Review on the NGS-based studies of microbiotas of artisanal and regional kinds of cheese with potential as functional foods: composition and functional analysis

Abstract

This review aims to reveal and summarize the advantages and disadvantages of culture-dependent and culture-independent methods for identifying and characterizing the microbiota of artisanal regional varieties of cheese and the health benefits of these kinds of cheese as a functional food. With the rise of metagenomics studies, the focus is now on using these techniques to explore the genetic diversity in different types of cheese, identifying both known and novel microorganisms. Based on literature data, in addition to the possible roles of microorganisms included in artisanal dairy products on cheese ripening, the health-promoting effects of these bacteria are highlighted as well. The analysis of the literature data available so far showed that metagenomic analysis would allow for obtaining more accurate and rapid results. In addition, cheese microbiota and microbial interactions play an essential role in acidification and flavour development in cheese. The production of microbial bioactive compounds, their role in inhibiting pathogens and spoilage agents, and the probiotic potential are also discussed.

Keywords:

• Microbiota
• Metagenomic analysis
• Artisanal regional types of cheese
• Functional foods

Introduction

Why do we study cheese microbiota?

Studies of the microbiota in artisanal regional products are essential, especially when using innovative, modern analysis methods because many types of cheese made from sheep, goat, and buffalo milk are considered functional foods that have proven beneficial effects on health [1,2]. In addition, fermented products from sheep’s milk have organoleptic, nutritional and antioxidant properties, as documented in the scientific literature [3,4]. They are a source of probiotic microorganisms. Some kinds of cheese produced from goat’s milk also have antioxidant properties and are a source of probiotic bacterial strains. They are known primarily due to the immunomodulatory role of their proteins and peptides, which are formed because of their fermentation. Buffalo cheese (in particular Bulgarian brined buffalo cheese) is also classified as a functional food, and its beneficial properties are due to the content of various organic acids, protein degradation products and certain types of fatty acids [1,5].

On the one hand, the above-mentioned beneficial properties of cheese are due to the characteristics of milk proteins, sugars and lipids in different milk types. On the other hand, it is due to the microorganisms involved in milk processing. Microorganisms play important roles during both cheese manufacturing and the cheese’s ripening. The starter and secondary flora can modify the physical and chemical properties of the cheese, i.e. the characteristics of cheese depend on microflora dynamics, which may, in turn, be affected by the microorganism interactions and the influence of environmental conditions, among others [6].

Classical methods for studying cheese microbiota

Cultivation-based microbiological methods

In the past, classical microbiological techniques were predominantly used for studying the growth activity of microorganisms. The culture-dependent approach relies on growing bacteria and implies the following steps: growing bacteria on microbial media before enumeration, isolation, identification at genus and species level, and characterization of biotype at intraspecific level [6,7]. The number of colony-forming units (CFU) within various microbial groups and the isolation of microorganisms can be evaluated through culturing methods and future applications of microorganisms in cheese as starter cultures [8]. The culture-dependent methods are time-consuming and include a culturing step that can lead to inaccuracies due to the target species being present in low numbers or simply uncultivable. Hence, they cannot be used as a unique community monitoring tool [9] Sometimes, culture-dependent methods fail, which may be explained by the overshadowing of minor populations by the predominant populations or by the transition to a VBNC (viable but nonculturable) state, probably caused by environmental stress [10].

Molecular methods

Since some microorganisms in cheese cannot be detected solely by cultivation-based microbiological methods, the use of molecular methods is more reliable and complementary to evaluating the microorganism composition in the examined kinds of cheese. There is a rapid increase of culture-independent methods based on the direct analysis of DNA. Applying these techniques represents a rapid, reliable, and effective way to detect and identify the microorganisms present in dairy products [9]. Methods such as PCR-DGGE, PCR-TTGE, SSCP-PCR, T-RFLP, LH-PCR, qPCR, RT-qPCR and DHPLC (which does not rely on PCR amplification) are commonly used. [11–14].

