of 0.01 mm Fe
2+
resulted in an increase in simple nitrile formation from
allyl- and benzylglucosinolate compared with assays
without added Fe
2+
, but did not result in the forma-
tion of the epithionitrile from allylglucosinolate.
Addition of Fe
3+
increases ESP activity but not
NSP activity
The activity of ESP measured in crude seed extracts of
Lepidium sativum has been shown to increase in the
presence of both Fe
2+
and Fe
3+
[24], whereas no

effect of Fe
2+
on the activity of ESP in Crambe abyssi-
nica seeds was found [25]. To investigate whether or
not the impact of iron on the activity of the A. thali-
ana ESP is restricted to Fe
2+
, assays were carried out
with both Fe
2+
or Fe
3+
(0.5 mm). Allylglucosinolate
was chosen as a substrate, because formation of the
corresponding epithionitrile is strictly dependent on
ESP and is therefore a better measure of ESP activity
than simple nitrile formation. Saturation of ESP with
its substrate under standard assay conditions was
insured by reducing the amounts of ESP (0.3 units),
such that the isothiocyanate was formed in consider-
able amounts even under conditions of elevated ESP
activity. In this series of assays, the formation of epith-
ionitrile accounted for a maximum of 85% of the
amount of total hydrolysis products. The results
showed that addition of Fe
2+
in various forms resulted
in a 16-fold increase in ESP activity, while a sevenfold
increase was observed for addition of Fe
3+

(Fig. 5A).
To study the effects of Fe
2+
and Fe
3+
on nitrile for-
mation by NSP, assays were carried out using benzyl-
glucosinolate as substrate and adding iron salts to a
final concentration of only 0.01 mm to avoid a high
background of NSP-independent nitrile formation. As
described previously, the conversion of benzylglucosi-
nolate to its corresponding simple nitrile by NSP was
slightly increased by 0.01 mm Fe
2+
compared with no
iron salt addition, but no changes in NSP activity were
measured in the presence of Fe
3+
(Fig. 5B). Without
the addition of ESP or NSP, the enhanced formation
of nitriles from myrosinase-catalysed hydrolysis of
both allyl- and benzylglucosinolate was observed only
upon addition of Fe
2+
, whereas Fe
3+
did not change
the proportion of nitriles produced by this reaction.
The anions of the iron salts used did not have any
effects on ESP, NSP or myrosinase.

ESP and NSP differ in their requirements
for iron species
To study the role of iron in the catalytic mechanisms
of ESP and NSP in more detail, different chelators
Table 2. Nitrile and epithionitrile formation from different glucosinolates in the presence of myrosinase, ESP and NSP in vitro. Assays were
performed in 50 m
M Mes, pH 6.0, containing 1 mM glucosinolate as substrate, with one exception: 3 mM 4-methylsulfinylbutylglucosinolate
was used in ESP assays in order to compensate for the low activity of the Sinapis alba myrosinase with this substrate. Ferrous ions were
added as (NH
4
)
2
[Fe(SO
4
)
2
] to minimize oxidation. Dichloromethane extracts of the assays were analysed by GC-MS and GC-FID. The
amounts of products are expressed as a percentage of the total amount of hydrolysis products measured in nmol. In each case, the remain-
der of the hydrolysis products were isothiocyanates. Data are the mean ± SD of results obtained in at least three independent experiments.
4-mtb-, 4-methylthiobutyl-; 4-msob-, 4-methylsulfinylbutyl-; CN, cyanide; myr, myrosinase; nd, not determined.
Glucosinolate Hydrolysis product allyl- epithio-propyl-CN allyl-CN 4-mtb- 4-mtb-CN 4-msob- 4-msob-CN benzyl- benzyl-CN
No Fe
2+
added
myr + ESP 27.0 (± 6.6) 1.5 (± 1.8) 45.4 (± 5.7) 14.7 (± 4.2) 1.6 (± 1.2)
myr + NSP 0 (± 0) 48.5 (± 8.6) 64.4 (± 10.4) 45.4 (± 19.2) 48.3 (± 1.6)
myr 0 (± 0) 0.3 (± 0.6) 3.4 (± 4.3) 0 (± 0) 0.1 (± 0.2)
0.01 m
M Fe
2+

myr + ESP 87.2 (± 11.5) 2.1 (± 1.4) nd 86.7 (± 0.5) 18.3 (± 0.9)
myr + NSP 0 (± 0) 74.2 (± 5.9) 77.7 (± 10.1) 55.3 (± 27.2) 73.6 (± 9.4)
myr 0 (± 0) 19.4 (± 5.2) 40.7 (± 13.6) 24.5 (± 10.3) 14.5 (± 4.4)
0.5 m
M Fe
2+
myr + ESP 93.9 (± 2.1) 4.9 (± 1.6) 98.0 (± 2.5) 97.9 (± 2.6) 95.2 (± 3.5)
myr + NSP nd nd nd nd 92.1 (± 2.6)
myr 0 (± 0) 75.1 (± 5.4) 82.7 (± 1.5) 83.8 (± 1.8) 78.5 (± 7.6)
M. Burow et al. Nitrile-forming proteins from plants and insects
FEBS Journal 273 (2006) 2432–2446 ª 2006 The Authors Journal compilation ª 2006 FEBS 2437
were added to the enzyme assays. ESP activity was
strongly reduced in the presence of 5 mm bathophen-
anthroline disulfonic acid (BPDS), a chelator of fer-
rous ions, and by 5 mm deferoxamine, a chelator of
ferric ions (Fig. 6A). The addition of 2 mm EDTA,
known to chelate Fe
2+
and Fe
3+
, as well as numerous
other cations, was sufficient to completely inhibit ESP-
dependent epithionitrile formation in favour of the
generation of isothiocyanate (Fig. 7). In assays per-
formed using 5 mm EDTA, ESP activity was restored
by addition of 0.5 mm Fe
2+
or 0.5 mm Fe
3+
(Fig. 8A), whereas Ca

