Characterization of Volatile Flavor Compounds in Dry-Rendered Beef Fat by Different Solvent-Assisted Flavor Evaporation (SAFE) Combined with GC–MS, GC–O, and OAV

To comprehensively understand the volatile flavor composition of dry-rendered beef fat, solvent-assisted flavor evaporation (SAFE) with four extraction solvents (dichloromethane, pentane, ethyl ether, and methanol) combined with gas chromatography–mass spectrometry (GC–MS) and gas chromatography–olfactormetry (GC–O) were performed. GC–MS analysis found 96 different volatile compounds in total using the four extraction solvents. According to the GC–MS results and the heat map and principal component analysis (PCA), most of the volatile compounds resulted from dichloromethane and pentane extraction, followed by ethyl ether. Methanol extraction found a few volatile compounds of higher polarity, which was supplementary to the analysis results. Moreover, GC–O analysis found 73 odor-active compounds in total using the four extraction solvents. The GC–O results found that pentane and dichloromethane extraction had a significantly larger number of odor-active compounds than ethyl ether and methanol extraction. This indicated that pentane and dichloromethane were more effective solvents for the extraction of odor-active compounds than the other two solvents. Finally, a total of 15 compounds of odor-active values (OAVs) ≥ 1 were determined to be the key aroma compounds in the dry-rendered beef fat, including 2–methyl–3–furanthiol, 3–methylthiopropanal, (E,E)–2,4–nonadienal, 12–methyltridecanal, and 1–octen–3–one.


Introduction
Beef is rich in B vitamins and minerals, and the protein content is up to 20% (w/w).Global bovine meat output was around 79.3 million tons in 2022, according to the Food and Agriculture Organization of the United Nations (FAO) statistical databases.Beef fat is the raw material of many meat foods, such as hot pot [1].In particular, dry-rendered beef fat is welcome in Chinese traditional cuisine due to its unique flavor.However, its acceptance is lowered in consumers due to its high cholesterol and calories.Therefore, an investigation of the flavor composition of dry-rendered beef fat is of significance for developing a substitute for its simulated flavor.
Currently, the analysis of volatile flavor compounds in lipid matrices is mostly reported by HS-SPME combined with GC-MS.For example, Mónica Bueno et al. [2] used HS-SPME-GC-MS technology to analyze the volatile components in fresh beef with different degrees of fat oxidation, and a total of 30 volatile compounds were identified.Da et al. [3] used SPME-GC-MS technology to compare differences in volatile components of the fatty and lean parts of braised pork.Lioupi et al. [4] used HS-SPME-GC-MS technology to determine the volatile compounds in olive oil.The HS-SPME method has the advantages of short time and simple operation [5].However, it has low extraction recovery for volatile flavor compounds in an oil or fat due to the lipophilic trait of volatile flavor compounds.Moreover, knowledge of the aroma of oil or fat is widely based on volatile compounds obtained using GC-MS rather than the odor-active compounds obtained using gas chromatographyolfactometry (GC-O) technology, though it is the odor-active compounds that actually contribute to food aroma.
Solvent-assisted flavor evaporation (SAFE) is known to be able to extract volatile fractions with a high recovery yield from the food matrix.It can avoid potential flavor alteration or pseudo-product formation as it works under high vacuum and mild temperatures.In particular, SAFE is versatile in its selection of extraction solvents to isolate the target volatile components from a lipid matrix.Gerlach et al. [6] used methanol as the extraction solvent for SAFE/GC-MS and GC-O analysis of the aroma components in pig fat, and 16 compounds such as 2,4-heptadienal, nonanal, and (E,E)-2,4-decadienal were determined as key aroma compounds.Zhou et al. [7] used dichloromethane as the extraction solvent during a SAFE/GC-MS analysis of aroma components in olive oil, while 54 volatile compounds, including aldehydes, ketones, acids, and esters, were identified.
In this study, the SAFE treatment by four extraction solvents, namely, dichloromethane, pentane, ethyl ether, and methanol, combined with GC-MS and GC-O, were performed to analyze aroma compounds in dry-rendered beef fat.The effects of these solvents on the analysis results were evaluated while the better ones were selected for quantitative analysis and odor-active values (OAVs) calculation to expose the key aroma compounds.The research results can provide guidance for the preparation of flavorings of dry-rendered beef fat flavor and improvable utilization of beef fat in the food industry.

