The allene oxide cyclase family of Arabidopsis thaliana – localization and cyclization Florian Schaller 1, * , Philipp Zerbe 1, *, Steffen Reinbothe 1 , Christiane Reinbothe 2 , Eckhard Hofmann 3 and Stephan Pollmann 1 1 Lehrstuhl fu ¨ r Pflanzenphysiologie, Ruhr-Universita ¨ t Bochum, Germany 2 Lehrstuhl fu ¨ r Pflanzenphysiologie, Universita ¨ t Bayreuth, Germany 3 Lehrstuhl fu ¨ r Biophysik, AG Ro ¨ ntgenstrukturanalyse an Proteinen, Ruhr-Universita ¨ t Bochum, Germany Since the initial discovery of methyl jasmonate as a secondary metabolite in essential oils of jasmine in 1962 [1], jasmonates have become accepted as a new class of plant hormones. In the early 1980s, their wide- spread occurrence throughout the plant kingdom [2] and their growth-inhibitory [3] and senescence-promot- ing activities [4] were established, and their route of biosynthesis was elucidated [5]. Research in recent years generally confirmed the Vick and Zimmerman pathway of jasmonic acid (JA) biosynthesis (the octadecanoid pathway) and brought considerable progress with respect to the biochemistry Keywords 12-oxo-phytodienoic acid; allene oxide cyclase; allene oxide synthase; oxylipins; signaling Correspondence S. Pollmann, Lehrstuhl fu ¨ r Pflanzenphysiologie, Ruhr-Universita ¨ t Bochum, Germany Fax: +49 234 32 14187 Tel: +49 234 32 24294 E-mail: stephan.pollmann@rub.de *These authors contributed equally to this work (Received 12 December 2007, revised 24 January 2008, accepted 10 March 2008) doi:10.1111/j.1742-4658.2008.06388.x Jasmonates are derived from oxygenated fatty acids (oxylipins) via the octadecanoid pathway and are characterized by a pentacyclic ring struc- ture. They have regulatory functions as signaling molecules in plant devel- opment and adaptation to environmental stress. Recently, we solved the structure of allene oxide cyclase 2 (AOC2) of Arabidopsis thaliana, which is, together with the other three AOCs, a key enzyme in the biosynthesis of jasmonates, in that it releases the first cyclic and biologically active metabolite – 12-oxo-phytodienoic acid (OPDA). On the basis of models for the bound substrate, 12,13(S)-epoxy-9(Z),11,15(Z)-octadecatrienoic acid, and the product, OPDA, we proposed that a conserved Glu promotes the reaction by anchimeric assistance. According to this hypothesis, the transition state with a pentadienyl carbocation and an oxyanion is stabi- lized by a strongly bound water molecule and favorable p–p interactions with aromatic residues in the cavity. Stereoselectivity results from steric restrictions to the necessary substrate isomerizations imposed by the pro- tein environment. Here, site-directed mutagenesis was used to explore and verify the proposed reaction mechanism. In a comparative analysis of the AOC family from A. thaliana involving enzymatic characterization, in vitro import, and transient expression of AOC–enhanced green fluorescent pro- tein fusion proteins for analysis of subcellular targeting, we demonstrate that all four AOC isoenzymes may contribute to jasmonate biosynthesis, as they are all located in chloroplasts and, in concert with the allene oxide synthase, they are all able to convert 13(S)-hydroperoxy-9(Z),11(E),15(Z)- octadecatrienoic acid into enantiomerically pure cis(+)-OPDA. Abbreviations 12,13-EOT, 12,13(S)-epoxy-9(Z),11,15(Z)-octadecatrienoic acid; AOC, allene oxide cyclase; AOS, allene oxide synthase; EGFP, enhanced green fluorescent protein; HPOD, 13(S)-hydroperoxy-9(Z),11(E)-octadecadienoic acid; HPOT, 13(S)-hydroperoxy-9(Z),11(E),15(Z)- octadecatrienoic acid; JA, jasmonic acid; OPC-8:0, 3-oxo-2[2¢(Z)-pentenyl]-cyclopentane-1-octanoic acid; OPDA, 12-oxo-phytodienoic acid; OPR, 12-oxo-phytodienoic acid reductase; RBCS, ribulose-1,5-bisphosphate carboxylase ⁄ oxygenase. 2428 FEBS Journal 275 (2008) 2428–2441 ª 2008 The Authors Journal compilation ª 2008 FEBS of the enzymes involved, as well as the molecular organization and regulation of the pathway [6,7]. Biosynthesis is believed to start with the oxygenation of a-linolenic acid, which is converted to 13(S)-hydro- peroxy-9(Z),11(E),15(Z)-octadecatrienoic acid (HPOT) in a reaction catalyzed by 13-lipoxygenase (Fig. 1). Allene oxide synthase (AOS) converts HPOT to the unstable epoxide 12,13(S)-epoxy-9(Z),11,15(Z)-octa- decatrienoic acid (12,13-EOT), which is cyclized by allene oxide cyclase (AOC) to give rise to the first cyc- lic and biologically active compound of the pathway, 12-oxo-phytodienoic acid (OPDA). Reduction of the 10,11-double bond by an NADPH-dependent OPDA reductase [8,9] then yields 3-oxo-2[2¢(Z)-pentenyl]- cyclopentane-1-octanoic acid (OPC-8:0), which under- goes three cycles of b-oxidation to yield the end product of the pathway, i.e. JA with a (3R,7S)-configu- ration [(+)-7-iso-JA] [10,11]. The biosynthesis of JA appears to involve two different compartments. The conversion of LA to OPDA is localized in chloroplasts [10–13], whereas the reduction of OPDA to OPC-8:0 [14,15] and the three steps of b-oxidation, i.e. conver- sion of OPC-8:0 to JA, occur in peroxisomes [11,15] (Fig. 1). OPDA transport into peroxisomes may be facilitated by ion trapping or via the ABC transporter CTS ⁄ PXA1 located in the peroxisomal membrane [16]. Whether OPDA is transported in its free acid form or activated by thioesterification to CoA, a pro- cess that may occur in the cytosol, the endoplasmic reticulum, or in the peroxisome, triggered by the