Carbocyclic Fatty Acids in Plants: Biochemical and Molecular Genetic Characterization of Cyclopropane Fatty Acid Synthesis of Sterculia Foetida

Home , Sterculia foetida, Sterculic acid

Carbocyclic fatty acids in plants: Biochemical and molecular genetic characterization of cyclopropane fatty acid synthesis of Sterculia foetida

Xiaoming Bao*, Sue Katz†, Mike Pollard*, and John Ohlrogge*‡

*Department of Plant Biology, Michigan State University, East Lansing, MI 48824; and †Montgomery Botanical Center, Miami, FL 33156

Contributed by Jan A. D. Zeevaart, Michigan State University, East Lansing, MI, March 14, 2002 (received for review January 23, 2002) Fatty acids containing three-member carbocyclic rings are found in a variety of desaturases (14, 15) in animals, which might be the bacteria and plants. Bacteria synthesize cyclopropane fatty acids underlying cause of some of the disorders. Because of the health (CPA-FAs) only by the addition of a methylene group from S- concerns, vegetable oils containing CPE-FAs need to be treated adenosylmethionine to the cis-double bond of monoenoic phos- with high temperature or hydrogenation before consumption pholipid-bound fatty acids. In plants CPA-FAs are usually minor (16). As a result, processing costs are increased whereas trans- components with cyclopropene fatty acids (CPE-FAs) more abun- fatty acids are also produced by the hydrogenation. Genetically dant. Sterculia foetida seed oil contains 65–78% CPE-FAs, princi- eliminating CPE-FAs from seed oils is therefore a nutritionally pally sterculic acid. To address carbocyclic fatty acid synthesis in desirable objective. plants, a cDNA library was constructed from developing seeds The biosynthetic pathway of CPA-FAs in bacteria is well during the period of maximum oil deposition. About 0.4% of 5,300 understood (17). The first CPA-FA synthase gene was cloned expressed sequence tags were derived from one gene, which from Escherichia coli on the basis of its ability to complement a shared similarities to the bacterial CPA-FA synthase. However, the CPA-FA-deficient mutant (18). It was demonstrated that bac- predicted protein is twice as large as the bacterial homolog and terial CPA-FAs were synthesized from monounsaturated fatty represents a fusion of an FAD-containing oxidase at the N terminus acids by addition of a methylene group, derived from S- and a methyltransferase at the C terminus. Functional analysis of adenosylmethionine, across the double bond. The monoenoic the isolated full-length cDNA was conducted in tobacco suspension fatty acyl substrates are esterified to phospholipids, most prom- cells where its expression resulted in the accumulation of up to inently phosphatidylethanolamine (17). After the identification 6.2% dihydrosterculate of total fatty acids. In addition, the dihy- of the E. coli CPA-FA synthase, three genes from Mycobacterium drosterculate was specifically labeled by [methyl-14C]methionine tuberculosis were found, which function as cyclopropane syn- and by [14C]oleic acid in the transgenic tobacco cells. In in vitro thases in mycolic acid biosynthesis (19, 20). By contrast, little assay of S. foetida seed extracts, S-adenosylmethionine served as attention has been paid to the biosynthesis of CPE-FAs in plants. a methylene donor for the synthesis of dihydrosterculate from Yano et al. (21) conducted in vivo labeling experiments with oleate. Dihydrosterculate accumulated largely in phosphatidylcho- several species of Malva and proposed that the pathway involved line in both systems. Together, a CPA-FA synthase was identified initial formation of dihydrosterculic acid from oleic acid with from S. foetida, and the pathway in higher plants that produce subsequent desaturation to sterculic acid (Fig. 1). carbocyclic fatty acids was defined as by transfer of C1 units, most In this study we examine the biochemical and molecular likely from S-adenosylmethionine to oleate. genetic basis for the biosynthesis of three-member ring carbocyclic fatty acids in plants. Sterculia foetida is a tropical tree wide array of unusual fatty acids is found in seed oils (1, 2). belonging to the Sterculiaceae family of order Malvales. The AAmong these are fatty acids containing three-member seeds of S. foetida are rich in oil (55% dry weight) and contain carbocyclic rings, namely cis-cyclopropane fatty acids (CPA- up to 78% CPE-FAs (2, 22), one of the highest levels of FAs) and cyclopropene fatty acids (CPE-FAs). Carbocyclic fatty carbocyclic fatty acids reported in nature. Here, we report the acids are distributed across several plant orders, most notably isolation, with an expressed sequence tag (EST) approach, and Malvales but including the Fabales and Sapindales (2–4). Ster- functional characterization of the eukaryotic CPA-FA synthase culic acid is often the prevalent CPE-FA, but malvalic acid, one gene. carbon shorter in chain length than sterculic acid, can be a significant component. In Litchi chinensis seed oil, dihydroster- Materials and Methods culic acid is the major carbocyclic fatty (5, 6). The CPA- and Plant Material and Chemicals. Developing seeds of S. foetida L. CPE-FAs are not confined to seeds. Long-chain CPA-FAs have were collected from July 20 to October 15, 1998 at Montgomery been described in various polar lipid classes of leaves of early Botanical Center (Miami, FL); the accession number of the tree spring plants (7), whereas both CPA- and CPE-FAs are found in sampled is MBC 78453A. On receiving the fresh seeds, the seed root, leaf, stem, and callus tissue in plants of the Malvaceae (8, coats were removed and the cotyledons and embryos were used 9). In addition to carbon and energy storage in seeds, the for labeling experiments or were frozen and stored at Ϫ80°C for function of CPA- and CPE-FAs in plants may involve resistance RNA extraction, enzymology, and lipid analysis. Tobacco sus- to fungal attack (10). Although CPA-FAs are widely distributed pension cells (Nicotiana tabacum L. cv. Bright yellow 2) were in bacterial lipids, no CPE-FAs have been reported in bacteria to date. The cyclopropene ring of sterculic acid is highly strained, and Abbreviations: FAME, fatty acid methyl ester; CPA-FA, cyclopropane fatty acid; CPE-FA, cyclopropene fatty acid; SfCPA-FAS, Sterculia foetida cyclopropane fatty acid synthase; therefore, the acid exhibits a unique and reactive chemistry (3) 18:1, oleic acid; 18:2, linolenic acid; EST, expressed sequence tag. suggesting commercial oleochemical applications were sterculic Data deposition: The sequence reported in this paper has been deposited in the GenBank acid available in sufficient quantity. Many seed lipids containing database (accession no. AF470622). CPE-FAs are extensively consumed by humans, especially in ‡To whom reprint requests should be addressed. E-mail: [email protected]. tropical areas (11). It is well documented that dietary CPE-FAs The publication costs of this article were defrayed in part by page charge payment. This lead to the accumulation of hard fats and other physiological article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. disorders in animals (12, 13). CPE-FAs are strong inhibitors of §1734 solely to indicate this fact.