2+
,Mg
2+
and K
+
had no effect.
However, when Ca
2+
or Mg
2+
was added to a final
concentration of 5 mm, the epithionitrile of allylglucos-
inolate was formed (data not shown).
The activity of NSP, measured by nitrile formation
from benzylglucosinolate, was changed only by chela-
tors that complex divalent cations. Addition of EDTA
and BPDS reduced the proportion of nitrile formed
from 36.3 to 16.9 and 15.0%, respectively (Fig. 6B).
In contrast, deferoxamine did not affect NSP activity,
A
0 0.05 0.5 2
0
20
40
60
80
100
120
140
160

L-ascorbate [mM]
products [nmol]
B
0 0.05 0.5 2
0
20
40
60
80
100
120
140
160
products [nmol]
L-ascorbate [mM]
Fig. 5. Effects of L-ascorbate on ESP and NSP. (A) ESP assays
were carried out in 50 m
M Mes, pH 6.0, containing 1 mM allylglu-
cosinolate, 0.3 units ESP and 2 units myrosinase (grey, epithiopro-
pyl cyanide; white, allyl isothiocyanate). (B) NSP assays were
performed under the same conditions using benzylglucosinolate,
1 unit NSP and 1 unit myrosinase (grey, benzyl cyanide; white, ben-
zyl isothiocyanate). Data are the mean ± SD of three independent
experiments.
CN
A
12 3
S
CN
S

NCS
CN
3
68
10 12
0
200
400
IS
2
1
B
abundance
IS
2
1
400
200
0
12
8
10
6
abundance
C
IS
2
400
200
0

126810
abundance
D
retention time [min]
Fig. 6. The effects of ESP and NSP on myrosinase-catalysed hydro-
lysis of allylglucosinolate. (A) Chemical structures of the hydrolysis
products of allylglucosinolate: 1, allyl cyanide; 2, allyl isothiocyanate;
3, epithiopropyl cyanide (two diastereomers). (B–D) Allylglucosino-
late was incubated with ESP and myrosinase (B), NSP and myrosin-
ase (C), or only myrosinase (D). Assays were performed in 50 m
M
Mes, pH 6.0, containing 1 mM allylglucosinolate. Shown are GC-FID
chromatograms of dichloromethane extracts (IS, internal standard).
Compound 3 was present as a 1:1 mixture of two diastereomers
as shown, which were separated only by gas chromatography
using a chiral column (data not shown).
Nitrile-forming proteins from plants and insects M. Burow et al.
2438 FEBS Journal 273 (2006) 2432–2446 ª 2006 The Authors Journal compilation ª 2006 FEBS
indicating that Fe
3+
does not play a major role in
NSP-catalysed nitrile formation. The reduction in NSP
activity observed in the presence of 5 mm EDTA
was compensated for by addition of 0.5 mm Fe
2+
(Fig. 8B). However, a slight increase in nitrile forma-
tion by NSP was also seen upon the addition of Fe
3+
(0.5 mm) in the presence of EDTA (5 mm). This result
can be explained by displacement of small amounts of

ferrous ions from EDTA-binding sites due to the smal-
ler equilibrium dissociation constant of the Fe
3
±
EDTA complex.
Further biochemical characterization of ESP
provides no support for any established
reaction mechanism
It has been suggested that ESP-catalysed epithionitrile
formation is similar to Fe-dependent epoxidations
mediated by cytochrome P450-dependent monooxygen-
ases [17], which are oxygen-requiring catalysts that
employ a free-radical mechanism. To test this proposi-
tion, reactions were carried out in the absence of oxy-
gen or with radical scavenging agents. However, no
changes in ESP activity were measured in assays per-
formed with oxygen excluded using an argon purge
(Table 1). Addition of sorbitol, l-cysteine or superox-
ide dismutase as radical scavengers also did not affect
the absolute or relative amounts of epithionitrile
formed by ESP.
To investigate whether sulfhydryl groups play a role
in the mechanism of epithionitrile formation, ESP
assays were carried out in the presence of dithiothreitol
and b-mercaptoethanol. Addition of dithiothreitol to a
final concentration of 0.1 mm resulted in a threefold
increase in the formation of epithionitrile from allyl-
glucosinolate. However, in the presence of 7 mm
b-mercaptoethanol, only a 20% increase in ESP activ-
ity was detected.

Discussion
The outcome of myrosinase-catalysed glucosinolate
hydrolysis can be altered by plant proteins that have
no hydrolytic activity on glucosinolates themselves.
Several species of the Brassicaceae including A. thali-
ana have been shown to produce epithionitriles
and simple nitriles upon tissue damage instead of
isothiocyanates due to the presence of an ESP
[12,13,15,16]. In this study, we characterized the ESP
from A. thaliana, ecotype Ler, in order to learn more
about the mechanism by which this protein controls
the outcome of glucosinolate hydrolysis. We compared
the biochemical properties of the A. thaliana ESP with
those of the functionally related NSP from the midgut
A
no Fe
added
0
20
40
60
80
100
120
140
40
35
30
25
20

15
10
0
5
FeCl
3
Fe
2
(SO
4
)
3
FeCl
2
(NH
4
)
2
[Fe(SO
4
)
2
] FeSO
4
FeCl
3
Fe
2
(SO
4

)
3
FeCl
2
(NH
4
)
2
[Fe(SO
4
)
2
] FeSO
4
no Fe
added
B
products [nmol]
products [nmol]
Fig. 7. Effects of iron salts on ESP and
NSP. (A) ESP activity was measured in
50 m
M Mes, pH 6.0, containing 0.3 units
ESP, 1 m
M allylglucosinolate and 2 units
myrosinase. Salts were added to a final con-
centration of 0.5 m
M. For each salt, the left
bar represents the hydrolysis products
formed in the presence of ESP (dark green:

epithiopropyl cyanide; brown: allyl cyanide;
light green: allyl isothiocyanate), while the
right bar represents the products formed by
myrosinase in the absence of ESP (dark
grey, allyl cyanide; light grey, allyl isothiocya-
nate). (B) NSP assays were carried out
under the same conditions using 1 unit
NSP, 1 unit myrosinase and 1 m
M benzylglu-
cosinolate as substrate. Salts were added to
a final concentration of 0.01 m
M. For each
salt, the left bar represents the hydrolysis
products formed in the presence of NSP
(brown: benzyl cyanide; light green: benzyl
isothiocyanate), while the right bar repre-
sents the products formed by myrosinase in
the absence of NSP (dark grey, benzyl cyan-
ide; light grey, benzyl isothiocyanate). Data
are the mean ± SD of results obtained in
three independent experiments.
M. Burow et al. Nitrile-forming proteins from plants and insects
FEBS Journal 273 (2006) 2432–2446 ª 2006 The Authors Journal compilation ª 2006 FEBS 2439
of larvae of P. rapae, a protein factor that redirects
the myrosinase-catalysed hydrolysis of glucosinolates
in ingested plant material towards the formation of
nitriles instead of toxic isothiocyanates [18]. Studies
were carried out using purified recombinant A. thaliana
ESP, P. rapae NSP purified from larval midgut
extracts, and a pure preparation of myrosinase isolated

