J Am Oil Chem Soc (2012) 89:1599–1609 DOI 10.1007/s11746-012-2071-5

ORIGINAL PAPER

Identification of the Unsaturated Heptadecyl Fatty Acids in the Seed Oils of Thespesia populnea and Gossypium hirsutum

Michael K. Dowd

Received: 29 December 2011 / Revised: 25 March 2012 / Accepted: 5 April 2012 / Published online: 4 May 2012 Ó AOCS (outside the USA) 2012

Abstract The fatty acid composition of the seed oils of (17:1) and heptadecadienoic (17:2) acids, which were Thespesia populnea and cotton variety SG-747 (Gossypium reported to be present at levels of about 0.1 % each. hirsutum) were studied to identity their 17-carbon fatty Because of their small concentrations, these components acids. With a combination of chemical derivatization, gas have largely been ignored in studies on cottonseed oil chromatography, and mass spectrometry, 8-heptadecenoic composition [3–5]. Consequently the positioning of the acid, 9-heptadecenoic acid, and 8,11-heptadecadienoic double bonds in these compounds has never been deter- acids were identified in both oils. Additionally, traces of mined, and their origins have not been discussed. 10-heptadecenoic acid were identified in the T. populnea As part of an in-progress survey of the seed fatty acid oil. Although these odd-carbon number fatty acids are composition of wild Gossypium species, seeds of several present in only minor amounts in cottonseed oil, they make related plant genera were also evaluated. Among these up about *2 % of the fatty acids in T. populnea seed oil. were seeds of Thespesia populnea. Chromatography of the The identification of these acids indicates that fatty acid methyl esters from this plant’s seed oil suggested that the a-oxidation is not restricted to cyclopropene fatty acids in oil contained 10- to 20-fold higher levels of these odd- these plants, but also occurs with unsaturated fatty acids. chain fatty acids, which have not been mentioned in prior Combined with malvalic acid (generally accepted as being analyses of the plant’s oil [6–8]. formed by a-oxidation of sterculic acid), *7 % of the fatty T. populnea is a small- to medium-sized flowering tree acids in T. populnea seed have under gone a-oxidization. that is widely distributed in the tropics. Its wood is used in The results should help clarify the composition of T. pop- many cultures for carving small ornamental and utilitarian ulnea seed oil, which has been reported inconsistently in objects. Like the cotton plant, the Thespesia genus is part the literature. of the Gossypieae tribe within the Malvoideae subfamily of Malvaceae plants. The plant is a gossypol producer, and its Keywords a-Oxidation Á Cottonseed Á Cyclopropenoid bark was the original source material for the isolation of fatty acids Á Heptadecyl fatty acids Á Malvaceae plants (?)-gossypol [9]. Other unusual terpenoid compounds have also been isolated from its hardwood [10]. T. populnea is thought to originate from India but has become established Introduction throughout the tropics. Consequently, the plant has many common names, including Portia tree (Caribbean), Milo In early communications, Fisher and coworkers [1, 2] noted (South Pacific), Tulip Tree (India), Pacific Rosewood that cottonseed oil contained small levels of heptadecenoic (Australia), and Seaside Mahoe (USA). To give a better account for the composition of this oil and to allow for comparisons between the seed oils of this plant M. K. Dowd (&) tribe, the identities of these acids were needed. These were Commodity Utilization Research Unit, Agricultural Research determined by a combination of chemical derivatization Service, US Department of Agriculture, 1100 Robert E. Lee Blvd, New Orleans, LA 70124, USA methods coupled with gas chromatography and mass spec- e-mail: [email protected] trometry. The origin of these components is also discussed. 123 1600 J Am Oil Chem Soc (2012) 89:1599–1609

