Enzymatic synthesis of a bicyclobutane fatty acidby a hemoprotein–lipoxygenase fusion proteinfrom the cyanobacterium Anabaena PCC 7120Claus Schneider*, Katrin Niisuke*, William E. Boeglin*, Markus Voehler†, Donald F. Stec†, Ned A. Porter†,and Alan R. Brash*‡
Departments of *Pharmacology and †Chemistry, Vanderbilt Institute of Chemical Biology, Vanderbilt University School of Medicine, Nashville, TN 37232
Edited by Judith P. Klinman, University of California, Berkeley, CA, and approved October 5, 2007 (received for review July 30, 2007)
Biological transformations of polyunsaturated fatty acids oftenlead to chemically unstable products, such as the prostaglandinendoperoxides and leukotriene A4 epoxide of mammalian biologyand the allene epoxides of plants. Here, we report on the enzy-matic production of a fatty acid containing a highly strained bicyclicfour-carbon ring, a moiety known previously only as a modelcompound for mechanistic studies in chemistry. Starting fromlinolenic acid (C18.3�3), a dual function protein from the cyanobac-terium Anabaena PCC 7120 forms 9R-hydroperoxy-C18.3�3 in alipoxygenase domain, then a catalase-related domain converts the9R-hydroperoxide to two unstable allylic epoxides. We isolatedand identified the major product as 9R,10R-epoxy-11trans-C18.1containing a bicyclo[1.1.0]butyl ring on carbons 13–16, and theminor product as 9R,10R-epoxy-11trans,13trans,15cis-C18.�3, anepoxide of the leukotriene A type. Synthesis of both epoxides canbe understood by initial transformation of the hydroperoxide to anepoxy allylic carbocation. Rearrangement to an intermediate bicy-clobutonium ion followed by deprotonation gives the bicyclobu-tane fatty acid. This enzymatic reaction has no parallel in aqueousor organic solvent, where ring-opened cyclopropanes, cyclobu-tanes, and homoallyl products are formed. Given the capabilityshown here for enzymatic formation of the highly strained andunstable bicyclobutane, our findings suggest that other transfor-mations involving carbocation rearrangement, in both chemistryand biology, should be examined for the production of the highenergy bicyclobutanes.
catalase � carbocation � epoxide � leukotriene � bicyclobutonium ion
The ability of lipoxygenase (LOX) enzymes to oxygenatepolyunsaturated fatty acids to specific fatty acid hydroper-
oxides is used throughout the eukaryotic world for the produc-tion of signaling molecules and other complex products (1–4).The initial hydroperoxy fatty acid product is often furthertransformed to a highly unstable biosynthetic intermediate.Thus, plants express specialized cytochrome P450 enzymes ofthe CYP74 family that convert hydroperoxy-C18 fatty acids toallene oxides, the best characterized of which is an intermediatein biosynthesis of the hormone jasmonic acid (5). In the leuko-cytes of higher animals, the 5-LOX enzyme forms the initial5-hydroperoxy-C20.4 product and converts it into the highlyunstable epoxide leukotriene A4 (LTA4), from which the otherleukotriene family members arise (6). As yet another facet of thistheme, marine corals express a natural fusion protein (7) inwhich a LOX domain converts arachidonic acid to its 8R-hydroperoxide and a catalase-related domain effects a furthertransformation to an unstable allene oxide, a potential interme-diate in formation of marine prostanoids (8). This catalase-related domain of the coral fusion protein is structurally similarto true catalases (9) yet quite distinct in function. Based on theknowledge that the plant CYP74 enzyme family exhibits aspectrum of catalytic reactions, including formation of alleneoxides, aldehydes, or vinyl ethers (10, 11), there is the possibility
that the catalase-related allene oxide synthase (AOS) is proto-typical of an enzyme family that also has diversified functions.Accordingly, the unusual catalytic activity of the catalase-relatedcoral AOS provided the impetus for the present investigation,namely to explore other possible occurrences of catalase-relatedproteins with novel functions in the biotransformation of poly-unsaturated fatty acids.