from seeds of S. alba. The use of purified enzymes
enabled us to perform all enzyme assays under strictly
defined conditions. It has been suggested that plant
ESPs are not enzymes, but rather allosteric protein co-
factors that bind to myrosinases and change their
product specificities thereby promoting the formation
of epithionitriles and simple nitriles upon glucosinolate
hydrolysis [14]. This proposal is supported by the lack
of stereospecificity in epithionitrile formation seen here
and in previous studies [26]. In addition, the predicted
structure of the A. thaliana ESP protein contains Kelch
motifs (Fig. 9) that are known to mediate protein–
protein interactions. However, our results did not con-
firm a stable interaction between ESP and myrosinase
during chromatographic separation (Fig. 10). In addi-
tion, the A. thaliana ESP migrated separately from
myrosinase on native PAGE gels (data not shown).
These findings are in agreement with earlier reports on
the separation of ESP and myrosinase from Crambe
abyssinica and Brassica napus by gel filtration chroma-
tography [13,19]. Nevertheless, some proximity of ESP
to myrosinase seems to be required for epithionitrile
formation because no ESP activity was detected when
the two proteins were spatially separated by a dialysis
membrane.
To investigate further whether ESP acts as cofactor
of myrosinase or a separate enzyme, the ratio of epith-
ionitrile to isothiocyanate formed from allylglucosino-
late in the presence of the A. thaliana ESP or the
P. rapae NSP was monitored as myrosinase activity

was increased by the addition of l-ascorbate [21–23].
For both ESP and NSP, the ratios of hydrolysis prod-
ucts changed markedly with an increase in myrosinase
activity (Fig. 3). These results do not support a role
for NSP and ESP as myrosinase cofactors, but are
more consistent with a catalytic role for these nitrile-
forming proteins in which the unstable product of the
myrosinase reaction serves as a direct substrate for
nitrile formation. In this interpretation, ESP and NSP
become quickly saturated as the activity of myrosinase
is increased upon ascorbate addition, and the excess
myrosinase product then rearranges to form additional
isothiocyanate. A catalytic function for ESP and NSP
A
Def.BPDSEDTA
no
addition
relative abundance
of epithiopropyl-CN [%]
50
40
30
20
10
0
relative abundance
of phenylacetonitrile [%]
B
50
40

30
20
10
0
no
addition
EDTA
BPDS
Def.
Fig. 8. Effects of metal ion chelators on ESP and NSP. (A) ESP
activity was assayed in 50 m
M Mes, pH 6.0, containing 1 mM allyl-
glucosinolate, 3 units ESP and 4 units myrosinase. (B) NSP assays
were performed under the same conditions using 1 unit NSP and
1 unit myrosinase. Chelators were added to a final concentration of
5m
M. Activities of ESP and NSP are expressed as the amount of
epithiopropyl cyanide (ESP) and benzyl cyanide (NSP) as a percent-
age of the total amount of hydrolysis products measured in nmol.
Data are the mean ± SD of results obtained in three independent
experiments. BPDS, bathaphenanthroline disulfonic acid; Def., defe-
roxamine; -CN, cyanide.
0
50
100
150
200
250
products [nmol]
1002468

EDTA [m
M]
Fig. 9. ESP activity in the presence of EDTA. Assays were per-
formed under standard ESP assay conditions using three units ESP
and allylglucosinolate as a substrate. All products formed are given
in nmol (m, epithiopropyl cyanide; h, allyl isothiocyanate; s, allyl
cyanide). All data are the mean ± SD from results obtained in three
independent experiments.
Nitrile-forming proteins from plants and insects M. Burow et al.
2440 FEBS Journal 273 (2006) 2432–2446 ª 2006 The Authors Journal compilation ª 2006 FEBS
is also supported by the pronounced substrate specifici-
ty of ESP (Table 2) and the increased ESP activity
observed in conjunction with reduced myrosinase
activity at lower temperatures (Table 1).
Because the putative substrates of ESP and NSP
are highly unstable, there is currently no way of
assaying ESP and NSP activity other than in com-
bined assays in which the substrates for ESP and
NSP are delivered by myrosinase through hydrolysis
of the corresponding glucosinolates. Although this
precludes rigorous determination of their kinetic
parameters, we compared a number of properties of
ESP and NSP and found some fundamental differ-
ences between these two nitrile-forming proteins.
Under natural conditions, both ESP and NSP
encounter a complex mixture of glucosinolates with
variable side chains. Therefore, we investigated the
influence of ESP and NSP on the myrosinase-cata-
lysed hydrolysis of different parent glucosinolates.
Glucosinolates without alkene function in their side

chains were converted to simple nitriles by both pro-
teins. However, when allylglucosinolate was used as a
substrate, ESP promoted the formation of the corres-
ponding epithionitrile, but the simple nitrile, allyl
cyanide (¼ but-3-ene nitrile) was not formed in signi-
ficant amounts with ESP under any assay conditions
tested. In the presence of NSP, only the simple nitri-
le, and not the epithionitrile, was formed from al-
lylglucosinolate (Fig. 4). This major difference in
product spectrum between ESP and NSP may reflect
a fundamental difference in their reaction mecha-
nisms.
Among different ecotypes of A. thaliana, the pres-
ence of a functional ESP was found to coincide with
the accumulation of alkenylglucosinolates [15]. ESPs
might therefore have a special role in plants in the
formation of epithionitriles, a class of glucosinolate
hydrolysis products that could be an effective defence
against insect herbivores due to the reactive thiirane
ring. When ESP assays were carried out using the
nonalkenyl aliphatic glucosinolates, 4-methylthiobutyl-
and 4-methylsulfinylbutylglucosinolate, the corres-
ponding simple nitriles were formed. However, the
aromatic benzylglucosinolate was mainly hydrolysed
to benzyl isothiocyanate under the same conditions
(Table 2). This distinct substrate specificity of
A. thaliana ESP suggests that the function of this pro-
tein lies in the formation of specific nitrile hydrolysis
products rather than an overall decrease in isothio-
cyanate formation. In contrast, the function of NSP