Experimental Methods For identifying fatty acids and determining fatty acid profiles, single seeds were dehulled and extracted in indi- Materials vidual microcentrifuge tubes as described above. Between 1 and 1.5 mL of hexane was used to produce miscellas Seeds of a commercial Upland cotton (Gossypium hirsu- containing 20–30 mg/mL of extracted crude oil. tum) variety (SG 747) were obtained from an ARS cotton research program in Stoneville, MS, USA. Seeds of Derivatization Chemistry T. populnea were collected from trees growing just above the shoreline along the gulf coast of Central Florida (USA). To prepare fatty acid methyl ester (FAME) derivatives, Seed of Asclepias syriaca (milkweed) were purchased from 200 lL of methanolic base was mixed with *1mLof Everwilde Farms (Sand Creek, WI, USA) and were used to freshly extracted miscella in a screw-capped test tube. With confirm the elution time of the methyl ester of 9,12-hex- periodic vortex mixing, each solution was heated at 70 °C adecadienoic acid (9,12-16:2), which was also identified in for 10 min. After allowing the tubes to cool to room tem- both seed oils. Methanolic base, dimethyl disulfide perature, 1 mL of hexane and 1 mL of a saturated NaCl (DMDS), 3-hydroxymethylpyridine (i.e., 3-pyridylcarbi- solution were added, and the samples were vortex mixed nol), tert-butoxide in tetrahydrofuran, dichloromethane, again. The contents were then allowed to settle into organic and N-methyl-N-(trimethylsilyl)fluoroacetamide were pur- and aqueous layers. The organic phase was recovered and chased from Sigma-Aldrich (St. Louis, MO, USA) or their was used directly for chromatography or was dried with Supelco subsidiary (Bellefonte, PA, USA). Methyl ester anhydrous sodium sulfate for further derivatization. standards of vernolic acid and dihydrosterculic acid To prepare thiomethyl adducts, a procedure similar to (CPA19:0) were purchased from Matreya, LLC (Pleasant that described in the AOCS Lipid Library was used [12]. Gap, PA, USA). A volume of the FAME mixture was taken to yield *10 mg of esters; the hexane was evaporated under a Oil Content and Iodine Value stream of dry nitrogen; and the esters were re-dissolved in 1 mL of DMDS. Two hundred microliters of a 60 mg/mL Seeds were sectioned with a razor blade and were dehulled solution of iodine in diethyl ether was then added to by hand. The kernels were ground in a Braun hand chopper catalyze the reaction. Each tube was capped and allowed to pass a #20-mesh sieve and were then freeze dried. to mix gently at room temperature overnight (*16 h). Approximately 5.0 g of dry ground tissue was prepared The next morning, 5 mL of hexane was added to each from each seed source. To determine the kernel oil content, sample, and the contents were mixed. Solutions were then *1.5 g of each sample was extracted with 40 mL of washed two-to-three times with 1 mL portions of a 5 % petroleum ether in a Soxtec extractor (Foss North America, (w/v) aqueous sodium thiosulfate solution until all of the St. Claire, MN). After recovering the miscella, the bulk of color of the organic phase cleared. Each organic phase the solvent was evaporated, and the recovered oil was dried was then recovered, dried over anhydrous sodium sulfate, in an oven at 130 °C for 30 min. The oil was then stored in and evaporated to dryness under a stream of dry nitrogen. a desiccator until the sample reached room temperature, Preparations were then taken up in 1 mL of hexane for and the amount of oil was determined gravimetrically. chromatography. To measure iodine values, *2.0 g of ground kernel To prepare picolinyl esters, freshly extracted oil was sample was extracted with hexane at room temperature in a trans-esterified with 3-hydroxymethylpyridine under basic series of microcentrifuge tubes (*200 mg seed tissue and conditions as described by Destaillats and Angers [13]. 1 mL of hexane per tube). Chrome-steel beads (2.3 mm Crude oil (*10 mg) was recovered from each miscella by diam.) were added to the tubes, which were then ground in evaporating the hexane under a stream of dry nitrogen. Oils a Biospec Products (Bartlesville, OK, USA) microcentri- were then treated with 1 mL of anhydrous dichlorometh- fuge bead mill (90 % maximum speed) to macerate the ane, followed by 200 lL of 3-hydroxymethylpyridine and kernel matrix. The tubes were then centrifuged for 5 min at 100 lL of a 1.0 M solution of potassium tert-butoxide in *10,000g to pellet the debris, and the supernatant misc- tetrahydrofuran. The mixtures were allowed to react at ellas were combined in a single screw-cap test tube. Oil 40 °C for 30 min. After cooling, 1 mL of a 2.5 % sodium was recovered by evaporating the hexane under a stream of bicarbonate solution was added to each sample. The sam- dry nitrogen until the tube’s weight loss was negligible. ples were vortex mixed, and the aqueous and organic The process yielded *600 mg of crude oil for each sam- phases were allowed to separate. Each organic (lower) ple. Iodine values were determined by AOCS Official phase was recovered, dried with anhydrous sodium sulfate, Method Cd 1–25 [11]. Both yield and iodine analyses were and centrifuged to pellet the drying agent. The supernatants conducted in duplicate. were recovered for chromatography. 123 J Am Oil Chem Soc (2012) 89:1599–1609 1601

Gas Chromatography/Mass Spectrometry Determination of Fatty Acid Composition

Most chromatography was conducted on a pair of Agilent The oils from ten individual seeds of T. populnea and ten 7890 gas chromatographs with split/splitless injectors and individual seeds of SG747 cottonseed were extracted and flame ionization detectors. One instrument was operated trans-methylated to form methyl esters. These were ana- with a Supelco SP-2380 capillary column (30 m 9 0.25 lyzed on both the SP-2380 and SP-2560 stationary phases. mm i.d. 9 0.20 lm film thickness) and the other instru- Common fatty acids were identified from expected elution ment was operated with a Supelco SP-2560 capillary col- times. Malvalic (CPE18:1) and sterculic (CPE19:1) acids umn (100 m 9 0.25 mm i.d. 9 0.20 lm film thickness). were identified from their reported mass spectra [14]. On Two columns were required as neither column could sep- the SP-2380 column, the methyl esters of CPA19:0 and arate all of the FAME components of interest (discussed linoleic acid (9,12-18:2) co-eluted, which was confirmed below). Both injectors were operated at 250 °C, and the by chromatography of a standard. On the SP-2560 column, detectors were operated at 300 °C. Helium was used as the the methyl ester of CPA19:0 eluted between the methyl carrier gas with a head pressure that was programmed to esters of CPE19:1 and 9,12-18:2. On this column, however, produce a constant linear velocity of *20 cm/s. The the unidentified 17:2 component of interest co-eluted with injectors were operated in split mode with a split ratio of the methyl ester of oleic acid (9-18:1). 1:100. Sample injection volumes were typically 1 lL. Fatty acid distributions were determined from the SP- To obtain mass ionization data, an Agilent 6890 chro- 2380 chromatograms. To account for CPA19:0, the ratio of matograph with an Agilent 5973N mass selective (MS) ion peak areas for the methyl esters of CPA19:0 and CPE19:1 detector was used. This instrument was fitted with a longer was determined from the SP-2560 chromatograms. The Supelco SP-2380 column (100 m 9 0.25 mm i.d. 9 0.20 expected peak area for the CPA19:0 methyl ester was then lm film thickness), which was used to compensate for the calculated from this area ratio and CPE19:1 methyl ester increased pressure drop associated with vacuum conditions peak area from the SP-2380 chromatographic data. This of the detector. Injector and gas flow conditions were as area was then subtracted from the area of the co-eluting described above, except that the split ratio was varied as 9,12-18:2 methyl ester peak. Individual peak areas were needed to generate suitable fragmentation data. Mass spec- corrected for FID response factor differences as described tral data were collected between 50 and 450 u. in AOCS Official Method Ce 1e-91 (AOCS, 1998) [15]. Oven conditions varied depending on the column, form of detection, and the type of derivative being analyzed. For FAME-FID analyses on the SP-2380 column, the initial Results oven temperature was 170 °C, which was held constant for 3 min; then the temperature was ramped at 1 °C/min to The yield of crude oil from the T. populnea kernels was 180 °C; then the temperature was ramped at 4 °C/min to 32.6 ± 0.5 %, which was comparable to the oil yield 240 °C, which was held constant for 15 min (43 min total obtained from SG 747 cottonseed kernels (31.2 ± 0.2 %). run time). For analysis on the SP-2560 column, the initial The iodine value of freshly extracted T. populnea seed oil oven temperature was 185 °C, which was held constant for was 98.8 ± 3.0, which was also comparable to that of the 35 min; then the temperature was ramped at 5 °C/min to SG-747 cottonseed oil (100.9 ± 4.3). The iodine value of 240 °C, which was held constant for 15 min (61 min total the SG-747 variety was toward the low end of the range of run time). values typically reported for cottonseed oils (98–118) [3]. For FAME-MS analysis with the SP-2380 column, the Gas chromatography of the fatty acids as methyl esters initial temperature of 170 °C was maintained for 11 min showed that T. populnea seed oil was composed mostly of and the final temperature was held for 20 min to account the same fatty acids found in cottonseed oils (Fig. 1). for the increased elution times associated with the longer Palmitic acid (16:0), 9-18:1, and 9,12-18:2 were the most column (56 min total run time). Higher oven temperatures abundant fatty acids. Also identified were myristic (14:0), were needed to separate the dithiomethyl and picolinyl palmitoleic (9-16:1), heptadecanoic (17:0), stearic (18:0), derivatives. For the esters treated with DMDS, the starting cis-vaccenic (11-18:1), a-linolenic (9,12,15-18:3), arachi- temperature was 240 °C, which was held for 11 min, fol- dic (20:0), 11-eicosenoic acid (11-20:1), behenic (22:0), lowed by a 0.5 °C/min temperature ramp to 260 °C, which and lignoceric (24:0) acids. CPE18:1, CPA19:0 and was held for 9 min (60 min total run time). For the picol- CPE19:1 were present in both samples. Additionally, a inyl esters, an isothermal oven temperature of 260 °C was peak eluting with an equivalent carbon number of *25.8 used (50 min total run time). on the SP-2380 stationary phase was apparent in both oils.