Materials and Reagents
Abdominal subcutaneous fat of 24 month old Inner Mongolia yellow beef was bought from a supermarket in Beijing, China.The acid value of the fat was determined to be 1.63 (mgKOH/g), according to the National Standard GB5009.229-2016.Before use, the subcutaneous fat was stored at −20 • C.

Dry-Rendered Beef Fat Preparation
The beef fat was first minced into about 1 × 1 × 1 cm 3 cubes and then heated at 130 ± 5 • C for 20 min with an induction cooker while manually stirring.Three replicates were performed.
The color values of the dry-rendered beef fat were L* = 80.39, a* = 2.04, and b* = 16.27,measured using a colorimeter (UltraScan PRO, Hunter Associates Laboratory, Inc., Reston, VA, USA).The acid value was 1.92 (mgKOH/g), determined according to the National Standard GB5009.229-2016.The prepared dry-rendered beef fat was stored at −20 • C before analysis.

Fatty Acid Analysis
First, the raw subcutaneous beef fat was homogenized and extracted using dichloromethane and methanol (2:1, v/v) to obtain the fat fraction.Then, the dry-rendered fat and the fat fraction were methyl esterified and analyzed by GC-MS, respectively, as described in our previous work [8].Briefly, an Agilent 7890A/5975B GC-MS system with a HP-5 capillary column (30 m × 0.25 mm × 0.25 µm) (Agilent Technologies, Santa Clara, CA, USA) was used.The fatty acid methyl esters were identified by the search through the NIST 11 mass spectra database and injection of the standards of fatty acid esters.The results were expressed as percentages (%) of total fatty acids, which were derived via normalization of peak areas of the fatty acids esters.Three replicates were performed.

SAFE Extraction
The solvents dichloromethane, n-pentane, ethyl ether, and methanol were purified by distillation.An aliquot of 500 g of the dry-rendered beef fat was extracted using 500 mL of each solvent three times (500 mL × 3).The extract was combined and then subject to SAFE treatment as described by Zhao et al., 2017 [9].The temperature of the water bath with a sample flask was 30 • C, the circulating water temperature was 50 • C, and the system pressure was less than 10 −5 Pa.The received distillate was dried with anhydrous sodium sulfate.Then, it was first concentrated to 5 mL using a Vigreux column (I.D. 50 cm × 1 cm) and then to 0.3 mL using a stream of gentle nitrogen gas.

GC-MS Analysis
The 7890A-5975C gas chromatography-mass spectrometry (Agilent, Santa Clara, CA, USA) with a DB-WAX (30 m × 0.25 mm × 0.25 µm) column was used.The column temperature was initially 40 • C, then it was raised at 2.5 • C/min to 200 • C, and finally raised at 6 • C/min to 240 • C. The carrier gas was He (99.999%) at 1 mL/min.The sample of 1 µL was injected at 250 • C at a split ratio of 20:1.
The mass detector was operated at 150 • C in electron impact mode at 70 eV.The ion source temperature was 230 • C. The transfer line temperature was 250 • C. The chromatograms were recorded by monitoring the total ion current in the 40~450 mass range.
Compounds were identified based on the NIST11 library, retention index, and authentic chemicals injection.The amount of volatile compounds was derived via the normalization of peak areas of the detected compounds.

GC-O Analysis
As described by Zhao et al., 2017 [9], an Agilent 7890A gas chromatograph (Agilent Technologies, Santa Clara, CA, USA) equipped with an FID detector and a DATU 2000 highresolution olfactometer (DATU Inc., New York, NY, USA) was used.The capillary column was DB-WAX (30 m × 0.25 mm × 0.25 µm).The column temperature was initially 40 • C and then was raised to 230 • C at 5 • C/min.The carrier gas was N 2 (purity of 99.999%) in the flow rate of 1 mL/min.The sample of 1 µL was injected at 250 • C in a splitless mode.
A panel of twelve graduate students were trained prior to GC-O analysis.The odor descriptors were determined through the discussion of the panelists.During the GC-O analysis, each panelist recorded the sniffed odors with retention times, which were then converted to RI values relative to the n-alkane series (C 5 ~C29 ).The results were expressed as NIF (%)-that is, a percentage of the summed frequencies of a compound perceived by all panelists divided by the total times of a sample analyzed.
The compounds were identified based on the GC-MS identification results, the sniffed retention indices and odor characteristics, and the injection of authentic chemicals.