from P. rapae may be to prevent the general forma-
tion of isothiocyanates, which have been shown to
reduce the survival and the growth of these insect
herbivores [27]. In vitro, P. rapae NSP promotes nitri-
le formation from all glucosinolates tested with com-
parable efficiencies (Table 2) indicating that the
hydrolysis of any glucosinolate present in the larval
host plants can probably be redirected towards nitrile
formation.
To compare the mechanism of epithionitrile and
simple nitrile formation by the plant ESP with the
mechanism of simple nitrile formation catalysed by the
insect NSP, we studied some of the biochemical prop-
erties of these two proteins. Plant ESPs from differ-
ent sources have previously been described as
relative abundance
of epithiopropyl cyanide [%]
A
relative abundance
of phenylacetonitrile [%]
B
45
40
35
30
25
20
15
10
5

0
no EDTA
no
metal ion
Fe
2+
Fe
3+
Ca
2+
Mg
2+
K
+
no EDTA
no
metal ion
Fe
2+
Fe
3+
Ca
2+
Mg
2+
K
+
45
40
35

30
25
20
15
10
5
0
Fig. 10. Effects of metal ions on ESP and NSP in the presence of
EDTA. (A) ESP assays were carried out in 50 m
M Mes, pH 6.0,
containing 1 m
M allylglucosinolate, 1 unit ESP and 4 units myrosin-
ase. (B) The activity of 1 unit NSP was measured as nitrile form-
ation from benzylglucosinolate in the presence of 1 unit myrosinase.
EDTA was used at a concentration of 5 m
M. Salts of metal ions
[(NH
4
)
2
Fe(SO
4
)
2
,NH
4
Fe(SO
4
)
2

CaCl
2
, MgCl
2
, KCl] were added to a
final concentration of 0.5 m
M. Activities of ESP and NSP are
expressed as the amount of nitrile formed as a percentage of the
total amount of hydrolysis products measured in nmol. Data are the
mean ± SD from three independent experiments.
M. Burow et al. Nitrile-forming proteins from plants and insects
FEBS Journal 273 (2006) 2432–2446 ª 2006 The Authors Journal compilation ª 2006 FEBS 2441
iron-dependent proteins [12]. The mechanism of epith-
ionitrile formation by ESP has been suggested to be
analogous to that of cytochrome P450-mediated epoxi-
dations and to require Fe
2+
⁄ Fe
3+
for intramolecular
sulfur transfer [17]. Epithionitrile formation from allyl-
glucosinolate and simple nitrile formation from non-
alkenyl aliphatic glucosinolates by A. thaliana ESP
was strongly increased by the addition of Fe
2+
at low
concentrations (0.01 mm) [Table 2, Fig. 5A]. The high-
est activity was reached at concentrations of 0.2–
0.5 mm Fe
2+

, in agreement with earlier measurements
made with crude extracts from E. coli overexpressing
ESP [15]. Addition of Fe
3+
(0.5 mm) to ESP assays
also resulted in a substantial increase in epithionitrile
formation from allylglucosinolate. To determine if
A. thaliana ESP is strictly dependent on the presence
of these ions, we studied the effects of chelators that
bind exclusively divalent cations (BPDS), exclusively
trivalent cations (deferoxamine) or both (EDTA) [Figs
6-8]. EDTA completely inhibited nitrile formation by
ESP at a concentration of 2 mm, whereas in the pres-
ence of BPDS or deferoxamine (5 mm ) only small
amounts of the epithionitrile of allylglucosinolate were
formed. In assays carried out with EDTA (5 mm ), ESP
activity was restored by addition of Fe
2+
or Fe
3+
(0.5 mm). These results suggest a requirement for the
presence of iron cations in the catalytic mechanism of
epithionitrile formation with either Fe
2+
or Fe
3+
being sufficient for activity. Because added iron species
did not promote the formation of the simple nitrile
from allylglucosinolate in the presence of ESP, forma-
tion of the thiirane ring appears to be inseparable from

formation of the cyanide group. Therefore, the require-
ments of ESP for the presence of iron species apply to
epithionitrile as well as simple nitrile formation. Reac-
tivation of ESP in the presence of EDTA was also
accomplished by the addition of other divalent cations,
such as Ca
2+
and Mg
2+
. But these had to be added to
a much higher concentration (5 mm) than the iron cati-
ons and therefore may not be directly involved in the
catalytic mechanism. To test the mechanism of epithio-
nitrile formation proposed previously [17], the impact
of anaerobic conditions and several radical scavengers
on ESP activity was also investigated. Because no
changes in ESP activity were observed in these experi-
ments (Table 1), a mechanism similar to the formation
of epoxides as catalysed by cytochrome P450s cannot
be supported. Nevertheless, more detailed studies are
needed to disprove or confirm this proposed mechan-
ism.
Unlike A. thaliana ESP, the NSP from P. rapae does
not require the addition of iron species for its catalytic
activity, although Fe
2+
(0.01 mm) slightly increased
the formation of the nitrile of benzylglucosinolate
(Fig. 5B). Addition of EDTA or BPDS, chelators of
divalent cations, led to reduction but not complete

inhibition of nitrile formation by NSP (Fig. 6B). The
reduced NSP activity caused by EDTA was restored
by Fe
2+
and to a lesser extent by Fe
3+
. However, the
addition of deferoxamine to chelate trivalent cations
did not affect nitrile formation in the presence of NSP.
These results suggest that added iron species are not
essential for the mechanism of nitrile formation by
NSP; a required metal ion cofactor may, however, be
covalently bound to the protein.
Although both A. thaliana ESP and P. rapae NSP
alter the outcome of myrosinase-catalysed glucosino-
late hydrolysis, this comparative biochemical charac-
terization revealed a number of differences between
these two proteins, which also show no significant
similarity in their amino acid sequences. ESP promotes
the formation of epithionitriles and simple nitriles,
whereas NSP redirects glucosinolate hydrolysis
towards the formation of simple nitriles only. ESP
exhibits considerable substrate specificity among differ-
ent glucosinolates, whereas NSP does not. ESP
requires either added Fe
2+
or Fe
3+
for activity,
whereas NSP does not. Therefore, these two proteins