123 1602 J Am Oil Chem Soc (2012) 89:1599–1609

This was a minor component in the T. populnea sample but significantly greater than the area of the second peak. An was more substantial in the cottonseed sample. The mass additional trace level component appeared in the T. pop- spectrum of this component suggested that it was the ulnea sample that eluted between the third and fifth peaks methyl ester of vernolic acid, which was confirmed by but was poorly resolved (Fig. 1 inset). Fragmentation data chromatography of a standard. Also, observed at even collected at times through this region again suggested this longer retention times were a few peaks that exhibited to be a 17:1 methyl ester. The last peak of this cluster, significant tailing on the polar stationary phases. These which appeared at trace levels in both samples, had a appeared to be methyl esters of hydroxylated fatty acids, as molecular ion at m/z 266 with fragmentation ions at silylation of the samples with N-methyl-N-(trimethyl- m/z 235 (M-31), 234 (M-32), 192 (M-74), and 150 (M-116) silyl)fluoroacetamide eliminated the peak tailing and in levels that suggested that this trace component was a reduced the elution times of these components. methyl ester of a hexadecadienoic acid. Chromatography of Two additional components were also present. These the methyl esters of A. syriaca seed oil, which is known to were comparable in size to several secondary fatty acids in have 1–2 % of 9,12-16:2 [17], yielded a peak of the the T. populnea sample, and they were apparent in the expected size with the same elution time, confirming this cottonseed oil albeit at much lower levels. From their peak’s identity. elution times, it appeared likely that these components The second heptadecyl component of interest eluted just corresponded to the heptadecyl monoene and diene acids before the 9-18:1 methyl ester (Fig. 1). Mass spectrometry that were mentioned by Fisher [1, 2] (Fig. 1). of this component produced a molecular ion of m/z 280 The first of these two components eluted between 17:0 with additional fragment ions at m/z 249 (M-31), 248 and 18:0 among a cluster of trace level components that (M-32), 206 (M-74), and 164 (M-116) in proportions were mostly present in both oils (Fig. 1 inset). Among this consistent with a 17:2 methyl ester. group of peaks, the first eluting component was detected in To determine the positions of the double bonds of these both samples, but mass spectrometry did not yield diag- components, DMDS adducts were prepared from the nostic peaks. Given its minor contribution to both oils, no FAME samples of both oils, and the components were additional work was conducted to determine its identity. separated on the SP-2380 column (Fig. 2). Because of the The mass spectra of the second and third peaks of the SG- mass added by the thiomethyl groups, higher chromato- 747 sample produced similar fragmentation patterns, with graphic temperatures were needed to elute these compo- molecular ions of m/z 282 and fragment ions at m/z 251 nents. This separates these derivatives from the unreacted (M-31), 250 (M-32), and 208 (M-74) in proportions that saturated esters but causes them to elute with other less suggested that these components were methyl esters of volatile materials and artifacts caused by the chemistry 17:1 [16]. Both peaks also appeared in the T. populnea [18]. The dithiomethyl heptadecyl methyl esters were sample, but in this sample the area of the first peak was found in the chromatograms from their expected mass ions

Fig. 1 FAME-FID chromatograms of T. populnea and SG-747 G. hirsutum 17:1 (cotton) seed oils. A Supelco 9-18:1 9,12-18:2 16:0 SP-2380 capillary column (30 cpe18:1 m 9 0.25 mm i.d. 9 0.20 lm film thickness) was used with a 9,12-16:2 temperature program starting at 18:0 170 °C (3 min), followed by a temperature ramp of 1 °C/min for 10 min, followed by a second temperature ramp of 17:2 ? 11-18:1

4 °C/min for 15 min to reach a 14:0 final temperature of 240 °C, 11-20:1 which was held for 15 min cpe19:1 17:1 ? 9-16:1 20:0 FID detector response

