Development of a Semi-Industrial Kefalotyri-Type Cheese Using Thermized Milk from Native Epirus Sheep Breeds and Autochthonous Starter and Adjunct Cultures

Abstract

Autochthonous starter and adjunct cultures have gained increasing practical research interest for use in traditional cheese-making in recent years. This study evaluated the performance of a native starter comprising Streptococcus thermophilus ST1 and two wild Lactococcus lactis strains, including the M78 nisin A-producer, during the semi-industrial production of Kefalotyri cooked hard cheese from thermized sheep milk (TM; 65 °C, 30 s) in the absence (C-cheese) or presence of the native adjunct strains Lactiplantibacillus plantarum H25 and Leuconostoc mesenteroides KFM7 + KFM9 (N-cheese). The growth of the native starter was optimal in all three cheese trials within the first 24 h of processing, but afterward, ST1 failed to exceed 8 log CFU/g in favor of total mesophilic LAB, comprising both L. lactis starter strains, and the prevalent (ca. 8.5 log CFU/g) H25 adjunct in the ripening N-cheeses. Instead, in the C-cheeses, indigenous non-Enterococcus NSLAB survivors from TM prevailed, whereas enterococci failed to increase above 6 log CFU/g in all cheeses. Although the mature (90-day-old) N-cheeses presented no statistically significant differences regarding the pH value, gross composition, and hydrophobic (HO)/hydrophilic (HI) peptide ratios from the mature C-cheeses, they had lower total LAB counts and contained less residual lactose, more acetate, and an overall less diversified volatilome (VOC) profile. The most abundant VOCs in both cheeses were acetone, butyric acid, methyl butyrate, ethyl ether, and ethanol. All mature cheeses were safe and graded of ‘excellent quality’ (i.e., moisture < 35%; fat-in-dry-matter > 47%).

1. Introduction

Greece has a very long and strong tradition in cheese manufacturing technologies, resulting in the production of many different cheeses, either in the country’s mainland or many islands [1,2]. From ancient Greek myths attributing cheese-making knowledge to the gods, including a reference in Homer’s Odyssey, to the Byzantine Empire’s cheese production reminiscent of feta, the evolution of Greek cheeses is interwoven with historical events [2]. In contemporary Greece, cheese holds a significant place in the daily diets of its inhabitants, with over 70 varieties recorded, 23 registered as having a protected designation of origin (PDO), and two as having a protected geographical indication (PGI) [3,4,5]. Among the rest Greek cheeses currently not listed in the EU geographical indications registers [6], Kefalotyri, a robust and venerable hard cheese [1], is of the highest popularity and total annual production, traded under the name of the region where it is manufactured [2,7]. Thus, Kefalotyri is regarded as a precursor to many of the distinguished Greek hard cheeses [6], with Kefalograviera PDO produced in Epirus and Western Macedonia from sheep milk of native breeds, permissively mixed with up to 10% goat’s milk sharing the highest technological, compositional, textural, and sensorial similarity [1,3,4].

Nestled within the heart of Greek culinary tradition, Kefalotyri cheese stands as a testament to the rich heritage and gastronomic diversity of the Mediterranean region. With a history dating back centuries, this distinctive cheese has woven itself into the fabric of Greek culture, gracing tables with its unique flavor and playing a pivotal role in the country’s traditional cuisine. Originating from the mountainous landscapes and sun-soaked pastures of Greece, Kefalotyri not only encapsulates the essence of artisanal craftsmanship but also reflects the resilience of time-honored cheese-making practices [1]. The etymology of its name is steeped in intriguing possibilities, ranging from associations with the Greek words “Kefali” (head) and “Tyri” (cheese) to its resemblance to a traditional Greek hat known as “Kefalo” [1,8]. With a flat, cylindrical shape, compact texture, and a distinctive salty and piquant taste, Kefalotyri weighs between 5 and 10 kg and finds its place as a versatile table cheese, mainly eaten as a delicatessen accompanist of wine or other traditional Greek alcoholic drinks. Its culinary applications extend to pie-filling, pan-frying, grilling, or grating, adding depth to various Greek dishes [6].

The production of Kefalotyri exhibits prominent variations in cheese manufacturing technologies across different regions of Greece, which have been against establishing strict specifications required for its PDO certification. Nonetheless, despite an initial PDO rejection by the EU, specifications established in 1994 serve as a foundation for many Kefalotyri cheese varieties on the market. An examination of Kefalotyri’s historical production reveals shifts from traditional, heavily salted cheeses created in unhygienic conditions to modern, industrial-scale productions that align with current consumer preferences [1,8,9]. Regional cheese variations, such as the Kefalotyris of Epirus, Crete, Naxos, Thessaly, or Kefalonia, highlight diverse cheese shapes, sizes, and qualities [1,2].

Traditionally, Kefalotyri cheese is produced from sheep or goat milk, or mixtures of them, but cow milk can also be used [1,2]. A high variability in the proportion of each milk type added in the mixture continues nowadays; however, while the artisanal methods utilize raw milk without starter cultures or with natural starter and rennet preparations, contemporary production employs pasteurized milk with the addition of commercial starter cultures and commercial rennet in powdered or liquid form, reflecting evolving industry standards [10,11,12,13]. Otherwise, Kefalotyri still undergoes different types of salting (brine, dry salting, or a combination of both techniques) sufficient to achieve a final salt content >3% to 4% after maturation for a minimum of 3 months, in accordance with Greek legislation [14]. The mature cheese has a salty, spicy taste and a pleasant aroma. Depending on the mixture of milk used in production, the cheese color varies between white and yellow. The Greek Codex Alimentarius requires hard cheeses, including Kefalotyri, to have a maximum moisture content of 38% and a minimum percentage of 40% fat-in-dry-matter (FDM) to be of first quality, while an excellent quality hard (Kefalotyri) cheese has a moisture content < 35% and FDM > 47% [14].

Despite its high popularity, production, and consumption, published research data on the non-PDO Kefalotyri cheese are limited compared to similar Greek PDO hard cheeses, namely Kefalograviera and three Graviera PDO varieties [4], particularly regarding its microbiology [2]. Chemical studies compared the gross composition and descriptive physicochemical characteristics of Kefalotyri with other Greek hard cheese-type samples collected from the market [7]. Additional studies on the nutritional profile of Kefalotyri reviewed most recently [6] reveal variations in moisture and highlight its cholesterol concentration, anti-atherogenic properties, and levels of selenium, nitrate, and fatty acids [6,15,16]. The presence of specific volatile compounds and fatty acids, such as butyric acid and 3-methyl butanoic acid, in artisanal Kefalotyri was emphasized [10]. Also, a closer look at the aging process of Kefalotyri unveils variations in lipid and protein oxidation, influenced by factors such as season of production (winter vs. spring milk cheeses), light exposure vs. storage in the dark, and storage conditions (in air vs. vacuum or MAP) [13,17].

The microbiota of traditional Kefalotyri cheeses made from raw [9] or pasteurized sheep [12], goat [18], or cow [19] milk without or with the use of natural or commercial starter cultures have been tested in very few studies, all published before 2005 and last reviewed a decade ago [2]. Streptococcus thermophilus and Lactobacillus bulgaricus, being naturally symbiotic in yogurt, were the best-performing starter cultures in Kefalotyri from pasteurized sheep milk, but Lactococcus lactis and Lacticaseibacillus (ex. Lactobacillus) casei also performed well [12]. However, in mature (4-month-old) Kefalotyri made from pasteurized (71.7 °C; 15–20 s) cow milk without starter cultures, Enterococcus faecium predominated, followed by Lactiplantibacillus (ex. Lactobacillus) plantarum and Lc. casei; whereas S. thermophilus and Leuconostoc (Ln.) lactis, which were initially prevalent during fermentation, disappeared early in ripening, as did indigenous lactococci and gas-forming lactobacilli [19]. The production of Kefalotyri-like cheeses from goat milk with yogurt as a starter, or with its partial replacement with the probiotic adjunct strains Lc. rhamnosus LC705 and Lc. paracasei subsp. paracasei DC412, was shown to be an effective way of producing a probiotic cheese, preferably with the DC412 strain [18].

The microbial ecology of artisanal Kefalotyri cheeses from raw milk remains largely unexplored [2]. A high lactic acid bacteria (LAB) diversity with dynamic changes in the dominant non-starter LAB (NSLAB) species from the fresh cheese (10-day-old; 15 different LAB species; Enterococcus faecalis was predominant) to the ripened cheese (90-day-old; 14 different species) was found in ‘Orinotyri’ (Orino + tyri = mountainous + cheese), a local name of a Kefalotyri variety produced from raw sheep milk in a mountainous village in Epirus [9]. An early outgrowth (>7–8 log CFU/g) of total Enterobacteriaceae and coliforms in the fresh cheese declined below the 5-log level in ‘Orinotyri’ [9] and in another type of mountainous raw sheep milk Kefalotyri of Pindos [10] after 90 and 60 days of ripening, respectively. However, despite the viability of spoilage or pathogenic Gram-negative bacteria being suppressed during ripening, the total quality and safety of high-pH (>5.7) artisanal hard cheeses, like Kefalotyri, may be compromised; on the other hand, when the milk is pasteurized and then inoculated with Direct Vat Set (DVS) commercial starter cultures at high initial cell numbers, the indigenous (autochthonous) NSLAB species diversity in the fermenting and ripening cheese is reduced or may be lost [20,21].

In this context, mild thermization treatments (57–68 °C; from 5 to 60 s holding time) empirically applied in traditional (hard) cheese technologies to moderate the thermal killing effects of pasteurization (72 °C; 15 s, or equivalent at 60 °C; 30 min, or 68 °C; 10 min; open-batch) on the beneficial raw milk LAB biota offer a valuable alternate option [20,22]. Overall, compared to most Gram-negative spoilage and pathogenic bacteria, which are vulnerable to thermization, dairy LAB are relatively heat-resistant and thus selectively survive post-thermally in the milk intended for (Kefalotyri and other hard-type) cheese production, mainly autochthonous enterococci, thermophilic streptococci, and non-starter lactobacilli [22,23]. Considering the shift towards industrialization and potential loss of biodiversity, efforts are underway to preserve the traditional Kefalotyri of Epirus. This research aims to restore its historical character through the exclusive use of local milk from native sheep breeds after thermization and native starter cultures.

