Anal. Chem. 2007, 79, 4696-4701
Technical Notes
Enantioselective Determination of Anteiso Fatty Acids in Food Samples Saskia Thurnhofer,† Georg Hottinger,‡ and Walter Vetter*,†
Institute of Food Chemistry, University of Hohenheim, Garbenstrasse 28, 70599 Stuttgart, Germany, and BGB Analytik, Lettenstrasse 97, CH-8134 Adliswil, Switzerland
Anteiso fatty acids (aFAs)slong-chain carboxylic acids with a methyl branch on the (n - 2)-carbonsare among the most simple fatty acids that are chiral. The most frequently occurring aFAs in food are 12-methyltetradecanoic acid (a15:0) and 14-methylhexadecanoic acid (a17:0), structures where the asymmetric carbon is more than 10 carbons separated from the polar head group. Previously, only enantioseparation of 4-methyl-substituted carboxylic fatty acids has been reported by gas chromatography. Here we present the first direct partial enantioresolution of synthesized racemic a15:0-a17:0 on a capillary column coated with 50% heptakis(6-O-tertbutyldimethylsilyl-2,3-di-O-methyl)-β-cyclodextrin diluted in OV1701. Synthesized (S)-(+)-enantiomers were used to demonstrate that the elution order was (R)- prior to (S)-enantiomers. Using this system, food samples (butter, goat’s milk fat, suet, human milk, seal oil, cod liver oil) known to contain aFAs were analyzed. Prior to the enantioselective gas chromatography, unsaturated fatty acids were preseparated by urea complexation, silver ion high performance liquid chromatography (Ag+-HPLC), or both from food samples. The fractions of the food samples enriched with methyl-branched fatty acids were then analyzed by GC/MS in the SIM mode. The measurements confirmed that the (S)-enantiomer of a15:0 (ee >96%), a16:0, and a17:0 (ee >90%, respectively) dominated in all samples. While the (R)-enantiomers could not be identified in samples from ruminants and human milk, their presence could be established in cod liver and seal oil (ee 99% was determined by TLC (toluene/2-propanol 7:2) and high-temperature gas chromatography as described previously.31 The pure, laboratory-made CSPs were diluted in polysiloxane and coated on fused-silica capillary columns according to Blum and Aichholz32 with the following three 30 m × 0.25 mm i.d. columns prepared and evaluated: (i) 20% β-TBDM in SE52, film thickness 0.2 µm (BGB 176SE, BGB Analytik, Adliswil, Switzerland); (ii) 20% β-TBDM in PS086, film thickness 0.1 µm; (iii) 50% β-TBDM in OV1701, film thickness 0.2 µm. The GC oven programs were optimized by lowering temperature stepwise until the best enantioresolution was obtained for a15:0-ME. After elution of a15:0-ME, the oven program was raised by 1 °C/min for elution of a16:0-ME and a17:0-ME within 650 min. For the 50% β-TBDM column, the oven program optimized for a15:0-ME was the following: 60 °C at injection for 1 min, then ramped at 10 °C/min to 115 °C (hold time 400 min), and finally ramped at 1 °C/min to 137 °C (hold time 221 min). A second GC oven program employed in this study was as follows: 60 °C (hold time 1 min), then ramped at 10 °C/min to 120 °C (hold time 525 min), finally ramped at 1 °C/min to 145 °C (hold time 90 min). RESULTS AND DISCUSSION Investigation of >10 capillary columns coated with different modified cyclodextrins as CSPs failed to separate racemic anteiso fatty acid standards (data not shown). Subsequent comprehensive testing of a13:0 and a14:0 methyl ester (a13:0-ME, a14:0-ME), respectively, in the lab of the late Winfried Ko¨nig brought the attention to heptakis(6-O-tert-butyldimethylsilyl-2,3-di-O-methyl)β-cyclodextrin (β-TBDM). Introduced by Mosandl and co-workers and applied to the enantiomeric resolution of a wide range of chiral compounds including flavor-relevant short chain γ-methylbranched carboxylic acids,16,17,30 it is the only CSP that has partially resolved a13:0-ME and a14:0-ME to date. In the present study we focused on the use of different columns coated with β-TBDM. Initial evaluation by GC/FID clarified that the commercially available 20% β-TBDM column (BGB 176SE) yielded slight partial resolution of a15:0-ME at retention times in the range of 400 min. (29) Campra-Madrid, P.; Guil-Guerrero, J. L. Chromatographia 2002, 56, 673677. (30) Dietrich, A.; Maas, B.; Messer, W.; Bruche, G.; Karl. V.; Kaunzinger, A.; Mosandl, A. J. High Resolut. Chromatogr. 1992, 15, 590-593. (31) Vetter, W.; Klobes, U.; Luckas, B.; Hottinger, G. J. Chromatogr., A 1999, 846, 375-381. (32) Blum, W.; Aichholz, R. J. High Resolut. Chromatogr. 1990, 13, 515-518.