activity of specific acetyl-CoA synthetases, is yet to be demonstrated. AOC, a soluble enzyme in corn [17], was biochemi- cally purified as an apparent dimer of 47 kDa from maize kernels [18] and characterized with respect to its substrate specificity. In contrast to AOS, which produces both allene oxides from the respective 13(S)-hydroperoxy fatty acids (18:3 and 18:2, respec- tively), the enzyme accepted 12,13(S)-epoxylinolenic acid but not 12,13(S)-epoxylinoleic acid as a substrate [19]. It thus appeared that AOC confers additional specificity to the octadecanoid biosynthetic pathway. AOC has been cloned as a single-copy gene from tomato [20] and barley [21] and as a small gene family (AOC1–4) from Arabidopsis thaliana [22]. Recently, we solved the structure of A. thaliana AOC2 trimers [23] and proposed a mechanism for the enzymatically catalyzed cyclization reaction (Fig. 2). In the AOC2 crystals, the competitive inhibitor (±)cis- 12,13-epoxy-9(Z)-octadecenoic acid (vernolic acid) was found to be positioned inside the barrel of one mono- mer. No induced fit mechanism could be observed. On the basis of the resulting arrangement of protein side chains around the bound substrate analog, the following reaction mechanism for the cyclization reac- tion inside the protein was postulated (Fig. 2). Subse- quent to the positioning of 12,13-EOT inside the binding pocket of AOC2, Glu23 introduces a negative charge and thus leads to delocalization of the C15 Fig. 1. JA biosynthesis in A. thaliana. LOX2, 13-lipoxygenase OPR3, 12-oxo-phytodienoate reductase 3; ACS, acyl-CoA syn- thetase; ACX, acyl-CoA oxidase; MFP, multifunctional protein; KAT, L-3-ketoacyl-CoA thiolase; CTS ⁄ PXA1, ABC transporter for OPDA or OPDA-CoA import. F. Schaller et al. Oxylipin cyclization FEBS Journal 275 (2008) 2428–2441 ª 2008 The Authors Journal compilation ª 2008 FEBS 2429 double bond. The delocalization of the positive charge between C13 and C16 promotes the opening of the oxirane (Fig. 2A) and the cyclization reaction by the mechanism of anchimeric assistance. The formed oxyanion is stabilized by a water molecule (water75), which is tightly bound and positioned by the forma- tion of hydrogen bonds with the protein environment (Fig. 2B). Asn25, Asn53 and Ser31, together with the main chain nitrogen of Pro32, build a polar patch that stabilizes the water molecule in its position. To facili- tate the subsequent pericyclic ring closure, a trans–cis isomerization around the C10–C11 bond is necessary, resulting in the formation of a nonplanar ring-like pentadienyl carbocation. This carbocation might be stabilized by p–p interaction with the aromatic elec- trons of Phe51 or by Cys71. Owing to steric restric- tions inside the protein cavity, the conformational change around the C10–C11 bond from trans to cis geometry has to be accompanied by a cis–trans rota- tion around the C8–C9 bond. The conformational change is further promoted by a hydrophobic effect, as it buries more of the hydrocarbon tail of the molecule inside the cavity. In contrast to the spontaneous cyclization of 12,13- EOT, which is likely to proceed via dipolar ring closure [24], the situation is not yet clear for the enzymatic cyclization reaction. Here, a classic conro- tary pericyclic ring closure, according to the Wood- ward–Hoffmann rules, seems to be more favorable. The absolute stereoselectivity of the reaction is steri- cally controlled by the protein environment and allows only the formation of the cis intermediate. A rotation in the opposite direction would automatically result in the opposite stereoisomer of the product. Phe43, Val45, Phe85 and Tyr105, in particular, seem to form a greasy slide that, at the same time, facili- tates and restricts the conformational change of the hydrocarbon tail. Both the spontaneous cyclization of allene oxides [25] and the AOC-catalyzed cyclization share a require- ment for the C15 double bond in the supposed anchi- meric assistance mechanism. This also explains the impact of the protein environment on the stereoselec- tivity of the cyclization reaction. As a result of the steric restrictions in the cavity, changes in the position of the epoxide group will affect the binding affinities of possible substrates. In the present study, we scrutinized the reaction mechanism proposed for AOC2, using site-directed mutagenesis and biochemical characterization of the mutant proteins. Moreover, we compared all four AOC isoenzymes of A. thaliana with respect to their specific activity and subcellular localization. Fig. 2. Schematic overview of the proposed cyclization reaction of A. thaliana AOC2. The AOC2-catalyzed cyclization reaction involves the opening of the oxirane ring and a subsequent conrotatory peri- cyclic ring closure (see text for details). The fatty acid moiety of 12,13-EOT [(CH 2 ) 7 COOH] is labeled as R, and possibly involved amino acids are presented in the one-letter code following the numbering of the recombinant protein (see Experimental proce- dures). Adapted from [48]. Oxylipin cyclization F. Schaller et al. 2430 FEBS Journal 275 (2008) 2428–2441 ª 2008 The Authors Journal compilation ª 2008 FEBS Results Overexpression, purification, and enzymatic characterization of AOC1–AOC4 Because of relatively weak expression levels of published constructs (C. Wasternack, unpublished results), we cloned differently N-terminally truncated versions of AOC2 in pET21b(+) (Novagen ⁄ Merck, Nottingham, UK; C-terminal His-tag) and pQE30 (Qiagen, Hilden, Germany; N-terminal His-tag), respectively, and ana- lyzed the expression levels of insoluble and soluble AOC2. Both N-terminally and C-terminally His 6 -tagged truncated versions of AOC2, beginning at amino acid 78 of the preprotein in the pET21b and the pQE30 vectors, showed the highest expression of soluble protein (data not shown). AOC1, AOC3 and AOC4 were expressed as N-terminally His-tagged proteins. All AOC isoforms could be easily purified via affinity chromatography on Ni–nitrilotriacetic acid agarose and were characterized biochemically (Fig. 3A). In a first experiment, the neces- sity of the C15 double bond for the enzymatic cycliza- tion was analyzed in a coupled assay that included recombinant Arabidopsis AOS [26]. Using HPOT as the substrate, AOS produced 12,13-EOT, which, in the absence of AOC activity, cyclized spontaneously [27]. When HPOD [13(S)-hydroperoxy-9(Z),11(E)-octadeca- dienoic acid], lacking the C15 double bond, was used as the substrate for AOS, only trace amounts of cyclopen- tenones could be detected, indicating that 12,13-EOD does not readily undergo spontaneous cyclization [24]. Also, in the coupled assay of AOS with the four AOC isoforms, no cyclization of 12,13-EOD was observed (data not shown). In contrast, OPDA produc- tion was detectable, and a shift in the enantiomeric composition towards the (+)-enantiomer occurred when HPOT was used as substrate (Fig. 3B). Furthermore, an increase in OPDA formation and stereoselectivity towards the cis(+)-enantiomer occurred when methylated HPOT was used as sub- strate (Fig. 3B); this effect was most probably caused by an improved positioning of the methylated allene oxide within the hydrophobic barrel of the enzyme, leading to a larger amount of substrate for the AOC reaction. Thus, the carboxylic group of HPOT plays no essential role in the cyclization reaction. Cellular distribution of AOS and AOC1–AOC4 All four AOCs of A. thaliana, as well as the Arabidop- sis AOS, are predicted to be localized in the chloro- plast and to contain appropriate N-terminal transit peptides (predicted by chlorop [28]). In previously performed immunocytochemical analyses [22], an anti- body capable of detecting AOC1–AOC4, although detecting AOC2 preferentially over the other isoen- zymes, proved the plastid localization of AOC. To val- idate these data and to investigate the subcellular localization specifically for each of the four AOC A B Fig. 3. Substrate specificity of A. thaliana AOC1–AOC4. (A) Coo- massie-stained SDS ⁄ PAGE (I) and western blot analysis using a monoclonal a-(His) 5 -antibody (II) of affinity-purified AOCs. (1) AOC1, (2) AOC2, (3) AOC3, and (4) AOC4 (B) Ten micrograms of AOS and 5 lg of AOC were incubated in 10 m M PP i buffer (pH 7.0) for 15 min with 100 lg of HPOT (left bars) and methylated HPOT (right bars), respectively. The enzymatic reaction was stopped by acidifi- cation and extraction with ethyl acetate. Product formation was quantified via chiral GC-MS, and is given as relative activity as com- pared to the yield with AOC2. (1) AOS; (2) AOS and AOC1; (3) AOS and AOC2; (4) AOS and AOC3; (5) AOS and AOC4. cis-(+)- OPDA and cis-())-OPDA are indicated by white and gray bars, respectively. F. Schaller et al. Oxylipin cyclization FEBS Journal 275 (2008) 2428–2441 ª 2008 The Authors Journal compilation ª 2008 FEBS 2431 isoforms, we carried out in vitro import studies in con- junction with transient expression assays using enhanced green fluorescent protein (EGFP) technol- ogy. First, 35 S-labeled precursor (p)AOC1–pAOC4 were synthesized from corresponding cDNAs and imported into isolated Arabidopsis leaf mesophyll chlo- roplasts. Figure 4 shows that [ 35 S]pAOC2 and all of the other 35 S-labeled AOC isoforms were faithfully taken up by chloroplasts and processed to mature size. The lack of salt extractability of the imported proteins revealed a membrane localization in all four cases (Fig. 4E–G). Similarly, 35 S-labeled precursor AOS was also imported into leaf mesophyll chloroplasts and processed to mature size, and accumulated in a salt-resistant form in total membranes (Fig. 4A–D). Second, DNAs encoding full-length AOC1–AOC4 precursors fused to EGFP were transformed into Arabidopsis leaf epidermis cells and coexpressed with a construct encoding the first 32 amino acids of the small subunit of ribulose-1,5-bisphosphate carboxyl- ase ⁄ oxygenase (RBCS) fused to the DsRed protein, selected as an internal control. Transient expression and subsequent confocal laser scanning microscopy clearly showed a plastid colocalization of all four AOC–EGFP fusion proteins as well as of an AOS– EGFP chimeric protein with the RBCS–DsRed protein (Fig. 5). Together, these results confirmed and extended findings for AOS (AF230372) of tomato [29] as well as AOC3 and AOC4 [7] of Arabidopsis, and showed that plastids have the ability to import all AOC preproteins (pAOC1–pAOC4) as well as pAOS. Substrate-binding site and biochemical analysis of mutated AOC2 The barrel part of AOC2 forms an elongated cavity that is lined mostly by aromatic and hydrophobic resi- dues and reaches about 14 A ˚ into the protein [23]. In particular, Val49, Phe43, Phe51, Phe85 and Tyr105 are part of this interior hydrophobic pocket, forming a greasy slide, and these residues, at the same time, impose the necessary conformational