Antibiotic Susceptibility of Autochthonous Enterococcus Strain Biotypes Prevailing in Sheep Milk from Native Epirus Breeds Before and After Mild Thermization in View of Their Inclusion in a Complex Natural Cheese Starter Culture

Abstract

Autochthonous enterococci surviving mild thermization of raw milk (RM) before traditional Greek cheese processing may simultaneously comprise safe and virulent thermoduric strains with multiple antibiotic resistances (ARs). Therefore, this study biotyped and then compared the ARs of 60 Enterococcus isolates from two antilisterial sheep milks of native Epirus breeds before (RM) and after (TM) thermization at 65 °C for 30 s; the RM isolates were previously genotyped and evaluated for primary safety traits, namely, hemolytic activity, vanA/vanB, cytolysin, and virulence genes, by molecular methods. Biochemically typical and atypical strains of Enterococcus faecium (six biotypes), E. durans (five biotypes), E. faecalis (two biotypes), and E. hirae (one biotype), which were subdominant to other LAB species in RM (19 isolates), prevailed in TM (41 isolates). E. faecium biotypes 1A, 1D, and 1H included multiple-Ent+ (entA/entB/entP or entA/entB) strains with strong antilisterial CFS activity, whereas E. faecium 1X (entA), E. durans 2A, 2B, 2C, and 2X (entA/entP or entP), E. faecalis 3B, and E. hirae 4A (entA) biotypes displayed direct in vitro antilisterial activity only. Biotypes 1D, 1X, and 2A were selected in TM. All E. faecium/durans isolates were susceptible to vancomycin, but the m-Ent + E. faecium biotype 1A and 1D strains were resistant to penicillin, erythromycin, ciprofloxacin, and ampicillin. In contrast, all biotype 1X isolates were susceptible to all antibiotics tested. All E. faecalis and most E. durans isolates were resistant to penicillin but susceptible to erythromycin and ciprofloxacin. Biotype 2X isolates and one virulent (ace; gelE) E. faecalis isolate from RM were tetracycline-resistant. A sporadic RM isolate of E. hirae that was resistant to penicillin and vancomycin was not retrieved from the counterpart TM, but the inclusion of three vancomycin-resistant isolates from TM in the primary biotype 3B of E. faecalis was a cause for concern. In conclusion, based on the results, antibiotic-susceptible representatives of all strain biotypes of the E. faecium/durans group, as well as antagonistic m-Ent+ E. faecium strains from sheep milk that were susceptible to vancomycin and ampicillin and lacking virulence genes, can be included in safe complex natural starters to be developed for onsite use in traditional Greek hard cheese technologies.

1. Introduction

The development, maintenance, and successive reproduction of natural starter cultures (NSCs) in the dairy industry have gained a strong global research interest in recent years [1,2]. Safe NSCs are particularly needed and applied in advanced traditional cheese manufacturing technologies to supplement or replace the commercial starter cultures (CSCs), which require frequent rotation primarily due to bacteriophage infections [3,4,5]. Additionally, NSCs are characterized by high biodiversity and strong adaptation to the raw food (milk) material to be fermented [6,7]. These traits offer technological advantages regarding the evolutionary complexity and complementary activities of the autochthonous (indigenous) microbial species and strains in traditional NSC-mediated milk fermentations [1,2,7,8,9,10]. For these reasons, NSCs are historically associated with artisanal cheese technologies and have an empirically transmitted preparation and use in the form of ‘selective natural concentrates’ of beneficial lactic acid bacteria (LAB) strains, i.e., as natural yogurt starters [10,11] or natural whey starters [12,13] mainly comprising thermophilic dairy streptococci and lactobacilli added to the milk to assure an optimal fermentation and curd acidification [1]. In principle, an NSC represents a complex microbial consortium consisting of an undefined number of thermophilic and mesophilic LAB and non-LAB species and strains naturally selected and coexisting in equilibrium during fermentation [14,15,16].

In practice, the complex microbiota of raw milk represents the most diversified and best performing NSC [17,18], reflected in the global production of excellent traditional raw milk cheeses with peculiar sensorial qualities and health-promoting properties [7,19,20]. Consequently, recovery of the entire beneficial microbial/LAB association from raw milk is needed to develop the most suitable NSCs for preserving artisan cheese qualities. However, raw milk often contains various undesirable, spoilage or pathogenic bacterial contaminants [21,22], which may cause serious deviations in artisan cheese fermentations or lead to faulty or unsafe final products [23,24], unless the milk is industrially (72 °C, 15 s) or traditionally pasteurized (open-batch at 60 °C, 30 min, or at 68 °C, 10 min) or, at least, thermized (57–68 °C for 0–60 s heat holding-time) before cheese processing [1,17,25]. However, apart from the primary non-LAB spoilers and pathogens, a major part of the dominant mesophilic raw milk LAB biota, comprising mainly Lactococcus and Leuconostoc species, is inactivated even by mild thermization treatments, which select for non-starter LAB (NSLAB) with an inherent heat resistance, primarily thermoduric Enterococcus and Streptococcus species amongst the most frequent raw milk contaminants [26,27]. This natural selection of enterococci and animal streptococci in thermized milks and/or craft-made NSCs should be a concern because of the controversial role in (dairy) foods of the above LAB genera [28,29], which include certain species or strains able to cause healthcare-associated infections [30,31].

Enterococci are common inhabitants of the human and animal gastrointestinal tract and thus persist in clinical and farm environments and are transferred to raw milk mainly due to deficiencies in milking hygiene, as well as from various contamination sources (i.e., animal skin, feces, water, etc.) on farms [28,29]. The genus Enterococcus, particularly the species E. faecium and E. faecalis, comprises numerous human pathogenic strains characterized by two alarming traits: harboring multiple antibiotic resistance and/or virulence genes and the ability to exchange them [32,33]. On the other hand, autochthonous dairy enterococci, mainly of the E. faecium/durans group, without virulence traits but instead capable of exerting desirable technological, bioprotective, or probiotic activities in situ, prevail in artisan cheeses and NSCs, having a long history of safe use [6,7,9,20,28,29]. However, despite their beneficial role in traditional cheese fermentations, no Enterococcus species is included in the list of QPS (qualified presumption of safety) [34] or GRAS microbes [29]; a safety evaluation at the strain level is a prerequisite for their potential use in (dairy) foods according to the current EFSA recommendations [34].

Alternate strategies are, therefore, required in the development and maintenance of NSCs with desirable technological traits to keep beneficial enterococci and other autochthonous LAB species and strains in balance in local dairy plants that are aiming to preserve artisan cheese-making practices [35,36]. On the need to remove the undesirable non-LAB biota but restore the entire LAB biota of raw milk without applying heating, an interesting strategy was described by Chessa et al. [35]. Raw Sardinian sheep milk aliquots were cultured anaerobically on acidified whey agar, and the total colony lawns were transferred to sterile whey for culture enrichment, followed by two successive transfers to sterile sheep milk at 37 °C with continuous acidification monitoring; the final coagulum, checked to be free of coliform bacteria, was freeze-dried to obtain a novel NSC [35]. The LAB diversity and safety of that complex NSC was assessed in a follow-up study, which revealed the presence of 33 distinct genotypes assigned to nine species, including E. faecium, E. durans, E. faecalis, and an additional five non-QPS thermophilic Streptococcus species. Although further molecular and antibiotic resistance investigations revealed that all the non-QPS strains could be considered safe in view of using that NSC for cheesemaking [37], a constantly safe outcome of the above strategy can neither be guaranteed nor generalized. Different raw sheep milks or other milk types may contain virulent or multidrug-resistant (MDR) Enterococcus, Streptococcus, or other non-QPS LAB strain contaminants [38,39,40,41], which could be highly enriched following treatment of the maternal milk in the above [35] or a similar manner, rendering the final NSC unsafe for use.

A recent molecular identification study by Sioziou et al. [42] conducted to retrieve promising autochthonous strains from Greek raw sheep milk provided consistent data, i.e., most of the non-QPS LAB species of the above NSC derived from Sardinian raw sheep milk [35] also occurred in two raw sheep milk batches from native Epirus breeds displaying direct in situ antilisterial activity [42]. Consistently, several enterocin-producing E. faecium/durans strains, lacking virulence traits and coexisting with E. faecalis strains harboring ace and/or gelE genes, were retrieved as subdominant members of the indigenous LAB biota of the above raw milk batches, which were dominated by Leuconostoc mesenteroides and pyogenic streptococci [42]. Next, Samelis et al. [43] assessed the selective effects of mild thermization (65 °C, 30 s) on the LAB biota of this raw sheep milk type and confirmed that enterococci (68.6%) were highly selected in all counterpart thermized milks, followed by thermophilic streptococci (8.6%), whereas the predominance of leuconostocs (48.4%) in raw milks was reduced to 5.7% post-thermally. Because the antibiotic resistance (AR) at the strain level of the above autochthonous LAB has yet to be assessed [42,43], this follow-up study was undertaken to determine the primary antagonistic Enterococcus strain biotypes prevailing in antilisterial sheep milks before and after thermization, and to compare their enzymatic activity profiles and primary ARs in vitro. These are basic traits for the selection of safe representatives from all beneficial Enterococcus biotypes to be combined with other LAB species in a novel NSC, with the aim to restore the original raw Epirus sheep milk biodiversity, excluding all virulent or MDR, non-QPS strain biotypes, during cheese fermentation and ripening.

