Volatilome of brine-related microorganisms in a curd-based medium

The possible contribution of brine-derived microflora to the sensory attributes of cheese is still a rather un-explored field. In this study, 365 bacteria and 105 yeast strains isolated from 11 cheese brines were qualitatively tested for proteolytic and lipolytic activities, and positive strains were identified by sequencing. Among bacteria, Staphylococcus equorum was the most frequent, followed by Macrococcus caseolyticus and Corynebacterium flavescens . As for yeasts, Debaryomyces hansenii , Clavispora lusitaniae , and Torulaspora delbrueckii were most frequently identified. A total of 38% of bacteria and 59% of yeasts showed at least 1 of the metabolic activities tested, with lipolytic activity being the most widespread (81% of bacteria and 95% of yeasts). Sub-sequently 15 strains of bacteria and 10 yeasts were inoculated in a curd-based medium and assessed via headspace-solid phase microextraction coupled with gas chromatography-mass spectrometry to determine their volatilome. After a 30-d incubation at 12°C, most strains showed a viability increase of about 2 log cfu/ mL, suggesting good adaptability to the cheese environment. A total of 26 compounds were detected in the headspace, carbonyl compounds and alcohols being the major contributors to the volatile profile of the curd-based medium. Multivariate analysis was carried out to elucidate the overall differences in volatiles produced by selected strains. Principal component analysis and hierarchical clustering analysis demonstrated that the brine-related microorganisms were separated into 3 different groups, suggesting their different abilities to produce volatile compounds. Some of the selected strains have been shown to have interesting aromatic potential and to possibly contribute to the sensory properties of cheese.


INTRODUCTION
The volatile profile is generally recognized as one of the most relevant aspects for evaluating cheese quality.
It reflects the flavor of cheese and is recognized as being useful in characterizing the cheese and defining its ties to the geographic area and cheesemaking process (Innocente et al., 2013).Several metabolic reactions are responsible for the production of flavor compounds in cheese-like products and biochemical pathways are largely influenced by the diversity, growth, and metabolism of the microbial community during ripening (Maifreni et al., 2002;Hassan et al., 2012;Pogačić et al., 2015).Despite the ingredients being the same for most cheese varieties (i.e., milk, rennet, and salt), the microbiota is a key factor in creating heterogeneous active ecosystems, with relevant effects on the final quality and authenticity of the product (Gobbetti et al., 2018).
The microbiota of cheese is certainly related to the autochthonous microorganisms, which biologically originate from raw milk and mainly consist of nonstarter lactic acid bacteria.Nevertheless, there is a constant need to improve cheese flavor through the addition of starter lactic acid bacteria with aroma-improving capability, especially for cheeses made from thermally treated milk (Pogačić et al., 2015).Therefore, the milk autochthonous microbiota, added starters, and adjunct cultures are the most important ripening agents that contribute to the development of specific intense flavors and mainly justify the differences among cheeses.
Unavoidably, cheese microbiota is also related to all the processing steps that can affect its biodiversity, e.g., pasteurization, coagulation, cooking temperature, brining and ripening conditions (Yeluri Jonnala et al., 2018).During cheesemaking and ripening, the modification of the physicochemical features of the curd plays a pivotal role in microbial dynamics.The concomitant contribution of biotic (e.g., microbial interplay) and abiotic (e.g., pH, salt, oxygen, and nutrient availability) factors continuously affect the microbial balance (Montel et al., 2014;Calasso et al., 2016).
The microbiota coming from the dairy environment, which includes utensils, brine containers, and ripening rooms, has often been referred to as an important source of microorganisms that have a central role in cheesemaking (Bokulich and Mills, 2013).During cheesemaking, such resident microbial populations, generally named the house microbiota, interact with the microbes coming from raw milk and starters (Somers et al., 2001;Calasso et al., 2016).A microbial transfer from the cheese and the environment occurs, generating a specific equilibrium, which affects the ripening kinetics and the cheese quality (Stellato et al., 2015).
Salting in brine is the most common procedure to increase the salt content of cheese, where a molded curd is normally soaked for a short time, up to several days, in a solution containing 5 to 25% sodium chloride.During this processing step, the difference in osmolarity drives the transfer of Na + and Cl − from the brine into the outer layers of curd, which in turn largely influences cell viability and metabolism, as well as enzyme activity (Fox et al., 2017).Brines are used for up to several months and for many batches; thus, an abundant house microbiota deriving from tanks, salt, water, and curd, is frequently present and gradually increases in concentration (Schirmer et al., 2014).Several factors drive the microbiological and chemical composition of brines, such as the type of cheese, the cheesemaking procedures, and the salt concentration.The microbiota of cheese brines is rarely studied, and studies have almost entirely centered on the potential presence of pathogens (Larson et al., 1999;Osaili et al., 2014).Indeed, brines can harbor Listeria monocytogenes, as well as Staphylococcus aureus (Larson et al., 1999;Wemmenhove et al., 2014).However, during cheese manufacture, the brines can also be enriched with microorganisms favorable to cheesemaking.Mesophilic bacteria can range between 2.0 and 6.5 log cfu/mL in used brines.The main microbial phyla are Firmicutes and Proteobacteria, as well as Actinobacteria and Bacteroidetes, with Micrococcaceae, Streptococcaceae, Lactobacillaceae, and even Bifidobacteriaceae being the most widespread families (Marino et al., 2017).These microorganisms come in contact with the cheese rind during soaking, and it is feasible they contribute to the cheese ripening through their metabolism, which may affect the characteristics and sensory profile of the product.In this context, most of the research is about smear-ripened cheeses (Mounier et al., 2006a,b).However, as regards the possible contribution of the microflora present in the brines used for cheeses other than the latter, to the best of our knowledge, no peer-reviewed literature exists.Considering this, the characterization of these microorganisms producing odor-active volatiles in cheese-related conditions is an ongoing challenge in this research field.
This study aimed at screening brine-related microorganisms for their aptitude to produce volatiles in cheese.To this end, 470 microbial strains isolated from cheese brines were qualitatively screened in terms of proteolytic and lipolytic activity.Positive strains (bacteria and yeasts) were identified, and 15 strains of bacteria and 10 yeasts were characterized for their capability to grow and produce volatile compounds in a curd-based medium under conditions simulating cheese manufacturing and ripening.