specificity to the hydrocarbon tail (Fig. 6). In addition to these hydro- phobic residues, the conserved Glu23 is positioned at the very bottom of the cavity, introducing a negative charge to initiate the opening of the oxirane ring, and the subsequent formation of the classic pentadienyl cation after substrate binding. An additional polar patch is formed by Ser31, Asn25 and Asn53 on one side of the cavity. These three residues, together with the main chain nitrogen from Pro32, are in appropri- ate positions for coordinating a tightly bound water molecule, which is found in all solved AOC2 struc- tures. At last we find Cys71 on the opposite wall of the cavity, also being in a favorable position to stabi- lize the pentadienyl cation. These residues are strictly conserved among all AOC sequences in the EBI UNIRef100 database. A B C D E F G Fig. 4. In vitro plastid import of AOS and AOC1-4. 35 S[Met]-labeled pAOS and pAOC1–pAOC4 were synthesized from respective cDNAs by coupled transcription ⁄ translation in a wheat germ lysate and imported into isolated Arabidopsis leaf mesophyll chloroplasts. After 15 min, intact plastids were reisolated on Percoll. Unimported precursors were degraded by thermolysin (Thl), and plastid protein was extracted and resolved by 10–20% polyacrylamide gradient gels containing SDS. (A) Precursor (P) and mature (m) AOS levels in intact chloroplasts; TP, translation product. (B) Salt extraction of [ 35 S]AOS from total membranes recovered after import from lysed chloroplasts with either 1 M NaCl or 0.1 M Na 2 CO 3 (pH 11), followed by centrifugation of the assays to obtain respective membrane (M) and supernatant (S) fractions. (C,D) Time course of [ 35 S]pAOC2 plastid import. P, precursor AOC2; m, mature AOC2. (E, F) Plastid import of [ 35 S]pAOC1, [ 35 S]pAOC2, [ 35 S]pAOC3 and [ 35 S]pAOC4. Precursor (P) and mature (m) AOC1–AOC4 protein lev- els are shown for reisolated, intact chloroplasts treated with (+) or without ()) Thl. (G) Salt extraction of total membranes containing imported [ 35 S]AOC1–AOC4 with 0.1 M Na 2 CO 3 (pH 11). After treat- ment, the assays were centrifuged. AOC1–AOC4 were detected in the resulting membrane and supernatant (Sups) fractions by SDS ⁄ PAGE and autoradiography. Oxylipin cyclization F. Schaller et al. 2432 FEBS Journal 275 (2008) 2428–2441 ª 2008 The Authors Journal compilation ª 2008 FEBS Fig. 5. Detection of chimeric fluorophores by confocal laser scanning microscopy. Chimeric fusion proteins were transformed into A. thaliana using biolistic transformation. Left row: AOC–EGFP fluorescence (500–530 nm). Middle row: RBCS–DsRed fluorescence (575–605 nm). Right row: superimposition of the GFP channel and the DsRed channel. F. Schaller et al. Oxylipin cyclization FEBS Journal 275 (2008) 2428–2441 ª 2008 The Authors Journal compilation ª 2008 FEBS 2433 The contribution of Phe85 to substrate positioning has been shown previously [23]. Substitution of Phe85 with either Ala or Leu resulted in moderately reduced reactivity and stereoselectivity of the reaction (Table 1 [23]). To further characterize the importance of the hydrophobic environment inside the binding pocket of AOC2, and to assess the role of the amino acids forming the polar patch, we generated point mutants of AOC2 and analyzed their ability to catalyze the cyclization reaction. Except for P32V and N53L, all of the different mutants could be affinity purified to homogeneity in similar amounts as the wild-type. CD spectroscopic analyses excluded overall folding defects (supplementary Figs S1 and S2). Any changes in enzymatic activity are thus caused by the effects of the single amino acid substitutions. With respect to the importance of the hydrophobic protein environ- ment inside the barrel, we analyzed the relevance of Val49, Phe43 and Phe51, in addition to the previously characterized Phe85. Single point mutations of Phe43 to Tyr, Val49 to Phe and Phe51 to Ala did not reduce AOC2 activity. In contrast, the V49F muta- tion augmented the enzymatic activity of the protein (Table 1). Mutation of Glu23 to Ala resulted in a complete loss of the enzymatic activity of AOC2 (Table 1). As the CD spectra of the mutant did not show significant alterations as compared to the wild-type protein, loss of function can be attributed specifically to the point mutation. The data support the proposed role for Glu23, which is to introduce a negative charge into the binding pocket, leading to a delocalization of the C15 double bond towards the oxirane, which promotes its opening and the cyclization reaction by the mechanism of anchimeric assistance (Fig. 2A). According to the proposed reaction scheme [23], the oxyanion in the transition state is stabilized by a water molecule (water75) (Fig. 2B). This water molecule is assumed to be tightly bound by either Ser31, Asn25, or Asn53, together with Pro32 [23]. Mutational experi- ments on these four amino acids clearly demonstrate that Pro32 and Asn25, which were substituted by Val and Leu, respectively, are essential for the enzymatic activity of AOC2. Ser31 (S31A) and Asn53 (N53L), on Fig. 6. Ligplot sketch showing the molecular interactions of AOC2 and the bound inhibitor vernolic acid. Contacts with the conserved water, water75, are shown as blue dashed lines with the corre- sponding distances (A ˚ ); hydrophobic contacts are represented by opposing gray spoked arcs. The influence of the negative charge of