3.1. Spontaneous Fermented Sausages Characterization
The three spontaneous fermented sausages, produced without the use of starter cultures or the addition of sugars to meat batter, were characterized and used as a source of isolation of LAB.
Table 3 shows the microbiological parameters, pH and a
w of the sausages at the end of ripening.
The products showed high pH values, because of low initial pH decrease due to slow and weak fermentation for the absence of added sugars and starter cultures. Differences among samples can be attributed to the absence, in BAS, of mould growth on the casing, and therefore not contributing to a rise in pH during ripening. The aw was particularly low in the BAS sample (0.831), while it was higher (0.908) in BRE, characterized by a larger diameter.
The microbiological analyses showed a high concentration of LAB in BRE and ROM, with a concentration of about 9 and 8 log CFU/g, respectively. On the contrary, this microbial population was present at a very low concentration (2.54 log CFU/g) in BAS, probably due to the low aw and storage conditions. In BAS, staphylococci and enterococci also showed low concentrations, being enterococci under the detection limit. In BRE and ROM, staphylococci were present at similar levels (about 7 log CFU/g) while enterococci were found in high concentration exclusively in the BRE sample (4.23 log CFU/g). This high number of enterococci could depend on several factors, such as the microbial quality of raw materials, the environment of production or the ripening conditions. The differences in the viable population in BAS can have many reasons. The BAS sample remained one month after ripening under fat, undergoing much longer osmotic stress than other samples that reached final aw values progressively over time during ripening. These stresses can have great influences on microbial populations and their cultivability. In addition, the absence of oxygen in BAS can affect the stress response mechanisms of several bacteria, in particular staphylococci.
3.2. Clustering of LAB Isolates and 16S rRNA Sequencing
Representative colonies were picked up from MRS plates deriving from ROM (97 colonies), BRE (50) and BAS (62). After purification, DNA was extracted from each isolate. RAPD fingerprinting and clustering analysis allowed the obtainment of a total of 41 biotypes, as 12 clusters and 10 single strains from ROM, 6 clusters and 6 single strains from BRE and 5 clusters and 5 single strains from BAS (results are shown in the
Supplementary Material Figure S1). From this clustering, different scenarios in the spontaneously fermented salami could be observed: (i) dominance of a specific strain in BAS, represented by the cluster containing more than 80% of all isolates; (ii) about 50% single isolate in BRE, with no clear dominance of a particular isolate in the remaining 50%; and (iii) co-dominance of two clusters that comprise more than 50% of the isolates in ROM.
A total of 41 biotypes, comprising one representative strain for each cluster and the single strains, showing good growth after sub-culturing, were planned for 16S rRNA sequencing and subsequent analyses (as schematized in
Figure 1). Three single isolates present in the clustering profiles (one from BRE and two form ROM) did not show growth after sub-culturing and were discarded. Therefore, the molecular identification was performed on 11 strains from BRE, 20 strains from ROM and 10 strains from BAS (
Table 4). Results showed that
Latilactobacillus sakei was the dominant species (
Table 4) in 80.5% of all strains. This confirmed its strong adaptation to the meat matrix and, in particular, to fermented sausages [
22,
23]. However, within
Lat. sakei, a great strain biodiversity could be appreciated in all sausages.
Latilactobacillus curvatus was the second most abundant species among the biotypes, representing 9.1% in BRE, 15.0% in ROM and 20.0% in BAS, whereas
Leuconostoc mesenteroides was detected only in BAS, representing 20.0% of all isolated strains.
Several authors reported that LAB species diversity in fermented sausages is limited. In fact,
Lat. sakei predominates during the ripening process, given its competitiveness and assertiveness in the meat matrix, and it was evidenced as the dominant species in some European spontaneously fermented sausages, together with
Lat. curvatus [
24,
25]. This predominance is mainly due to its physiological and technological characteristics and to its peculiar metabolic pathways for the meat ecological niche, including the arginine deiminase pathway and the utilization of nucleosides [
26,
27,
28].
Lat. sakei can show a high biodiversity degree and, interestingly, strain-specific differences in species performance in meat environments were demonstrated [
25].
The present work highlighted the presence of different biotypes in the same sample, but it is known that the co-dominance of these strains can confer competitive exclusion of other strains in the sausage environment and can lead to the peculiar characteristic of the products [
29].
The scheme in
Figure 1 was used both for the screening procedure and for safety and technological characterization of the studied biotypes.