2. Materials and Methods

2.1. Sheep Milk Isolates, Reference-Control Strains, and Culture Conditions

Nineteen autochthonous Enterococcus spp. previously isolated from two ‘winter’ raw milk batches (RM1, RM2) and collected from two yards with mixed native sheep breeds, Karamaniko and Karagouniko, crossbred in the area of Arta, Epirus, Greece, and identified at the species level by molecular methods [42], were included in this study (

Table 1. Molecular identification and genotypic antagonistic and safety characteristics of 19 raw milk (RM) Enterococcus isolates from native sheep breeds included in this study ¹.

Species	No of Isolates	Molecular Id-Method/s	Antilisterial Enterocin Activity			Safety Characteristics
			M17 agar Overlay 2	CFS/Well Assay 2	Bac-gene/s Possessed 3	Hemolytic Activity	Cytolysin vanA/vanB	Virulence gene/s 4
Enterococcus faecium	6	IGS/16S rRNA	++ (3) + (3)	++ (3) ng (3)	entA/entB/entP (3) entA (3)	A	No/No	None
Enterococcus durans	9	IGS/16S rRNA	+ (1) +w (2) ng (6)	ng (9)	entA/entP (1) entP (2) None (6)	A	No/No	None
Enterococcus hirae	1	IGS/16S rRNA	+w (1)	ng (1)	entA (1)	A	No/No	None
Enterococcus faecalis	3	IGS	ng (3)	ng (3)	None	A	No/No	ace/gelE (1) ace (1), gelE (1)

¹ All data in Table 1 are adapted from Sioziou et al. [42]. ² ++, strong inhibition; +, good inhibition; +w, weak inhibition; ng, negative. ³ All isolates were tested for possession of the enterocin A, B, P, 50A/50B, Bac31, and AS-48 genes, plus the cytolysin gene strongly associated with virulent strains. ⁴ All isolates were tested for possession of the virulence genes agg, ace, espA, hyl, IS16, and gelE.

). Because both RM1 and RM2 batches displayed strong in situ, bacteriocin-mediated antilisterial activity in well diffusion assays following incubation at 37 °C for 48 h, the above selected isolates, coded with the prefix KFM, were further tested by Sioziou et al. [42] for structural enterocin-encoding genes and primary safety traits (i.e., hemolytic activity, cytolysin, vancomycin resistance, and virulence genes). The corresponding previous findings for each species are summarized in Table 1.

Because both RM batches retained their strong in situ antilisterial activity after mild thermization (65 °C; for 30 s holding-time) in a local dairy plant (Skarfi E.P.E., Pappas Bros. Traditional Dairy, Filippiada, Epirus), an additional 41 Enterococcus spp. isolates representing 83.7% of the total LAB survivors retrieved from the counterpart thermized milk (TM1 and TM2) batches, coded with the prefix KTM [43], were included herein, too, to be assessed comparatively with the respective RM isolates listed in Table 1. Amongst them, four Enterococcus isolates (KTM46, KTM47, KTM49, and KTM55) from batch TM2 displayed strong, enterocin-mediated antilisterial CFS activity in well diffusion assays, as illustrated by Samelis et al. [43].

Additionally, 12 reference clinical or virulent and non-virulent (safe) dairy strains of the species E. faecium, E. durans, and E. faecalis [44,45] were used as controls, including E. faecalis ATCC 29212^TM, a quality control strain for antibiotic resistance tests [46]. The origin and the primary traits of the control strains are supplemented in Table S1.

All sheep milk Enterococcus isolates and the control strains were resuscitated by transferring 0.1 mL of frozen (−30 °C) working stock cultures with 20% (w/v) glycerol, in 10 mL MRS broth (Neogen Culture Media, Lab M, Heywood, UK), incubated at 30 °C for 24 h, and subcultured twice, as above. Afterward, the second culture of each isolate or reference strain was streaked for growth on MRS agar (Neogen) or M17 agar (Biolife, Italiana, S.r.l., Milano, Italy) at 30 °C or 37 °C for 48–72 h, to check colony growth appearance and ensure purity by transferring one single colony to new 10 mL MRS or M17 tubes, incubated as above, for use in the experiments, as appropriate for each test.

2.2. Biochemical Identification and Biotyping of the Sheep Milk Isolates

All sheep milk isolates and the control strains were first rechecked rapidly for their Gram-positive and catalase-negative reactions, and then for basic phenotypic traits (microscopic appearance, CO₂ production from glucose; NH₃ from arginine hydrolysis; growth at 10 °C and 45 °C; growth in 4% and 6.5% salt, and on KAA agar), according to Samelis et al. [43], to confirm their assignment to the genus Enterococcus. Next, all 60 isolates and the control strains in Table S1 were tested for the fermentation of 13 basic (key taxonomic) sugars, L-arabinose, cellobiose, galactose, lactose, maltose, mannitol, melibiose, raffinose, ribose, sorbitol, sucrose, trehalose, and xylose (Sigma-Aldrich Chemie GmbH, Steinheim, Germany), in 96-well sterile mini-plates in order to identify them biochemically at the species level and obtain the basic strain biotypes. The biochemical identification was primarily based on the main sugar fermentation profiles, using the phenotypic criteria tabulated for Enterococcus spp. in the Bergey’s Manual of Systematic Bacteriology [47]. In particular, the sheep milk isolates were distinguished into different strain biotypes within each Enterococcus species, according to the criteria and the biotype coding list described by Vandera et al. [10]; the respective biotypes of the control Enterococcus strains are included in Table S1.

Next, selected isolates of each Enterococcus species and strain biotype were tested for their entire sugar fermentation patterns and/or enzymatic activity profiles with the API 50 CHL, API 20 STREP, and API ZYM kits, according to the manufacturer’s instructions (BioMerieux, Marcy l’ Etoile, France). Strain selection was based on the polyphasic molecular identification and safety evaluation conducted by Sioziou et al. [42] in association with the basic strain biotype profiles obtained during this study (viz. Results). Reference Enterococcus strains with identical basic biotypes (Table S1) served as controls. Particularly for the API 20 STREP analysis, a positive or negative β-hemolysis reaction of each tested strain was required as an additional external test to generate the 7-digit API identification code. Except for the three virulent Enterococcus strains in Table S1, the remaining control strains and all sheep milk isolates were not β-hemolytic [42,43]; therefore, the hemolysis test was not repeated.

Also, all 19 Enterococcus isolates from the two antilisterial RM batches (Table 1), as well as all 41 respective isolates from the counterpart TM batches included in this study, were previously assayed for direct antilisterial activity by the M17 agar overlay technique followed by well assay testing of the enterocin-mediated CFS activity of the positive isolates, as reported and illustrated by Samelis et al. [43]. Accordingly, in this study, the structural enterocin genes possessed by the antagonistic Enterococcus TM isolates were detected by PCR, as described for the respective RM isolates by Sioziou et al. [42].

2.3. Antibiotic Susceptibility of the Sheep Milk Isolates

All sheep milk isolates were assessed for susceptibility to eight common antibiotics amongst those specified by the Clinical and Laboratory Standards Institute (CLSI) [48] for enterococci. Specifically, all 60 Enterococcus isolates, and the control strains in Table S1, were screened by the Kirby–Bauer disk diffusion method [49] for resistance to the following(μg per disk): ampicillin (AMP; 10), chloramphenicol (CHL; 30), ciprofloxacin (CIP; 5), erythromycin (ERY; 15), gentamicin (GEN; 10), penicillin G (PEN; 10 units per disk), tetracycline (TET; 30), and vancomycin (VAN; 30), as described by Tsanasidou et al. [45]. Briefly, commercial antibiotic disks (BioMaxima SA, Lublin, Poland) were used. Two M17 agar plates were spread-inoculated with 100 μL of fresh (24 h) M17 broth cultures of each tested strain; the inocula were left to absorb for 5–10 min, and then 4 different antibiotic disks were placed over the cell lawn on each agar plate. All plates were incubated at 37 °C overnight. On the next morning, the diameters of the inhibition zones formed around each antibiotic disk, including the diameter (6 mm) of the disk, according to CLSI [48], were measured with a micrometer (model D15, Mitutoyo, Kanagawa, Japan). The above process was repeated using new, fresh strains’ cultures, and the two measurements were averaged. M17 agar was applied because enterococci grow abundantly on/in this medium containing lactose (i.e., the main sugar fermented by LAB in milk and dairy products), instead of glucose. However, to account for potential media effects on the size and clearness of the inhibition zones, the antibiotic susceptibility test was repeated, as above, by seeding the fresh cell culture lawn of several strains on Müller–Hinton agar (Neogen) plates, compared to the M17 agar plates. The results were interpreted according to the breakpoints recommended by CLSI for enterococci [48]. E. faecium 315VR (vancomycin-resistant) and E. faecalis ATCC 29212^TM (vancomycin-sensitive QC strain) in Table S1 were the primary reference strains used as positive and negative controls, respectively. The EFSA guidance on the safety assessment of E. faecium was followed [50].

3. Results

3.1. Comparative Species Identification and Biotyping of the Enterococcus Isolates Prevailing in Antilisterial Raw Milks and Surviving in the Same Milks After Thermization

Table 2. Differentiation of 60 Enterococcus isolates from two counterpart raw (RM) and thermized (TM) sheep milks into 4 species and 14 strain biotypes.