7172–7177 ͉ PNAS ͉ May 14, 2002 ͉ vol. 99 ͉ no. 10 www.pnas.org͞cgi͞doi͞10.1073͞pnas.092152999 Downloaded by guest on September 27, 2021 activity was located and quantified by using an Instant-Imager. Finally, the labeled spots were recovered for GC͞MS analysis.

Assay of CPA-FA Synthase. Frozen cotyledon tissue from S. foetida developing seeds was ground to a fine powder in liquid nitrogen. A cell-free extract was prepared by thawing the powder and briefly homogenizing in four volumes of chilled buffer containing 0.1 M Na Tricine⅐NaOH (pH 7.0), 1% wt͞vol defatted BSA, 1% wt͞vol polyvinylpyrrolidone-40, 15% vol͞vol glycerol, and 1 mM 2-mercaptoethanol, then filtering the slurry through mira- cloth. CPA-FA synthase assays contained 0.1 ml of cell-free homogenate, 0.02–0.05 mM oleoyl-CoA, and 0.02 mM S-[methyl- 14C]adenosylmethionine substrate in a total volume of 0.2 ml and were incubated at 30°C for 1 h. The assay was terminated by the addition of 0.5 ml of 10% aqueous KOH and 1.0 ml of ethanol, and allowed to stand overnight to give complete saponification of the lipids. On acidification, the labeled free fatty acids were extracted into hexane, then the hexane phase washed with water and evaporated to dryness. An aliquot of the organic phase was Fig. 1. Proposed pathway for the biosynthesis of sterculic acid from oleic acid (21). assayed for radioactivity. The remainder of the product was analyzed by TLC directly or after treatment with an ether solution of diazomethane. maintained in liquid medium containing Murashige and Skoog basal salts (GIBCO), 3% sucrose, 2.5 mM Mes⅐KOH (pH 5.7), Library Construction and Sequencing. Equal amounts of developing 1 mg/ml thiamine, 1 mg/ml myo-inositol, and 1 ␮M 2,4- seeds from collections of August 5, August 25, September 10, dichlorophenoxyacetic acid. Cultures were subcultured weekly and September 30, 1998, were pooled. During this period, the with 5% (vol͞vol) inoculum from a 7-day-old culture and shaken deposition of carbocyclic fatty acids showed a linear increase. at 28°C in 200-ml flasks. L-[methyl-14C]methionine (55 mCi/ Ten grams of the developing seeds were ground into fine powder mmol) and [1-14C]oleic acid (50 mCi/mmol) were purchased in liquid nitrogen and RNA extracted as described by Schultz et from American Radiolabeled Chemicals (St. Louis). al. (26). A cDNA library of S. foetida developing seeds was constructed by Stratagene. The cDNAs were directionally cloned Lipid Analysis. into the Uni-ZAP vector at EcoRI and XhoI sites. The library To determine fatty acid accumulation during S. ϫ 7 foetida seed development an internal standard, tritridecanoin, consisted of 2.6 10 plaque-forming units of primary plaques was added to seed tissue before the lipid extraction. Lipids were with average insert size of 1.7 kb. Mass excisions were performed extracted according to Bligh and Dyer (23). Lipids were trans- according to the protocol provided by Stratagene. Clones (21,120) were picked and grown in two hundred twenty 96-well PLANT BIOLOGY methylated by vortexing at room temperature for 3 min with plates. Twenty-eight of the 220 plates were directly sequenced at 0.5% sodium methoxide in methanol͞heptane (1:1, vol͞vol). the Michigan State University sequencing facility. The remain- After extraction, the fatty acid methyl esters (FAMEs) were ing 192 plates (18,432 clones) were spotted on nylon filters for analyzed by GC͞MS by using a Hewlett Packard 5890 gas library subtraction as described by White et al. (27). Based on the chromatograph MSD 