Isolation and Partial Characterization of Lactic Acid Bacteria from Natural Whey Starter Culture

Abstract

Natural whey starter (NWS) cultures are microbial consortia characterized by high microbial diversity in terms of genus and species, as well as strains, accounting for the variety of different characteristics and quality of the artisanal fermented food. By means of a combined approach, using plate counts, bacterial isolation, molecular identification, and genotyping, we analyzed 41 colonies isolated from NWS of cow milk used in the production of caciocavallo, a typical pasta filata Italian cheese. Results revealed that 27 of them were lactic acid bacteria (LAB), including Lactococcus lactis as the dominant species, followed by Streptococcus thermophilus, Enterococcus faecium, Limosilactobacillus fermentum, Lactobacillus helveticus, and Lacticaseibacillus rhamnosus. The remaining isolates were taxonomically identified as non-LAB, probably due to environmental contamination. These results were mostly confirmed by metagenomic analysis, with the exception of only three species. Finally, small-scale fermentation experiments were performed in both standard media and skimmed milk to further characterize the newly isolated LAB strains. Overall, our results show that, except for four of the Lactococcus isolates and one Streptococcus, which show multi-drug resistance, the isolated strains under study exhibit levels of acidifying, metabolic properties, and safety parameters, suggesting their potential as starter cultures in cheese production.

1. Introduction

Lactic acid bacteria (LAB) are the main bacteria used in the production of fermented food, including dairy, fish, meat, and vegetables. Based on their functional roles in fermentation, LAB can be divided into two groups: starter LAB (SLAB), and non-starter LAB (NSLAB). Starter cultures are selected microorganisms that beneficially influence fermentation or maturation in food processing. They comprise living microorganisms intentionally used to trigger fermentation, thereby altering the substrate’s chemical composition and sensory characteristics and ensuring greater product consistency. In contrast, NSLAB typically arise as part of the spontaneous microbiota that are naturally present in the production and processing environment [1]. In the food ecosystem, microorganisms interact directly at the nutrient and metabolomic level through physical contact, quorum sensing (QS), symbiosis, inhibition, and others. For this reason, the success and outcome of fermentation rely mostly on the choice of starter cultures [2]. Although lactic acid bacteria have been employed in the natural fermentation of foods for millennia, extensive research continues to focus on isolating and cultivating pure LAB strains for the development of defined starter cultures. These tailored cultures are designed to achieve specific fermentation characteristics—such as acidity, flavor, and aroma—across a variety of raw materials [1,3,4,5]. In addition to their importance for food processing and preservation, LAB species have also received great attention in recent years for their probiotic properties, health effects, and production of functional foods [6].

Probiotics have been investigated for treating lactose intolerance, food allergies, high serum cholesterol levels, infectious diarrhea in infants, and for strengthening the mucosal barrier. More recently, they have attracted significant attention for their antimicrobial, immunomodulatory, anticancer, and antihypertensive effects [7]. Because the use of lactic acid bacteria as starter cultures or probiotics depends on strain-specific traits, there is strong motivation to isolate and characterize new LAB strains from natural sources. Considering the extensive microbial diversity associated with traditional fermentations, numerous studies emphasize the significance of artisanal products being rich sources of LAB strains, exhibiting distinctive technological properties and potential probiotic functionalities [1,3,6,8]. Many studies in recent years have described the microbial characterization of the natural whey starter (NWS) culture, which is commonly used in Italy and in other regions of Europe to produce different varieties of cheese, such as Comté cheese [9], different pasta filata cheeses [10,11,12], Grana Padano, Parmigiano Reggiano [13,14,15,16,17], and São Jorge cheese [18]. NWSs represent complex microbial ecosystems composed of multiple, often undefined, strains within each species. This microbial diversity plays a fundamental role in shaping the distinctive flavor and aroma profiles that are characteristic of these cheeses. Generally dominant species in NWSs include yeast and thermophilic LAB, such as Streptococcus thermophilus, Lactobacillus helveticus, Lactobacillus delbrueckii, and Limosilactobacillus fermentum, even though mesophilic LAB, such as Lactococcus lactis, Lactiplantibacillus plantarum, Lacticaseibacillus casei, Enterococcus faecium, and Enterococcus faecalis are also encountered, depending on the source of the milk and the technology of cheese-making [7,10]. Moreover, NWS microbial communities include not only numerous species but also various biotypes within the same species: a distinguishing feature that differentiates natural cultures from selected ones.

The primary objective of this study was to establish a defined collection of naturally occurring LAB isolates that are suitable for developing targeted starter cultures intended for use in fermented dairy products and functional foods. Among the many cheeses traditionally produced in southern Italy, caciocavallo remains relatively little-explored, with only a few studies focusing mainly on its microbial ecology and community dynamics rather than strain-level isolation and characterization [12,19,20,21]. In our previous works, we isolated several LAB strains from NWS of buffalo and cow’s milk, primarily investigating the probiotic properties of certain Lactococcus lactis strains [10,11]. Moreover, caciocavallo offers a technologically relevant model system, as it harbors a microbial consortium that has naturally adapted to the specific processing conditions of pasta filata cheesemaking. Its well-established acidification capacity and contribution to flavor development make it an ideal reference matrix for exploring strain-level diversity and functional traits within traditional dairy fermentations [22].

Here, we report an integrated approach, adopted to deeply investigate the microbial diversity of NWSs used for the artisanal production of caciocavallo cheese, as well as the technological potential of the newly isolated strains in dairy fermentation. We combine culture-dependent analysis (sequencing of the 16S rDNA, species/subspecies-specific PCR, polymerase chain reaction, and clustering by RAPD-PCR, random amplified polymorphic DNA–polymerase chain reaction) with culture-independent assessment (NGS analysis), to obtain a comprehensive understanding of both taxonomic and functional traits of autochthonous LAB populations. Indeed, small-scale experiments utilizing skimmed milk as a substrate were performed to evaluate the acidification potential and technological functionalities of the isolated species for their application in dairy (cheese) manufacturing [23]. The isolation of genetically different strains, coupled with the evaluation of their safety and technological features, provides novel insights into the microbial ecology of traditional dairy fermentations in southern Italy. Furthermore, the identification of LAB strains with desirable acidification and metabolic profiles highlights their potential for the development of region-specific starter cultures, thereby contributing to the preservation of microbial biodiversity and the valorization of artisanal cheese-making practices.