MATERIALS AND METHODS
Because no human or animal subjects were used, this analysis did not require approval by an Institutional Animal Care and Use Committee or Institutional Review Board.

Microbial Strains
A total of 470 microbial strains (365 bacteria and 105 yeasts) were isolated from 11 brines from dairies located in the northeast area of Italy (Marino et al., 2017).The selected brines were in use for at least 2 mo but were not in contact with curd at the time of sampling.Isolation was carried out from the count plates of gelatin sugar-free agar (Oxoid, Milan, Italy), mannitol salt agar (Oxoid), and oxytetracycline glucose yeast extract agar (Oxoid).From each brine, at least 15 morphologically different cultures were isolated, based on cultural criteria (shape, size, and color of colonies).Isolated bacterial and yeast strains were stored at −80°C in brain heart infusion (BHI; Oxoid) with 30% glycerol added and malt extract broth (Oxoid) with 30% glycerol, respectively.

Qualitative Screening of Proteolytic and Esterase Activity
Each strain was subcultured in BHI (bacteria) or malt extract broth (yeasts) for 24 h.To detect proteolytic activity, a loopful of culture was streaked onto skim milk agar (Oxoid), and the plates were incubated at 30°C.After 48 h, the plates were checked for the presence of a clear zone around the culture, which was taken as a positive indicator of proteolysis (Pereira et al., 2001).Esterase activity was evaluated by streaking a loopful of each culture onto tributyrin agar (Fluka Analytical, Buchs, Switzerland).The plates were incubated at 30°C and, after 48 h, observed for the presence of a clarification halo around the colonies, which was considered a positive indicator of esterase activity (Morandi et al., 2013).