Species	Biotype	RM Isolates	TM Isolates	Total Isolates	Isolate Code Number 1	Reference Strain
Enterococccus faecium	1A	3	1	4	KFM17 **, KFM28 **, KFM29 **, KTM21	KE118 **
	1D		6	6	KTM46 **, KTM47 **, KTM50, KTM53, KTM54, KTM55 **	KE82 **, KE85
	1E		2	2	KTM20, KTM37	KE86
	1G		1	1	KTM7
	1H		1	1	KTM49 **	KE118 ** (1A/1H)
	1X	3	11	14	KFM57 *, KFM58 *, KFM59 *, KTM3, KTM6, KTM10, KTM11, KTM12, KTM32, KTM33, KTM36, KTM40, KTM41, KTM52	KE77 *
Enterococcus durans	2A		5	5	KTM19, KTM25, KTM38, KTM45 *, KTM48 *	KE100 *
	2B	5	2	7	KFM16, KFM18, KFM20, KFM27, KFM30, KTM24, KTM31 *	KE108 *
	2C	2	1	3	KFM6 *, KFM19, KTM17	KE106
	2D		4	4	KTM23, KTM42, KTM43, KTM44	KE66
	2X	2		2	KFM46 *, KFM49 *
Enterococcus faecalis	3B	2	6	8	KFM47, KFM50, KTM13, KTM22, KTM34 *, KTM35 *, KTM39, KTM51	GL322 *
	3D	1		1	KFM48
Enterococcus hirae	4A	1	1	2	KFM56 *, KTM4

¹ Strain codes previously identified by 16S rRNA gene sequencing and/or IGS are written in bold; strain codes bearing an asterisk showed moderate-to-strong, direct enterocin-mediated antilisterial activity in the M17 agar overlays; strain codes bearing a double asterisk gave CFS with strong enterocin activity in the well diffusion assays. The reference (control) strain KE118 was confirmed to be intermediate of the typical biotype 1A and the new atypical biotype 1H of E. faecium (viz. Section 3.1), in accordance with its API 50 CHL profile [44].

summarizes the species identification and biotyping of the 19 autochthonous Enterococcus isolates from the two antilisterial ‘winter’ RM1 and RM2 batches (prefixed KFM) compared to the 41 Enterococcus survivors isolated from the counterpart TM1 and TM2 batches (prefixed KTM). All isolates were characterized and identified correspondingly to the previous molecular species identifications of the 19 KFM isolates from RM (i.e., bold-written in Table 2), followed by biotyping at the strain level of the 60 typical Enterococcus spp. isolates (i.e., all of them were homofermentative, arginine-positive cocci, able to grow at 10 °C and 45 °C in 6.5% salt, and abundantly on KAA agar) assigned to the E. faecium/durans/hirae group or E. faecalis (Figure 1). As mentioned, biotyping was performed correspondingly to the strain biotypes described by Vandera et al. [10] for 113 autochthonous E. faecium, E. durans, and E. faecalis isolates from naturally fermented and ripened Graviera cheeses. The basic sugar fermentation profiles of all Enterococcus strain biotypes are presented in Figure 1. Within each different species or strain biotype, the isolates that displayed direct antilisterial activity in the M17 agar overlay assays bear an asterisk in Table 2, with those retaining a strong enterocin-mediated activity in their CFS bearing a double asterisk; the structural enterocin genes detected by PCR in the genome of all antilisterial isolates are summarized in

Table 3. Antilisterial activity of Enterococcus sheep milk isolates detected by the agar overlay technique, and confirmation of the presence of active enterocin/s in their cell-free supernatants (CFS) linked with the PCR detection of enterocin genes in their genome.

Species	Biotype	M17 Agar Overlay 1	CFS/Well Assay 2,3	Enterocin Gene/s Detected by PCR 4,5				Sheep Milk Isolates with the Same Ent+ Profile 6
				entA	entB	entP	bac31
E. faecium	1A	++	++	+	+	+	−	KFM17, KFM28, KFM29
	1D	++	++	+	+	+	−	KTM46, KTM47
	1D	++	++	+	−	−	−	KTM55
	1H	++	++	+	+	−	−	KTM49
	1X	+	−	+	−	−	−	KFM57, KFM58, KFM59
E. durans	2A	+	−	−	−	+	−	KTM45, KTM48
	2B	+	−	+	−	+	+	KTM31
	2C	+	−	+	−	+	−	KFM6
	2X	(+)	−	−	−	+	−	KFM46, KFM49
E. faecalis	3B	+	−	−	−	−	−	KTM34, KTM35
E. hirae	4A	(+)/−	−	+	−	−	−	KFM56

¹ ++, strong clearness zone/inhibition of L. monocytogenes; +, moderate clearness zone; (+) faint clearness zone; −, no clearness zone around the streaked colonies. ² ++, large inhibition zone (>5–15 mm depending on the CFS (MRS or M17) medium; −, no inhibition zone. ³ The neutralized CFS lost their antilisterial activity after treatment with 1 mg/mL of proteinase K, α-chymotrypsin, or trypsin. ⁴ +, presence of the enterocin gene tested; −, absence of the enterocin gene tested. ⁵ None of the strains possessed the enterocin genes ent50A, ent50B, entAS-48, or cytolysin (CytL_L). ⁶ All data in Table 3 for the KFM isolates are adapted from Sioziou et al. [42] to compare with the KTM isolates assayed in this study.

.

So, according to their basic/key sugar fermentation profiles (Figure 1), the 60 typical (arginine-positive) Enterococcus sheep milk isolates were differentiated into six E. faecium, five E. durans, and two E. faecalis strain biotypes, with each species comprising a total of 28, 21, and 9 strains, respectively (Table 2). Additionally, two sporadic isolates, one from RM (KFM56) and the other from TM (KTM4), were assigned to E. hirae (Table 2) and shared the biotype 4A strain profile (Figure 1).

All E. faecium isolates were L-arabinose positive and formed two typical (1A and 1D) and four atypical (1E, 1G, 1H, and 1X) strain biotypes. Surprisingly, no typical biotype 1B or 1C or atypical biotype 1F isolates of E. faecium [10] were detected in sheep milk before or after thermization (Figure 1). Instead, two additional atypical E. faecium strain biotypes, 1H and 1X, unable to ferment trehalose, were isolated (Table 2, Figure 1). Biotype 1H included one single m-Ent+ (entA/entB) E. faecium strain (KTM49) (Table 3), isolated from TM (Table 2), which had the same basic fermentation profile as the m-Ent+ (entA/entB/entP) E. faecium KE118 control strain (biotype 1A), except for its trehalose-negative reaction (Table S1). However, KE118, which was originally tabulated to represent the typical E. faecium biotype 1A isolates from Graviera cheese [10], was later reassessed to ferment trehalose very weakly, or not at all, in the API 50 CH strip [44]. Therefore, KE118 is an E. faecium strain with an ‘intermediate’ 1A/1H biotype and probably a closer relative of the KTM49 strain (biotype 1H) rather than of the m-Ent+ (entA/entB/entP) E. faecium KFM17, KFM28, and KFM29 isolates, which are typical trehalose-positive biotype 1A strains displaying a strong antilisterial activity as well (Table 2 and Table 3).

Meanwhile, the 14 most numerous isolates in Table 2 represent another distinct trehalose-negative, but melibiose-positive, E. faecium biotype 1X (Figure 1), which was frequent in RM from native Epirus sheep breeds, comprised weakly antilisterial (entA) strains, i.e., KFM57 to KFM59 [42] (Table 3), and, most importantly, showed the highest isolation frequency in both TM1 and TM2 batches (Table 2). The latter finding suggests that E. faecium biotype 1X strains may have a pronounced ability to survive thermization but may be outcompeted during cheese fermentation by more antagonistic biotype 1A, 1B, and 1D strains [10].

All E. durans isolates from RM or TM were assigned to the typical biotypes 2A to 2D [10], which comprise strains negative for all five key sugars, L-arabinose, mannitol, sorbitol, raffinose, and sucrose (Figure 1). Two novel atypical, mannitol-positive E. durans isolates retrieved from the RM2/TM2 samples were assigned to biotype 2X and fermented sucrose, melibiose, and trehalose. Of note, neither the biotype 4A isolates (Figure 1), molecularly identified as E. hirae [42], another mannitol-negative species of the E. faecium/durans group [47], were previously detected in mature Graviera cheese [10]. Also, none of the E. faecalis sheep milk isolates were assigned to biotype 3A, the only typical, melibiose-positive E. faecalis strains found to be subdominant of the E. faecium/durans in mature Graviera cheese [10]. Instead, all of them, except of the single RM isolate KFM48, belonged to another typical E. faecalis biotype 3B, which comprises strains unable to produce acid from melibiose (Figure 1), like the GL322 control strain isolated from artisan Galotyri cheese (Table 2; Table S1). Strain KFM48 was a strange isolate assigned to the new and most atypical E. faecalis biotype 3D. It fermented all 13 basic (key) sugars in Figure 1, including L-arabinose, which is not fermented by E. faecalis [47], and xylose, which is fermented by the E. gallinarum group, rarely by E. faecium, but not by E. faecalis [47].

Notably, strain KFM48 was found to possess the gelE gene and thus may represent a virulent, opportunistically pathogenic E. faecalis genotype with a unique phenotype, although it is α-hemolytic and lacks cytolysin [42].

