Queijo Serra da Estrela PDO Cheese: Investigation into Its Morpho-Textural Traits, Microbiota, and Volatilome

Queijo Serra da Estrela is a PDO Portuguese cheese produced through coagulation of raw ewe’s milk using cardoon (Cynara cardunculus L.) flowers. The present research was aimed at depicting an up-to-date and comprehensive overview of the microbiota of Queijo Serra da Estrela cheese. To this end, viable counting and metataxonomic analysis were carried out on cheeses sampled from four Portuguese artisan producers. Physico-chemical and morpho-textural analyses were also performed, together with the analysis of volatile organic compounds (VOCs). Finally, non-starter lactic acid bacteria (NSLAB) isolated from the cheeses were characterized for their enzymatic activities using a semi-quantitative method. According to the metataxonomic analysis, Lactococcus lactis and Lactococcus piscium were the species occurring at the highest relative abundance. The isolates collected from the cheeses were assigned to Enterococcus durans, Enterococcus faecalis, Enterococcus faecium, Enterococcus lactis, Levilactobacillus brevis, Latilactobacillus graminis, Leuconostoc mesenteroides, and the Lacticaseibacillus casei group. The enzymatic characterization of these cultures highlighted esterase, aminopeptidase, acid phosphatase, beta-galactosidase, alpha-glucosidase, and beta-glucosidase among the major enzymatic activities. Fungal populations were dominated by Debaryomyces hansenii and Kurtzmaniella zeylanoides; however, species rarely found in cheese (e.g., Candida boidinii, Vishniacozyma victoriae, and Starmerella) were also detected. The volatile compounds characterizing the analyzed cheeses were carboxylic acids and esters, followed by carbonyl compounds and alcohols.


Introduction
Cheese represents one of the most studied sources of protein that are part of the Mediterranean diet [1]. Indeed, in the Mediterranean biogeographical Region, that includes among others, France, Italy, Greece, Cyprus, Morocco, Tunisia, Spain, and Portugal, a plenty of traditional cheeses is produced [1]. In these countries, traditional cheeses are still manufactured in accordance with ancient traditions that should be preserved and valorized to maintain the uniqueness and quality of these products [2].
Cheese is the result of the coagulation of milk proteins (caseins) through the addition of animal rennet or other milk coagulants (e.g., vegetable rennet) [3]. Cheese is mostly obtained from cow's, ewe's, or goat's milk, whether raw or pasteurized. Among milk coagulants, calf rennet, which is obtained from the fourth stomach of suckling calves, is the most widely exploited at industrial level [4]. In addition, extracts from plants ascribed to Euphorbia, Lactuca, Solanum, Streblus, and to members of the Asteraceae family, commonly Hence, the present research was aimed at depicting an up-to-date comprehensive overview of the microbiota of Queijo Serra da Estrela PDO cheeses sampled from four producers located in Portugal. To this end, bacterial and fungal communities naturally occurring in the sampled cheeses were studied through viable counting and metataxonomic analysis. Moreover, a pool of autochthonous NSLAB was isolated from the cheeses, identified, and further characterized for their enzymatic activity using a semi-quantitative method. Finally, physico-chemical and morpho-textural analyses were also carried out on the cheeses, together with the analysis of volatile organic compounds (VOCs).

Sampling of Cheese
Eight samples of Queijo Serra da Estrela PDO cheese were collected from four Portuguese artisan producers. Two samples of the same batch were collected from each producer, as follows: samples E1 and E2 from producer 1, located in Vila Nova de Tazem; samples E3 and E4 from producer 2, located in Minhocal; samples E5 and E6 from producer 3 and samples E7 and E8 from producer 4, both producers being in São Cosmado. All sampled cheeses had the same height (~5 cm), diameter (~10 cm), and weight (~500 g). The ripening time was as follows: approximately~60 days for cheeses collected from producer 1, and 35-40 days for cheeses collected from producer 2, 3, and 4.
No starter cultures were used for the manufacturing of these cheeses. All samples were transported to the laboratory under refrigerated conditions and stored at +4 • C until analysis.

Evaluation of Physico-Chemical Features
A pHmeter (Hanna Instruments, Padova, Italy) equipped with a HI2031 solid electrode (Hanna Instruments) was used for pH determination.
For measurement of total titratable acidity (TTA), 10 g of each sample was added with 90 mL of deionized water and homogenized using a Stomacher 400 Circulator (VWR International PBI, Milan, Italy) at 260 rpm for 5 min. The results were expressed as amount (mL) of 0.1 N NaOH solution needed to titrate the pH of the cheese homogenates back to 8.3 [5].
The cheese content of lactic and acetic acid was determined using the D-/L-Lactic Acid (D-/L-Lactate) (Rapid) test kit and the Acetic Acid (Acetate Kinase Manual Format) test kit (Megazyme, Bray, Ireland), respectively [22].
Specific volume, expressed as g mL −1 , was calculated as previously detailed by Cardinali et al. [5].
For each physico-chemical feature considered, the results of three independent measurements were reported as mean ± standard deviation.

Analysis of Morpho-Textural Parameters
Cheese color was assessed using a Konica Minolta CR-310 chroma meter (Ramsey, NJ, USA). Color parameters (L*, a*, b*, chroma, and hue) were taken in triplicate, and expressed as mean ± standard deviation [23].
Cheese texture was evaluated through TPA test (Texture Profile Analysis) using an AXIS texture analyzer FC200STAV500 (AXIS, Gdansk, Poland) provided with the software "AXIS FM" as previously reported by Cardinali et al. [5]. In more detail, an aluminium 20 mm diameter cylindrical probe was used in a double compression test (TPA) to penetrate to 50% depth, at 1 mm s −1 speed test. Hardness (N) was the force at the maximum deformation, whereas cohesiveness, springiness, chewiness, and resilience were calculated from the peaks. Analysis was carried out in quadruplicate at 25 • C on cheese cylinders (20 mm height and 20 mm diameter) cut from a cheese slice of each sample. The results were reported as mean ± standard deviation.