By using BLAST searches for sequences similar to the coralcatalase-related domain, one of the top matching hits besidesother coral homologues was identified in the cyanobacteriumAnabaena sp. strain PCC 7120. The Anabaena genus of cya-nobacteria are photosynthetic prokaryotes that grow in longstrings or filaments and that can develop a nitrogen-fixing abilityin specialized heterocysts. They are studied as a model forprokaryotic developmental biology (12). Anabaena PCC 7120has a genome of 6.4 Mb, and the cells also contain several largeplasmids. The novel gene resides on the 102 kb gamma plasmid.Enticingly, this small catalase-related sequence was found in thesame ORF as a LOX-like sequence, albeit a highly unusual one,much smaller than any previously known member of the LOXsuperfamily. In a separate study, we show that this C-terminaldomain of the fusion protein is a catalytically complete lipoxy-genase that specifically forms 9R-hydroperoxides from C18polyunsaturated fatty acid substrates (Y. Zheng, W.E.B., C.S.,A.R.B., unpublished data). Here, we report characterization ofthe catalytic activities of the N terminus of the fusion protein, theheme-containing domain with sequence similarity to catalase.This unusual enzyme utilizes the 9R-hydroperoxylinolenic acid(C18.3�3) product of the LOX domain as a substrate andconverts it to two epoxy fatty acids, the major one of whichcontains a bicyclic four-carbon ring. Its synthesis has importantimplications for the possible existence of novel carbocationrearrangements in both chemistry and biology.
ResultsProtein Sequences and Alignments. The novel hemoprotein fromAnabaena and the AOS domain from the coral Plexaura homomallashare an overall 35% amino acid identity. Particularly significantmatches are conserved around the distal heme His residue and thedistal heme Asn [supporting information (SI) Fig. 5]. A verysignificant mismatch occurs around the heme proximal ligand,
Author contributions: A.R.B. designed research; C.S., K.N., W.E.B., M.V., and D.F.S. per-formed research; C.S., K.N., W.E.B., M.V., D.F.S., N.A.P., and A.R.B. analyzed data; and C.S.,N.A.P., and A.R.B. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
‡To whom correspondence should be addressed at: Department of Pharmacology, Vander-bilt University School of Medicine, 23rd Avenue South at Pierce, Nashville, TN 37232-6602.E-mail: [email protected].
This article contains supporting information online at www.pnas.org/cgi/content/full/0707148104/DC1.
which is a Tyr in the coral AOS, as is characteristic of all catalasefamily members, yet by alignment this residue is replaced by His inAnabaena. Remarkably, therefore, the Anabaena sequence appearsto represent a His-ligated heme in the context of a catalase-relatedprotein framework. We should note that, whereas Anabaena is aprokaryotic cyanobacterium and other cyanobacteria are found assymbionts in corals (13), the P. homomalla AOS–LOX is unam-biguously eukaryotic based on the presence of multiple introns inthe DNA (unpublished observations). Nonetheless, such coexist-ence could have provided the opportunity for an earlier genetransfer one way or the other.
Expression and Purification of the Anabaena Fusion Protein. Weexpressed the whole Anabaena fusion protein as well as theisolated LOX domain in Escherichia coli and partially purifiedthe proteins by nickel affinity chromatography by using N-terminal His6-tags; (expression of the catalase-related domain byitself gave protein containing no heme and exhibiting no cata-lytic activity). Anabaena contains abundant polyunsaturatedfatty acids and is particularly rich in linolenic acid (C18.3�3)(14–16). We found that this fatty acid is oxygenated by the LOXdomain to the corresponding 9R-hydroperoxide (a contrast withplant 9-LOX enzymes, which have S stereospecificity) (5). Thelipoxygenase is not involved in further metabolism, a point weestablished by using the separately expressed LOX domain.However, the catalase-related domain of the fusion proteinavidly metabolizes the 9R-hydroperoxide to a complex spectrumof stable end products, including triols, diols, and epoxyalcohols(data not shown). It appeared likely that the primary enzymicproduct of the catalase-like hemoprotein is an unstable epoxideor epoxides, and we reasoned that if this could be analyzeddirectly it would greatly simplify the product profile and helpclarify the fundamental mechanism of biosynthesis.