Genus- and species-specific PCR

PCR is one of the most sensitive techniques available for detecting bacterial targets. The 16S rRNA gene is an ideal target for bacteria because it is highly specific to each species. The standard method involves PCR amplification of the 16S rRNA gene or some other species- or genus-specific gene, sequencing, and comparison to known databases for identification. PCR-based methods are not only faster than conventional culture-based methods but are helpful in the identification of bacteria that are difficult to grow in laboratory conditions [15]. Some authors studied Lactiplantibacillus plantarum strains (formerly known as Lactobacillus plantarum) isolated from the traditional Slovak raw sheep milk cheese by using genus-specific and species-specific PCR for their inhibitory potential against a broad spectrum of foodborne pathogens, including wild strains of Staphylococcus aureus [16]. In another study [17], the PCR technique was applied using Cb1-Cb2R and species-specific primers, utilized on various French soft flowered or washed rind types of cheese to identify Carnobacterium species. Thirty kinds of cheese made from cow’s, ewe’s, or goat’s milk (raw or pasteurized) were studied in both autumn and spring. The PCR results with species-specific primers of Carnobacterium showed that ten varieties of cheese contained only the species Carnobacterium maltaromaticum. Six different fermentation patterns were found, and three of the ten types of cheese contained C. maltaromaticum isolates with antilisterial activity.

Another research team found that Rep-PCR typing, combined with the Limosilactobacillus fermentum, Lentilactobacillus parabuchneri and Levilactobacillus brevis species-specific duplex PCRs, has been extremely useful to classify a large set of presumed obligately heterofermentative lactobacilli (OHL) dairy isolates rapidly and reliably at species level. This study confirmed the prevalence of Limosilactobacillus fermentum and Levilactobacillus brevis strains during ripening in different cheese varieties, and that of Lentilactobacillus parabuchneri was highlighted [18]. RAPD-PCR (Random amplified polymorphic DNA PCR) and partial sequencing of the 16S rRNA gene were used by Merchan et al. (2022) for the identification of seventeen species in the Spanish sheep’s milk varieties of cheese ‘Torta del Casar’ and ‘Queso de la Serena’. Examination of the enterobacteria involved in the ripening of these types of cheese was carried out. It showed that Hafnia paralvei was the predominant species, and H. alvei and Lelliottia amnigena were present to a lesser extent. The investigators proved Hafnia spp. strains did not produce cytolytic toxin, were active against HeLa cells, nor contained the virulence genes for its synthesis. H. alvei strains 544 and 1142 were proposed to be included in the industrial application to promote cheese homogeneity [19].

Quantitative PCR

Quantitative PCR or real-time PCR (qPCR) provides higher sensitivity and accuracy, as well as the ability to monitor DNA amplification in real-time, relying on fluorescence intensity by utilizing the Cq value (cycle number at which fluorescence intensity rises above the detectable level) [15]. The main drawback of the method in microbiota research is that it relies on primers targeting specific organisms, so it cannot give a clear picture of the microbiome’s composition. Instead, it can be used to determine the presence and quantification of specific microorganisms. It is used mainly for detecting pathogens and spoiling agents, especially in its multiplex form, which allows for the simultaneous search of several microorganisms [20]. For example, the method has been successfully used for the detection of clostridia in some Austrian [21], Turkish [22], Greek [23] and other kinds of cheese. An interesting advantage of the methods is that it not only allows the detection of the presence of specific microorganisms but, upon the primer designs, the assessment of their pathogenicity potential when it deals with bacteria with ambiguous nature, which could both play a positive or negative role during ripening [24]. An example of such an application of the qPCR technique is the detection of Enterobacteriales’ genes conferring resistance to critically important antibiotics for human medicine in some French kinds of cheese [25].