5972 mass analyzer. Separations of information from the first 1,500 sequences from the nonsub- FAMEs were performed on a 30 m ϫ 0.25 mm i.d. DB-23 ͞ tracted library, the three most abundant sequences were chosen column. Lipid extraction, preparation of FAMEs, and GC MS to subtract the library on the filters. A total of seventy 96-well analysis for tobacco suspension cells used the same procedure as plates were re-racked and sequenced. described for S. foetida seeds, except that no internal standard was added. To identify dihydrosterculic acid in transgenic to- Constructs for Tobacco Expression. A putative S. foetida CPA-FA bacco suspension cells, saturated and unsaturated FAMEs were synthase (SfCPA-FAS) clone was identified on the basis of its separated by argentation TLC (24). Argentation plates (15% Ϫ abundance and similarity to bacterial CPA-FA synthase. The silver nitrate in silica) were developed sequentially at 20°Cto complete cDNA of the putative SfCPA-FAS is 2,977 bp long heights of 8, 13, and 19 cm in toluene. FAME bands were located ͞ Ј Ј compiled from two overlapping clones R50-D5 (1–1,732 bp) and by spraying with 0.2% (wt vol) 2 ,7 -dichlorofluorescein in R15-C3 (933–2,977 bp). Both clones were sequenced from both ethanol and viewing under UV light. FAMEs were recovered strands. To assemble a full-length SfCPA-FAS clone the se- ͞ from the scraped silica bands by elution with hexane diethyl quence encoding amino acids 1–335 was amplified from clone ͞ ͞ ether (2:1, vol vol), and analyzed by GC MS. A racemic R50-D5 with primers JO886 (TCCTCTAGACTCGAGCCC- methyl cis-dihydrosterculate standard was prepared by reaction GGGATGGGAGTGGCTGTGATCGGTGGTGGGAATC) of methyl oleate with diiodomethane and diethylzinc in and JO883 (GTTGTAAGACGTCGTGTAACTCGGT- hexane (25). CATACAATTCG). The sequence encoding amino acids 336– 864 with the stop codon was amplified from clone R15-C3 with Radiolabeling Transgenic Tobacco Cells. The independent trans- primers JO884 (CAATGTGCTGCAGAATGTTGGGAAAA- genic callus was transferred back into liquid medium and sub- CAAGTCA GCC) and JO885 (GGGAGGATCTCGAGC- cultured as described above. After 3 days of subculture, 5 ␮Ci of CTATTTACTTTTGATAAAGTTAATAGGC). For the con- L-[methyl-14C] methionine or [1-14C]oleic acid was added to the venience of cloning, a silent mutation was made at the third medium and culture continued for 24 h. Cells were collected by position of the codon for lysine 335, from CTA to CTG, so that brief centrifugation, lipids extracted, and FAMEs prepared as a PstI site could be created. The N-terminal sequence was described above. Saturated FAMEs were separated from other inserted into pBluescript KS at XhoI and PstI sites, after which FAMEs by argentation TLC. The saturated FAMEs were re- the C-terminal section was inserted at PstI and XbaI sites. The covered and fractionated by C18 reverse-phase TLC by using resulting construct was named pBluescript KS-SfCPA-FAS and acetonitrile͞methanol͞water (75:25:0.5, vol͞vol͞vol). The radio- the sequence was confirmed from both strands. The full-length

Bao et al. PNAS ͉ May 14, 2002 ͉ vol. 99 ͉ no. 10 ͉ 7173 Downloaded by guest on September 27, 2021 Fig. 2. Alignment of the S. foetida CPA-FA synthase C-terminal portion with other known cyclopropane synthases. E. coli represents the CPA-FA synthase from E. coli. CMAS1, CMAS2, and MMAS2 are from M. tuberculosis (20). The dashed line below the sequences indicates the S-adenosylmethionine-binding site, and the * indicates the catalytically important cysteine.