17:0 Thespesia 9,12,15-18:3 populnea seed oil

SG-747 cotton seed oil

4681012 Elution time, min

123 J Am Oil Chem Soc (2012) 89:1599–1609 1603 of m/z 376 (Fig. 2). As noted previously, thiomethylation an ion at m/z 199 (Fig. 3c). This identified the third com- can modestly improve the chromatographic resolution of ponent as 10-17:1. monoene isomers [19]. This was apparent in that the two The relatively high level of 8-17:1 in T. populnea oil and heptadecyl isomers in the cottonseed sample were almost the lack of the 10-17:1 in cottonseed oil indicated that the baseline separated after treatment with DMDS (Fig. 2), bis-thiomethyl derivatives elute in the same relative order whereas they were incompletely separated as FAME as the unsaturated 17:1 methyl esters on the SP-2380 sta- derivatives (Fig. 1 inset). Because of the much greater tionary phase. For both types of compounds, isomers elute concentration of one isomer in the T. populnea oil sample, more quickly when the double bond or the dithiomethyl the same components were not completely separated for moieties are positioned closer to the ester end of the mol- this sample (Fig. 2). As suggested by mass spectrometry of ecule. This elution order is the same as that observed for the FAME derivatives, a third minor dithiomethyl hepta- isomeric heptadecenoic methyl esters on the SP-2560 sta- decyl ester was observed in the T. populnea sample, but tionary phase [20]. was not observed in the cottonseed sample (Fig. 2). Reaction of DMDS with diene esters is generally more The mass spectrum of the first eluting DMDS-treated complicated than with monoene esters, and different 17:1 isomer had pronounced fragment ions of m/z 173 and products can be expected when the diene double bonds are 203 (Fig. 3a). Also apparent was an ion at m/z 171 char- positioned closer than three methylene carbon atoms from acteristic of the loss of methanol from the ester end the each other [21]. However, the first step of this chemistry m/z 203 fragment, which is expected for fragmentation of must be the reaction of one of the double bonds, and if the bis-dimethylthio fatty acid esters [18]. These ions place the reaction is stopped quickly enough, it is possible to detect thiomethyl groups at the 8 and 9 carbon atoms, indicating these products. Yamamoto et al. [22] have reported mass that the original fatty acid was 8-17:1. The second eluting spectra for these initial products from the reaction of DMDS-treated 17:1 component, the shoulder peak in the DMDS with 9,12-18:2. For the T. populnea oil sample T. populnea chromatogram but baseline separated in the treated with DMDS for 16 h, two pairs of dithiomethyl- cottonseed chromatogram (Fig. 2), had pronounced frag- monoene ester peaks were apparent that appeared to result ments at m/z 217 and 159 with a secondary ion at m/z 185, from the 17:2 and 18:2 components. Spectra identical to indicating the loss of methanol from the m/z 217 ion those reported by Yamamoto et al. [22] identified the pair (Fig. 3b). This places the thiomethyl groups at the 9 and 10 of derivatives derived from 9,12-18:2 (data not shown). acyl positions and identifies the original fatty acid as The second pair of peaks, which were incompletely sepa- 9-17:1. The third peak (present only in the T. populnea rated, eluted before the dithiomethyl derivative of 9-18:1 sample) had pronounced ions at m/z 231 and 145 with an (Fig. 2). Suitable spectra were obtained from their leading additional loss of methanol from the m/z 231 peak to give and trailing edges (Fig. 4). The first eluting component had

Fig. 2 Total ion chromatogram Elution of dithiomethyl FAME of the dimethyl disulfide- derivatives of: derivatized fatty acid methyl 9-18:1 esters for T. populnea seed and SG-747 cottonseed oils.

A Supelco SP-2380 capillary 8-17:1 column (100 m 9 0.25 mm 9-16:1 i.d. 9 0.20 lm film thickness) was used with a temperature program starting at 240 °C (11 min), followed by a 17:2 N/A temperature ramp of 0.5 °C/min for 40 min to reach a final temperature of 260 °C, which 9-17:1 was held for 15 min T. populnea

10-17:1 N/A 11-16:1 seed oil Total ion response

SG 747 cottonseed oil

32 33 34 35 36 37 38 Elution time, min

123 1604 J Am Oil Chem Soc (2012) 89:1599–1609

SCH Ave. 34.896 to 34.960 min 203 243 3 A100% A 12 CH3OOC 173 11 203 SCH3 SCH 131 9 3 CH3OOC 8 SCH 173 3 Ave. 37.226 to 37.300 min 279 100%

171 131 376 69

Relative abundance Relative 95 154 123 55

163 326 281 95 329 191 243 40 80 120 160 200 240 280320 360 400 m/z abundance Relative 374

217 B 100% Ave. 35.097 to 35.189 min 40 80 120 160 200 240 280 320 360 400 SCH 3 159 m/z 9 CH3OOC 10 217 SCH 3 SCH 203 3 159 B 9 CH3OOC 8 171 SCH3

279 376 100% Ave. 37.419 to 37.529 min 61 109 185 87 123 Relative abundance Relative

281 345 171

40 80 120 160 200 240 280 320 360 400 m/z 67 203 123

231 95 326 C 100% Ave. 35.244 to 35.318 min 231 SCH3 11 CH3OOC

10 abundance Relative 154 145 SCH3 343 374 229 247 145 199 40 80 120 160 200 240 280 320 360 400 m/z

61 95 376 Fig. 4 Mass spectra of the initial dimethyl disulfide products formed 170 from the 17:2 fatty acid methyl ester of T. populnea seed oil.

Relative abundance Relative 78 a Spectrum for 11,12-dithiomethyl-8-17:1. b Spectrum for 8,9- 281 dithiomethyl-11-17:1 40 80 120 160 200 240 280 320 360 400 m/z Although these partially reacted heptadecyl peaks were Fig. 3 Mass spectra of dimethyl disulfide-derivatized 17:1 fatty acid apparent in the T. populnea sample, they were not readily methyl ester peaks of T. populnea and G. hirsutum (cotton) seed oils. observed in the cottonseed sample (Fig. 3). This was likely a Spectrum for 8,9-dithiomethyl-17:0. Compound identified in both T. populnea and G. hirsutum seed oils. b Spectrum for 9,10- due to the much smaller concentration of 17:2 in this oil dithiomethyl-17:0. Compound identified in both T. populnea and combined with the loss in concentration that occurs G. hirsutum seed oils. c Spectrum for 10,11-dithiomethyl-17:0. because of the splitting of the component between two Compound identified only in T. populnea seed oil initial products and additional component loss due to incomplete and secondary reactions. To provide further confirmation of the identity of these a small mass ion of m/z 374 with a pair of fragment ions fatty acids, picolinyl esters were prepared and separated on that equaled the compound’s mass at m/z 131 and 243 the same capillary column (Fig. 5). The picolinyl esters of (Fig. 4a). These ions place the dithiomethyl groups at the 16:0, 9-16:1, 18:0, 9-18:1, 11-18:1, 9,12-18:2, 20:0, 11 and 12 acyl positions. The trailing peak also exhibited a CPE18:1, and CPE19:1 were identified from their known mass ion of m/z 374 but with fragment ions of m/z 171 and mass spectra [23, 24]. The 17:1 and 17:2 components m/z 203 (Fig. 4b), indicating that the thiomethyl groups were found from their expected mass ions of m/z 359 and were located at the 8 and 9 acyl positions. Hence, the m/z 357, respectively (Fig. 5). It was, however, only pos- heptadecyl diene component appeared to have its double sible to get diagnostic spectra for the largest of the 17:1 bonds between the 8 and 9 carbon atoms and between the isomers, as these components eluted on the tail of the 11 and 12 carbon atoms. picolinyl ester peak for 18:0, and for the 17:2 component,

