Molecular Identification and VOMs Characterization of Saccharomyces cerevisiae Strains Isolated from Madeira Region Winery Environments

The quality and typical characteristic of wines depends, among other factors, on the volatile organic metabolites (VOMs) that are biosynthesized by yeasts, mainly Saccharomyces cerevisiae species. The yeast strain influences the diversity and proportions of the VOMs produced during the fermentation process, as the genetic predisposition of the strains is a by-product of selective adaptation to the ecosystem. The present work reports the characterization of S. cerevisiae strains isolated from grape must, used in the Demarcated Region of Madeira (DRM) for winemaking. Yeast species were identified by amplification and by restriction fragment length polymorphism (RFLP) analysis of the region 5.8S-internal transcribed spacers (PCR-RFLP of 5.8S-ITS) of ribosomal DNA (rDNA). The strains identification was performed by analyzing the RFLP pattern of mitochondrial DNA (RFLP-mtDNA). The representative strains were selected for the characterization of the volatile profile through headspace solid-phase microextraction (HS-SPME) followed by gas chromatography-mass spectrometry (GC-MS) analysis. A total of 77 VOMs were identified. Higher alcohols, esters, and fatty acids were the major chemical families representing 63%, 16%, and 9%, respectively, in strain A and 54%, 23%, and 15% in strain B. The results indicate the influence of the strain metabolism in the production of VOMs, many of which probably participate in the aroma of the corresponding wines.


Introduction
The quality and typicity of wines result from the conjugation of several natural and human factors [1,2]. In the Demarcated Region of Madeira (DRM), specific agro-edaphoclimatic conditions such as insular Atlantic territory, volcanic origin, combined with selective noble grapes and specific agronomic techniques adapted to the area and orography of the island, in close relationship with traditional and sui generis vinification process, confer to the wines produced in the DRM unique organoleptic properties, among which aroma is an important parameter that defines its quality and typicity [3].
The contribution of S. cerevisiae to the aroma of the wine is well known. During alcoholic fermentation, as a secondary metabolomic pathway, S. cerevisiae biosynthesizes semi-volatile and non-volatile organic metabolites, especially a large number of different volatile organic metabolites

Microfermentation and S. cerevisiae Isolation
Ten different grape musts (500 mL) were submitted to spontaneous microfermentation according to Schuller et al. [9], with some modification, namely, room temperature (27.9 °C; ±1) and, stirring or orbital shaker (160 rpm). The microfermentations progress was monitored daily through the determination of must mass loss until weight stabilization. Serial dilutions (10 −2 to 10 −8 ) were spread on YPD plates. After incubation (48 h, 30 °C), 10 isolates were selected based on colonies morphology (corresponding to S. cerevisiae) and purified. The pure isolates (100) were stored in YPD broth supplemented with glycerol (30%, v/v) at −42 °C until analysis.

DNA Isolation
For the yeast species identification, DNA was isolated and purified according to Lõoke et al. [10], with slight modifications. YPD plates were divided in six sections. Isolates (each pure culture) were spread on each section and incubated (48 h, 30 °C). Lithium acetate-sodium dodecyl sulphate (LiOAc-SDS) solution was used for yeast cell lysis. Stock solution of LiOAc was prepared (200 mM), autoclaved and 1% of SDS was added to this solution. A pure culture was picked from the YPD section plate, suspended in 100 µL of the solution described previously, and incubated at 70 °C, for 20 min. Then, DNA was precipitated adding 300 µL of ethanol (96%), mixed briefly by vortex and then centrifuged at ≈15,000× g (approximately) for 3 min. Supernatant was removed and pellet was washed with 500 µL of ethanol (70%). Finally, the dried DNA was suspended in 50 µL of ultrapure autoclaved water. DNA concentration was determined using the NanoDrop 2000c Spectrophotometer (Thermo Scientific TM ).

Microfermentation and S. cerevisiae Isolation
Ten different grape musts (500 mL) were submitted to spontaneous microfermentation according to Schuller et al. [9], with some modification, namely, room temperature (27.9 • C; ±1) and, stirring or orbital shaker (160 rpm). The microfermentations progress was monitored daily through the determination of must mass loss until weight stabilization. Serial dilutions (10 −2 to 10 −8 ) were spread on YPD plates. After incubation (48 h, 30 • C), 10 isolates were selected based on colonies morphology (corresponding to S. cerevisiae) and purified. The pure isolates (100) were stored in YPD broth supplemented with glycerol (30%, v/v) at −42 • C until analysis.