SfCPA-FAS was subcloned from pBluescript KS into the binary amounts of developing seeds from the collections during this vector pE1776 between SmaI and XbaI sites. pE1776 carries the period were mixed together and used for RNA extraction and constitutive promoter (derived from mannopine synthase gene cDNA library construction. After mass excision of the library, of Agrobacterium tumefaciens) and agropine synthase terminator clones were picked and sequencing of the first 1,500 clones gave (28). The resulting construct was named as pE1776-SfCPA-FAS Legumin A, Legumin B, and a nonspecific lipid transfer protein and transferred into Agrobacterium strain LBA4404 for tobacco as the three most abundant sequences. These were chosen for suspension cell transformation. library subtraction because they represented about 30% of all of the clones in the library. After subtraction by filter hybridization Expression of SfCPA-FAS Clone in Tobacco Suspension Cells. Agrobac- and re-racking, another 3,800 sequences of 500 bp or greater terium-mediated tobacco transformation was conducted as de- were obtained. BLAST searches identified several ESTs showing scribed by Rempel and Nelson (29). A 100-␮l aliquot of over- similarities with bacterial CPA-FA synthase. On compiling these night Agrobacterium culture containing the construct was added EST sequences it was clear that 23 overlapping ESTs were to 4 ml of 3-day-old tobacco suspension cells and kept at 28°C derived from the same gene. Assembly of sequences led to a for 3 days. The cells were harvested by brief centrifugation and predicted putative CPA-FA synthase protein of 864 aa with washed three more times with NT medium containing 100 ␮g/ml molecular weight of 98,046. Even though no significant trans- kanamycin and 500 ␮g/ml carbenicillin. The cells were spread on membrane domains were identified, the S. foetida CPA-FA selection plates (NT medium with addition 0.7% phytagar, 100 synthase is likely a membrane-associated protein, because 70% mg/liter of kanamycin, and 500 mg/liter of carbenicillin). After of the total CPA-FA synthase activity was recovered in the 3–4 weeks, independent transformants were transferred to new microsomal pellet after centrifugation at 100,000 ϫ g for1h.In plates. Tissue (0.5–1 g fresh weight) was collected, lipids were contrast, the bacterial cyclopropane synthase is a soluble protein extracted, and FAMEs were prepared for GC͞MS analysis. and about half the size of the putative S. foetida CPA-FA synthase at 382 aa long. The bacterial protein shares significant Results homology only with the carboxyl-terminal half of the S. foetida Lipid Deposition During Seed Development of S. foetida. In S. foetida clone with 32% (122 of 376) identity and 49% (188 of 376) seed oils, sterculic acid is the major CPE-FA (55–78%) with similarity (Fig. 2). malvalic acid present at about 10% of the amount of sterculic acid (2, 3, 30). To define the appropriate stage of seeds for Functional Analysis of the Putative SfCPA-FAS Sequence in Tobacco studying the biosynthesis of CPE-FAs, developing seed pods of Suspension Cells. To test its function, the putative SfCPA-FAS was S. foetida were collected from July 20 to October 15, 1998, at subcloned into a plant transformation vector pE1776 behind a Montgomery Botanical Center (Miami, FL). Each pod con- constitutive promoter (28) and transformed into tobacco sus- tained 10–20 seeds. The pods were dark green and the seeds pension cells. These cells do not contain any cyclopropane fatty white in July, and the color of the pod gradually turned red and acids. After the transformation and antibiotic selection, inde- the seed coats turned brown with the progress of seed develop- pendent transformants were subcultured at 20-day intervals. ment. The cotyledons of the first collection on July 20 were Lipids were extracted from 15 transformants with pE1776- watery and filled progressively until seed desiccation began in SfCPA-FAS, and from 12 transformants with only empty vector late October. Both total fatty acids and CPE-FAs accumulated pE1776. Fatty acid compositions of independent transformants at linear rates from August 5 to October 14. During the same with the SfCPA-FAS clone contained dihydrosterculic acid from period the percent of CPE-FAs in total fatty acid increased from 3.2 to 6.2% of total fatty acids, with an average of 4.0%. No 40 to 60%. CPE-FA (sterculate) was detected in the unfractionated tobacco lipid extract and no dihydrosterculate was detected in control Identification of CPA-FA Synthase from S. foetida Developing Seeds transformants. To confirm the identity of dihydrosterculic acid, Through an EST Approach. To maximize the representation of FAMEs prepared from lipid extracts were fractionated by CPA-FA synthase and desaturase clones in the cDNA library, argentation TLC to give a saturated FAME fraction, which was developing seeds at stages when CPE-FA accumulate at the then analyzed by GC͞MS. Fig. 3A shows the total ion current highest rate were used to prepare the library. Because CPE-FA chromatogram of saturated fatty acids from control tobacco accumulated at a linear rate from August 5 to October 14, equal callus transformed with empty vector of pE1776. The major

7174 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.092152999 Bao et al. Downloaded by guest on September 27, 2021 Fig. 4. Comparison of the mass spectra of the methyl dihydrosterculate standard and the additional peak found in transgenic cells expressing S. foetida CPA-FA synthase construct. (A) Mass spectrum of the peak with retention time 35.69 from Fig. 2B.(B) Mass spectrum of methyl dihydrosterculate standard.