Identification of Strains
Genomic DNA was isolated using the InstaGene Matrix (Bio-Rad Laboratories, Hercules, CA).The isolates were then submitted to partial 16S rRNA (bacteria) or 23S (yeasts) gene amplification (Kurtzman and Robnett, 1997;Carraro et al., 2011).The amplified fragments (~700 bp of the V1-V3 region of the 16 S rRNA gene for bacteria, and ~600 bp of the D1/D2 domain of the 26S rRNA gene for yeasts) were sequenced, and the resulting sequences were aligned with the closest sequences available in the GenBank database (≥98% homology; http: / / www .ncbi.nml.nih.gov/BLAST).

Growth in the Curd-Based Medium
A curd-based medium was prepared using a 10-d curd of an Italian semi-hard cheese according to the modified protocol of Pogačić et al. (2015).Because 10-d curd is depleted of most of its lactose due to the metabolism of starter lactic acid bacteria, the curd-based medium was supplemented with a low concentration of lactose to stimulate the growth of microbes that use milk lactose as carbon sources during the first steps of cheese ripening.Additionally, to deliver the nitrogen fraction released in cheese during maturation, peptone was added to the medium (Pogačić et al., 2015(Pogačić et al., , 2016)).Moreover, NaCl was added in addition to that naturally present in the 10-d curd, to counteract the dilution effect during preparation of the curd-based medium and to simulate the conditions of cheese during ripening.
Then, 100 g of curd was finely chopped and added to 200 mL of a solution containing 1.2 g/L lactose (Sigma, Milan, Italy), 1.2 g/L bacteriological peptone (Oxoid), and 18 g/L NaCl (Sigma), and then homogenized in a Stomacher Lab-Blender (PBI International, Milan, Italy) for 30 s.The homogenate was coarsely filtered with sterile gauze and then with Whatman 1 filter paper (Sigma).Aliquots of 10 mL were then portioned into 20-mL vials, which were capped with Teflon septum (Agilent Technologies, Santa Clara, CA) and sealed with an aluminum cap before being sterilized at 110°C for 15 min.Each selected strain was streaked on BHI or malt extract broth agar (for bacteria and yeasts, respectively) and incubated at 30°C for 48 h.For each strain, 5 to 10 colonies were used to inoculate 250 µL of maximum recovery diluent (Oxoid).Then, the cultures were counted using a Burker chamber and, if necessary, diluted to about 10 9 cfu/mL with maximum recovery diluent.A total of 100 µL of each culture was used to inoculate the curd-based medium in 3 biological replicates (final concentration about 10 7 cfu/mL), and vials were incubated at 12°C for 30 d.The viability was quantified at the time of inoculation and after 15 and 30 d. Non-inoculated control medium (CTR) was also prepared and analyzed under the same conditions.Viable counts were evaluated by spreading decimal dilutions of the inoculated curd-based medium onto BHI and oxytetracycline glucose yeast extract agar plates for bacteria and yeasts, respectively.Plates were incubated at 30°C for 48 to 72 h.

