3.2. Microbiological Characteristics of the Cheeses
The populations of
L. lactis and
S. thermophilus decreased gradually over the ripening time, but no differences (
p > 0.05) were found between the CS and the LS cheeses in relation to their counts throughout the ripening period (
Table 1). The concentration of lactobacilli decreased significantly between day 1 and day 60 of ripening, from 8.72 to 6.54 log cfu/g while it remained stable from day 60 to the end of ripening, and no significant differences were found throughout the rest of ripening.
In relation to the metataxonomic analysis of three CS cheeses at days 1, 120 and 240 and three LS cheeses at the same sampling times, the sequencing of the six cheese samples yielded 1,502,505 high-quality reads (median = 237,910.5 reads/sample, ranging from 200,787.0 to 284,112.8 reads/sample). The two diversity indices evaluated in this work (Shannon and Simpson indices) increased over time in both types of cheeses. For CS cheeses, the Shannon indices were 0.53, 0.66 and 0.99 at days 1, 120 and 240, respectively, while the values at the same sampling times where 0.60, 0.77 and 0.97 for LS cheeses. In relation to the Simpson index, the values were 0.33, 0.32 and 0.54 at days 1, 120 and 240, respectively, among the CS samples and 0.35, 0.36 and 0.49, respectively, among the LS ones. No differences were found when CS and LS samples were compared within each sampling time, but there was a significant increase for both diversity indices and both types of cheeses when samples of days 1 and 240 were compared (
p < 0.01). It has been previously found that diversity in the microbiota of cheeses is a major driver in the organoleptic characteristics found in raw-milk cheeses [
22].
The genera detected in the metataxonomic analysis of the samples are shown in
Table 2. Among them, the genus Lactococcus was the most abundant in both cheese types although its concentration decreased from day 1 to day 240. The genus Streptococcus was the second genus in order of abundance in both cheese types at days 1 and 120, and its percentage also decreased throughout the ripening period. The genus
Ligilactobacillus was present from the beginning in the LS cheeses and its relative abundance increased during ripening, probably because of the decrease in the relative abundances of genera
Lactococcus and
Streptococcus. In contrast,
Ligilactobacillus sequences were sparsely detected in CS cheeses. Interistingly,
Lactiplantibacillus sequences, probably with an environmental origin, were not detected on day 1 in any of the cheese types but its abundance increased over time, and in fact, it was the second most abundant genus in both cheese types at the end of the ripening period.
Lactiplantibacillus plantarum and
Lactiplantibacillus pentosus are common and often dominant members of the non-starter LAB communities in cheeses [
12,
23,
24]. These microbes are present at low or even not detectable concentrations in the curd, but their levels increase notably during ripening [
24], which is in agreement with the results of this study. Despite the great abundance of
Lactiplantibacillus sequences in both cheeses, these microorganisms were not detected in the microbiological analyses, probably because they do not grow in the medium used and/or the conditions tested.
3.5. Analysis of Volatile Compounds
A total of 82 volatile compounds were identified by SPME/GC-MS in the cheeses, including 5 hydrocarbons, 5 sulphur compounds, 4 aldehydes, 9 ketones, 11 esters, 20 alcohols, 6 benzene compounds and 12 carboxylic acids. The use of
L. salivarius SP36 as an adjunct culture significantly increased (
p < 0.01) the formation of 25 of these volatile compounds (
Table 3,
Table 4 and
Table 5).
L. salivarius favoured the formation of 6 ethyl-, 2 propyl-, 1 butyl- and 1 branched esters and their relative abundances increased significantly (
p < 0.001) during LS cheese ripening (
Table 3). The biosynthesis of aroma-active esters in dairy systems proceeds through two enzymatic mechanisms: esterification, a reaction in which esters are formed from alcohols and carboxylic acids, and alcoholysis, a transferase reaction in which fatty acyl groups from acylglycerols and acyl-CoA derivatives are directly transferred to alcohols. Evidence has been provided that esterases of lactic acid bacteria catalyse not only the hydrolysis of milk fat glycerides to release free fatty acids but also the synthesis of esters from glycerides and alcohols via alcoholysis in cheese [
28]. The observed differences in ester concentration among cheeses could be related to the different lactic acid bacteria compositions; it has been described that
L. salivarius strains isolated from human milk displayed esterase activities on most assayed substrates when tested for dairy technological properties [
29]. Esters, which usually increase during cheese ripening, have a very low perception threshold, playing an important role in the aroma profiles of different cheese varieties [
30,
31,
32,
33,
34]. Most esters encountered in cheese are described as having sweet, fruity and floral notes [
30,
34].