Fig. 3. Total ion current gas chromatograms of the saturated FAME fraction PLANT BIOLOGY from transgenic tobacco cells and the methyl dihydrosterculate standard. (A) with the SfCPA-FAS were used for in vivo labeling experiments. Saturated FAMEs from transgenic tobacco cells with the control construct 14 pE1776. (B) Saturated FAMEs from transgenic tobacco cells expressing the S. Suspension cells were labeled with [1- C]oleic acid and L-[meth- 14 foetida CPA-FA synthase clone, pE1776-SfCPA-FAS-2. A major additional peak yl- C]methionine separately for 24 h. Fig. 5A shows the distri- with retention time of 35.69 min is observed. (C) Methyl dihydrosterculate bution of radioactivity from the transmethylation products of 14 standard (retention time of 35.69 min). total lipids labeled with [1- C]oleic acid and analyzed by argentation TLC. In the transformant carrying the SfCPA-FAS about 3.5% of the total radioactivity was found in the upper band saturated fatty acids in tobacco callus are 16:0 and 18:0, with containing only saturated FAMEs, whereas no label was seen in some 20:0. Fig. 3B shows the saturated fatty acids from one of this band in the vector control transformant. This result dem- the 15 tobacco lines transformed with the putative SfCPA-FAS onstrated that SfCPA-FAS-transformed lines converted oleate clone. An additional prominent peak is present in the transfor- into a saturated fatty acid, most likely dihydrosterculate. Fig. 5B mant, with a retention time identical with the methyl cis- gives the radioactivity distribution from cells incubated with dihydrosterculic acid standard (Fig. 3C). From GC equivalent L-[methyl-14C]methionine. Again, radioactivity was only found in chain-length data on cyclopropane FAMEs (31, 32), it is clear the saturated FAME band from the cells transformed with the that trans-cyclopropane isomers would elute significantly before SfCPA-FAS. To confirm that the radioactivity in the upper cis-isomers. However, a cis-11,12-isomer would run later than saturated FAME band does indeed represent dihydrosterculate, and probably be incompletely resolved from the cis-9,10-isomer. the saturated FAMEs of pE1776-SfCPA-FAS-2 were eluted Inspection of peak shape and width suggested a single compo- from the argentation TLC plate and analyzed by C18 reverse- nent eluting in the additional peak. The mass spectrum of the phase TLC. As shown in Fig. 5C, the radioactive FAMEs labeled additional peak (Fig. 4A) matches the dihydrosterculate stan- by both [1-14C]oleic acid and L-[methyl-14C]methionine migrated dard (Fig. 4B). A weak molecular ion at m͞z ϭ 310 along with at the same position as the dihydrosterculate standard. The mass (M-32) and (M-74) peaks at 278 and 236 atomic mass units, coeluting with the labeled bands in Fig. 5C was confirmed as respectively define the molecular weight of the new compound dihydrosterculate by elution and GC͞MS analysis. These results (3), and fingerprinting of the two spectra shows a very close demonstrated by in vivo labeling that the SfCPA-FAS gene match. On the basis of the following criteria for the new fatty acid product is responsible for synthesis of dihydrosterculic acid by found in tobacco callus transformed with putative SfCPA-FAS transferring a methylene group from methionine to oleic acid. clone, namely being isolated as a saturate and having a retention Additional in vivo labeling experiments (data not shown) showed time and mass spectrum identical with the standard, we conclude that the dihydrosterculate product labeled from methionine with certainty it is dihydrosterculic acid. accumulates mainly in phosphatidylcholine.

In Vivo Labeling of Tobacco Suspension Cells Transformed with the S-Adenosylmethionine Is a Methyl Donor for CPA-FA Synthase in SfCPA-FAS Clone. To confirm both the origin of and the acceptor Extracts of S. foetida Seeds. To test whether S-adenosylmethionine for the methylene group, tobacco suspension cells transformed is a potential C1 donor for plant carbocyclic fatty acid synthesis

