Characterization and synthetic applications of recombinant AtNIT1 from Arabidopsis thaliana Steffen Osswald, 1 Harald Wajant 2 and Franz Effenberger 1 1 Institut fu È r Organische Chemie, and 2 Institut fu È r Zellbiologie und Immunologie, Universita È t Stuttgart, Germany The nitrilase AtNIT1 from Arabidopsis thaliana was over- expressed in Escherichia coli with an N-terminal His 6 tag and puri®ed by zinc chelate anity chromatography in a single step almost to homogeneity in a 68% yield with a speci®c activity of 34.1 U ámg )1 . The native enzyme ( 450 k Da) consists of 11±13 subunits (38 kDa). The temperature optimum was determined to be 35 °C, and a pH optimum of 9 was fo und. Thus, recombinant AtNIT1 resembles in its properties the native enzyme and the nitrilase from Brassica napus. The stability of AtNIT1 could be signi®cantly i mproved by t he addition of dithiothreitol and E DTA. The substrate range of AtNIT1 diers considerably from those of bacterial nitrilases. Aliphatic nitriles are the most eective substrates, showing increasing rates of hydrolysis with increasing size of the residues, a s d emonstrated in the series butyronitrile, octanenitrile, phenylpropionitrile. In comparison with 3-indolylacetonitrile, the rate of hydrolysis of 3-phenyl- propionitrile is increased by a factor of 330, and the K m value is reduced by a factor of 23. With the exception of ¯uoro, substituents in the a position to the nitrile function completely inhibit the hydrolysis. Keywords: Arabidopsis thaliana; enzymatic properties; nitrile; substrate s peci®city. Nitriles are found in a variety of naturally occurring compounds such as cyanolipids, cyanoglucosides, and simple aliphatic or aromatic nitriles as metabolites of micro-organisms [1]. In nature, the hydrolysis of nitriles to the corresponding carboxylic acid and NH 3 is catalyzed by nitrilases ( EC 3.5.5.1) or based on the sequential action of a nitrile hydratase (EC 4.2.1.84)±amidase (EC 3.5.1.4) system [2,3]. Most nitrilases described so far have been isolated from fungi or bacteria. In recent years, however, four nitrilases (AtNIT1±AtNIT4) have been cloned from Arabidopsis thaliana, a member of the brassicaceae family [4,5]. The genes of AtNIT1±3 are clustered on chromosome 3 and have sequence identities of more than 80% at the amino acid level, whereas AtNIT4 has a distinct chromo- somal localization a nd is only 65% identical with AtNIT1±3 [5]. The subdivision of the Arabidopsis nitrilases i nto AtNIT1±3 and AtNIT4 is also re¯ected by functional differences between these enzymes. Whereas AtNIT1±3 convert 3-indolylacetonitrile (IAN) into the plant hormone 3-indolylacetic acid, IAN is not a substrate for AtNIT4 [6,7]. Moreover, homologs of AtNIT1±3 have exclusively been found in Arabidopsis and other members of the brassicaceae, whereas AtNIT4 isoforms have also been reported in species from other taxonomic groups such as tobacco [8] and rice [7]. In accordance with the brassi- caceae-restricted occurrence of nitrilases of the AtNIT1±3 type, t hese en zymes seem to be involved in the degradation of nitriles released from glucosinolates, which can be found in high concentrations in various species of the brassicaceae [9]. Recent studies have shown that AtNIT4 and two related n itrilases f rom t obacco are b-cyano-( L )-alanine nitrilases [7]. As nitrilases of the AtNIT4 type have been found in taxonomically quite distinct groups, it seems likely that AtNIT4 homologs may exist in all higher plants. In accordance with this is the fact that the substrate of the AtNIT4-type nitrilases, b-cyano-( L )-alanine, seems to occur in all plants as t he result of detoxi®cation of cyanide, which is inevitably produced during biosynthesis of the plant hormone ethylene [10]. In general, nitriles are synthetically more acce ssible than the corresponding carboxylic acids. Chemical hydrolysis of nitriles to carboxylic acids, however, requires drastic conditions (strong mineral acids and bases and relatively high reaction temperature). Biocatalysts for the transfor- mation of nitriles to carboxylic acids are therefore of particular interest. Up until now, hydratase±amidase systems, not nitri- lases, have mainly been used in practice as nitrile- hydrolyzing enzymes [3,11±14]. In this paper, we report on basic investigations of the nitrilase AtNIT1 from A. thaliana, in particular, the substrate range required for the hydrolysis of nitriles to carboxylic acids. Cloning and overexpression of AtNIT1 [4,5] (EC 3.5.5.1) will provide an interesting plant nitrilase in suf®cient quantities for synthetic applications. The application of AtNIT1 to the hydrolysis of several speci®c substrates such as aliphatic dinitriles and 2-¯uoroarylacetonitriles has been published in detail [15,16]. Correspondence to F. Eenberger, Institut fu È r Organische Chemie, Universita È t Stuttgart, Pfaenwaldring 55, D- 7 0569 St uttgart, Germany. Fax: + 49 711685 4269, Tel.: + 49 711685 4265, E-mail: franz.eenberger@po.uni-stuttgart.de Abbreviations: AtNIT1, nitrilase from Arabidopsis thaliana;IAN, 3-indolylacetonitrile. Note: This is part 42 of the series of publications Enzyme catalyzed reactions. Part 41 is Eenberger , F. & Osswald, S. (2001) Select ive hydrolysis of aliphatic dinitriles to monocarboxylic acids by a nitrilase from Arabidopsis thaliana. Synthesis 1866±1872. (Received 3 September 2001, revised 23 November 2001, accepted 26 November 2001) Eur. J. Biochem. 