2. Materials and Methods

2.1. Native Starter/Adjunct LAB Strain Combinations and Culture Conditions

Six native (indigenous) LAB strains preserved at the Dairy Research Department, Applied Microbiology Laboratory (Katsikas, Ioannina, Greece), were selected for use, namely [21], S. thermophilus ST1, a natural commercial strain previously evaluated as the primary starter for optimal cheese milk acidification; L. lactis subsp. cremoris M78, a wild nisin A (NisA+)-producing lactococcal strain genotype (strain M78 = M104) isolated from Epirus raw sheep/goat milk and routinely applied as a co-starter and bioprotective (anti-listerial) adjunct culture in commercial Graviera cheese production at our collaborating semi-industrial plant (SKARFI E.P.E.; Pappas Bros. Traditional Dairy, Filippiada, Epirus, Greece) since 2015 [24]; L. lactis subsp. lactis KE109, another wild-type lactococcal strain isolated from a naturally fermented Graviera Pappas Bros. cheese [25] and characterized by good skimmed milk acidification (pH 4.5; 30 °C; 24 h) in vitro; Lp. plantarum H25, an indigenous NSLAB strain isolated from a ripening CSC-fermented Graviera Pappas Bros. cheese with enriched enzymatic and fermentation activity profiles in vitro compared to the ST1 and M78 strains [21,25]; and Ln. mesenteroides KFM7 and KFM9, two novel NSLAB strains recently isolated from raw sheep milk of native Epirus breeds [26]. A very good growth compatibility of the ST1, M78, and H25 strains was previously recorded in model sheep milk co-cultures [27], as well as under pilot-scale artisan fresh cheese manufacturing conditions at the SKARFI plant [21]; whereas this is the first time the strains KE109, KFM7, and KFM9 were evaluated in a cheese-making process.

Specifically, for the purposes of this study, the ST1 + M78 + KE109 strain combination served as the basic mixed thermophilic/mesophilic native starter culture to produce an artisan-type Kefalotyri product (C-cheese; control), while the H25 + KFM7 + KFM9 strain combination served as a native mesophilic NSLAB adjunct culture, supplementary to the basic starter, to produce a new innovative Kefalotyri cheese type (N-cheese) under direct, comparable semi-industrial conditions. In particular, the two autochthonous gas-forming strains KFM7 and KFM9 were used to replace the predominant Leuconostoc, mainly Ln. mesenteroides biota of raw sheep milk [23,26], which was reduced (1.7% survival only) by the preceding thermization [23], thus aiming to enhance eye-(hole)-opening and possibly improve the taste and aroma development in Kefalotyri cheese by their interaction with the basic starter and the primary NSLAB ripening strain Lp. plantarum H25 [21,25].

All strains were sub-cultured twice for reactivation by transferring 0.1 mL of stock (−30 °C) culture into 10 mL of MRS broth (Neogen Culture Media, Lab M, Heywood, UK) incubated at 30 °C for 24 h, except for S. thermophilus ST1, which was reactivated and cultured in M17 broth (Merck, Darmstadt, Germany) incubated at 37 °C for 24 h. For the preparation of the Kefalotyri cheese milk inocula, all strains were cultured in 10 mL of heat-sterilized (121 °C, 5 min) reconstituted skimmed milk (RSM, Lab M), as above. Fresh RSM cultures were used to develop pure cultures of each strain in pre-sterilized portions (400–1000 mL) of sheep milk incubated for 24 h at the optimal growth temperature of each strain. These cultures were used for milk inoculation on each cheese production day at SKARFI.

2.2. Traditional Kefalotyri Cheese Processing with Native Starter and Adjunct Cultures

All the traditional Kefalotyri cheese products in this study were made of ‘summer’ bulk (tank) milks collected in June from two sheep yards with mixed native Karamaniko and Karagouniko breeds, located around Arta, Epirus, Greece. Three raw sheep milk batches (coded RM3, RM4, and RM5) were used to produce three individual Kefalotyri cheese trials at different processing days at the SKARFI plant. All trials were processed after mild (65 °C; 30 s) thermization of the milk; the counterpart thermized milk batches were coded TM3, TM4, and TM5, as previously reported by Samelis et al. [23], who evaluated the microbiological quality and safety of the above milks before and after thermization.

Both Kefalotyri types (C-cheese; N-cheese) in each trial were processed according to the plant’s routine manufacturing protocol (Figure S1), suitably modified to inoculate the fresh native starter (ST1 + M78 + KE109) and adjunct (H25 + KFM7 + KFM9) culture mixtures (i.e., instead of adding commercial starter cultures) in the mildly thermized (i.e., instead of using pasteurized) sheep milk under semi-industrial cheese processing conditions.

At this point, it should be noted that mild thermization (65 °C; 30 sec heat-holding time) of the maternal RM3, RM4, and RM5 batches in the plant’s ‘closed’-type pasteurizer (Schmibt, Sigma 17S BN, Bretten, Germany) was intentionally selected for use in order to enhance the survival of the indigenous NSLAB biota while suppressing the spoilage, mainly Gram-negative, and pathogenic bacteria present in the raw sheep milk, as shown by Samelis et al. [23]. The primary aim was to produce two new artisan-type Kefalotyri cheese varieties with native starter LAB cultures, simultaneously characterized by a more diversified indigenous NSLAB biota, compared to the CSC-mediated Kefalotyri cheese.

For cheese-making, the bulk TM of each trial (TM3, TM4, and TM5 batches; 1400 L each) was divided into two 700L portions automatically pumped from the pasteurizer into separate, stainless-steel open vat tanks with double-wall steam circulation under continuous stirring. The first portion was used to make the C-cheese inoculated with the basic starter culture only; thus, it was always processed first, followed by the N-cheese inoculated with both cultures. Each starter or adjunct LAB strain was separately added to each of the bulk milks kept constant at 35 °C in the vat, in an amount sufficient to provide an initial inoculation level of the milk in the range of 6.0–6.5 log CFU/mL for strains ST1, M78, and KE109 and 5.5–6.0 log CFU/mL for strains H25, KFM7, and KFM9.

After LAB inoculation, liquid rennet (80% chymosin, 20% pepsin; strength 1:15,000, Semi Picante 80PS15, Bioren, Corato, Italy) and CaCl₂ were added, and the cheese milk was left to curdle for 35 min (Figure S1). After milk curdling and cutting of the curd, the final reheating temperature was reduced from 45 to 43 °C by, respectively, increasing the total reheating time from 20 to 48 min (i.e., a critical modification of the manufacturing protocol in Figure S1) in order to facilitate the native mesophilic starter’s and adjunct LAB strains’ initiation of growth in the curd while ensuring sufficient whey expulsion.

After cooking, the curd was not handled as described in Figure S1 but, instead, the cooked curd mass, together with the whey, was transferred by pumping from the double-jacketed cheese vat to an industrial stainless-steel pressing machine equipped with a draining basin underneath (Kalt Söhne AG, KäsereiMaschinen, CH9604 Lütisburg, Switzerland). The pump evenly distributed the curd into the perforated, stainless-steel Kefalotyri cheese molds (30 cm diameter), which then were covered with the toplids and pressed for two subsequent times (30 min each). The first pressing (20 kg/cm²) was initiated with the molds immersed in whey, while the second pressing (40 kg/cm²) was applied after the whey was drained out, and the cheeses were turned over in the molds by hand. After pressing, the fresh cheeses were de-molded from the pressure molds, placed in new pre-disinfected plastic molds (30 cm diameter) and left to equilibrate at 18 °C for 24 h before brining. During the above operations, the temperature and the pH of the cheese milks, curds, or fresh cheeses was recorded with a probe thermometer and a portable digital pH meter equipped with a pointed electrode, respectively.

Next, the fresh fermenting cheeses were removed from the plastic molds, immersed in 20% brine at 12 °C for 48 h, left for an additional 2 days at 12 °C to drain (Figure S1), and then were transferred to the ripening (maturation) chamber, which at SKARFI is a fully controlled industrial room operating at 16.5 ± 0.5 °C and 91.0% ± 1.0% relative humidity under a standard, temperature-based sequential air ventilation program [24]. The brined (6-day-old) cheeses were ripened side-by-side on the wooden shelves for an additional 24 days (30-day process). During this period, 6 dry saltings (3 sequent times × 2 cheese mold surfaces) were implemented as follows: on the surface of each cheese, 50 g of coarse sea salt the size of rice grains were distributed, left to melt slowly overnight, and spread on the surface and sides with cheesecloth the next day; the cheese was then flipped and salted from on other surface the following day. This process was repeated thrice, starting on day 3 in the ripening room for 18 days to complete dry salting. The primary ‘controlled-ripening’ maturation process was concluded one month (30-day-old) after cheese manufacture. From day30 onwards, the cheeses continued to mature in cryovac vacuum bags at 4 °C for an additional two months at a minimum (90-day-old) before consumption or further storage at 4 °C until their distribution in the market.

Each trial resulted in the production of a total of 24 Kefalotyri cheese molds (28 cm diameter; 11–12 cm height; 8–10 kg each); 12 for each cheese (C or N) treatment.

2.3. Sampling of Traditional Kefalotyri Cheeses During Processing and Ripening

All cheese trials (C1 to C3 and N1 to N3) were analyzed for their physicochemical and microbiological profiles at constant time intervals during processing and ripening. Samples were taken on day 1 (unsalted, fermenting cheese), day 6 (brined cheese before ripening), day 18 (middle of ripening; for conducting microbiological analyses only), day 30 (end of ripening in the maturation chamber) and day 90 (fully ripened cheese).

On each sampling day, two separate molds of each cheese treatment (C or N) were sampled either aseptically with the aid of sterile cork-borers or with a clean/disinfected stainless-steel knife; the former samples were kept separate in pre-sterilized pouches for conducting the microbiological analyses whereas the latter samples were cut in cubes and pooled in plastic bags for the chemical analyses. Additionally, the RM/TM samples before or after inoculation and cooked curd samples from all trials and treatments were poured into pre-sterilized Duran bottles during cheese manufacture (day 0). All samples were transferred to the Dairy Research Department in insulated ice boxes within 45 min. Samples were subjected to the microbiological analyses, pH measurement, and chemical gross composition (moisture, fat, solids, salt, etc.) analyses on the day of sampling, or they were divided into portions and stored in a freezer (−30 °C) for later biochemical analyses.

2.4. Microbiological Analyses of the Kefalotyri Cheese Samples

On each sampling occasion, 25 g of cheese milk or curd samples were homogenized with 225 mL of sterile quarter-strength Ringer solution (Lab M) in stomacher bags (Lab Blender, Seward, London, UK) for 60 s at room temperature. Afterward, 1 mL portions of the homogenates were decimally diluted in tubes with 9 mL of Ringer, and appropriate dilutions were poured (1 mL samples) or spread (0.1 mL samples) in duplicate on total and selective agar plates. Unless otherwise stated, all agar media and their supplements were purchased from Neogen Culture Media (formerly Lab M).

All cheese samples were analyzed for total viable counts (TVCs), total mesophilic and thermophilic LAB, total mesophilic and thermophilic dairy (lactose-fermenting) LAB (presumptive lactococci and streptococci, respectively), enterococci, kanamycin-resistant/aesculin-positive lactobacilli, total staphylococci, pseudomonad-like bacteria, coliform bacteria, and yeasts and molds, according to the procedures and by using the enumeration agar media and incubation conditions previously described and tabulated by Samelis et al. [23] for the analysis of the corresponding sheep milk batches before and after thermization. The only analytical difference was that, after sample spreading, the M17 agar (Conda S.A., Madrid, Spain) plates used for enumerating the total thermophilic dairy LAB populations were incubated at 45 °C, instead of 42 °C, for 48 h in order to become highly selective for the enumeration of the primary starter strain S. thermophilus ST1—none of the mesophilic native LAB strains M78, KE109, H25, KFM7, and KFM9 can grow at 45 °C [21,25,27]—while the contamination and growth of autochthonous LAB—mainly enterococci—able to grow at 45 °C remained, at 100–1000-fold lower levels in all the inoculated TM, curd, and cheese samples across processing. The latter observation was confirmed by the selective enumerations of enterococci on Slanetz and Bartley (SB) agar and Kanamycin Aesculin Azide (KAA) agar, incubated at 37 °C for 48 h. Neither ST1 nor any of the mesophilic native LAB strains herein used for cheese milk inoculation can promote growth on SB (or KAA) agar media, except for Lp. plantarum H25 promoting an unhindered growth in the form of whitish colonies which were easily discriminated from the reddish-brown Enterococcus colonies on the SB agar plates [21].