Bao et al. PNAS ͉ May 14, 2002 ͉ vol. 99 ͉ no. 10 ͉ 7175 Downloaded by guest on September 27, 2021 dihydrosterculic acid was synthesized at levels as high as 6.2% of total fatty acids. By using the transgenic tobacco cells, dihydrosterculate was effectively labeled from both [1-14C]oleic acid and L-[methyl-14C]methionine (Fig. 5). This result defined the biosynthesis of dihydrosterculic acid in S. foetida as by addition of a methylene group to the double bond of oleate. The methylene group is derived from methionine, and in vitro assays with S. foetida seed extracts show that S-adenosylmethionine is indeed a methylene donor. Although the amino acid sequence does not display obvious transmembrane domains, the concentration of this activity in a microsomal fraction is consistent with the CPA-FA synthase being either a membrane-associated or an integral membrane protein. In summary, these data clearly demonstrate that the S. foetida clone homologous to the bacterial CPA-FA synthases encodes a functional CPA-FA synthase. On the basis of in vivo labeling of Malva tissues Yano et al. (21) proposed that the biosynthesis of sterculic acid was a two-step reaction; first the addition of a methylene group to oleate, then desaturation of dihydrosterculate to sterculate (Fig. 1). They considered sterculic acid not to be derived by methylene addition across the 9,10-triple bond of stearolic acid, because no conversion of [1-14C]stearolic acid to sterculic acid was observed. Fig. 5. In vivo labeling of fatty acids from transgenic tobacco suspension Dihydrosterculate was clearly desaturated to sterculate, but the cells. Transgenic tobacco suspension cells were labeled for 24 h with [1-14C]- conversion rates were very slow, which agrees with our obser- oleic acid (A)orL-[methyl-14C]methionine (B). Labeled FAMEs were analyzed vations after labeling studies of S. foetida seeds (data not shown). by argentation TLC. The control (lane 1) is tobacco transformed with the A pathway for the biosynthesis of sterculic acid as a two-step empty vector pE1776. SfCPA-FAS-2 (lane 2) is a cell line transformed with the reaction by way of dihydrosterculate is strongly supported by the S. foetida CPA-FA synthase construct, pE1776-SfCPA-FAS. In the saturated FAME band at the top of the plate radioactivity was detected only in trans- isolation and functional demonstration of S. foetida CPA-FA genic line SfCPA-FAS-2. (C) To identify the radioactivity in the saturated FAME synthase described here, but at this stage we cannot unequivo- fraction further, the radioactive spots of SfCPA-FAS-2 (lane 2) from A and B cally rule out an additional route by way of an acetylenic were recovered and separated by C18 reverse-phase TLC. Radioactivity from precursor. An unusual feature associated with sterculic acid both samples migrated with the methyl dihydrosterculate standard. biosynthesis in both seeds and other plant tissues is substantial ␣-oxidation. Often malvalic acid is a major CPE-FA. Other noncarbocyclic fatty acids only rarely show limited amounts of cell-free extracts of developing seeds were optimized for activity. ␣-oxidation products. Thus, extensive ␣-oxidation seems unique S-[methyl-14C]Adenosylmethionine was tested as substrate, and to CPE-FA biosynthesis, and it is possible to speculate that it may under optimal conditions CPA-FA synthase activities in the Ϫ Ϫ be associated, either by a biosynthetic linkage or because of a range of 0.25–0.5 nmol⅐min 1⅐g 1 (fresh weight) of seed tissue particular propensity of CPE-FAs to undergo and͞or to induce were measured. The optimum pH was 7.0 with Tricine buffer ␣-oxidation. No chain-shortened CPA-FAs were found in any of preferred over Tris or sodium phosphate. The addition of BSA the transgenic tobacco cell lines screened. The latter is an and polyvinylpyrrolidone-40 together enhanced activity 2-fold. indication that the presence of CPA-FAs did not induce ␣- Oleoyl-CoA in the assay at 20–50 ␮M enhanced activity about oxidation in tobacco. In L. chinensis, the only seed in which 2-fold; higher concentrations were inhibitory. The apparent Km CPA-FAs accumulate without CPE-FA, no ␣-oxidation products for S-adenosylmethionine in the cell-free extract was 20 ␮M. of dihydrosterculic acid exist, although traces of 17- and 15- After saponification and hexane extraction, Ͼ95% radioactivity carbon ␤-oxidation products are seen (6). in the hexane extract comigrated with free fatty acids. Analysis So far, one gene from E. coli (33) and three closely related of labeled FAMEs by C18 reversed-phase TLC showed a single genes (cmas-1, cmas-2, and mmas-2) from Mycobacterium (20, radioactive spot coeluting with the methyl dihydrosterculate 34) have been functionally proven as cis-CPA-FA synthases. The standard. Thus, the radioactivity recovered in the hexane phase S. foetida CPA-FA synthase reported here is the first enzyme of after saponification is a good measure of the total label incor- this kind that has been isolated from eukaryotes. A comparison porated into [methylene-9,10-14C]dihydrosterculate. When the of the amino acid sequence of S. foetida CPA-FA synthase with reaction with the cell-free homogenate was terminated by lipid these CPA-FAS genes is shown in Fig. 2. A striking difference is extraction most of the labeled dihydrosterculate was found in the that all of the bacterial amino acid sequences are about half the phosphatidylcholine fraction, which suggests that oleoyl phos- length of S. foetida CPA-FA synthase and share significant phatidylcholine may be a substrate for the enzyme in vivo. similarity only to the C terminus half of the S. foetida sequence. The proposed S-adenosylmethionine binding motif (amino acid Discussion 171–179, with the E. coli numbering) and the catalytically This study is a step toward understanding the biosynthesis of important cysteine 354 are absolutely conserved for all of the CPA- and CPE-FAs in plants at molecular and biochemical proteins (33). The S. foetida and E. coli sequences are more levels. The EST approach has yielded a CPA-FA synthase clone closely related to each other than either are to the Mycobacte- from the cDNA library of S. foetida developing seeds. Twenty- rium genes. The C-terminal portion of the S. foetida CPA-FA three ESTs derived from CPA-FA synthase clone give a relative synthase shares a high degree of similarity with a cluster of abundance in the library of Ϸ0.4%. This level is higher than most Arabidopsis genes on chromosome 3 and with other plant enzymes of fatty acid metabolism in oilseed cDNA libraries. sequences, although no evidence of CPA-FA synthesis is known Thus, the expectation that the high oil and sterculic acid contents for these species. These genes possibly encode enzymes involved found in S. foetida seeds would result in a high expression level in methyl transfer reactions for other hydrophobic substrates. for an enzyme of the pathway proved correct in this instance. On The N-terminal half (amino acids 1–438) is unique for the S. expressing a full-length construct in tobacco suspension cells, foetida CPA-FA synthase sequence in that no other known