269, 680±687 (2002) Ó FEBS 2002 MATERIALS AND METHODS Expression cloning of AtNIT1 AtNIT1 cDNA was cloned in the expression vector pQE10 (Qiagen), which allows isopropyl b- D -thiogalactoside- induced expression of N-terminally His-tagged recombinant protein. In brief, the c oding region and p art of the 3¢-noncoding region of AtNIT1 cDNA were ampli®ed from an A. thaliana cDNA library (Stratagene) with an advanced polymerase system (Clontech) using the primers AtNIT1-for (5¢-GCTGCTAGATCTTATGTC AACTGT CCAAAA CGCAACTCCTTTTAACGGCGTTGCCCC ATCCACC -3¢; start codon according to [4] in bold) and AtNIT1-rev (5¢-ACAATTGATGATTCAACGCCCAAC 3¢). Using the BglII sites in the 5¢ overhang of AtNIT1-for and the 3¢-noncoding region of the cDNA, the AtNIT1 cDNA was inserted in-frame in the BamHI site of pQE10. The resulting expression plasmid pQE10-AtNIT1 was sequenced to con®rm the identity of the AtNIT1 sequence after PCR ampli®cation. pQE10-AtNIT1 was transformed in Escherichia coli M15[pREP4] cells (Qiagen) for over- expression of AtNIT1. For induction of recombinant AtNIT1, an overnight culture was performed at 37 °Cin Luria±Bertani medium supplemented with ampillicin (50 lgámL )1 ) and kanamycin (20 lgámL )1 ), diluted 1 : 20 with Luria±Bertani medium supplemented again with ampillicin and kanamycin, and grown at 30 °C. After 4 h, isopropyl b- D -thiogalactoside was added to a ®nal concen- tration of 0.5 m M for in duction of AtNIT1 expression. After an additional 6 h, cells were harvested. Preparation of the crude extract and puri®cation of recombinant AtNIT1 Cells were separated from the nutrient medium by centri- fugation (30 min, 4 °C, 5700 g), and washed with sodium phosphate buffer A (50 m M , pH 7.8). The p ellet was resuspended in buffer A (100 mL per 10 g wet weight) andsonicated(3´ 5min, 0°C). The homogenate was centrifuged (40 min, 4 °C, 186 000 g). The supernatant (100 mL) was degassed with argon, ®ltered through a membrane (70 lm)andappliedtoaZn 2+ -charged HiTrap metal c helate af®nity chromatography column (Pharmacia). The column was rinsed successively with 20 mL each of sodium p hosphate buffer B (50 m M ,100 m M NaCl, p H 7.8) and buffer A until the absorbance reached the base line of column equilibration. Nonspeci®cally bound proteins were eluted at a ¯ow rate of 2 mLámin )1 in a 22.5-mL linear gradient of 0±100 m M imidazole in buffer A, and succes- sively in 5 mL of sodium phosphate buffer C (50 m M , 100 m M imidazole, pH 7.8). After additional rinsing with 11.25 mL buffer A, AtNIT1 was eluted with 11.25 m L sodium phosphate buffer (50 m M ,100 m M EDTA, pH 7.8). To the collected fractions (2.5 mL), 25 lL sodium phos- phate buffer (50 m M , 100 m M dithiothreitol, pH 7.8) was added, and after measurement of enzyme activity, fractions were pooled. Gel-®ltration analysis Recombinant puri®ed AtNIT1 (200 lL) was separated by size-exclusion c hromatography on a Superdex 200 HR10/30 column (Pharmacia) in 50 m M sodium phosphate buffer, containing 100 m M EDTA and 1 m M dithiothreitol, pH 7.8, at a ¯ow rate of 0.5 m Lámin )1 . For calibration of the column, thyroglobulin (663 kDa), apoferritin (443 kDa), alcohol dehydrogenase (150 kDa), BSA (66 k Da), carbonic anhydrase ( 29 kDa) and cytochrome c (12.4 k Da) (all from Sigma) were used. Enzyme assay Enzyme activity towards 3-phenylpropionitrile was assayed using bacterial protein (0.34±135 mg) in 5 mL Tris/HCl buffer (70 m M ,pH8.5)and50lL 3-phenylpropionitrile in methanol (0.25 M ). The reaction was carried out for 1 h at 35 °C. An aliquot of 1 mL was acidi®ed with 50 lLHCl (5 M ) and ex tracted with d iethyl ether (5 mL). After centrifugation (5 min, 2000 g) and cooling at )30 °Cfor 30 min to freeze the aqueous layer, the organic layer was decanted and derivatized with ethereal diazomethane (0.2 M ). After concentration, the residue was taken up in 1 m L d iethyl ether and subjected to gas chromatography on a C arlo Erba Fractovap 4160 wit h FID and Spectra Physics minigrator using a capillary glass column ( 50 m) with PS086 and carrier gas 50 kPa hydrogen. Peak areas were calibrated as follows. A volume of 5 mL each of a solution of 3-phenylpropionitrile (181.5 mg) and 3-phenylpropionic acid (205.2 mg) in methanol (10 mL), and Tris/HCl buffer (990 mL, 70 m M , pH 8.5) were mixed, and 5 mL from this mixture was added to 5 mL of the 3-phenylpropionitrile solution. This procedure w as repeated