Microbiological Analyses
Ten grams of each sample were homogenized with 90 mL of sterile peptone water (Oxoid, Basingstoke, UK) in a stomacher apparatus for 2 min at 260 rpm [24]. Serial ten-fold dilutions were prepared from the cheese homogenates and used for viable counting of the following microbial groups: presumptive mesophilic lactococci, presumptive thermophilic streptococci, presumptive mesophilic lactobacilli, coagulase-negative cocci, enterococci, and eumycetes (yeasts and molds). Growth media and incubation temperatures were those already detailed by Cardinali et al. [25].
The results of viable counting, expressed as Log of colony-forming units (cfu) per gram of sample, were reported as mean of two biological replicates ± standard deviation.

DNA Extraction and Sequencing
The E.Z.N.A. soil DNA kit (Omega Bio-tek, Norcross, GA, USA) was used for the extraction of total microbial DNA from the cell pellets obtained by centrifugation of 1 mL of each cheese homogenate (dilution 10 −1 ) prepared as previously described. DNA extracts were then checked for quantity and purity using a Nanodrop ND 1000 (Thermo Fisher Scientific, Wilmington, DE, USA), and further standardized using the Qubit ds Kits.
For each sample, the DNA extracts obtained from each biological replicate were pooled to reduce the inter-sample variability [26].
A metataxonomic approach was applied to analyze pooled DNAs. The 16S rRNA gene (V3-V4 regions) was amplified using the primers and procedures described previously [27]. The Illumina guidelines were used for purification of amplicons, taggering and pooling. The Illumina MiSeq platform with V2 chemistry was used to generate 250-bp paired-end reads and the raw fastq files obtained were elaborated by QIIME 2 software [28]. Cutapter was used to remove prime sequences, and DADA2 algorithm was used to denoise the obtained reads by using the q2-dada2 plugin in QIIME 2 [29]. Taxonomy classification was performed against the SILVA database by means of the QIIME2 feature-classifier. The amplicon sequence variants (ASVs) with less than five read counts in at least two samples were excluded to increase the confidence of sequence reads.
The raw read data were deposited in the Sequence Read Archive of NCBI under the bioproject accession number PRJNA822519. Colonies grown on the agar plates used for viable counting of lactic acid bacteria were randomly selected, sub-cultured to purity on the same media used for enumeration and stored at −80 • C for long-term maintenance.
Total genomic DNA was extracted from the collected isolates according to Osimani et al. [30] and checked for purity and quantity using a NanoDrop ND 1000 (Thermo Fisher Scientific, Wilmington, DE, USA). DNA extracts were standardized to a final concentration of 100 ng/µL and subjected to PCR in a Mastercycler X50a (Eppendorf, Hamburg, Germany) using the universal prokaryotic primers 27f and 1495r, as described by Osimani et al. [30]. The amplification was verified by electrophoresis in 1.5% (w/v) agarose gel and 0.5X Tris/Borate/EDTA (TBE) buffer containing 0.5 µg/mL GelRed ® Nucleic Acid Gel Stain, 10,000X in water (Biotium, San Francisco, CA, USA). The amplicons were then sent to Azenta Life Sciences (Leipzig, Germany) for purification and sequencing.
The raw sequences, represented in FASTA format, were analyzed with UCHIME2 software tool to uncover chimeras [31] and were trimmed to remove NNNs and misleading data from the terminations. Afterwards, a BLAST search was exploited to compare the obtained sequences with 16S rRNA sequences of type or reference strains deposited at the GenBank DNA database (http://www.ncbi.nlm.nih.gov/, 12 November 2022). Hence, the sequences of the 60 pure cultures were submitted to the GenBank DNA database to acquire the respective accession numbers.
Briefly, a suspension of each isolate in 2 mL of API Suspension Medium was prepared to reach a turbidity of 5-6 McFarland. Sixty-five µL of the resulting suspension was used for the inoculation of each cupule of the API ® ZYM strips (bioMérieux). After incubation at 37 • C for 4 h, 1 drop of ZYM A reagent (bioMérieux) and 1 drop of ZYM B reagent (bioMérieux) were added to each cupule. After resting (at least 5 min), the resulting colours were recorded based on intensity and assigned a semiquantitative notation using a colour code ranging from 0 (no color development) to 5 (maximum color intensity); 1 to 4 corresponded to intermediate color intensities, with "3", "4" and "5" being considered as positive reactions.

Cheese Volatile Profile
Solid phase microextraction (SPME) was used to collect VOCs, as previously described by Belleggia et al. [32]. A Trace 1300 gas chromatograph coupled with a ISQ 7000 single quadrupole mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA) equipped with a Zebron ZB-5ms capillary column 30 m × 0.25 mm i.d., 0.25 µm film thickness (Phenomenex, Torrance, CA, USA) were used according to the procedure previously detailed by Foligni et al. [33]. According to Mozzon et al. [34], the identification of VOCs was made by matching the mass spectral data with NIST/EPA/NIH Mass Spectral Library 2020, and the chromatographic behaviour with published Kovats retention indices (RIs). An automated spreadsheet was used to simplify the calculation of RIs of the unknown components [35].