7176 ͉ www.pnas.org͞cgi͞doi͞10.1073͞pnas.092152999 Bao et al. Downloaded by guest on September 27, 2021 cyclopropane synthases possess this portion. However, the N- more likely in the ␣-oxidation that accompanies the formation terminal half of S. foetida CPA-FA synthase shares significant of sterculic acid. Such speculations can only be addressed by homology with Arabidopsis gene At3g23500 (and to a lesser future experiments. The fact that in the transgenic tobacco cells extent with At3g23520). These putative genes encode products neither CPE-FA nor ␣-oxidation products were observed sug- related to tryptophan 2-monooxygenases and belong to a flavin- gests additional gene products are required for these functions. containing group of oxidases. The tryptophan 2-monoxygenase In conclusion, the CPA-FA synthase gene was isolated from gene product itself produces indole acetamide, which can be developing seeds of S. foetida. The gene product catalyzed the converted to the auxin indole acetic acid by hydrolysis. Most addition of a methylene group (possibly derived from S- flavin-containing proteins have a highly conserved motif located adenosylmethionine) to oleic acid to form dihydrosterculic acid. at the N terminus of the protein and involved in binding of the Experiments are underway to determine the substrate specificity ADP moiety of FAD (35–37). The proposed motif (G-X-G-X- of the S. foetida CPA-FA synthase, both with respect to acyl X-G-X-X-X-A) is preceded by three or four hydrophobic resi- chain unsaturation and length, and to the head group of the dues (38). The conserved FAD-binding motif is present in the phospholipid acceptor. The functional expression of this clone first 15 aa of S. foetida CPA-FA synthase (MGGGGIQGLV- gives us the essential tool to identify the gene(s), most likely SAYVLAKAGVNVVVYE). Because the mechanism of cyclo- desaturases, that will produce sterculic acid. From such studies propane ring formation is believed to proceed by a carbocation we may eventually define the role for the FAD-containing, mechanism, with OCHCH OCHϩO the proposed intermedi- oxidase-like N-terminal domain of the S. foetida CPA-FA syn- 3 thase. The homology between certain Arabidopsis thaliana pu- ate formed by addition of the methyl group from S- tative methyltransferase and oxidase gene products and the S. adenosylmethionine (17), it is hard to see how a redox system foetida CPA-FA synthase raises interesting questions about such as a FAD-containing protein is involved in the catalytic alternative roles of these genes. reaction of methylene addition. Thus, the discovery that the CPA-FA synthase protein contains a fused redox domain, prob- Dr. S. B. Gelvin (Purdue University) graciously provided vector pE1776 ably an oxidase, remains a puzzle. We cannot be sure that this for plant transformation. This work was funded in part by The Dow domain is functional, but its potential functionality raises ques- Chemical Company and Dow AgroSciences. The Michigan Agricultural tions about a role in either the desaturation to sterculic acid, or Experiment Station is also acknowledged for its support of this research.