Glu23 is represented as a red arrow. Adapted from [23]. Table 1. Analysis of point mutants of AOC2 of A. thaliana. The results are presented as relative amount of total OPDA formed, and the specific formation of the cis(+)-enantiomer as compared to the yield with AOC2. Standard deviations have been calculated on the basis of three independent triplicate measurements. Discrepan- cies in the CD spectropolarimetric analyses as compared to the wild-type are marked as ‘)’. Consistency between wild-type and mutant CD spectra are indicated by ‘+’. Protein Purity (%) Activity analysis CD spectra Total OPDA (%) cis-(+)-OPDA (%) AOS  90 28.8 ± 1.7 56.5 ± 1.3 + AOC1 > 90 92.9 ± 10.0 91.3 ± 0.2 + AOC2 > 90 100.0 ± 0.0 93.6 ± 0.8 + AOC3 > 90 90.2 ± 7.5 94.1 ± 0.9 + AOC4 > 90 94.6 ± 13.2 93.9 ± 0.9 + E23A > 90 29.0 ± 4.2 57.6 ± 3.0 + N25L  90 38.6 ± 1.8 63.4 ± 1.6 + L27R > 90 101.3 ± 0.2 92.8 ± 0.5 + S31A > 90 75.8 ± 2.3 71.8 ± 2.8 + P32V  90 61.0 ± 2.2 57.1 ± 2.6 ) F43Y > 90 90.4 ± 12.2 93.5 ± 2.1 + V49F > 90 130.4 ± 20.1 93.0 ± 2.3 + F51A > 90 128.4 ± 25.9 94.7 ± 0.8 + N53L Refolded 48.0 ± 5.4 83.4 ± 0.1 (+) C71S > 90 97.1 ± 13.4 93.9 ± 0.6 + C71Y > 90 33.3 ± 3.2 67.4 ± 1.4 + C71A > 90 97.7 ± 10.1 94.1 ± 0.4 + F85L > 90 79.6 ± 4.9 87.7 ± 6.5 + F85A  90 54.7 ± 6.0 71.7 ± 10.0 + S31A ⁄ N53L > 90 93.6 ± 11.6 84.9 ± 1.4 + Oxylipin cyclization F. Schaller et al. 2434 FEBS Journal 275 (2008) 2428–2441 ª 2008 The Authors Journal compilation ª 2008 FEBS the other hand, seemed to be of minor importance for AOC2 functionality (Table 1). As the N53L mutant could not be expressed as soluble protein, it was puri- fied from inclusion bodies and successfully refolded (supplementary Fig. S3). The S31A ⁄ N53L double mutant exhibited almost wild-type activity (Table 1), confirming that neither amino acid is involved in water stabilization. The mutation P32V resulted in complete loss of stereoselectivity of AOC2, whereas the residual activity of total racemic OPDA formation was still higher than in the AOS control reaction, reflecting AOC-independent, autocatalytic cyclization of 12,13- EOT. However, this protein could not be purified to apparent homogeneity, and the CD spectrum showed slight alterations relative to wild-type AOC2, so the absolute activity values have to be viewed with cau- tion. In the N25L mutant, stereoselectivity was influ- enced to a minor degree, but the capability for total OPDA formation was decreased as compared to the P32V mutant. This could be due to the different nature of the side chains of Asn and Leu and steric con- straints imposed by the two methyl groups of Leu on the substrate. The finding that the location of the epoxy group in the 12,13-position of vernolic acid, an inhibitor of AOC (see below), is essential for binding to AOC from corn [18] is consistent with the observed binding position in corn AOC [23]. Any shift of the epoxy group would inhibit the formation of the hydro- gen bond to the conserved water molecule. We next tested the effect of (±)cis-12,13-epoxy-9(Z)- octadecenoic acid (vernolic acid) on AOC1–AOC4 and the respective AOC2 mutant proteins. Intrigu- ingly, the total activity of wild-type AOC1 and AOC2 was inhibited by 40% by 0.64 lm vernolic acid, with a slight loss of stereoselectivity, whereas AOC3 and AOC4 were only marginally inhibited. The sensitivity of the AOC2 L27R mutant towards vernolic acid was not enhanced as compared to wild-type AOC2 (Table 2). Discussion Substrate-binding site and biochemical analysis of mutated AOC2 protein In the present study, a structure–function analysis was performed for AOC2 of A. thaliana. According to the proposed reaction mechanism of cyclization for the pericyclic ring closure, a conformational change around the C10–C11 bond from trans to cis geometry is necessary, producing a nonpolar ring-like pentadie- nyl carbocation (Fig. 2B). It was previously assumed that Phe51 would be positioned to stabilize this carbo- cation by p–p interaction. Here we have demonstrated that Phe51 is not involved in carbocation stabilization, because the F51A mutant shows wild-type-like AOC2 activity, which is reflected by slightly enhanced total OPDA formation and wild-type-like stereoselectivity. Alternatively, Cys71 had been postulated to be in a favorable position to stabilize the positive carbocation. We mutated Cys71 to Ala, Ser and Tyr to assess the importance of the sulfhydryl group of Cys. The muta- tion of Cys71 to Ser and Ala resulted in fully active proteins. Only the mutation to Tyr led to nearly com- plete loss of enzymatic activity, which is most likely explained by steric hindrance invoked by the bulky phenyl ring system. In the proposed catalytic cavity, Phe51, in conjunc- tion with Phe43, Phe85 and Val49, is supposed to form a greasy slide that helps to coordinate the hydrocarbon tail of the substrate, 12,13-EOT. Except for Phe85, none of these amino acids appeared to be essential for enzy- matic activity. Rather, the sum of the hydrophobic resi- dues located within the protein’s catalytic cavity must contribute to the stereoselectivity and cyclization mech- anism. In fact, only the mutation of Phe85 to Leu or Ala significantly reduced enzymatic activity (Table 1). In the coupled AOS ⁄ AOC2 activity test, the F85L mutant showed moderately reduced total activity and Table 2. Studies on the inhibitory effect of vernolic acid on the Arabidopsis AOC. The results are presented as relative