three times. A sample of 1 mL from each solution was treated as described above and analyzed by gas chromatography. The conversion factor was determined from the plot of ratio areas vs. ratio concentrations. One unit is de®ned as 1 lmol convert- edámin )1 . Determination of temperature and pH optimum of AtNIT1 Temperature dependence. Nitrilase activity towards 3-phenylpropionitrile was assayed as described above using puri®ed enzyme (55.2 UámL )1 ,1.98mgproteinámL )1 )ina 1 : 5000 dilution with Tris/HCl buff er ( 70 m M ,pH8.5)and 50 lL 3-phenylpropionitrile in methanol (0.25 M ). The reaction was initiated by the addition of substrate either directly after preliminary heating at the respective tempera- ture for 10 min or cooling at 7 °C f or 30 min and after 24 h, respectively. PH dependence. Enzyme activity was assayed as described above using puri®ed enzyme (56.8 UámL )1 ,1.78mgpro- teinámL )1 ) in phosphate buffer (50 m M , pH 7.8), which w as diluted (1 : 5000) at 4 °C w ith the respectiv e buffer. Aft er preliminary w arming at room temperature, the reaction w as initiated by the addition of 50 lL 3-phenylpropionitrile in methanol. RESULTS Puri®cation and determination of K m values Recombinant AtNIT1 was puri®ed from E. coli lysates by metal chelate af®nity chromatography using a Zn 2+ -charged Ó FEBS 2002 Synthetic applications of recombinant AtNIT1 (Eur. J. Biochem. 269) 681 HiTrap column. After a wash step with 100 m M imidazole, the tightly bound AtNIT1 was eluted with h igh recovery by 100 m M EDTA (Fig. 1A). This single-step puri®cation yielded almost pure AtNIT1 (Fig. 1B) with a speci®c activity of 34.1 Uámg )1 (Table 1) and a subunit mass of 38 kDa (Fig. 1 B). Recombinant AtNIT1 was e luted during gel-®ltration chromatography (Fig. 1C) in fractions corre- sponding to a molecular m ass of  450 kDa, s uggesting that native AtNIT1 occurs as a homomeric protein complex of 11±13 subunits (data not shown). Recombinant AtNIT1 showed a K m value of 3.67 m M for 3-indolylacetonitrile and 0.159 m M for 3-phenylpropionit- rile (Fig. 2A,B). The K m value for 3-indolylacetonitrile is in good agreement with a reported value of 5 m M [4]. Enzyme stability As a crude extract o f recombinant AtNIT1 had a half-time of 2 d ays at pH 8 and 4 °C, the i n¯uence of antioxidants and protease inhibitors on enzyme stability was investi- gated. Of the applied thiols that p roved to be good an tioxidants (mercaptoethanol and dithiothreitol) [4,17±21], dithiothre- itol had the better stabilizing effect (Table 2). The loss of enzyme activity on the addition of 2 m M dithiothreitol was 20% compared with 63% for the reference without thiol. However, increasing the dithiothreitol concentration to 5m M did not further improve enzyme stability. The best result with protease inhibitors was achieved using EDTA at a concentration of 2 m M (Table 2). Thus, all buffers used for cell disintegration and conversions were supplemented with dithiothreitol a nd EDTA (2 m M each). In this way, we succeeded in signi®cantly increasing the enzyme stability of both crude extract and puri®ed enzyme: after 2 days at room temperature and 3 months at 4 °C, 95% and 90% enzyme activity, respectively, remained. Temperature optimum The nitrilases investigated so far generally show highest activity in the temperature range 35±40 °C, no matter what the enzyme source [18,21±23]. However, as little is known about their stability a t higher temperatures, which is a decisive factor in their application as biocatalysts in chemical reactions, the effect of temperature on AtNIT1 stability was investigated. Recombinant AtNIT1 shows a sharp temperature optimum at 35 °C, determined after 1 h of incubation, with a gentle slope at < 35 °C a nd a steeper slope at > 35 °C (Fig. 3). E nzyme stability at different temperatures was determined after 24 h of incubation. At 25 °Cand35°C, only a slight decrease in activity was found. At 35 °C, the relative enzyme activity amounts to  80%, whereas the enzyme was almost completely deac- tivated at 40 °C. The highest absolute enzyme activity, Fig. 1. Puri®cation and characterization of recombinant AtNIT1 in E. c oli. (A) Lysate of isopropyl b- D -thiogalactopyranoside-induced E. c oli-pQE10-AtNIT1 was applied to a Zn 2+ -charged HiTrap column. The column was washed w ith 100 m M imidazole (- - - -), and bound AtNIT1 was eluted with 100 m M EDTA (± ± ± ±). Fractions were analyzed for nitrilase a ctivity as described in M aterials and methods. The elution pro ®le was detected at 280 nm. (B) Fractions o btaine d in (A) were separat ed by SDS /PAGE a nd stain ed with Coo massie . Last lane, molecular masses of standards (kDa); lanes 17±19, active fractions after HiTrap chromatography. (C) E stimation of native m olecular mass of recombinant AtNIT1 by gel-®ltration chromatography on a Superdex 200 HR10/30 column. A 200 lL volume of puri®ed AtNIT1 (d) and various m ass standards [s; thyroglobulin ( 663 kDa), apoferritin (443 kDa), alcohol dehydrogenase (150 kDa), BSA (66 kDa), carbo nic anhydrase ( 29 kDa) and c ytochrom e c (12.4 kDa)] were separated at a ¯ow rate of 0.5 mLámin )1 .AtNIT1waseluted.The AtNIT1 elution volume corresponds to m olec ular mass of  450 kDa. The dotted line shows the 95% con®dence interval of the linear regression of the s emilogarithmic molecular m ass±elution volume plot. Table 1. Summary of puri®cation of the nitrilase from A. thaliana. Fraction Total activity (U) Total protein (mg) Speci®c activity (Uámg )1 ) Puri®cation (fold) Yield (%) Crude extract 459 4350 0.10 1 100 HiTrap 312 9.15 34.10 326 68 682 S. Osswald et al.