Additionally, accurate selective enumerations of the NisA+ M78 colonies on the M17/22 °C agar plates of both cheese-type (C or N) samples by the simple agar overlay anti-listerial assay and of the large convex yellowish H25 colonies macroscopically on the MRS/30 °C agar plates of the N-cheese samples were performed, as described in the relevant pilot-scale Galotyri PDO production study by Samelis et al. [21]. The large colony growth of the KE109 starter strain intermixed with other indigenous lactococcal colonies were also easily discriminated on the M17/22 °C plates of all cheese samples from the pin-point colony growth of the ST1 and H25 strains on the same plates after 48–72 h of incubation at 22 °C.

Direct selective enumerations were difficult to conduct only for the Ln. mesenteroides strains, KFM7 and KFM9, in the N-cheeses; both strains grow optimally in the form of indistinguishable whitish LAB colonies on Milk Plate Count (MPCA), MRS, and M17 agar media at 22 to 37 °C. Therefore, to conclude whether sufficient competitive growth of the Ln. mesenteroides inocula occurred in the N-cheeses, indirect culture techniques were applied post-enumeration of the rest starter/adjunct LAB strains (viz. Results).

Pathogenic (coagulase-positive) staphylococci were discriminated within the total staphylococcal colonies on Baird–Parker agar plates with egg yolk tellurite at 37 °C for 48 h by a rapid latex agglutination test. Lastly, in all trials, the potential presence of Listeria/L. monocytogenes and Salmonella spp. contamination was assessed by the culture enrichment of 25g cheese samples taken before (day 6) and after (day 30) ripening, using the selective culture media and pathogen identification kits reported by Samelis et al. [23].

2.5. Measurement of pH and Gross Composition of the Kefalotyri Cheese Samples

The pH of all Kefalotyri cheese, milk, or curd samples was measured with a Jenway 3510 digital pH meter (Essex, UK) after plating for microbiological analysis. All fresh and ripened Kefalotyri cheese samples were analyzed for moisture, fat, protein, salt, and ash with FOODSCAN 2 Dairy Analyzer (Foss, Hillerøed, Denmark).

2.6. Determination of Sugar and Organic Acid Concentrations

Sugars (lactose, galactose, and glucose) and organic acids (citric, succinic, lactic, pyruvic, formic, acetic, and propionic acid) were extracted from the Kefalotyri samples by following the method described by Bergamini et al. [28]. Briefly, 5 g of cheese were homogenized (Ultra-Turrax blender, Τ25 basic IKA Labor Technik, Breisgau, Germany) with 15 mL of HPLC-grade water, then warmed up to 40 °C and maintained for 1 h. The suspension was centrifuged at 3000× g for 30 min at 4 °C, filtered through Whatman No.1 filter paper, and the supernatant was adjusted with HPLC-grade water to a final volume of 25 mL.

The extracts (20 μL), filtered through a 0.22 μm filter (PTFE, Membrane Solutions), were analyzed on an LC-20AT high-performance liquid chromatograph (Shimadzu, Tokyo, Japan) equipped with a thermostated autosampler (SIL-20A), a high-pressure mixing binary pump (LC-20AT), a CBM-20A controller, a column oven (CTO-20AC), and a diode array detector (SPD-M20A) in-line with a RID-10A refractive index serial detector (Shimadzu, Kyoto, Japan). Separation of the extracts was carried out on a Repromer H column (9 μm, 300 mm × 7.8 mm I.D., Dr Maisch GmbH, Tubingen, Germany) connected with a Repromer H precolumn (9 μm, 30 mm × 4.6 mm I.D., Dr Maisch GmbH) Sugars and organic acids were separated using an isocratic elution program with H₂SO₄0.01 M as the mobile phase at a flow rate of 0.6 mL/min. The column temperature was set at 65 °C and the peaks were detected at 210 nm. The total running time was 35 min, in which the last 10 min were used to equilibrate the detectors for the next injection. Identification of the sugars and organic acids was based on their retention times, according to standard sugar and organic acid curves, respectively. For the standard curves, standard mixtures of sugars and organic acids of 1000–10,000 mg/L were used.

2.7. RP-HPLC Peptide Profiles

The cheese extracts, according to the method of Bergamini et al. [28], were filtered using 0.45 μm PVDF filters (Millipore Corporation, Bedford, MA, USA), and 20 μL were used for HPLC analysis. The HPLC system consisted of an LC-20AT high-performance liquid chromatograph (Shimadzu, Tokyo, Japan), a thermostated autosampler (SIL-20A), a high-pressure mixing binary pump (LC-20AT), a CBM-20A controller, a column oven (CTO-20AC), and a diode array detector (SPD-M20A) in-line with a RID-10A refractive index serial detector (Shimadzu, Kyoto, Japan). Samples were analyzed using a Shim-pack GIST C18 column (3 μm, 100 mm × 3 mm I.D., Shimazdu, Kyoto, Japan). Eluent A was 0.1% (v/v) trifluoroacetic acid (TFA) in HPLC-grade water and eluent B was 0.085% (v/v) TFA in 60:40 (v/v) acetonitrile/HPLC-grade water. Separations were conducted at room temperature at a flow rate of 0.8 mL/min with 100% eluent A for 10 min and a linear gradient from 0% to 80% eluent B for 80 min, and finally with 100% eluent B for 20 min. The absorbance of the eluate was monitored at 214 nm.

The peptide profiles of the cheese samples were evaluated as described by Mallatou et al. [29]. Peptides were divided into two groups based on the elution times of peaks. The hydrophilic peptides consisted of the peaks with retention times from 0 to 67.5 min (0–55% eluent B, the gradient of elution solvent). The hydrophobic peptides were eluted from 67.6 to 110 min (55.1–100% eluent B, the gradient of elution solvent). The ratio of hydrophobic-to-hydrophilic peptides was obtained by dividing the total area of the peaks in the hydrophobic peptide portion by the total area of the peaks in the hydrophilic peptide portion of the HPLC run.

2.8. Volatile Compounds by SPME-GCMS

The determination of volatile compounds (VOCs) in the fully ripened (90-day-old) Kefalotyri C-cheese and N-cheese samples was conducted according to the method described by Thodis et al. [30], using solid-phase microextraction combined with gas chromatography–mass spectrometry (SPME–GC–MS). For each sample, 3 g of homogenized cheese were placed in a 20 mL glass vial together with 10 µL of an internal standard (IS) solution (0.134 mg/L, 4-methyl-2-pentanone). The vials were sealed with Teflon-lined septa and aluminum caps, and the SPME fiber (50–30 nm divinylbenzene–carboxen on polydimethylsiloxane, bonded to a flexible fused silica core; Supelco, Bellefonte, PA, USA) was inserted through the septum. The samples were placed in a 50 °C water bath for 15 min to equilibrate, and then the fiber was exposed to the headspace of the vial for another 15 min to collect volatile compounds. The extracted compounds were analyzed using a 60 m × 320 μm × 1 μm DB-5MS capillary column [(5%–phenyl)-methylpolysiloxane; J&W Scientific, Folsom, CA, USA] installed in an Agilent 7890A gas chromatograph connected to a 5975C mass spectrometer (Agilent Technologies, Wilmington, DE, USA). Helium was used as the carrier gas at a flow rate of 1.5 mL/min, and the injector operated in split mode (2:1) at 260 °C. The oven temperature was programmed as follows: 45 °C for 5 min, increased to 80 °C at 10 °C/min, and then to 240 °C at 5 °C/min, where it was held for 10 min. MS conditions were as follows: source 230 °C; quadrupole 150 °C; electron impact (EI) mode (70 eV) with 3 scans s-1 and the mass range used was m/z: 29–350,ion source and quadrupole temperatures were set at 230 °C and 150 °C, respectively, and the transfer line was kept at 270 °C. Retention indices (RIs) were determined using n-alkane standards (C8–C20; Fluka, Buchs, Switzerland).

2.9. Statistical Analysis

Three independent cheese trials were statistically analyzed (n = 3). Within each trial, the values reported for each microbial and physicochemical parameter of the C- and N-cheese treatments were the means of two individual measurements. The microbial counts were converted to log CFU/g and, along with the data for the physicochemical parameters, were subjected to a one-way analysis of variance using the software Statgraphics Centurion XVII (Statgraphics Technologies, Inc., The Plains, VA, USA). The means were separated by the Least Square Difference (LSD) procedure at the 95% confidence level (p< 0.05) for determining the significance of differences in each cheese treatment with time, and between the two cheese treatments on each sampling day.

3. Results

3.1. Physicochemical and Microbiological Characteristics of Sheep Milk After Thermization and After Inoculation with the Native Starter/Adjunct Strain Combinations

The physicochemical results of the sheep milk samples used for the semi-industrial Kefalotyri cheese processing showed that neither thermization nor the LAB inoculation affected the pH of the three ‘summer’ milk batches, which fluctuated between pH 6.58 and 6.63, irrespective of treatment (

Table 1. Physicochemical analyses of sheep milk from native Epirus breeds used for traditional Kefalotyri cheese production under semi-industrial processing conditions in a local dairy plant.

Milk Type	pH	Milk Composition (%)					FPD 2
		Fat	Protein	Lactose	SRWF 1	Total Solids
Raw milk (RM)	6.58 ± 0.12 a	6.08 ± 0.51 a	5.08 ± 0.16 a	4.53 ± 0.15 a	10.43 ± 0.19 a	16.60 ± 0.47 a	−0.55 ± 0.01 ab
Thermized milk (TM)	6.63 ± 0.07 a	5.66 ± 1.07 a	5.04 ± 0.14 a	4.56 ± 0.21 a	10.46 ± 0.25 a	16.27 ±0.86 a	−0.56 ± 0.01 a
Inoculated milk with control culture (IMC)	6.63 ± 0.04 a	5.61 ± 0.93 a	4.93 ± 0.16 a	4.48 ± 0.21 a	10.22 ± 0.24 a	15.94 ± 0.74 a	−0.54 ± 0.02 ab
Inoculated milk with novel culture (IMN)	6.58 ± 0.07 a	5.76 ± 0.73 a	4.96 ± 0.16 a	4.41 ± 0.10 a	10.20 ± 0.18 a	16.02 ± 0.67 a	−0.52 ± 0.02 b

Values are presented as the means ± standard deviation of three milk batches (n = 3). Statistically significant differences are indicated with different superscripted letters. ¹ Solid-residue-without-fat; ² freezing point depression values.

