Running Title: Determining the Relative Amounts of Positional Isomers in Complex Mixtures of Triglycerides Using Reverse-Phase HPLC-MS-MS


A reverse-phase high performance liquid chromatography-tandem mass spectrometry (RP-HPLC-MS-MS) method has been refined for the positional analysis of complex mixtures of triglycerides (TAGs). This method has the advantages of speed, ease of automation, and specificity over traditional digestion-based methods for the positional analysis of TAGs. Collisional-induced decomposition (CID) of ammoniated TAGs in an ion trap mass spectrometer produces spectra that are dependant on fatty acid position. Dominant diglycerol (DAG) fragments are formed from the loss of a fatty acid moiety from the ammoniated TAG species. Loss of fatty acids in the outer positions is favored over the loss of fatty acids in the central position. The combination of RP-HPLC and CID produce spectra that are free of isotope effects that can complicate spectral interpretation in existing methods. The combination also provides selectivity based on the chromatographic fractionation of TAGs, in addition, to the selectivity inherent in the CID process. Proof-of-concept experiments were performed with binary mixtures of TAGs from the SOS/SSO, OSO/OOS, and the PSO/POS/SPO positional isomer systems. Plots of fractional DAG fragment intensities vs. fractional composition of the binary mixtures are linear. These plots were used to determine the fractional composition of each of these isomeric systems in a variety of vegetable oils and animal fats. Current limitations, future developments, and applications of this method are discussed. Introduction Triglycerides (TAGs) consist of three fatty acid moieties attached to a glycerol backbone. The main biological function of TAGs is to serve as an energy source. Dietary TAGs are digested, reconstituted, and packaged as chylomicrons prior to entering the blood stream. Ultimately, the TAGs are delivered to cells in need of energy or stored as reserves in adipose tissue. Evidence is mounting that suggests TAG absorption , metabolism 1, , and atherogenic potential 2, 3, 13-23 (tendency for deposition of lipoproteins on the artery walls) may be influenced by fatty acid position. Additional work is needed in these areas to obtain a more complete understanding of the relationship between dietary lipids and heart disease. Development of efficient methods for the positional analysis of individual TAG species will facilitate the advance of these studies. Positional analysis of TAGs has traditionally been performed through digestion of the outer two fatty acids and subsequent HPLC analysis of the resulting 2monoglycerides and free fatty acids 2, 8, . Commonly, complex mixtures of TAGs are digested in a single step and the overall fatty acid composition is compared to the position-specific fatty acid composition. These data have revealed general patterns, such that mono-unsaturated fatty acids are overwhelmingly favored in the center position and saturated fatty acids are favored in the outer positions for most animal and vegetable oils, suggesting that fatty acids are selectively attached to the glycerol backbone. Extensive work on the details of TAG biosynthesis 28-33 has explained the positional dependence of fatty acid composition. These studies have established that unique enzymes catalyze the attachment of fatty acids onto each of the positions, and that these enzymes possess different fatty acid selectivities . The digestion-based methods of positional analysis discussed above have been plagued with problems associated with fatty acid migration during digestion 43, . In addition, these methods are cumbersome and time-consuming for investigations focusing on individual TAG species, since the mixtures must be fractionated prior to hydrolysis. Furthermore, the analysis of the hydrolysis products of co-eluting TAG species often produces ambiguous results. Mass spectrometric methods have recently been developed that are less labor intensive, more conducive to performing positional analyses on individual TAG species, and more easily automated. Evershed and coworkers have used high performance liquid chromatography atmospheric pressure chemical ionization mass spectrometry (HPLCAPCI-MS) for this purpose . The protonated TAGs formed during the APCI process acquire sufficient energy to fragment in the source. The major ions formed in the APCI process are the diglycerol (DAG) fragments, in which one of the fatty acid groups leaves the protonated TAG as a neutral fatty acid. Evershed’s data show that fragmentation is less likely to occur at the center position than at the outer positions. The authors were able to predict which positional isomer was most abundant based on the DAG fragment of lowest relative abundance. The drawback of the APCI method is that many co-eluting TAGs produce common DAG fragments. As a result, it is often difficult to assign peaks to specific TAGs. In addition, it is not possible to de-convolute peak intensities, which would be necessary for quantification of positional isomers. These difficulties are overcome to a great extent by combining RP-HPLC with the use tandem mass spectrometry (MS-MS). Kallio and coworkers have used ammonia negative ion chemical ionization in conjunction with tandem mass spectrometry (NICI-MS-MS) for their analysis 