The API 50 CHL sugar fermentation patterns of the most promising antilisterial Ent+ sheep milk isolates were tested in comparison with the corresponding controls because the basic biotypes in Figure 1 may include Enterococcus strains differing in secondary fermentation and/or metabolic (enzymatic) reactions. Indeed, based on the API results in Table S2, at least the primary E. faecium biotype 1D is heterogeneous. Specifically, the patterns of the m-Ent+ (entA/entB/entP) TM isolates KTM46 and KTM47 and the KE82 control strain match but differ from the TM isolate KTM55 and the other control strain GL31, which are two different strains possessing entA only (Tables S1, S2, and Table 3). Conversely, the m-Ent+ KTM49 TM strain (1H) and the KE118 control (1A/1H) strain possess the same API 50 CH sugar fermentation pattern but differ in their enterocin gene profile; both are trehalose-negative variants of the biotype 1A (KFM28 and KFM29) strains. All of them share the ability to ferment gentibiose, D-tagatose, and gluconate. E. faecium displays high strain-dependent variability in the above three sugars (Table S2), as also shown by the antagonistic E. faecium isolates from fermented TM distinguished into biotypes I to III previously [51]. The biotype 3B E. faecalis isolates and the control GL322 fermented gentibiose, D-tagatose, and gluconate, too (Table S2). However, apart from their basic fermentation differences (Figure 1), all E. faecalis biotypes produced acid from glycerol and melezitose, unlike all E. faecium biotypes (Table S2). Lastly, the three virulent Enterococcus strains displayed unique API 50 CHL profiles. As anticipated, the CLSI QC strain ATCC 29212^TM was L-arabinose negative and produced acid from most sugars typically fermented by E. faecalis, including starch; however, to our surprise, it was lactose-negative. In contrast, the β-hemolytic and cytolysin-positive E. faecalis GL320 strain (biotype 3C) fermented lactose, but it was unable to ferment sucrose, which is typically fermented by all E. faecalis, including all sheep milk isolates and the safe strain GL322 (biotype 3B). The virulent E. faecium 315VR possessed the most enriched sugar fermentation profile; it was negative with rhamnose and starch only; moreover, it was the only strain with a strong positive methyl-α-D-mannopyranoside reaction (Table S2).

Additionally, 13 of the Enterococcus KTM isolates from TM were characterized and validated further by the API20 STREP identification method (

Table 4. API20 STREP-based identification and strain biotyping of selected Enterococcus spp. isolates from raw or thermized sheep milk of native Epirus breeds ¹.

Test	Reaction/Enzymes	Enterococcus faecium				Enterococcus faecalis	Enterococcus hirae
		1D	1G	1H	1X	3B	4A
VP	Acetoin production	+	+	+	+	+	+
HIP	Hydrolysis (Hipuric acid)	−	−	−	−	−	−
ESC	β-glucosidase hydrolysis	+	+	+	+	−	+
PYRA	Pyrolidonyl arylamidase	+	+	+	+	+	+
αGAL	α-galactosidase	−	−	−	−	−	−
βGUR	β-glucuronidase	−	−	−	−	−	−
βGAL	β-galactosidase	+	+	+	+/− *	−	+
PAL	Alkaline phosphatase	−	−	−	−	−	−
LAP	Leucine aminopeptidase	+	+	+	+	+	+
ADH	Arginine dihydrolase	+	+	+	+	+	+
RIB	D-ribose (acidification)	+	+	+	+	+	+
ARA	L-arabinose (acidification)	+	+	+	+	−	−
MAN	D-mannitol (acidification)	+	−	+	+/(+)	+	−
SOR	D-sorbitol (acidification)	−	−	−	−	+	−
LAC	D-lactose (acidification)	+	+	+	+	+	+
TRE	D-trehalose (acidification)	+	−	−	−	+	+
INU	Inulin (acidification)	−	−	−	−	−	−
RAF	D-raffinose (acidification)	−	−	−	−	−	−
AMD	Starch (acidification)	−	−	−	−	+	−
GLYG	Glycogen (acidification)	−	−	−	−	−	−
βHEM	β-hemolysis	−	−	−	−	−	−
API code		5157510	5157400	5157500	5157500 5147500 *	1143711	5153410
API-ID		E. faecium 1 Very good	E. durans Good	E. faecium Good	E. faecium Good	E. faecalis Good	E. durans Good

¹ The isolates of each biotype were as follows: 1D/KTM46, KTM47, KTM55; 1G/KTM7; 1H/KTM49; 1X/KTM3, KTM6, KTM10, KTM11, KTM12, KTM36 *; 3B/KTM51; and 4A/KTM4. * KTM36 was the only β-galactosidase-negative isolate among the 1X isolates.

).

They were selected as representatives of the E. faecium biotypes 1D (three isolates), 1G (one), 1H (one), and 1X (six), E. faecalis biotype 3B (one), and E. hirae 4A (one), either because they were promising Ent+ or m-Ent+ antilisterial strains (1D, 1H), or sporadic TM isolates (1G, 4A), or their prevalence and thus biotechnological significance as NSLAB increased in TM compared to RM (biotypes 1X and 3B in Table 2). The corresponding results, which are summarized in Table 4, showed that the API20 STREP identification of all Enterococcus biotype representatives from TM was from good to very good, matching the genomic identification of the corresponding RM isolates of the E. faecium or E. faecalis biotypes sequenced by Sioziou et al. [42]. The only exceptions were the atypical mannitol and trehalose-negative strain KTM7 in the E. faecium biotype 1G and the mannitol-negative E. hirae KTM4 isolate, which both were identified as E. durans (Table 4). This is not surprising because the above two species are very close relatives in the E. faecium/durans genomic group [47]. Otherwise, the reliability of the strain biotyping data was high; all key taxonomic sugar fermentation reactions of all Enterococcus isolates in Table 4 matched those obtained by the conventional mini-plate (Figure 1) and the API50 CHL (Table S2) methods. A few inconsistencies occurred only between certain enzymatic reactions of the Enterococcus isolates included in the API 20 STREP strip and those obtained for the same isolates by the API ZYM method, which are specified and discussed below.

3.2. Species-Dependent or Strain Biotype-Dependent Differences of the Enterococcus Sheep Milk Isolates in Their Enzymatic Activity Profiles

The API ZYM semi-quantitative micro-method was used to assay for constitutive enzymes in the washed cell biomass of 26 selected sheep milk Enterococcus isolates pre-cultured for 24 h in MRS broth at 30 °C. Their enzymatic activity grading was expressed as before [45,52,53]. We focused on the most promising Ent+ and m-Ent+ strain biotypes lacking virulence traits.

Table 5. Enzymatic activity (API ZYM) profiles of selected antilisterial Enterococcus strains isolated from sheep milk of native Epirus breeds before and after thermization.

Species	Biotype	Enzymes Tested 1
		1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17	18	19
E. faecium
KE 118 **	1A	20	20	20	0	20	10	30	0	0	40	30	0	0	0	0	0	0	0	0
KFM 28 **		20	30	20	0	40	20	40	0	0	40	30	0	30	0	0	0	0	0	0
KFM 29 **		10	20	30	0	40	30	30	0	10	40	30	0	30	0	0	0	0	0	0
KE 64 **	1B	10	20	20	0	20	20	20	0	5	30	20	0	10	0	0	0	0	0	0
KE 85	1D	20	30	20	0	20	0	0	0	0	10	0	0	0	0	0	0	0	0	0
KE 82 **		20	30	30	0	40	20	30	0	30	40	20	0	0	0	0	0	0	0	0
KTM 46 **		10	30	20	0	40	20	40	0	0	40	20	0	40	0	0	0	0	0	0
KTM 47 **		10	20	20	0	40	10	30	0	5	30	20	0	30	0	0	0	0	0	0
KTM 54		30	30	20	0	20	10	40	0	0	40	20	0	30	0	0	0	0	0	0
KTM 55 **		10	20	20	0	30	20	40	0	5	40	30	0	20	0	0	0	0	0	0
KTM 37	1E	10	30	20	0	30	10	30	0	0	40	30	0	20	0	0	0	0	0	0
KTM 7	1G	10	20	20	0	40	30	30	0	10	40	20	0	0	0	0	0	0	0	0
KTM 49 **	1H	10	30	20	0	30	5	40	0	0	40	20	0	40	0	0	0	0	0	0
KE 77 *	1X	30	20	20	0	30	10	20	0	10	30	20	10	0	0	10	0	0	0	0
KTM 32		10	30	20	0	40	20	30	0	0	40	40	0	30	0	0	0	0	0	0
KFM 57 *		0	30	20	0	40	20	20	0	5	40	40	0	0	0	0	0	0	0	0
KFM 58 *		10	20	20	0	40	10	20	0	30	40	40	0	0	0	0	0	0	0	0
E. durans
KE 100 *	2A	20	30	30	0	40	30	30	0	0	40	20	0	0	0	0	0	0	0	0
KTM 45 *		20	20	30	0	40	30	30	0	0	40	20	0	0	0	0	0	0	0	0
KTM 48 *		20	30	30	0	40	30	30	0	0	40	20	0	0	0	0	0	0	0	0
KE 108 *	2B	20	20	30	0	30	30	30	0	0	40	20	0	0	0	0	0	0	0	0
KFM 20		10	30	30	0	40	40	30	0	5	40	20	0	0	0	0	0	0	0	0
KTM 31 *		20	20	20	0	30	30	30	0	0	40	20	0	0	0	0	0	0	0	0
KFM 6 *	2C	20	30	30	0	40	30	40	0	30	40	20	0	0	0	0	0	0	0	0
KTM 42	2D	5	30	30	0	40	30	40	0	0	40	20	0	0	0	0	0	0	0	0
KFM 46 *	2X	10	10	30	0	40	40	30	0	5	40	30	0	0	0	0	0	0	0	0
KFM 49 *		30	30	30	0	40	40	30	0	5	40	30	0	0	0	0	0	0	0	0
E. faecalis
GL 322 *	3B	20	30	30	0	40	20	30	0	40	40	30	0	0	0	40	30	0	0	0
KTM 34 *		20	30	30	0	40	10	30	0	30	40	40	0	0	0	40	0	0	0	0
KTM 35 *		20	30	30	0	40	10	20	0	30	40	40	0	0	0	40	0	0	0	0
KTM 39		10	30	20	0	10	5	5	0	5	40	40	0	0	0	10	0	0	0	0
KTM 51		10	30	20	0	10	5	5	0	5	30	40	0	0	0	5	0	0	0	0
GL 320 **	3C	10	30	30	0	30	0	5	0	40	20	40	0	20	0	10	0	0	0	0
KFM 48	3D	10	20	20	0	20	0	0	0	30	40	40	0	0	0	0	0	0	0	0
E. hirae
KFM 56 *	4A	0	20	20	0	30	0	0	0	20	40	30	0	0	0	0	0	0	0	0