may promote nitrile formation by very different cata-
lytic mechanisms, but more research is needed to eluci-
date the details of these mechanisms. Although
functionally related, the plant ESP and the insect NSP
may not have the same biological role. Nitrile forma-
tion in P. rapae enables the larvae to overcome the
chemical defence system of their host plants. However,
because nitriles appear to be less toxic to most herbiv-
ores than isothiocyanates [7], it is currently not known
how plants benefit from ESP activity. Interestingly,
both proteins may come together in nature when larvae
of P. rapae feed on an ESP-containing plant species.
Experimental procedures
Intact glucosinolates and standards for analysis
of glucosinolate hydrolysis products
Allylglucosinolate was purchased from Sigma-Aldrich Che-
mie (Schnelldorf, Germany). All other intact glucosinolates
were isolated as described previously [28,29]. Allyl isothio-
cyanate (3-isothiocyanatoprop-1-ene) and allyl cyanide
(but-3-enenitrile) standards were obtained from Fluka
(Taufkirchen, Germany); benzyl isothiocyanate [1-(isothio-
cyanatomethyl)benzene], benzyl cyanide (phenylacetonitri-
le), and propyl isothiocyanate (1-isothiocyanatopropane)
were procured from Sigma-Aldrich Chemie; benzonitrile
was from Merck (Darmstadt, Germany).
Nitrile-forming proteins from plants and insects M. Burow et al.
2442 FEBS Journal 273 (2006) 2432–2446 ª 2006 The Authors Journal compilation ª 2006 FEBS
Analysis of glucosinolate hydrolysis products
by GC
Dichloromethane extracts of ESP and NSP assays were ana-

lysed by GC-MS and by GC with flame ionization detection
(FID) using an Agilent 6890 series gas chromatograph
with an HP5MS column (Agilent, Waldbronn, Germany)
(30 m · 0.25 mm · 0.25 lm), splitless injection at 200 °C
(injection volume 1 lL), and temperature programmes I
(35 °C for 3 min, 12 °CÆmin
)1
to 96 °C, 18 °CÆmin
)1
to
150 °C, 60 ° CÆmin
)1
to 300 °C, 4 min final hold), II (35 °C
for 3 min, 10 °CÆmin
)1
to 250 °C, 60 °CÆmin
)1
to 300 °C,
3 min final hold), or III (45 °C for 3 min, 12 °CÆmin
)1
to
96 °C, 18 °CÆ min
)1
to 150 °C, 60 °CÆmin
)1
to 270 °C, 3 min
final hold). MS and FID were carried out as described
previously [15]. Products were identified by comparison of
mass spectra and retention times with those of authentic
standards and with published MS spectra [30]. Quantifica-

tion of the products by GC-FID was carried out as
described previously [15]. For measurements of NSP
activity, benzyl isothiocyanate and benzyl cyanide were
quantified by GC-MS [18]. For separation of the diastereo-
mers of epithiopropyl cyanide, a chiral column (FS-Hydro-
dex b-3P, Macherey-Nagel, Du
¨
ren, Germany) and the
following temperature programme was used: 45 °C for
3 min, 4 °CÆmin
)1
to 100 °C, 18 °CÆmin
)1
to 180 °C, 1 min
final hold.
Purification of myrosinase from Sinapis alba
Thirty-five grams of white mustard seeds were homogenized
in 200 mL water using a Polytron PT3100 (Kinematica Inc.,
Newark, NJ, USA). Insoluble material was removed by cen-
trifugation at 20 000 g for 30 min. Proteins precipitated by
addition of (NH
4
)
2
SO
4
)70% saturation were dissolved in
25 mL water and dialysed overnight against 50 mm
Tris ⁄ HCl, pH 8.0, containing 1 mm EDTA. The dialysed
sample was centrifuged at 20 000 g for 30 min, and the

supernatant was concentrated by ultrafiltration through an
Amicon Ultrafiltration Disc PM30 (Millipore, Schwalbach,
Germany). The sample was applied to a column
(2.7 · 8 cm) of Fast Flow DEAE-Sepharose (Sigma-Aldrich
Chemie) equilibrated with 50 mm Tris ⁄ HCl, pH 8.0, con-
taining 1 mm EDTA. Proteins were eluted with a linear gra-
dient of 0–1 m NaCl in the starting buffer. Fractions with
myrosinase activity [31] were combined, ultrafiltered to
exchange the buffer to 20 mm Tris ⁄ HCl, pH 7.5, containing
0.5 m NaCl, 1 mm MnCl
2
, and 1 mm CaCl
2
, and loaded on
a column (1.5 · 6 cm) of Con A Sepharose 4B (Amersham
Pharmacia Biosciences) equilibrated with the same buffer.
Active fractions were eluted with 20 mm Tris ⁄ HCl, pH 7.5,
containing 0.5 m NaCl and 0.5 m methyl-a-d-glucoside.
Fractions were analysed by myrosinase assays [31] and
SDS ⁄ PAGE on Tris-SDS ⁄ PAGE gels (12%) after boiling
for 5 min with standard loading buffer. One unit of myros-
inase was defined as the amount of enzyme that hydrolyses
33 nmol benzylglucosinolate per min at room temperature
in 50 mm Mes buffer, pH 6.0, containing 1 mm substrate. In
fractions of several independent myrosinase purifications,
one unit corresponded to 50–75 ng of protein. Under the
conditions described for ESP and NSP assays, the hydrolysis
of different glucosinolates by myrosinase was linear with
protein concentration and with time for at least 2 h.
Generation of the expression construct for ESP