Isolation of Two Allylic Epoxides Formed by the Catalase-RelatedDomain. We explored conditions under which the Anabaenaenzyme reacted with pure 9R-hydroperoxylinolenic acid in abiphasic hexane/pH 8 aqueous system that would simultaneouslyextract the hydroperoxy substrate from the hexane into thewater, allow enzymic metabolism, and then instantly extract theless polar epoxide product(s) back into the hexane, thus afford-ing protection from hydrolysis, an approach similar to the one wedeveloped for isolation of allene oxides (17). After 2 min ofvortex mixing at 0°C, UV spectroscopy of the hexane showeddisappearance of substrate and appearance of new chro-
mophores, one near 200 nm and the other with �max at 278 nmcharacteristic of the leukotriene A class of allylic epoxides (Fig.1A) (18). The resulting hexane extract was treated with dia-zomethane for 10 s at 0°C to form the methyl ester derivative (19)and the fatty acid derivatives subsequently analyzed by reversed-phase HPLC using conditions adapted from a method foranalysis of synthetic leukotriene A4 (Fig. 1B) (20). HPLCanalysis showed near quantitative conversion to two products,present in a 2:1 ratio as determined by using a 14C substrate. Themore prominent product 1 displays a UV spectrum with endabsorbance extending beyond 230 nm, and product 2 has thespectrum of a conjugated triene, �max 278 nm (Fig. 1C). LC-MSanalysis using positive ion electrospray ionization revealed thatthe two products have the same molecular weight (306 for themethyl ester) as indicated by their identical adduct ions withsodium, potassium, and triethylamine.
Identification of the Major Allylic Epoxide Product. The 1H-NMRspectrum of product 1 methyl ester, together with the COSYanalysis of cross-couplings, outlined a structure of a 9,10trans-epoxy-11trans-C18.1 derivative (Fig. 2 and SI Dataset 1). Withonly one double bond and yet the same molecular weight as theleukotriene A-type epoxide, it follows that the structure of 1must contain two rings. The arrangement of the carbon atomswas further investigated through conventional proton–protondecoupling experiments, heteronuclear single quantum correla-tion (HSQC), and heteronuclear multiple bond correlation(HMBC) to identify the carbon–proton couplings, the distor-tionless enhancement polarization transfer (DEPT) experimentto confirm the CH2, CH, and C carbons, and NOESY forthrough-space couplings (SI Figs. 6–9). The key structural datafor establishing the bicyclobutane ring was the fact that all fourcarbons of C13 through C16 appeared as CH signals in theHSQC and DEPT experiments, suggesting that, given theirchemical shift values, each of these carbons is bound to threeother carbons and one proton. The C14 and C15 carbons, whichexhibit almost identical chemical shifts at 9.3 ppm, are clearlyresolved in the carbon spectrum at 800 MHz (SI Fig. 8). TheCOSY analysis showed strong correlation among the protonsignals of H14, H15, and H16 of the ring and extending to H17and H18 of the �-carbon chain. H13 (a doublet at 2.36 ppm)couples only to H12 on the double bond; its couplings with theneighboring protons H14 and H15 thus must be very small,indicating that H14 and H15 are held at almost right angles in thebicyclic ring system. The through space cross-peaks in the
ω λA B C
Fig. 1. Preparation, purification, and UV analysis of unstable epoxides. An ice-cold solution of 9R-hydroperoxylinolenic acid (90 �M) in 10 ml of ice-cold hexanewas vortex-mixed for 2 min with Anabaena catalase–LOX enzyme (0.8 nmol) in 0.2 ml of phosphate buffer, pH 8; reaction with the same sample of enzyme wasrepeated twice more using fresh substrate in ice-cold hexane. The combined hexane phases were evaporated to �2 ml under a strong stream of nitrogen, treatedwith diazomethane and 1% ethanol for 10 s at 0°C, and then evaporated to dryness and stored in hexane at �70°C. (A) UV spectrum of the hydroperoxy substratein hexane before reaction and the hexane phase after mixing with enzyme. (B) Reversed-phase HPLC analysis of the product methyl esters with UV detectionat 205 and 270 nm. A Waters Symmetry C18 column (25 � 0.46 cm) was eluted with methanol/20 mM aqueous triethylamine at pH 8 [80:20 (vol/vol)] at a flowrate of 1 ml/min. (C) Normalized UV spectra of the two main products.
18942 � www.pnas.org�cgi�doi�10.1073�pnas.0707148104 Schneider et al.
NOESY spectrum confirm the covalent structure and confirmedassignment of the configuration of the side chains (C1–C12, C17,and C18) as exo–endo (Fig. 3). Thus, the complete covalentstructure of product 1 was established as 9R,10Rtrans-epoxyoctadeca-11trans-(13,14,15,16-bicyclo[1.1.0]butyl)-enoicacid.