Cheeses made with the
L. salivarius strain showed higher concentrations of 3-methyl butanal than control (CS) cheeses (
Table 4). This branched-chain aldehyde originates from leucine by transamination, leading to an intermediate imide that can be decarboxylated, or by Strecker degradation, and has been identified as a potent odourant in several cheese varieties [
30]. A high aminopeptidase activity has been described in
L. salivarius strains, including Leu-pNA as a substrate [
29]. 3-Methyl butanal presents a green malty odour, but at low concentration, the odour becomes fruity and rather pleasant.
The levels of 1-propanol, 1-hexanol, 1-octanol, 1,2-propanediol and 2-furanmethanol were significantly (
p < 0.01) higher in cheeses made with
L. salivarius SP36 throughout ripening (
Table 4). However, the levels of ethanol and 1-butanol were only higher (
p < 0.01) at the end of ripening, while the levels of 2-pentanol were higher (
p < 0.01) at 60 and 120 d in cheeses made with
L. salivarius. The relative abundances of all these alcohols increased (
p < 0.01) with cheese age. Many metabolic pathways are involved in the biosynthesis of the alcohols commonly found in cheese, such as lactose metabolism, methyl ketone reduction, amino acid metabolism, as well as degradation of linoleic and linolenic acids [
30]. In general, primary alcohols have a high perception threshold, so scarcely contribute to the aroma of cheese. However, they are the limiting factor in ester formation in hard cheeses [
34]. With respect to 1,2-propanediol, Fröhlich-Wyder et al. [
35] described the production of this compound when
Lactobacillus buchneri and
Lactobacillus parabuchneri were used as adjunct cultures in the manufacture of model cheese. On the other hand, 2-furanmethanol has been found to contribute to the nutty and roasted aroma of Parmigiano-Reggiano cheese [
36].
The use of
L. salivarius SP36 as an adjunct culture in cheese manufacturing also promoted the formation of six carboxylic acids at the end of ripening (
p < 0.01) (
Table 5). As with the rest of volatile compounds mentioned above, the concentrations of these acids increased (
p < 0.001) with cheese age. During cheese ripening, most of the acids having between two and six carbon atoms originate from the degradation of lactose and amino acids, but they can also be derived from ketones, esters and aldehydes by oxidation [
30]. Short-chain carboxylic acids have low perception thresholds and are major odourants of different cheese varieties. Propanoic acid has a typical vinegar odour, and pentanoic and hexanoic acids contribute significantly to the aroma of aged cheeses [
31,
32,
33]. On the other hand, branched-chain carboxylic acids such as 2-methyl propanoic acid and 2- and 3-methyl butanoic acids are characteristic odour impact compounds of goat and sheep cheeses. They derived from valine, isoleucine and leucine, respectively [
30].
To our knowledge, this is the first time that a
L. salivarius strain has been used as an adjunct culture in the manufacture of sheep cheeses. In a previous work, the potential of two human milk
L. salivarius strains in fresh model cheeses to develop a probiotic cheese was evaluated [
37]. However, the effect of these strains on the volatile profile of cheeses was limited. More recently, the use of
L. salivarius AR809 as an adjunct in the manufacture of Monascus-ripened cheese (a mould-ripened cheese) significantly promoted the formation of alcohols, acids and ketones but reduced the formation of certain ethyl esters [
38].