Statistical Analysis
Significant differences among cheeses were determined by one-way analysis of variance (ANOVA) using the software JMP ® Version 11.0.0 (SAS Institute Inc., Cary, NC, USA) and the Tukey-Kramer's Honest Significant Difference (HSD) test (α = 0.05).
The diversity script of QIIME2 was used to calculate alpha diversity indices. The non-parametric Kruskal-Wallis test in R environment was used to analyze differences between alpha diversity parameters and ASVs frequency.

Physico-Chemical Characterization
The results of physico-chemical analyses are reported in Table 1. As for pH, the overall means were comprised between 5.46 and 5.83, with samples of producers 3 and 4 showing the lowest values and those of producers 1 and 2 the highest ones.
Regarding TTA, overall mean values ranged between 7.52 and 13.97 mL of 0.1 NaOH, with samples from producer 1 showing the lowest values and those from producer 3 the highest ones.
As for organic acids, mean content of acetic acid was comprised between 0.11 and 0.27 g/100 g with no statistically significant differences among samples; regarding lactic acid, the overall mean content ranged between 0.63 and 1.13 g/100 g with samples from producer 1 showing the lowest values.
A w was comprised between 0.940 and 0.975, again with samples from producer 1 showing the lowest values.
Finally, mean values for specific volume ranged between 0.98 and 1.11 g/mL, with samples from producer 4 showing the lowest values and those from producer 1 the highest ones.
The average lightness (L*) values varied from 106.6 to 117.8 resulting in significant differences among the producers. The darkest samples were those collected from producer 1, whereas the lightest ones were those collected from producer 2. Differences in tones were also visible in all the samples; cheeses from producers 2 and 3 revealed greenish shades (a*) varying from −0.76 to −2.15. Cheeses sampled from producers 3 and 1 were characterized by red tones up to 1.10. No significant differences were seen in the average values of yellowish shade (b*) of cheeses collected from producers 1, 3, and 4, whereas those from producer 2 were characterized by significantly lower average b* and chroma values.
For hue, cheeses from producers 1 and 4 significantly differed from those sampled from producers 2 and 3.
The results of texture analysis are reported in Table 3. An overall high level of similarity was seen among cheeses, except for those collected from producer 1, which showed the highest mean value for hardness. Cheeses sampled from producers 1 and 2 differed for cohesiveness, whereas just those from producer 1 showed the highest springiness and chewiness, with mean values of 0.48 and 6.79, respectively. Resilience ranged between 0.16 and 0.18 mean values, with no statistically significant differences among cheeses.

Viable Counting
The results of viable counting are reported in Table 4. Regarding presumptive lactococci, overall mean counts ranged between 7.98 and 8.66 log cfu/g, with samples from producers 3 and 2 showing the lowest and highest counts, respectively.
Overall mean counts of presumptive thermophilic cocci ranged between 7.16 and 8.28 log cfu/g, with cheeses from producer 1 showing the lowest loads.
As for presumptive lactobacilli, overall mean counts were comprised between 8.13 and 8.50 log cfu/g, with no statistically significant differences among producers.
Coagulase-negative cocci showed overall mean counts comprised between 5.34 and 6.42 log cfu/g, with cheeses from producer 4 showing the highest loads.
Overall mean counts of enterococci ranged between 6.56 and 7.84 log cfu/g, with cheeses from producer 3 and 4 showing the lowest and highest counts, respectively.
As for yeasts, overall mean counts were comprised between 1.27 and 4.24 log cfu/g, with cheeses from producer 3 showing the lowest counts; finally, overall mean counts of molds were always < 1 log cfu/g, except for cheeses sampled from producer 2, whose mold counts attested at 4.07 log cfu/g.

Microbiota Composition
The incidence (%) of the bacterial taxonomic groups detected in the cheeses herein analyzed is reported in Figure 1 and Supplementary Table S1.
Briefly, Lactococcus lactis dominated in almost all cheeses (attesting at~25-30% of the relative frequency) except for those collected from producer 4, where this ASV was found with the lowest relative frequency (~3%). Other minor ASVs were also detected; these latter included Leuconostoc mesenteroides (attesting at 5 to 14% of the relative frequency), Latilactobacillus sakei (attesting at 2 to 21% of the relative frequency), Leuconostoc lactis (attesting at 1 to 4% of the relative frequency), and Serratia (attesting at 6 to 8% of the relative frequency) were found in cheeses from producers 2 and 3, whereas Lactococcus piscium was detected in cheeses from producers 3 and 4 (attesting at 37 to 42% of the relative frequency). Enterococcus was also detected in cheeses from producer 4, at a high relative frequency (~16%).
The incidence (%) of the fungal taxonomic groups is reported in Figure 2 and Supplementary Table S2.