48, . The TAG sample is directly inserted into an ammonia CI source of a triple-quadrupole mass spectrometer via a probe and all of the TAGs are ionized under negative ion ammonia CI conditions, simultaneously. De-protonated TAGs are formed in the source and mass selected by the first quadrupole. These ions are transmitted to the second quadrupole for collision-induced dissociation (CID). The product ions are analyzed by a third quadrupole to produce a mass spectrum of the CID products. The most abundant fragment ions are deprotonated fatty acid fragments, but DAG-containing fragments (the information rich fragments) are fairly abundant peaks. CID affords a high level of selectivity, since ions of a particular mass-to-charge ratio are selected by the mass analyzer and subjected to CID. As a result, the analysis can be targeted to TAGs of a particular molecular weight. Direct analysis by tandem mass spectrometry without any chromatography to fractionate the TAGs can provide a wealth of information in a relatively short period of time (15-20 min.), even for very complex extracts. In this time frame a series of CID mass spectra can be acquired on several [M-H] peaks of interest. However, this method has two main limitations. First of all, C isotope contributions from TAGs that are lower in molecular weight by two mass units must be meticulously subtracted from the measured intensities, limiting the precision of the method. The DAG ions of interest in the CID spectra are often overwhelmed by other C isotope-containing DAG fragments. Secondly, TAGs with the same molecular weight and one of the three fatty acids in common, such as OOS and POA1 will yield common DAG CID fragments, interfering with positional analysis. Nevertheless, Kallio was able to construct plots for some positional isomers that correlated the relative abundances of the DAG ions with the fractional composition. By contrast, the use of RP-HPLC prior to CID analysis completely eliminates the problems associated with C isotope peaks because TAGs that differ by two mass units are generally completely separated chromatographically. The second limitation is also avoided in many cases by adequate separation of the interfering species. Data highlighting these advantages are presented in this work. The analysis of TAGs by electrospray (ESI) mass spectrometry was first reported by Duffin and Henion . They showed that the addition of ammonium acetate to the TAG sample could produce intense [M+NH4] ions, that are efficiently dissociated in CID experiments to form dominant DAG fragments indicative of the loss of the fatty acid moieties. Using an ESI-double focusing sector instrument, Cheng and Pittenauer observed that CID spectra of [M+NH4] from positional isomers were indistinguishable (PPO and POP produced the same CID spectra) . Han and Gross obtained similar results with a triple-quadrupole instrument analyzing at the CID products of lithium adducts . However, in an HPLC-ESI-MS-MS study with a triple-quadrupole instrument, Hvattum observed that the relative abundances of DAG fragments from ammoniated TAGs are dependant on fatty acid position , similar to Kallio’s NICI experiments. Our experiments using an ion trap mass spectrometer also show that the CID spectra ammoniated TAGs are dependant on fatty acid position. We report the use of RP-HPLC-ESI-MS-MS for fractional composition analysis of TAG positional isomers. This method combines the advantages of efficient fractionation of the TAGs via RP-HPLC and enhanced selectivity provided by the CID process (DAG fragments derive from TAGs of only the selected molecular weight). This added degree of selectivity enables the quantification of many systems of positional isomers that would otherwise be difficult. The relative intensities of the DAG fragments in the CID spectrum of mixtures of positional isomers can be used to measure their relative abundances. Standard mixtures of positional isomers were analyzed and calibration plots of fractional DAG fragment intensities vs. fractional composition of three sets of positional isomers were constructed and used to determine the fraction composition of positional isomers in various vegetable oils and animal fats. Experimental Procedures The TAG designations that will be used throughout this paper consist of three letters, each indicating the presence of a particular fatty acid. The middle letter designates which fatty acid is in the center position. No distinction is made between the outer two positions. The standard symbols and one letter abbreviations used throughout the paper are listed as a footnote in the beginning of the paper. HPLC grade methanol, n-propanol, n-butanol, methyl t-butyl ether (MTBE) were purchased from Acros (NJ ). Ammonium formate (99%) was also purchased from Acros. Purified TAGs (LnLnLn, LLL, OOO, PPO, POP, SOS, SSO, POS, PSO, SPO, SSO, SOS, APO, POA, and PAO) were purchased from Larodan (Malmo, Sweeden). The unsaturated fatty acids contained in purchased TAGs were all in the cis configuration. The current study did not investigate possible differences in the relative intensities of diglycerol fragments between TAGs containing cis vs. trans fatty acids. Various vegetable oils were purchased at the local supermarket. Fats from pork, chicken, and beef were carved from meat products purchased at the local supermarket. Standards