¹ Enzymes: 1—Alkaline phosphatase; 2—Esterase (C 4); 3—Esterase Lipase (C 8); 4—Lipase (C 14); 5—Leucine arylamidase; 6—Valine arylamidase; 7—Cystine arylamidase; 8—Trypsin; 9—α-chymotrypsin; 10—Acid phosphatase; 11—Naphthol-AS-BI-Phosphohydrolase; 12—α-galactosidase; 13—β-galactosidase; 14—β-glucuronidase; 15—α-glucosidase; 16—β-glucosidase; 17—N-acetyl-β-glucosaminidase; 18—α-mannosidase; and 19—α-fucosidase. The reactions were graded from 0 (negative) to 5 (strongly positive) based on the change in the color intensity after the addition of the ZYM A+ZYM B reagents in each well. Activity (approximate values) was expressed as arbitrary units of substrate hydrolyzed; 0: negative; 1: liberation of 5 nmol; 2: 10 nmol; 3: 20 nmol; 4: 30 nmol; 5: 40 nmol or more [52,53]. Reference/control strains within each biotype are indicated in bold on the species column. All strains with direct (*) or CFS (**) antilisterial activity bear a single or double asterisk, as explained in Table 2.

presents their API-ZYM profiles compared with the respective profiles of the control strains in each biotype, if available. Particularly for the E. faecium biotype 1D, two different control strains, KE85 and KE82 (Tables S1 and S2), were used. The EntA+ E. faecium KE64 (biotype 1B) and the cytolysin-positive E. faecalis GL320 (biotype 3C) strains were included as controls for comparison only, because no biotype 1B or 3C sheep milk isolates were detected (Figure 1). Major similarities, but also major species-dependent or strain-dependent differences, were noted. Regarding similarities, all Enterococcus isolates, including the control strains, showed moderate-to-high esterase (C4) and esterase–lipase (C8) activities, irrespective of strain biotype, whereas none showed lipase (C14) activity. Also, all of them and the control strains displayed strong acid phosphatase and moderate-to-strong napthol-AS-BI-phosphohydrolase activities, except for the enterocin-negative E. faecium KE85 control strain (Table 5).

Additionally, 88.5% (23/26) of the E. faecium isolates and all control strains showed a weakly positive to moderately positive (grade 3; maximum liberation of 20 nmol substrate) alkaline phosphatase activity, except for three different negative strains: E. hirae KFM56 (4A), E. faecium KFM57 (1X), and E. durans KTM42 (2D). However, of the 23 sheep milk isolates considered positive with alkaline phosphatase, 60.9% displayed a weakly colored (grade 2; up to 10 nmol) reaction (Table 5), which, according to the API ZYM manufacturer, is not positive. In agreement, among them, six E. faecium or E. faecalis TM isolates, KTM46, KTM47, KTM55, KTM49, KTM7, and KTM51, were found to be negative for alkaline phosphatase in the API20 STREP kit (Table 4). From the nine clearly positive isolates, according to the API ZYM method, only two, E. durans KFM49 (2X) and E. faecium KTM54 (1D), gave a strong (grade 4; 30 nmol) reaction. Hence, the alkaline phosphatase activity seems to be strictly strain-dependent in enterococci, and, probably, most strains shift to negative, depending on the sensitivity of the analytical method.

With regard to the proteolytic activities, all amino acid (leucine, valine, and cystine) arylamidase reactions were moderate to strong in most Enterococcus sheep milk isolates and the control strains, as was anticipated [45]. Prominent exceptions were, again, the Ent-negative E. faecium KE85 control strain and the E. hirae KFM56, as well as the strange multi-fermenting E. faecalis KFM48 strain in biotype 3D; all were positive with leucine but negative with valine and cystine. The β-hemolytic E. faecalis GL320 control strain in biotype 3C was very similar to KFM48. However, the greatest heterogeneity in proteolytic activity was noted between the biotype 3B isolates of E. faecalis, with the antilisterial strains GL322, KTM34, and KTM35 being more proteolytic than the other two strains tested. In contrast, all E. durans strain biotypes were very homogeneous, displaying the strongest positive reactions with leucine, valine, and cystine. Thus, high amino acid arylamidase activities may be a species-specific trait in E. durans. Conversely, in E. faecium, the proteolytic activities were strain-dependent, with most sheep milk isolates showing weaker positive reactions than the E. durans isolates, particularly with valine (Table 5).

Prominent differences occurred in the α-chymotrypsin activity, too; they appeared to be species-dependent and secondarily strain-dependent. Indeed, both control strains and all sheep milk isolates of E. faecalis possessed strong α-chymotrypsin activity, except for the two non-antagonistic isolates KTM39 and KTM51 in biotype 3B. In contrast, the two control strains and all isolates of E. durans lacked α-chymotrypsin activity, except for the m-Ent+ (entA/entP) KFM6 strain in biotype 2C. Similarly, the sheep milk isolates of E. faecium either were completely lacking or a few of them showed very weak α-chymotrypsin activity. Of note, the m-Ent+ (entA-entB-entP) KE82 control in biotype 1D was the only E. faecium strain highly active for α-chymotrypsin, whereas the m-Ent+ (entA-entB-entP) E. faecium KE118 (1A/1H) and all m-Ent+ E. faecium sheep milk strains, KTM46, KTM47 (1D), and KTM49 (1H), lacked α-chymotrypsin activity. The very high similarity of the KE118 and KTM49 strains (biotype 1H) was reflected by their API ZYM profiles, too (Table 5).

Two strong species-dependent differences between the Enterococcus isolates were noted in the glycolytic enzyme reactions: all E. durans isolates and E. hirae KFM56 displayed negative reactions for all glycolytic enzymes. In contrast, all E. faecium sheep milk isolates in biotypes 1A, 1D, 1E, and 1H, comprising the most antagonistic m-Ent+ isolates, showed strong β-galactosidase activity, despite the fact that the respective control strains derived from Graviera cheese were negative (Table 5). The isolates in biotypes 1G and 1X were also negative for β-galactosidase, except for the KTM32 strain, and all E. faecalis sheep milk and control isolates, except for the virulent β-hemolytic strain GL320. On the other hand, all E. faecalis were positive, or weakly positive, for α-glucosidase, except for the multi-fermenting KFM48 strain (3D). Particularly, the control strain and both antilisterial isolates KTM34 and KTM35 in biotype 3B showed very strong α-glucosidase activity. The following findings are notable: (i) GL322 was the only strain that also showed strong β-glucosidase activity; (ii) no Enterococcus isolate or control strain showed α-galactosidase activity, except for the 1X E. faecium KE77 Graviera cheese strain, which showed weak α-galactosidase and α-glucosidase activities. Lastly, no strain showed trypsin, or β-glucuronidase, N-acetyl-β-glucosaminidase, α-mannosidase, or α-fucosidase activities (Table 5).

3.3. Antibiotic Resistances of the Enterococcus Species and Strain Biotypes Isolated from Sheep Milk of Native Epirus Breeds Before and After Thermization

The results of the ARs of the total 60 Enterococcus sheep milk isolates are split into

Table 6. Antibiotic susceptibility of autochthonous Enterococcus faecalis and Enterococcus hirae strain biotypes isolated from sheep milk of native Epirus breeds before (raw milk; KFM isolates) and after (thermized milk; KTM isolates) thermization in comparison with antibiotic-resistant and/or virulent, β-hemolytic Enterococcus control strains.