).Thermization did not affect the FPD values or the fat, protein, lactose, or the solid-residue-without-fat (SRWF) contents and the total solids of the RM either. However, the high standard deviation values of the fat content may reflect compositional variations in the sheep milk from each yard comprising each separate RM batch, or inconsistencies in the milk skimming operation generally applied before hard cheese processing at the SKARFI plant. Indeed, the last raw sheep milk batch (coded RM5; collected from the Filippiada yard by late June) was characterized by a higher fat content (5.8–6.0% fat) after its partial skimming before mild thermization.

Regarding the different microbial populations enumerated in each sheep milk batch before (RM3, RM4, and RM5) and after thermization at 65 °C for 30 s (TM3, TM4, and TM5), the corresponding data were presented and discussed in detail by Samelis et al. [23]. For the research purposes of this study, we should adapt the mean post-thermal contamination levels of the above TM batches before inoculation with the native starter and adjunct LAB strains, which were (log CFU/mL) TVC, 5.10 ± 0.09; total mesophilic LAB, 4.27 ± 0.40; total thermophilic LAB, 3.73 ± 0.26; mesophilic dairy (on M17 agar) LAB, 4.68 ± 0.01; thermophilic dairy LAB (i.e., counted at 42 °C, as noted in the Methods; Section 2.4), 4.50 ± 0.05; enterococci (SB) 3.95 ± 0.25; enterococci plus kanamycin-resistant lactobacilli (KAA) 4.39 ± 0.57; total staphylococci, 1.70 ± 0.20; and coliform bacteria, 0.82 ± 0.38. In all TM batch samples, Pseudomonas-like bacteria and yeasts were <100 CFU/mL, coagulase-positive (RFP+) staphylococci were absent in 1 mL, and Listeria spp./L. monocytogenes and Salmonella were absent in 25 mL of TM after culture enrichment.

Based on the above post-thermal contamination levels, all three TM batches used for Kefalotyri production were microbiologically safe and of good microbial quality, having a total bacterial load of ca. 100,000 CFU/mL, consisting primarily of LAB and comprising similar mesophilic and thermophilic NSLAB levels, following the selection of autochthonous heat-resistant LAB types, mainly Enterococcus spp., by thermization [23].

3.2. Evolution of the Native Starter/Adjunct LAB Strains, Indigenous NSLAB and Non-LAB During Processing, Fermentation, and Ripening of Kefalotyri Cheese Samples

Table 2. Total microbial (TVC) and lactic acid bacteria (LAB) population changes (log CFU/g) in Kefalotyri cheeses from milk inoculation to full maturation ¹.

Cheese Type	Production Stage/Day	TVC (MPCA/ 37 °C)	Total Mesophilic LAB (MRS/30 °C)	Total Thermophilic LAB (MRS/45 °C)	Dairy Mesophilic LAB (M17/22 °C)	Dairy Thermophilic LAB (M17/45 °C)	Enterococci (SB/37 °C)	Lc. lactis subsp. cremoris M78	Lp. plantarum H25 (MRS/30 °C)
C-cheese (control)	Inoculated milk	6.56 ± 0.20 a,A	6.17 ± 0.15 a,A	3.53 ± 0.49 a,A	6.24 ± 0.10 a,A	6.06 ± 0.03 abc,A	3.82 ± 0.25 ab,A	6.21 ± 0.06 bc,A	N/A
	Fresh curd	7.91 ± 0.30 cd,A	7.15 ± 0.26 c,A	3.77 ± 0.18 ab,A	6.85 ± 0.30 b,A	7.67 ± 0.34 e,A	3.65 ± 0.35 a,A	6.81 ± 0.26 de,A	N/A
	1	8.37 ± 0.72 de,A	7.42 ± 0.19 c,A	4.26 ± 0.51 bcd,A	7.41 ± 0.23 de,A	7.29 ± 0.22 de,A	4.24 ± 0.52 abc,A	6.99 ± 0.26 def,A	N/A
	6	8.58 ± 0.15 ef,A	8.24 ± 0.35 def,A	5.01 ± 0.37 ef,A	8.02 ± 0.28 fg,A	7.28 ± 0.23 de,A	4.51 ± 0.17 bcd,A	7.24 ± 0.24 ef,A	N/A
	18	8.81 ± 0.06 ef,A	8.67 ± 0.15 g,A	5.37 ± 0.29 fg,A	8.27 ± 0.15 gh,A	6.90 ± 0.86 bcde,A	5.01 ± 0.45 d,A	7.44 ± 0.27 f,A	N/A
	30	8.45 ± 0.07 ef,A	8.64 ± 0.32 g,A	5.60 ± 0.38 g,A	8.22 ± 0.15 gh,A	6.11 ± 0.66 abc,A	5.09 ± 0.55 d,A	6.89 ± 0.24 de,A	N/A
	90	7.57 ± 0.19 c,A	8.03 ± 0.22 d,A	4.29 ± 0.41 bcd,A	7.16 ± 0.13 bcd,A	6.20 ± 0.55 abc,A	5.18 ± 0.57 d,A	5.90 ± 0.41 b,A	N/A
N-cheese (novel)	Inoculated milk	6.41 ± 0.07 a,A	6.16 ± 0.14 a,A	3.44 ± 0.37 a,A	6.19 ± 0.09 a,A	5.89 ± 0.26 ab,A	3.71 ± 0.16 a,A	5.82 ± 0.18 b,A	5.04 ± 0.71 a
	Fresh curd	7.79 ± 0.17 c,A	6.73 ± 0.15 b,B	4.17 ± 0.19 bc,A	6.73 ± 0.11 b,A	7.64 ± 0.17 e,A	3.80 ± 0.32 ab,A	6.82 ± 0.21 de,A	4.67 ± 0.82 a
	1	8.54 ± 0.35 ef,A	7.36 ± 0.25 c,A	4.47 ± 0.25 cde,A	7.38 ± 0.25 cde,A	6.54 ± 1.20 abcd,A	4.20 ± 0.41 abc,A	6.93 ± 0.20 de,A	6.32 ± 0.44 b
	6	8.83 ± 0.02 ef,A	8.51 ± 0.15 efg,A	4.82 ± 0.35 def,A	8.54 ± 0.14 h,B	7.01 ± 0.32 cde,A	4.56 ± 0.43 cd,A	7.02 ± 0.16 ef,A	8.39 ± 0.15 c
	18	8.88 ± 0.05 f,A	8.62 ± 0.16 fg,A	5.11 ± 0.21 fg,A	8.09 ± 0.62 fg,A	6.71 ± 0.92 abcde,A	4.71 ± 0.61 cd,A	7.13 ± 0.20 ef,A	8.53 ± 0.18 c
	30	8.37 ± 0.27 de,A	8.14 ± 0.20 de,B	5.11 ± 0.16 fg,A	7.73 ± 0.15 ef,B	5.66 ± 0.52 aA	4.71 ± 0.45 cd,A	6.53 ± 0.43 cd,A	7.30 ± 0.42 b
	90	7.09 ± 0.12 b,B	7.44 ± 0.30 c,B	3.41 ± 0.38 a,B	6.97 ± 0.25 bc,A	5.74 ± 0.63 a,A	5.12 ± 0.30 d,A	5.12 ± 0.67 a,B	<5.00/5.65/<5.00 *

¹ Values are presented as the mean ± standard deviation of three Kefalotyri cheese-making trials (n = 3). Lowercase letters indicate statistically significant differences within a column. Uppercase letters within a column highlight the statistically significant differences between the control and novel cheese type on the same production day (p < 0.05).* It was not possible to calculate a mean log CFU/g value ± standard deviation for the strain H25 populations on day 90 because, in the cheese trials N1 and N3, its viability fell below the 5-log level which was the lowest detection limit for the MRS/30 °C agar plating analyses; N/A, not applicable.

shows the total microbial (TVC) and LAB population changes in the semi-industrial Kefalotyri C-cheese and N-cheese products during processing, i.e., from the TM after inoculation (day 0) to the end of cheese maturation (day 90).

As anticipated, the inoculation of both cheese milk types with the basic starter culture (ST1 + M78 + KE109) at ca. 6.0–6.5 CFU/mL for each native strain increased the TVC, the total and dairy mesophilic LAB, and the thermophilic dairy LAB populations by 1–2 log units compared to the corresponding populations in TM (viz. Section 3.1). In contrast, the total thermophilic LAB populations remained at an approximate 1000-fold-lower level (ca. 3.5 log CFU/g) in both inoculated cheese milk types, because none of the starter or the adjunct strains could promote growth at 45 °C on MRS agar, as reported in Section 2.4. Consequently, the levels of total autochthonous thermophilic LAB and enterococci nearly matched in all inoculated cheese milks (Table 2 and Table S1) due to the enhanced selective survival of Enterococcus spp. in TM [23] and the absence of native thermophilic starter lactobacilli (i.e., Lb. delbrueckii or Lb. helveticus) strains. Lastly, the inoculation of the N-cheese milks with the adjunct culture did not further increase the TVC or total mesophilic LAB counts (Table 2) because the adjunct strains H25, KFM7, and KFM9 were intentionally seeded at an approximate 10-fold-lower level in TM.

After cooking (day 0), major (ca. 1.5 log units) increases in the TVCs and the total dairy thermophilic LAB populations were noted in all samples, reflecting the prevalent growth of the S. thermophilus ST1 starter in the fresh cheese curds (Table 2). The increases in the total mesophilic LAB populations were ca. 0.5–1.0 log lower than ST1, presumably due to significant growth in the NisA+ M78 and KE109 starter strains in all fresh curds. In contrast, neither enterococci nor the adjunct strain H25 grew significantly in any of the fresh curds (Table 2 and Table S1), indicating that the competition by the growing basic starter strains, mainly ST1, was strong enough to prevent an outgrowth of adventitious enterococci highly selected in the milk post-thermally [23]. The growth of indigenous thermophilic lactobacilli was also restricted (Table 2), whereas indigenous mesophilic, kanamycin-resistant LAB increased 10-fold in the fresh C-cheese curds only (Table S1).

After the first 24 h of fermentation, TVCs exceeded 8 log CFU/g in all cheeses, except for the C1-cheese, which underwent a slower fermentation until day 6 and before ripening (data not tabulated separately). However, at mid-ripening (day 18), all cheeses had an almost 9 log CFU/g TVC which decreased slightly at the end of ripening (day 30) and drastically after an additional two-month maturation at 4 °C (Table 2).

Selective LAB enumerations revealed that the primary aciduric thermophilic starter strain ST1 failed to predominate in both fresh cheeses after the first 24 h of fermentation (day 1), despite the fact that it grew earlier and faster in all curds compared to the mesophilic starter (M78 + KE109) and adjunct (H25 + KFM7 + KFM9) strains, which were enumerated more representatively as the total mesophilic LAB fraction on MRS agar at 30 °C (Table 2).