123 J Am Oil Chem Soc (2012) 89:1599–1609 1605 both from the T. populnea sample (Fig. 6). The spectra of included in cottonseed oil surveys [4, 5], were also detec- the picolinyl derivative of the 17:1 peak exhibited a 26 u ted. In the cottonseed oil, these components were present at gap between its m/z 220 and 246 ions and had relatively a combined level of *1.8 %, whereas in T. populnea oils pronounced m/z 260 and 274 peaks (Fig. 6a), characteris- these components were present at levels of around tics that are typical of a picolinyl monoene ester [23–25]. *0.2 %. These features place this component’s double bond between the 8 and 9 carbon atoms. For the heptadecyl diene ester, 26 u gaps were apparent between the m/z 220 Discussion and 246 ions and between the m/z 260 and 286 ions (Fig. 6b), which places this component’s double bonds A few reports have appeared that discuss the composition between the 8 and 9 and 11 and 12 carbon atoms. These of T. populnea seed oil. Subbaram [6] reported an iodine results confirm the identifications of 8-17:1 and 8,11-17:2. value for T. populnea seed oil that seems low (71.5) given With the 17:1 and 17:2 components identified, analysis his and other reported fatty acid profiles for the oil [6–8]. of ten individual T. populnea and SG 747 seeds yielded The value obtained in this work (98.8) is in better agree- average oil compositions (Table 1). Expected levels of ment with the reported fatty acid data. The relatively low 16:0 and 9,12-18:2 were apparent in both oils. Compared iodine value for the SG-747 cottonseed oil was likely due with the SG 747 oil, T. populnea seed oil had an order of to the relatively low level of 9,12-18:2 present in the magnitude greater level of cyclopropenoid acids. Corre- variety (43 %) compared with the typical levels for this spondingly, the level of 9:18:1 was reduced in the T. pop- acid in cottonseed oils (48–56 %) [3]. The higher oil yields ulnea oil compared with the SG 747 oil. T. populnea had reported here compared with prior reports probably relate levels of 8-17:1 and 8,11-17:1 of 0.49 and 1.45 %, to the state of the seed, as the values were determined on respectively; whereas in the SG-747 oil sample, both of freeze-dried dehulled kernels. these acids were present at levels around 0.06 %. Levels of Subbaram reported the fatty acid composition of 14:0, 9-16:1, 18:0, 20:0, 22:0, and 24:0 in the T. populnea T. populnea seed oil to be 43.2 % 9,12-18:2, 32.5 % oil were similar to expected values for these acids in cot- 9-18:1, 21.4 % 16:0, 1.9 % 18:0 and 1.0 % 14:0 [6]. tonseed oils [3–5]. The vernolic acid concentration of 3 % Cornelius and coworkers later reported the oil composition was relatively high in the cottonseed oil, while it was just as 44 % 9,12-18:2, 29 % 16:0, 16 % 9-18:1; 2.0 % 18:0, detectable in the T. populnea seed oil at a level of 0.11 %. 1.0 % 14:0, and 8 % cyclopropenoid acids (measured as Hydroxylated fatty acids, which have been noted in CPE19:1 by hydrobromic acid titration) [7]. The main Malvaceae seed oils [26] but have not generally been difference in these reports is in the level of 9-18:1 and the

Fig. 5 Total ion chromatogram of the picolinyl fatty acid esters Fatty acids 9-18:1 of T. populnea seed and SG 747

as picolinyl esters 18:0 cottonseed oils. Results are for a Supelco SP-2380 capillary column (100 m 9 0.25 mm i.d. 9 0.20 lm film thickness) with an isothermal oven temperature of 260 °C 8,11-17:2 11-18:1 8-17:1 9-16:1 T. populnea

seed oil N/A 16:2 Total ion response Total

SG 747 cottonseed oil

20 21 22 23 24 25 26 27 Elution time, min

123 1606 J Am Oil Chem Soc (2012) 89:1599–1609

Fig. 6 Spectra of the A 220 260 unsaturated heptadecyl picolinyl CH OOC Ave. 22.849 to 22.922 min 2 esters from T. populnea seed oil. N 89 a Spectrum for the picolinyl 92 246 ester of 8-17:1. b Spectrum for 100% 359 the picolinyl ester of 8,11-17:2

164

274 108 260 55

26 u 316 Relative abundance Relative 302 206 246 330 130 288 186 220 344

40 80 120 160 200 240 280 320 360 400 m/z

B 246 286 Ave. 26.345 to 26.446 min CH2OOC 92 N 8 9 11 12 100% 220 260

357 164 108

67

55 81

Relative abundance Relative 26 u 26 u 151 260 300 122 206 314 178 246 220 286 328 342