Isolation Campaign and Semi-Quantitative Assessment of Enzymatic Activity of Pure Cultures
The closest relatives, the percent identities, and the accession numbers of the sequences obtained from the pool of NSLAB isolated from the cheeses are reported in Table 5. Lacticaseibacillus casei group 99.72% NR_025880 * Percentage of identical nucleotides in the sequence obtained from the non-starter lactic acid bacteria (NSLAB) isolates and the sequence of the closest relative found in the GenBank database. ** Accession number of the sequence of the closest relative found by BLAST search.
The results of semi-quantitative assessment of the enzymatic activities of the pool of isolates are reported in Table 6.
Latilactobacillus graminis Enterococcus faecalis According to the manufacturer's instructions, only the isolates that showed a positive reaction (coded as "3", "4", or "5") were considered as positive for the tested enzymatic activity.
Regarding enterococci, just one E. faecalis isolate (SE12) showed the esterase (C 4) activity, whereas two isolates of E. durans (SE28 and SE29) were positive for esterase lipase (C 8); the same isolates were also positive for leucine arylamidase and beta-glucosidase. SE7 was positive for acid phosphatase, whereas SE22 was positive for naphthol-AS-BI-phosphohydrolase. Six isolates being SE5, SE6, SE13, SE14, SE16, and SE28 showed a positive reaction for betagalactosidase, whereas SE7 and SE12 were positive for alpha-glucosidase.
As for L. brevis, six out of seven isolates, being SE2, SE3, SE11, SE19, SE20, and SE24, were positive for leucine arylamidase, whereas the same isolates but SE20 were also positive for valine arylamidase and acid phosphatase. SE3, SE11, SE19, and SE24 were positive for naphthol-AS-BI-phosphohydrolase, whereas the same isolates but SE19 were also positive for alpha-galactosidase. All the L. brevis isolates were positive for beta-galactosidase and beta-glucosidase, whereas the isolates SE19 and SE24 were positive for beta-glucuronidase; finally, SE11, SE20, and SE24 were also positive for alpha-glucosidase.
Regarding the L. casei group, three out of six isolates, being SE8, SE9, and SE17, were positive for esterase (C4), whereas the same isolates but SE9 were also positive for esterase lipase (C8). Five out of six isolates belonging to this clade, namely SE8, SE10, SE17, SE25, and SE30, showed a strong leucine arylamidase activity, whereas all the six isolates assayed were positive for valine arylamidase. SE25 was also positive for cystine arylamidase; SE10, SE17, and SE25 for acid phosphatase; and SE8, SE9, SE10, SE17, and SE25 for naphthol-AS-BI-phosphohydrolase. Finally, the isolates SE8, SE25, and SE30 showed a positive reaction for beta-galactosidase, whereas the isolates SE17 and SE25 were positive for alpha-glucosidase, and only SE17 was positive for beta-glucosidase.
As for L. graminis, only the isolate SE23 was positive for alkaline phosphatase, whereas all the three isolates SE4, SE23, and SE26 were positive for leucine arylamidase and valine arylamidase.
Finally, only one out of the two L. mesenteroides subsp. dextranicum isolates (SE21) was positive for esterase (C 4), whereas SE27 was positive for alpha-galactosidase. Both these isolates were positive for beta-galactosidase, whereas only SE27 was positive for beta-glucosidase.

Volatile Profile
The detailed identification of the VOCS found in the cheeses is reported in Table 7.

Physico-Chemical and Morpho-Textural Traits of Cheeses
Regarding pH, data were comparable with those already reported by Fogeiro et al. [14] and Macedo et. al. [15] for the same cheese, with values comprised between 5.24 and 5.30.
The reduction of pH during cheese manufacturing is principally due to the metabolic activity of lactic acid bacteria producing organic acids through fermentation of lactose. Of note, acidification improves sensory profile and safety of cheeses, with a remarkable reduction in the load of pathogenic (e.g., Enterobacteriaceae and Staphylococcaceae) or spoilage (e.g., Pseudomonadaceae) microorganisms naturally occurring in the raw milk [36]. pH is strictly related with TTA; this latter parameter represents a better predictor than pH of how organic acids impact on flavor of foods. Unlike strong acids that are fully dissociated, organic acids are only partially ionized. While pH measures the concentration of free protons in a solution, total titratable acidity measures the sum of free protons and undissociated acids [37]. In cheeses, TTA is influenced by organic acids produced during fermentation as well as by milk proteins and acid phosphates [38]. In the present study, a high variation for TTA was seen among producers, although, to the author's knowledge, no previous studies reporting TTA values of Queijo Serra da Estrela PDO cheese are currently available for further comparison of data. In any case, the TTA values herein measured were slightly lower than those reported by Cardinali et al. [5,25,39] for the Portuguese thistle-curdled cheeses Queijo de Azeitão PDO, Queijo de Nisa PDO and Queijo da Beira Baixa PDO.
In the cheeses herein analyzed, as expected, lactic and acetic acids prevailed among organic acids. The release of lactic acid is principally due to the metabolism of homofermentative lactic acid bacteria species, whereas acetic acid is mostly produced by heterofermentative lactic acid bacteria species [40]. Together with the already mentioned protective effect against pathogenic and spoilage bacteria [41], lactic and acetic acid contribute to cheese aroma formation. In the Queijo Serra da Estrela PDO cheeses herein analyzed levels of lactic acid were notably higher than those of acetic acid, thus confirming the prevalence of homofermentative species. To the authors' knowledge, no data are available on the content of these two organic acids in Queijo Serra da Estrela PDO cheese for further comparison of the results. However, the levels herein recorded for these two organic acids were higher than those found in Queijo de Azeitão PDO [39], and comparable to those found in Queijo de Nisa PDO and Queijo da Beira Baixa PDO [5,25].
Regarding a w , this parameter is strongly affected by ripening; together with pH, it represents a key parameter for cheese preservation. The differences emerged among cheeses might reflect the slight variations in ripening time and conditions at the four dairy plants. However, again no previous data on a w of Queijo Serra da Estrela PDO cheese are available for further comparison of results.
The evaluation of specific volume of cheese allows the presence of any porous or microporous structure to be revealed. Although Queijo Serra da Estrela PDO is recognized as a cheese with no or low porosity, some differences were seen in the specific volume of cheeses herein analyzed, with samples from producer 1 being characterized by the highest specific volume, feasibly due to the longer ripening time (approximately 60 days) in respect with the other sampled cheeses.
For cheese color and texture, both parameters depend on multiple factors, including the origin of milk, the pasture location, and the grazing seasons [25]. All Queijo Serra da Estrela PDO cheeses herein analyzed were characterized by comparable color and texture, except for those sampled from producer 1, which where darker and harder than those sampled from the other three producers. Again, this finding might be ascribed to the longer maturation of cheeses from producer 1.