in Escherichia coli
To obtain the full-length ESP cDNA from A. thaliana, eco-
type Landsberg erecta (GenBank accession number
AF416787), RNA extracted from rosette leaves was tran-
scribed into cDNA using Superscript II Reverse Transcrip-
tase (Invitrogen), according to the manufacturer’s
instructions. The ESP cDNA was amplified from the RT
reaction by PCR using the forward primer (5¢-GCAGCCA
TGGCTCCGAC) and the reverse primer (5¢-TCATCTAG
ATTAAGCTGAATTGACCGCATAG). PCR was per-
formed in a total volume of 50 lL PCR buffer containing
2.5 units Pwo DNA Polymerase (Roche Molecular Bio-
chemicals), 1 mm MgSO
4
, 200 ng of template DNA, 200 lm
dNTPs, and 50 pmol of each primer. After incubation for
3 min at 94 °C, 30 cycles of 94 °C for 30 s, 60 ° C for 50 s,
72 °C for 60 s were carried out, followed by a final elonga-
tion phase at 72 °C for 5 min. The PCR product was diges-
ted with NcoI and XbaI and ligated into the pTrc99A vector
(Amersham Pharmacia Biotech). The ESP cDNA was
amplified from this construct by PCR using the forward pri-
mer (5¢-ATGGTAGGTCTCAGCGCGCTCCGACTTTGC
AAGGCCAG) and the reverse primer (5¢-ATGGTAGG
TCTCATATCATGAATTGACCGCATAGAAGTAGAG).
PCR was performed as described above, but the cycling
conditions were 94 °C for 3 min, 20 cycles of 94 °C for 30 s,
55 °C for 30 s, 72 °C for 60 s, and finally 72 °C for 5 min.
The BsaI-digested PCR product was ligated into the pASK-
IBA7 vector (IBA GmbH, Go

¨
ttingen) in frame with the
N-terminal Strep-tag II. The ESP cDNA including the tag
was subsequently cloned into the pCRT7 ⁄ CT-TOPO vector
(Invitrogen) by PCR using the forward primer (5¢-ATG
GCTAGCTGGAGCCACCCG) and the reverse primer (5¢-
TCATCTAGATTAAGCTGAATTGACCGCATAG). PCR
was performed as described for the cloning of the ESP
cDNA into pASK-IBA7. After cloning into the
pCRT7 ⁄ CT-TOPO expression vector, the correct insert
sequence was verified by sequencing.
Heterologous expression in E. coli and
purification of the recombinant protein
E. coli strain BL21(DE3)pLysS (Invitrogen) was trans-
formed with the ESP expression construct or the empty
M. Burow et al. Nitrile-forming proteins from plants and insects
FEBS Journal 273 (2006) 2432–2446 ª 2006 The Authors Journal compilation ª 2006 FEBS 2443
pCRT7 ⁄ CT-TOPO vector. Single colonies selected on
Luria–Bertani medium supplemented with 100 lgÆmL
)1
ampicillin and 34 lgÆmL
)1
chloramphenicol were used to
inoculate 20 mL of terrific broth (TB) medium, containing
the same antibiotics. After growth at 18 °C and 220 r.p.m.
for 62 h, 10 mL of the precultures were transferred to 1 L
of TB medium without antibiotics. The cultures were
grown at 18 °C and 220 r.p.m. to an D
600
of 0.4–0.6 and

protein expression was induced by addition of isopropyl-1-
thio-b-d-galactopyranoside at a final concentration of
1mm. Cells were pelleted 16 h after induction, resuspended
in 20 mL of 100 mm Tris ⁄ HCl, pH 8.0, containing 150 mm
NaCl, and extracted by sonication. The recombinant ESP
was purified from the 10 000 g supernatant by affinity chro-
matography using Strep-Tactin Sepharose resin (IBA
GmbH, Go
¨
ttingen, Germany), with one deviation from the
manufacturer’s instructions: no EDTA was added to the
buffer used for protein purification. The purity of the frac-
tions was analysed by Tris-SDS ⁄ PAGE as described above.
Protein content was determined with the Pierce BCA Protein
Reagent (Pierce, Rockford, IL) using bovine plasma c-glob-
ulin (Biorad, Munich, Germany) as a standard. Purified
ESP was stable for 12 h when stored at 4 °C.
ESP assay
ESP assays were carried out in 500 lL50mm Mes buffer,
pH 6.0, with 1 unit of ESP (unless otherwise stated).
Glucosinolates were added to the assay mixture to a final
concentration of 1 mm (allyl-, 4-methylthiobutyl-, and ben-
zylglucosinolate) or 3 mm (4-methylsulfinylbutylglucosino-
late). The reaction was started by the addition of 4 units of
myrosinase. After incubation at room temperature (22–
24 °C) for 1 h, 50 lL of benzonitrile (100 ngÆlL
)1
in meth-
anol) were added as internal standard. The assay mixtures
were extracted with 2 · 750 lL dichloromethane, and the

organic phases were combined, dried over Na
2
SO
4
, and
concentrated under a nitrogen stream. Glucosinolate hydro-
lysis products were analysed by GC-FID with temperature
programme I (allylglucosinolate) or with temperature pro-
gramme II (other glucosinolates). One unit of ESP was
defined as the amount of enzyme that forms 30 nmol epith-
iopropyl cyanide [2-(thiiran-2-yl)acetonitrile] per min from
allylglucosinolate under the conditions described above. In
fractions of independent purifications, one unit correspon-
ded to 4–6 lg purified protein. Under the conditions des-
cribed above, epithiopropyl cyanide formation by ESP was
linear with time for 2 h. All data presented are the
mean ± SD of results obtained in at least three independ-
ent experiments.
ESP assays in dialysis cassette
Purified recombinant ESP (25 units) was added to a Slide-
A-Lyser dialysis cassette (0.5 mL, 10 000 MWCO, Pierce).
The cassette was placed into a beaker with 80 mL of
50 mm Mes buffer, pH 6.0, containing 50 units of myrosin-
ase. After equilibration for 10 min on a magnetic stirrer,
the reaction was started by addition of allylglucosinolate to
the outer solution at a final concentration of 1 mm. After
incubation at room temperature for 90 min, 50 lL of ben-
zonitrile (100 ngÆlL
)1
in methanol) were added to 10 mL