DNA Isolation
For the yeast species identification, DNA was isolated and purified according to Lõoke et al. [10], with slight modifications. YPD plates were divided in six sections. Isolates (each pure culture) were spread on each section and incubated (48 h, 30 • C). Lithium acetate-sodium dodecyl sulphate (LiOAc-SDS) solution was used for yeast cell lysis. Stock solution of LiOAc was prepared (200 mM), autoclaved and 1% of SDS was added to this solution. A pure culture was picked from the YPD section plate, suspended in 100 µL of the solution described previously, and incubated at 70 • C, for 20 min. Then, DNA was precipitated adding 300 µL of ethanol (96%), mixed briefly by vortex and then centrifuged at ≈15,000× g (approximately) for 3 min. Supernatant was removed and pellet was washed with 500 µL of ethanol (70%). Finally, the dried DNA was suspended in 50 µL of ultrapure autoclaved water. DNA concentration was determined using the NanoDrop 2000c Spectrophotometer (Thermo Scientific TM ).
PCR amplification was carried out according to Esteve-Zarzoso et al. [13]: an initial denaturation at 95 • C for 5 min, followed by 35 cycles of denaturation at 94 • C for 1 min, primers annealing at 55.5 • C for 2 min and extension at 72 • C for 2 min. The final extension was carried out at 72 • C for 10 min, followed by cooled at 4 • C. The fragments were digested with the restriction endonucleases HaeIII (BsuRI) and Hinf I as recommended by the manufacturer. The PCR products and their restriction fragments were analysed by agarose gel electrophoresis (1.4 and 3% w/v respectively). Gels were stained with ethidium bromide and observed under UV light. Digital images were acquired using a Kodak 290C camera.

RFLP-mtDNA Analysis
The digestion was performed according to Shuller et al. [14]. The mix reaction contained 17 µL of DNA (extracted by enzymatic method), 0.5 µL (10 U/µL) of restriction endonucleases Hinf I, 2 µL (10X) of R buffer and 0.5 µL (10 mg/mL) of RNase was incubated overnight at 37 • C. The restrictions fragments were separated on agarose gel electrophoresis (1.5% w/v) and a 1-kb DNA ladder was used as a molecular size marker. Gels were stained with ethidium bromide and observed under UV light. Digital images were acquired using a Kodak 290C camera.

Inoculated Microfermentations
Two dominant S. cerevisiae strains were selected for the VOMs characterization. Commercial strain QA23 (Lavin) was selected as positive control. 10 5 and 10 6 cells/mL were inoculated in commercial rectified concentrated grape must clarified (MC), diluted in mQ to 12.5% of probable alcohol. Before the assay, the MC was subject to microbiological control on YPD, PDA and NA plates. Inoculated microfermentations were carried out in homologues conditions (28 • C, 160 rpm). The evolution of the microfermentations was monitored by weight loss until stabilization. At the end of microfermentations, the wines were clarified and refrigerated at 4 • C and analyzed for no more than 12 h.

HS-SPME Conditions
The extractions conditions, namely, sample volume, sample temperature, ionic strength, fibre type and contact time between fibre and headspace were carried out according to Perestrelo et al. [15] with some modifications. 3-octanol (1:1, v/v) was selected as internal standard, 10 mL for sample volume, 20% NaCl and 10 µL of internal standard. The VOMs extraction was performed with a DVB/CAR/PDMS StableFex fiber in headspace mode at 40 • C during 40 min under continuous stirring (9000 rpm).

GC-MS Conditions
The VOMs separation was performed in an Agilent Technologies 6890N (Palo Alto, CA, USA) gas chromatography, equipped with a fused silica capillary column BP20 (SGE, Dortmund, Germany) of 60 m × 0.25 mm I.D. × 0.25 µm film thickness. Helium at a flow of 1 mL/min was used as carrier gas. The temperature gradient used was: 45 • C during 1 min, increasing 2 • C/min until 100 • C maintaining this temperature for 3 min, following an increase of 5 • C/min to 130 • C and remain for 5 min, rising again 20 • C/min until 220 • C for 15 min. The total run time was 72 min [15].
The VOMs detection was performed using an Agilent Technologies 5975N quadrupole mass selective detector operated in electronic impact ionization (EI) mode at 70 eV and mass spectra was recorded in the range of 30-300 m/z. The ions source, transfer line, and quadrupole temperatures were 220, 250, and 150 • C, respectively.
The VOMs identification was performed by comparison of mass spectra with NIST 05 Al library. The VOMs with percent of similarity >75% were considered for the present analysis. Additionally, some VOMs were identified and confirmed by comparison of Kovats retention index (KI) value with the reported in the literature for similar chromatographic columns. The KI was calculated using the Equation (1), where "n" is the number of carbon atoms present in the alkane with i carbon atoms (i), N, the number of carbon atoms relative to alkane i + 1, t'r (n) the retention time adjusted for minor alkane, and t'r (N) the retention time adjusted for major alkane. The retention times used in the equation correspond to the times obtained by direct injection of the C8-C20 n-alkane series.