DGGE, TGGE, RFLP, SSCP, PFGE

In denaturing gradient gel electrophoresis (DGGE), the electrophoretic mobility of a partially melted double-strand DNA molecule in polyacrylamide gel is implemented. There is a different melting behaviour of the molecules with different sequences; therefore, a molecular migration at different positions in the gel would be expected [26]. In the temperature-gradient gel electrophoresis technique (TGGE), the biopolymers migrate perpendicular to a temperature gradient so that every individual molecule is at a constant temperature throughout electrophoresis [27]. Denaturing gradient gel electrophoresis (DGGE) and temperature gradient gel electrophoresis (TGGE) were routinely used in many microbiological laboratories worldwide as molecular tools to compare the diversity of microbial communities and to monitor population dynamics [28]. In SSCP (single-strand conformation polymorphisms), genomic DNA is digested by restriction endonucleases, followed by denaturation in an alkaline solution and electrophoresis on a neutral polyacrylamide gel [29]. In addition, DGGE/TGGE and SSCP were suitable for qualitatively studying microbial communities’ temporal and spatial variation. Combining the profiles of the 16S rRNA gene and the functional gene could allow the relation of a structure to a function of an ecosystem. Another variant of this approach is the use of oligonucleotide microarrays, which offer more applications in microbial ecology. It would be essential to study the role of the bacterial community in cheese ecosystems [30]. PCR-DGGE analysis may be used to give information on concentration ratios among species that occur in a particular environment and can be employed to rapidly identify dominant microorganisms as an alternative to traditional tools [31]. Dolci et al. used PCR-DGGE and qPCR-DGGE to study the succession of bacterial communities encountered in Fontina PDO cheese. Coryneform bacteria were actively present and could be considered determinants of rind formation. qPCR-DGGE gels showed a richer band profile than the one obtained based on DNA analysis, thus indicating that RNA analysis can highlight bacterial species that DNA analysis is not able to show, as the biodiversity of the Fontina PDO surface is described better through qPCR [32]. Parayre et al. developed a rapid and easy method to extract DNA from various kinds of cheese and optimize the separation of low- and high-GC-content bacteria by PCR-Temporal Temperature Gel Electrophoresis (PCR-TTGE). Seventy-six strains belonging to fifty of the most common bacterial species in dairy products were used to construct a database. They created PCR-TTGE specific ladders that contained seventeen species, forming a regular scale. The investigators indicated that the method appears to be particularly effective in characterizing the cheese rind ecosystem [9].

Restriction fragment length polymorphisms (RFLP) identify bacterial strains via unique DNA fingerprints created by variations in homologous DNA sequences. It employs restriction enzymes to cut PCR-amplified DNA into fragments of varying lengths. These fragments are then separated through agarose gel electrophoresis, producing distinct banding patterns for each strain. Differences in these patterns indicate bacterial strain diversity, with closely related strains exhibiting identical or similar patterns [15]. PCR-RFLP has been used for cheese microbiota research [33]. Some examples are the traditional Portuguese Serpa cheese [34], several traditional kinds of cheese from North-West Iran [35], and some other varities of cheese worldwide.

In pulsed-field gel electrophoresis (PFGE), alternating an electric field in more than one direction is observed through a solid matrix to achieve the separation of DNA fragments. This method requires the preparation of unsheared DNA, digestion of the DNA by a rare-cutting restriction endonuclease, separation of fragments by PFGE, and the visualization and interpretation of banding patterns. Fragments as large as ten megabases (Mb) can be separated. The time required for DNA fragments of different sizes to reorientate to the new electric field is proportional to their molecular weight [36]. Lactic acid bacteria (LAB) isolates from the interior of six traditional Pecorino Siciliano cheese samples during ripening (for one, thirty and ninety days) were characterized genotypically using PCR and PFGE. PFGE analysis allowed the identification of different strains of the species Lactiplantibacillus plantarum and Lacticaseibacillus paracasei. It indicated 52 different band patterns for enterococci, 9 for lactococci, 5 for leuconostocs, and 1 for streptococci and pediococci [37].

Although the molecular methods give decent information for identifying microorganisms in cheese, their applications have some limitations. Those could be poor DNA extraction yield, PCR inhibition by various extraction by-products or substances coming from the cheese matrix itself, or the possibility of differential PCR amplification. Another point is that culture-independent methods often fail to identify species that culture-dependent methods can detect [38,39]. These two methods provide different perspectives on the same community. Thus, combining culture-dependent and culture-independent methods offers a more accurate view of the microbial composition [40].

Main genera and species participating in cheese ripening

Streptococcaceae family

Lactococcus genus

Lactococci are coccoid Gram-positive, anaerobic bacteria that produce L(+)-lactic acid from lactose in spontaneously fermented raw milk. The microbial nature of lactic fermentation was recognized in 1857 by Louis Pasteur. Joseph Lister obtained and scientifically described the first pure bacterial culture on Earth in 1873, Lactococcus lactis [41]. The Lactococcus genus comprises 13 species: Lactococcus lactis, L. taiwanensis, L. hirilactis, L. fujiensis, L. nasutitermitis, L. garvieae, L. petauri, L. formosensis, L. piscium, L. raffinolactis, L. reticulitermitis, L. chungangensis, L. plantarum, and L. laudensis [42].