Strain Code (Biotype)	Antibiotic Tested (μg/Disk) 1
	AMP 10	CHL 30	CIP 5	ERY 15	GEN 10	PEN10U	TET 30	VAN 30
Control Strains
E. faecium 315VR ** (1F)	R (6.0)	I (16.1)	R (6.0)	R (6.0)	R (6.0)	R (6.0)	R (9.4)	R (7.6)
E. faecalis ATCC 29212TM	S (20.6)	S (22.0)	I (19.1)	I (16.8)	I (9.4)	R (12.7)	I (15.1)	I (16.1)
E. faecalis GL320 ** (3C) (virulent, β-hemolytic)	S (22.8)	S (26.2)	I (19.1)	I (21.5)	S (11.8)	R (12.6)	S (26.9)	I (16.1)
E. faecalis GL322 * (3B) (control, non-virulent)	S (26.2)	S (22.2)	I (15.8)	I (19.8)	NT	S (17.6)	S (24.5)	R (12.3)
Sheep milk isolates
E. faecalis KFM47 (3B)	S (19.5)	S (21.8)	I (16.6)	I (16.9)	R (6.5)	R (9.2)	R (9.7)	I (15.7)
E. faecalis KFM50 (3B)	S (19.4)	S (24.2)	I (19.0)	S (22.7)	S (10.3)	R (13.3)	S (30.1)	I (16.3)
E. faecalis KTM13 (3B)	S (22.5)	S (23.0)	S (22.0)	I (20.9)	I (8.0)	R (12.1)	S (26.6)	I (16.4)
E. faecalis KTM22 (3B)	S (23.3)	S (23.2)	S (20.0)	I (19.4)	I (7.7)	R (10.7)	S (25.9)	R (13.9)
E. faecalis KTM34 * (3B)	S (22.5)	S (22.5)	I (17.5)	I (19.8)	I (7.9)	R (11.2)	S (27.7)	R (13.9)
E. faecalis KTM35 * (3B)	S (20.4)	S (22.8)	I (17.8)	I (19.4)	I (7.6)	R (13.3)	S (26.3)	R (14.2)
E. faecalis KTM39 (3B)	R (14.4)	S (24.9)	I (19.1)	S (24.5)	S (11.3)	R (13.3)	S (30.0)	I (15.7)
E. faecalis KTM51 (3B)	S (19.1)	S (27.7)	I (18.4)	S (25.7)	I (9.2)	R (8.6)	S (32.0)	S (16.8)
E. faecalis KFM48 (3D)	S (18.0)	S (22.0)	I (19.2)	I (20.2)	S (11.5)	R (8.3)	S (19.6)	S (16.9)
E. hirae KFM56 * (4A)	S (18.7)	S (23.5)	S (22.7)	I (21.4)	I (8.4)	R (12.0)	S (24.8)	R (13.4)
E. hirae KTM4 (4A)	S (21.0)	S (23.9)	I (19.8)	I (21.2)	S (12.5)	S (16.2)	S (26.2)	S (17.9)

¹ AMP, ampicillin; CHL, chloramphenicol; CIP, ciprofloxacin; ERY, erythromycin; GEN, gentamicin; PEN, penicillin; TET, tetracycline; VAN, vancomycin. The numbers near the antibiotics indicate μg per disk, or units per disk for penicillin; S (susceptible), I (intermediate), R (resistant); the numbers in brackets indicate the inhibition zone size (mm). The single and double asterisks indicate the type of antilisterial activity of the antagonistic strains, as specified in Table 2. NT, not tested.

,

Table 7. Antibiotic susceptibility of 28 Enterococcus faecium isolates (differentiated into 6 strain biotypes) from sheep milk of native Epirus breeds before (raw milk; KFM isolates) and after (thermized milk; KTM isolates) thermization.

Strain (Isolate)	Antibiotic Tested (μg/Disk) 1
	AMP 10	CHL 30	CIP 5	ERY 15	GEN 10	PEN10U	TET 30	VAN 30
Biotype 1A
E. faecium KE118 ** (control; 1A/1H)	S (17.0)	S (22.7)	R (13.0)	R (12.7)	S (11.6)	S (14.5)	S (25.9)	S (18.6)
E. faecium KFM17 **	R (14.7)	S (21.8)	I (17.5)	R (13.5)	S (12.8)	R (10.1)	S (25.6)	S (18.1)
E. faecium KFM28 **	R (16.5)	S (24.4)	I (16.7)	R (9.2)	S (13.5)	R (7.6)	S (31.9)	S (21.1)
E. faecium KFM29 **	S (17.2)	S (24.9)	I (16.8)	I (14.1)	S (11.5)	R (9.8)	S (25.8)	S (19.5)
E. faecium KTM21	S (24.3)	S (21.2)	I (17.6)	I (21.1)	S (12.5)	S (19.3)	S (24.5)	S (17.2)
Biotype 1D
E. faecium KE82 ** (control)	S (17.3)	S (24.9)	R (12.0)	R (11.0)	S (11.3)	R (11.3)	S (28.3)	S (19.4)
E. faecium KE85 (control)	S (20.8)	S (25.7)	R (13.3)	I (16.5)	I (8.1)	R (12.2)	R (10.3)	S (21.4)
E. faecium KTM46 **	R (13.8)	S (29.0)	R (13.6)	R (9.9)	S (11.5)	R (8.9)	S (25.5)	S (20.0)
E. faecium KTM47 **	R (13.8)	S (27.3)	R (14.9)	R (11.0)	S (12.0)	R (7.5)	S (27.6)	S (19.6)
Four remaining E. faecium 1D isolates	S (20.5 ± 2.0)	S (23.3 ± 1.7)	I (17.1 ± 1.3)	I (20.0 ± 1.3)	S (12.3 ± 1.6)	S (18.3 ± 1.1)	S (27.0 ± 1.7)	S (18.3 ± 0.8)
Biotype 1E
E. faecium KTM37	S (19.7)	S (23.9)	I (16.6)	R (13.1)	S (12.5)	S (27.5)	S (27.8)	S (21.4)
E. faecium KTM20	S (20.8)	S (23.3)	S (21.2)	I (21.9)	S (13.2)	S (16.0)	S (26.0)	S (18.3)
Biotype 1G
E. faecium KTM7	S (22.9)	S (23.7)	I (16.7)	I (14.6)	S (12.0)	S (25.1)	S (26.9)	S (17.8)
Biotype 1H
E. faecium KTM49 **	S (22.7)	S (21.5)	R (13.3)	R (11.2)	S (10.5)	S (22.0)	S (26.2)	S (19.2)
Biotype 1X
E. faecium KE77 * (control)	S (26.4)	S (25.5)	I (18.2)	I (18.5)	S (14.4)	S (22.8)	S (28.1)	S (19.8)
All 14 sheep milk isolates	S (21.4 ± 2.7)	S (25.8 ± 2.2)	I (18.2 ± 1.7)	I (20.1 ± 3.5)	S (13.5 ± 1.3)	S (20.6 ± 3.5)	S (28.4 ± 3.5)	S (19.8 ± 1.9)

¹ AMP, ampicillin; CHL, chloramphenicol; CIP, ciprofloxacin; ERY, erythromycin; GEN, gentamicin; PEN, penicillin; TET, tetracycline; VAN, vancomycin. The numbers near the antibiotics indicate μg per disk, or units per disk for penicillin. S (susceptible), I (intermediate), R (resistant); the numbers in brackets indicate the inhibition zone size (mm). The single and double asterisks indicate the type of antilisterial activity of the antagonistic strains as specified in Table 2.

and

Table 8. Antibiotic susceptibility of 21 Enterococcus durans isolates (differentiated into 5 strain biotypes) from sheep milk of native Epirus breeds before (raw milk; KFM isolates) and after (thermized milk; KTM isolates) thermization.

Strain (Isolate)	Antibiotic Tested (μg/Disk) 1
	AMP 10	CHL 30	CIP 5	ERY 15	GEN 10	PEN10U	TET 30	VAN 30
Biotype 2A
E. durans KE100 * (control)	S (20.0)	S (21.7)	I (20.4)	I (20.9)	S (10.7)	S (17.1)	S (23.0)	S (18.0)
E. durans KTM19	S (23.1)	S (23.4)	S (22.5)	S (23.6)	S (16.0)	R (14.3)	S (29.3)	S (18.9)
E. durans KTM38	S (20.3)	S (23.4)	S (22.3)	R (10.5)	S (13.9)	R (13.3)	S (24.9)	S (18.4)
Three remaining E. durans 2A isolates	S (21.2 ± 1.4)	S (22.3 ± 1.2)	S (21.2 ± 1.7)	S (22.1 ± 2.1)	S (12.3 ± 3.3)	S (16.9 ± 1.7)	S (24.2 ± 2.7)	S (19.2 ± 2.0)
Biotype 2B
E. durans KE108 * (control)	S (23.4)	S (26.9)	I (19.8)	S (24.0)	S (10.2)	S (19.3)	S (24.0)	S (19.0)
E. durans KFM16	S (19.8)	S (25.0)	S (23.2)	S (23.6)	S (13.4)	R (14.2)	S (27.0)	S (19.0)
E. durans KFM18	S (21.0)	S (25.7)	S (24.0)	S (24.2)	S (13.5)	R (13.0)	S (30.0)	S (19.9)
E. durans KFM20	S (20.2)	S (23.9)	S (24.2)	S (24.8)	S (13.4)	R (14.5)	S (26.8)	S (20.5)
E. durans KTM24	S (19.6)	S (21.4)	I (20.1)	S (23.5)	S (11.2)	R (13.7)	S (26.8)	S (19.4)
Three remaining E. durans 2B isolates	S (21.3 ± 2.5)	S (24.9 ± 2.1)	S (21.7 ± 1.8)	S (23.5 ± 1.2)	S (11.8 ± 2.7)	S (17.2 ± 2.2)	S (25.8 ± 2.5)	S (20.2 ± 0.8)
Biotype 2C
E. durans KFM6 *	R (15.9)	S (25.1)	S (22.3)	R (7.8)	S (14.7)	S (17.7)	S (32.8)	S (21.8)
E. durans KFM19	R (16.1)	S (23.7)	I (18.4)	S (23.7)	S (12.4)	R (10.6)	S (25.9)	S (19.5)
E. durans KTM17	S (20.3)	S (22.5)	S (22.6)	I (20.2)	S (12.1)	R (13.8)	S (25.6)	S (17.5)
Biotype 2D
E. durans KTM44	S (18.0)	S (25.9)	S (24.3)	S (23.5)	S (13.8)	R (13.8)	S (26.8)	S (17.9)
Three remaining E. durans 2D isolates	S (18.4 ± 1.2)	S (24.9 ± 1.7)	S (22.5 ± 1.1)	S (24.7 ± 1.9)	S (14.8 ± 1.6)	S (15.5 ± 0.5)	S (25.1 ± 1.9)	S (18.2 ± 0.9)
Biotype 2X
E. durans KFM46 *	S (23.7)	S (23.7)	S (21.5)	S (23.4)	S (13.1)	S (16.4)	R (13.0)	S (19.4)
E. durans KFM49 *	S (23.6)	S (21.7)	S (24.5)	S (24.5)	S (14.5)	R (13.9)	R (12.0)	S (20.8)

