HPLC is one of the most widely used methods for TAG analysis in diff erent foodstuff s [43,44,46,62]. Because milk fat is so complex, developing effi cient methods of HPLC analysis has required improvement and optimization of the chromatographic system components (solvents, elution gradients, detector type, and column characteristics). Normal phase HPLC (NP-HPLC), RP-HPLC, and silver ion HPLC (Ag+-HPLC) modes have all been used appropriately, with the latter two being the most eff ective for analyzing complex mixtures of TAGs.
RP-HPLC of TAGs is performed on nonpolar stationary phases using almost exclusively col-umns packed with silica gel chemically bonded with octadecyl groups. Th e solvent system (usually mixtures of acetonitrile with acetone or methanol) should be more polar than the stationary phase, which reverses the elution order of the solutes. Th e elution order of TAGs in RP-HPLC is based on the combined eff ect of chain length and the degree of unsaturation of the FA moieties, i.e., each double bond is equivalent to about two methylene groups in its eff ect on the retention properties.
Th e term equivalent carbon number (ECN) was introduced for this reason [63] and was defi ned as ECN = CN – 2DB, where CN is the total carbon number and DB is the total number of double bonds in the TAG molecule. Within a sample of TAGs having the same ECN, the elution order is fi rst of all those TAG molecules containing polyunsaturated fatty acids (PUFAs), followed by those containing monounsaturated FAs, and lastly those containing saturated FAs. Furthermore, there is greater retention of TAGs containing short-chain FAs on the columns [64–66]. Th e main diffi culty with this type of approach when analyzing complex mixtures to separate TAGs is that TAG species with the same ECN value, known as “critical pairs,” coelute on columns fi lled with a packing with a 10 mm particle size and cannot be resolved. Development of column stationary phases with 3–5 mm particle size packing improved the separation of ECN critical pairs [67].
Th eir new retention properties were expressed more specifi c as theoretical carbon number (TCN) determined experimentally for each fatty acyl moiety on the TAG moiety.
In RP-HPLC, the mobile phase has a main eff ect on the separation of TAGs through com-petition between the mobile phase and the TAGs for the stationary phase and the increase or decrease in TAG solubility in the mobile phase, which can enhance or retard their separation. Th e most widely employed solvent system is a mixture of acetonitrile and acetone. Acetone improves the separation of critical pairs of TAGs, and acetonitrile is able to interact with the π electrons of unsaturated FAs, thereby exerting an eff ect on the separation of unsaturated species. Combining diff erent eluting solvents (acetone, acetonitrile, benzene, dichloromethane, ethanol, hexane, isooc-tane, isopropanol, methanol, etc.) could be suitable for separation, but the choice of mobile phase depends mainly on the solvents’ aptness for the detection mode.
7.3.1 Detection Systems
Th e choice of detection system has proved to be of great importance in milk fat analysis. Refraction index (RI), ultraviolet (UV), and MS detectors were the fi rst detectors used for HPLC analysis of dairy TAGs [68]. RI detection was widely used at fi rst [69–73], but certain diffi culties arise against its use, e.g., the diff ering response to saturated and unsaturated compounds, low sensitivity, and
its unsuitability for gradient elution. UV detection [74,75] does allow the use of gradients and has been used to detect TAG molecules containing FAs with conjugated double bonds that absorb at selective wavelengths [76]. However, UV detection is incompatible with mobile phases that contain acetone, which absorbs in the same regions of the spectrum where TAGs absorb. For such cases, the use of hexane, n-propanol, ethanol, and other solvents as the mobile phase has been proposed [62]. FID detection has been tried by way of an alternative [77], but light scattering (LS) detection could be a more suitable option [66,76,78–83], since it can be operated with gradients and at the same time allows all organic solvents to be used as the mobile phase.
As previously mentioned in the section dealing with GC, MS is the most suitable detection system for qualitative analysis and thus for identifying diff erent TAG species. Th e effi cient separa-tions achieved using HPLC and the structural information provided by MS suggest that HPLC-MS has considerable potential for elucidating the composition of dairy fat. Chemical ionization (CI)-based methods are the most widely employed MS procedures for ionizing and characteriz-ing the molecular fragments of milk fat TAGs separated by HPLC. Uscharacteriz-ing RP-HPLC with both positive and negative CI detection, Kuksis et al. [64] were able to analyze the TAGs present in some butter oil fractions. Chloroform attachment negative ion CI produced [M + Cl]− ions, which enabled the molecular mass of each TAG species to be determined. Analysis of the same fractions by positive CI yielded both protonated molecular ions and diacylglycerol ions, allowing the FAs present in each TAG to be determined. Molecular species of milk fat TAGs were identifi ed with this method [64]. In subsequent work [65,79,84,85], this ionization method has been a key factor in characterizing numerous species of milk fat TAGs.
HPLC in association with atmospheric pressure CI (APCI)-MS has also been shown to be a powerful tool for TAG analysis [86,87], and this approach has been very useful in elucidating the structure of cow’s milk [88] and donkey’s milk [89,90] TAGs. APCI-MS of TAGs typically yields protonated molecular ions [M + H]+, diacylglycerol ions resulting from loss of a fatty acyl moiety, and the FAs themselves in the form of acylium ions. Whereas protonated molecular ion abundance is low in highly saturated TAG species, the diacylglycerol ions enable the FAs in the sn-2 position to be diff erentiated from those in the sn-1 and sn-3 positions.