3.6. Sensory Evaluation
The evaluation of the sensorial properties of the cheeses from 60 days of ripening was initially performed following the methodology of the ISO Standard Nº 13299:2016 and the PDO Manchego Cheese Sensory Guide (Procedure PAS-02). This evaluation concluded that, in both types of cheeses, the external descriptors (shape, physical integrity, colour of the rind, uniformity of the engraving of the pleita, uniformity of the engraving of the flower, hardness of the rind, degree of rind distinction, uniformity of the colour of the cheese paste, colour of the cheese paste, distribution and size of the eyes, and integrity of the paste) as well as the internal descriptors (odour, texture and flavour) were within the ranges typical of Manchego cheese. To the appreciation of the trained panelists, both cheeses had similar sensorial properties overall (
Figure S1), but LS cheeses had more complex flavours than CS ones. This is a logical finding since CS cheeses were made with a starter culture that is well tested by dairy companies, including those producing Manchego cheeses, and selected to provide cheese uniformity, which is an essential issue for consumers’ reliability. Therefore, it is not strange that the addition of a
L. salivarius strain with the ability to produce sensorial-related compounds led to the appreciation of “wilder” flavours; spicier touches; and greater firmness, microstructure and friability (
Figure S1).
Subsequently, the cheeses were also evaluated and compared using a descriptive test based on the guidelines of Bérodier et al. [
21]. The cheeses were evaluated for taste, odour and aroma quality and intensity, and for texture and colour quality, throughout ripening (
Table 6). The taste intensity and aroma intensity, and texture quality (
p < 0.01) of the cheeses increased significantly with cheese age (
p < 0.001), whereas taste quality, odour quality, aroma quality and colour quality, and odour intensity remained stable. Odour intensity and quality scores were higher in cheeses made with the
L. salivarius strain, but significant differences were not detected (
p > 0.01). However, the panelists gave significantly higher aroma intensity scores to cheeses made with
L. salivarius throughout ripening (
p < 0.01) (
Table 6). Aroma quality scores were also higher for cheeses made with
L. salivarius and were significantly higher after 120 and 240 d of ripening (
p < 0.01). The higher aroma intensity and aroma quality scores obtained for cheeses made with the
L. salivarius strain are in accordance with the higher levels of volatile compounds detected in these cheeses (
Table 3,
Table 4 and
Table 5).
In addition, cheeses made with
L. salivarius SP36 obtained significantly higher “sheepy”, “pungent”, “fruity” and “nutty” odour and aroma scores than control cheeses (
p < 0.001). The higher “pungent” odour and aroma scores obtained by cheeses made with
L. salivarius SP36 could be related to the higher levels of carboxylic acids found in these cheeses since many of them have been associated with “pungent” odours notes [
30]. The increase in “fruity” and “nutty” aroma scores in cheeses including the
L. salivarius strain could be associated with the higher levels of esters and furanmethanol found in these cheeses, respectively. Fruity and nutty flavour notes have been described for these compounds, respectively [
30,
36]. In a previous study, the use of two human milk
L. salivarius strains in fresh cheeses did not have a relevant impact in the sensory quality and acceptance of the cheese, which was in agreement with their limited effect on their volatile profiles [
37]. In contrast, the use of
L. salivarius AR809 as an adjunct in the manufacture of Monascus-ripened cheese had a positive effect on sensory properties [
38]. This strain significantly promoted the formation of several volatile compounds in cheese.
Taste intensity and quality scores were also higher for cheeses made with
L. salivarius and were significantly (
p < 0.01) higher for 60-, 120- and 180-day-old cheeses and 120-day-old cheeses, respectively (
Table 6). Although some taste descriptors, such as “sour”, “bitter”, “sweet” or “salty”, were not influenced by the use of
L. salivarius SP36 in cheese manufacture, “umami” descriptor scores were significantly higher in cheeses made with the
L. salivarius strain (4.37 vs. 3.82,
p < 0.01). There is a large number of factors that may exert an influence on the sensorial properties of cheeses, but among them, their microbial composition has a paramount relevance [
39]. Overall, changes in the parameters cited above may be closely related to the impact of the strain in cheese maturation, most probably related to the proteolytic and lipolytic activities of the strain and its ability to participate in the biosynthesis of volatile compounds. The analysis of the genome of
L. salivarius SP3 has already revealed the presence of a wide variety of genes potentially involved in such properties [
15]. They are relevant traits since the flavour-forming potential is one of the LAB traits with the highest interest to the dairy industry [
40]. Work is in progress to further elucidate the metabolic potential of such a strain for dairy applications.