¹ AMP, ampicillin; CHL, chloramphenicol; CIP, ciprofloxacin; ERY, erythromycin; GEN, gentamicin; PEN, penicillin; TET, tetracycline; VAN, vancomycin. The numbers near the antibiotics indicate μg per disk, or units per disk for penicillin. S (susceptible), I (intermediate), R (resistant); the numbers in brackets indicate the inhibition zone size (mm). The single asterisks indicate the type of antilisterial activity of the antagonistic strains as specified in Table 2.

in accordance with their species identification and strain biotype within each species. Because AR is strain-dependent and, as noted in Section 3.1, each basic strain biotype in Table 2 may comprise similar strains differing in secondary properties, every sheep milk isolate found resistant to at least one of the antibiotics tested is tabulated separately; otherwise, the isolates within each biotype are combined by averaging the diameter of their inhibition zones. On this basis, the AR reactions of the isolates identified as different strain biotypes of E. faecalis, E. faecium, and E. durans are shown in Table 6, Table 7 and Table 8, respectively. The ARs of the three virulent, β-hemolytic Enterococcus reference strains used as positive controls (Table S1) and of the two sporadic sheep milk isolates assigned to E. hirae are included in Table 6, together with E. faecalis.

Initially, the reliability and reproducibility of the Kirby–Bauer disk diffusion method was confirmed by the strong MDRs of the clinical vanA + E. faecium 315VR strain, which was not inhibited by ampicillin, ciprofloxacin, erythromycin, gentamicin, or penicillin (i.e., each disk had a 6 mm diameter), while it was inhibited slightly by tetracycline, and very slightly by vancomycin; the latter was expected. Strain 315VR was of intermediate susceptibility to chloramphenicol only (Table 6). Also consistent with its reference data [48], the CLSI-QC strain E. faecalis ATCC 29212^TM was resistant to penicillin and partially susceptible to most of the other antibiotics tested, including vancomycin [46], and the virulent, β-hemolytic strain GL320 genotype of E. faecalis (biotype 3C) was also susceptible (Table 6). Similar ARs were manifested by the E. faecalis sheep milk isolates: all were resistant to penicillin and partially susceptible to ciprofloxacin, erythromycin, and/or gentamicin; however, the two antilisterial isolates KTM34 and KTM35, together with KTM22 in biotype 3B, were resistant to vancomycin. CytL_L-negative E. faecalis control strain GL322 in biotype 3B was also resistant to vancomycin, which, on the other hand, was the only E. faecalis strain sensitive to penicillin (Table 6). Meanwhile, only one strain in biotype 3B, the RM isolate KFM47, was MDR to penicillin, tetracycline, and gentamicin (Table 6); it is noteworthy that KFM47 was the only Enterococcus isolate from RM (Table 1) with two virulence genes (ace, gelE) detected in its genome [42]. Only E. faecalis KTM39 was resistant to ampicillin. Regarding the sporadic E. hirae isolates in biotype 4A, KTM4 from TM1 was fully or partially susceptible to all antibiotics tested, whereas KFM56 from RM2 was resistant to penicillin and vancomycin, like the three E. faecalis 3B isolates (Table 6).

The sheep milk isolates of the diverse E. faecium strain biotypes displayed major differences in their AR profiles (Table 7). In general consistency with their control strains, biotypes 1A, 1D, and 1H comprised all the E. faecium isolates that resisted at least two, and up to four, antibiotics tested. Specifically, the m-Ent+ E. faecium RM isolates KFM17 and KFM28 in biotype 1A and the m-Ent+ TM isolates KTM46 and KTM47 in biotype 1D were resistant to ampicillin, erythromycin, and penicillin, while the 1D isolates were further resistant to ciprofloxacin, as the m-Ent+ E. faecium KE82 control strain in biotype 1D also was. Resistant to ciprofloxacin and erythromycin, but sensitive to ampicillin and penicillin, was the single m-Ent+ isolate KTM49, sharing its AR profile with the Ent+ E. faecium KE118. Thus, consistent with the biotyping data, KE118 belongs to the atypical biotype 1H, along with the m-Ent+ KTM49 strain; the absence of entP from the genome of KTM49 (Table 3) stands as the only notable difference between them. Lastly, compared to the MDR isolates KFM17 and KFM28, the third m-Ent+ isolate KFM29 in biotype 1A was resistant to penicillin only, a result in alignment with our previous 16S rRNA-based findings that KFM29 is a different strain from the KFM17=KFM28 strain genotype, despite both possessing the entA/entB/entP gene profile [42] and having nearly identical sugar fermentation (Table S2) and enzymatic activity (Table 5) patterns.

In contrast with the above findings, all E. faecium sheep milk isolates in biotypes 1E, 1G, and 1X did not comprise AR strains, except for the TM isolate KTM37 in biotype 1E, which was resistant to erythromycin only (Table 7). Particularly, the total absence of ARs within the 14 isolates of the E. faecium biotype 1X prevailing (11 isolates; Table 2) in TM was a positive finding regarding cheese milk safety. Of course, E. faecium strains, which were totally or partially susceptible to the antibiotics tested, occurred within the biotype 1A and mainly in the biotype 1D isolates, too, including the antilisterial entA strain KTM55 in Table 2. Actually, all E. faecium sheep milk isolates of this study were of high sensitivity to chloramphenicol, gentamicin, tetracycline, and, mainly, vancomycin (Table 7).

Compared to E. faecium, the E. durans biotypes comprised isolates with relatively fewer ARs; no MDR strains were found. Only 4 out of the total of 21 E. durans isolates were resistant to two antibiotics, and none to three or more (Table 8). However, similar to E. faecium, the most pronounced AR for 47.6% (10/21) of the E. durans isolates was to penicillin, distributed within all five strain biotypes. Beyond that, the two biotype 2C E. durans isolates from RM, including the entA/entP strain KFM6 (Table 3), were resistant to ampicillin, while an additional two antilisterial (entP) isolates from RM (KFM46, KFM49) assigned to the new atypical biotype 2X, were resistant to tetracycline.

The antilisterial KFM6 strain was also resistant to erythromycin, as well as one TM isolate (KTM38) of the most typical E. durans biotype 2A. However, KFM6 and KFM46 strains were susceptible to penicillin, contrary to the tendency of ca. 50% of the E. durans sheep milk isolates. On the other hand, all E. durans strain biotypes were susceptible to chloramphenicol, ciprofloxacin, gentamicin, and, mainly, to vancomycin (Table 8). Meanwhile, both typical E. durans antilisterial control strains, KE100 (2A; entP) and KE108 (2B; entA/entP/bac31), from mature Graviera cheese (Table S1) were susceptible to all antibiotics tested, and they were particularly sensitive to penicillin, as were a major (42.9%) part of the E. durans sheep milk isolates (Table 8).

4. Discussion

Mild thermization treatments of raw sheep/goat milks applied in traditional Greek cheese technologies select for thermoduric, non-starter LAB (NSLAB) survivors, primarily autochthonous enterococci in TM [26,43]. This natural selection is very critical from a practical, technological, and safety point of view because salt-tolerant enterococci have a high growth potential during the fermentation and ripening of Greek TM cheeses [9,45], especially when NSCs of low cell density, slow growth, and/or low competitive ability are used [10]. Globally, the prevalence of autochthonous antagonistic, Ent+/m-Ent+ Enterococcus strain biotypes with high aminopeptidase and esterase activities in fresh curds has established beneficial effects on the taste and aroma development of traditional RM or TM cheeses, as well as on the in situ growth inhibition of pathogenic bacteria, such as L. monocytogenes and S. aureus [7,20,28,29,54]. However, cheese safety risks may arise when the naturally selected Enterococcus biota in TM [26,43] comprises high numbers of opportunistically pathogenic strains possessing one or more of the following unwanted traits: β-hemolytic activity, cytolysin genes/cytolytic activity, virulence genes (especially IS16, esp, hyl-like), AR genes, vancomycin resistance, MDR, and undesired enzymatic activities, such as β-glucoronidase and β-glucosidase, which are known to be potential mediators of colon carcinogenesis [32,33,50,54,55,56].