Spontaneous Microfermentation and Yeast Isolation
All the samples fermented spontaneously. The microfermentations had a loss of mass between 31.25 and 80.50 g/L and a duration between six and 12 days. Three types of fermentative profiles were obtained as shown in Figure 2. The VOMs separation was performed in an Agilent Technologies 6890N (Palo Alto, California, EUA) gas chromatography, equipped with a fused silica capillary column BP20 (SGE, Dortmund, Germany) of 60 m × 0.25 mm I.D. × 0.25 µm film thickness. Helium at a flow of 1 mL/min was used as carrier gas.
The temperature gradient used was: 45 °C during 1 min, increasing 2 °C/min until 100 °C maintaining this temperature for 3 min, following an increase of 5 °C/min to 130 °C and remain for 5 min, rising again 20 °C/min until 220 °C for 15 min. The total run time was 72 min [15].
The VOMs detection was performed using an Agilent Technologies 5975N quadrupole mass selective detector operated in electronic impact ionization (EI) mode at 70 eV and mass spectra was recorded in the range of 30-300 m/z. The ions source, transfer line, and quadrupole temperatures were 220, 250, and 150 °C, respectively.
The VOMs identification was performed by comparison of mass spectra with NIST 05 Al library. The VOMs with percent of similarity >75% were considered for the present analysis. Additionally, some VOMs were identified and confirmed by comparison of Kovats retention index (KI) value with the reported in the literature for similar chromatographic columns. The KI was calculated using the Equation (1), where "n" is the number of carbon atoms present in the alkane with i carbon atoms (i), N, the number of carbon atoms relative to alkane i + 1, t'r (n) the retention time adjusted for minor alkane, and t'r (N) the retention time adjusted for major alkane. The retention times used in the equation correspond to the times obtained by direct injection of the C8-C20 n-alkane series.

Spontaneous Microfermentation and Yeast Isolation
All the samples fermented spontaneously. The microfermentations had a loss of mass between 31.25 and 80.50 g/L and a duration between six and 12 days. Three types of fermentative profiles were obtained as shown in Figure 2. In the first type, the fermentations started in the 3rd or 4th day followed by a significant weight loss, this profile was observed in Malvasia Cândida, Verdelho 1 (Prazeres) and Verdelho 2 (São In the first type, the fermentations started in the 3rd or 4th day followed by a significant weight loss, this profile was observed in Malvasia Cândida, Verdelho 1 (Prazeres) and Verdelho 2 (São Vicente) varieties. The second type showed regular loss weight during all fermentation process. This pattern was observed in Malvasia de São Jorge variety. In the third type, the fermentations started quickly with significant weight loss until stabilization. This profile was observed in the remaining samples ( Figure 2).
After microfermentations, the number of viable yeasts per milliliter was estimated for each wine (Table 1). The lowest concentration of yeasts per mililitre was obtained for Boal variety (470 CFU/mL), followed by Verdelho 1 (530,000 CFU/mL) and Malvasia Cândida (770,000 CFU/mL). In the remaining wines, yeasts were quantified in an order of magnitude in the millions of CFU/mL. Sercial 1, Sercial 2, Tinta Negra and Malvasia de São Jorge were quantified in 4.2, 4.7, 12, and 15.5 million CFU/mL, respectively. However, the highest concentration of yeasts was found in the Bastardo (1260 million CFU/mL) followed by Terrantez and Verdelho 2 in which 137 and 970 million CFU/mL were quantified.