Analysis of Volatile Compounds
The volatilome of samples at 30 d of incubation was determined by headspace-solid phase microextraction (HS-SPME) coupled with GC-MS.Briefly, 5 g of the cheese-based medium was placed in a 20-mL vial, which was crimped with a Teflon septum (Agilent Technologies) and an aluminum seal and then set in a water bath at 60°C for 30 min to accelerate equilibrium of headspace volatile compounds between the cheesebased medium and the headspace.A fiber assembly coated with divinylbenzene/carboxen/polydimethylsiloxane (DVB/CAR/PDMS), 30-µm film thickness, 2 cm long (Supelco, Bellefonte, PA) was exposed for 20 min in the sample and kept in the water bath at 60°C (Innocente et al., 2007(Innocente et al., , 2013)).Preliminary optimization of SPME analytical conditions was carried out by testing the following parameters: conditioning and extraction temperature (40°C or 60°C), equilibration time (20, 30, 40, or 60 min), fiber coating (CAR/PDMS or DVB/CAR/PDMS), and fiber exposure time (10, 20, 30, or 40 min).The DVB/CAR/PDMS fiber performed best in terms of sensitivity and repeatability, extraction temperature of 60°C allowed the highest analytical response, the minimum equilibrium time was found to be 30 min, and a 20-min fiber exposure time was chosen to avoid volatile compounds competition phenomena.
A system consisting of a GC2010 gas chromatograph and a QP2010 Ultra quadrupole mass spectrometer (Shimadzu, Kyoto, Japan) was used to separate and identify the volatile compounds, which were thermally desorbed in splitless mode from the SPME fiber in the injector port (set up at 250°C).The split valve was left open 3 min after injection.Helium was used as the carrier gas, with a flow rate of 1 mL/min.Separation was performed on an Econo-Cap Ec-Wax column (30 m long, 0.25-mm inner diameter, 0.25-µm film thickness; Alltech, State College, PA).The oven temperature was set at 50°C for 5 min and then increased to 230°C (10°C/min, held 10 min) to a final temperature of 250°C (10°C/min, held 10 min).The ion trap detector MS conditions were as follows: the temperatures of the manifold and the transfer line were 170°C and 250°C, respectively; electron impact mass spectra were recorded at 70 eV; the ionization current was 10 µA; the scan rate was 1.5 scans/s with a mass range from 30 to 350 m/z.Compounds were identified by matching their mass spectra with the reference mass spectra of the NIST library (NIST/EPA/NIH 20 Mass Spectral Library, John Wiley & Sons Inc., Hoboken, NJ) and by comparing their Kovats retention indexes with those reported in the literature.Kovats retention index was calculated in relation to the GC/MS retention time of n-alkanes (C7-C30).Data were expressed as absolute peak areas.

Statistical Analysis
For each strain and for each analysis time, 3 replicates were prepared.As for SPME analysis, for each of the 26 volatile compounds identified, the average absolute peak area was calculated and the coefficient of variation was in all cases <10%.An ANOVA was processed using Statistica for Windows 8.0 (StatSoft, Tulsa, OK).Tukey's honestly significant difference test was used for multiple comparisons between means (P < 0.05).Principal component analysis (PCA) and hierarchical clustering analysis were conducted using Origin Pro 9 software (OriginLab, Northampton, MA).Data for PCA and clustering analysis were pre-processed, in the specific log10 [x] -transformed and scaled, and Euclidean distances and Ward's minimum variance method were used for clustering.

Isolation, Identification, and Characterization of Strains
From 11 brines used for the salting of soft and semihard cheeses produced in Italian dairy plants, 365 bacteria and 105 yeast were isolated.The isolates underwent a qualitative test to verify their proteolytic and lipolytic ability.In all, 140 out of 365 (38%) bacteria and 62 out of 105 (59%) yeasts were positive for at least 1 of the metabolic activities tested.Lipolytic activity was the most widespread among positive strains, being present in 81% of bacteria and 95% of yeasts (Figure 1).Proteolytic activity was present in only 27 bacteria and 3 yeasts.Proteolytic activity is generally less widespread than lipolytic activity among the dairy secondary non-lactic acid bacteria microbiota (Ruaro et al., 2013;Tulini et al., 2016;Merchán et al., 2022).We found 30 bacteria and 4 yeasts positive for both activities.
Among positive bacteria, most belonged to the Staphylococcus genus, presumably due to the halotolerant characteristics of the members of this microbial group (Table 1).37 isolates were Staphylococcus equorum, which is a species frequently found in the processing environment (Irlinger et al., 1997;Ruaro et al., 2013).The other isolates were Staphylococcus epidermidis, Staphylococcus saprophyticus, Staphylococcus sciuri, Staphylococcus aureus, Staphylococcus auricularis, and Staphylococcus hominis.Presence of S. aureus was detected in 4 out of 11 brines, which suggests that they may constitute a possible source of contamination by this microorganism (Marino et al., 2017;Al-Nabulsi et al., 2020).The presence of potentially pathogenic microorganisms constitutes a warning for immediate regeneration of the brines, especially when these are used for salting fresh or soft cheese.Another rather frequent species in brines was Corynebacterium flavescens (19 strains), followed by Corynebacterium variable (12).These species belong to the group of so-called coryneforms, which also includes the genera Arthrobacter, Brevibacterium, and Microbacterium, also isolated in this study.These are microorganisms that often produce pigmented compounds and which are widespread on the surfaces of washed-rind cheeses, where the halophilic and halotolerant microflora has intense lipolytic and proteolytic activity (Brennan et al., 2002).
They owe their presence in cheese brines to the fact of being particularly halotolerant or psychrotrophic, or both (Coton et al., 2012;Haastrup et al., 2018;Lauková et al., 2021).Among the lactic bacteria, the only species isolated in this study was Lactococcus lactis.
Regarding yeast, Debaryomyces hansenii was most frequently isolated, followed by Clavispora lusitaniae and Torulaspora delbrueckii.Debaryomyces hansenii and T. delbrueckii have already been identified in cheese brine (Seiler and Busse, 1990), whereas Clav.lusitaniae has been found in cheese (El-Sharoud et al., 2009).Other less frequently isolated species belonged to the genera Candida, Rhodotorula, Trichosporon, and Kluyveromyces.The presence of many yeast species in the brines is presumably related to their halotolerance, the ability to grow at low temperatures and low pH values, which are especially common in brines that have been in use for a long time.