2. Materials and Methods

2.1. Natural Whey Starter Culture Sampling

One batch of natural whey starter (NWS) culture derived from cow’s milk for the production of caciocavallo was obtained from a dairy artisanal factory located in the Campania region (southern Italy). The NWS culture was preserved at −80 °C, after the addition of 20% (v/v) glycerol for subsequent analysis.

2.2. Bacteria Enumeration and Isolation from NWS Culture

NWS (20 milliliters) was diluted with 2% trisodium citrate and homogenized using a stomacher (Stomacher 400 Circulator, Avantor, Radnor, PA, USA) for 2 min. Serial tenfold dilutions of NWS samples were prepared in 0.9% NaCl solution. Cell suspensions were plated and incubated as follows: (I) on Man, Rogosa, Sharpe (MRS) agar media, incubated aerobically for 48 h at 30 °C, for mesophilic LAB rods; (II) on ESTY agar, incubated aerobically for 48 h at 30 and 42 °C, for mesophilic and thermophilic LAB cocci, respectively; (III) on brain heart infusion (BHI) agar, incubated aerobically for 48 h at 37 °C, for enterococci and other fastidious or spoilage bacteria; (IV) on Rogosa agar, incubated anaerobically for 48 h at 37 °C (anaerobic jar system, OXOID, Basingstoke, UK), for the lactobacilli growth. All media were supplied by Condalab (Madrid, Spain). After incubation, colonies were counted, and the results were expressed as the logarithm (base 10) of colony-forming units per milliliter of NWS sample (log CFU/mL). Ten random colonies from each growth condition were selected from the lowest dilution plate and purified by triple streaking on agar media. All cultures were stored at −80 °C in 20% (v/v) glycerol.

2.3. Enrichment Cultures

Enrichment for Lactobacilli was performed in Rogosa broth, which contains a higher sodium acetate (15 g/L) concentration, enhancing its selectivity for Lactobacilli [24,25]. One mL of NWS homogenate was used to inoculate 10 mL of Rogosa broth and, after anaerobic incubation at 37 °C for 72 h, the viable count was performed through serial decimal dilutions (as described above). Ten colonies were randomly isolated from the lowest plate dilution, and the cultures obtained from them were stored at −80 °C in 20% (v/v) glycerol.

2.4. Genotypic Identification by Partial 16S rDNA Gene Sequence Analysis

Genomic DNA extraction from isolated colonies was carried out using the Bacterial Genomic DNA Isolation Kit (Norgen Biotek Corp., Thorold, ON, Canada), according to the manufacturer’s instructions. Amplification of the V1-to-V3 region of the 16S rDNA gene and sequencing were performed as previously described [10].

2.5. Species- and Subspecies-Specific PCR

PCR reactions were performed by using a SimpliAmp thermal cycler (Applied Biosystems, Waltham, MA, USA), as already described [10]. Annealing temperatures were optimized for each target species: 60 °C for L. lactis subsp. cremoris, 54 °C for E. faecium, 51 °C for L. helveticus, 50 °C for L. rhamnosus, and 58 °C for S. thermophilus, L. lactis subsp. lactis, and L. fermentum. When possible, the specificity of primers was evaluated by a PCR of one reference strain from the laboratory collection, and negative controls were included as well. Synthetic primers used in species-specific PCR are reported in

Table 1. Primers for species and subspecies-specific PCR.

Genus/Species/Subspecies	Primer	Sequence 5′ > 3′	Target Gene	Amplicon Size (bp)	Ref.
S. thermophilus	Str-THER-F2116 Str-THER-R2693	GCTTGGTTCTGAGGGAAGC CTTTCTTCTGCACCGTATCCA	lacZ	578	[10]
L. lactis subsp. lactis	Lactis F 23 SRevLac	GCTGAAGGTTGGTACTTGTA AGTGCCAAGGCATCCACC	16S rRNA (V1) and 16S/23S spacer region	1824	[10]
L. lactis subsp. cremoris	PLc(A)For 23SRevLac	GGTGCTTGCACCAATTTGAA AGTGCCAAGGCATCCACC	16S rRNA (V1) and 16S/23S spacer region	1814	[10]
L. fermentum	Lac-FER-F753 Lac-FER-R1062	CCAGATCAGCCAACTTCACA GGCAAACTTCAAGAGGACCA	Arginine-ornitine antiporter	310	[10]
L. rhamnosus	PrI RhaII	CAGACTGAAAGTCTGACGG GCGATGCGAATTTCTATTATT	16S rRNA (V1) and 16S/23S spacer region	512 283	[26]
L. helveticus	LHELV_F LHELV_R	ACGATGCTACTCACTCACAC GCACCTAATACTTCGATCCAAC	Phenylalanine tRNA ligase	256	This study
E. faecium	EfaeciumF EfaeciumR	GCAAGGCTTCTTAGAGA CATCGTGTAAGCTAACTTC	D-Ala Ala ligase	575	[10]

.

2.6. Random Amplified Polymorphic DNA–Polymerase Chain Reaction (RAPD-PCR) Analysis

Genomic DNA from each strain was extracted as described above. RAPD1 (5′-AGCAGGGTCG-3′) and M13 (5′-GAGGGTGGCGGTTCT-3′) primers were employed. RAPD reaction mixture and amplification programs were performed as previously described [10]. PCR amplicons were separated by electrophoresis on 2.5% agarose gel (100 V, 4 h), stained with ethidium bromide (0.5 mcg mL⁻¹), and subsequently visualized under UV light with Typhoon Biomolecular Imager (Amersham, UK). Fragment sizes were estimated against a 1 kb Plus-DNA marker (abm, Richmond, BC, Canada; 100–10,000 bp). For strain similarity assessments, only distinct amplification bands, regardless of staining intensity, were scored in a binary format as present (1) or absent (0). The binary matrix was analyzed using Jaccard’s coefficient, and the resulting similarity values were employed to construct a UPGMA dendrogram with DendroUPGMA (https://usuaris.tinet.cat/debb/UPGMA, accessed on 17 November 2023).