Consequently, strain ST1 failed to exceed 8 log CFU/g in all cheeses, while from day 18 to full ripening (day 90) it was essentially lost and replaced mainly by enterococci on M17 agar at 45 °C. Enterococci also grew selectively at very similar levels of 4.5 to 5.5 log CFU/g on MRS/45 °C SB (Table 2) and KAA (Table S1) agar plates, but they failed to increase above 6 log CFU/g in all cheeses throughout processing. Evidently, enterococci were suppressed by the prevalent competitive growth of the starter strains M78 and KE109 from day 1 to 18, and the gradual prevalence of the adjunct strain H25 in the N-cheeses during brining to the middle of ripening (Table 2 and Table S1). In particular, the NisA+ strain M78 gradually evolved well above 7 log CFU/g until day 18 of ripening in all cheeses, but afterward its viable counts showed drastic decreases (Table 2), possibly due to autolysis or because M78 was overgrown by salt-tolerant, kanamycin-resistant, mesophilic NSLAB, especially in the C-cheeses (Table S1). Correspondingly, in the N-cheeses, a strong (>8.3 log CFU/g) prevalence of the kanamycin-resistant adjunct strain Lp. plantarum H25 was established until the middle of ripening (day 18) (Table 2 and Table S1).

However, after one month of ripening at 16–17 °C, the dominant mesophilic LAB, including M78 and H25, populations showed drastic decreases, which maximized after cold ripening at 4 °C for an additional 2 months, correspondingly to the decreases in TVC. The most impressive and quite unexpected decrease was that of strain H25, whose viability fell below 5.0 log CFU/g in the N1 and N3 cheeses on day 90; viable H25 counts (5.65 ± 0.49 log CFU/g) were detected in the N2-cheese samples only (Table 2 and Table S1).

Meanwhile, compared to the C-cheeses, the N-cheeses developed numerous holes (eyes) by the middle (day 18) to the end (day 30) of ripening, suggesting that sufficient subdominant growth of the gas-forming KFM7 + KFM9 adjunct strains occurred in them. To validate this, 10 typical whitish (non-H25) LAB colonies randomly picked from the highest (−7) dilution MRS/30 °C agar plates after counting of the day 18 and day 30 N1, N2, and N3 cheese samples were cultured separately in 10 mL sterile, modified MRS broth (i.e., with sodium citrate replacing ammonium citrate) vials with inverted Durham tubes at 30 °C. Only the colony cultures that accumulated gas (CO₂) after 24 to a maximum of 48 h of incubation were further tested microscopically and for vancomycin (30 μg/disk) susceptibility; i.e., an inherent antibiotic resistance of Leuconostoc spp. [31]. On each testing occasion, 1 to a maximum of 3 out of the 10 mesophilic N-cheese (MRS) colonies were obligatory heterofermentative, vancomycin-resistant coccoid LAB; therefore, they were considered as representatives of either the KFM7 or KFM9 strain inocula. Thus, although accurate selective enumerations of Leuconostoc spp./Ln. mesenteroides were not feasible throughout processing, good evidence was provided that the joint KFM7 + KFM9 adjunct strain populations increased to ca. 7.0–7.5 log CFU/g levels in the ripening N-cheeses.

Regarding the evolution of the indigenous non-LAB biota during fermentation and ripening, it was suppressed by the dominant LAB growth; therefore, the corresponding data are not included in Table 2. Only a gradual increase in total staphylococci and coliforms in the fresh cheese (day 1) and at the beginning (day 6) to the middle of ripening (day 18) occurred. Specifically, in the C-cheeses, staphylococci increased from 2.49 ± 0.61 log CFU/g in the fresh cheese (day 1) to 5.18 ± 0.27 log CFU/g by day 18, then decreased to 3.49 ± 0.94 log CFU/g by day 30 and finally fell to <100 CFU/g in the fully ripened cheese (day 90). The growth profile of total staphylococci was similar in the N-cheeses, where their respective populations were 3.28 ± 0.92 log CFU/g (day 1), 5.14 ± 0.42 log CFU/g (day 18), 3.84 ± 1.13 log CFU/g (day 30), and <100 CFU/g (day 90).

Similarly, coliform bacteria, which were always below 10 cells/mL in the inoculated milk, increased to 2.85 ± 0.64 log CFU/g and 3.71 ± 0.39 log CFU/g in the fresh (day 1) C-cheeses and N-cheeses, respectively, remained at approximately the same levels, 3.10 ± 0.55 log CFU/g (C-cheeses) and 3.22 ± 0.68 log CFU/g (N-cheeses) until the beginning of ripening (day 6), decreased to 1.88 ± 0.76 log CFU/g (C-cheeses) and 2.15 ± 0.76 log CFU/g log (N-cheeses) at the end of ripening (day 30), and fell below 10 cells/g in all final cheeses after cold ripening for an additional two months at 4 °C (day 90).

The psychrotrophic aerobic Pseudomonas bacteria and yeasts remained at <1000 cells/g during fermentation and ripening of both cheese types. Pathogenic staphylococci never exceeded 100 cells/g in all cheese samples from day 0 to day 90, without exception. In general, except for LAB, none of the other categories of microorganisms studied were detected in populations exceeding 100 cells/g on day 90. Also, all cheeses were free of L. monocytogenes and Salmonella spp. at the start (day 6) and the end (day 30) of ripening.

3.3. pH and Gross Composition of the Kefalotyri Cheeses

The physicochemical characteristics of the Kefalotyri cheese samples at various stages of production and maturation are presented in

Table 3. Physicochemical characteristics of Kefalotyri cheeses at different production stages.

Cheese Type	Day	Fat (%)	Moisture (%)	Protein (%)	Salt (%)	SFA (%)	FDM (%)	TS (%)	pH
Control (C)	1	30.98 ± 1.13 a,A	42.77 ± 0.74 d,A	0.52 ± 0.14 a,A	0.52 ± 0.14 a,A	20.14 ± 0.91 ab,A	54.07 ± 1.35 b,A	57.23 ± 0.74 a,A	5.73 ± 0.11 b,A
	6	33.38 ± 0.93 bc,A	36.09 ± 0.31 bc,A	3.19 ± 0.11 cd,A	3.19 ± 0.11 cd,A	20.92 ± 0.48 bc,A	52.23 ± 1.17 ab,A	63.91 ± 0.31 bc,A	5.52 ± 0.15 a,A
	30	34.94 ± 1.59 c,A	34.78 ± 2.16 ab,A	2.76 ± 0.29 bc,A	2.76 ± 0.29 bc,A	22.19 ± 0.82 d,A	53.57 ± 0.90 ab,A	65.22 ± 2.16 cd,A	5.38 ± 0.10 a,A
	90	35.50 ± 1.24 c,A	33.17 ±0.81 a,A	3.42 ± 0.09 de,A	3.42 ± 0.09 de,A	21.52 ± 0.50 cd,A	53.10 ± 1.21 ab,A	66.83 ± 0.81 d,A	5.46 ± 0.14 a,A
Novel (N)	1	30.65 ± 1.44 a,A	43.50 ± 0.73 d,A	0.56 ± 0.05 a,A	0.56 ± 0.05 a,A	19.66 ± 0.59 a,A	54.23 ± 1.86 b,A	56.50 ± 0.73 a,A	5.73 ± 0.15 b,A
	6	32.61 ± 1.54 ab,A	36.45 ± 0.49 bc,A	3.44 ± 0.29 de,A	3.44 ± 0.29 de,A	20.48 ± 0.84 abc,A	51.30 ± 2.07 a,A	63.55 ± 0.49 bc,A	5.48 ± 0.09 a,A
	30	33.55 ± 1.56 bc,A	36.72 ± 1.54 c,B	2.48 ± 0.72 b,A	2.48 ± 0.72 b,A	21.40 ± 0.56 cd,A	53.03 ± 2.24 ab,A	63.28 ± 1.54 b,B	5.34 ± 0.03 a,A
	90	35.19 ± 0.71 c,A	33.67 ± 0.23 a,A	3.95 ± 0.46 e,A	3.95 ± 0.46 e,A	21.01 ± 0.28 bc,A	53.07 ± 1.24 ab,A	66.33 ± 0.23 d,A	5.42 ± 0.04 a,A

Values are presented as the means ± standard deviation of three cheese trials (n = 3). Lowercase letters indicate statistically significant differences within a column. Uppercase letters within each column highlight statistically significant differences between the control and novel cheese types on the same production day (p < 0.05). FDM, fat-in-dry matter; TS, total solids.

. Consistent with the major growth increases in the native starter LAB, mainly strain ST1, during the first 24 h of fermentation (Table 2), the fresh (day 1) cheeses had a mean pH of 5.73, which was lower (p < 0.05) than the TM pH after inoculation (IMC pH 6.63 and IMN pH 6.58; Table 1). This major (ca. one pH-unit) reduction reflected the starter LAB acidification process, which was progressive, from the milk curdling (pH 6.52–6.54) to the second pressing (pH 5.67–6.19) on the previous day of Kefalotyri manufacture. No significant differences in the pH reduction pattern during fermentation were observed between the C-cheese and the N-cheese samples, a result indicating that the adjunct culture (H25 + KFM7 + KFM9) had minor, if any, effects on milk acidification. During brining and dry salting in the ripening room, the pH of both cheese types decreased further, reaching its lowest value range (pH 5.3–5.4) by the middle of ripening (day 18). Thereafter, the pH of both cheese types remained stable, and it was about 5.4 after 90 days (Table 3), assuming that the adjunct culture did not significantly affect the pH of the mature Kefalotyri cheese.

The gross composition of the C-cheese and the N-cheese samples after 90 days of ripening did not differ significantly (Table 3). Only the moisture (%) and TS (%) varied between the two cheese types after 30 days of ripening, with the N-cheese displaying a ca. 2% higher moisture retention than the control C-cheese without the adjunct culture (p < 0.05). Significant changes in each chemical parameter were noted in both cheese types across processing: moisture decreased while fat, protein, and total solids increased from day 1 to day 90 (Table 3). Also, the salt content increased by 2.5–3.0% after brining in all cheese samples. However, remarkable salt (%) variations were noted between cheese trials or individual cheese molds within each trial, irrespective of the type of LAB culture added, a result suggesting that the traditionally used dry salting process was fairly unbalanced in stabilizing the salt content.

3.4. Sugar and Organic Acid Concentrations

The amounts of sugars found in the Kefalotyri cheese samples from day 1 to day 90 are shown in

Table 4. Changes in basic milk sugars during the production and ripening of Kefalotyri cheeses.

Cheese Type	Production Day	Lactose mg/g Cheese	Glucose mg/g Cheese	Galactose mg/g Cheese
Control (C)	1	15.06 ± 1.33 cde,A	14.13 ± 4.96 ab,A	22.15 ± 13.19 bc,A
	6	15.28 ± 2.94 de,A	17.30 ± 1.75 ab,A	7.38 ± 12.79 cb,A
	30	11.85 ± 0.90 bcd,A	18.67 ± 1.79 ab,A	0.11 ± 0.20 a,A
	90	9.16 ± 1.71 b,A	12.12 ± 3.04 a,A	3.31 ± 4.56 a,A
Novel (N)	1	18.91 ± 1.55 e,A	18.83 ± 5.65 b,A	26.36 ± 13.44 c,A
	6	11.13 ± 4.79 bc,B	17.68 ± 5.15 ab,A	11.93 ± 10.65 abc,A
	30	0.96 ± 1.66 a,B	14.26 ± 4.18 ab,A	0.63 ± 1.09 a,A
	90	0.50 ± 0.86 a,B	15.83 ± 0.96 ab,A	0.00 ± 0.00 a,A

Values are presented as the means ± standard deviation of three cheese trials (n = 3). Lowercase letters indicate statistically significant differences within a column. Uppercase letters within each column highlight the statistically significant differences between the control and novel cheese types on the same production day (p < 0.05).