Growth in Curd-Based Medium
A total of 15 strains of bacteria and 10 yeasts were selected based on considerations relating to their frequency in brines, relevance in the dairy field, and absence of pathogenic traits.The strains were characterized by their ability to grow and produce volatile compounds in a curd-based medium.To hypothesize a possible dairy role of the isolated strains, it was necessary to grow them in a system as similar as possible, at least from a chemical point of view, to the cheese during ripening.For this reason, curd was used as the major ingredient of the medium developed for analysis.After inoculation, the cultures were incubated for 30 d at 12°C, which is a condition mimicking the initial steps of cheese ripening.During the incubation, the abilities to grow and produce volatiles were evaluated.
Most strains grew during 30-d incubation, with final populations ranging from 7.58 to 8.98 log cfu/mL (Table 2).The population remained at similar levels during incubation in 6 bacterial strains.
Regarding yeasts, the highest populations after 30 d were observed for Candida zeylanoides and Clav.lusitaniae, which exhibited about a 2-log increase.The ability of these species to grow in cheeses has already been observed (Tofalo et al., 2014;Ceugniez et al., 2017a).As for bacteria, S. equorum, Idiomarina loihiensis, and Cor.flavescens reached the highest final viabilities.Staphylococcus equorum is present in raw milk and is the prevalent species on the wooden boards used in ripening rooms of traditional Sicilian cheeses (Meugnier et al., 1996;Settanni et al., 2021).On the surface of hard raw milk cheeses, S. equorum is one of the "first colonizers" of the rind and can contribute to the formation of volatile substances (Wolfe et al., 2014;Quijada et al., 2020).As for I. loihiensis, it is a marine deep-sea species generally not related to dairy products, except for its presence in brines (Vermote et al., 2018).As far as we know, this is the first time that its ability to grow on a dairy substrate has been documented (Galperin, 2005).However, Cor. flavescens is of dairy origin and is frequently found on the surface of cheeses (Ceugniez et al., 2017b).It is a microbial species capable of growing at 10°C and in the presence of high salt concentrations.Indeed, this species has been isolated in cheese until d 16 of ripening (Brennan et al., 2002).In general, in this study, the presence of proteolytic or lipolytic activity was not related to the ability to grow in the curdbased medium.For example, strains B3 and Y1, both  negative for proteolytic activity, had growth capacities comparable to those of strains B4 and Y6, which lacked proteolytic activity (see Table 2 for strain abbreviation definitions).