2.7. High-Throughput Sequencing

Microbial DNA extraction from NWS was performed, and the next generation sequencing analysis was realized at the BMR Genomics srl service (https://www.bmr-genomics.it, accessed on 17 November 2023), as previously described [10].

2.8. Determination of the Safety Parameters

2.8.1. Antibiotic Susceptibility

Genetically distinct strains were tested for antimicrobial susceptibility by the Kirby–Bauer disk diffusion method, as previously described [27]. Antimicrobial disks were supplied by Condalab (Madrid, Spain). After 24 h of incubation, the inhibition zone diameters were measured and results were expressed as sensitive (S), intermediate (I), and resistant (R), following the criteria recommended by the disk manufacturer, based on CLSI and EUCAST guidelines.

2.8.2. Hemolytic Activity

Hemolytic activity was assessed, following the protocol described by Maragkoudakis et al. [28]. Colonies were streaked onto Columbia agar supplemented with 5% defibrinated sheep blood (VWR, Leuven, Belgium) and incubated under conditions specific to each strain. Plates were inspected after 48 h, and then seven days after incubation. Hemolytic activity was classified as α-hemolysis, β-hemolysis, or γ-hemolysis. S. aureus ATCC 6538 and L. plantarum WCFS1 were used as positive controls of β and γ-hemolysis, respectively.

2.9. Homo- and Heterofermentative Properties

2.9.1. Small-Scale Experiments for Biomass and Acid Production

Frozen stocks of each strain were prepared from cultures in the exponential phase, grown in MRS, M17, or BHI broth (VWR International S.r.l., Milan, Italy), and stored at −80 °C after the addition of a 20% v/v glycerol solution. Small-scale experiments were performed in 50 mL tubes with a working volume of 40–45 mL, incubated at 30, 37, or 44 °C in static conditions or −150 rpm in a rotary shaker incubator (model Minitron, Infors, Bottmingen, Switzerland) for 24 h (

Table 2. Growth conditions of each strain.

Strain	Origin	Respiratory Metabolism	Temperature (°C)	Media	Volume of Media	Agitation (RPM)
S. thermophilus BA1	Cow	Aerobic	44	M17	40	150
S. thermophilus BB6	Cow	Aerobic	44	M17	40	150
L. lactis subsp. lactis BA7	Cow	Aerobic	30	M17	40	150
L. lactis subsp. lactis BB8	Cow	Aerobic	30	M17	40	150
L. lactis subsp. lactis BC9	Cow	Aerobic	30	M17	40	150
L. lactis subsp. lactis BC10	Cow	Aerobic	30	M17	40	150
L. lactis subsp. lactis BD1	Cow	Aerobic	30	M17	40	150
L. lactis subsp. lactis BD5	Cow	Aerobic	30	M17	40	150
L. fermentum BF1	Cow	Anaerobic	37	MRS	45	0
L. fermentum BF2	Cow	Anaerobic	37	MRS	45	0
L. fermentum BE2	Cow	Anaerobic	37	MRS	45	0
L. rhamnosus BF3	Cow	Anaerobic	37	MRS	45	0
L. helveticus BE3	Cow	Anaerobic	37	MRS	45	0
E. faecium BD6	Cow	Aerobic	37	BHI	40	150

). Optical density (O.D.) at 600 nm (using DU730 line science UV/VIS spectrophotometer, Beckman culture, Milan, Italy); cell dry weight, obtained by oven drying cell pellets for 24 h at 40 °C; carbon sources consumption; and acid and ethanol production (UHPLC analyses) were analyzed at time 0, and after 8 and 24 h. Viability was evaluated by serial decimal dilutions and plating on MRS, M17, or BHI agar medium [29]. Plates were incubated for 36–48 h before counting viable cells. Each experiment was performed in triplicate.

2.9.2. Analytical Methods: UHPLC Analyses

Samples were analyzed for lactose and glucose residue after 8 and 24 h, as well as lactic acid and ethanol production using a Dionex Ultimate 3000+ UHPLC chromatograph (Thermofisher, Milan, Italy) with a UV/Vis and RI detector. The standards and samples were first ultrafiltered on Centricons with a 3 kDa cut-off, centrifuged at 12,000 rpm for 15 min, and then the permeate was recovered. The samples were then injected onto a Phenomenex Rezex ROA-organic acid column under the following conditions: isocratic elution with 0.1% sulfuric acid in water at 0.6 mL/min, temperature of 35 °C, concentration range of 20–0.2 mg/mL, and an acquisition time of 25 min [30]. During analyses, a reference standard (e.g., glucose, lactose, lactic acid) at various concentrations is periodically analyzed, along with the samples, to verify repeatability and accuracy.

3. Results

3.1. Enumeration and Isolation of Lactic Acid Bacteria from Natural Whey Starter Culture

The concentration of the major microbial cultivable groups in NWS was assessed by plate counts, using media supporting the growth of lactobacilli, lactococci/streptococci, enterococci, and other fastidious microorganisms (

Table 3. Cell numbers * (log CFU mL⁻¹) of presumptive lactic acid bacteria groups isolated from the natural whey starter culture (NWS) of cow milk.