Tandem MS is another detection method used to analyze milk fat TAGs on HPLC systems. In this method, an initial positive or negative ionization step is followed by a subsequent step to ionize the fragments resulting from the fi rst step. Th e initial fragmentation products are used to determine the CNs and the double bonds of the FAs making up the TAGs, while the second fragmentation can yield mass spectra helpful in determining the regioisomers. Applying this method to human milk fat [91–93] revealed palmitic acid to be the most abundant FA in the sn-2 position and the 18:1–16:0–
18:1 to be the most important TAG quantitatively (10% of the total TAGs). Similarly, tandem MS has also been used successfully to identify minor TAGs with branched-chain FAs or odd-numbered carbon FAs in cow’s milk [66,94]. Still, this ionization method is not entirely problem-free. Studies using tandem MS in which diacylglycerol ions and FA ions are formed from TAG ions have shown that the diacylglycerol ions were not representative of the expected random distribution of diacylg-lycerols but rather contained more of the FAs at the sn-2 position [91,95]. In other words, cleavage of FAs from the sn-2 position was less than that from the sn-1 and sn-3 positions.
7.3.2 Determination of Molecular Species of TAGs
HPLC has not been used for quality control and milk fat authenticity studies like GC but has been employed to study changes in the TAG profi le due to such factors as seasonality and diet [71–73].
TAGs have been tentatively identifi ed based on the TCN and relative retention times. However,
Triacylglycerols in Dairy Foods 䡲 179
none of these studies using a single HPLC column was able to reliably determine individual TAGs, because most of the chromatographic peaks contained a number of TAG species. Barrón et al. [74]
used two HPLC columns connected in series and GC to determine the FA composition of the resulting fractions. Th ese authors [74] identifi ed 116 TAG molecular components of milk fats from diff erent ruminant species. Later, up to 181 TAG species were identifi ed using LS detection and deconvolution software for determining diff erent molecular components in each chromatographic peak [82].
7.3.2.1 Fractionation Methods
As already discussed in the section on GC, most of the procedures that have been put forward for determining individual TAG species in dairy fat have been based on prefractionation of samples followed by HPLC determination (Table 7.1). TLC was one of the fi rst methods tested. Th is method separated the milk fat TAGs into distinct bands according to TAG CN. Each of these fractions was then analyzed by HPLC [76,83,88,96,97]. Gas permeation chromatography (GPC) to separate milk fat TAG fractions prior to HPLC analysis has also been reported [88]. Separation was observed to take place by polarity rather than by molecular size, suggesting that the mode of separation was more similar to RP-HPLC than to true size-exclusion chromatography.
While Ag+-TLC has been used [65,76,79,83], Ag+-HPLC columns have ultimately carried the day as a prefractionation method [78,80,81,84,85,90,94]. Stable Ag+ columns for HPLC in which the silver ions are linked, via ionic bonds, to phenylsulfonic acid moieties bound to a silica matrix are commercially available (ChromSpher Lipids™, Chrompack, Middelburg, the Netherlands). Th ese Ag+-HPLC columns have well-defi ned chemical properties, and since most chromatographic conditions (temperature, mobile phase composition, and fl ow rate) can be very precisely controlled, reproducible data can be obtained. Trisaturated species elute fi rst, followed by disaturated-monoenoic species and saturated-dimonoenoic species, as expected. Indeed, not only the usual fractions with saturated and cis-monoenoic residues, but also those with trans double bonds could be separated on these columns. Th ese Ag+-HPLC columns were used by Adlof [98]
and have since been applied extensively to separate the TAGs in dairy fats.
Ag+-HPLC followed by RP-HPLC with MS detection proved to be an eff ective tool for characterizing the TAGs in milk fat [84,85,90,94]. Successful fractionation by Ag+-HPLC of TAGs with confi gurations diff ering in one FA is signifi cant, because afterwards geometric (cis and trans) isomers are not diff erentiable by MS. Using this method, Laakso and Kallio [84]
were able to discriminate between TAGs with two saturated FAs and one monounsaturated FA diff ering only in the geometric confi guration of this last-mentioned FA. Th ey found a higher proportion of cis FAs on TAG molecules containing short-chain FAs. Th is was attributed to steric hindrance produced by esterifi cation of long-chain FAs on molecules that already con-tained a trans FA. A similar analytical approach subsequently developed by Kallio et al. [85]
added more information about the location of cis and trans FAs at the primary and the second-ary positions of TAGs. Th eir results suggest that the sn-position of cis- and trans-monoenoic FAs depends on the two other FAs present on the molecule. Such study [85] suggests that cis- and trans-FAs are processed in milk fat biosynthesis with other FAs aff ecting the regiospecifi c position of the monoenoic C18 FAs.
Prefractionation by RP-HPLC followed by GC analysis of the fractions thus obtained is another method that has yielded abundant information on the molecular structure of milk fat TAGs (Table 7.1). Maniongui et al. [99] and Gresti et al. [100] carried out a comprehensive study combining these two procedures to determine the proportions of 223 individual TAGs composed
of the 14 major FAs that made up 80% of the total in bovine milk fat. Th is work was of decisive importance in proving the nonrandom distribution of FAs in the TAGs of dairy fats.