Molecular Identification
From 10 microfermentations, 100 isolates whose macromorphological colony characteristics corresponding to the description of S. cerevisiae were tested. The DNA concentration obtained was between 700-100 ng/µL and an average degree of purity of 1.9 (d s 0.05) for the genus and species identification. The DNA concentration for the strain identification was diluted until 5500-6000 ng/µL and an average degree of purity of 2.0 (d s 0.10). DNA concentration and purity were measured by NanoDrop 2000c Spectrophotometer.
Regarding genus identification by 5.8S-ITS of rDNA amplification, the amplicon obtained in the isolate analysis corresponded to a fragment between 850-880 pb, after visualization in agarose gel ( Figure 3). Vicente) varieties. The second type showed regular loss weight during all fermentation process. This pattern was observed in Malvasia de São Jorge variety. In the third type, the fermentations started quickly with significant weight loss until stabilization. This profile was observed in the remaining samples ( Figure 2). After microfermentations, the number of viable yeasts per milliliter was estimated for each wine (Table 1). The lowest concentration of yeasts per mililitre was obtained for Boal variety (470 CFU/mL), followed by Verdelho 1 (530,000 CFU/mL) and Malvasia Cândida (770,000 CFU/mL). In the remaining wines, yeasts were quantified in an order of magnitude in the millions of CFU/mL. Sercial 1, Sercial 2, Tinta Negra and Malvasia de São Jorge were quantified in 4.2, 4.7, 12, and 15.5 million CFU/mL, respectively. However, the highest concentration of yeasts was found in the Bastardo (1260 million CFU/mL) followed by Terrantez and Verdelho 2 in which 137 and 970 million CFU/mL were quantified.

Molecular Identification
From 10 microfermentations, 100 isolates whose macromorphological colony characteristics corresponding to the description of S. cerevisiae were tested. The DNA concentration obtained was between 700-100 ng/µL and an average degree of purity of 1.9 (ds 0.05) for the genus and species identification. The DNA concentration for the strain identification was diluted until 5500-6000 ng/µL and an average degree of purity of 2.0 (ds 0.10). DNA concentration and purity were measured by NanoDrop 2000c Spectrophotometer.
Regarding genus identification by 5.8S-ITS of rDNA amplification, the amplicon obtained in the isolate analysis corresponded to a fragment between 850-880 pb, after visualization in agarose gel ( Figure 3). Regarding species identification, the amplicon digested with restriction endonucleases resulted in restriction pattern with fragments with approximately 320, 230, 180, and 150 pb with HaeIII enzyme, and 365 and 155 pb with the HinfI enzyme (data no presented).
From 100 isolates of S. cerevisiae, 8 RFLP-mtDNA patters were identified (assigned as A to O), each pattern was considered as a strain. Pattern B occurred in 81% of the isolates and pattern A was identified in 11% ( Figure 4). The pattern C and L were identified in 2% while other four patterns occurred only in 1% of the isolates. We also identified the presence of two or three isolates in most microfermentations: strains B and O in Malvasia (São Jorge), B and A in Tinta Negra (São Vicente), and B and C in Verdelho (Prazeres), all with 9:1 proportion, B and A in Sercial 2 (Jardim de Serra) with 6:4 proportion, and B with K in Malvasia Cândida (Fajã dos Padres) with 6:4 proportion. In two microfermentations, three strains were identified: B, A and L in Sercial 1 (Seixal) with 7:2:1 proportion and B, A and M in Verdelho (São Vicente) with 6:3:1 proportion, respectively. In Boal and Bastardo varieties, strain B was identified, and on Terrantez B, A, L, and N strains were identified in 7:1:1:1 proportion, respectively. Regarding species identification, the amplicon digested with restriction endonucleases resulted in restriction pattern with fragments with approximately 320, 230, 180, and 150 pb with HaeIII enzyme, and 365 and 155 pb with the HinfI enzyme (data no presented).
From 100 isolates of S. cerevisiae, 8 RFLP-mtDNA patters were identified (assigned as A to O), each pattern was considered as a strain. Pattern B occurred in 81% of the isolates and pattern A was identified in 11% ( Figure 4). The pattern C and L were identified in 2% while other four patterns occurred only in 1% of the isolates. We also identified the presence of two or three isolates in most microfermentations: strains B and O in Malvasia (São Jorge), B and A in Tinta Negra (São Vicente), and B and C in Verdelho (Prazeres), all with 9:1 proportion, B and A in Sercial 2 (Jardim de Serra) with 6:4 proportion, and B with K in Malvasia Cândida (Fajã dos Padres) with 6:4 proportion. In two microfermentations, three strains were identified: B, A and L in Sercial 1 (Seixal) with 7:2:1 proportion and B, A and M in Verdelho (São Vicente) with 6:3:1 proportion, respectively. In Boal and Bastardo varieties, strain B was identified, and on Terrantez B, A, L, and N strains were identified in 7:1:1:1 proportion, respectively.
The dominant strains were A and B. Strain A was identified at Sercial 1, Tinta Negra, Verdelho 1, Terrantez and Sercial 2 grape musts microfermentations. Strain B was identified in all grape must microfermentations.