The proof that lactococci are one of the central role players in cheese ripening is that they are omnipresent in cheese worldwide. Lactococcus lactis strains were isolated and characterized from raw-milk Oaxaca cheese from the Tulancingo Valley, Mexico [43], the traditional Sardinian Casizolu pasta filata cheese, where it is often accompanied by Lactococcus raffinolactis [44], the very soft Bulgarian cheese Krokmach [45] etc. In addition, more than one lactococcal species within the cheese microbiotas can be found within one cheese batch. The genotypic and phenotypic variability of 40 Lactococcus lactis isolates from Batzos cheese was investigated across three cheese-making trials conducted in winter, spring, and summer. RAPD-PCR, plasmid profiling, and PFGE were employed to study genetic variability and differentiate closely related isolates. Results indicated significant heterogeneity among strains. Notably, 42.5% of isolates exhibited high acidifying ability in milk after 24 hours of ripening. In winter cheese samples, 75% of isolates showed higher Lys-aminopeptidase activity than Leu-aminopeptidase, while approximately 67% of summer cheese samples showed the opposite pattern. Their caseinolytic activity after growth in milk for 24 h was significant, with a preference for α_s-casein degradation. The majority (90%) of the strains formed methanethiol from methionine. These results suggest valuable strains among the wild lactococcal population from Batzos cheese that are appropriate to be used as starters for the dairy industry [46]. Odamaki et al. 2011 provided an efficient method for identifying lactococcal strains of industrial importance. They designed a multiplex PCR primer set based on the nucleotide sequences of the 16S rRNA gene of the seven lactococcal species. Thus, seven species were taxonomically classified (Lactococcus lactis, Lactococcus garvieae, Lactococcus piscium, Lactococcus plantarum, Lactococcus raffinolactis, Lactococcus chungangensis and Lactococcus fujiensis). The investigators accented that the one-step multiplex PCR enables the identification and speciation of bacterial strains that belong to the genus Lactococcus, as well as the differentiation of strains of L. lactis subsp. lactis and L. lactis subsp. cremoris [47]. A rich lactococcal microbiota was also found in the Robiola di Roccaverano cheese from the Piedmont region of Italy, which was investigated by RAPD-PCR where Lactococcus lactis ssp. lactis and Lactococcus lactis ssp. cremoris were reported [48]. The investigation of the natural bacterial population used in the production of Toma piemontese cheese showed that lactococci constituted 67% of the coccal isolates, identified as Lactococcus lactis and L. garvieae [49]. Morea et al. studied bacterial populations of Mozzarella cheese by utilizing physiological analyses and molecular techniques. An analysis of RAPD fingerprints revealed twenty-five different biotypes. The investigators emphasized that Lactococcus lactis was highly acidifying and proteolytic [50].

Streptococcus genus

Streptococci are Gram-positive cocci and facultatively anaerobic chemo-organotrophs with complex nutritional requirements and a fermentative metabolism resulting in L (+) lactic acid as the primary product of their glucose fermentation [51]. The group comprises three genetically similar species - Streptococcus salivarius, Streptococcus vestibularis and Streptococcus thermophilus. S. salivarius and S. vestibularis are commensal organisms that may occasionally cause opportunistic infections in humans, whereas S. thermophilus is a food bacterium widely used in dairy production [52]. Streptococcus thermophilus is a primary starter for the dairy industry and is highly economically significant [53].

In an investigation of the bacterial diversity in traditional Minas cheese, made from raw milk, produced in four different regions of Minas Gerais state in Brazil, PCR–DGGE analysis of the V3 region of the bacterial 16S rDNA was used [54]. Samples of Minas cheese made with pasteurized milk from the Serro region were used as a control. The results suggest that specific band profiles from traditional types of cheese from each region may be used as a biological barcode to disclose their origins. DGGE band sequencing analysis showed that species that belong to the genera Streptococcus may predominate in the traditional Minas cheese [55]. Antibiotic susceptibility, antimicrobial activity, and genotypic and technological properties of 52 Streptococcus thermophilus isolates collected from four northern Italian traditional cheese samples were investigated by Morandi et al. [56]. The method of RAPD-PCR was applied to study genetic variability and distinguish closely related strains. The results showed a high degree of heterogeneity among isolates.