Thus, to develop novel NSCs of high LAB species diversity and complexity derived from sheep milk of native Epirus breeds, the selected strains should be free of the above potential or additional (i.e., histamine formation) pathogenic traits. Conversely, specific Enterococcus strains with β-galactosidase, esterase–lipase, and aminopeptidase activities, moderate proteolytic activity, high diacetyl/acetoin production, which are safe and also manifest high in situ enterocin gene expression and activity, are amongst the best LAB species candidates for inclusion in complex NSCs for traditional (hard) cheese-making [44,45,56]. In this context, the results of this study demonstrate an additional enterococcal controversy which, to our knowledge, has yet to be addressed adequately: the two most promising m-Ent+ (entA/entB/entP) bioprotective E. faecium strains in biotypes 1A and 1D (Table 2 and Table 3), which also produce acetoin and possess high β-galactosidase, esterase, esterase–lipase and aminopeptidase activities (Table 4 and Table 5), are simultaneously MDR strains, particularly to ampicillin (Table 7). According to the EFSA [50] guidelines, any E. faecium strain demonstrating a resistance to ampicillin greater than 2 mg/L or possessing any of the three marker genes, IS16, esp, hylEfm, should not be used as a feed additive in animal nutrition and, consequently, in human food [54]. IS16, in particular, is an insertion sequence involved in ampicillin resistance, as well as a novel identifier for emerging MDR strains of E. faecium [54,57]. Hence, both m-Ent+ E. faecium native bioprotective strain biotypes, KFM17=KFM28 and KTM46=KTM47, are free of the above virulence markers, mainly IS16, and also are α-hemolytic, susceptible to vancomycin, and lack β-glucoronidase and β-glucosidase; their phenotypic resistance to ampicillin is of serious concern, hampering their inclusion in complex NSCs without testing for ARGs in addition to the vanA/vanB testing conducted before [42]. Nevertheless, the above controversy may be associated with induced stress hardening of the most antagonistic m-Ent+ Enterococcus strains for survival in natural niches, such as in animal farms and raw milk environments [28,29], where the use of antibiotics along with other chemical or physical stressors may trigger several horizontal gene transfers among the best-adapted enterococcal strains and typical pathogenic bacteria in the field [28,32,33]. This hypothesis aligns with the fact that most sheep milk isolates with direct antilisterial activity were also resistant to the least two antibiotics; the most active E. durans KFM6 (2C) strain was ampicillin-resistant, too (Table 8), and E. faecalis KTM34, E. faecalis KTM35, and E. hirae KFM56 were vancomycin-resistant (Table 6). A recent epidemiological survey by Ayobami et al. [58] reported that the proportions of vancomycin-resistant E. faecium in Europe increased from 8.1% in 2012 to 19.0% in 2018, while the mean proportion of vancomycin-resistant E. faecalis remains low, 1.1%; this is clearly an emerging safety concern and an ongoing challenge [58]. Notably, except for one E. faecium, all vancomycin-resistant/MDR strains isolated from Egyptian fresh raw milk cheese were E. faecalis [40]. Likewise, in this study, the E. faecalis biotype 3B control strain GL322 from fresh Galotyri PDO cheese was vancomycin-resistant (Table 6), despite the fact that it was previously shown to be a-hemolytic and free of virulence genes [42]. Also, apart from the three resistant survivors in TM, another four RM/TM E. faecalis in biotype 3B were partially susceptible to vancomycin.

The profound phenotypic resistance of all E. faecalis, the m-Ent+ (entA/entB/entP) E. faecium biotype 1A and 1D strains, and most E. durans sheep milk isolates to penicillin is an additional concern, consistent with our previous findings regarding the profound penicillin-resistance of the m-Ent+ (entA/entB/entP) E. faecium KE82 and the Ent+ (entA) E. faecium KE64=KE67 control strains from naturally fermented Graviera cheese [45]. Overall, the in vitro phenotypic susceptibility of dairy isolates of the E. faecium/durans genomic group to different antibiotic groups is a strongly strain-specific trait that displays major variations from country to country, from the raw materials (i.e., milk) to the final cheese product, and between cheese varieties and research studies. Generally, apart from their susceptibility to vancomycin (i.e., a primary prerequisite for NSC development), most dairy isolates of E. faecium are susceptible to ampicillin, chloramphenicol, gentamicin, penicillin, and tetracycline, too [59,60,61,62,63].

However, many penicillin-resistant dairy Enterococcus spp., specifically E. faecium strains that display resistance to penicillin, gentamicin, and/or tetracycline and present the respective ARGs, have been reported to occur in milk and traditional cheeses by others [40,64,65,66]. Sometimes, contradictive findings were reported between milk types and countries; for instance, all E. faecium/hirae and E. faecalis isolates from Spanish goat milk were susceptible to gentamicin (low level; 10 μg/disk) [41], as they were our isolates, too, but showed a high percentage of resistance to tetracycline [41]. In contrast, all antilisterial E. faecium strains from Turkish raw cow milk were resistant to gentamicin (10 μg/disk), but all were susceptible to tetracycline [67]. In Egyptian RM cheese, 55% of the Enterococcus isolates were gentamicin-resistant (low level), but only one E. faecium strain showed high level (120 μg) gentamicin (aminoglycoside) resistance [40]. Also, many studies emphasize an increased proportional resistance of enterococci to erythromycin and ciprofloxacin. In particular, most E. faecium isolates from dairy foods display an increased resistance or intermediate susceptibility to erythromycin and ciprofloxacin [64,65,68,69]. The above trend of E. faecium was strongly confirmed, specifically for the most antagonistic strains with strong CFS antilisterial activity, including the single m-Ent+ (entA/entB) E. faecium KTM49 and the m-Ent+ (entA/entB/entP) control strain KE118 in biotype 1H, both of which were susceptible to ampicillin and penicillin (Table 7). E. faecium with MDR profiles were also frequent in Balkan cheese, while E. durans presented an overall higher antibiotic susceptibility than their E. faecium co-isolates [69], as is also shown in Table 8. In agreement, the most recent comprehensive genomic analysis study of Enterococcus food isolates by Tadesse et al. [54] reported that based on their MIC (μg/mL) values, E. faecium BT0194, five selected E. durans, and 12 selected E. lactis isolates were all susceptible (i.e., below the EFSA cutoff limits for each antibiotic tested) to ampicillin, vancomycin, tetracycline, chloramphenicol, kanamycin, streptomycin, and erythromycin, except for one being resistant, and six out of the twelve E. lactis strains showing intermediate susceptibility, to erythromycin.

On the other hand, the enriched enzymatic activity profiles of the 20 sheep milk strains of the E. faecium/durans group (Table 5) are desired in case all, or some, of them are selected for inclusion in the complex NSC to be developed for onsite use in future cheese-making studies. We suggest that the moderate-to-strong esterase, esterase–lipase, and arylamidase activities of all strains of the E. faecium/durans group should be expressed in situ along with the moderate-to-strong β-galactosidase activities of nearly all E. faecium strains, as they can act complementarily to the strong or moderate α-galactosidase and other beneficial enzymatic activities of primary NSC strains, such as Streptococcus thermophilus ST1, Lactococcus lactis subsp. cremoris M78, and Lactiplantibacillus plantarum H25 are already used effectively in traditional Greek cheese technologies [36,45]. Based on the results and the preceding discussion, we conclude that antibiotic-susceptible representatives of all strain biotypes of the E. faecium/durans/hirae group, especially of the prevalent biotypes 1X, 1D, 2B, 2A, and 1A, can be beneficial and safe constituents of a novel complex NSC. Safe, non-MDR representatives of the above or less prevalent E. faecium and E. durans biotypes showing in vitro antilisterial activity, such as KTM55 (1D), KTM49 (1H), KFM57, KFM58, and KFM59 (1X), KTM45 and KTM48 (2A), KTM31 (2B), KFM6 (2C), KFM 46, and KFM49 (2X), may be selected, depending on their entire functional traits. The most antagonistic, non-virulent Ent+ or m-Ent+, but otherwise phenotypically MDR, E. faecium strain genotypes (i.e., KFM17=KFM28 and KTM46=KTM47) can be included, too, provided that there is a complete genomic safety evaluation for possession of ARGs, whereas all vancomycin-resistant and/or virulent E. faecalis and E. hirae strains must be excluded.

Overall, the inclusion of safe antilisterial strains of the E. faecium/durans group in novel complex NSCs can contribute to eliminating L. monocytogenes in traditional cheeses, including the cheese types made of sheep milk. This is needed because the incidence of L. monocytogenes in sheep milk cheeses has been reported by meta-analysis at 3.61% (95% CI: 0.70–16.4%), with the natural L. monocytogenes contamination in raw sheep milk lying at similar levels (3.56%; 95% CI: 1.53–8.11%) [24]. Also, announced recalls of market sheep milk cheeses contaminated with L. monocytogenes remain frequent, as, for example, an Italian semi-hard sheep cheese in Slovenia [70] and a French and a German-type sheep cheese in the United States [71,72] were recalled in 2023–2025. Notably, although the use of raw (sheep) milk is a primary route for the transfer of L. monocytogenes in the resultant cheese, recent systematic reviews on the prevalence of the pathogen in various cheese types have emphasized that, globally, pasteurized and/or thermized milk cheeses may also contain L. monocytogenes survivors at comparable contamination levels [23,73,74]. Particularly, in European cheeses for the period 2005–2015, meta-analysis revealed that the mean pooled prevalence of L. monocytogenes in soft/semi-soft cheeses did not differ significantly for cheeses produced from pasteurized (0.9%; CI: 0.4–1.9%) or unpasteurized (1.0%; CI: 0.4–2.2%) milk [74]. Thus, although pasteurization or thermization kills the normally few pathogen cells [17,26,43], post-thermal contamination of the milk with L. monocytogenes occurs to result in final RTE cheeses, mainly soft, smear, or brined cheese types, harboring viable pathogen counts, often at levels above the 100 cfu/g threshold specified in EC 2073/2005 Regulation; occasionally, the Listeria-positive samples are at contamination rates as high as 40–46% of the tested samples, including soft and semisoft (sheep/goat) cheeses from the Greek market [73,74]. To minimize the above serious safety risks in traditional (artisanal) cheese-making technologies, the addition of CSCs or, preferably, diverse NSCs comprising Nis+ Lc. lactis and/or safe m-Ent+ E. faecium/durans as starter adjunct strains, has been recommended. Selected antagonistic NSCs in particular have potential to restore the natural antilisterial activity of the autochthonous raw sheep milk biota reduced by pasteurization or thermization [27,75,76] and provide additional biotechnological benefits toward an extended shelf-life, an enhanced cheese proteolysis and aroma development, and probiotic, health-promoting effects [7,28,29,76].