Quantitation of Odor-Active Compounds
The GC-MS instrument and analytical conditions used were as same as those described in the above section on GC-MS analysis, whereas the selective ion monitoring mode (SIM) was adopted.The analyzed samples were the concentrates after SAFE treatment using pentane and dichloromethane as the solvents.The ratio of the peak area of compound to that of the internal standard 1,2-dichlorobenzene (30 µg/mL) was used in the calculation of calibration curves.The amounts of compounds were expressed as ng/g in the dry-rendered beef fat.

Statistical Analysis
The results were means of three replicates.ANOVA was used to identify significant difference (p < 0.05) among the means.Tables were drawn using Microsoft Excel 2016 software.Principal component analysis (PCA) was performed using SIMCA13.0 (Version 13.0, Umetrics AB, Umeå, Sweden).Heat maps were plotted using Heml1.0(Huazhong University of Science and Technology, Wuhan, China).

Fatty Acid Composition
As shown in Table 1, a total of 12 fatty acids were found in the raw beef fat, while oleic acid (C18:1), palmitic acid (C16:0), and stearic acid (C18:0) were the major fatty acids.This agreed with the reports of Onopiuk et al. [10] on beef fat fatty acid composition analysis.Compared to the raw beef fat, in the dry-rendered beef fat, the content of polyunsaturated fatty acid (PUFA) of C18:2 was decreased significantly.This was because the PUFAs were apt to undergo lipid oxidization and degradation during the heating processing, which could produce volatile flavor compounds [11,12].
In Table 2, the detected sulfur-containing compounds, nitrogen-containing heterocyclic compounds, and oxygen-containing heterocyclic compounds were mainly produced via the Maillard reaction.Usually, the sulfur-containing compounds have the characteristics of a meaty aroma, and the nitrogen-containing compounds (pyrazine) have the characteristics of a roasted aroma.These compounds play important roles in cooked meat flavor due to their low odor thresholds.
Moreover, GC-MS results (Table 2) were processed via heat map and principal component analysis (PCA), as shown in Figures 1 and 2, respectively.As illustrated in the legend (Figure 1), the colors of the block from red to blue indicated that the compounds were present from higher to lower levels.It could be seen from Figure 1 that the volatile components found by SAFE/GC-MS using dichloromethane, pentane, and ethyl ether as the extraction solvents were similar, which had a markedly greater number of identifications than by methanol extraction.Methanol only extracted a small number of compounds due to its high polarity.Figure 2 shows that the contribution rate of the first principal component was 57.1%, the second principal component was 19.1%, and the cumulative contribution rate was 76.2%, indicating that the PCA analysis represented most of the information of the original variables and the analysis results were reliable.According to the score plot (Figure 2), the samples of pentane, ethyl ether, and dichloromethane were distributed closer, indicating that volatile composition extracted using the three solvents had a high similarity.In comparison, the position of the sample point for methanol extraction was far from the position of the points for pentane, ethyl ether, and dichloromethane extraction, which demonstrated the significant difference in extraction effects between them.
Moreover, GC-MS results (Table 2) were processed via heat map and principal component analysis (PCA), as shown in Figures 1 and 2, respectively.As illustrated in the legend (Figure 1), the colors of the block from red to blue indicated that the compounds were present from higher to lower levels.It could be seen from Figure 1 that the volatile components found by SAFE/GC-MS using dichloromethane, pentane, and ethyl ether as the extraction solvents were similar, which had a markedly greater number of identifications than by methanol extraction.Methanol only extracted a small number of compounds due to its high polarity.Figure 2 shows that the contribution rate of the first principal component was 57.1%, the second principal component was 19.1%, and the cumulative contribution rate was 76.2%, indicating that the PCA analysis represented most of the information of the original variables and the analysis results were reliable.According to the score plot (Figure 2), the samples of pentane, ethyl ether, and dichloromethane were distributed closer, indicating that volatile composition extracted using the three solvents had a high similarity.In comparison, the position of the sample point for methanol extraction was far from the position of the points for pentane, ethyl ether, and dichloromethane extraction, which demonstrated the significant difference in extraction effects between them.It could be seen from Figure 3 that the position of the most volatile compounds was well correlated with the position of dichloromethane and pentane on the score plot of Figure 2.This indicated that dichloromethane and pentane extraction had a strong positive correlation with the most volatile compounds.Moreover, the position of ethyl ether in Figure 2 also showed as being somewhat well correlated with the most volatile compounds in Figure 3, indicating that ethyl ether extraction also had certain positive corre- It could be seen from Figure 3 that the position of the most volatile compounds was well correlated with the position of dichloromethane and pentane on the score plot of In other words, the aforementioned three solvents displayed good performance in the extraction of the most volatile compounds.However, methanol was not a good extraction solvent, because it was far away from the most volatile compounds on the loading plot (Figure 3).The above results were consistent with the above heat map analysis and GC−MS analysis results.It could be seen from Figure 3 that the position of the most volatile compounds was well correlated with the position of dichloromethane and pentane on the score plot of Figure 2.This indicated that dichloromethane and pentane extraction had a strong positive correlation with the most volatile compounds.Moreover, the position of ethyl ether in Figure 2 also showed as being somewhat well correlated with the most volatile compounds in Figure 3, indicating that ethyl ether extraction also had certain positive correlation.In other words, the aforementioned three solvents displayed good performance in the extraction of the most volatile compounds.However, methanol was not a good extraction solvent, because it was far away from the most volatile compounds on the loading plot (Figure 3).The above results were consistent with the above heat map analysis and GC−MS analysis results.