Volatile Compounds
Overall, about 26 different volatile compounds were identified, consisting of 5 alcohols, 3 acids, 9 carbonyl compounds (aldehydes and ketones), 3 heterocycles (furans), 2 esters, 3 sulfur compounds, and 1 lactone (Table 3).These compounds have already been identified in numerous hard and semi-hard cheeses (Curioni and Bosset, 2002;Smit et al., 2005).However, it should be noted that the ability of a volatile compound to contribute to the aroma of a complex system as a cheese depends on various factors, such as its volatility and liquid/vapor partition equilibrium, competition for the partition equilibrium in the mixture with other compounds, and characteristics of the matrix, as well as its sensory threshold (Bertuzzi et al., 2018).
Among the chemical classes identified, the most numerous were carbonyl compounds (primarily ketones) and alcohols, which are principally produced from fatty acids and amino acids (Marilley and Casey, 2004).
Multivariate analysis and hierarchical clustering analysis were carried out to visualize the global differences in volatile compounds at 30 d of incubation in a curd-based medium (Figure 2 and Figure 3).Considering clustering analysis, the microbial strains were clustered into 3 different groups (I, II, and III), suggesting discrimination based on the volatilome (Figure 3).The cumulative contribution rate of the first (PC1) and second (PC2) principal components was 51.15% (PC1 32.98%, PC2 18.17%).The CTR was situated in the right negative region of the PCA biplot and clustered within strains belonging to group II (B1, B7, B8, B10, B13, Y1, Y4, and Y6).
Both CTR and the strains belonging to group II were characterized by low absolute areas of the total volatile components at 30 d of incubation (Figure 2).These results lead to the hypothesis that these bacteria may slightly influence the volatile profile of cheese.However, a higher absolute area of acetic and isobutyric acids was found in group II strains, especially in Macrococcus caseolyticus (B8) and C. zeylanoides (Y4) samples.Generally, in cheese environments, acetic acid is produced from heterofermentation of lactose, whereas isobutyric acid is produced from valine catabolism (Curioni and Bosset, 2002).Moreover, the headspaces of strains belonging to the II cluster were characterized by the presence of heterocyclic compounds, including 2-ethyl furan, 2-pentyl furan, and 2-hexyl furan.In processed foods, furans are chemically generated during the thermal process as a consequence of the Maillard reaction (Barbieri et al., 1994;Nájera-Domínguez et al., 2014).Otherwise, in cheese, these compounds could also originate enzymatically from microorganisms, even if their formation during cheese ripening is not yet well understood (Curioni and Bosset, 2002;Ndagijimana et al., 2006).
In the left positive quadrant and a very limited region of the left negative quadrant of the PCA biplot (Figure 2) were situated strains most different from CTR, which were clustered in group I (Figure 3).Three bacterial strains (B11, B14, and B15) and 7 yeast strains (Y2, Y3, Y5, Y7, Y8, Y9, Y10) showed higher areas than CTR at 30 d of incubation.The PCA and heatmap revealed that alcohols and esters and a higher total area of the volatile compound were mainly related to these samples.Regarding bacteria, the prevalent alcohols were ethanol, 2-phenylethanol, and 2-octanol (Figure 2).Ethanol, which is mainly produced by the metabolism of lactose, was present with a larger absolute area in the headspaces of samples inoculated with Corynebacterium variabile (B4), Psychrobacter halophilus (B14), and S. equorum (B15; Figure 3).2-Phenyl ethanol derives mainly from aromatic amino acids and was detected in relatively high amounts in curd-based media inoculated with Cor.variabile (B4) strain.Secondary alcohols, such as 2-octanol, are notoriously produced from the degradation of the released fatty acids and reduction of methyl-ketones (Xiao et al., 2020).As regards yeasts, in contrast, greater presence of ethanol was detected in Candida pseudoglaebosa (Y2), Candida thasoenensis (Y3), Clav.lusitaniae (Y5), Kluyveromyces marxianus (Y7), Rhodotorula mucilaginosa (Y8), T. delbrueckii (Y9), and Trichosporon coremiiforme (Y10).These yeasts were also able to produce moderate amounts of 2-octanol.
Ethyl acetate and ethyl propanoate were the prevalent esters in samples inoculated with yeast strains belonging to the first group, except for Y5 and Y7.These compounds derive from the esterification reaction between an alcohol and free fatty acids.This is the first report that suggests a possible contribution of C. thasoenensis (Y3) to volatile compound production.In general, yeasts showed greater ability to influence the aromatic profile of the curd medium compared with bacteria (Figure 3).Overall, it was not possible to demonstrate a reliable relationship between high values of total area present in the headspace of samples inoculated with different microorganisms and their ability to actively grow in the cheese medium.For example, 6 out of the 15 tested bacterial strains did not significantly grow (P > 0.05), but they contributed to increase the total absolute area of volatile compounds.We cannot exclude the possibility that, in this case, the increase in the volatile fraction could be due to the activity of intracellular enzymes released by autolysis (Hannon et al., 2007;Lazzi et al., 2016).Conversely, 3 strains (B3, B8, and B10) grew but did not change the total absolute area of volatile compounds.
The strains placed in the right positive quadrant of the PCA biplot were characterized by the presence of sulfur and carbonyl compounds (Figure 2).These strains were clustered by hierarchical analysis in group III (Figure 3).In particular, the most representative sulfur compounds were dimethyl disulfide, dimethyl trisulfide, and methanethiol, which all originated from metabolism of the amino acid methionine (McSweeney, 2004).Curd media inoculated with Chryseobacterium indologenes (B2), Microbacterium lacticum (B9), and Psychrobacter alimentarius (B12) had large prevalence of these compounds in the headspace and were also characterized by the presence of small relative area amounts of δ-decalactone and notable presence of ketones.δ-Lactones are formed from the β-oxidation of fatty acids (Marilley and Casey, 2004).More precisely, β-oxidation leads to the release of hydroxy acids, which then undergo intramolecular esterification with the formation of lactones, which are cyclic compounds (McSweeney and Sousa, 2000).Headspaces of group III strains were also characterized by the presence of alkan-2-ones compounds, such as 2-heptanone and 2-nonanone (Figure 3), which are ketones produced from the oxidation and decarboxylation of free fatty acids (McSweeney and Sousa, 2000).As for Chr.indologenes, Halomonas alkaliphila, Microb.lacticum, and Psychr.alimentarius, these are the first data relating to their ability to produce volatile substances.