Identification of Product 2 as a LTA4 Analogue. The 1H-NMR andH,H-COSY spectra of product 2 established the structure of an
allylic epoxide with a conjugated triene double bond system (SIDataset 2). Based on the coupling constants, the configurationof the 11,12, 13,14, and 15,16 double bonds is defined astrans,trans,cis (J � 15.2, 14.8, and 11.0 Hz, respectively). Con-sidering also that the configuration of C9 should not changeduring the transformation of the 9Rhydroperoxide to the 9,10-epoxide, product 2 was identified as a leukotriene A-typeepoxide, 9R,10R-trans-epoxyoctadeca-11E,13E,15Z-trienoicacid. This structure is directly analogous to the structure of themammalian 5-LOX product, LTA4, a 5S,6S-trans-epoxy-20.4�6fatty acid with a trans,trans,cis (7E,9E,11Z) conjugated triene.Interestingly, biosynthetically produced LTA4 has not beensubject to a complete and direct structural analysis. Knowledgeof its structure rests on comparison of its biotransformation andhydrolytic reactions with synthetic LTA4, and on an understand-ing of its mechanism of biosynthesis by the leukocyte 5-LOXenzyme (21, 22).
DiscussionBicyclobutane is the simplest bicyclic hydrocarbon (Scheme 1).It has a bond strain energy of 66 kcal/mol, more than double thevalue of either cyclopropane or cyclobutane (�26 kcal/mol) (23).Extensive efforts have been made during the past 40 years todevelop synthetic approaches to molecules containing a bicy-clobutane subunit and to understand the chemistry of thisstructure (24–27). Although bicyclobutanes have been the focus
10,11
12,13 9,10
8,9
2,3
3,4
12 1110
913
2
163 8
7
184,5,6
171415
163 8
7a,b
18
4,5,6
17
14,15
-OCH3
109
13 2
17,18
16,1716,14/15
H3CO2CH
HH
HO
8
9
2 10 12
1118
1315
1416
-OCH3
12 13 1 2 1 2 36 2 6integration:
Fig. 2. NMR analysis of product 1. Spectra were recorded in d6-benzene at283 K using a Bruker 600 MHz spectrometer equipped with a cryoprobe. Thetwo-dimensional H,H COSY spectrum is shown below, with an expanded viewof two areas of the spectrum depicted above. On the COSY spectrum, thecorrelations are clear from H8 through the trans epoxide (H9, H10, J � 2.0 Hz)to the trans double bond (H11, H12, J � 15.5 Hz) to H13 of the bicyclobutanering, a doublet at 2.36 ppm. H14 and H15 are in a very similar chemicalenvironment and, therefore, have similar chemical shifts in the proton (1.29–1.36 ppm) and carbon (9.2 and 9.3 ppm) spectra (SI Fig. 8). Both H14 and H15are coupled to H16, which extends the correlation into the ethyl substituent(H17 and H18). Although the correlations through H13 are not evident here,H12 shows a three-bond correlation to both C14 and C15 in the HMBCspectrum (SI Fig. 7).
Fig. 3. Configuration of the bicyclobutane ring of product 1. The NOESYNMR spectrum of product 1 was recorded in d6-benzene at 600 MHz. Thepartial chemical structure illustrates the through-space couplings of the singleprotons at H13 and H16, the two ends of the bicyclobutyl ring. The 13-exo,16-endo configuration of the ring can be deduced from the observation that thecoupling of H13 to H14/15 is weak, whereas H16-H14/15 is strong, and that thecoupling of H13 to H17 is strong, whereas there is no detectable NOE betweenH12 and H16 (SI Fig. 9).
HH
H H
HH
Scheme 1. Bicyclo [1.1.D] butane.
Schneider et al. PNAS � November 27, 2007 � vol. 104 � no. 48 � 18943
of many fundamental studies in organic chemistry, this structureis, to our knowledge, heretofore unknown in nature. We reporthere on the isolation and identification of a substituted bicy-clobutane formed in biosynthesis from a fatty acid hydroperox-ide precursor. The proposed biosynthetic mechanism proceedsthrough the rearrangements of carbocations and is completedwith an enzymatic step of deprotonation of a bicyclobutoniumion yielding a bicyclobutane ring via a route not observed insolution chemistry.