of the assay mixture, which was then extracted
with 2 · 10 mL dichloromethane. Organic phases were
combined, dried over Na
2
SO
4
, concentrated to 200 lL and
analysed by GC-FID with temperature programme I. Con-
trol reactions were carried out under the same conditions
without a dialysis cassette.
NSP assay
NSP was purified from larval midgut extracts of P. rapae
as described previously [18]. NSP assays were performed
in 50 mm Mes buffer, pH 6.0, containing 1 mm glucosino-
late and 1 unit NSP. The reaction was started by the addi-
tion of 1 unit of myrosinase. After incubation at room
temperature for 40 min, 50 l L of benzonitrile (100 ngÆlL
)1
in methanol) were added. Hydrolysis products were extrac-
ted with dichloromethane as described for the ESP assay
and analysed by GC-MS (benzylglucosinolate, pro-
gramme III) or GC-FID (allylglucosinolate, temperature
programme I; 4-methylthiobutyl- and 4-methylsulfinylbu-
tylglucosinolate, temperature programme II; benzylglucosi-
nolate, temperature programme III). One unit of NSP was
defined as the amount of enzyme that catalyses the forma-
tion of 15 nmol phenylacetonitrile per min from benzylglu-
cosinolate under these conditions. Under standard assay
conditions, benzyl cyanide formation by NSP was linear
with time for 40 min. Data presented are the mean ± SD

of results obtained in at least three independent experi-
ments.
Anti-ESP serum
A rabbit polyclonal antibody against ESP was raised using
the synthetic peptide H
2
N-RDENRNFENFRSYDTV-CO-
NH
2
which corresponds to the amino acids 94–109 of the
native protein (Eurogentec, Seraing, Belgium). The peptide
was N-terminally coupled with keyhole limpet hemocyanin
as a carrier.
ESP-affinity chromatography
The anti-ESP serum was purified from the immune serum
by protein A binding using the Nab Spin Purification Kit
(Pierce) and subsequently dialysed against NaCl ⁄ P
i
. ESP-
affinity chromatography was carried out with the Seize Pri-
mary Immunoprecipitation Kit (Pierce). The antibody was
covalently coupled to AminoLink Plus Coupling Gel
Nitrile-forming proteins from plants and insects M. Burow et al.
2444 FEBS Journal 273 (2006) 2432–2446 ª 2006 The Authors Journal compilation ª 2006 FEBS
according to the manufacturer’s instructions. The antibody-
coupled resin (0.2 mL) was equilibrated with 50 mm
Tris ⁄ HCl, pH 7.5, before 50 lg of freshly purified ESP
were loaded. The ESP-affinity resin was obtained after
incubation at 4 °C overnight with end-over-end mixing,
and unbound ESP was removed by washing three times

with 0.4 mL 50 mm Tris ⁄ HCl, pH 7.5. Crude protein
extracts were prepared from rosette leaves of A. thaliana
Col-0 by grinding 500 mg of leaf material with 500 lLof
50 mm Tris ⁄ HCl, pH 7.5, followed by centrifugation at
10 000 g for 10 min. A 500 lL aliquot of the supernatant
(4 mg total protein) was combined with the ESP-affinity
resin and incubated at room temperature for 2 h with end-
over-end mixing. After washing three times with 0.4 mL
25 mm Tris ⁄ HCl, pH 7.2, containing 150 mm NaCl, to
remove unbound components of the plant extract, the anti-
gen (ESP) and bound proteins from the plant extract were
eluted according to the manufacturer’s instructions. Frac-
tions were separated on 12% Tris-SDS ⁄ PAGE gels and
transferred to nitrocellulose membranes (Schleicher &
Schuell, Dassel, Germany). Membranes were blocked with
2% dry milk powder in TTBS (20 mm Tris-HCL, pH 7.5,
150 mm NaCl, 0.1% Tween 20) for 30 min before addition
of the anti-ESP serum (1 : 10.000 diluted with TTBS, con-
taining 2% dry milk powder). After incubation at 4 °C
overnight and washing with TTBS (3 · 7 min), the
secondary antibody [anti-(rabbit IgG) alkaline phosphatase-
conjugate, Sigma-Aldrich Chemie, 1:10 000] was added in
TTBS, containing 2% dry milk powder. Membranes were
incubated at room temperature for 1 h, washed with TTBS
(3 · 7 min) and rinsed with detection buffer (100 mm
Tris ⁄ HCl, pH 9.5, containing 100 mm NaCl and 5 mm
MgCl
2
). Detection was carried out with 0.085 lgÆmL
)1

NBT and 0.17 lgÆmL
)1
BCIP in detection buffer. Myrosin-
ase activity was measured as described for ESP assays using
allylglucosinolate as a substrate. Hydrolysis products
formed were analysed and quantified by GC-FID (tempera-
ture programme I). Addition of 50 lg purified anti-ESP
serum to ESP assays carried out under standard assay con-
ditions did not affect ESP activity.
Acknowledgements
We thank Andrea Bergner for technical assistance,
Michael Reichelt for providing intact glucosinolates,
and the Max Planck Society for financial support.
References
1 Kessler A & Baldwin IT (2002) Plant responses to insect
herbivory: the emerging molecular analysis. Annu Rev
Plant Biol 53, 299–328.
2 Wittstock U & Gershenzon J (2002) Constitutive
plant toxins and their role in defense against herbi-
vores and pathogens. Curr Opin Plant Biol 5, 300–
307.
3 Fahey JW, Zalcmann AT & Talalay P (2001) The che-
mical diversity and distribution of glucosinolates and
isothiocyanates among plants. Phytochemistry 56, 5–51.
4 Wittstock U & Halkier BA (2002) Glucosinolate
research in the Arabidopsis era. Trends Plant Sci 7, 263–
270.
5 Matile P (1980) The mustard oil bomb – compartmenta-
tion of the myrosinase system. Biochemie Physiologie
Pflanzen 175, 722–731.