NWS Sample	Mesophilic Lactococci and Streptococci (ESTY)	Termophilic Lactococci and Streptococci (ESTY)	Enterococci and Others (BHI)	Mesophilic Lactobacilli (MRS)	Mesophilic Lactobacilli (Rogosa)	Mesophilic Lactobacilli (Enriched Culture)
Cow’s milk	7.1 ± 0.1	5.9 ± 0.2	7.07 ± 0.1	6.1 ± 0.2	6.5 ± 0.1	9.6 ± 0.1

* Mean values of viable count performed in triplicate.

). The presumed mesophilic lactococci and streptococci, as well as enterococci and other fastidious microorganisms, were prevalent (7.1 log CFU mL⁻¹ and 7.07 log CFU mL⁻¹, respectively), with respect to lactobacilli and thermophilic cocci (6.1–6.5 log CFU mL⁻¹ and 5.9 log CFU mL⁻¹, respectively). Despite the high number of putative lactobacilli counted in the MRS medium, generally, the isolation of well-growing Lactobacillus spp. from natural sources often fails [10,16,31]. For this reason, we also explored the possibility of increasing the amounts of growing Lactobacillus strains by incubating the NWS sample for 72 h in Rogosa broth, a selective medium for Lactobacillus [25]. This enrichment step increased the viable cell count from 6.1 to 9.6 log CFU mL⁻¹ (Table 3). As expected, the NWS culture under investigation exhibited a high abundance of lactic acid bacteria. A total of 60 colonies, 10 from each medium and growth condition, were randomly selected from the plates with the lowest dilution for further examination. Of these, 41 isolates were successfully re-cultured and subjected to taxonomic identification.

3.2. Molecular Identification of Species Isolated from Natural Whey Starter Cultures

3.2.1. Sequencing of V1–V3 rDNA Region

Sequencing of highly variable V1-V3 rDNA regions has been widely used for the identification and detection of lactic acid bacteria [10,16,32]. According to our previous studies [10], we carried out taxonomic identification of the isolated colonies through analysis of the V1–V3 region of the 16S rRNA gene. A 520 bp amplicon was generated from each isolate, using degenerate oligonucleotides targeting conserved bacterial domains. The PCR products were purified, sequenced, and analyzed using the BLAST 2.17.0 algorithm for taxonomic assignment. Forty-one clones were analyzed and 27 of them were identified as LAB, including 3 S. thermophilus, 6 L. lactis subsp. lactis, 4 E. faecium, 2 L. helveticus, 11 L. fermentum, and 1 L. rhamnosus. The latter and seven clones identified as L. fermentum were isolated after the enrichment step in the Rogosa broth. The remaining 14 isolates were classified as non-LAB, such as Rothia kristinae, Klebsiella pneumoniae, Enterobacter cloacae, Exiguobacterium acetylicum, Fibrobacter succinogenes, Citrobacter braakii, Kocuria kristinae, Pseudomonas putida, and Pseudomonas gessardi. The overall abundance of species identified by the VI-V3 sequence is reported in Figure 1. The name of the isolates, growth conditions, and BLAST results are reported in Supplementary Table S1. The BLAST analysis of the LAB isolates was confirmed by species-specific PCR.

3.2.2. Species- and Subspecies-Specific PCR

Bacteria cannot always be discriminated at the species level using partial 16S rDNA sequencing [33,34,35]. Indeed, various methodologies can be combined to achieve selective identification at the species or subspecies level. In this study, a species- and subspecies-specific PCR strategy was employed, targeting genes that contain hypervariable regions, which provide higher discriminatory power than 16S rDNA for differentiating closely related bacterial taxa. For this purpose, we targeted the following genes: lacZ, coding beta-galactosidase (S. thermophilus), the 16S/23S rDNA region (L. lactis subsp. lactis/subsp. cremoris and L. rhamnosus), the D-Ala/D-Ala ligase coding gene (E. faecium), the arginine–ornithine antiporter coding gene (L. fermentum), and the phenylalanine tRNA ligase coding gene (L. helveticus). Species- and subspecies-specific PCR confirmed the taxonomic identification of all LAB isolates (Figure 2).

3.2.3. Random Amplified Polymorphic DNA–Polymerase Chain Reaction (RAPD-PCR) Analysis

RAPD-PCR analysis was performed to test the genetic diversity of LAB isolates belonging to the same species. We employed two different primers: M13 (5′-GAGGGTGGCGGTTCT-3′) for L. rhamnosus, L. fermentum, L. helveticus, S. thermophilus, and E. faecium, and RAPD1 (5′-AGCAGGGTCG-3′), for L. lactis subsp. lactis, whose efficiency in terms of genotypic diversity, discrimination, and reproducibility was already demonstrated [10]. According to UPGMA (unweighted pair group method algorithm with arithmetic mean) analysis, RAPD profiles were clustered in OTU (operational taxonomic units), taking into account that a similarity coefficient of 1.000 indicates that these isolates belong to the same strain. In detail, RAPD profiles of S. thermophilus were clustered in OTU1 (BA1 and BA3 isolates) and OTU2 (BB6 isolate), sharing a low similarity coefficient (0.111), which indicated high genetic diversity (Figure 3A). Interestingly, electrophoretic profiles obtained with the six L. lactis subsp. lactis isolates were very different from each other. Indeed, they were clustered in six different OTUs (OTU 3–8, corresponding to BA7, BB8, BC9, BC10, BD1, and BD5 isolates, respectively) sharing a similarity coefficient between 0.154 and 0.714, demonstrating high genetic diversity for lactococci too (Figure 3B). Contrarily, RAPD profiles of the two L. helveticus clones (BE3, BE5) were clustered in the same OTU (OTU 14), sharing a similarity coefficient of 1.000, indicating no genetic diversity between the two isolates (Figure 3C). Similarly, fingerprints of the L. fermentum group show identical patterns, with a similarity coefficient of 1.000 for almost all the isolates, grouped in OTU 10 (BE4, BE8, BF1, BF4, BF6, BF7, BF8, and BF10), except for BE2 and BF5 (OTU 9) and BF2 (OTU 11), which share similarity coefficients of 0.938 and 0.455 with OTU10, respectively (Figure 3D). Finally, RAPD profiles of the four E. faecium isolates (BD2, BD4, BD6, BD10) show a similarity coefficient of 1.000, indicating that these isolates are representative of one single strain, hereby clustered in OTU 12 (Figure 3E).