3.3. Determination of Antibiotic Susceptibility (MIC)
The minimal inhibition concentration of eight antibiotics was examined in the 41 LAB biotypes previously identified, following the indications given by EFSA [
18].
Table 5 shows the results obtained.
No strains were resistant to gentamycin, streptomycin, erythromycin and clindamycin, while the resistance against the other antibiotics was 14.6% for tetracycline and chloramphenicol, 7.3% for ampicillin and 2.4% for kanamycin.
Resistances to tetracycline and erythromycin are the most widely observed and studied among lactobacilli from fermented meats. Tetracycline is a commonly used antibiotic in pig farming, explaining the abundance of
tet resistance genes in the pig microbiome and, consequently, in the raw meat used for fermented sausages [
30]. In many cases, this specific resistance involved the lactic acid bacteria responsible for meat fermentation; however, the prevalence of resistant strains varied highly in products from different geographical areas [
31,
32]. In this study, all strains were susceptible to erythromycin and 6 out of 41 (14.6%) were tetracycline resistant. Three of these latter strains were isolated from ROM while the other three were from BAS. According to Fontana et al. [
33], 26.4% of
Lat. sakei and
Lactiplantibacillus plantarum strains from fermented sausages differing in meat and geographical origin were resistant to tetracycline while the percentage was reduced at 10.7% for erythromycin. Although, according to Abriouel et al. [
34], many lactobacilli are intrinsically resistant to kanamycin, streptomycin and gentamicin, in this study, no strains were resistant to streptomycin and gentamicin, and only one strain, isolated from BAS and belonging to
Leuc. mesenteroides species, was resistant to kanamycin. Danielsen and Wind [
35] found higher resistance for
Lat. sakei/curvatus to kanamycin (66.7%), gentamicin (50%) and streptomycin (100%). Resistant strains were also not found for clindamycin, confirming the data reported by Danielsen and Wind [
35] and Fontana et al. [
27]. Three ampicillin-resistant strains were found, two from ROM and one from BAS. A higher percentage of resistant strains to this antibiotic were found by Aymerich et al. [
36] and by Federici et al. [
37].
A higher resistance frequency was observed for chloramphenicol. Five strains out of twenty isolated from ROM were resistant, while only one strain from BRE and none from BAS showed a MIC higher than the cut-off value. Aymerich et al. [
36] described a low percentage (1.2%) of
Lat. sakei resistant to this antibiotic, while Danielsen and Wind [
35] found all the strains of
Lat. sakei/
curvatus susceptible.
Interestingly, three strains from ROM showed multiple resistances:
Lat. sakei C12G (tetracycline and chloramphenicol), C22G (tetracycline and ampicillin) and E3G (tetracycline, chloramphenicol and ampicillin). It was demonstrated that genes responsible for tetracycline resistance (
tetM and
tetS, which encode ribosomal protection proteins) can mediate the resistance to other antimicrobials such as erythromycin, clindamycin and chloramphenicol [
34].
3.5. Antimicrobial Activity Assay
Twenty-nine strains lacking antimicrobial resistance traits and amino-biogenic potential were included in further screening steps (
Figure 1). The LAB culture antagonistic activity against target foodborne pathogens showed that all strains were able to inhibit at least one pathogen to a different extent. Some strains, namely
Lat. sakei E13G, E15G and C10B, were the most performant antagonistic strains against all four screened pathogens (
Table 6). Conversely, non-neutralized and neutralized LAB supernatants did not exert any antimicrobial activity against the same pathogens.
However, two
Lat. sakei strains, namely C21B and E23B, showed an anti-LAB activity against
Lat. curvatus DSM 20019
T and
Lat. sakei subsp.
sakei LMG 13558
T in the well diffusion assay (data not shown). Although strain-specific monitoring systems are rarely investigated, a recent study demonstrated how the bacteriogenic strain
Lat. curvatus TMW 1.624 could exclude some of the
Lat. sakei strains during sausage fermentation competition, while coexisting in the same environment with bacteriocin resistant
Lat. sakei 1.417 [
43]. The innate resistance against bacteriocin produced by closely related competitors was hypothesised as useful competitive fitness in
Lat. sakei 23K [
26]. Further investigation of bacteriocins in the
Lat. sakei C21B and
Lat. sakei E23B genome would elucidate the nature of this anti-LAB activity.