6 Rask L, Andreasson E, Ekbom B, Eriksson S, Pon-
toppidan B & Meijer J (2000) Myrosinase: gene family
evolution and herbivore defense in Brassicaceae. Plant
Mol Biol 42, 93–113.
7 Wittstock U, Kliebenstein D, Lambrix VM, Reichelt M
& Gershenszon J (2003) Glucosinolate hydrolysis and
its impact on generalist and specialist insect herbivores.
In Recent Advances in Phytochemistry – Integrative
Phytochemistry. From Ethnobotany to Molecular Ecology
(Romeo JT, ed.), pp. 101–126. Elsevier, Amsterdam.
8 Fahey JW, Zhang YS & Talalay P (1997) Broccoli
sprouts: an exceptionally rich source of inducers of
enzymes that protect against chemical carcinogens. Proc
Natl Acad Sci USA 94, 10367–10372.
9 Hecht SS (2000) Inhibition of carcinogenesis by isothio-
cyanates. Drug Metab Rev 32, 395–411.
10 Schlueter M & Gmelin R (1972) Abnormale enzyma-
tische Spaltung von 4-Methylthiobutylglucosinolat in
Frischpflanzen von Eruca sativa. Phytochemistry 11,
3427–3431.
11 Hasapis X & MacLeod AJ (1982) Benzylglucosinolate
degradation in heat-treated Lepidium sativum seeds and
detection of a thiocyanate-forming factor. Phytochemis-
try 21, 1009–1013.
12 MacLeod AJ & Rossiter JT (1985) The occurrence and
activity of epithiospecifier protein in some Cruciferae
seeds. Phytochemistry 24, 1895–1898.
13 Tookey HL (1973) Crambe thioglucoside glucohydro-
lase (EC 3.2.3.1) – separation of a protein required
for epithiobutane formation. Can J Biochem 51, 1654–

1660.
14 Petroski RJ & Kwolek WF (1985) Interaction of a fun-
gal thioglucoside glucohydrolase and cruciferous plant
epithiospecifier protein to form 1-cyanoepithioalkanes:
implications of an allosteric mechanism. Phytochemistry
24, 213–216.
15 Lambrix VM, Reichelt M, Mitchell-Olds T, Kliebenstein
DJ & Gershenzon J (2001) The Arabidopsis epithiospeci-
fier protein promotes the hydrolysis of glucosinolates to
nitriles and influences Trichoplusia ni herbivory. Plant
Cell 13, 2793–2807.
16 Brocker ER & Benn MH (1983) The intramolecular for-
mation of epithioalkanenitriles from alkenylglucosino-
M. Burow et al. Nitrile-forming proteins from plants and insects
FEBS Journal 273 (2006) 2432–2446 ª 2006 The Authors Journal compilation ª 2006 FEBS 2445
lates by Crambe abyssinica seed flour. Phytochemistry
22, 770–772.
17 Foo HL, Gronning LM, Goodenough L, Bones AM,
Danielsen BE, Whiting DA & Rossiter JT (2000) Purifica-
tion and characterisation of epithiospecifier protein from
Brassica napus: enzymic intramolecular sulphur addition
within alkenyl thiohydroximates derived from alkenyl
glucosinolate hydrolysis. FEBS Lett 468, 243–246.
18 Wittstock U, Agerbirk N, Stauber EJ, Olsen CE, Hip-
pler M, Mitchell-Olds T, Gershenson J & Vogel H
(2004) Successful herbivore attack due to metabolic
diversion of a plant chemical defense. Proc Natl Acad
Sci USA 101, 4859–4864.
19 Bernardi R, Negri A, Ronchi S & Palmieri S (2000) Iso-
lation of the epithiospecifier protein from oil-rape (Bras-

sica napus ssp. oleifera) seed and its characterization.
FEBS Lett 467, 296–298.
20 Adams J, Kelso R & Cooley L (2000) The Kelch repeat
superfamily of proteins: propellers of cell function.
Trends Cell Biol 10, 17–24.
21 Burmeister WP, Cottaz S, Rollin P, Vasella A & Henris-
sat B (2000) High resolution X-ray crystallography
shows that ascorbate is a cofactor for myrosinase and
substitutes for the function of the catalytic base. J Biol
Chem 275, 39385–39393.
22 Ohtsuru M & Hata T (1979) Interaction of l -ascorbic-
acid with the active-center of myrosinase. Biochim
Biophys Acta 567, 384–391.
23 Shikita M, Fahey JW, Golden TR, Holtzclaw WD &
Talalay P (1999) An unusual case of ‘uncompetitive
activation’ by ascorbic acid: purification and kinetic
properties of a myrosinase from Raphanus sativus seed-
lings. Biochem J 341, 725–732.
24 Hasapis X & MacLeod AJ (1982) Effects of metal ions
on benzylglucosinolate degradation in Lepidium sativum
seed autolysates. Phytochemistry 21, 559–563.
25 Tookey HL & Wolff IA (1970) Effect of organic redu-
cing agents and ferrous ion on thioglucosidase activity
of Crambe abyssinica seed. Can J Biochem 48, 1024–&.
26 Petroski RJ (1986) Stereoselectivity of the interactions
of thioglucoside glucohydrolase and epithiospecifier pro-
tein from various sources. Plant Sci 44, 85–88.
27 Agrawal AA & Kurashige NS (2003) A role for isothio-
cyanates in plant resistance against the specialist herbi-
vore Pieris rapae. J Chem Ecol 29, 1403–1415.

28 Zrybko CL, Fukuda EK & Rosen RT (1997) Determi-
nation of glucosinolates in domestic and wild mustard
by high-performance liquid chromatography with confir-
mation by electrospray mass spectrometry and photo-
diode-array detection. J Chromatogr A 767, 43–52.
29 Thies W (1979) Detection and utilization of a glucosino-
late sulfohydrolase in the edible snail, Helix pomatia.
Naturwissenschaften 66, 364–365.
30 Spencer GF & Daxenbichler ME (1980) Gas chromato-
graphy-mass spectrometry of nitriles, isothiocyanates
and oxazolidinethiones derived from cruciferous glucosi-
nolates. J Sci Food Agric 31, 359–367.
31 Zhang YS, Wade KL, Prestera T & Talalay P (1996)
Quantitative determination of isothiocyanates, dithiocar-
bamates, carbon disulfide, and related thiocarbonyl
compounds by cyclocondensation with 1,2-benzene-
dithiol. Anal Biochem 239, 160–167.
Nitrile-forming proteins from plants and insects M. Burow et al.
2446 FEBS Journal 273 (2006) 2432–2446 ª 2006 The Authors Journal compilation ª 2006 FEBS