Bacterial Biota
Microbial counts highlighted the occurrence of viable populations of thermophilic and mesophilic cocci as well as mesophilic lactobacilli. The contribution of these microbial groups to cheese fermentation and ripening is well acknowledged. First, as already mentioned, these pro-technological bacteria are responsible for fermentation of lactose into lactic acid and other secondary metabolites. Moreover, lactic acid bacteria can improve the rheological, textural, microstructural, and sensory properties of cheese by producing longchain polysaccharides composed of repeated units of mono-sugars or their derivatives [42]. These microorganisms can also contribute to enhancing the safety of cheese by producing bacteriocins; these latter are peptides with bacteriostatic, bactericidal, and/or bacteriolytic effect [43]. Finally, the enzymatic activities exerted by lactic acid bacteria strongly affect the taste and flavour of cheese [9].
The counts of these microorganisms in the Queijo Serra da Estrela PDO cheeses herein analyzed were in accordance with those reported by Tavaria and Malcata [16] for the same cheese.
Although enterococci are considered as members of this heterogeneous bacterial group, their role in cheese is still controversial since strains within the genus Enteroccocus can carry transferable antibiotic resistances or virulence genes [44]. Moreover, some strains can cause bacteremia, endocarditis, and other infections in humans [44]. However, in past studies, enterococci isolated from cheese whey showed a high biotechnological potential as starter cultures [44]. As even reviewed by Foulquié Moreno et al. [45], in Mediterranean cheeses produced with raw ewes' or goats' milk, enterococci contribute to the typical taste and flavor of cheeses through proteolysis, lipolysis, and citrate breakdown, thus explaining the interest of the dairy industry for this bacterial group [46]. The viable counts of enterococci in Queijo Serra da Estrela PDO cheeses herein analyzed were in accordance with those reported by Tavaria and Malcata [16] in the same cheese, thus confirming the importance of this bacterial group in such a Mediterranean cheese.
Regarding the coagulase-negative cocci, as reviewed by Khusro and Aarti [47], these microorganisms can improve the color and flavor of cheese during ripening, through lipolysis and proteolysis [47]. To the author's knowledge, the scientific literature on staphylococci in Queijo Serra da Estrela PDO cheese only deals with the occurrence of coagulase-positive species; hence, data obtained in the present study represent an advancement in the field. However, a massive presence of coagulase-negative cocci has already been detected in other Portuguese cheeses clotted with vegetable rennet [5,25,39], thus suggesting a role of this bacterial group in these cheeses.
Metataxonomic analysis undoubtedly represents a powerful tool to highlight the relative abundance of bacterial and fungal taxa in cheeses. Surprisingly, only one study has applied a metataxonomic approach to the investigation of Queijo Serra da Estrela PDO cheese microbiota so far [21], hence, the results of the present investigation undoubtedly represent an advancement in the knowledge of the microbial species harbored by this renowned Portuguese cheese.
Among the most represented ASVs, L. lactis was detected in all the analyzed samples. L. lactis is a mesophilic spherical or ovoid-shaped bacterium that can grow at 10 • C but not at 45 • C; it is considered a key species in fermented dairy products since it produces lactic acid from lactose and contributes to the breakdown of milk proteins during fermentation [48,49]. The metabolic activity of L. lactis is also essential for flavor formation through the conversion of amino acids into volatile aroma compounds [50]. Finally, L. lactis contributes to refine the texture characteristics of cheese being able to produce exopolysaccharides, these latter directly affecting the rheological properties of cheeses [50]. The occurrence of this species in both Portuguese and Italian thistle-curdled cheeses is well documented [5,25,39,51,52]. Regarding Queijo Serra da Estrela PDO cheese, strains ascribed to L. lactis had previously been isolated [53,54]; this species was also detected by Rocha et al. [21] and Tavaria & Malcata [19] by sequencing of rRNA gene amplicons, thus confirming the crucial role of this microorganism in fermentation and ripening of Queijo Serra da Estrela PDO cheese.
As for leuconostocs, the presence of these bacteria in cheese is mainly related to contamination from the dairy environment, being considered as NSLAB [55]. In cheese, leuconostocs are responsible for the development of flavor compounds as diacetyl and acetoin. A synergistic relationship between Leuconostoc and L. lactis is usually established in cheese, since Leuconostoc metabolizes citrate and produces flavor compounds under acidic pH, this latter affected by the lactic acid production by L. lactis [55]. Some Leuconostoc strains are also able to produce heat-stable bacteriocins (e.g., mesentericin) active against Listeria spp. [55]. Moreover, L. mesenteroides is known to produce exopolysaccharides (e.g., dextran) with texturizing function (e.g., increase of viscosity and strengthening of casein network) [55].
L. sakei is one of the key species in meat fermentation [56]; hence, its high occurrence in the cheeses herein analyzed is quite uncommon and requires further investigation. However, closest relatives to L. sakei have already been detected by Aquilanti et al. [51] in a Caciotta cheese manufacture curdled with C. cardunculus L., as well as in other Portuguese cheeses produced with vegetable rennet, as Queijo de Nisa PDO, Queijo da Beira Baixa PDO, and Queijo de Azeitão PDO [5,25,39], with a possible role of this microorganism as a NSLAB species [57].
As for L. piscium, this Lactococcus species is usually detected among spoilage microorganisms in packaged meat, although recent insights into the microbiota of Portuguese thistle-curdled cheeses reported its presence among the autochthonous NSLAB [5,25,39]. Interestingly, no ASVs belonging to L. piscium had previously been detected in Queijo Serra da Estrela PDO by Rocha et al. [21] using a metataxonomic approach, thus confirming the need for further metataxonomic studies on this Portuguese cheese to better disclose its biodiversity.
Enterococcus was massively detected in Queijo Serra da Estrela PDO cheeses from producer 4 and, to a lesser extent, from the other producers. The occurrence of enterococci in Queijo Serra da Estrela PDO cheese had already been reported by Foulquié Moreno et al. [45] and Rocha et al. [21], the latter authors using a metataxonomic approach; members of the genus Enterococcus have also been found by several authors in thistle-curdled cheeses [5,25,39,51,52,58]. In the present study, Enterococcus spp. represented the most frequently isolated NSLAB whose enzymatic activities will be thoroughly discussed below.

