DETERMINATION OF VOLATILE ORGANIC COMPOUNDS IN SLOVAK BRYNDZA CHEESE BY THE ELECTRONIC NOSE AND THE HEADSPACE SOLID-PHASE MICROEXTRACTION GAS CHROMATOGRAPHY-MASS SPECTROMETRY

The aim of the present study was to describe volatile organic compounds of the traditional Slovak bryndza cheese determined by using an electronic nose (e-nose) and a gas chromatography mass spectrometry (GCMS) with head-space solid phase microextraction (HS-SPME). For the first time, e-nose based on the gas chromatography principle with a flame ionization detector was described to identify and quantify aroma active compounds of bryndza cheese from Slovakia. The e-nose detects aroma compounds of very small concentrations in real-time of a few minutes and identifies them by comparing Kovats ́ retention indices with the NIST library. Bryndza cheese produced from unpasteurized ewe ́s milk and from a mixture of raw ewe ́s and pasteurized cow ́s types of milk were collected from 2 different Slovak farms beginning in May through to September 2019. The flavour and aroma of bryndza cheese are apparently composed of compounds contained in milk and the products of fermentation of the substrate by bacteria and fungi. Regarding volatile organic compounds, 25 compounds were detected and identified by an electronic nose with a discriminant >0.900 with ethyl acetate, isopentyl acetate, 2-butanone, acetic acid, butanoic acid, and butane-2,3-dione confirmed by gas chromatography. We confirm the suitability of the electronic nose to be used for monitoring of bryndza cheese quality.


INTRODUCTION
One of the traditional ewe´s milk production is bryndza cheese or Oštiepok cheese ( Several volatile organic compounds (VOC) of cheese, including raw milk-based ewe´s cheese, are formed by proteolysis and by the subsequent transformation of amino acids (Ozturkoglu-Budak et al., 2016) to α-keto acids (Čaplová et al., 2018). Two different major pathways of amino acid degradation have been identified in Lactococcus lactis (Yvon and Rijnen, 2001). The first pathway is initiated by an elimination reaction of methionine catalyzed by amino acid lyases and leads to major sulphur aroma compounds (Dias and Weimer, 1998a; Dias and Weimer, 1998b). The second pathway is initiated by a transamination reaction catalyzed by aminotransferases, and has been observed especially for aromatic amino acids, branched chain amino acids, and methionine (Rijnen et al., 1999; Bourdat-Deschamps et

MATERIAL AND METHODOLOGY Bryndza cheese samples
Samples of bryndza cheese were provided by 2 different producers. The first sample was produced from unpasteurized ewe´s milk by farm dairy. The second one was produced from a mixture of raw ewe´s (min. 50%) and pasteurized cow´s milk by industrial dairy. Samples were collected from May to September 2019. All samples (10) were places in sterile sample containers and transported to the laboratory on ice. Fresh samples were analyzed by head-space solid phase microextraction gas chromatography mass spectrometry (HS-SPME GC-MS) and electronic nose (e-nose) within one day after the delivery.

E-nose analysis
The electronic nose method (e-nose Heracles II, Alpha M.O.S., Toulouse, France) previously described by

HP-SPME-GC-MS analysis
The head-space solid phase microextraction method was used for a sample extraction according to Sádecká et al. (2014) in a modified version. For each analysis, 2.5 g of sample was incubated statically in a 20 mL vial in a thermostat block at 50 °C for 30 min (CombiPal Autosampler 120, CTC Analytics AG, Zwingen, Switzerland), with an SPME fibre (1 cm; DVB/CAR/PDMS) (Supelco, Bellefonte, PA, USA) placed in the CombiPal.
Semi-quantitative composition of samples was determined by gas chromatography coupled with mass spectrometry (GC-MS) using an Agilent 7890B oven coupled with Agilent 5977A mass detector (Agilent Technologies Inc., Palo Alto, CA, USA) equipped with column DB-WAXms (30 m × 0.32 mm × 0.25 µm; Agilent Technologies) operating with a temperature program and MS conditions according to Sádecká et al. (2014).
The identification of compounds was carried out by comparison of mass spectra (over 80% match) with the NIST® 2017 library and retention times of reference standards (ethyl acetate, hexanoic acid, and isopentanol). The semi-quantitative content of determining compounds was calculated by dividing the individual peak area by total area of all peaks. Peaks under 1 % were not counted.

Statistical analysis
Compounds identified by e-nose with a discriminant >0.900 were selected, based on which the semi-qualitative evaluation was performed and PC analysis (Principal Component Analysis) was made by Alpha Soft V14 (Alpha M.O.S.) software. Descriptors were analyzed using single factor analysis of variance and significance was at p <0.05.
The STATGRAPHICS Centurion (© StatPoint Technologies, Inc., USA) and GraphPad Prism 6.01 (GraphPad Software Incorporated, San Diego, California, USA) were used for statistical GC-MS analysis. The ANOVA method complemented by the Test of Tukey´s Multiple Comparison Test with a value of p <0.05 was applied.