Regarding the technological properties, after 6 h of incubation in milk, 25% of the streptococcal strains could be defined as fast acid producers. However, after 24 h, most isolates (79%) showed weak acidification activity. Reduction activity was generally low. All the studied S. thermophilus were susceptible to ciprofloxacin, levofloxacin, penicillin G, ampicillin, mupirocin, nitrofurantoin, quinupristin/dalfopristin and rifampicin [56]. Randomly amplified polymorphic DNA (RAPD)-PCR was used to identify 218 Streptococcus thermophilus isolates from four Italian PDO cheese types. RAPD-PCR allowed the discrimination of part of the S. thermophilus isolates according to the cheese origin. Isolates that show RAPD-PCR profiles very similar to those of two Streptococcus macedonicus reference strains were detected, assuming the presence of these streptococci in Italian kinds of cheese as well [57]. Pacini et al. developed a new approach for detecting and enumerating Streptococcus macedonicus in cheese. The method is based on a first screening of the cheese by a PCR assay specific for S. macedonicus, followed by plating positive samples on a differential SM medium. It was applied to 51 samples derived from PDO and traditional Italian varieties of cheese. Streptococcus macedonicus was found in 16 of the 51 samples. The system developed was particularly useful for the differential count of S. macedonicus in cheese and allowed the evaluation of the occurrence of this species within the complex microbial lactic acid bacteria (LAB) population, which is typical for traditional kinds of cheese. The results showed that in the examined cheese samples, S. macedonicus cannot be considered a dominant LAB species [58]. In a study of the development of the dominant bacterial populations during traditional Mozzarella cheese production, the analysis of RAPD fingerprints revealed that the dominant bacterial community was composed of 25 different biotypes. Streptococcus was found to be highly acidifying [50].

Lactobacillaceae family

Lactic Acid Bacteria (LAB) are microorganisms comprising Gram-positive, catalase-negative bacteria that produce lactic acid as the major metabolic end-product of carbohydrate fermentation. Among them, members of the former Lactobacillus genus play a central role in the ripening of fermented dairy products (before its taxonomic re-classification in 2020 [59]). Many strains of the former genus are used as probiotics. The former genus Lactobacillus included over 200 species widely used in fermented food preservation and biotechnological processes or explored for their beneficial effects on health [60]. The most frequently encountered representatives are Lacticaseibacillus casei, Lacticaseibacillus paracasei, Lactiplantibacillus plantarum, Lacticaseibacillus rhamnosus and Latilactobacillus curvatus [61]. Lacticaseibacillus casei, Lacticaseibacillus paracasei, and Lacticaseibacillus rhamnosus are phenotypically and genotypically closely related and together comprise the former Lactobacillus casei group, which is now reclassified as the Lacticaseibacillus genus. The strains of this genus are commercially valuable as probiotics [62].