Enzymatic Activity of Lactic Acid Bacteria Isolates
The pool of isolates collected from the cheeses herein analyzed reflected the occurrence of some NSLAB found via metataxonomic analysis (e.g., Leuconostoc mesenteroides, Enterococcus, Lacticaseibacillus). For most of these isolates, an unambiguous classification to the species level was achieved, whereas 6 isolates could be just assigned to the Lacticaseibacillus casei group. For this latter clade, which includes Lacticaseibacillus casei, Lacticaseibacillus paracasei, and Lacticaseibacillus rhamnosus (basonym Lactobacillus casei, Lactobacillus paracasei, and Lactobacillus rhamnosus, respectively) the analysis of the 16S rRNA gene does not allow unknown isolates to be identified to the species level [59,60].
The semi-quantitative assessment of the enzymatic activities of the isolates allowed a better comprehension of the bacterial interaction with the cheese matrix to be obtained.

Enterococcus spp.
As for E. faecalis SE12, which was positive for esterase (C 4), this enzymatic activity has already been observed by Cebrián et al. [61] in one E. faecalis strain isolated from Spanish raw ewe's milk cheese. Esterase is a hydrolytic enzyme that is involved in the enterococci life cycle (e.g., to provide nutrients, to assist in biofilm formation, and to start the proinflammatory responses) [62]. In cheese, esterases produced by E. faecalis promote the transformation of fats into fatty acids and glycerol [61].
The isolate E. faecalis SE7 was positive for acid phosphatase; this enzyme is a nonspecific hydrolase that liberates phosphate ions from organic esters at pH values ranging from ∼4.5 to 6.0 and is essential for the hydrolysis of phosphopeptides during cheese ripening [63,64]. Of note, the activity of this enzyme has already been found in E. faecalis strains isolated from Pecorino Abruzzese cheese [64].
The alpha-glucosidase activity was observed in only two E. faecalis isolates herein assayed. This hydrolase was among the most active enzymes in E. faecalis strains isolated from ewe's milk Roncal and Idiazabal cheeses (Spain) [65].
As for the isolates E. durans SE28 and SE29, the esterase lipase (C 8) activity was observed. In this regard, Tsanasidou et al. [66] have already reported the strong esteraselipase activity of E. durans strains isolated from Greek cheese. The presence of esterase lipase is a beneficial trait during cheese ripening for production of cheese flavors. Both the abovementioned isolates were also positive for leucine arylamidase and beta-glucosidase activity. Leucine arylamidase is an aminopeptidase that has already been detected in enterococci isolated from artisanal productions of Azorean Pico cheese [67], whereas beta-glucosidase is an enzyme belonging to the hydrolases class, whose presence has already been reported in enterococci [68]. This enzyme acts on the beta-glycosidic bonds of polysaccharides, hydrolyzing the terminal residues of beta-D-glucose.
Finally, most of the isolated enterococci showed a strong beta-galactosidase activity. This glycosidase, usually known as lactase, catalyzes the hydrolysis of the beta-glycosidic bond between glucose and galactose [69], thus suggesting the high adaptation of the assayed Enterococcus isolates to the dairy environment.

Levilactobacillus brevis
L. brevis represents one of the most common NSLAB species detected in Mediterranean and Middle Eastern cheeses obtained from raw ewes' and goats' milk [70].
In the present study, most isolates ascribed to L. brevis were positive for leucine arylamidase and valine arylamidase. The presence of these aminopeptidases has already been reported for this species by Son et al. [71] and Song et al. [72].
The acid phosphatase showed by most of the L. brevis isolates herein assayed has already been reported for the same species by Okoth et al. [73].
A great part of these L. brevis isolates was also positive for alpha-galactosidase. This enzyme is an exoglycosidase that hydrolyzes the terminal alpha-galactosyl portions from glycoproteins, glycolipids, galactomannans, and galactolipids [74]. Moreover, alphagalactosidase allows bacteria to use polysaccharides as a carbon source since it cleaves the alpha-1,6 bond between galactose and glucose [74,75].
As for lactase (beta-galactosidase), all the tested L. brevis isolates showed this enzymatic activity, although this heterofermentative species rarely ferments milk.
Regarding beta-glucosidase, this enzymatic activity has already been observed in L. brevis strains of vegetable origin [71,72]. Beta-glucosidase and beta-glucuronidase can act as carcinogens [76], thus suggesting the need for further investigations into L. brevis isolates that exhibited these enzymatic activities.
As for the alpha-glucosidase activity found in the isolates L. brevis SE11, SE20, and SE24, it is known that, in this species, this enzyme is mainly found in the intracellular form, and it principally degrades disaccharides and oligosaccharides [77].