40 80 120 160 200 240 280 320 360 400 m/z accounting for the cyclopropanoid acids. Subbaram’s that their 9,10-CPA17:0 acid is the 8-17:1 component higher level of 9-18:1 appears to be due to the inclusion of identified in this work. These compounds have the same the cyclopropenoid acids with this component. molecular mass and as methyl esters they would have More recently, Knothe et al. [8] have reported a com- similar mass fragmentation patterns. Yet, the ionization position that is similar regarding the common components, spectrum for the picolinyl derivative of these two compo- but also identifies a 9,10-methylene-hexadecyl fatty acid nents (which was also used by Knothe et al. [8]), would (9,10-CPA17:0) and CPA19:0 but does not report CPE19:1. have produced notable differences. Specifically, the mass This profile diverges somewhat from what is generally spectra of the picolinyl ester of 9,10-CPA17:0 would reported for Malvaceae plants [26], where CPE18:1 and exhibit an ion at m/z 247. This unusual ion occurs because CPE19:1 are usually observed and CPA19:0 is often noted. of fragmentation through the cyclopropyl ring, and it has The absence of CPE19:1 is somewhat surprising, especially been noted in the picolinyl spectra of both 9,10-CPA17:0 noting its presence in the middle of the biosynthesis path- and 9,10-CPA19:0 [23]. In this work, single ion monitoring way between CPA19:0 and CPE18:1 [27, 28]. for this ion identified the CPA19:0 picolinyl ester peak but As these authors did not discuss their rationale for it did not reveal a peak for the shorter CPA17:0 compo- making their component assignments [8], it is difficult to nent. The possible equivalence of these components is also comment on these differences. However, it seems possible suggested by the elution order reported by Knothe et al. [8]

123 J Am Oil Chem Soc (2012) 89:1599–1609 1607

Table 1 Fatty acid composition of T. populnea seed and SG-747 9,12-18:2 methyl ester (Fig. 7a). This was confirmed by cottonseed oils both the mass spectrum of the first peak (Fig. 7a), which Fatty acid T. populnea SG-747 cottonseed exhibited a small mass ion of m/z 308 and a fragment ion of seed oil oil m/z 277 as would be expected for the CPE19:1 methyl ester [14], and by chromatography of a standard of the CPA19:0 14:0 0.598 ± 0.034 1.08 ± 0.08 methyl ester, which eluted at the same time as the 9,12- 16:0 30.2 ± 0.5 27.1 ± 0.5 18:2 methyl ester. Initially, it was confirmed that CPA19:0 9-16:1 0.306 ± 0.015 0.650 ± 0.064 was present in both seed oils by single ion monitoring of 17:0 0.094 ± 0.003 0.083 ± 0.004 the FAME samples, i.e., the expected CPA19:0 mass ion of 8-17:1 0.49 ± 0.024 0.058 ± 0.005 m/z 310 and fragment ion of m/z 278 were detected under 9-17:1 tr tr the 9,12-18:2 methyl ester peak in both samples. (With 10-17:1 tr nd 9,12-18:2 methyl ester having a mass of 294 u, neither ion 9,12-16:2 tr tr could have been contributed by this ester.) Subsequent 18:0 2.61 ± 0.26 2.33 ± 0.18 analysis with the more polar SP-2560 stationary phase CPE18:1 4.90 ± 0.24 0.469 ± 0.081 allowed for the resolution and quantification of both 8,11-17:2 1.45 ± 0.07 0.067 ± 0.008 CPE19:1 and CPA19:0 (Fig. 7b). Knothe et al. [8], in 9-18:1 10.4 ± 0.5 17.8 ± 1.9 contrast, report that the CPA19:0 methyl ester elutes after 11-18:1 1.11 ± 0.06 0.926 ± 0.093 the 9,12-18:2 methyl ester on the SP-2380 phase; hence, it CPE19:1 1.53 ± 0.06 0.266 ± 0.009 seems possible that the CPE19:1 methyl ester could have CPA19:0 0.586 ± 0.074 0.315 ± 0.011 co-eluted under their 9,12-18:2 methyl ester peak, which 9,12-18:2 44.7 ± 1.0 43.1 ± 3.0 would account for their lack of mention of this component. 20:0 0.329 ± 0.016 0.311 ± 0.024 To test for this shift in elution behavior, the chromato- 9,12,15-18:3 0.06 ± 0.00 0.101 ± 0.018 graphic conditions indicated by Knothe et al. [8] were 11-20:1 tr 0.061 ± 0.005 repeated as part of this study. These conditions, however, 22:0 0.124 ± 0.005 0.152 ± 0.021 produced the same elution pattern reported above, i.e., with 24:0 0.094 ± 0.004 0.124 ± 0.012 the CPE19:1 methyl ester eluting before the co-eluting Vernolic acid 0.113 ± 0.002 3.03 ± 0.68 CPA19:0 and 9,12-18:2 methyl esters. Hence, the reasons Hydroxylated fatty acids 0.208 ± 0.021 1.84 ± 0.60 for the differences in elution profile and component iden- Total Sat 34.0 ± 0.7 31.2 ± 0.6 tity are unclear. Total Unsat 58.6 ± 0.8 62.9 ± 1.4 Fisher and Cherry [2] also mentioned the existence of Total CPE 6.43 ± 0.30 0.735 ± 0.150 epoxy acids in cottonseed oil. One of these acids, described Total a-oxidized acids 6.86 ± 0.28 0.593 ± 0.087 as an epoxide of 9,12-18:2, was likely vernolic acid, which was found in both the T. populnea and cottonseed oil Also detected in both oils were 15:0, 11-16:1, 21:0, 23:0, 25:0, and 26:0 samples and has been identified in other Malvaceae seed tr trace = \0.05 %; nd not detected oils [26]. The level of this acid in the SG-747 oil was 3.0 %, which is greater than the 0.85 % level noted by Fisher and Cherry [2], but is well within the range of values observed within seeds of Malvaceae plant species and by the roughly similar levels of these components in (0.0–9.5 %) [26]. The second epoxide mentioned by Fisher the reported distributions. and Cherry [2], an epoxide of 9:18:1, was not readily Knothe et al. [8] also noted the presence of an unknown apparent in either oil, although one cannot rule out the component in T. populnea oil. As the table caption in that possibility that this component is present at trace levels or work indicates that this unknown has an elution time that it co-elutes under another peak on both stationary greater than 24:0, this component may be vernolic acid. phases. Alternatively, if this component was included as an add-on Hydroxylated fatty acids have been noted previously in to the end of the table and its elution was faster, it may be Malvaceae seed oils [26]. Silylation of the FAME mixtures the 8,11-17:2 acid identified in this work. The relative level of both the cottonseed and T. populnea oil samples caused of this acid (1.2 %) also suggests one of these possibilities. a few tailing components to elute faster and yield sharper Differences in the reporting of CPA19:0 and CPE19:1 peaks (data not shown). Preliminary MS analysis of these were also unexpected, especially as both studies used the shifted peaks yielded spectra identical to those reported for same SP-2380 capillary column [8]. In this work, the 18-trimethylsiloxy-octadecenoic, 13-trimethylsiloxy-9,11- CPE19:1 methyl ester eluted before the 9,12-18:2 methyl octadecadienoic, and 9-trimethylsiloxy-10,12-octadecadie- ester, and the CPA19:0 methyl ester co-eluted with the noic methyl esters [14, 29]. Additional work, however, is 123 1608 J Am Oil Chem Soc (2012) 89:1599–1609