amount of total OPDA formed, and the specific formation of the cis(+)-enantiomer as compared to the yield with AOC2 in the absence of the inhibitor. Stan- dard deviations have been calculated on the basis of two independent duplicate measurements. HPOT HPOT + vernolic acid HPOT + methylated vernolic acid Total OPDA (%) cis-(+)-OPDA (%) Total OPDA (%) cis-(+)-OPDA (%) Total OPDA (%) cis-(+)-OPDA (%) AOS 31.4 ± 3.4 53.3 ± 2.1 29.1 ± 2.1 52.6 ± 1.0 30.7 ± 3.6 51.2 ± 0.4 AOC1 110.3 ± 4.7 98.3 ± 0.2 60.8 ± 3.7 96.5 ± 0.6 132.0 ± 12.9 98.8 ± 0.2 AOC2 100.0 ± 0.0 96.4 ± 0.6 63.7 ± 0.4 93.7 ± 0.2 105.2 ± 3.5 97.2 ± 0.2 AOC3 102.4 ± 4.0 96.4 ± 0.0 85.4 ± 4.1 93.7 ± 0.2 113.9 ± 8.8 97.8 ± 0.1 AOC4 102.4 ± 2.8 97.7 ± 0.4 100.7 ± 1.9 94.4 ± 0.3 105.2 ± 12.8 98.3 ± 0.2 L27R 75.8 ± 11.4 96.8 ± 1.2 81.8 ± 3.7 91.4 ± 1.1 91.5 ± 1.2 96.2 ± 0.0 F. Schaller et al. Oxylipin cyclization FEBS Journal 275 (2008) 2428–2441 ª 2008 The Authors Journal compilation ª 2008 FEBS 2435 a slight loss of stereoselectivity (Table 1). For the F85A mutant, this result was even more pronounced: here, total activity was reduced by about 50%, and the stereoselectivity was reduced to a ratio of 1 : 4 [23]. As this mutant could not be as highly purified as the other proteins, the absolute activity values have to be treated with caution. Consistent with the proposed function of Glu23 in the delocalization of the C15 double bond towards the oxirane (Fig. 2A), this residue was found to be essential for enzymatic activity: substitution of Glu23 with Ala resulted in complete loss of activity. The intermediate oxyanion of the cyclization reaction was proposed to be stabilized by a tightly bound water molecule (Fig. 2B). Two amino acids previously impli- cated in water binding, i.e. Pro32 and Asn25, were found to be essential for the enzymatic activity of AOC2, presumably by providing a water scaffold to guide oxylipin cyclization for the enantioselective for- mation of OPDA. Taking into account the high degree of sequence conservation among AOCs of Arabidopsis and other plant species, and considering the lack of sub- stantial differences in their enzymatic activity, our find- ings implicate a general oxylipin cyclization mechanism. Substrate-binding site of AOC2 of Arabidopsis in comparison to the AOC binding site of corn Vernolic acid is a substrate analog of 12,13-EOT lack- ing the C11 double bond, and is a strong inhibitor of corn AOC but not of Arabidopsis AOC2. The binding mechanism of the corn enzyme supposes an interaction between Arg27, which is absent from the Arabidopsis enzyme and replaced by a Leu, and the carboxylic acid group of the inhibitor. Arg27 in corn AOC is in a favorable position to form a strong salt bridge to the inhibitor [22]. Its lack in case of the Arabidopsis enzyme could explain why the AOC2 crystal structure soaked with vernolic acid did not provide a defined electron density indicative of strong interactions for the first five carbon units of the inhibitor [23]. There- fore, we tested vernolic acid and its methyl ester for their inhibitory activity on A. thaliana AOS, AOC1– AOC4, and the L27R AOC2 mutant (Table 2). As compared to the AOC of corn, the Arabidopsis AOCs displayed a much lower affinity for the inhibitor, even at high inhibitor concentrations (Table 2). Suprisingly, the substitution of Leu27 by Arg in Arabidopsis AOC2 did not result in enhanced vernolic acid sensitivity, arguing against the proposed role for Arg in the bind- ing of the inhibitor’s carboxylate moiety. Likewise, the increased stereoselectivity of the reac- tion, when using methylated HPOT as substrate, which is at variance with findings of Ziegler et al. [22], seems to disprove a role for the inhibitor’s carboxylate moi- ety in enzyme binding. Consistent with previous work, we conclude that the epoxy group in the 12,13-posi- tion, in conjunction with other, yet to be identified, structural elements of the inhibitor, may play the strongest role in inhibitor binding to the enzyme [18,19]. However, further effort is needed to resolve the issue of vernolic acid binding to AOC. Comparison of AOC1–AOC4 and their physiological functions The amino acid sequence of mature AOCs is highly similar, with all amino acids postulated to be involved in the cyclization reaction being conserved [23], and their three-dimensional structures, including the sub- strate-binding pocket, are almost identical. The simi- larity in structure is consistent with the observed similarity with respect to their biochemical proper- ties. All Arabidopsis AOCs convert the allene oxide resulting from HPOT with similar efficiency, and none of the isoforms accepts the HPOD-derived epoxide as a substrate, supporting the importance of the double bond at position C15 for both the spontaneous and the enzyme-catalyzed cyclization reactions [24,25]. Moreover, the increased formation of cis(+)-OPDA with methylated HPOT as substrate demonstrates that the carboxylic moiety of 12,13-EOT is not essential for the cyclization reaction. This finding is consistent with the disordered state of the carboxylate tail of vernolic acid in the structure of the AOC2–inhibitor complex (accession code 2DIO). The enhanced activity is due to the improved positioning of the epoxide within the hydrophobic barrel of AOC rather than to enhanced stability of the substrate, as we have observed no con- centration dependency. This finding is further