. Overall, as was expected, lactose was gradually decomposed to glucose and galactose and concomitantly almost completely consumed by the dominant LAB biota after 90 days. However, there were significant differences in the residual lactose concentrations between the two Kefalotyri cheese types by the end of ripening (day 30) and in the final mature cheese (day 90), as only trace amounts of lactose were found in the N-cheeses produced with the adjunct LAB strain (H25 + KFM7 + KFM9) combination. Similarly, galactose was not detected or was contained in small quantities in both cheese types after 90 days. Glucose, on the other hand, was neither depleted nor replenished due to lactose decomposition (Table 4), possibly reflecting the preferential catabolism of galactose or the effect of factors such as salt-to-moisture ratio and calcium and phosphorus contents, which have been reported to affect glucose catabolism [32].

Table 5. Organic acid concentrations (mg/g of cheese) in Kefalotyri cheese samples during fermentation and ripening.

Cheese Type	Production Day	Citric	Succinate	Lactic	Pyruvic	Formic	Acetic	Propionic
Control (C)	1	4.14 ± 0.44 c,A	0.82 ± 0.02 a,A	27.11 ± 4.50 ab,A	1.28 ± 0.12 a,A	1.82 ± 1.64 b,A	2.26 ± 0.28 a,A	1.38 ± 1.23 ab,A
	6	5.44 ± 1.60 cd,A	0.88 ± 0.07 a,A	30.46 ± 9.54 b,A	1.20 ± 0.59 a,A	0.92 ± 0.50 ab,A	3.04 ± 2.66 ab,A	1.78 ± 1.91 ab,A
	30	3.98 ± 0.87 bc,A	0.84 ± 0.03 a,A	28.25 ± 7.11 ab,A	3.90 ± 3.34 b,A	1.65 ± 0.99 ab,A	4.96 ± 0.68 b,A	1.77 ± 1.21 ab,A
	90	2.19 ± 1.90 ab,A	0.52 ± 0.45 a,A	19.38 ± 2.77 a,A	1.67 ± 1.01 a,A	0.31 ± 0.28 a,A	1.01 ± 1.23 a,A	2.66 ± 2.83 abc,A
Novel (N)	1	6.76 ± 0.52 d,B	0.96 ± 0.04 a,A	43.73 ± 1.40 c,B	0.79 ± 0.24 a,A	1.17 ± 0.49 ab,A	3.55 ± 0.48 ab,A	0.37 ± 0.24 a,A
	6	4.34 ± 1.50 c,A	0.94 ± 0.10 a,A	25.98 ± 6.38 ab,A	1.24 ± 0.59 a,A	0.66 ± 0.10 ab,A	1.26 ± 0.30 a,A	0.80 ± 0.58 ab,A
	30	1.28 ± 0.60 a,B	1.90 ± 1.05 b,B	27.66 ± 7.53 ab,A	2.59 ± 0.28 ab,A	1.43 ± 0.47 ab,A	5.23 ± 2.97 b,A	3.23 ± 1.92 bc,A
	90	1.20 ± 0.23 a,A	1.39 ± 0.95 ab,A	24.25 ± 3.97 ab,A	1.22 ± 0.49 a,A	1.26 ± 0.57 ab,A	5.07 ± 0.95 b,B	4.85 ± 0.30 c,A

Values are presented as the means ± standard deviation of three cheese trials (n = 3). Lowercase letters indicate statistically significant differences within a column. Uppercase letters within each column highlight the statistically significant differences between the control and novel cheese types on the same production day (p < 0.05).

shows that the concentration of all studied organic acids remained stable throughout ripening except for citric acid, whose concentration decreased over time in both cheese (C and N) types; conversely, the concentrations of propionic and succinic acids increased in the N-cheeses after ripening. Overall, the concentrations of organic acids presented no statistically significant differences between the two types of Kefalotyri cheeses after 90 days of ripening, apart from acetic acid, whose concentration was higher (p < 0.05) in the N-cheese type. As expected, lactate was the most abundant acid throughout cheese processing, ranging from 6- to 20-fold-higher levels than the other acids, and with its highest concentration (p < 0.05) recorded in the N-cheeses after the first 24 h of fermentation (Table 5).

3.5. Peptide Profile of Kefalotyri Cheeses

The peptide profiles of Kefalotyri cheese samples across production (from day 1 to day 90) are presented in

Table 6. Peptide profile in Kefalotyri cheeses during fermentation and ripening for 90 days after production.

Cheese Type	Production Day	HO	HI	HO/HI	Part 0	Part I	Part II	Part III
C-cheese (control)	1	32.19 ± 12.77 bc,A	67.81 ± 12.77 ab,A	0.51 ± 0.26 bc,A	37.02 ± 2.29 a,A	11.89 ± 2.09 a,A	21.29 ± 11.64 b,A	29.80 ± 14.43 b,A
	6	34.96 ± 14.37 c,A	65.04 ± 14.37 a,A	0.58 ± 0.31 c,A	43.40 ± 14.21 ab,A	10.58 ± 4.29 a,A	14.29 ± 2.53 ab,A	31.73 ± 13.75 b,A
	30	17.81 ± 3.54 a,A	82.19 ± 3.54 c,A	0.22 ± 0.05 a,A	49.80 ± 3.75 bcd	20.23 ± 2.58 bc,A	14.50 ± 2.02 ab,A	15.47 ± 2.56 a,A
	90	20.91 ± 4.58 ab,A	79.09 ± 4.58 bc,A	0.27 ± 0.08 ab,A	55.97 ± 5.13 cd,A	15.60 ± 1.38 ab,A	13.11 ± 1.91 ab,A	15.33 ± 4.56 a,A
N-cheese (novel)	1	40.67 ± 4.54 c,A	59.33 ± 4.54 a,A	0.69 ± 0.13 c,A	32.54 ± 4.35 a,A	10.79 ± 3.24 a,A	19.99 ± 4.74 ab,A	36.68 ± 4.08 b,A
	6	36.73 ± 2.81 c,A	63.27 ± 2.81 a,A	0.59 ± 0.07 c,A	39.69 ± 1.30 ab,A	10.71 ± 2.80 a,A	16.93 ± 1.73 ab,A	32.67 ± 2.20 b,A
	30	16.96 ± 4.30 a,A	83.04 ± 4.30 c,A	0.21 ± 0.06 a,A	44.09 ± 9.72 abc,A	23.45 ± 2.70 c,A	19.70 ± 6.09 ab,A	12.76 ± 3.47 a,A
	90	14.94 ± 1.37 a,A	85.06 ± 1.37 c,A	0.18 ± 0.02 a,A	60.83 ± 4.31 d,A	17.40 ± 3.89 b,A	11.41 ± 1.41 a,A	10.35 ± 0.51 a,A

Values are presented as the means ± standard deviation of three cheese trials (n = 3). Lowercase letters indicate statistically significant differences within a column. Uppercase letters within each column highlight the statistically significant differences between the control and novel cheese types on the same production day (p < 0.05). HO (% TA = total area): hydrophobic, HI (% TA): hydrophilic, Part 0: 0–10 min, Part I: 10–40 min, Part II: 40–70 min, and Part III: 70–110 min.

. With RP-HPLC, the different-sized peptides and free amino acids contained in the water-soluble extract of cheeses are “classified” in relation to their hydrophobicity and size. Free amino acids and small–medium hydrophilic peptides are usually eluted in the first parts of the chromatograms, while in the second medium–large hydrophobic peptides as well as serum proteins are eluted. The large peaks in the 80–100 min area correspond to serum proteins. The quantitative differences between the profiles were determined based on the area of their distinct areas and the ratio of the 10–55 and 55–100 min areas. The ratio between the surfaces of these two regions is called the ratio of hydrophobic (HO)-to-hydrophilic (HI) peptides, and it is a ripening indicator for cheeses. In the present study, a decrease in hydrophobic peptides and an increase in hydrophilic groups over the ripening time of all Kefalotyri cheese trials were observed (Table 6). The HO/HI ratio decreased similarly in both cheese types during ripening.

3.6. Volatile Compounds Profile of the Kefalotyri Cheeses After Maturation for 90 Days

The results from the SPME- GCMS analysis are presented in

Table 7. Volatile profiles of the fully ripened (90-day-old) Kefalotyri cheeses.

Retention Time	RIexp *	RIlit **	Volatile Compound	Cheese (C) Ppm	Cheese (N) Ppm
			Alcohols
4.003	<500	427	Ethyl Alcohol	0.214 ± 0.19 a	0.294 ± 0.05 a
5.635	538	554	1-Propanol	0.008 na	0.004 ± 0.01 na
8.498	650	669	1-Butanol	0.018 na	0.018 ± 0.01 na
10.993	730	735	3-methyl-Butan-1-ol	0.071 ± 0.008 a	0.06 ± 0.01 a
			Ketones
4.481	<500	500	2-Propanone (Acetone)	1.976 ± 0.280 a	1.532 ± 0.65 a
6.385	574	580	2,3-Butanedione	0.039 na	ND
6.551	613	622	2-Butanone	0.012 na	ND
16.866	881	888	2-Heptanone	0.123 ± 0.071 a	0.069 ± 0.01 a
24.072	1092	1095	2-Nonanone	0.053 ± 0.02 a	0.018 ± 0.01 a
			Ethers
4.655	<500	485	Ethyl ether	0.340 ± 0.02 a	0.351 ± 0.06 a
			Esters
5.077	512	522	Methyl acetate	0.025 ± 0.015 a	0.019 ± 0.01 a
6.979	601	610	Ethyl acetate	0.020 ± 0.004 na	0.034 na
7.485	610	621	Methyl priopionate	0.030 na	ND
10.518	710	735	Methyl butyrate	0.758 ± 0.76 na	0.418 ± 0.091 na
13.408	790	798	Ethyl butyrate	0.109 ± 0.088 a	0.168 ± 0.04 a
18.082	792	798	Methyl hexanoate	0.262 ± 0.314 a	0.087 ± 0.001 a
20.757	980	996	Ethyl hexanoate	0.070 ± 0.044 a	0.070 ± 0.02 a
25.063	1113	1120	Methyl caprylate	0.110 ± 0.160 na	0.015 ± 0.01 na
27.356	1182	1193	Ethyl octanoate	0.015 na	0.008 na
30.869	1320	1324	Methyl decanoate	0.144 na	ND
			Alkane
5.862			2-methyl-Pentane	0.06 ± 0.01 a	0.05 ± 0.006 a
6.615	592	600	Hexane	0.043 ± 0.025 na	0.053 na
9.748	701	700	Heptane	0.983 ± 0.18 a	0.888 ± 0.101 a
13.474	799	800	Octane	0.018 na	ND
			Acids
6.23	570	595	Acetic acid	0.241 ± 0.190 a	0.440 ± 0.12 a
11.272	761	753	Isobutyric acid	0.064 ± 0.015 na	0.121 ± 0.06 na
12.682	780	784	Butanoic acid	0.922 ± 0.713 a	1.548 ± 0.26 a
14.428	830	838	3-Methylbutanoic acid	0.168 ± 0.148 na	ND
14.556	880	878	Pentanoic acid	0.194 na	0.308 ± 0.097 na
14.887	891	898	2-Methylbutanoicacid	0.116 ± 0.055 a	0.149 ± 0.053 a
19.565	960	965	Caproic acid	0.207 ± 0.164 na	0.154 ± 0.109 na
			Aldehydes
8.366	640	650	3-methylbutanal	0.006 na	ND
			Miscellaneous
11.597	740	747	Dimethyl disulfide	0.004 na	ND
17.271	890	895	Styrene	0.042 ± 0.003 a	0.035 ± 0.005 a
18.381	801	-	1,3-Dihydroxy-6-methoxy-1,2,3,4,-tetrahydroquinolin-2-one	0.013 na	ND
18.925	937	940	a-pinene	0.016 ± 0.002 a	0.012 ± 0.001 a
22.329	1030	1039	dl-Limonene	0.005 ± 0.001 na	ND