No
Heat map analysis of the odorants was carried out, as shown in Figure 4, where the colors of blocks from red to blue indicated the NIF values ranged from higher to lower levels.Combined with GC-O results (Table 3), this analysis showed that pentane and dichloromethane had better performance for odorants extraction.Ethyl ether as the extraction solvent was not as good, since a smaller number of compounds were detected with high detection frequency (NIF ≥ 80%).Methanol as the extraction solvent was poor, since few odor-active compounds with high detection frequency (NIF ≥ 80%) were found.Furthermore, it could be noted that those odorants found through ethyl ether and methanol extraction were likewise able to be found using pentane and dichloromethane extraction.Therefore, pentane and dichloromethane were selected to be extraction solvents for the quantitative analysis.
Foods 2023, 12, x FOR PEER REVIEW 13 of 16 chloromethane had better performance for odorants extraction.Ethyl ether as the extraction solvent was not as good, since a smaller number of compounds were detected with high detection frequency (NIF ≥ 80%).Methanol as the extraction solvent was poor, since few odor-active compounds with high detection frequency (NIF ≥ 80%) were found.Furthermore, it could be noted that those odorants found through ethyl ether and methanol extraction were likewise able to be found using pentane and dichloromethane extraction.Therefore, pentane and dichloromethane were selected to be extraction solvents for the quantitative analysis.

Quantitation and OAV Calculation
Normally, potential higher-contribution compounds, screened using GC-O, are considered as key aroma compounds when their odor-active values (OAVs) exceed or are equal to one (OAVs ≥ 1).Therefore, those compounds mentioned above of NIFs ≥ 80% are mainly quantitated via GC-MS in SIM detection mode (Table 4).

Quantitation and OAV Calculation
Normally, potential higher-contribution compounds, screened using GC-O, are considered as key aroma compounds when their odor-active values (OAVs) exceed or are equal to one (OAVs ≥ 1).Therefore, those compounds mentioned above of NIFs ≥ 80% are mainly quantitated via GC-MS in SIM detection mode (Table 4).
Table 5. Concentrations and odor-active values (OAVs) for the odorants in the dry-rendered beef fat using pentane and dichloromethane extraction for solvent-assisted flavor evaporation (SAFE) and selected ion monitoring (SIM) detection of gas chromatography-mass spectrometry (GC-MS) for quantitative analysis.

Figure 1 .
Figure 1.Heat map of volatile compounds identified in the dry-rendered beef fat using four extraction solvents (pentane, ethyl ether, dichloromethane, and methanol) for solvent-assisted flavor evaporation combined with gas chromatography-mass spectrometry (SAFE/GC−MS) analysis.