(Eur. J. Biochem. 269 ) Ó FEBS 2002 however, w as found at 35 °C, and, moreover, t he stability at this temperature is suf®cient for applications in longer lasting biotransformations. PH dependence of AtNIT1 The pH dependence of AtNIT1 was investigated with different buffer systems in order to guarantee suf®cient buffering capacity in the range pH 6±10 (Fig. 4). As can be seen from Fig. 4, the choice of the buffer system affects the enzyme activity slightly, changing from Tris/HCl to glycine/NaOH. With both buffer systems, however, an activity optimum of pH 9.0 was found, with 97% of the maximum activity being measured at pH 8.5. The decrease in enzyme activity at pH values > 9 is not an irreversible process: acidifying an enzyme solution with pH 10 back to pH 9 r esulted in > 80% r ecovery of activity. The pH optimum measured in this way is in slight contrast with the value of pH 7.5 reported for the nitrilase from A. thaliana [4], possibly arising from the deviant structure at the N-terminus. However, several bacterial nitrilases also clearly have a basic pH optimum [23±27]. Substrate range of recombinant AtNIT1 The substrate range of recombinant AtNIT1 was investi- gated using structurally varied aromatic a nd aliphatic nitriles (Table 3). The a ctivities g iven in Table 3 are referred to the speci®c nitrilase activity towards butyronitrile. As c an be seen, aliphatic nitriles are the most effective substrates, showing increased rates of hydrolysis with increasing size of the hydrophobic residue, in the order butyronitrile, octanenitrile, phenylpropionitrile. In contrast with 3-phe- nylpropionitrile, arylacetonitriles s uch as benzyl cyanide were converted 20 times more slowly. Aromatic nitriles, such as benzonitrile, were converted even more slowly (270 times) than phenylpropionitrile. The assumed natural substrate of AtNIT1, 3-indolylacetonitrile [4], was found Fig. 3. Determination of the temperature optimum of AtNIT1. Table 2. Eect of antioxidants and protease inhibitors on enzyme activity. Enzyme activity was determined after incubation of 5 mL crude enzyme extract in Tris/HCl buer (70 m M , pH 8.0) with the respective antioxidant (neat) or a st ock s olution of p rotease inhibitors (50-fold concentration; Protease-Inhibitor-Set, Boehringer-Mann- heim). T he reaction was carried out fo r 48 h at room t emperature with vigorous stirring. The initial activity of 112 UáL )1 is 100% in the case of antioxidants and 97 UáL )1 in the case of inhibitors. Reagent Concn (m M ) Relative activity (%) Mercaptoethanol a 162 264 554 Dithiothreitol a 161 280 578 Aminophenylmethanesulfonyl ¯uoride b 176 523 EDTA b 188 296 10 94 100 97 a Activity of the reference, 37%; b activity of the reference, 82%. Fig. 2. Experimentally determined rates for the hydrolysis of 3-ind- olylacetonitrile (100 mg proteináL )1 ) (A) an d 3 -phenylpropionitrile (0.3 mg protein áL )1 ) (B) plotted against initial substrate c oncentrations. A Lineweaver±Burk plot of the data was used to calculate K m values: 3.67 m M for 3-indolylacetonitrile and 0.159 m M for 3-phenyl- propionitrile. Fig. 4. Determination of the pH optimum of AtNIT1 using dierent buer systems. (j) Tris/HCl; (m)glycine/NaOH;(.)KH 2 PO 4 / K 2 HPO 4 . Ó FEBS 2002 Synthetic applications of recombinant AtNIT1 (Eur. J. Biochem. 269) 683 to be a poor substrate (Table 3). Also the hydrolytic rate of cinnamonitrile, an a,b-unsaturated system, is signi®cantly diminished compared with the corresponding saturated phenylpropionitrile. A double bond in th e b,c-position, however, has almost no effect on enzyme activity, as can be seen if 4-phenyl-3-butenenitrile is compared with 4-phenyl- butyronitrile. Suitability as a substrate is strongly in¯ue nced by the substituents in the 2-position. All substituents other than ¯uoro inhibit enzymatic hydrolysis almost completely (Table 3). Nitriles with substituents in the 3-position, for example 3-methylbutyronitrile, a re also poor substrates, b ut the decrease in the hydrolytic rate is less pronounced. Interestingly, benzoylglycine nitrile is a much better substrate for AtNIT1 than glycine nitrile itself. Acid amides as byproducts of AtNIT1-catalyzed nitrile hydrolysis Acid amide was ®rst detected as a major product