Lacticaseibacillus casei Group
As for esterase activity showed by some isolates ascribed to the L. casei group, it is noteworthy that Fenster et al. [78] have already characterized intracellular esterase from L. casei. Hence, the same authors suggested that this enzyme might play a crucial role in cheese flavor development by modulating ester profiles in cheese.
The isolates ascribed to the L. casei group were also positive for the leucine arylamilase and valine arylamilase activity; both these enzymatic activities have already been observed by Colombo et al. [79] in L. casei isolates collected from the dairy environment. In the study herein carried out, the low number of isolates positive for cystine arylamilase was in accordance with previous investigations reporting a weak or even no activity for dairy isolates ascribed to this clade [79,80]. Some isolates ascribed to the L. casei group herein assayed were also positive for acid phosphatase; of note, the production of this latter enzyme in L. casei has already been reported by Colombo et al. [79]. Acid phosphatase of L. casei can be useful in metabolizing phosphates in the acidic environment occurring in cheese during maturation [81].
Finally, half of the L. casei group herein assayed showed beta-galactosidase. As observed by de Souza et al. [82], beta-galactosidase activity in L. casei is strain depended; this finding likely explains the absence of this activity in all the L. casei group isolates obtained from Queijo Serra da Estrela PDO cheeses.

Latilactobacillus graminis
L. graminis was the sole isolate that showed a positive reaction for alkaline phosphatase. This hydrolase, whose activity has widely been reported in lactic acid bacteria, removes phosphate groups from various molecules, such as nucleotides, proteins, and alkaloids [83].
All isolates of L. graminis also showed leucine arylamidase and valine arylamidase activity. To the author's knowledge, a paucity of data regarding these enzymatic activities in L. graminis is available in the scientific literature for further comparison of data. However, the presence of these aminopeptidases suggests a contribution of this NSLAB species to cheese flavor formation.

Leuconostoc mesenteroides
As for L. mesenteroides, only one isolate showed esterase activity; this result is in accordance with that reported by González et al. [84], who found weak or no esterolytic activity in leuconostocs used as starters to produce Genestoso cheese, a traditional dairy product from the North of Spain.
The alpha-galactosidase activity showed by the isolate L. mesenteroides SE27 was in accordance with the results obtained by Kamarinou [80] that observed the same enzymatic activity in autochthonous L. mesenteroides strains isolated from Greek traditional dairy products.
As for beta-galactosidase, the presence of this enzymatic activity in Leuconostoc depends on the strain or the substrate (glucose or lactose) [85]. Huang et al. [85] reported that the optimum pH for L. mesenteroides beta-galactosidase was comprised between 7.2 and 7.5; the same authors also observed an inhibitory effect of Ca 2+ on the activity of this enzyme, depending on the CaC1 2 concentration.

Mycobiota
In artisanal cheeses, yeasts constitute an important part of the microbiota since they grow well in acidified environments and do usually not compete with bacteria for nutrients [86]. As reviewed by Bintsis [86], in cheese, yeasts exert esterase or lipase activity as well as peptidase activity, thus contributing to flavor formation. Moreover, yeasts use lactose produced by lactic acid bacteria with a subsequent increase of pH in cheeses [86].
To the authors' knowledge, to date, yeast populations of Queijo Serra da Estrela PDO cheese have only partly been investigated; hence, the present study represents a further advancement in the knowledge of the mycobiota occurring in this dairy product.
Although with wide variations, likely due to the artisanal practice, the counts of yeasts observed in cheeses herein analyzed were in accordance with those reported by Freitas and Malcata [87] in the same cheese, attesting at about 3-4 log cfu/g.
Regarding molds, only cheeses from producer 2 were characterized by high viable counts, whereas the remaining cheeses showed loads < 1 log cfu/g. To the author's knowledge, no data on molds occurring in Queijo Serra da Estrela PDO cheese are available in the scientific literature for further comparison of results.
In the present study, the metataxonomic analysis of fungal populations allowed major and minor taxa to be detected.
Regarding D. hansenii, this species advantages of low a w , acidic environment, and high salt content, thus explaining its dominance over other yeasts [88]. D. hansenii has a high proteolytic activity (casein hydrolysis) and a moderate lipolytic activity, being able to hydrolyze palmitic acid and stearic acid esters and, to a lesser extent, oleic acid ester [88]. The occurrence of D. hansenii in PDO cheeses manufactured from ovine and caprine milks in the Iberian Peninsula, including Queijo Serra da Estrela PDO, has already been reported [5,21,25,39,87].
V. victoriae is a yeast species able to grow at low temperatures; moreover, it attracted the attention of the food industry for its ability to secrete killer toxins and hydrolytic enzymes (protease, chitinase, and glucanase) [89]. Furthermore, V. victoriae can grow using lactose, thus explaining its presence in the analyzed cheeses [89]. To the authors' knowledge, V. victoriae has never been detected in Queijo Serra da Estrela PDO cheese before, although it has already been identified among the minority yeasts occurring in Queijo da Beira Baixa PDO [25].
As for Naganishia, this extremophilic fungus, previously ascribed to the Cryptococcus albidus clade, was found to dominate the eumycete populations of soil-like material (tephra) on volcanoes [90]. To the authors' knowledge, N. albidosimilis has never been detected in cheese, before, although Cryptococcus species have already been identified in milk as well as surface mold-ripened and blue cheeses [86].
G. geotrichum is a mold that can be found in soil, plants, and fruits [91]. This microorganism is also part of the autochthonous mycobiota inhabiting dairy products, since it has been detected in raw milk and on the surface of semi-hard, mold-ripened, and semi-soft cheeses [92]. Interestingly, Grygier et al. [92] have recently observed a high polyunsaturated fatty acids (PUFA) production carried out by strains of G. geotrichum, thus suggesting the use of this mold to naturally enrich cheeses in PUFA. To the authors' knowledge, the presence of G. geotrichum has never been reported in Queijo Serra da Estrela PDO cheese before; however, its occurrence in Queijo de Azeitão PDO cheese has been reported by Cardinali et al. [39].
In fermented dairy products, Starmerella has rarely been detected, although this microorganism has already been identified among the minority species occurring in Queijo de Azeitão PDO and Queijo da Beira Baixa PDO cheeses [25,39]. Of note, the extracellular biosurfactants (sophorolipids) produced by Starmerella spp. showed to be active against pathogenic foodborne fungi [93], thus suggesting the need for further investigation on the contribution of this yeast for improvement of cheese safety.
C. boidinii has rarely been detected in cheese; notwithstanding, the presence of this yeast has already been reported by Papademas and Robinson [94] in mature commercial Halloumi cheeses. The enzymatic profile of C. boidinii isolates collected from Halloumi cheeses showed alkaline phosphatase, esterase (C4), esterase lipase (C8), leucine aminopeptidase, acid phosphatase, and phosphoamidase activity [94], thus suggesting a key role of this yeast in cheese flavor formation. C. boidinii has also been detected among the minority yeasts in Queijo de Azeitão PDO cheese [39].
Finally, K. zeylanoides (formerly known to as Candida zeylanoides) has already been detected in Queijo Serra da Estrela PDO by Rocha et al. [21]. As suggested by Rocha et al. [21], the occurrence of this yeast in cheese might be ascribed to latent infection (mastitis) of lactating bovine females, thus suggesting the need for a constant monitoring of the health conditions of these dairy animals.