Figure 1 .
Figure 1.Heat map of volatile compounds identified in the dry-rendered beef fat using four extraction solvents (pentane, ethyl ether, dichloromethane, and methanol) for solvent-assisted flavor evaporation combined with gas chromatography-mass spectrometry (SAFE/GC−MS) analysis.

Foods 2023 , 16 Figure 2 .
Figure 2. Score plot of volatile compounds in the dry-rendered beef fat using four extraction solvents (pentane, ethyl ether, dichloromethane, and methanol) for solvent-assisted flavor evaporation combined with gas chromatography-mass spectrometry (SAFE/GC−MS) analysis.

Figure 2 .
Figure 2. Score plot of volatile compounds in the dry-rendered beef fat using four extraction solvents (pentane, ethyl ether, dichloromethane, and methanol) for solvent-assisted flavor evaporation combined with gas chromatography-mass spectrometry (SAFE/GC−MS) analysis.

FoodsFigure 2 .
Figure 2.This indicated that dichloromethane and pentane extraction had a strong positive correlation with the most volatile compounds.Moreover, the position of ethyl ether in Figure 2 also showed as being somewhat well correlated with the most volatile compounds in Figure 3, indicating that ethyl ether extraction also had certain positive correlation.In other words, the aforementioned three solvents displayed good performance in the extraction of the most volatile compounds.However, methanol was not a good extraction solvent, because it was far away from the most volatile compounds on the loading plot (Figure3).The above results were consistent with the above heat map analysis and GC−MS analysis results.

Figure 2 .
Figure 2. Score plot of volatile compounds in the dry-rendered beef fat using four extraction solvents (pentane, ethyl ether, dichloromethane, and methanol) for solvent-assisted flavor evaporation combined with gas chromatography-mass spectrometry (SAFE/GC−MS) analysis.

Figure 3 .
Figure 3. Loading plot of volatile compounds in the dry-rendered beef fat using four extraction solvents (pentane, ethyl ether, dichloromethane, and methanol) for solvent-assisted flavor evaporation combined with gas chromatography−mass spectrometry (SAFE/GC−MS) during the principal component analysis (PCA) analysis.

Figure 4 .
Figure 4. Heat map of odor-active compounds identified in the dry-rendered beef fat using four extraction solvents (pentane, ethyl ether, dichloromethane, and methanol) for solvent-assisted flavor evaporation combined with gas chromatography-olfactometry (SAFE/GC−O) analysis.

Figure 4 .
Figure 4. Heat map of odor-active compounds identified in the dry-rendered beef fat using four extraction solvents (pentane, ethyl ether, dichloromethane, and methanol) for solvent-assisted flavor evaporation combined with gas chromatography-olfactometry (SAFE/GC−O) analysis.

Table 2 .
Volatile compounds identified in the dry-rendered beef fat using four extraction solvents (pentane, ethyl ether, dichloromethane, and methanol) for solvent-assisted flavor evaporation combined with gas chromatography-mass spectrometry (SAFE/GC-MS) analysis.

Table 2 .
Cont.The linear retention indices (RI) detected in the gas chromatography-mass spectrometry (GC-MS) on a DB-WAX column. 2 Means ± standard deviations (n = 3), means within the same row with different letters indicate significant differences (p < 0.05); "-", undetected.3MS,identified by NIST 11 mass spectral database; RI, agreed with the retention indices published in the literature; S, the MS and RI agreed with those of the available authentic chemicals of standards.

Table 3 .
Cont.The linear retention indices (RIs) sniffed in the GC−O analysis on a DB-WAX column. 2 Means of the NIF values (%), that is, percentages of the summed frequencies of a compound perceived by all panelists divided by the total times of a sample analyzed (n = 3), "-", not detected.3MS,based on GC-MS identification; RI, agreed with the retention indices published in the literature; O, agreed with the odor descriptors published in the literature; S, all the analytical parameters (RI, MS, and O) agreed with the available authentic chemicals of standards.

Table 4 .
Scanned ions, diluted concentrations of the standards, and the obtained calibration curves during quantitation of the odorants by gas chromatography-mass spectrometry (GC-MS) in selected ion monitoring (SIM) detection.