4698 Analytical Chemistry, Vol. 79, No. 12, June 15, 2007
Table 1. r-Values and Retention Times of Anteiso Fatty Acids and Nonchiral Isomers on 50% β-TBDM in OV1701 Using Two Different GC Oven Programsa GC oven program 1
GC oven program 2
FAMEs
retention time (min)
retention time (min)
(R)-a15:0/(S)-a15:0 i15:0 15:0 (R)-a16:0/(S)-a16:0 i16:0 16:0 (R)-a17:0/(S)-a17:0 i17:0 17:0
408.1/411.0 417.8 438.8 493.5/494.8 499.0 528.6 618.3/619.6 626.6 not eluted
R-value 1.007 1.003 1.002
290.6/ 293.0 303.2 357.4 525.8/ 530.7 542.6 565.4 619.8/621.0 624.8 not eluted
R-value 1.008 1.009 1.002
a GC oven program 1: 60 °C (1 min)-10 °C/min-115 °C (400 min)-1 °C/min-137 °C (221 min). GC oven program 2: 60 0 °C (1 min)-10 °C/min-120 °C (525 min)-1 °C/min-145 °C (90 min).
Figure 2. Elution of a15:0. (a) highly enantiopure (S)-a15:0 as well as (b) 5 and (c) 10% racemate spiked into (S)-a15:0.
Moreover, the methyl esters provided better chiral resolution than the less volatile ethyl ethers. We also studied d3-labeled methyl esters but the slightly higher volatility did not improve the enantioseparation on the medium polar β-TBDM columns.25 Based on this result two further β-TBDM columns were produced. With a thinner film, the first column was designed for lower aFA retention times, thus allowing for lower oven temperatures, a
Figure 3. Enantioselective determination of anteiso fatty acids in (a) butter and (b) butter spiked with racemic anteiso fatty acids. Branchedchain fatty acids were enriched by urea complexation and Ag+-HPLC prior to injection.
parameter known to have crucial influence on enantioselective analysis. However, no resolution of aFA enantiomers was achieved by this column. By contrast, the column with 50% CSP yielded a better enantioseparation than the 20% β-TBDM initially tested. The 50% β-TBDM was thus deemed best and was selected for further investigation using GC/EI-MS in the SIM mode. We screened m/z 74 and m/z 87 which are the most abundant fragment ions in the GC/EI mass spectra of the methyl esters of all saturated and monoenoic fatty acids.24 In addition, the molecular ions and other fragment ions were recorded as well (see below). The GC oven temperature was lowered stepwise and optimized. Our goal was the elution of a15:0-ME, a16:0-ME, and a17:0-ME within 650 min (maximum run time on the GC/MS system). Along with two fast programs between the runs to exclude memory peaks, two analyses could be run daily. Two GC oven programs were developed using these criteria (Table 1). With the ultimate goal of determining the enantiomeric composition of aFAs in food samples in mind, we also recorded the retention times of isomeric fatty acids, which, if coeluting, would prevent the determination (Table 1). As mentioned before, m/z 74 and 87 are suited to identify the methyl esters of all saturated fatty acids whereas the low-abundant molecular ions are specific for isomeric fatty acids. The data in Table 1 demonstrate that iso- and straight-chain isomers were well separated from anteiso fatty acids despite the isothermal elution. On β-TBDM, the retention times of isomers increased in the order anteiso isomers < iso isomer < straight-chain isomer (Table 1).
This elution order was different from both more polar (100% cyanopropyl) and nonpolar (DB-5-like) phases from which iso fatty acids are less retarded than anteiso fatty acids. Since iso fatty acids eluted very shortly after the anteiso isomers and an isothermal elution was best-suited for the enantioseparation of racemic aFAs, we concluded that shortening of the column could not be carried out because this would have led to a (partial) coelution of anteiso and iso isomers. This can be seen from the fact that faster GC oven programs led to a partial coelution of (S)-a15:0-ME and i15:0-ME (see Supporting Information). Elution orders of enantiomers cannot be predicted on β-TBDM. Pure synthesized (S)-enantiomers and pure (S)-enantiomers spiked with racemate were used to determine that the (R)enantiomer eluted prior to the (S)-aFA. Since (S)-enantiomers were expected to be more relevant in food, this elution order was favorable for us because of the slight peak tailing observed (Figure 2). Another interesting phenomenon was observed. Although the neat (S)-enantiomer of a15:0 was injected, we observed a small preshoulder peak (Figure 2a). In order to determine what traces of (R)-enantiomer can be identified under these chromatographic conditions, we spiked low amounts of the racemate into the (S)-enantiomer (Figure 2b,c). As can be seen from the central chromatogram in Figure 2, ee ) 95% can be distinguished from pure (S)-a15:0. At ee ) 90% (Figure 2c), a significant valley was obtained between the enantiomeric peaks. Thus, we concluded that proportions as low as 2% (R)-15:0 can be identified in samples Analytical Chemistry, Vol. 79, No. 12, June 15, 2007
4699
despite the long retention times (∼5 to ∼7 h) and broad peak shapes (∼5 min) obtained for a15:0. The chiral resolution of the higher anteiso fatty acids was decreased, but similar spiking experiments (see below) indicated that 5% or less of the (R)-enantiomers of a16:0 and a17:0 could be identified in samples dominated by the (S)- enantiomer. On the basis of this initial evaluation, we attempted to determine the enantiomeric composition of aFAs in food samples. To maintain the best quality control achievable, fatty acid methyl esters obtained from the food samples by transesterification were not analyzed directly by GC/MS. Even though isomer-selective masses could be used, anteiso fatty acids are only present as minor compounds in food (96% in favor of the (S)-enantiomers for a15:0 and a17:0 in butter. It is also noteworthy that the low-abundant peak of a16:0 was only detected in fractions enriched after urea complexation and Ag+-HPLC. Based on the peak areas of aFAs in the butter sample, the concentration of a16:0 was
