Development of a Liquid Chromatography-High Resolution Mass Spectrometry Method for the Determination of Free Fatty Acids in Milk.

The determination of free fatty acids (FFAs) in milk is of importance for quality control, legislative purposes, authentication and product development. We present herein a liquid chromatography–high resolution mass spectrometry method for the direct determination of FFAs in milk. The method involves mild sample preparation, avoids time-consuming derivatization and allows the direct quantification of twenty-two FFAs in a 10-min single run. It was validated and applied in thirteen cow milk and seven goat milk samples. Saturated fatty acids C16:0, C18:0 and unsaturated C18:1 (n-9) were found to be the major components of milk FFAs at concentrations of 33.1 ± 8.2 μg/mL, 16.5 ± 5.3 μg/mL and 14.8 ± 3.8 μg/mL, respectively, in cow milk and at concentrations of 22.8 ± 1.8 μg/mL, 12.7 ± 2.8 μg/mL and 13.3 ± 0.3 μg/mL, respectively, in goat milk. Other saturated and unsaturated FFAs were found in significantly lower quantities. Saturated fatty acids C6:0, C8:0 and C10:0 were found in higher quantities in goat milk than in cow milk. The levels of the important (for human health) odd-chain FFAs C15:0 and C17:0 were estimated in cow and goat milk.


Introduction
Triacylglycerols (TAGs) are the dominant component of milk fat (accounting for >95% of the total lipid content), accompanied by small amounts of di-and mono-acylglycerols, free fatty acids (FFAs), phospholipids and cholesterol [1,2]. The presence of FFAs in milk is of great importance, because FFAs have strong sensory properties contributing to the flavor and aroma of milk. FFAs are primarily formed in milk and other dairy products through the breakdown of TAGs, due to the enzymatic hydrolysis by lipoprotein lipase and other lipolytic enzymes [3][4][5]. Elevated levels of FFAs are responsible for rancidity in milk and FFAs levels exceeding 1.5 mmol/L are generally unacceptable for the consumer [5]. Furthermore, some of the beneficial health and nutritional effects of milk may be attributed to particular FAs [6]. Therefore, the determination of FFAs in milk is of importance for quality control, legislative purposes, authentication and product development.
The most common approach to quantify FFAs in dairy products is the use of gas chromatography flame ionization detection (GC-FID), involving the conversion of fatty acids into the corresponding methyl esters (FAME) [7,8]. A recent report compares three acid-or alkaline-catalyzed transesterification methods and proposes a simple one-step protocol based on 0.2 M methanolic KOH, a short reaction time (20 min) and a mild reaction temperature (50 • C) for milk FAME preparation [9]. The quantification of FFAs in milk has also been reported using an in-solution derivatization to ethyl esters by GC/MS [10]. Most recently, a butyl ester method was reported for the determination of FFAs in dairy products, where extracted free fatty acids were converted to butyl esters prior to GC-FID [11]. To our knowledge, liquid chromatography-mass spectrometry (LC/MS) methods on the determination of FFAs in milk that avoid a derivatization step do not exist, while LC/MS studies applied after derivatization are also very limited. La Nasa et al. have described the determination of FFAs in food samples, including milk, after derivatization with 2-hydrazinoquinoline, using HPLC-ESI-Q-ToF [12]. The determination of FFAs in infant milk powder employing a 2,4-dimethoxy-6-piperazin-1-yl pyrimidine (DMPP)-based isotope derivatization technique and UHPLC-ESI-MS/MS has been reported [13]. The direct determination of ten FFAs in cheese by applying matrix solid-phase dispersion followed by UPLC and tandem MS analysis has been described [14].
For decades, guidelines have suggested the reduction of dietary saturated FAs to lower the risk of metabolic syndrome and cardiovascular disease (CVD). In recent years, growing evidence from scientific studies has modified this notion for milk, leading to disestablishment of these recommendations in a healthy population [6]. The 2018 World Health Organization (WHO) draft guidelines on dietary saturated FAs recommend reducing total intake of saturated fat and replacing it with polyunsaturated and monounsaturated FAs [15]. However, recently Astrup and colleagues argued that these guidelines fail to take into account considerable evidence that the health effects vary for different types of saturated FAs and that the composition of the food source is crucially important [16]. It has been demonstrated that the high circulating odd-chain saturated fatty acid C17:0 is inversely associated with CVD and stroke mortality and potentially associated with higher risk of non-CVD death [17]. A recent large meta-analysis, which pooled the findings from 16 prospective cohort studies, has shown that higher levels of odd-chain saturated fatty acids C15:0 and C17:0 are associated with a lower risk of type 2 diabetes (T2D) [18].
The aim of our work was the development of a straightforward method for the rapid determination of a big set of FFAs in milk, permitting the estimation of the content of each particular FA. Herein, we describe a liquid chromatography-high resolution mass spectrometry (LC-HRMS) method for the direct determination of FFAs, avoiding any derivatization step and following a simple and mild liquid/liquid extraction protocol for sample preparation. The present method allows the simultaneous determination of twenty-two FFAs in a 10-min single run.