solutions of each of the pure TAGs were prepared in n-propanol at concentrations of 100±2 μM. Diluted standards (10.00 μM) for each were prepared in methanol/ammonium formate. These diluted standards were used to prepare standard solutions for a variety of different experiments that were performed in this work. Extracts from various vegetable oils were prepared in 20 ml vials by dissolving a drop of oil in 15 ml n-PrOH. The extract was subsequently diluted 1:100 with methanol. A mixed extract was also prepared in this manner from a one drop each of peanut oil and corn oil. The fats were extracted into MTBE at 120 °C for one hour using the 20 ml vials and a variable temperature heating manifold. A drop of the extract was dissolved in 15 ml n-butanol and thoroughly dissolved. The resulting solution was diluted 1:100 in methanol in preparation for analysis. Mass Spectrometer Parameters A Thermofinnigan LCQ Advantage ion trap mass spectrometer (Sunnyvale, CA) was used to detect and characterize the TAGs. The connection from the syringe pump (direct injection experiments) or the HPLC column to the ESI source was made through a 1/16′′ stainless steel zero-dead volume union and a 30 cm long, 50 μm ID, 185 μm OD, segment of fused silica capillary. The end of the fused silica capillary was fed into the ESI interface through a metal sheath. The tip of the capillary was carefully cut to provide a uniformly shaped tip. The tip of the capillary was positioned so that it was at the edge of the metal sheath. The electrospray cone is formed by applying a potential difference between the metal sheath and the ion transfer tube that focuses the ions into the mass analyzer. The operating parameters of the ion trap mass spectrometer were as follows; capillary temperature (280 °C), spray voltage (4.00 kV), sheath gas (30 cm/min). CID was performed at a relative collision energy of 28 (unitless quantity) unless otherwise stated. This value should be applicable to other LCQ systems, assuming that the instrumental calibration procedures described by the manufacturer are carefully followed. HPLC Parameters A low-flow Shimadzu (Kyoto, Japan) HPLC system, which included a SCL-10A vp controller, two LC-10AD vp pumps, a SIL-10AD vp auto injector, and a 10 cm, 1 mm ID, 3 μm particle size, C18 BetaBasic column from ThermoElectron Corporation (Sunnyvale, CA) was used to separate the TAGs. The HPLC was operated at a flow rate of 35 μl/min (low flow pumps/no splitting necessary). A gradient elution was utilized, consisting of mobile phase A {methanol/n-propanol (80:20, by vol), saturated ammonium formate (≈ 1 mM), pH 7} and mobile phase B {methanol/n-propanol (20:80, by vol), saturated ammonium formate, pH 7. The Binary Gradient was as follows; 0 min/0 % A, 4 min/10 %, 36 min/60 % A, 38-40 min/85 % A. Injection volume was 5 μl for all samples (0.5-5.0 pmol of TAG on-column). CID spectra of the Standards by direct infusion The 10 μM diluted standards were each analyzed by direct infusion ESI-MS-MS (no HPLC column) using a syringe pump with a 500 μl syringe. The flow rate was 5 μl/min. The CID spectra were acquired under the conditions described above. Analysis of a standard mixture and an oil extract by RP-HPLC-ESI-MS-MS A mixture containing each of the standard TAGs at concentrations of about 0.7 μM was prepared in methanol using the diluted standards described above. This mixture was analyzed by RP-HPLC-MS in the MS-only mode and in a targeted RP-HPLC-MSMS mode. In the MS-only mode a mass spectrum of the ions formed in the electrospray process was produced at a rate of about 1 spectrum per second throughout the course of the chromatographic run. The targeted MS-MS analysis was developed based on the retention times determined from the experiment performed in the MS-only mode. In this targeted mode the mass spectrometer was programmed to select the appropriate m/z ratio for CID analysis during a 2-3 minute chromatographic time window corresponding to the eluting TAG. The mixed extract of peanut/corn oil was also analyzed in the MS-only targeted MS-MS modes. Experiments performed to construct calibration plots The 10 μM standards SOS, SSO, OSO, OOS, PSO, SPO, and POS were used to prepare known binary mixtures of positional isomers. Standard mixtures of the following pairs of positional isomers were prepared; SOS/SSO, OSO/OOS, POS/SPO, PSO/POS, and SPO/PSO. The sum of the concentrations for the pair of positional isomers for each of the standard mixtures was 1 μM. The fractions of a positional isomer in the binary mixtures ranged from 0.00-1.00 in increments of 0.10. Each of the standard mixtures was analyzed by a targeted RP-HPLC-MS-MS method designed specifically for these three systems of positional isomers. The oil and fat extracts described above were also analyzed by this method in efforts to perform a positional analysis of various oils and fats for these three systems of positional isomers. Data Analysis The spectra used in this work to define the intensities of the DAG fragments are the composite average of 50 spectra, unless otherwise stated. Regression parameters for the calibration plots were used to calculate the fractional composition of various vegetable based oils and animal fats. For the SSO/SOS and OOS/OSO systems the linear regression data from the calibration plots were used directly. Repeated