Overall, RAPD analysis revealed the presence of two genotypically different S. thermophilus strains, six L. lactis subsp. lactis, one L. helveticus, three L. fermentum, and one E. faecium.

3.2.4. 16S rDNA Next-Generation Sequencing Analysis (NGS)

Given the well-documented limits of culture-dependent methods to characterize microbial communities, we also applied a culture-independent metagenomics approach, based on high throughput sequencing (HST) of the variable regions of the 16S rRNA gene. The sequencing results showed 30,838 validated reads, whose analysis revealed the presence of a bacterial community constituted by Firmicutes and Proteobacteria in the NWS. The latter mainly includes bacteria belonging to the Moraxellaceae, Pseudomonadaceae, Enterobacteriaceae, and Erwiniaceae families. Based on the purpose of this work, we focused our attention on the Firmicutes, including the LAB belonging to the Streptococcaceae, Lactobacillaceae, Leuconostocaceae, and Enterococcaceae families. Interestingly, the LAB species identified by NGS analysis mostly overlap with those identified by the culture-dependent approach, with differences observed only in terms of the relative abundance of LAB (Figure 4). Moreover, the Leuconostoc mesenteroides, Lactobacillus delbrueckii, and Lactococcus raffinolactis found by NGS were not detected by culture-dependent strategies. This discrepancy may result from the inherent limitations of traditional culture-based taxonomic approaches [37]. However, it is worth considering that the statistical analysis of NGS data showed a good confidence value of the taxonomic results only for genus-level identification and not for species-level identification.

3.3. Safety Parameters (Antibiotic Susceptibility and Hemolytic Activity)

The isolated strains were evaluated for their antibiotic susceptibility: an essential criterion for assessing the safety of microorganisms intended for human or animal use. Antibiotic susceptibility testing was conducted, using the agar diffusion method (Kirby–Bauer assay) against five antimicrobial agents: ampicillin (AM), penicillin (P), gentamicin (CN), vancomycin (VA), and tetracycline (TE). The results of these assays, summarized in

Table 4. Antibiotic sensitivity test (AM, ampicillin; TE, tetracycline; P, penicillin; CN, gentamicin; VA, vancomycin; S, sensitivity; and R, resistance).

Strain	AM 10 μg	TE 30 μg	P 10 μg	CN 10 μg	VA 30 μg
L. lactis subsp. lactis BC9	S	S	S	S	S
L. lactis subsp. lactis BB8	R	R	R	R	R
L. lactis subsp. lactis BC10	R	R	R	R	R
L. lactis subsp. lactis BD5	R	R	R	R	R
L. lactis subsp. lactis BA7	S	S	S	S	S
L. lactis subsp. lactis BD1	R	R	R	R	R
S. thermophilus BA1	R	R	R	R	R
S. thermophilus BB6	S	S	S	R	S
L. fermentum BE2	S	S	S	S	R
L. fermentum BF1	S	S	S	S	R
L. fermentum BF2	S	S	S	S	R
L. helveticus BE3	S	S	S	R	S
L. rhamnosus BF3	S	S	S	S	R
E. faecium BD6	S	S	S	R	S

, are expressed as resistant (R), sensitive (S), and intermediate (I), according to that described by the disk manufacturer. All isolates, except L. lactis BB8, BC10, BD5, and BD1, and S. thermophilus BA1, were susceptible to most antibiotics tested. In detail, among L. lactis subsp. lactis, strains BC9 and BA7 were susceptible to all the antibiotics tested; S. thermophilus BB6, L. helveticus BE3, and E. faecium BD6 showed resistance to gentamicin. Finally, all Lactobacillus strains, except L. helveticus BE3, exhibited resistance only to vancomycin (Table 4).

The absence of hemolytic activity represents another key criterion in the selection of food-grade microbial strains, as such activity may indicate potential bacterial virulence [38]. In the present study, hemolytic activity was evaluated on Columbia blood agar plates. None of the tested isolates exhibited hemolytic activity, confirming their non-virulent behavior under in vitro conditions.

3.4. Physiological Studies in Small-Scale Fermentation Experiments

Small-scale experiments were performed on fourteen isolated strains in skimmed milk and media (MRS, M17, or BHI) to evaluate acidification (pH), growth (OD₆₀₀ and cell dry weight), viability, substrate consumption (glucose and lactose), organic acids, and ethanol production.

Table 5. pH and viability in skimmed milk experiments. The initial pH of milk is 6.8–6.9.

Strain	pH (4 h)	pH (6 h)	Final pH (24 h)	Viability (CFU/mL)
S. thermophiles BA1	6.8 ± 0.1	6.7 ± 0.1	6.6 ± 0.1	5 × 102
S. thermophiles BB6	5.8 ± 0.1	5.5 ± 0.1	5.3 ± 0.1	6 × 106
L. lactis subsp. lactis BA7	5.6 ± 0.2	5.4 ± 0.1	5.3 ± 0.1	7 × 107
L. lactis subsp. lactis BB8	6.5 ± 0.1	6.0 ± 0.1	5.8 ± 0.1	4 × 107
L. lactis subsp. lactis BC9	6.2 ± 0.2	6.1 ± 0.1	5.9 ± 0.1	4 × 107
L. lactis subsp. lactis BC10	6.4 ± 0.1	6.1 ± 0.1	5.9 ± 0.2	9 × 107
L. lactis subsp. lactis BD1	6.2 ± 0.1	6.0 ± 0.2	5.2 ± 0.1	2 × 106
L. lactis subsp. lactis BD5	6.7 ± 0.1	6.6 ± 0.2	5.8 ± 0.2	4 × 106
L. fermentum BF1	6.1 ± 0.1	5.9 ± 0.2	5.2 ± 0.2	6 × 107
L. fermentum BF2	5.8 ± 0.1	5.6 ± 0.1	5.1 ± 0.1	9 × 108
L. fermentum BE2	5.8 ± 0.2	5.5 ± 0.1	5.1 ± 0.2	7 × 108
L. rhamnosus BF3	5.6 ± 0.2	5.3 ± 0.1	4.8 ± 0.2	9 × 108
L. helveticus BE3	6.3 ± 0.1	5.5 ± 0.1	5.0 ± 0.1	9 × 108
E. faecium BD6	6.3 ± 0.1	6.0 ± 0.1	5.4 ± 0.2	7 × 107

reports the results obtained from the milk samples, including pH measurements and cell viability. For viability, test cells were plated on MRS, M17, or BHI media.