Transformation of the linolenate hydroperoxide to the allylicepoxides 1 and 2 can be understood by a mechanism involvingintermediate carbocations (Fig. 4). The conversion of conju-gated diene hydroperoxides to allyl carbocations like 3 has ampleprecedent, being implicated in the enzymatic synthesis of alleneoxides, vinyl ethers, and other fatty acid derivatives (28), as wellas related transformations of prostaglandin endoperoxides toprostacyclin and thromboxane A2 (29). Removal of a protonfrom the carbocation 3 gives the leukotriene-type epoxide,product 2. Although the structure of 2 is directly analogous tothat of the leukocyte 5-LOX product LTA4, its mechanism ofbiosynthesis is quite different. Whereas epoxide 2 is formed viainitial activation of the hydroperoxide and subsequent rear-rangements of an epoxy allylic carbocation with a final elimina-tion of H� (Fig. 4), the biosynthesis of LTA4 involves an initialLOX-catalyzed hydrogen abstraction from the carbon chain,followed by radical rearrangements that lead to cleavage of thehydroperoxide and epoxide formation at the ultimate step.
Carbocation 3 is homoallylic and, as such, provides access tothe cyclopropylcarbinyl and bicyclobutonium ions (30), two ofwhich are shown in Fig. 4. The nature of these carbocationintermediates has been an important chapter in the ‘‘classical–nonclassical’’ ion debate (31, 32). It is generally agreed that themost stable carbocation intermediates present are separated bylow-energy barriers (33) and that cyclopropanes, cyclobutanes,and homoallylic structures are the predictable set of productsthat form from these intermediates and the equivalent alcoholderivatives in aqueous media (34). Such reactions occur inenzymatic transformations to a wide array of natural products(e.g., refs. 28 and 35–38). In the case of the Anabaena catalase-related protein, we suggest that generation of bicyclobutoniumion 5 in the vicinity of an enzymatic base provides an additionalproduct-forming route, deprotonation to give the bicyclobutane,product 1. This is a most unusual outcome that is not reproducedin solution chemistry. The enzyme must provide an environmentin which proton abstraction from the bicyclobutonium ionfacilitates production of the high-energy bicyclobutyl moiety, aspecies we then recovered by its immediate extraction intohexane. Notably, a potential epoxide product arising via protonabstraction from the cyclopropylcarbinyl ion 4 was not detected,
and we isolated a single bicyclobutyl isomer, arguing for stringentreaction control in the enzyme active site.
Historically, the catalase gene family is known for its role inprotection from oxidative stress via the breakdown of hydrogenperoxide, yet clearly the Anabaena gene has a biosyntheticfunction. A role in biosynthesis is precedented by the catalase-related domain of the coral AOS–LOX fusion protein thatconverts 8R-hydroperoxy-arachidonic acid to an allene oxide, areaction equivalent to that of the P450 AOS of plants (10). Withthe recent successful x-ray structural analysis of the coral cata-lase-related domain, its modest sequence similarity to truecatalases evolved into remarkable parallels in three-dimensionalstructure (9). The heme-binding pocket and surrounding proteinnetwork of true catalases are well conserved, whereas thestructural features involved in melding together the catalasehomotetramer are absent. The coral AOS domain is 43 kDa insize and crystallizes as a dimer (9), whereas true catalases usuallyare tetrameric or hexameric with subunits of 55–69 kDa(‘‘small’’) or 75–84 kDa (‘‘large’’) (39). The Anabaena hemo-protein domain is �41 kDa in size, with strong homology to thecoral AOS, except near the end of the sequence where theproximal heme ligand appears to be substituted with a histidine(SI Fig. 5). The 9R-hydroxy metabolites of linolenic and linoleicacids have been detected in �5:1 ratio in extracts of a species ofAnabaena (40), but these organisms have yet to be examined forthe more complex products that might be expected to arise fromthe epoxy-bicyclobutyl linolenate. The Anabaena hemoprotein–lipoxygenase fusion protein resides on the �-plasmid inAnabaena PCC 7120, and, by analogy with the plasmid-encodedantibiotic resistance genes, it may have a specialized role andconfer an advantage in a selected environment. Other smallcatalase analogues reside in the genomes of many microorgan-isms and have the potential for additional functions in themetabolism of natural peroxides.
Given the capability shown here for enzymatic transformation tothe highly strained and unstable bicyclobutane, other enzymictransformations involving carbocation rearrangements should beexamined for the production of the high-energy bicyclobutanes.Immediate extraction of unstable reaction products may also beapplicable to purely chemical transformations and may allow for theisolation of products believed not to be existent or considered toounstable for characterization. We show that, with appropriatemethodology, even previously intractable biological products, suchas the leukotriene A-type epoxide, and novel ones, like the bicy-clobutane, are amenable to recovery and structural analysis.