Inoculated Microfermentations
Supplementary Figure S1 shows the overlap of the chromatograms obtained for A and B strains. For better visualization of the VOMs biosynthesized by each strain, the MC chromatogram was included. The chromatogram of strain B is more complex with a greater number and more intense peaks, followed by the control whose chromatogram (data no shown) presents the same number of peaks than B, although somewhat less intense. Finally, the chromatogram of strain A showed the least number of peaks and the lowest intensity.
Regarding to total GC peak areas, the results expressed in arbitrary units (a.u.) were 2.28 × 10 8 for MC; 8.53 × 10 8 for strain A; 1.72 × 10 9 for strain B and 1.23 × 10 9 for the control strain. Figure 5 shows the total areas of VOMs identified from each strain and positive (C) and negatives (MC) controls. The dominant strains were A and B. Strain A was identified at Sercial 1, Tinta Negra, Verdelho 1, Terrantez and Sercial 2 grape musts microfermentations. Strain B was identified in all grape must microfermentations.

Inoculated Microfermentations
Supplementary Figure S1 shows the overlap of the chromatograms obtained for A and B strains. For better visualization of the VOMs biosynthesized by each strain, the MC chromatogram was included. The chromatogram of strain B is more complex with a greater number and more intense peaks, followed by the control whose chromatogram (data no shown) presents the same number of peaks than B, although somewhat less intense. Finally, the chromatogram of strain A showed the least number of peaks and the lowest intensity.
Regarding to total GC peak areas, the results expressed in arbitrary units (a.u.) were 2.28 × 10 8 for MC; 8.53 × 10 8 for strain A; 1.72 × 10 9 for strain B and 1.23 × 10 9 for the control strain. Figure 5 shows the total areas of VOMs identified from each strain and positive (C) and negatives (MC) controls. As expected, the volatile fraction increased after fermentation. The A, B and C registered an increase of 374%, 540%, and 754% respectively, in comparison to the volatile fraction of MC. The highest production of secondary metabolites was obtained by strain B whose volatile fraction exceeded about 100% the volatile fraction of A and approximately 27% of the control strain C.
The VOMs identified were grouped by chemical families as follows: esters, high alcohols, carbonyl compounds, fatty acids, terpenes and miscellaneous. Table 2 shows the contribution of each chemical family to the total volatile profile and the number of compounds identified.  (7) 7.38 (6) Before fermentation, MC volatile profile showed that carbonyl compounds are the major chemical family, representing 89%, followed by esters (4.10%) and terpenes that represent approximately 3%. Miscellaneous and higher alcohols represent 1.92 and 1.11%, respectively. In a lesser extent, fatty acids (0.78%). The most abundant volatile compound is 3-octanone (74.76%), followed by cyclohexane representing 9.82% of the total volatile fraction. Between 1% and 2% are benzaldehyde, 1,1-diethoxy-ethane, and 4-ethylresorcinol ( Table 3). The remaining compounds represent less than 1% each. As expected, the volatile fraction increased after fermentation. The A, B and C registered an increase of 374%, 540%, and 754% respectively, in comparison to the volatile fraction of MC. The highest production of secondary metabolites was obtained by strain B whose volatile fraction exceeded about 100% the volatile fraction of A and approximately 27% of the control strain C.
The VOMs identified were grouped by chemical families as follows: esters, high alcohols, carbonyl compounds, fatty acids, terpenes and miscellaneous. Table 2 shows the contribution of each chemical family to the total volatile profile and the number of compounds identified.  (7) 7.38 (6) Before fermentation, MC volatile profile showed that carbonyl compounds are the major chemical family, representing 89%, followed by esters (4.10%) and terpenes that represent approximately 3%. Miscellaneous and higher alcohols represent 1.92 and 1.11%, respectively. In a lesser extent, fatty acids (0.78%). The most abundant volatile compound is 3-octanone (74.76%), followed by cyclohexane representing 9.82% of the total volatile fraction. Between 1% and 2% are benzaldehyde, 1,1-diethoxy-ethane, and 4-ethylresorcinol ( Table 3). The remaining compounds represent less than 1% each. Table 3. VOMs identified by HS-SPME/GC-MS analysis, obtained from wines inoculated with pure strains (RFLP-mtDNA), indicating the contribution of each compound in the chemical family and in the total volatile fraction of the wine, as well as the olfactory descriptor.    After fermentations, between 77% and 79% of the identified VOMs belong to higher alcohols and esters. Higher alcohols are found in a greater percentage, especially in strain A, representing 63.20%, while in B and C they are found in a similar proportion (54% approximately). In relation to esters, strains B and C have the highest proportion (about 23%), while A has about 16%.
Minor chemical families represent between 20 and 23% of the identified VOMs. Of this fraction, fatty acids are the most abundant VOMs, representing approximately 15.0, 10.5 and 9.2 in strains B, C and A, respectively. The miscellaneous compounds represent 7.4% and 6.1% in C and A respectively, and in strain B they represent only 2.9%. The carbonyl compounds showed more heterogeneous results, representing 4.8%, 3.8%, and 2.5% in A, B, and C respectively. Terpenes are the minor chemical family with the lowest percentage (<1.2%) in all samples.
Regarding the number of VOMs, were identified 78, from which 22 were carbonyl compounds, 18 esters, 12 higher alcohols, 11 terpenes, seven fatty acids, and eight miscellaneous. In the MC, 33 VOMs were identified and, in each wine fermented by pure strain, a total of 53 VOMs were identified in the A, and 60 in the B and C, respectively.
It is verified that the studied volatile profile is composed of three majority VOMs, representing approximately 57% of the total volatile fraction in B and 61% in A and C. The chemical family with the greatest contribution is the higher alcohols, representing 55.7% of the total volatile fraction of A, approximately 47% of B and 50.8% of control. Moreover, 3-methyl-1-butanol was the major compound, mainly in A (44%) and in a smaller percentage in B and C (between 34% and 36%). The second major VOM was 2-phenylethanol (flowers odour), which represents about 12%, 13%, and 15% of the total volatile fraction of A, B, and C, respectively. The third major identified volatile was ethyl octanoate, contributing approximately 10-11% to the total volatile profile of B and C. However, in A it is present in a smaller percentage (5.3%). The fourth most abundant VOM was acetic acid, contributing 8% of the total volatile fraction of B and approximately 5.8% of A and C (Table 3).