of AtNIT1-catalyzed nitrile hydrolysis with fumaronitrile as substrate. In this reaction, which was followed by gas chromatography, less than 10% of the expected amount of 3-cyanoacrylic acid, estimated from the calibration, was formed. A product mixture of an unidenti®ed product and 3-cyanoacrylic acid in the ratio 93 : 7 (Table 4) was found by HPLC. As demonstrated by co-injection, fumaric acid was not formed in the r eaction. After extractive separation from 3-cyanoacrylic acid and subsequent recrystallization from chloroform, the unknown product was isolated in 68% yield and unambiguously characterized as 3-cyano- acrylamide by elemental analysis and NMR spectroscopy. The a mide to acid ratio was independent of conversion. The isolated amide was not hydrolyzed to 3-cyanoacrylic acid under these reaction conditions. Moreover, the hydrolytic rate and r atio of acid to amide did not depend on enzyme purity, giving similar results with both crude extract and highly puri®ed enzyme. From blank experiments, it could also be excluded that it was an impurity of nitrile hydratase. As 3-cyanoacrylamide h ad been identi®ed as a byproduct of fumaronitrile hydrolysis, the AtNIT1-catalyzed hydro- lysis of other substrates with donor and acceptor substi- tuents was investigated with respect to amide formation Table 4. Product distribution of amide and acid in the AtNIT1-cata- lyzed nitrile hydrolysis. The reactions were performed as described (Table 3) in Tris/HCl buer (70 m M , containing dithiothreitol and EDTA, 2 m M each, pH 8) a t r oom temperature with 10 m M substrate concentration. The samples were analyzed by gas chromatography and/or HPLC (RP C 18 Vertex column; 4 ´ 250 mm; Nucleosil 100; 5 lm; Knaur ( ); ¯ow rate 1 mLámin )1 ; 220 nm detection wavelength). Relative activities are referred to the speci®c n itrilase activity t owards butyronitrile [8.931 lmolámin )1 á(mg protein) )1  100% at pH 8.0; 1.736 lmolámin )1 á(mg protein) )1  100% at pH 6.0]. Substrate Relative activity (%) Product distribution (amide : acid) a Relative activity referred to the E-isomer. b Conversion at pH 6.0, relative activity referred to butyronitrile under these conditions. Table 3. Relative activities of recombinant AtNIT1-catalyzed hydrolysis of nitriles. The reactions were performed in Tris/HCl buer (7 0 m M ,with dithiothreitol and EDTA, 2 m M each, pH 8) at room temperatu re. At a concen tratio n of 1.25 m M , all substrates were completely soluble; the enzyme concentration was varied so that the reaction time for all substrates w as in the range 2±4 h for c onversion of 15± 40%. Relative activities are referred to the speci®c nitrilase activity towards butyronitrile [1.393 lmo lámin )1 á(mg protein) )1  100%]. Substrate Relative activity (%) Substrate Relative activity (%) Butyronitrile 100 2-Methylbutyronitrile < 0.01 b Octanenitrile 291 2-Fluoropentanenitrile 131 3-Indolylacetonitrile a 2.2 2-Phenylpropionitrile < 0.01 b Benzonitrile 2.7 3-Methylbutyronitrile 4.0 Benzyl cyanide a 31 Cyclopropylacetonitrile 15 3-Phenylpropionitrile a 729 2-Hydroxypentanenitrile 0.2 c 4-Phenylbutyronitrile a 154 Glycine nitrile 0.4 Cinnamonitrile 48 2-Amino-4-methylpentanenitrile < 0.03 b 4-Phenylbut-3-enenitrile 188 Benzoylglycine nitrile 65 a See also literature data [9]. b 24 h reaction time. c Hydrolysis at pH 7.0. 684 S. Osswald et al.(Eur. J. Biochem. 269 ) Ó FEBS 2002 (Table 4). Amides have also been found as major products in the AtNIT1-catalyzed hydrolysis of a-¯uoroarylaceto- nitriles [15]. An a-¯uoro substituent, however, does not conclusively result in amide formation as can be seen in the hydrolysis of a-¯uorobutyronitrile, y ielding 95% of the corresponding acid (Table 4). Nevertheless, both a ¯uoro substituent in the a position and a second nitrile group conjugated to the nitrile (fumaronitrile) seem to play a decisive role in amide formatio n. The assumption that electron-withdrawing substituents favor the formation of amides was supported by the hydrolysis of differently 3-substituted acrylonitriles (Table 4). Whereas 3-nitroacrylonitrile was hydrolyzed to 3-nitroacrylamide as sole product, in the case of the donor- substituted 3-methoxyacrylonitrile and crotononitrile, the corresponding acids were formed almost quantitatively (Table 4). As 3-nitroacrylonitrile tends to decompose under basic conditions, the reaction was performed at pH 6 (Table 4), where the amide was formed only by enzyme- catalyzed hydrolysis and not by chemical reaction, as con®rmed by a blank experiment. Table 4 reveals that the relative activity is almost completely independent of the kind of substituent. DISCUSSION Substrate range Analysis of the substrate range w ith a var iety of structurally different aromatic and aliphatic nitriles revealed that aliphatic nitriles are hydrolyzed more ef®ciently than the natural substrate IAN or structurally related aromatic nitriles. With a relative a ctivity of only 2.2%, compared with butyronitrile, 3-indolylacetonitrile is a poorer substrate for AtNIT1 than benzyl cyanide (31% relative activity). This ®nding is in agreement with literature data, showing that IAN is one o f the weakest substrates [9]. The order of the relative AtNIT1 activity towards the substrates 3-phenyl- propionitrile, 4-phenylbutyronitrile and benzyl cyanide (Table 3) also corresponds to that just recently reported [9]. 