Materials and MethodsMaterials. Fatty acids was purchased from NuChek Prep.[1-14C]Linolenic acid was purchased from NEN Life ScienceProducts.
ω
Fig. 4. Transformation of 9R-hydroperoxy-C18.3�3 by the Anabaena catalase-related enzyme.
18944 � www.pnas.org�cgi�doi�10.1073�pnas.0707148104 Schneider et al.
Cloning, Expression, and Purification of Anabaena Enzyme Constructs.The cDNA for the full-length fusion protein was cloned by PCRfrom Anabaena sp. strain PCC 7120 genomic DNA, a kind giftfrom James W. Golden (Texas A&M University, College Sta-tion, TX). Sequencing confirmed the identity to the publishedsequence in the National Center for Biotechnology Informationdatabase (NP�478445) and at CyanoBase (http://bacteria.kazusa.or.jp/cyanobase). Several different cDNA expressionconstructs were prepared: each listed below was cloned intopET17b for expression in E. coli. Full-length constructs includedthe native cDNA sequence (amino acid code MDLNTY—LMMSINI.) and the same sequence with a His6 tag on the Nterminus (MHHHHHHDLNTY—LMMSINI.). These con-structs expressed with similar heme content (with the main Soretband observed at 406 nm) and were used for enzyme quantifi-cation assuming � � 100,000. The affinity-purified His-taggedexpression construct was used throughout these experiments.The proteins were expressed in E. coli BL21 (DE3) cells (No-vagen) using methodology we described previously (41), and theHis6-tagged protein was purified on Ni-NTA agarose (Qiagen)according to the manufacturer’s instructions. The lipoxygenase-only domain (starting at amino acid 344 with the amino acidsequence KDDLPGK. . . and comprising the last 430 aa of thefull-length construct) was expressed with an N-terminal His6 tagand purified by nickel affinity chromatography. We also at-tempted to express several constructs encoding the N-terminaldomain only (amino acids 1–344). These constructs had the Histag placed at either the N or C terminus (or no His tag) andincluded constructs with the C terminus extended a further 20 aaalong the fusion protein. However, these constructs expressedwith no heme and exhibited no catalytic activity.
Preparation of Hydroperoxides. 9R-Hydroperoxy-C18.3�3 was pre-pared by using the LOX domain of the Anabaena enzyme in pH7.5 Tris buffer according to the methods described previously forother fatty acid hydroperoxides (42). The product was extractedinto dichloromethane and purified by SP-HPLC (Beckman 5�silica column; 25 � 0.46 cm) using a solvent system of hexane/isopropanol/acetic acid 100:1:0.1 by volume (flow rate of 1ml/min), with detection of the product by UV detection at 235nm (42). The product was quantified by UV spectroscopy (� �23,000 at 235 nm) and stored at �20°C in ethanol.
Activity Assays. Small-scale incubations with the purified enzymeswere typically conducted in a 0.5-ml UV cuvette and analyzed byUV spectrometry in an incubation buffer (50 mM Tris, 150 mMNaCl, pH 7.5). Enzyme activity was monitored by repetitive scan-ning in the range 350–200 nm or by monitoring disappearance ofthe signal at 235 nm in the time-drive mode. To measure the rate
of reaction at a higher concentration of substrate (100–250 �M),reactions were conducted in a 2-mm path-length microcuvette.
Extraction and HPLC Analysis of Unstable Products. Enzyme reactionswere conducted at 0°C, with the substrate initially in hexane (10 ml,�100 �M 9-hydroperoxide) layered over the Anabaena enzyme(0.8 nmol) in 200 �l of phosphate buffer, pH 8. The reaction wasinitiated by vigorous vortex mixing of the two phases, which wascontinued for 2 min; then the test tube was placed back on ice. Thehexane phase was scanned from 200–350 nm in UV light by usinga Perkin-Elmer Lambda-35 spectrophotometer, and, if all substratewas consumed (as in Fig. 1A), the reaction was repeated with freshsubstrate in hexane mixed with the same batch of enzyme. Thecombined hexane phases were evaporated to �2 ml by using avigorous stream of nitrogen. The sample was then treated withethanol (20 �l) and ethereal diazomethane for 10 s at 0°C and thenrapidly evaporated to dryness and stored in hexane at �20°C untilfurther analysis. The same procedures but omitting the methylationstep gave samples of the free acids.