Values are presented as the means ± standard deviation of three cheese trials (n = 3). Lowercase letters indicate statistically significant differences between the two cheese types (p <0.05); ND, not detected; na, statistical analysis was not applicable since this compound was only detected in 1 or 2 cheese batches, * experimental retention indices values based on the calculations using the standard mixture of alkanes; and ** retention indices of the identified compounds according to the literature data cited in the NIST MS library.

. In general, there were remarkable differences in the volatile profiles between the two Kefalotyri cheese types. Qualitatively, most of these differences were attributed to compounds that were not detected in one cheese type but were detected in the other type, as well as to compounds that were not detected in all batches of the same cheese (C or N) type. The compounds that presented both features were exclusively detected in the C-cheeses and were 2,3-butanedione, 2-butanone, methyl propionate, methyl decanoate, octane, 3-methyl butanoate, 3-methyl butanal, dimethyl sulfide, 1,3-Dihydroxy-6-methoxy-1,2,3,4,-tetrahydroquinolin-2-one, and dl-limonene. The presence of methyl octanoate was not consistent in all batches of N-cheese, while 1-propanol, ethyl hexanoate, pentanoic acid, 2-methyl butanoate, and hexanoic acid were not detected in all batches of C-cheese. Similarly, 1-butanol, ethyl acetate, ethyl octanoate, hexane, and 2-methyl propanoate were not detected in all batches of both cheese types. From a quantitative perspective, and when statistical analysis was enabled (i.e., for the compounds that were present in all batches of both cheese types), no statistically significant differences were detected.

4. Discussion

This study assessed traditional technological methods, i.e., mild thermization of raw milk and the application of native starter and adjunct cultures, as alternatives for the production of Kefalotyri under semi-industrial conditions. Three independent artisan-type cheese trials were made from thermized milk (65 °C; 30 s) from native Epirus sheep breeds and were fermented with the aid of S. thermophilus and two wild L. lactis starter strains (C-cheese), supplemented, for the first time in Greece, with an autochthonous adjunct culture (N-cheese) consisting of a Lp. plantarum hard cheese-ripening strain and two atypical raw sheep milk strain genotypes of Ln. mesenteroides. Both Kefalotyri cheese product types were analyzed and compared microbiologically, as well as for their pH, gross composition, main sugar and organic acid contents, and peptide profiles across processing, and for their VOC profiles at full maturation (90-day-old cheeses).

The gross composition of both cheese (C and N) types at full maturation was within the limits defined for traditional Greek hard cheeses of first quality (moisture < 38%; FDM > 40%). According to the Greek Codex Alimentarius [14], all final cheeses were graded of ‘excellent quality’ in terms of having a moisture content < 35% and FDM > 47%, while most of them had already attained this high grade after 30 days of ripening. The physicochemical characteristics of the fully ripened (90-day-old) control (C) and novel (N) Kefalotyri cheeses are similar to or even better than those reported in the literature [1,6,7,10,12,13,16,33]. More specifically, their final moisture values of 33.17–33.67% recorded after 90 days of ripening (Table 3) are lower than the moisture values of 34.77–36.53% of the ripened (90-day-old) Kefalotyri experimental cheeses produced from pasteurized sheep milk with commercial starters [12], or of the 36.8% [7] and 35.8–37.2% moisture values of commercial (90-day-old) Kefalotyri cheeses tested recently [6,13,16]. Differences in moisture can be attributed to various technological factors, including the pressing and salting methods employed for moisture removal from the cheese molds. For instance, the high (40.4%) moisture of the mature (90-day-old) artisanal Kefalotyri of Pindos cheese from raw sheep milk without starters was attributed, among others factors, to the old manual pressing method [10]. Similarly, the final moisture of artisanal “Orinotyri” cheese from raw sheep milk without starters fluctuated at 36.4 ± 3.05% [9]. Accordingly, the fat content of 35.19–35.50% of the present Kefalotyri cheeses after 90 days of ripening (Table 3) is higher than the corresponding fat content of 28.7–32.09% reported in most of the above studies [1,7,10,12,13,16] and can be assigned partly to the reduced moisture and partly to the sheep breed and their nutritional management [34,35].

Conversely, the final salt content of 3.42–3.95% of the cheeses (Table 3) was within the 3.1–4.1% salt range of ripened (90-day-old) Kefalotyri cheeses recorded in previous studies [1,6,7,10,12]. Although the mean salt differences in the same-age cheeses were not significant due to dry salting fluctuations between trials (viz. Results), the N-cheeses tended to be saltier than the control C-cheeses after brining (day 6) and ripening (day 90) (Table 3). The intermediate (day 30) reductions in the mean salt content (i.e., 2.48–2.76%) in both cheese types was likely due to salt redistribution and binding within the matrix during proteolysis [36]. Overall, salt variations are normal as different salting methods (brining and variable dry saltings) are involved in Kefalotyri processing, aiming to remove moisture and preserve the cheeses by high SM (salt-in-moisture %) values [7,8,9].

Regarding the pH of mature (90-day-old) Kefalotyri cheese, although the literature consistently reports pH values ranging from 4.78 to 5.30 [1,6,7,10,12,13,16,33], in this study, the respective pHs of the semi-industrial cheeses were remarkably higher, at 5.42 to 5.46, which possibly reflects the inferior acidification capacity of the total native LAB biota during the present fermentation process. Even the artisan Kefalotyri cheeses from raw sheep and goat milk without starters produced in Pindos [10] or Naxos [33] were recorded to have pH values of 5.2 and 4.95–5.2, respectively, after 90 days of ripening. Only the artisanal “Orinotyri” of Epirus from raw sheep milk had a high pH, 5.79 ± 0.28, after 90 days [9], consistently with the technological bulletin for the traditional Kefalotyri by Zerfyridis [8], who stressed the gradual increase in pH from 5.0 to 5.3 in the fermenting cheese (day 1 to 15) to 5.6 and 5.8 after 1 and 3 to 6 months of ripening, respectively. Similarly, Pappa et al. [10] reported a late increase in the pH to 5.7 in artisanal Kefalotyri cheese after 180 days of ripening. Generally, the pH increases during hard cheese ripening are attributed to proteolysis along with reductions in acidity, because significant amounts of the lactate that accumulates post-fermentation are catabolized to serve as carbon energy sources for LAB (mainly NSLAB), surface bacteria, or yeasts [8,37].

The NSLAB evolution in the present Kefalotyri cheeses, including the enterococci in both cheeses and, mainly, the adjunct H25 and KFM7 + KFM9 strains in the N-cheeses, was in general accord with the literature. NSLAB constitute an essential, occasionally predominant, portion of the microbiota in traditional European hard cheeses produced from RM or TM with added natural (NSC; yogurt, whey) or commercial (CSC) starter cultures [37,38], including the best-known Greek cooked hard cheeses, Graviera (PDO and non-PDO varieties) [25,39], Kefalograviera PDO [40], and Kefalotyri [19,33]. Apart from thermoduric, salt-tolerant, non-aciduric enterococci, NSLAB comprise facultatively heterofermentative members of the genera Lactiplantibacillus (Lp. plantarum, Lp. pentosus), Lacticaseibacillus (Lc. casei, Lc. paracasei, Lc. rhamnosus), Latilactobacillus (Lt. curvatus), and less frequently, Pediococcus (P. pentosaceous), as well as obligatory heterofermentative members of the genera Levilactobacillus (Lv. brevis), Leuconostoc (Ln. mesenteroides, Ln. lactis), and to a lesser extent, Lentilactobacillus (Lnt. buchneri, Lnt. parabuchneri) and Weissella (W. paramesenteroides) [2,25,31,37,38,39,40]. All of them are derived from raw milk or the creamery environment, and most of them are aciduric, salt-tolerant, can grow at 42 °C but not at 45 °C, and survive thermization less prominently than enterococci [22,25,37,38,40]. Specifically, a recent MALDI-TOF profiling analysis of the NSLAB, grown on Rogosa agar at 30 °C, identified Lv. brevis as the most frequent (23/46 isolates) species in the Kefalotyri of Naxos, followed by Lp. plantarum (12/46) and Ln. mesenteroides (8/46) [33]. Moreover, numerous Lactiplantibacillus (Lp. plantarum) and Levilactobacillus (Lv. brevis) strains are kanamycin-resistant and thus interfere with the enumeration of Enterococcus spp. on KAA agar [23], as also occurred in this study, (Table S1).

Generally, mesophilic non-starter lactobacilli, primarily Lp. plantarum, Lc. paracasei, and Lv. brevis, can promote significant ‘secondary’ slow growth increases in the (cooked) hard cheese interior to succeed the starter LAB and numerically dominate in the final product. Similarly to the starter LAB at earlier ripening days, the viability of NSLAB may decline 100-fold or more with prolonged ripening due to autolysis and the combined stresses imposed by the hostile cheese environment, especially when sheep milk is used for making cheese. For instance, Gantzias et al. [33] found both 90-day-old Kefalotyri of Naxos (Damarionas area) samples solely made of sheep milk to have NSLAB (mainly Lp. plantarum and Lv. brevis) counts as low as 4.52–5.91 log CFU/g, whereas the respective counts of the remaining six Kefalotyri samples from mixed sheep and goat milk were much higher, at 6.8 to 7.92 log CFU/g. The sole use of sheep milk that contains more fat and total solids than cow and goat milk (thus yielding cheeses of higher quality) may lead to more stressful hard cheese interiors for LAB survival after moisture removal during ripening, especially when the cheese is salted with 3–4% salt to be preserved by the SM% content, like traditional Kefalotyri [8]. Thus, the major (≥3-log) declines in Lp. plantarum H25, as well as the greater (p < 0.05) declines in L. lactis M78, total mesophilic LAB, TVCs, and total thermophilic LAB, except of enterococci, in the saltier N-cheeses compared to the C-cheeses on day-90 (Table 2) are consistent with the literature.