Volatile Organic Compounds
Regarding the short-chain fatty acids, butanoic, hexanoic, 2-methyl and 3-methyl butanoic are usually recognized as key molecules in the characterization of the aroma profile of different cheese varieties, due to their low perception threshold [95,96].
Moreover, aldehydes, esters, and methyl ketones greatly contribute to the production of other aroma compounds [98]. In the Queijo Serra da Estrela PDO cheeses herein analyzed, 2-butanone was the most abundant methyl ketone detected, thus suggesting a role of this compound to the enrichment of cheese aroma with a butterscotch odor note. Methyl ketones were among the most abundant compounds identified in Queijo da Beira Baixa PDO cheese and Cheddar [25,81].
In the present study, phenylacetaldehyde and benzaldehyde might likely be originated by the microbial metabolism of phenylamine and tryptophan [99].
The occurrence of isovaleric acid (which is a product of leucine catabolism) in the samples herein analyzed was in accordance with what reported by Tavaria et al. [100] about the abundance of volatile fatty acids in Queijo Serra da Estrela PDO cheese.
Ester fraction was mainly represented by ethyl esters of fatty acids as ethyl esters of octanoic and decanoic acids, which confer floral and fruity notes to the analyzed cheeses.
The high amounts of alcohols detected in the Queijo Serra da Estrela PDO cheeses herein analyzed were in accordance with those already found in other Portuguese cheeses [5,25]. It is well known that ethanol, which is responsible for the formation of esters, is produced by heterofermentative lactic acid bacteria and its contribution to the aroma is generally limited.

Conclusions
Raw milk PDO cheeses represent a source of still partly undisclosed microbial diversity, where well known microbial species driving fermentation co-exist with other co-occurring microorganisms that are likely to have a strong impact in the sensory traits of cheeses. In the present study, L. lactis and L. piscium represented the lactic acid bacteria detected at the highest relative abundance in Queijo Serra da Estrela PDO cheese. At this regard, the leading role of L. lactis in cheese fermentation is widely acknowledged, whereas the influence of L. piscium in cheese maturation has still to be clarified. Moreover, E. durans, E. faecalis, E. faecium, E. lactis, L. brevis, L. casei group, L. graminis, and Leuconostoc mesenteroides subsp. dextranicum were isolated from the cheeses. The enzymatic characterization of the pool of NSLAB allowed the comprehension of the role that these microorganisms might play in the definition of sensory and volatile traits of Queijo Serra da Estrela PDO cheese. The results overall collected suggested the need for further investigations on the pro-technological features (e.g., production of bacteriocins and exopolysaccharides) of the isolates for their possible exploitation as adjunct cultures in cheese manufacturing. In the cheeses herein analyzed, fungal populations were dominated by D. hansenii and K. zeylanoides, however species rarely found in cheese (e.g., C. boidinii, V. victoriae, and Starmerella) were surprisingly detected, thus representing an advancement in the knowledge of the mycobiota naturally occurring in Queijo Serra da Estrela PDO cheese. The volatile compounds characterizing the analyzed cheese were carboxylic acids and esters, followed by carbonyl compounds and alcohols.
Differences among most of the morpho-textural traits were observed in cheeses collected from producer 1 in respect with those collected from the other producers. The observed differences were likely affected by the prolonged ripening time of cheeses from producer 1.
The data overall collected allowed an up-to-date and a comprehensive overview of morpho-textural traits, microbiota, and volatilome of Queijo Serra da Estrela PDO cheese, thus contributing to improve the knowledge of this renowned Portuguese cheese and providing objective quality parameters for product evaluation and valorization.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/foods12010169/s1, Table S1: Incidence (%) of the bacterial taxonomic groups in Queijo Serra da Estrela PDO cheese samples according to producers. Table S2: Incidence (%) of the fungal taxonomic groups in Queijo Serra da Estrela PDO cheese samples according to producers.