RESULTS AND DISCUSSION
In this study, the aromatic profiles of ten bryndza cheese samples by e-nose and HS-SPME GC-MS were evaluated. Bryndza cheese samples were collected from the Slovak dairies. Bryndza cheese is a soft spreadable cheese, made from unpasteurized ewe´s milk or a mixture of ewe´s and cow´s milk. The identification of the compounds determined by enose was performed by matching the measured peaks with Kovats ŕetention indices with the NIST library. Aroma compounds identified in bryndza cheeses by e-nose are shown in Table 1. Total, 25 compounds with a discriminant >0.900 from class alcohols, esters, fatty acids, terpenes ketones, and aldehydes were identified. Ten compounds -3-methyl butanal, 2-methyl-1-propanol, 2butanone, 2-pentanone, ethyl acetate, isopentyl acetate, ethyl butyrate, acetic acid, butanoic acid, and butane-2,3dione were identified in this study by e-nose and by gas chromatography-olfactometry previously described in Identified compounds by e-nose with a discriminant >0.900 were selected, based on which the semi-qualitative evaluation was performed by the principal component analysis (PCA). Figure 1 displays result processed by the PCA technique of the aroma profile of bryndza cheese samples. The first dimension (PC1 86.927%) allows the separation of July2 and August2 (positive score) from the other samples of cheese (negative score). Among samples May1, July2 and September2 (negative score of the second dimension PC2; 9.402%) and other samples of cheese (positive score PC2) statistically significant differences (p <0.05) were evident in their aroma profiles. We confirmed differences in aroma profiles between May bryndza and summer or winter bryndza produced from ewe´s milk by the first dairy. On the contrary, more aroma profile differences were in bryndza cheese produced from the mixture of ewe´s and cow´s milk by the second dairy.  The samples collected in May and June showed no significant differences in aroma profiles when compared to each other but the aroma profiles were significantly different in comparison with July, August, and September samples and at the same time, the samples collected from July, August, and September showed statistically different (p <0.05) aroma profiles compared to each other. On the contrary, the position of May2, June2, June1 -September1 samples on the negative score of axis 1 and a positive score of axis 2 could be explained by its higher proportions of acetic acid, benzyl alcohol, butyl acetate, butanal, benzaldehyde, n-butanol, 2-pentanone, and hexanol. The August2 sample is positioned at the opposite on axis 1 due to its higher proportions in butanoic acid, 2propanol, 2-methyl propanal, 3-methyl butanal, α-pinene, and ethyl acetate. The sample of July2, positioned on the positive score of axis 1 and the negative score of axis 2, contains higher proportions of isopentyl acetate, 2butanone, 2-methyl propanol, furfural, and propanoic acid and the position of May1 and September2 samples on the negative scores of axes 1 and 2 could be explained by their higher proportions of ethyl butyrate, ethyl propanoate, propanal, 2,3-butandione, and acetaldehyde.
The bryndza cheese from the first dairy contains about 15% of acetoin collected from May and June, and, on the contrary, the next months its content decreased (10.6%, 7.04%, and 10.5% respectively). The content of acetoin in samples from the second dairy was more variable, its amount was decreased (3.88%) in samples collected from July production and increased in samples collected from August production (27.1%). in at least one bryndza cheese sample. Letters in superscript indicates statistically significant difference: a -among samples depending on month of production, b -among samples depending on kind of milk.
The content of acetic acid was in a range of 8.55% -13.3% in samples produced in May-July, it was decreased in August (1.77% and 6.92%, respectively) and increased in September (18.0% and 18.7%, respectively). While acetoin and acetic acid were the most representative compounds in samples produced in May and June, samples produced in July and August had different profiles. While the samples from the first dairy had a higher content of acetoin and acetic acid (July) and 2-phenethyl acetate and 2-phenyl ethanol (August), the samples from the second dairy had a higher content of 2-butanone and acetic acid (July) and 2-butanone and acetoin (August). Bryndza cheese produced in September had a weaker aroma, but acetic acid and acetoin were identified in the higher amount than in the May and June samples. Statistical analysis by the Test of Tukey´s Multiple Comparison Test confirmed that the amount of eight aromatic compounds ( Table 2) was influenced by the month of production. On the other hand, only two aroma substances were notably influenced by the kind of milk.

CONCLUSION
This study has proved for the first time the possibility of bryndza cheese quality evaluation using an e-nose with GC columns and FID detectors. The results were compared with gas chromatography with mass spectrometry. The e-nose method can determine the aroma profile of samples in a short time and the results may be supplemented by aensory evaluation by the assessors. The e-nose may take great advantages over GC-MS in distinguishing the integral aroma profiles, although it cannot identify the explicit volatile compounds of different samples.