Table 6. pH, growth, and viability in media experiments at 24 h.

Strain	OD600nm	Cell Dry Weight (g/L)	Final pH	Viability (CFU/mL)	Cell Dry WEIGHT/O.D.	Median with Standard Deviation (Cell Dry Weight/OD) for Strain
S. thermophiles BA1	1.4 ± 0.1	0.3 ± 0.1	6.0 ± 0.1	7 × 103	0.2 ± 0.1	0.25 ± 0.1
S. thermophiles BB6	2.2 ± 0.1	0.6 ± 0.1	5.8 ± 0.1	7 × 107	0.3 ± 0.1
L. lactis subsp. lactis BA7	3.7 ± 0.3	1.5 ± 0.1	5.4 ± 0.1	3 × 108	0.4 ± 0.1	0.41 ± 0.1
L. lactis subsp. lactis BB8	2.8 ± 0.2	1.2 ± 0.1	5.9 ± 0.1	4 × 108	0.4 ± 0.1
L. lactis subsp. lactis BC9	2.6 ± 0.2	1.1 ± 0.1	5.4 ± 0.1	1 × 108	0.4 ± 0.1
L. lactis subsp. lactis BC10	2.9 ± 0.2	1.4 ± 0.1	6.1 ± 0.1	4 × 108	0.4 ± 0.1
L. lactis subsp. lactis BD1	2.8 ± 0.2	0.7 ± 0.1	5.4 ± 0.1	3 × 108	0.6 ± 0.1
L. lactis subsp. lactis BD5	2.1 ± 0.1	0.6 ± 0.1	5.4 ± 0.1	5 × 109	0.2 ± 0.1
L. fermentum BF1	4.9 ± 0.2	1.6 ± 0.1	4.5 ± 0.1	9 × 107	0.3 ± 0.1	0.33 ± 0.05
L. fermentum BF2	3.9 ± 0.1	1.7 ± 0.1	4.7 ± 0.1	9 × 107	0.4 ± 0.1
L. fermentum BE2	5.1 ± 0.2	1.7 ± 0.1	4.5 ± 0.1	8 × 107	0.3 ± 0.1
L. rhamnosus BF3	6.5 ± 0.2	3.3 ± 0.1	4.0 ± 0.1	7 × 107	0.5 ± 0.1	0.5 ± 0.1
L. helveticus BE3	4.7 ± 0.1	1.8 ± 0.1	4.6 ± 0.1	4 × 107	0.4 ± 0.1	0.4 ± 0.1
E. faecium BD6	2.2 ± 0.1	0.9 ± 0.1	5.5 ± 0.1	6 × 107	0.4 ± 0.1	0.4 ± 0.1

and

Table 7. Evaluation of the organic acid contents by HPLC analysis at 8 and 24 h. Lactic acid concentrations were adjusted based on its presence in the control samples. The lowest concentration of the standards used is ≤0.20. Yeald (Yp/s) was calculated at 24 h.

Strain	8 h		24 h
	Lactic Acid (g/L)	Ethanol (g/L)	Lactic Acid (g/L)	Ethanol (g/L)	Yield Yp/s (Lactic Acid Produced/Glucose–Lactose Consumed g/g)
L. lactis subsp. lactis BC9	5.8 ± 0.2	0.4 ± 0.1	6.4 ± 0.1	0.5 ± 0.1	1.2 ± 0.2
L. lactis subsp. lactis BC10	6.8 ± 0.2	0.2 ± 0.2	7.0 ± 0.2	0.6 ± 0.1	1.4 ± 0.1
L. lactis subsp. lactis BD1	11.2 ± 0.3	0.4 ± 0.1	11.2 ± 0.1	0.5 ± 0.1	1.7 ± 0.1
L. lactis subsp. lactis BD5	5.1 ± 0.2	0.5 ± 0.1	7.1 ± 0.2	0.5 ± 0.1	1.0 ± 0.1
L. fermentum BF1	12.5 ± 0.3	3.6 ± 0.2	12.5 ± 0.1	3.6 ± 0.2	0.7 ± 0.1
L. fermentum BF2	12.9 ± 0.3	3.3 ± 0.2	12.7 ± 0.2	3.5 ± 0.1	0.8 ± 0.1
L. fermentum BE2	12.8 ± 0.3	3.4 ± 0.2	12.5 ± 0.2	3.6 ± 0.2	0.8 ± 0.1
L. rhamnosus BF3	14.0 ± 0.2	≤0.20	15.6 ± 0.2	≤0.20	1.1 ± 0.1
L. helveticus BE3	13.2 ± 0.2	0.3 ± 0.1	13.4 ± 0.1	0.3 ± 0.1	0.8 ± 0.1
E. faecium BD6	3.2 ± 0.2	0.3 ± 0.1	3.3 ± 0.1	0.2 ± 0.1	0.6 ± 0.1

show the results obtained using standard media. L. fermentum strains displayed a heterofermentative metabolism, generating roughly 3.5 g/L of ethanol, whereas S. thermophilus, L. lactis, and L. rhamnosus strains demonstrated a homofermentative metabolic profile. The MRS medium did not exhibit the same buffering capacity as M17. This difference may be attributed to the presence of beta disodium glycerophosphate in M17 and to the lower initial carbohydrate concentration: 1% (w/v) lactose in M17 compared to 2% (w/v) glucose in MRS. Indeed L. lactis subsp. lactis BD5 reached a pH of 5.4, with the highest viability of 5 × 10⁹ CFU/mL., while L. rhamnosus BF3 decreased the pH to a final value of four, with a lower viability of 7 × 10⁷ CFU/mL (Table 6). In fact, the highest lactic acid concentration was produced by L. rhamnosus BF3, reaching 15.6 g/L after 24 h (Table 7), which represents a notable titer at this scale. However, when considering productivity calculated over the first 8 h of fermentation, L. lactis subsp. lactis BD1, L. fermentum BF1, and L. rhamnosus BF3 exhibited high rates, ranging from 1.4 to 2.0 g/L·h.