AB double bonds of these heptadecyl fatty acids would have cis configurations. In addition to the measurable levels of 8-17:1 and 8,11- 17:2, trace levels of 10-17:1 were detected in the seed oil of

9,12-18:2 + cpa19:0 T. populnea, which suggests that 11-18:1 (present at 9,12-18:2

cpe19:1 cpe19:1 *1.0 % level) also undergoes a-oxidation. In contrast, the small levels of 9-17:1 observed in both oils do not appear to be formed by a-oxidation, as there is no 10-18:1 pre- cursor acid. This acid is more likely formed by cross T. populnea reactivity of 17:0 with fatty acid desaturase 1, as occurs seed oil with 16:0 to form 9-16:1. The lack of trace levels of 20:0 10-17:1 and the lower levels of 8-17:1 and 8,11-17:2 in 20:0 cpa19:0 cottonseed oil compared with T. populnea seed oil indi- cates that a-oxidation of unsaturated fatty acids is less prone to occur in this plant. This difference exists despite the relatively high level of cyclopropenoid fatty acid a-oxidation that occurs in both plants. Because monoene and diene fatty acids are more stable than cyclopropenoid SG 747 fatty acids, T. populnea might prove to be useful as a model cottonseed system in studies of fatty acid a-oxidation. oil Unsaturated heptadecyl fatty acids with a double bond in the 8-position have been noted in a few plants. In the seed oil of Pachira aquatica, a Malvaceae plant from the 10 11 12 25 27 29 Bombacoideae subfamily, 8-17:1 and 8,11-17:2 have been Elution times, min noted to be present at around 0.1 % each [30]. 8-17:1, 8,11- 17:2, and 8,11,14-17:3 have been reported in the seed oil of Fig. 7 Partial FAME-FID chromatograms showing elution profiles about the 9,12-18:2 methyl ester region for T. populnea and SG-747 Salvia nilotica at a total concentration of about 0.5 % [31], (cotton) seed oils. a Separation on a Supelco SP-2380 column (30 m and the trienoic acid has been reported in Thymus vulgaris 9 0.25 mm i.d. 9 0.20 lm film thickness) with a temperature at a concentration of *2%[32]. In these latter Lamiaceae program starting at 170 °C (3 min), followed by a temperature ramp family plants, 2-hydroxyl octadecyl fatty acids are also of 1 °C/min for 10 min, followed by a second temperature ramp of 4 °C/min for 15 min to reach the final temperature of 240 °C, which present and occur at levels greater than the corresponding was held for 15 min. b Separation on a Supelco SP-2560 column 17-carbon a-oxidation products [31, 32]. Their presence (100 m 9 0.25 mm i.d. 9 0.20 lm film thickness) with a temperature supports the proposal that these acids have a role as program starting at 185 °C (35 min), followed by a temperature ramp intermediates or side products in a-oxidation [33]. That of 0.5 °C/min to 240 °C, which was held for 15 min these 2-hydroxy-octadecyl fatty acids were not readily observed in this study and have not been noted in most needed to confirm these identifications. Consequently, for prior Malvaceae seed oils studies suggests that a funda- the moment these components are included in the distri- mental differences exist between a-oxidation mechanisms butions only as total hydroxylated fatty acids (Table 1). in these plant families. In a few prior studies of Malvaceae The existence of double bonds at the 8-position strongly plant seed oils, 17:1 and, less frequently, 17:2 fatty acids suggests that 8-17:1 and 8,11-17:2 are formed by oxidation have been mentioned, but without any consideration of the of the a-carbon atoms of 9-18:1 and 9,12-18:2, respec- positions of the double bonds [34–36]. In these reports, tively. It is generally accepted that CPE18:1 is formed by levels of 17:1 as high as 2.6 % and levels of 17:2 as high as a-oxidation of CPE19:1 [27, 28]. Some of the original 4.3 % have been noted. While is seems likely that at least work on these transformations in Sterculia foetida seed oil some portion of these fatty acids correspond to the noted that while a-oxidation occurs extensive with cyclo- monoene and diene isomers reported here, a re-survey of propenoid fatty acids, it generally is not observed with these plants will be needed to confirm this. other classes of fatty acids [28]. The identification of these heptadecyl acids in T. populnea indicates that a-oxidation Acknowledgments The author thanks C. Grimm and S. Lloyd of unsaturated fatty acid classes does occur to a measurable (SRRC) for extended access to the mass spectrometer. Mention of trade names or commercial products in this paper is solely for the degree, at least within some seed types of the Gossypieae purpose of providing specific information and does not imply rec- tribe. This synthesis mechanism also indicates that the ommendation or endorsement by the US Department of Agriculture.