2-Substituted substrates such as 2-methylbutyronitrile and 2 -phenylpropionitrile, however, w ere almost c ompletely unacceptable for AtNIT1, indicating that substituents in the 2-position, other than ¯uoro, inhibit the hydrolysis. The broad substrate range observed for AtNIT1 in this study is in good agreement with reports showing that AtNIT1 acts on a variety of aliphatic a nd aromatic substrates [4,9] and is in contrast with the high speci®city of AtNIT4 for b-cyano- ( L )-alanine [7]. Its broad substrate range, recombinant accessibility and reasonable stability make AtNIT1 a promising c andidate for applications in organic chemistry, in particular the synthesis of optically active 2-¯uorocarb- oxylic acids, which are very useful as analogs o f pheromones and antirheumatics, for example [15]. Also the mono- hydrolysis of aliphatic dinitriles to monocarboxylic acids is of great industrial interest because selective chemical hydrolysis is virtually impossible [16]. Amide formation The formation of amides as byproducts of nitrilase- catalyzed reactions was ®rst reported as early as 1964 [28,29]. Furthermore, in subsequent publications [19,30± 32], small amounts of amides (< 15%) could be detected besides the carboxylic acids during nitrilase catalysis. In all cases, the amide to acid ratio was indepen dent of reaction conditions (temperature and pH) and the applied enzyme concentrations. In their basic work on the four A. thaliana nitrilases NIT1±4, Bartel & Fink [5] described the conversion of IAN into 3-indolylacetic acid and indole- 3-acetamide and found that the latter is not a substrate for these enzymes. For the hydrolysis of b-cyano- ( L )-alanine, catalyzed by NIT4, Piotrowski et al.[7] reported the simultaneous formation of asparagine and aspartic acid in a ratio of 1.5 : 1, independent of reaction conditions. A dependence of the amide to acid ratio on the substituents, however, has not been reported in the literature so far. Until now the reaction mechanism of n itrilase-catalyzed hydrolysis has not been con®rmed experimentally. The mechanism postulated [19,28,29,33] involves the donation of a cysteine from the enzyme to the nitrile group to yield a thioimidate, which s ubsequently forms a tetrahedral inter- mediate A by addition of water. Generally, NH 3 is eliminated from this intermediate A to give a thioester, which reacts with a further w ater molecule to give the carboxylic acid (Fig. 5). Therefore, the formation of the acid amide from A logically arises from the elimination of cysteine. It has been shown for the chemical hydrolysis o f thioimidate esters [34,35] that the formation of thiol ester is favored in acidic medium (pH < 2.7), whereas at higher pH values (pH > 2.7) the formation o f a mide d ominates. This result was explained by a facilitated elimination of NH 3 caused by protonation of the a mino group in the tetrahedral intermediate. Although, a s m entioned, some p apers have dealt with t he mechanism of the nitrilase-catalyzed hydrolysis of nitriles, a relationship between the chemical structure of the substrate and t he amount of acid amide formation has not so far b een described. For A tNIT1-catalyzed nitrile hydrolysis, w e could demonstrate for the ®rst time such a structural Fig. 5. Postulated mechanism for acid amide formation in the n itrilase- catalyzed hydrolysis of nitriles. Ó FEBS 2002 Synthetic applications of recombinant AtNIT1 (Eur. J. Biochem. 269) 685 relationship, because the amide formation clearly d epends on the kind of substituent. The preference for acid amide formation by a-¯uoro substituents or by acceptor groups (CN, NO 2 )inp-conjugated nitriles is clear evidence of an electronically preferred formation and stabilization of the tetrahedral intermediate A in the enzyme±substrate com- plex. Becau se the crystal stru cture o f the active site of AtNIT1 is not known, how the stabilization of the tetrahedral intermediate assists the elimination of cysteine to yield the acid amide cannot be explained. CONCLUSIONS Chemical hyd rolysis of many nitriles with labile substituents catalyzed by acid or base is virtually impossible because of the drastic reaction conditions required. Therefore, over the last few y ears, b iocatalysts capable of hydrolyzing nitriles to carboxylic acids have been intensively investigate d [36]. In most cases, however, nitrile hydratase±amidase s ystems have been described, although not exclusively [36]. The nitrilase AtNIT1 from A. thaliana is the ®rst plant nitrilase to be investigated with respect to its synthetic potential. Because of optimized expression, the enzyme is now accessible in suf®cient quantities. Clear optimization of enzyme stability under the reaction conditions, which is important for practical a pplication, could be achieved b y addition of the protease inhibitor EDTA (Table 2). Therefore, slowly reacting nitriles can also be hydrolyzed without problem. The most important criteria for practical applications, however, are the substrate range and selectivity of an enzyme. In contrast with other nitrile-hydrolyzing enzymes, the nitrilase AtNIT1 stands out as having a very broad substrate range (Table 3). Although longer-chain aliphatic nitriles are the most effective substrates, hydrolysis of aromatic nitriles is also catalyzed. Because of the clearly improved enzyme stability, AtNIT1-catalyzed hydrolysis is also applicable to aromatic nitriles. Moreover, AtNIT1 shows a very interesting stereoselectivity and chemoselec- tivity. The in¯uence of substituents in the a position to the nitrile function has already been mentioned. Because the enzyme does not accept any substituents at the a position except ¯uoro, compounds bearing several different cyano groups can be selectively hydrolyzed. Hydrolysis of racemic 2-¯uoroarylacetonitriles proceeds enantioselectively [15]. In dinitriles with chemically comparable cyano groups (e.g. adiponitrile), only one cyano group is hydrolyzed exclu- sively to give the corresponding cyanocarboxylic acids [16,31,37], opening up interesting possibilities for organic synthesis, for example the preparation of certain lactams [31]. Furthermore, A tNIT1 e xhibits cis /trans selectivity w ith a,b-unsaturated nitriles [38], as also reported for other enzymes [19,39,40]. Because of i ts broad s ubstrate r ange on the one hand and unusual regioselectivities and stereoselectivities, the nitrilase AtNIT1 from A. thaliana is a very interesting biocatalyst in organic synthesis. ACKNOWLEDGEMENTS This work was generously supported by the Fonds der Chemischen Industrie. We acknowledge Dr K. Trummler for assistance in enzyme puri®cation, Dr S. Fo È rster for fermentation, and Dr A. Baro for preparing the manuscript. REFERENCES 1. Legras, J.I., Chuzel, G ., Arnaud, A. & Galzy, P. (1990) Natural nitriles and their metabolism. World J. Microbiol. Biotechnol. 6, 83±108. 2. Faber, K. (1995) Biotransformations in Organic Chemistry,2nd edn. Springer, Berlin . 3. Drauz, K. & Waldmann, H. (1995) Enzyme Catalysis in Organic Synthesis. Verlag Chemie, Weinheim. 4. Bartling, D., Seedorf, M., Mitho È fer, A. & Weiler, E.W. (1992) Cloning and expression of an Arabidopsis nitrilase which can convert indole-3-acetonitrile to t he plant hormone, indole-3-acetic acid. Eur. J. Biochem. 205, 417±424. 5. Bartel,B.&Fink,G.R.(1994)Dierentialregulationofanauxin- producing nitrilase gene family in Arabidopsis thaliana. Proc. Natl Acad. Sci. USA 91, 6649±6653. 6. Schmidt, R.C., Mu È ller, A., H ain, R., Bartling, D. & Weiler, E.W. (1996) Transgenic tobacco plants expressing the Arabidopsis thaliana nitrilase II enzyme. Plant J. 9, 683±691. 7. Piotrowski, M., Sc ho È nfelder, S. & Weiler, E.W. (2001) The Arabidopsis thaliana iso ge n e NIT4 and its orthologs in tobacco encode b-cyano- L -alanine hydratase/nitrilase. J. Biol. Chem. 276, 2616±2621. 8. Tsunoda, H. & Yamaguchi, K. (1995) The cDNA sequence o f an auxin-producing nitrilase homolog in t o bacco (GenBank D63331) (PGR 95-058). Plant Physiol. 109, 339. 9. Vorwerk,S.,Biernacki,S.,Hillebrand,H.,Janzik,I.,Mu È ller, A., Weiler, E.W. & P iotrowski, M. (2 001) Enzymatic c haracterization of the recombinant Arabidopsis thaliana nitrilase subfamily encoded by the NIT2/NIT1/NIT3-gene cluster. Planta 212, 508±516. 10. Kende, H . (1989) Enzymes o f ethylene biosynthesis. Plant Physiol. 91, 1±4. 11. Nagasawa, T., Shim izu, H. & Yamada, H . (1 993) The superiority of the third-generation c atalyst, Rhodococcus rhodochrous J1 nitrile hydratase, for industrial production of acrylamide. Appl. Microbiol. Biotechnol. 40, 189±195. 12. Kobayashi, M., Nagasawa, T. & Yamada, H. (1992) Enzymic synthesis of acrylamide: a success story not yet over. Trends Biotechnol. 10, 402±408. 13. Layh,N.,Stolz,A.,Bo È hme,J.,Eenberger,F.&Knackmuss,H J. (1994) Enantioselective hydrolysis of racemic naproxen nitrile and naproxen amide to S-naproxen by new bacterial isolates. J. Biotechnol. 33, 175±182. 14. Eenberger, F. & Graef, B.W. (1998) Chemo- and enantioselective hydrolysis of nitriles and acid amides, respectively, with re sting cells of Rhodococcus sp. C3II and Rhodococcus erythropolis MP50. J. Biotechnol. 60, 165±174. 15. Eenberger, F. & Oûwald, S. (2001) Enantioselective hydrolysis of (RS)-2-¯uoroarylacetonitriles using nitrilase from Arabidopsis thaliana. Tetrahedron: Asymmetry 12, 279±285. 16.Eenberger,F.&Oûwald,S.