under- scored by the high rate of the catalyzed HPOT conversion, combined with a moderate stability of HPOT in aqueous solutions; collectively, these factors eclipse the aspect of substrate stability in this particu- lar context. The similarity of Arabidopsis AOCs is not limited to structure and activity but extends to subcel- lular localization. All four isoforms were localized to plastids (Fig. 5), where they are potentially involved in JA biosynthesis. Confocal laser scanning microscopy revealed a plastid localization also for EGFP-tagged Arabidopsis AOS (Fig. 5), making an interaction with all AOC isoforms in planta possible. These findings are consistent with a recent study showing a plastid locali- zation for AOS and AOC of Solanum tuberosum. Potato AOS was found to be associated with thylakoid membranes, whereas AOC was identified as a predomi- nantly soluble protein in the plastid stroma, with only Oxylipin cyclization F. Schaller et al. 2436 FEBS Journal 275 (2008) 2428–2441 ª 2008 The Authors Journal compilation ª 2008 FEBS a minor thylakoid association [30]. This is somewhat contradictory to the situation found for A. thaliana,as AOS and AOC2 were shown to be part of the Arabid- opsis inner envelope membrane (Fig. 4). However, our observations corroborate previous findings [31,32]. Although complex formation of Arabidopsis AOS and AOC is not obligatory for cis(+)-OPDA formation in vitro [26], both the enhanced enzymatic activity when the polypeptides come into close vicinity and the plastid colocalization of both emzymes suggest novel means and regulatory mechanisms for controlling the flow of metabolites through the Vick and Zimmerman pathway in planta. Considering the similarity of Arabidopsis AOCs with respect to structure, catalytic activity, and subcellular localization, the question arises as to what the specific functions of individual isoforms may be. All enzymes are likely to perform the same reaction in the JA bio- synthetic pathway, and different functionalities of the Arabidopsis AOC isoforms may thus result from developmental and ⁄ or tissue-specific expression of the individual genes. Indeed, the expression of AOC genes was shown to be transiently and differentially upregu- lated upon wounding, both locally and systemically, and is induced by JA treatment [22]. Furthermore, the recent analysis of transgenic lines carrying the GUS reporter gene under the control of the individual AOC promoters revealed nonredundant promoter activities in different tissues and during distinct stages of development [7] (C. Delker, unpublished results). The characterization of the specific role of each of the AOC isozymes concomitant with JA production during plant growth and development and in response to stress will thus require the detailed analysis of loss- of-function mutants for each of the AOC genes in Arabidopsis. Experimental procedures Protein expression and purification A truncated version of AOC2 from A. thaliana was cloned into the vector pQE30 (Qiagen) using standard protocols according to [33] or [34], to yield a fusion protein (20.2 kDa) in which the first 77 amino acids (the predicted transit peptide for chloroplast targeting) are replaced by a His 6 -tag (MRGSHHHHHHRS). A second construct was generated in pET21b(+) (Novagen), in which the His 6 -tag is linked to the C-terminus of the truncated protein. Simi- larly, truncated versions of AOC1, AOC3 and AOC4 were amplified by PCR with the full-length cDNA (see below) as the template. These fragments were cloned into the BamHI and SalI restriction sites of pQE30. The constructs were transformed into Escherichia coli strains M15 and BL21AI, respectively. Cells were grown in 2YT medium, and expres- sion was induced at D 600 nm 0.5–0.6 by addition of 0.2 mm isopropyl thio-b-d-galactoside or 0.2% arabinose, respec- tively. Bacteria harboring the pQE30 constructs were har- vested after an induction time of about 5 h at 37 °C and 220 r.p.m. Bacteria with the pET21b construct were induced for an additional 30–40 h at 25 °C. Cells were lysed by soni- cation, and the fusion proteins were purified by affinity chromatography (Ni–nitrilotriacetic acid; Qiagen) and con- centrated by ultrafiltration (Centricon concentrators, 5000 Da cutoff; Millipore, Billerica, MA, USA). The yield of purified proteins exceeded 10 mgÆL )1 culture. If not sta- ted otherwise, the mutated proteins of AOC2 were expressed in the pQE30 vector under the same conditions as the native AOC2. Primers were as follows: AOC1,5¢-TATGGATCCC CAAGCAAAGTTCAAGAACTG-3¢, and 5¢-TATGTCGA CGACTAATTTTATTCACTAATT-3¢; AOC3,5¢-TATGG ATCCCCAAGTAAGATCCAAGAACTA-3¢,and5¢-TATG TCGACTAAACAGCTAATTACTTAATT-3¢; and AOC4, 5¢-TATGGAT CCCCAACTAAGATCCAA GAGCTT-3¢, and 5¢-TATGTCGACACAAAGATTTAGATTTCAATT-3¢. PCR conditions comprised 30 cycles of 94 °C for 30 min, 54 °C for 45 min, and 72 °C for 100 min. Recombinant AOS [35] was expressed according to Oh & Murofushi [36] in TB medium after induction by addition of 0.4 mm isopropyl thio-b-d-galactoside at a D 600 nm of 0.5–0.6. Cells were sedimented after 20 h at 16 °C and 150 r.p.m., and lysed by sonication, and recombinant AOS was purified as described by Zerbe et al. [26]. Purification of inclusion bodies and refolding The N53L mutant of AOC2 failed to express as soluble protein but accumulated in inclusion bodies, which were purified from bacterial lysates; the recombinant protein was subsequently