123 J Am Oil Chem Soc (2012) 89:1599–1609 1609

References 19. Christie WW (1997) Structural analysis of fatty acids. In: Christie WW (ed) Advances in lipid methodology. vol 4. Oily Press, 1. Fisher GS, Schuller WH (1981) Gas chromatographic analysis of Dundee pp 119–69 cyclopropenoid acids in cottonseed oils. J Am Oil Chem Soc 20. Delmonte P, Hu Q, Kia A-RF, Rader JI (2008) Preparation, 58:943–946 chromatographic separation and relative retention times of cis/ 2. Fisher GS, Cherry JP (1983) Variation of cyclopropenoid fatty trans heptadecaenoic (17:1) fatty acids. J Chromatogr A acids in cottonseed lipids. Lipids 18:589–594 1214:30–36 3. O’Brien RD, Jones LA, King CC, Wakelyn PJ, Wan PJ (2005) 21. Vincenti M, Guglielmeti G, Cassani G, Tonini C (1987) Deter- Cottonseed oil. In: Shahidi F (ed) Bailey’s industrial oil and fat mination of double bond position in diunsaturated compounds by products, Edible oil & fat products: edible oils, 6th edn., vol 2. mass spectroscopy of dimethyl disulfide derivatives. Anal Chem Wiley, Hoboken, pp 173–279 59:694–699 4. Lukonge E, Labuschagne MT, Hugo A (2007) The evaluation of 22. Yamamoto K, Shibahara A, Nakayama T, Kajimoto G (1991) oil and fatty acid composition in seeds of cotton accessions from Determination of double-bond positions in methylene-interrupted various countries. J Sci Food Agric 87:1732–1743 dienoic fatty acids by GC-MS as their dimethyl disulfide adducts. 5. Dowd MK, Boykin DL, Meredith WR Jr, Campbell BT, Bourland Chem Phys Lipids 60:39–50 FM, Gannaway JR, Glass KM, Zhang J (2010) Fatty acid profiles 23. The AOCS Lipid Library, Picolinyl esters of fatty acids—archive of of cottonseed genotypes from the national cotton variety trials. mass spectra. http://lipidlibrary.aocs.org/ms/arch_pic/index.htm J Cotton Sci 14:64–73 (checked 3/20/2012) 6. Subbaram MR (1954) Analysis and characterization of the oil 24. Christie WW, Brechany EY, Johnson SB, Holman RT (1986) A from the seeds of Thespesia populnea. Proc Ind Acad Sci comparison of pyrrolidide and picolinyl ester derivatives for the 39A:301–304 identification of fatty acids in natural samples by gas chroma- 7. Cornelius JA, Hammonds TW, Leicester JB, Ndabahweji JK, tography-mass spectrometry. Lipids 21:657–661 Rosie DA, Shone GG (1970) New tropical seed oils. III. Com- 25. Christie WW, Brechany EY, Holman RT (1987) Mass spectra of ponent acids of leguminous and other seed oils (continued). J Sci the picolinyl esters of isomeric mono- and dienoic fatty acids. Food Agric 21:49–50 Lipids 22:224–228 8. Knothe G, Rashid U, Yusup S, Anwar F (2011) Fatty acids of 26. Badami RC, Patil KB (1980) Structure and occurrence of unusual Thespesia populnea: mass spectrometry of picolinyl esters of fatty acids in minor seed oils. Prog Lipid Res 19:119–153 cyclopropene fatty acids. Eur J Lipid Sci Technol 113:980–984 27. Yano I, Morris LJ, Nichols BW, James AT (1971) The biosyn- 9. King TJ, de Silva LB (1968) Optically active gossypol from thesis of cyclopropane and cyclopropene fatty acids in higher Thespesia populnea. Tetrahedron 3:261–263 plants. Lipids 7:35–45 10. Boonsri S, Karalai C, Ponglimanont C, Chantrapromma S, 28. Bao X, Katz S, Pollard M, Ohlrogge J (2002) Carbocyclic fatty Kanjana-opas A (2008) Cytotoxic and antibacterial sesquiter- acids in plants: biochemical and molecular genetic characteriza- penes from Thespesia populnea. J Nat Prod 71:1173–1177 tion of cyclopropane fatty acid synthesis of Sterculia foetida. 11. AOCS (1998) Method Cd 1b-87. In: Firestone D (ed) Official Proc Natl Acad Sci 99:7172–7177 methods and recommended practices of the AOCS, 5th edn. 29. Sessa DJ, Gardner R, Kleiman R, Weisleder D (1977) Oxygen- AOCS, Champaign ated fatty acid constituents of soybean phosphatidylcholines. 12. The AOCS Lipid Library, Mass spectra of methyl esters of fatty Lipids 12:613–619 acids. Part 7. Derivatization of the double bonds. http://lipidlibrary. 30. Spitzer V (1991) GC-MS characterization (chemical ionization aocs.org/ms/ms04/index.htm (checked 3/20/2012) and electron impact modes) of the methyl esters and oxazoline 13. Destaillats F, Angers P (2002) One-step methodology for the derivatives of cyclopropenoid fatty acids. J Am Oil Chem Soc synthesis of FA picolinyl esters from intact lipids. J Am Oil 68:963–969 Chem Soc 79:253–256 31. Bohannon MB, Kleiman R (1975) Unsaturated C18 a-hydroxy 14. The AOCS Lipid Library, Methyl esters of fatty acids—archive acids in Salvia nilotica. Lipids 10:703–705 of mass spectra. http://lipidlibrary.aocs.org/ms/arch_me/index.htm 32. Smith CR Jr, Wolff IA (1969) Characterization of naturally (checked 3/20/2012) occurring a-hydroxylinolenic acid. Lipids 4:9–14 15. AOCS (1998) Method Ce 1e-91. In: Firestone D (ed) Official 33. Harwood JL, Russell NL (1984) Lipids in Plants and Microbes, methods and recommended practices of the AOCS, 5th edn. chap 5, pp 118–129. Allen and Unwin, London AOCS, Champaign 34. Bohannon MB, Kleiman R (1978) Cyclopropene fatty acids of 16. McCloskey JA (1970) Mass spectrometry of fatty acid deriva- selected seed oils from Bombacaceae, Malvaceae, and Sterculi- tives. In: Gunstone FD (ed) Topics in lipid chemistry, vol 1. aceae. Lipids 11:270–273 Wiley, New York, pp 369–440 35. Ralaimanarivo A, Gaydou EM, Bianchini J-P (1982) Fatty acid 17. Chisholm MJ, Hopkins CY (1960) 11-Octadecenoic acid and composition of seed oils from six Adansonia species with par- other fatty acids of Asclepias syriaca seed oil. Can J Chem ticular reference to cyclopropane and cyclopropene acids. Lipids 38:805–812 17:1–10 18. Francis GW (1981) Alkylthiolation for the determination of 36. Gaydou EM, Ramanoelina ARP, Rasoarahona JRE, Combres A double-bond position in unsaturated fatty acid esters. Chem Phys (1993) Fatty acid composition of Sterculia seeds and oils from Lipids 29:369–374 Madagascar. J Agric Food Chem 41:64–66

123