Discussion
Several demarcated regions of the world present in Portugal, Spain, and Italy [9,19,20], and emerging wine regions in Chile and Australia [21], characterized molecular and technologically endemic strains of S. cerevisiae. However, this is the first study carried out at the DRM.
Since the must grapes were fermented under similar conditions, it appears that the different fermentative profiles may be due to the chemical composition of the must [11] and/or the association of strains during fermentation. The fermentation process is completed by one to three strains as reported by Versavaud et al. [27] and with some results obtained by Schuller et al. [9]. The concentration of viable cells/mL of waste varies in all varieties studied, probably agro-edafoclimatic factors may be directly related to this CFU difference. Fermentation can be affected not only by the substrate, in this case by the grape variety, but also by the flora endemic on the grape surface and symbiosis, variety/ecosystem.

Intraspecific Variability
The success of spontaneous microfermentation (100%) in this study was the same reported by Cappello et al. [22] and Blanco et al. [28] while this value was much lower (60%) in the work of Schuller et al. [9]. On the other hand, they reported a higher rate of intraspecific variability (18.3%) compared to the present study (8%) and to those reported by Cappello et al. [22] and Blanco et al. [28] (2.3% and 4.3%, respectively).
These differences may be due to the sampling methodology used in the different studies. It appears that S. cerevisiae from cellar environments have a very high rate of microfermentation but a lower rate of intraspecific variability. These results possibly suggest that some (endemic) strains may be more susceptible to enological treatment, making a first selection in an important technological parameter.
The RFLP-mtDNA restriction patterns obtained in the present work do not coincide with the RFLP-mtDNA patters in the consulted bibliography, until the date of completion of this work.