The presence of an excess of alcohol over water in a slightlybasic solution greatly prolongs the half-life of allylic epoxides,such as leukotriene A4, allowing their analysis by reversed phaseHPLC at room temperature (20). The hexane extracts wereanalyzed and purified by using a Waters Symmetry C18 5-�mcolumn (0.46 � 25 cm) eluted at a flow rate of 1 ml/min withmethanol/20 mM potassium phosphate, pH 8 (replaced withtriethylamine for LC-MS analysis; see below), in the proportions80:20 (vol/vol), with UV light detection at 205, 220, 235, and 270nm using an Agilent 1100 series diode array detector. The mainproducts were recovered by extraction with cold hexane followedby evaporation to dryness under a strong stream of nitrogen.
LC-MS and NMR analysis. LC-MS of the allylic epoxides wasperformed using a Thermo Finnigan LC Quantum instrument.A Waters Symmetry C18 column (0.2 � 15 cm) was eluted withmethanol/20 mM aqueous triethylamine adjusted to pH 8 withacetic acid [80:20 (vol/vol)] at 0.2 ml/min. The heated capillaryion lens was operated at 220°C. Nitrogen was used as a nebuli-zation and desolvation gas. The electrospray potential was heldat 4 kV. Source-induced dissociation was set at �10 eV. Massspectra were acquired over the mass range m/z 100–500 at 2 s perscan. Collision-induced dissociation was performed at �15 eV.
NMR spectra were recorded on a Bruker 800-MHz (13C andDEPT) or 600-MHz instrument at 283 K. The samples weredissolved in d6-benzene, and the chemical shifts are reportedrelative to the benzene signal (� 7.16 ppm for hydrogen and 128.0ppm for carbon). Both instruments were equipped with a BrukerTCI cryoprobe.
We thank Dr. Thomas M. Harris for careful review of the NMR data. Thiswork was supported by National Institutes of Health Grant GM-74888.
1. Gerwick WH (1994) Biochim Biophys Acta 1211:243–255.2. Grechkin A (1998) Prog Lipid Res 37:317–352.3. Brash AR (1999) J Biol Chem 274:23679–23682.4. Serhan CN, Arita M, Hong S, Gotlinger K (2004) Lipids 39:1125–1132.5. Feussner I, Wasternack C (2002) Annu Rev Plant Biol 53:275–297.6. Samuelsson B, Dahlen SE, Lindgren JA, Rouzer CA, Serhan CN (1987) Science 237:1171–1176.7. Koljak R, Boutaud O, Shieh B-H, Samel N, Brash AR (1997) Science 277:1994–1996.8. Corey EJ, d’Alarcao M, Matsuda SPT, Lansbury PT, Jr (1987) J Am Chem Soc 109:289–290.9. Oldham ML, Brash AR, Newcomer ME (2005) Proc Natl Acad Sci USA 102:297–302.
10. Tijet N, Brash AR (2002) Prostaglandins Other Lipid Mediat 68–69:423–431.11. Grechkin AN (2002) Prostaglandins Other Lipid Mediat 68–69:457–470.12. Golden JW, Yoon HS (2003) Curr Opin Microbiol 6:557–563.13. Lesser MP, Mazel CH, Gorbunov MY, Falkowski PG (2004) Science 305:997–1000.14. Parker PL, van Baalen C, Maaurer L (1967) Science 155:707–708.15. Sato N, Murata N, Miura Y, Ueta N (1979) Biochim Biophys Acta 572:19–28.16. Li R, Watanabe MM (2004) Curr Microbiol 49:376–380.17. Brash AR (1989) J Am Chem Soc 111:1891–1892.18. Corey EJ, Arai Y, Mioskowski C (1979) J Am Chem Soc 101:6748–6749.19. Brash AR, Baertschi SW, Ingram CD, Harris TM (1988) Proc Natl Acad Sci USA 85:3382–3386.20. Fitzpatrick FA, Morton DR, Wynalda MA (1982) J Biol Chem 257:4680–4683.21. Borgeat P, Samuelsson B (1979) Proc Natl Acad Sci USA 76:3213–3217.22. Shimizu T, Izumi T, Seyama Y, Tadokoro K, Radmark O, Samuelsson B (1986) Proc Natl
