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: u.wittstock@tu-bs.de (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 (http://www.ebi.ac.uk/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 [...].. .Nitrile-forming proteins from plants and insects M Burow et al iron-dependent proteins [12] The mechanism of epithionitrile formation by ESP has been suggested to be analogous to that of cytochrome P450-mediated epoxidations and to require Fe2+ ⁄ Fe3+ for intramolecular sulfur transfer [17] Epithionitrile formation from allylglucosinolate and simple nitrile formation from nonalkenyl... 2443 Nitrile-forming proteins from plants and insects M Burow et al 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... 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 glucosinolate hydrolysis, ... 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 Chemie (Schnelldorf, Germany) All... 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] Quantification 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... 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- Nitrile-forming proteins from plants. .. Nitrile-forming proteins from plants and insects 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... myrosinase-catalysed glucosinolate hydrolysis, this comparative biochemical characterization 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... Addition of 50 lg purified anti-ESP serum to ESP assays carried out under standard assay conditions did not affect ESP activity Acknowledgements Nitrile-forming proteins from plants and insects 3 4 5 6 7 8 9 10 11 12 13 14 We thank Andrea Bergner for technical assistance, Michael Reichelt for providing intact glucosinolates, and the Max Planck Society for financial support 15 References 1 Kessler A &... formation of epithioalkanenitriles from alkenylglucosino- FEBS Journal 273 (2006) 2432–2446 ª 2006 The Authors Journal compilation ª 2006 FEBS 2445 Nitrile-forming proteins from plants and insects 17 18 19 20 21 22 23 M Burow et al lates by Crambe abyssinica seed flour Phytochemistry 22, 770–772 Foo HL, Gronning LM, Goodenough L, Bones AM, Danielsen BE, Whiting DA & Rossiter JT (2000) Purification and characterisation . Comparative biochemical characterization of nitrile-forming proteins from plants and insects that alter myrosinase-catalysed hydrolysis of glucosinolates Meike. 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