Screening of Volatile Profile
In total, 46 VOMs were biosynthesized by yeasts, mainly higher alcohols and esters, and to a lesser extent, fatty acids, miscellaneous, and some carbonyl compounds. In addition, 8 esters present in the wort in a low percentage, were obtained in wines in higher percentage, so it is inferred that the surplus was the product of biosynthesis by the respective yeast. Also, three VOMs are possibly used as precursors in different metabolic pathways, since they were in a high percentage in the wort, decreasing drastically after fermentation as is the case of 3-octanone. In the particular case of terpenes, it appears that they were possibly subject to modifications, probably metabolic, since they are present in the must and absent in the wine.

Higher Alcohols
After fermentation, the volatile fraction corresponding to the higher alcohols is predominantly the most representative [5,29]. This result agrees with those obtained in the present study (Table 3). Higher alcohols are biosynthesized by yeasts during alcoholic fermentation, in two ways, namely via the amino acid catabolism or Ehrlich mechanism [4,7,30], and through the anabolic route from α-keto acids [4,7].
It was found that the production of higher alcohols, both in quantitative terms and in diversity of VOMs, it is strongly conditioned by the S. cerevisiae strain that carried out the fermentation process, since, on identical substrates and similar fermentation conditions, different profiles were obtained, both in the contribution of alcohols to the total volatile fraction (about 54% in B and C, and 64% in A), and in the number of higher alcohols synthesized by each strain (10, 12, and 11 in A, B, and C respectively ( Table 2)), and also in the individual proportion of each alcohol within the chemical family itself, and in the total volatile profile (Table 3).
Generally, 3-methyl-1-butanol, the main aliphatic alcohol, represents between 40-72% of the volatile fraction of higher alcohols. In this study, it represented 62-70% of this fraction, concluding that the results obtained is in accordance with the percentage range described in the literature [4,29]. Moreover, 3-Methyl-1-butanol mimics the smell of bitter, balsamic, and astringent and is produced from leucine catabolism [18,30].
MW in the manufacturing process (younger non-marketable), was also found as the majority VOC, extracted and identified using the same technique used in the present study [31].
The second VOM in percentage is 2-phenylethanol (Table 3). This phenolic alcohol is biosynthesized from phenylalanine of great oenological interest since it is associated with aromatic descriptors that imitate the smell of roses [4,31] evolving into honey, contributing to the bouquet of some aged wines [4].
Regarding the screening of the volatile profile from VM of the Malvasia variety [32] it was verified through the respective chromatogram, that the peak referring to 3-methyl-1-butanol and 2-methyl-1-butanol is the seventh most intense peak in wines aged for 20 years (1988), and the fifth most intense peak in wines aged for two years (2006). Additionally, it is still the second most intense peak belonging to the chemical family of higher alcohols. In the same study, 2-phenylethanol is the third most intense peak in wines aged for 20 years, and the most intense peak in wines aged for two years, being the most representative of higher alcohols. In this sense, it was inferred that its biosynthesis possibly occurs from the initial stage of alcoholic fermentation, since VM Malvasia belongs to the type of sweet wine [32]. The literature also reports that the characteristic heating process in VM does not substantially modify the percentage of this chemical family [33].
Higher alcohols are also important in esterification reactions with carboxylic acids, precursors responsible for the formation of odorant esters, with a positive impact on the wine's aroma [4,30], especially in the bouquet.

Esters
A chemical family with positive aromatic descriptors, which mimic aroma of fruits and flowers, sometimes found in small percentages in the grape. Its abundance in wine results mainly from the secondary metabolism of S. cerevisiae during alcoholic fermentation [4].
In the present study, fatty acid ethyl esters are the most common. Ethyl octanoate, which gives a fruity aroma and anise [16], is the major ester in the three strains analyzed, representing approximately 10% of the total volatile fraction in wines produced by strains B and C. However, in wine produced with A, it represents just over 5% confirming the influence of the strain on the biosynthesis of this VOM of oenological interest. Ethyl decanoate obtained the second highest percentage in B and C (about 3.2%), however, in A the value was <1%. Ethyl hexanoate, on the other hand, represented about 2.3% in A and C, reducing its synthesis in B.
Ethyl acetate, isoamyl acetate and 2-phenylethyl acetate were the only esters identified, belonging to the group of higher alcohol acetates, being strain C with the highest percentages of these volatiles (5.1%). These results can be explained by the enzymatic esterification reaction between acetic acid and the corresponding higher alcohol, as well as the influence of the strain on the biosynthesis of these compounds [5].