(2001)Selectivehydrolysisof aliphatic dinitriles to monocarboxylic acids by a nitrilase from Arabidopsis thaliana. Synthesis 1866±1872. 17. Harper, D.B. (1977) Microbial metabolism of aromatic nitriles. Enzymology of C±N cleavage by Nocardia sp. (Rhodococcus group) N.C.I.B. 11216. Biochem. J. 165, 309±319. 18. Bhalla, T.C., Miura, A., Wakamoto, A., Ohba, Y. & Furuhashi, K. (1992) Asymmetric hydrolysis of a-aminonitriles to optically active amino acids by a nitrilase of Rhodococcus rhodochrous PA-34. Appl. Microbiol. Biotechnol. 37, 184±190. 19. Stevenson,D.E.,Feng,R.,Dumas,F.,Groleau,D.,Mihoc,A.& Storer, A.C. (1992) Mechanistic and structural studies on Rhodococcus ATCC 39484 nitrilase. Biotechnol. Appl. Biochem. 15, 283±302. 20. Kobayashi, M., Yanaka, N ., Nagasawa, T. & Y amada, H. (1990) Puri®cation and characterization of a novel nitrilase of 686 S. Osswald et al.(Eur. J. Biochem. 269 ) Ó FEBS 2002 Rhodococcus rhodochrous K22 that acts on aliphatic nitriles. J. Bacteriol. 172, 4807±4815. 21. Bestwick, L.A., Gronning, L.M., James, D.C., Bones, A. & Rossiter, J.T. (1993) Puri®cation and characterization of a nitrilase from Brassica napus. Physiol. Plant. 89, 811±816. 22. Nagasawa, T., Mauger, J. & Yamada, H. (1990) A novel nitril ase, arylacetonitrilase, of Alcaligenes faecalis JM3: puri®cation and characterization. Eur. J. Biochem. 194, 765±772. 23. Kobayashi, M., Nagasawa, T. & Yamada, H . (1989) Nitrilase of Rhodococcus rhodochrous J1. Puri®cation and characterization. Eur. J. Bioc hem. 182, 349±356. 24. Bandyopadhyay, A.K ., Nagasawa, T., Asano, Y., Fujishiro, K., Tani, Y. & Yamada, H. (1986) Puri®cation and characterization of benzonitrilases from Arthrobacter sp. strain J -1. Appl. Environ. Microbiol. 51, 302±306. 25. Harper, D.B. (1977) Fungal degradation of aromatic nitriles. Enzymology of C-N cleavage by Fusarium so lani. Biochem. J. 167, 685±692. 26. Stalker, D.M., Malyj, L.D. & McBride, K.F. (1988) Puri®cation and properties of a nitrilase speci®c for the herbicide bromoxynil and corresponding nucle otide sequence analysis of the bxn gene. J. Biol. Chem. 263, 6310±6314. 27. Robinson, W. & Hook, R. (1964) Ricinine nitrilase. I. Reaction product and substrate s peci®city. J. Biol. Chem. 239, 4257±4262. 28. Hook, R. & Robinson, W. (1964) Ricine nitrilase. II. Puri®cation and properties. J. Biol. Che m. 239, 4263±4267. 29. Thimann, K.V. & Mahadevan, S. (1964) Nitrilase I. Occurrence, preparation, and general properties of the enzym e. Arch. B iochem. Biophys. 105, 133±141. 30. Layh, N ., Stolz, A., Fo È rster, S., Eenberger, F. & Knackmuss, H J. (1992) Enantioselective hydrolysis of O-acetylmandelonitrile to O-acetylmandelic acid by bacterial nitrilases. Arch. Microbiol. 158, 405±411. 31. Gavagan, J.E., Fager, S.K ., F allon, R.D., F olsom, P.W., Herkes, F.E., Eisenberg, A., Hann, E.C. & DiCosimo, R. (1998) Chemo- enzymic production of lactams from aliphatic a,x-dinitriles. J. Org. Chem. 63, 4792±4801. 32. Goldlust, A. & Bohak, Z. (1989) Induction, puri®cation, and characterization of the nitrilase of Fusarium oxysporum f. sp. melonis. Biotechnol. Appl. Biochem. 11, 581±601. 33. Dufour, E., Storer, A .C. & Me  nard, R. ( 1995) Engineering n itrile hydratase a ctivity into a cysteine protease by a single mutation. Biochemistry 34, 16382±16388. 34. Chaturvedi, R.K., MacMahon, A.E. & Schmir, G.L. (1967) The hydrolysis of thioimidate esters. Tetrahedral intermediates and general acid catalysis. J. Am. Chem. Soc. 89, 6984±6993. 35. Chaturvedi, R.K. & Schmir, G.L. (1969) The hydrolysis of thioimidate esters. II. Evidence for the formation of three species of the tetrahedral intermediate. J. Am. C hem. Soc . 91, 737±746. 36. Wieser, M. & Nagasawa, T. (2000) Stereoselective nitrile-con- verting enzymes. In Stereoselective Biocatalysis (Patel, R.N ., ed.), pp. 463±465. Marcel Dekker, New York. 37. Bengis-Garber, C. & Gutman, A.L. (1988) Bacteria in organic synthesis: se lective conversion o f 1,3-dicyanobenzene into 3-cyanobenzoic acid. Tetrahedron Lett. 29, 2589±2590. 38. Eenberger, F. & Oûwald, S. (2001) E-Selective hydrolysis of E,Z a,b-unsaturated nitrile s by the recombinant nitrilase A tN IT1 from Arabidopsis thaliana. Tetrahedron: Asymmetry 13, 2581± 2587. 39. Quiro  s, M., Astorga, C., Rebolledo, F. & Gotor, V. (1995) Enzymic selective transformations of diethyl fumarate. Tetra- hedron 51, 7715±7720. 40. Klibanov, A.M. & Siegel, E.H. (1982) Geometric speci®city of porcine liver carboxylesterase and its application for the produc- tion of cis-arylacrylic esters. Enzyme Microb. T echnol. 4, 172±175. Ó FEBS 2002 Synthetic applications of recombinant AtNIT1 (Eur. J. Biochem. 269) 687 . Characterization and synthetic applications of recombinant AtNIT1 from Arabidopsis thaliana Steffen Osswald, 1 Harald Wajant 2 and Franz Effenberger 1 1 Institut. expression of N-terminally His-tagged recombinant protein. In brief, the c oding region and p art of the 3¢-noncoding region of AtNIT1 cDNA were ampli®ed from