3.6. Growth and Acidification Kinetics
The growth and acidification behaviour of 29 safe LAB strains not presenting antibiotic resistance or amino biogenic potential and studied for their antimicrobial activity, were monitored through the turbidity increase (measured on the McFarland scale) and pH variation under different salt concentrations. The growth data for each strain were modelled with the Gompertz equation [
21].
Table S1 and
Figure 2 show the biotype growth performances at different NaCl concentrations (0, 2, 4, 6 and 8%). The strains were grouped according to their origin and, for each condition, the estimated parameters of the Gompertz equation (
A,
λ and
μmax) were reported as mean (and standard deviation).
The maximum cell density of the McFarland scale (
A) reached after the incubation was lower, as expected, in the presence of higher salt concentrations. No significant differences were observed in relation to the isolation source in cells grown at 0 and 2% of NaCl. In general, the strains isolated from BRE and BAS showed higher
A estimates if compared with the strains isolated from ROM. It is noteworthy that comparable standard deviations characterize these data. However, this is not true for the means concerning the
λ values. In fact, higher NaCl concentration not only determined longer
λ time but also greater standard deviation of the means, indicating a wide variability in the responses of strains subjected to more severe stress conditions. Analogously to what was observed for
A, the values of λ did not present significant differences in the samples added with 0 and 2% salt. Shorter λ characterized the strains deriving from BRE and BAS at a higher concentration than those isolated from ROM. The estimates of
λ followed the same behaviour of
A, showing higher values for BRE and BAS strains, and were characterized by homogeneous standard deviations. In fermented sausage production, the rate at which starter cultures decrease the pH is crucial for safety and technological aspects [
44].
In the same samples analysed for the determination of the McFarland values, pH was also periodically monitored. In fact, the LAB metabolism increased the acidity, and the data were expressed as differences of pH with respect to the initial value. The curve obtained showed a behaviour similar to those obtained measuring the change of the turbidity on the McFarland scale. In this case, the results for the three parameters estimated with the Gompertz equation and obtained in relation to the isolation source and NaCl concentration are presented as a box and whiskers plot (
Figure 2a–c).
Regarding the maximum pH decrease (parameter A of the model), data showed a great variability within the strains in relation to the source of isolation. In particular, biotypes isolated from ROM were generally characterized by lower pH decreases if compared with BRE and BAS, independently of the NaCl concentration. Even if the extent of the acid production was affected by the NaCl concentration, it is interesting to observe that the variability of the results within each group was of the same order.
Regarding the rate of acidification (the parameter
µmax of the Gompertz equation) (
Figure 2b), the strains isolated from BAS presented median values generally higher with respect to the other isolation sources, while the ROM isolates showed lower performances at higher salt concentrations.
Another important characteristic of the starter culture is the time needed to modify the initial pH of the medium (the λ value). Also, in this case, the strains isolated from BAS were the more competitive, especially at increasing NaCl concentrations. As observed for the λ value for the McFarland scale, the variability within each group markedly increased when the higher content of salts was considered. Highly performant strains could be observed at 6% of NaCl with a short lag phase of 7.5 h. A high heterogeneity among strains can be observed when higher osmotic stress was applied, with a lag phase ranging from 16.92 to 53.55 h. This again confirms the high physiological heterogeneity and potential technological traits of the indigenous meat-borne strains.
The growth and acidification performances of starter cultures are important parameters for ensuring the safety and quality of fermented meats. The rapidity with which the cells can multiply in defined conditions, colonizing the environment and producing organic acids, results in a more efficient inhibition of pathogens or undesirable microbial outgrowth [
45]. In addition, in the case of fermented meats, a fast acidification performance represents a crucial technological point allowing the correct initial dehydration process of sausages [
46].
The different behaviour shown in the metabolic activity in relation to different NaCl concentrations is often linked to the isolation source. It is noteworthy that the strains isolated from BAS, which were the most performant also at high NaCl concentrations, were those isolated from the harsher environment, whose final a
w was 0.831. The great diversity within the species
Lat. sakei is well known and justifies the differences in adaptation and competition in natural systems, such as fermented sausages [
47]. In addition, there can be at least two relationships during meat colonization by members of this species, according to Janßen et al. (2018) [
25]. In the first case, a strain outcompetes the other members of microbiota (assertiveness), based on peculiar metabolic characteristics (lag phase, growth rate, antimicrobials, stress responses). Alternatively, a co-dominance or a cooperation between different strains is established, with the consequent creation of a colonization resistance. According to the data reported here, the second strategy characterized the ROM sausages, while BAS and BRE presented a selection of a reduced number of strains at the end of production.