CONCLUSIONS
Data confirmed a wide microbial diversity among cheesemaking brines used to salt semi-hard and hard Italian cheeses, with many bacterial and yeast strains showing proteolytic or lipolytic activities.The isolated species coincide in part with those that have been already identified on the surfaces of ripened cheese, and it is feasible that their metabolic activity can deeply influence the characteristics and sensory profiles of the cheese.Most of the strains were able to grow in the curd-based medium used in this study at 12°C and to produce volatiles.However, even strains that showed the least ability to grow in the curd medium were sometimes found to be able to influence the volatile profile in the headspace.This preliminary study showed that brines can represent a reservoir of microorganisms that inevitably encounter cheese and could contribute to the evolution of its characteristics during ripening, which could help in selecting new microbial cultures for dairy use.
Figure 1.Numbers of proteolytic and lipolytic strains isolated from 11 cheese brines.

Figure 2 .
Figure 2. Principal component (PC) analysis biplot of volatile compounds in curd-cheese medium-inoculated samples after 30 d of incubation.Red dots: group I; blue dots: group II; green dots: group III.SeeTable 2 for definitions of strain abbreviations.

Figure 3 .
Figure 3. Hierarchical clustering analysis with heatmap representation based on Euclidean distances and Ward's minimum variance of volatile compounds in curd-cheese medium-inoculated samples after 30 d of incubation.See Table2for definitions of strain abbreviations.

Table 1 .
Innocente et al.: VOLATILOME OF BRINE-RELATED MICROORGANISMS Microbial species and numbers of isolates detected and their frequency in 11 Italian cheese brines

Table 2 .
Proteolytic and lipolytic activities of bacteria and yeast strains (+ or − indicates presence or absence of activity) and growth (log cfu/ mL) in curd-based medium, incubated for 30 d at 12°C a-c Mean values (mean ± SD) with different letters differ significantly at P ≤ 0.05 according to Tukey's test.

Table 3 .
Innocente et al.: VOLATILOME OF BRINE-RELATED MICROORGANISMS List of volatile compounds, grouped according to chemical classes, identified in the headspace of curd-related medium samples inoculated with selected bacteria and yeast strains and incubated for 30 d at 12°C