The data of HPLC analyses showed higher lactic acid production for the strains growing on MRS, as expected, since this contains a greater amount of glucose (Table 7). The calculated corresponding yield coefficient (Yp/s) ranged from 0.7 to 1.1 lactic acid from glucose conversion. However, the yield coefficient was higher for the strains grown in M17 (Streptococcus and Lactococcus), likely due to their simultaneous utilization of amino acid-based substrates (i.e., peptides/proteins), which are present in higher amounts and in a higher ratio to carbohydrates in M17, respective to MRS (Supplementary Table S2).

4. Discussion

This study provides an approach to investigate the microbial composition of natural whey starter culture, taking advantage of the traditional cheesemaking production that is still diffusely present in Italy, particularly in rural territory. In this context, the work aimed to identify new LAB strains that can be employed in food processing and biotechnology. In recent years, numerous studies have investigated the microecosystem of NWS culture used in artisanal dairy production, where Firmicutes represent the dominant phylum, although Proteobacteria, including Acinetobacter, Pseudomonas, Aeromonas, and others, have also been found. The microbial composition of NWS depends on the origin of the milk and environmental contamination [13,16,39]. The diversity of microbial species within these cultures is essential for producing cheeses with the typical sensorial properties of traditional dairy products. Moreover, NWS cultures may serve as a valuable reservoir of interesting LAB, including novel taxa or strains possessing distinctive functional traits, which could be exploited for the development of new starter cultures and innovative fermented food products. In this study, we carried out a targeted search for LAB from natural whey starters derived from cow’s milk used in the production of caciocavallo, a traditional pasta filata cheese that is widely produced in southern Italy, and whose microbial diversity has been relatively described. [10,12,39]. Molecular analyses performed on 41 isolates showed that 27 of them were LAB, with L. lactis subsp. lactis emerging as the most prevalent species, with 6 clones isolated, followed by E. faecium and L. fermentum with 4 clones each, S. thermophilus with 3 clones, L. helveticus with 2 clones, and finally, L. rhamnosus with 1 clone. In addition, one strain of L. rhamnosus and seven other strains of L. fermentum were isolated by an enrichment step in Rogosa broth [20,21]. The remaining 14 isolates were taxonomically identified as non-LAB, including R. kristinae, K. pneumoniae, E. cloacae, E. acetylicum, F. succinogenes, C. braakii, K. kristinae, P. putida, and P. gessardi. The presence of some of them is commonly found in natural starters and/or raw cow milk, due to environmental contamination. This might reveal poor hygiene and sanitation conditions, as well as an insufficient heating process [1,40,41]. Genotypic analysis of the isolates showed high microbial diversity at the strain level. In particular, RAPD-PCR profiles revealed significant genetic diversity among the isolated clones of L. lactis, with six different OTUs, L. fermentum with three OTUs, and S. thermophilus with two OTUs, sharing a very low similarity coefficient (0.1). The genotypic diversity of microbial species within these cultures is essential for the production of cheeses with unique characteristics and for preventing the dominance of bacteriophages that can disrupt the cheese-making process [42].

Interestingly, the results obtained by classical identification, based on culture-dependent methods, were mostly in agreement with those obtained with NGS analysis. Both methods revealed Lactococcus as the most abundant genus, followed by Lactobacillus, Enterococcus, and Streptococcus. However, Leuconostoc mesenteroidetes, Lactobacillus delbrueckii, and Lactococcus raffinolactis were only detected by NGS analysis. This outcome is likely attributable to the limited cultivability of the bacteria from natural samples, characterized by high microbial diversity [31], and/or to the reduced viability of these bacteria after storage of the NWS sample at −80 °C. In fact, it has been largely discussed in the general context of cheese production that LAB cells could be in a viable but not cultivable state (VBNC) [43,44]. These findings highlight the importance of integrating both culture-dependent and culture-independent methods to achieve a more precise and comprehensive identification of LAB species in dairy products [33].

Based on their historical use in food fermentation, many LAB species are “generally recognized as safe” (GRAS). However, genus and/or species allocation is often not considered to be enough to guarantee safety [7]. Hence, the selection of microorganisms to be used in food requires assessing some safety parameters, such as antibiotic sensitivity and the absence of hemolytic activity [7]. Indeed, antibiotic resistance is a growing concern for public health and the global economy. Lactic acid bacteria (LAB) associated with dairy products and naturally inhabiting the agro-zootechnical environments can act as reservoirs of antibiotic resistance genes, with the ability to acquire such genes or transfer them to other microorganisms [45]. We performed antibiotic susceptibility tests on LAB isolates against five antimicrobial agents: ampicillin (AM), penicillin (P), gentamicin (CN), vancomycin (VA), and tetracycline (TE). The selection of these antibiotics was based on both technological relevance and health considerations. In particular, beta-lactams (ampicillin and penicillin) and tetracycline are commonly used in veterinary practice, thereby contributing to the spread of antibiotic resistance through the food chain. Moreover, these antibiotics, together with gentamicin and vancomycin, are among the most commonly used to assess the safety parameters of new isolates from the natural environment, given their widespread use in clinical practice [46,47,48,49].