1. Van de Loo, F. J., Fox, B. G. & Somerville, C. (1993) in Lipid Metabolism in 21. Yano, I., Morris, L. J., Nichols, B. W. & James, A. T. (1972) Lipids 7, 35–45. Plants, ed. Moore, T.S., Jr. (CRC, Boca Raton, FL), pp. 91–126. 22. Pasha, M. K. & Ahmad, F. (1992) J. Agric. Food Chem. 40, 626–629. 2. Badami, R. C. & Patil, K. B. (1981) Prog. Lipid Res. 19, 119–153. 23. Blight, E. G. & Dyer, W. J. (1959) Can. J. Biochem. Physiol. 37, 911–917. 3. Christie, W.W. (1970) in Topics in Lipid Chemistry, ed. Gunstone, F.D. (Logos, 24. Morris, L. J., Wharry, D. M. & Hammond, E. W. (1967) J. Chromatogr. 31, London), Vol. 1, pp. 1–49. 69–76. 4. Smith, C.R., Jr. (1970) Prog. Chem. Fats Other Lipids 11, 139–177. 25. Furukawa, J., Kawabata, N. & Nishimura, J. (1968) Tetrahedron 24, 53–58. 5. Lie Ken Jie, M. S. F. & Chan, M. F. (1977) J. Chem. Soc. Chem. Commun.78. 26. Schultz, D. J., Craig, R., Cox-Foster, D. L., Mumma, R. O. & Medford, J. I. 6. Gaydou, E. M., Ralaimanarivo, A. & Bianchini, J.-P. (1993) J. Agric. Food (1994) Plant Mol. Biol. Rep. 12, 310–316. Chem. 41, 886–890. 27. White, J. A., Todd, J., Newman, T., Focks, N., Girke, T., Martı´nez de Ila´rduya,

7. Kuiper, P. J. C. & Stuiver, B. (1972) Plant Physiol. 49, 307–309. O., Jaworski, J. G., Ohlrogge, J. & Benning, C. (2000) Plant Physiol. 124, PLANT BIOLOGY 8. Yano, I., Nichols, B. W., Morris, L. J. & James, A. T. (1972) Lipids 7, 30–34. 1582–1594. 9. Schmid, K. M. & Patterson, G. W. (1988) Phytochemistry 27, 2831–2834. 28. Gelvin, S. B., Hauptmann, R., Ni, M. & Cui, D. (1999) U.S. Patent 5,955,646. 10. Schmid, K. M. & Patterson, G. W. (1988) Lipids 23, 248–252. 29. Rempel, H. C. & Nelson, L. M. (1991) Transgenic Res. 4, 199–207. 11. Ralaimanarivo, A., Gaydou, E. M. & Bianchini, J. P. (1982) Lipids 17, 1–10. 30. Bohannon, M. B. & Kleiman, R. (1978) Lipids 13, 270–273. 12. Phelps, R. A., Shenstone, F. S., Kemmerer, A. R. & Evans, R. J. (1965) Poultry 31. Christie, W. W. & Holman, R. T. (1966) Lipids 1, 176–182. Sci. 44, 358–394. 32. Christie, W. W., Gunstone, F. D., Ismail, I. A. & Wade, L. (1968) Chem. Phys. 13. Page, A. M., Sturdivant, C. A., Lunt, D. K. & Smith, S. B. (1997) Comp. Lipids 2, 196–202. Biochem. Physiol. B Biochem. Mol. Biol. 118, 79–84. 33. Wang, A. Y., Grogan, D. W. & Cronan, J. E. (1992) Biochemistry 31, 14. Fogerty, A. C., Johnson, A. R. & Pearson, J. A. (1972) Lipids 7, 335–338. 11020–11028. 15. Cao, J. M., Blond, J. P. & Bezard, J. (1993) Biochim. Biophys. Acta 1210, 27–34. 34. George, K. M., Yuan, Y., Sherman, D. R. & Barry, C. E. (1995) J. Biol. Chem. 16. Bianchini, J. P., Ralaimanarivo, A. & Gaydou, E. M. (1983) J. Food Sci. 48, 270, 27292–27298. 253–255. 35. Eggink, G., Engel, H., Vriend, G., Terpstra, P. & Witholt, B. (1990) J. Mol. Biol. 17. Grogan, D. W. & Cronan, J. E. (1997) Microbiol. Mol. Biol. Rev. 61, 429–441. 212, 135–142. 18. Grogan, D. W. & Cronan, J. E. (1984) J. Bacteriol. 158, 286–295. 36. Eberhardt, S., Richter, G., Gimbel, W., Werner, T. & Bacher, A. (1996) Eur. 19. Yuan, Y., Lee, R. E., Besra, G. S., Belisle, J. T. & Barry, C. E. (1995) Proc. Natl. J. Biochem. 242, 712–719. Acad. Sci. USA 92, 6630–6634. 37. Haigler, B. E., Suen, W. C. & Spain, J. C. (1996) J. Bacteriol. 178, 6019–6024. 20. Yuan, Y. & Barry, C. E. (1996) Proc. Natl. Acad. Sci. USA 93, 12828–12833. 38. Russel, M. & Model, P. (1988) J. Biol. Chem. 263, 9015–9019.

Bao et al. PNAS ͉ May 14, 2002 ͉ vol. 99 ͉ no. 10 ͉ 7177 Downloaded by guest on September 27, 2021