measurements of samples over the course of several months suggest that this method measures the ratio of DAG fragment ions are reproducible to within ± 5 %. The average relative standard error of the OOS/OSO and SSO/SOS calibration plots was 0.005 and 0.007, respectively, giving relative errors in the predicted fractions of ≈ 2-5 % (≈ standard error/slope). For the POS/SPO/PSO system, the linear regression data from the three calibration plots were used to construct three simultaneous equations that were used to solve for the fractional composition data for each of the three positional isomers in the each of the samples. The errors in the fractions were found through the application of a Monte Carlo analysis. Our Monte Carlo approach uses repeated iterations (N=10000) of solving the simultaneous equations, while randomly modulating the error in the input ion ratios obtained from the analysis of the oil and fat samples within a 5 % tolerance. The results give Gaussian distributions for each of the fractions. The standard deviations of these distributions are reported as reasonable estimates of the errors. This Monte Carlo analysis was performed using Mathematica 4.2. Results Direct infusion CID spectra of each of the standard TAGs The CID spectrum for each of the standard TAGs was acquired via direct infusion ESI-MS-MS. Figure 1 shows the CID spectrum of POS. This spectrum is zoomed in on the DAG fragments ([M + NH4 – FA+NH3]), which are the most intense fragment ions under these experimental conditions. These ions are presumably formed via the loss of a neutral fatty acid and ammonia. The fragment ions resulting from the loss of the oleate moiety are of significantly lower abundance than the fragment ions resulting from loss of the palmitate and stearate moieties. It is speculated that fragmentation initiates at the ammoniated fatty acid group and that adduct formation at the center fatty acid position is less probable, perhaps as consequence of steric hindrance. The relative intensities of diglyceride fragments in the CID spectra of all of the TAG standards studied in this work are listed in Table 1. The less favorable loss of the fatty acid from the center position appears to be universal for all TAGs. Other trends that can be observed from careful inspection of Table 1 is that fragmentation at a particular position increases with increasing degree of unsaturation, and, though to a lesser extent, with increasing chain length. Assuming that the mechanism for fragmentation given above is correct (fragmentation occurs at the ammoniated fatty acid), these trends can be rationalized on the basis of ammonium ion affinities, which are likely to increase with increasing degree of unsaturation and fatty acid chain length. Chromatography Figure 2a shows a chromatogram for the standard mixture of TAGs. Figure 2b shows a chromatogram of the peanut oil/corn oil extract. The data was collected in MS only mode. Our data show that retention increases for TAGs with FAs of longer chain lengths and lower degrees of unsaturation. These observations agree with previously reported data illustrating that retention of TAGs species in RP-HPLC can be predicted reasonably well based on the sum of empirically-derived retention factors for the constituent fatty acid groups. The major TAG species present in the standard mixture are labeled on the chromatogram shown in Figure 2a. In addition, targeted CID experiments were performed on the standard mixture and the peanut oil/corn oil extract. For the standard mixture the CID spectra of the m/z 878 and m/z 904 peaks, which correspond to the positional isomer systems of POS/PSO/SPO and SOO/OSO, respectively, were examined to investigate the extent at which positional isomers co-elute. In both cases it was found the ratios of the relative intensities of the DAG fragments do not vary as one moves across a chromatographic peak. This indicates little to no fractionation of positional isomers of TAGs occurs in our reverse-phase HPLC method. The masses and relative intensities of the DAG fragments from the targeted MSMS experiments were used to identify the major and minor TAG species in the peanut oil/corn oil extract. Labels indicating these identified species have noted above the chromatographic peaks in Figure 2b. Each labeled species is the most abundant positional isomer of its system. The dominance of the unsaturated fatty acids in the center position is apparent. Figure 3a is also data from the targeted CID experiment on the peanut/corn oil extract. It shows the total ion chromatogram of the CID products of TAGs that have a MW of 884.5 amu ([M+NH4] ion at m/z 902.6). The chromatogram indicates that TAGs of the same molecular weight have been partially fractionated by the RP-HPLC method. Figure 3b shows the average CID spectrum across these fractionated chromatographic peaks. The intense peak at m/z 603.3 corresponds to OO or LS, and the peaks at m/z 601.3 and m/z 605.4 correspond to LO and SO, respectively. These DAG fragments indicate the dominant presence of OOO and LOS/LSO/SLO in the oil

7 Figures and Tables

Cite this paper

@inproceedings{Malone2004RunningTD, title={Running Title: Determining the Relative Amounts of Positional Isomers in Complex Mixtures of Triglycerides Using Reverse-Phase HPLC-MS-MS}, author={Michael Terence Malone and Jason J Evans}, year={2004} }