The S. thermophilus BA3 and Lactococcus lactis BB8, BC10, BD5, and BD1 strains show resistance towards all the antibiotics tested, probably due to the presence of multidrug resistance pumps (MDR) [45,50]. Poelarends and co-workers [51] demonstrated that L. lactis is often resistant to clinically relevant antibiotics, including aminoglycosides, due to the presence of the LmrA transporter. Contrarily, the L. lactis strains BC9 and BA7 are sensitive to all tested antibiotics, and S. thermophilus BB6 only shows resistance to gentamicin, but the latter appears to be an intrinsic resistance and, therefore, does not represent a danger to human health [52]. Finally, all the L. fermentum isolates (BE2, BF1, BF2) and L. rhamnosus BF3 show intrinsic resistance to vancomycin, but sensitivity to all the other antibiotics tested; similarly, L. helveticus BE3 and E. faecium BD6 only showed resistance to gentamicin at the concentration tested. Generally, lactobacilli and enterococci are sensitive to beta-lactam antibiotics, such as ampicillin and penicillin, and broad-spectrum antibiotics such as tetracycline. On the other hand, they are resistant to vancomycin and gentamicin; however, this is an intrinsic resistance and is therefore not transferable to other species [53,54]. These results will be further investigated, including the assessment of minimum inhibitory concentrations (MICs) with the application of EFSA 2018 breakpoints, as well as the evaluation of intrinsic resistance traits at both the genus and species levels.

Analysis of hemolytic activity is strongly recommended when isolated strains are intended for use in food products, even if they possess GRAS or QPS (Quality Presumption of Safety) status (FAO) [38]. Hemolytic activity was tested on Columbia blood agar plates, revealing γ-hemolytic or no hemolytic activity, as already reported by many authors [54,55,56]. When the safety of newly isolated bacteria is concerned, the absence of hemolytic activity represents a key selection parameter, as hemolysis is associated with potential bacterial virulence. Strains that do not exhibit hemolytic activity are regarded as safe and non-virulent, since the absence of hemolysin production minimizes the risk of opportunistic pathogenicity [47].

The major technological traits associated with the LAB employed in food processing include their metabolic characteristics, productivity and biomass yield, and their ability to produce lactic acid through carbohydrate fermentation, particularly lactose in dairy manufacturing. Lactic acid (LA) causes a pH reduction, contributing to the cheese ripening and therefore generating a stressful condition for pathogenic or spoilage microorganisms, thus improving the hygienic properties and prolonging the shelf life of the final product [1,7]. The present study accomplished a full array characterization, both in skimmed milk and in semi-defined media, to analyze growth, viability, and metabolic activity with specific reference to LA biosynthesis. The obtained results show a marked pH decrease and, therefore, a higher acidification for the L. rhamnosus, L. fermentum, and L. helveticus strains. Thus, they can be indicated as better starters for dairy processing.

Growth and viability at 24 h prove to have a very consistent biomass/OD₆₀₀ ratio for all strains, demonstrating the reproducibility of the data obtained. Furthermore, for almost all strains, a decrease in viability is associated with an increased LA production. In fact, the pH lowering cannot be controlled in small-scale tube cultivation, unless there are buffering agents in the medium. Indeed, the presence in the M17 media of di-sodium β-glycerophosphate increases the buffering capacity, maintaining the stable pH. This also correlates with a higher viability of the strains grown on M17, as evaluated by colony-forming unit counting [57]. In agreement with the updated literature, the newly isolated strains of L. fermentum proved heterofermentative, with an ethanol production of about 3.5 g/L in 24 h [58], while S. thermophilus, L. lactis, L. rhamnosus, L.helveticus, and E. faecium showed a homofermentative metabolism, as expected [59,60,61]. Furthermore, the results obtained for the different strains are consistent with those reported in the literature, which indicate that most lactic acid bacteria exhibit slower acidification kinetics in milk compared to standard culture media such as MRS and M17 [62,63]. The production of lactic acid from glucose and lactose represents a key biochemical process during dairy manufacture. A rapid decrease in pH plays a crucial role, as it promotes coagulation and contributes to the inhibition of undesirable microorganisms [64]. Moreover, Ayad and collaborators indicate that both rapidly and slowly acidifying strains can be employed in dairy processing [62]. Rapid acidifiers are suitable candidates for primary starter cultures, as they ensure a prompt reduction in pH required for coagulation and microbial control. Conversely, slow-acidifying strains may be used as adjunct cultures, contributing to the development of sensory attributes and the overall complexity of the final product.

The 14 strains isolated showed growth and metabolites production in agreement with what was expected for these genera. In fact, their LA biosynthesis occurs within a few hours of cultivation, resulting in a reduction in the growth rate. In particular, the L. fermentum, L. rhamnosus, and L. helveticus strains showed an increase in LA up to 24 h cultivation. For starter culture applications, it may be advantageous to combine faster-growing strains (e.g., L. fermentum and L. rhamnosus or L. helveticus) with slower-growing strains, including at least one strain that is tolerant to high LA concentrations. This strategy would allow for further acidification of the milk or curd, even after certain fast-growing strains have ceased active growth due to pH decline and LA accumulation.

5. Conclusions

In this study, we report the isolation and molecular characterization of 14 autochthonous LAB strains, belonging to the Streptococcus, Lactococcus, and Lactobacillus genera. Despite the promising technological properties observed, four of the Lactococcus isolates and one Streptococcus cannot be considered for starter culture formulation due to multidrug resistance. For the remaining strains, levels of acidifying, metabolic properties, and safety parameters suggest their potential as starter cultures in dairy processing. This work paves the way to further investigation aimed at confirming the technological properties at a pilot scale and the safety properties of the isolated strains to use as adjunct cultures for fermented dairy products and functional food. The potential use of the selected LAB strains as autochthonous starters represents a promising strategy to guarantee the reproducibility and quality of the final product, while preserving the authenticity, flavor, and overall quality of traditional regional cheeses.