ESI-MS and LC Data
Twenty-two FAs, including short, medium and long chain saturated fatty acids, as well as monounsaturated and polyunsaturated fatty acids, were studied. The high-resolution mass spectra of these FAs were recorded in ESI negative mode. The structures of the standard compounds along with the exact masses of the deprotonated molecules, theoretical and measured (single measurement), and the mass errors are summarized in Table 1.               The extracted ion chromatograms (EICs) of the analytes in a standard solution (500 ng/mL) are presented in Figure 1. The retention times of the various analytes are included in Table 3.
Molecules 2020, 25, x 1 of 13 The extracted ion chromatograms (EICs) of the analytes in a standard solution (500 ng/mL) are presented in Figure 1. The retention times of the various analytes are included in Table 3.

Method Validation
The method was comprehensively validated to establish good linearity values for the analytes (R 2 > 0.99), limits of detection (0.4-1.6 ng/mL) and limits of quantification (1.1-4.8 ng/mL). The calibration curve data as well as limits of detection (LOD) and quantification (LOQ) are presented in Table 2.

Method Validation
The method was comprehensively validated to establish good linearity values for the analytes (R 2 > 0.99), limits of detection (0.4-1.6 ng/mL) and limits of quantification (1.1-4.8 ng/mL). The calibration curve data as well as limits of detection (LOD) and quantification (LOQ) are presented in Table 2. A simple liquid/liquid extraction protocol was used involving the addition of methanol for the protein precipitation. The EU Commission decision 202/657/EC was followed as a guideline for the verification of the accuracy and precision. Milk samples were spiked at three different concentration levels with three replicates for each fortification level. The proposed method was found to be accurate with satisfactory recoveries ranging from 82% to 97% for the low spike level and from 91% to 102% for the high spike level ( Table 3). The precision was investigated by means of the relative standard deviation (%RSD). The %RSD values that were obtained for intra-day (RSDr) and inter-day (RSD R ) variations ranged from 0.31 to 18.51 and from 1.15 to 19.12, respectively depending on the FA (Table 3).