refolded using the iFOLD refolding system (Invitrogen). The denatured protein was purified via affinity chromatography on Ni–nitrilotriacetic acid agarose and allowed to refold at room temperature during dialysis against 50 mm Tris (pH 7.5), 250 mm NaCl, 12.5 mm b-cyclodextrine, 1 mm Tris(2-carboxyethyl)-phosphine hydrochloride, and 0.5 ml-Arg. Gel electrophoresis and protein immunoblotting Denaturing gel electrophoresis was performed according to [37]. The discontinuous systems consisted of 4% stacking gels and 12.5% resolving gels. Protein blotting onto nitro- cellulose was carried out electrophoretically overnight (4 °C, 60 mA) as described by Towbin et al. [38]. Immun- odetection followed standard procedures [39], with either goat anti-(rabbit IgG)-conjugated or goat anti-(mouse IgG)-conjugated alkaline phosphatase as the secondary F. Schaller et al. Oxylipin cyclization FEBS Journal 275 (2008) 2428–2441 ª 2008 The Authors Journal compilation ª 2008 FEBS 2437 [...]... biosynthesis and the allene oxide cyclase family of Arabidopsis thaliana Plant Mol Biol 51, 89 5–9 11 23 Hofmann E, Zerbe P & Schaller F (2006) The crystal structure of Arabidopsis thaliana allene oxide cyclase – insights into the oxylipin cyclisation reaction Plant Cell 18, 320 1–3 217 24 Grechkin AN, Chechetkin IR, Mukhtarova LS & Hamberg M (2002) Role of structure and pH in cyclisation of allene oxide. .. Molecular cloning of allene oxide cyclase The enzyme establishing the stereochemistry of octadecanoids and jasmonates J Biol Chem 275, 1913 2–1 9138 21 Maucher H, Stenzel I, Miersch O, Stein N, Prasad M, Zierold U, Schweizer P, Dorer C, Hause B & Wasternack C (2004) The allene oxide cyclase of barley (Hordeum vulgare L.) – cloning and organ-specific expression Phytochemistry 65, 80 1–8 11 2440 22 Stenzel... 83 5–8 40 17 Hamberg M & Fahlstadius P (1990) Allene oxide cyclase: a new enzyme in plant lipid metabolism Arch Biochem Biophys 276, 51 8–5 26 18 Ziegler J, Hamberg M, Miersch O & Parthier B (1997) Purification and characterisation of allene oxide cyclase from dry corn seeds Plant Physiol 114, 56 5–5 73 19 Ziegler J, Wasternack C & Hamberg M (1999) On the specificity of allene oxide cyclase Lipids 34, 100 5–1 015... 242 8–2 441 ª 2008 The Authors Journal compilation ª 2008 FEBS F Schaller et al 35 Laudert D, Pfannschmidt U, Lottspeich F, Hollander¨ Czytko H & Weiler EW (1996) Cloning, molecular and functional characterisation of Arabidopsis thaliana allene oxide synthase (CYP 74), the first enzyme of the octadecanoid pathway to jasmonates Plant Mol Biol 31, 32 3–3 35 36 Oh K & Murofushi N (2002) Design and synthesis of. .. of O2-free sodium borate buffer, the reaction proceeded for 2 min at 4 °C, and was then stopped by acidification with concentrated HCl to pH 2. 0– 3.0 Synthesized hydroperoxides were extracted twice with two volumes of diethyl ether Anhydrous sodium sulfate was added, and the solution was swirled and filtered to remove residual water The ether phase was taken to dryness in a rotary evaporator For further... analysis, 1–2 nmol of 2[H]5OPDA was added before extraction as an internal standard Dried samples were incubated in 250 lL of 0.1 m KOH for 45 min at room temperature to convert the synthesized OPDA into the trans-configuration After a second extraction step, OPDA was dissolved in methanol and methylated with ethereal diazomethane [13] The dried fractions were then redissolved in 5 0–1 00 lL of chloroform, and. .. for the reaction mechanism Chem Phys Lipids 120, 8 7–9 9 25 Grechkin AN (1994) Cyclisation of natural allene oxide fatty acids The anchimeric assistance of beta, gammadouble bond beside the oxirane and the reaction mechanism Biochim Biophys Acta 14, 19 9–2 06 26 Zerbe P, Weiler EW & Schaller FS (2007) Preparative enzymatic solid phase synthesis of cis-(+)-12-oxo-phytodienoic acid – physical interaction of. .. cycles of 94 °C for 30 min, 56 °C for 45 min, and 72 °C for 100 min To create a control construct (pSP-rbcS-DsRed) for proteins with a plastid localization, the first exon of the small subunit of the RBCS preprotein (At1g67090) was amplified; this encodes the 66 N-terminal amino acids In the course of PCR, restriction sites for SacI and KpnI were added and used to fuse the rbcs fragment with the DsRed... biosynthesis in Arabidopsis thaliana – enzymes, products, regulation Plant Biol 8, 1–1 0 8 Schaller F & Weiler EW (1997) Enzymes of octadecanoid biosynthesis in plants 12-Oxo-phytodienoate 10,11-reductase Eur J Biochem 245, 29 4–2 99 9 Schaller F & Weiler EW (1997) Molecular cloning and characterisation of 12-oxophytodienoate reductase, an enzyme of the octadecanoid signalling pathway from Arabidopsis thaliana. .. Jackes LW (1990) Preparation and purification of soybean lipoxygenase-derived unsaturated hydroperoxy and hydroxyl fatty acids and determination of molar absorptivities of hydroxyl fatty acids Anal Biochem 188, 3 8–4 7 Hofmann E & Pollmann S (2008) Molecular mechanism of enzymatic allene oxide cyclization in plants Plant Physiol Biochem 46, 30 2–3 08 Supplementary material The following supplementary material . The allene oxide cyclase family of Arabidopsis thaliana – localization and cyclization Florian Schaller 1, * , Philipp Zerbe 1, *, Steffen Reinbothe 1 ,. biosynthesis and the allene oxide cyclase family of Arabidopsis thaliana. Plant Mol Biol 51, 89 5–9 11. 23 Hofmann E, Zerbe P & Schaller F (2006) The crystal structure