Fatty Acids
In general, strain A has the lowest percentage of fatty acids and, in opposite, strain B showed the highest percentage and the largest number of these VOMs (Table 3).
Acetic acid is the largest volatile fatty acid in wines, representing about 90% of this chemical family [4]. However, in the present study only about 54% of the fatty acid fraction was identified in strains B and C. The same was verified by Jiang et al. [29] which obtained between 45% and 60% of this compound, differing slightly with the results obtained in A (63%). Mauriello et al. [5], analyzed 36 strains of S. cerevisiae, in which acetic acid represented between 10% and 100% of the chemical family of fatty acids, and in other cases was not identified, with a relationship between the geographical area from which they were isolated and the production of acetic acid. The contribution of acetic acid to the total volatile profile is approximately 5.8% in A and C, whereas in B, a percentage of about 8% was registered.
In cellar environments, the conditions that provide the greatest biosynthesis of this compound are reported in wines whose musts have been clarified and fermented under anaerobic conditions [4]. According to Gayon et al. [7] and Garden-Cedal et al. [34], S. cerevisiae biosynthesizes low percentages of acetic acid in musts produced from grapes in good phytosanitary conditions and moderate levels of initial sugar (<13% v/v of probable alcohol), depending on the strain. The octanoic acid is the second fatty acid in percentage, registered in the three strains studied, whereas, the decanoic acid was biosynthesized in a greater percentage in the strains B and C. Regarding hexanoic acid, it was biosynthesized to a greater extent by strain B.

Carbonyl Compounds
A chemical family composed mainly of aldehydes and ketones, in which 3-octanone is the major carbonyl compound in MC and the most representative volatile organic compound identified in the wines under study. 3-Octanone mimics the aroma of mushrooms and herbaceous aromas [17].
Probably, the presence of 3-octanone in wines is due to the substrate and not to biochemical or chemical reactions.
Considering the carbonylated compounds product of the metabolism of S. cerevisiae, they are found in a very low percentage, representing about 1.7%, 1.0%, and 1.2% of the total volatile fraction of A, B, and C, respectively, with acetaldehyde being the major carbonyl compound, followed by benzaldehyde.

Terpenes
This chemical family constitute the varietal or primary aroma, associated with positive aromatic descriptors, mainly floral aromas [35].
Except for linalool identified only in strain C, it is observed that the terpenes identified in the MC were not found in wines. In this sense, it is concluded that these VOMs underwent rearrangements, resulting in a product of chemical reactions due to fermentation conditions (pH, ethanol content, etc.) [35]. On the other hand, evidence shows the biotransformation and biosynthesis of specific terpenes by S. cerevisiae under winemaking conditions [36][37][38].

Conclusions
The strains of S. cerevisiae play a fundamental role in the quality and typicality of wines produced in Demarcated Regions, through biosynthesis of VOMs with an impact on wine aroma, whose diversity and proportion depend closely on the yeast strain that drives the fermentative process.
Strains residing in wine-growing environments (grapes and wineries) have genetic variability adapted to the specificities of the ecosystem, which are reflected in the final product. Thus, in order to produce high quality wines, highlighting the typical characteristics, several studies have been carried out in the molecular identification and VOMs characterization of indigenous strains of S. cerevisiae from several demarcated regions. To date, to the best of our knowledge, no scientific study has been reported on this subject in DRM.
In this first approach, most of the VOMs (extracted and identified through HS-SPME/GC-MS) produced during alcoholic fermentation, persist in aged MWs. It is verified the importance of the identification and selection of the strain, considering its relevant impact during alcoholic fermentation and in the following vinification processes.
Eight strains of S. cerevisiae were identified by analyzing 100 RFLP-mtDNA restriction patterns. Comparing with those reported in scientific studies, it is possible to infer that the strains identified in the present study are probably indigenous from DRM.
The two predominant strains (B and A) were selected to characterize the volatile profile. The microfermentation tests inoculated with the pure strain showed that each of the strain of S. cerevisiae influenced (biosynthetized) differently the volatile profile with possible impact on the wine aroma. Higher alcohols, esters, and fatty acids were the dominant chemical families, and can be used as a differentiating parameter of the strain. Therefore, VOMs biosynthesized during fermentation can be called "the raw material of the bouquet" and "the fingerprint of wine".