Analysis of Samples
Thirteen cow milk samples and seven goat milk samples, which were purchased from the local market, were analyzed. The extracted ion chromatograms (EICs) of a cow milk sample (A) and a goat milk (B) are presented in Figure 2. The present method allows the simultaneous determination of twenty-two fatty acids in a 10-min single run. The contents of the twenty-two analytes in milk samples (in triplicate) are summarized in Table 4. The contents of FFA are expressed as µg of fatty acid per mL of milk. market, were analyzed. The extracted ion chromatograms (EICs) of a cow milk sample (A) and a goat milk (B) are presented in Figure 2. The present method allows the simultaneous determination of twenty-two fatty acids in a 10-min single run. The contents of the twenty-two analytes in milk samples (in triplicate) are summarized in Table 4. The contents of FFA are expressed as μg of fatty acid per mL of milk.
Other fatty acids, either saturated or unsaturated, were found in lower quantities. The relative average contents of FFAs are illustrated in Figure 3A (cow milk) and Figure 3B (goat milk). Comparing goat milk with cow milk, we observed that the content of C18:1 (n-9) is lower in goat milk than in cow milk, while the short chain saturated FAs C6:0, C8:0 and C10:0 were found in higher quantities in goat milk than in cow milk. In contrast, the long chain saturated FAs C20:0 and C24:0, not previously included in any study, are present in higher quantities in cow milk than in goat milk.
Other fatty acids, either saturated or unsaturated, were found in lower quantities. The relative average contents of FFAs are illustrated in Figure 3A (cow milk) and Figure 3B (goat milk). Comparing goat milk with cow milk, we observed that the content of C18:1 (n-9) is lower in goat milk than in cow milk, while the short chain saturated FAs C6:0, C8:0 and C10:0 were found in higher quantities in goat milk than in cow milk. In contrast, the long chain saturated FAs C20:0 and C24:0, not previously included in any study, are present in higher quantities in cow milk than in goat milk.
The present method has the advantage of avoiding any derivatization and is the first LC-HRMS method used for this application, which provides a rapid and simultaneous determination of twentytwo FFAs in milk. These are clear advantages over the previously reported methods, which require a derivatization step before analysis and have been applied for the study of a limited number of FAs The contents of FAs C18:2 (n-6) and total C18:3 (n-3) and (n-6) are similar in cow milk and goat milk (2.0 ± 1.2 µg/mL and 0.4 ± 0.3 µg/mL in cow milk and 1.8 ± 0.6 µg/mL and 0.6 ± 0.2 µg/mL in goat milk, respectively). Polyunsaturated FAs, such as C20:3 (n-6), C20:4 (n-6), C20:5 (n-3), C22:5 (n-3), C22:6 (n-6), were essentially absent from both cow and goat milk samples.
The present method has the advantage of avoiding any derivatization and is the first LC-HRMS method used for this application, which provides a rapid and simultaneous determination of twenty-two FFAs in milk. These are clear advantages over the previously reported methods, which require a derivatization step before analysis and have been applied for the study of a limited number of FAs in milk. The method developed in this work may also find applications for the determination of FFAs in various food sources, after the appropriate validation in each food matrix. time was set within a margin of ±5% for caproic and caprylic acid and a margin of ±2.5% for all the rest fatty acids.

Sample Preparation
One milliliter of milk was weighed in a screw cap glass centrifuge tube and methanol (4 mL) was added. The sample was stirred for about 30 s and then centrifuged at 4000× g for 10 min. The supernatant (500 µL) was then mixed with 500 µL of water in a vial and this mixture was used for the LC-MS/MS analysis AB SCIEX (ABSciex, Darmstadt, Germany).
For the recovery and intra-day and inter-day precision, milk samples were spiked at three different concentration levels.

Milk Samples
Twenty brand products of fresh whole milk were collected from the local market in Athens, Greece. Thirteen of them were cow milk products and seven of them goat milk products.

Conclusions
In conclusion, we present herein the first LC/HRMS method for the determination of FFAs in milk. Our method involves mild sample preparation conditions, excluding the hydrolysis of esterified fatty acids of triacylglycerols or other lipid classes and avoids time-consuming extraction pre-separation, or derivatization procedures. It is rapid and robust, permitting the quantification of twenty-two FFAs in a 10-min single run. The method was applied to thirteen cow milk samples and seven goat milk samples. The long chain saturated FAs C16:0 and C18:0, as well as mono unsaturated C18:1 (n-9), were found as the most abundant FFAs in both cow and goat milk samples. FAs with shorter chains, including C6:0, C8:0 and C10:0, were found in higher quantities in goat milk than in cow milk. The levels of odd-chain FFAs C15:0 and C17:0, which have been shown very recently to be important for human health by lowering the risk of type 2 diabetes, were estimated in cow and goat milk. The method developed in this work may also be used for other applications involving the determination of FFAs in various food sources, after the appropriate validation in each food matrix.