A slow but prevalent growth of non-starter lactobacilli in hard cheese occurs even if their counts in raw or thermized milk and curd are low (10–1000 CFU/mL or g), as has been addressed in a concise review by Bottari et al. [41] with the allegorical title “How the fewer become the greatest”, focused on Lc. casei/paracasei impact on long-ripened cheeses. Their growth occurs primarily at the expense of galactose after lactose depletion by the starter LAB accompanied or followed by lactate and citrate catabolism [41,42]. Additionally, the strong in situ proteinase and/or peptidase activities by most facultative or obligatory heterofermentative lactobacilli contribute to their slow growth and long-term survival in hard cheeses [43]. Accordingly, the higher uptake of lactose (p < 0.05) and galactose, close to depletion, after 30 and 90 days of ripening in the N-cheeses (Table 4), and the higher (p < 0.05) acetic acid content of the N-cheeses compared to the C-cheeses on day-90 (Table 5) can be attributed to the predominant growth and glycolytic activities of the H25 adjunct strain. Moreover, compared to the C-cheeses, the lower (p < 0.05) citrate and higher (p < 0.05) succinate concentrations of the N-cheeses after 30 days of ripening might also associate with the H25 catabolic activity [28], whereas an early potential growth of the Ln. mesenteroides adjunct strains might have contributed to the higher (p < 0.05) citrate concentration in the fresh N-cheeses [31] compared to C-cheeses on day 1 (Table 5).

To date, the major biotechnological importance of NSLAB, mainly Lactiplantibacillus and Lacticaseibacillus, for the flavor development and other peculiar traits in traditionally (long-) ripened hard cheeses of excellent quality, is well established [37,41]. Therefore, selected Lp. plantarum, Lc. paracasei, and Lc. rhamnosus strains have been included in advanced CSCs or they are marketed as single-strain bioprotective or probiotic cultures [44]. Particularly for Lp. plantarum, its multi-functional activities have been recognized in numerous recent studies, e.g., by the combined use of plantaricin-producing strains in multi-bacteriocin commercial cheese starter systems with nisin and lacticin 3147 Lc. lactis strains [45], or with autochthonous L. lactis strains to effectively inhibit L. monocytogenes in fresh cheese model systems [46], or by the expanded use of commercial Lp. plantarum strains as functional adjunct starters to commercial Greek PDO fresh acid-curd cheeses, like Feta cheese [47]. In contrast, the use of CSCs containing Lp. plantarum, Lc. paracasei, and Lc. rhamnosus in brand-name Kefalotyri cheese products has neither been explored nor set as a trademark yet. Only a few early experimental studies exist, which focused on comparing the physicochemical and sensory quality attributes of Kefalotyri cheese-making trials using thermophilic yogurt starters supplemented with Lc. casei [11,12] or Lc. paracasei subsp. paracasei and Lc. rhamnosus [18] as commercial co-starter or probiotic adjunct strains, respectively. The cheeses made with Lc. casei accumulated more diacetyl until the 60th day of ripening [11], while both probiotic adjuncts retarded the growth of enterococci, causing a greater decrease in pH and enhanced alpha (S)-casein and beta-casein degradation, free amino acid release, lipolysis, and the sensorial quality of the probiotic cheeses compared to the control cheese containing the yogurt starter only [18].

Meanwhile, since 2015, we have established that the routine commercial application of the native nisA+ L. lactis subsp. cremoris M78/M104 genotype as an adjunct starter in CSC-mediated Graviera Pappas from either pasteurized or thermized sheep/goat milk consistently leads to more aromatic cheeses during ripening than counterpart cheeses produced with the CSC only. Although the specific biochemical mechanisms underlying this phenomenon remain largely unexplored, it seems that the early (day 1) in situ nisin A production by the M78 strain [24] accelerates proteolysis by enhancing the lysis of the CSC cells during Graviera fermentation and early ripening, as previously reported for other bacteriocin-producing L. lactis strains [48,49]. In this study, the application of strain M78 as an in situ nisA+ co-starter was extended to the semi-industrial Kefalotyri manufacture with native starters only. Based on previous model co-cultures in sterile raw milk (SRM) [27], the growth compatibility of the ST1 + M78 starter in the C-cheese milks, and with the adjunct ripening strain H25 in the N-cheese milks, was anticipated to be high and suitable for enhancing proteolysis on a HO/HI peptide ratio and volatilome basis.

Indeed, because the growth pattern of the nisA+ M78 producer was similar in both Kefalotyri cheese types (Table 2), the proteolysis was progressive and generally occurred in a similar quantitative and qualitative manner during ripening, as revealed by the peptide profile (parts 0 to III) and the gradual decrease in the HO/HI ratio (Table 6), which serves as an indicator of the cheese maturation progress [49]. An increase in the percentage of the hydrophilic (HI) compared to the hydrophobic (HO) peptides signifies the breakdown of proteins, the opening of their hydrophobic structure, and the advancement of their proteolysis [50]. Several processing factors contribute to casein breakdown and HO peptide release during primary proteolysis [51], which afterward is succeeded by the secondary proteolysis, and the content of HO peptides decreases due to their conversion into free amino acids [52], as also occurred in all the Kefalotyri cheeses (Table 6). Because HO peptides have been associated with cheese bitterness [49,52], in this study, it was positive that HOs were reduced during ripening and reached similar low levels in both cheese types after 90 days of ripening, with an apparent tendency of the N-cheese samples to attain a lower final HO/HI ratio (Table 6), which can be partly assigned to the major declines in the viable H25 adjunct cells, probably due to autolysis.

The volatiles in the headspace of mature Kefalotyri cheese samples were grouped by chemical class to simplify comparisons (Table 7): first, it is worth noting that the total volatile profile of the N-cheeses (27 compounds) was less diversified than that of the C-cheeses (37 compounds),probably because the catabolic pathways giving rise to aromatic compounds [52] were less variable due to the prevalence of the H25 adjunct strain during ripening. In both cheese types and in decreasing order, the most abundant volatile groups were free fatty acids (FFAs), ketones, esters, and alkanes, with only the FFAs occurring at higher levels in the N-cheeses, owing mainly to the butanoic (butyric) and acetic acids (Table 7). Both, and especially acetic acid (also quantified in Table 5), are associated with primary heterolactic glycolytic pathways and the esterase (C4) activities of the mixed Lp. plantarum/Ln. mesenteroides adjunct culture. Conversely, in the C-cheeses, the most abundant VOCs were acetone, methyl-butyrate, and heptane, which occurred at moderately lower levels, but in the same decreasing order as in the N-cheeses. The most diversified VOC group in both the C- and N-cheeses (10 and 8 peaks) were esters, followed by FFAs (7 and 6 peaks), in general agreement with the literature; generally, ester accumulation in ripened cheese enhances aroma [52]. Specifically, esters like ethyl-butanoate and ethyl-hexanoate, also detected at similar joint levels lower than methyl-butyrate in both Kefalotyri cheese types (Table 7), have been reported to enhance the sweet and fruity aromas of ripened cheeses [53]. Conversely, alcohols, ethers, and aldehydes were less diversified in both cheeses, with the latter two groups represented by one VOC only, namely ethyl ether and 3-methyl butanal, which was traced in only one C-cheese trial (Table 7). In contrast, 3-methyl-butanol was detected in both cheese types at similar levels, which, however, were 3 to 4-fold lower than ethanol herein, and 5 to 6-fold lower than the 3-methyl-butanol levels found in artisanal Kefalotyri cheese from raw sheep milk without starters after 90 days of ripening [10].

To sum up, the volatile profiles of the present cheeses show high consistency with the artisanal Kefalotyri regarding the abundance of total FFAs, and with butyric acid, ethanol, and 3-methyl butanoic acid as the most abundant volatiles [10], the latter linked with the corresponding ester, methyl butyrate (Table 7). It is well documented that FFAs released from milk fat triglycerides owing to the activity of various lipases mainly during cheese ripening serve as precursors for aroma compounds such as esters, methyl ketones, lactones, and alcohols [52]. However, fat hydrolysis is not the only FFA source, as the catabolic products of lactose fermentation, deamination of amino acids, and lipid oxidation may also contribute to the overall fatty acid pool [52,53,54]. Butyric acid was also found to be an important volatile in commercial Kefalotyri cheeses [15], and abundant in other cheese types [54]. Furthermore, it should stressed that the levels of acetone (having acetoin as a precursor) in the present semi-industrial Kefalotyri cheeses fermented with native starter and adjunct strains were 8 to 10-fold higher than in the artisanal Kefalotyri without starters [10], possibly due to the strong acetoin-forming capacity of the M78, KE109, and H25 strains shown in vitro and, most probably, of the NSLAB ‘adjunct’ TM biota in the C-cheeses with the highest acetoin content. Kondyli et al. [55] also found that low-fat Kefalograviera cheeses produced with a commercial Lc. rhamnosus adjunct had higher levels of acetone. Conversely, 2-butanone, which occurred at quite high levels along with acetoin in artisanal Kefalotyri [10], was a minor volatile compound in the C-cheeses and undetectable in all fully ripened N-cheeses, unlike 2-heptanone and 2-nonanone. Meanwhile, a high buildup of lipid oxidation products, such as heptane (Table 7) and other alkanes (benzene and hexane), may cause flavor defects in ripened cheeses, as dimethyl sulphide also does, even at the low contents herein detected in the C-cheeses only. From the rest of the miscellaneous VOCs detected, styrene should not be present in cheese in detectable quantities unless there is contact with an external source that contains it, such as the plastic containers used to store the present cheese products after sampling. Overall, the reduced VOC diversity in the N-cheeses could be attributed to the prevalent growth of the Lp. plantarum H25 adjunct strain, which ultimately replaced the growth of autochthonous enterococci and other adventitious NSLAB species during ripening.

5. Conclusions

Two types (C and N) of commercial, semi-industrial, artisan-type Kefalotyri cheeses of high quality and safety were produced from mildly thermized (65 °C, 30 s) sheep milk inoculated with native starter strains of S. thermophilus and L. lactis, including the wild nisA+ M78 strain as a potential accelerator of proteolysis during ripening. Compared to the mature C-cheeses, the novel N-cheeses with the same starter added and three native Lp. plantarum and Ln. mesenteroides adjunct strains contained less residual lactose but more acetate. Both cheese types had similar peptide profiles, HO/HI peptide ratios, and volatilome (VOC) profiles. However, several VOCs (3-methylbutanal, 3-methylbutanoic acid, methyl propionate, 2,3-butanedione, 2-butanone, and dimethyl disulfide) that were undetectable in the N-cheeses existed in the C-cheeses. Additional studies are needed to evaluate the primary proteolysis, lipolysis, textural changes, sensory perception, and consumer acceptance, along with genomic approaches to map the evolution of NSLAB that survive in Epirus sheep milk after mild thermization. The latter biota can promote an unpredictably high growth to interfere, mask, or even counteract the beneficial effects anticipated by the native starter or adjunct strains in the resultant Kefalotyri cheeses.