Antonsson et al. studied the Non-Starter Lactic Acid Bacteria (NSLAB) from six ripened Danbo cheese samples of different ages and of different brands. The role of the members of the former genus Lactobacillus in cheese maturation was explored. Thirty-three isolates were differentiated by the PCR-based method, RAPD. The different RAPD types were identified at the species level byTTGE. The isolates were mainly revealed to be Lacticaseibacillus paracasei (76%), Latilactobacillus curvatus, Lactiplantibacillus plantarum, Lacticaseibacillus rhamnosus, and taxa originating from the starter culture, were detected. In one cheese, no lactobacilli were found. The investigators used different isolates as adjunct cultures in a cheese model system, showing that some Lacticaseibacillus paracasei strains are beneficial for the cheese flavour. Lactiplantibacillus plantarum introduced off-flavours in young model cheese [63]. Lactic acid bacteria from 18 Spanish goat cheese types, produced by seven dairies, were isolated to evaluate the genetic diversity of this bacterial community by phenotyping and RAPD-PCR analysis. There was a predominance of Lacticaseibacillus paracasei subsp. paracasei. The presence of Latilactobacillus curvatus, Lactiplantibacillus pentosus, Limosilactobacillus fermentum and Lacticaseibacillus rhamnosus was reported in Spanish goat cheese varieties for the first time. Some identified strains displayed strong acidifying and proteolytic capacities [64]. The lactobacilli intra- and inter- species diversity in a Piedmont hard cheese was also explored. The product is made of raw milk without thermal treatment and without the addition of an industrial starter. A screening for potential functional properties was performed for the first time. The isolates were collected during the cheese production and identified through molecular methods: 304 were identified as Lacticaseibacillus rhamnosus, 240 as Lactobacillus helveticus, 26 as Lactiplantibacillus fermentum, 11 as Lactobacillus delbrueckii, 3 as Limosilacto­bacillus pontis, and 2 as Lactobacillus gasseri and Limosilactobacillus reuteri respectively. The usage of repetitive bacterial DNA element fingerprinting (rep-PCR) with (GTG)₅ primer resulted in eight clusters of Lactobacillus helveticus and sixteen clusters of Lacticaseibacillus rhamnosus. Most isolates showed high auto-aggregation propriety, low hydrophobicity values, and a generally low survival rate in the simulated digestion process. Sixteen isolates showed promising functional characteristics [65]. Identification of lactobacilli in eight different traditional kinds of cheese from West Azerbaijan revealed the presence of Lactiplantibacillus plantarum (24%), Lacticaseibacillus casei (20%) and Ligilactobacillus agilis (18%) from facultative heterofermentative species and Lactobacillus delbrueckii (21%), Lactobacillus helveticus (14%) and Ligilactobacillus salivarius (3%) from the obligate homofermentative bacilli. The investigators highlighted the achievement of organoleptic characteristics of traditional cheese types in industrial productions, for which mixed starters including dominant Lactobacillaceae microflora would be necessary [66]. Strains of Lacto­bacillus delbrueckii subsp. lactis were isolated from Italian hard and semi-hard kinds of cheese and artisan starter cultures and were characterized by phenotypic and genotypic methods. Genotypic diversity was conducted by RAPD-PCR and PFGE. Phenotypic characterization indicated a wide variability in Lactobacillus delbrueckii subsp. lactis acidifying activity. Concerning peptidase activity, Lactobacillus delbrueckii subsp. lactis showed a homogeneously high x-prolil-dipeptidil-aminopeptidase activity and a considerable but more heterogeneous lysil-aminopeptidase activity [67]. Culture-dependent and culture-independent approaches were compared to assess the lactobacilli community biodiversity and evolution during the production of RDO Camembert in three cheese-making factories by usingTGGE to analyze total microbial DNA and DNA from single isolates. The two approaches provided complementary information. Lacticaseibacillus paracasei subsp. paracasei was the dominant species in the three factories, and Lactiplantibacillus plantarum was a dominant species in one. The presence of Lactobacillus delbrueckii susbp. lactis, Lactobacillus acidophilus, Lactobacillus delbrueckii susbp. bulgaricus and Lacticaseibacillus casei subsp. casei was detected as well [68].

Other Firmicutes

The bacterial Phylum Firmicutes contains the three Classes: Bacilli, Clostridia, and Mollicutes. It includes a total of 235 genera and all species of lactic acid bacteria. Members of this Phylum are highly diverse in morphology, physiology, and Gram-staining characteristics [69]. They occupy a wide range of habitats and can be beneficial or detrimental in diverse settings, including food and beverage-related industries [70]. Endospore formation is used by members of the phylum Firmicutes to withstand extreme environmental conditions [71]. Most other Firmicutes members are generally related to rather adverse effects in cheese ripening, including spoilage and unpleasant organoleptic properties. Some Firmicutes negatively impacting cheese production belong to Clostridium and Staph­ylococcus genera.

Clostridia

Le Bourhis et al. [72] applied for the first time differentiation of all clostridial species present in cheese with a single test. A nested-PCR TTGE approach was developed to detect bacteria belonging to phylogenetic cluster I of the genus Clostridium in cheese suspected of late blowing. PCR-TTGE was applied to analyze commercial kinds of cheese with defects. In all cheese samples with a high amount of butyric acid, the presence of C. tyrobutyricum DNA was confirmed by PCR-TTGE, suggesting the involvement of this species in butyric acid fermentation. These results demonstrated the PCR-TTGE method’s efficacy in identifying Clostridium in cheese. The experimental procedure developed in this study could be implemented as a routine detection method in the industry to screen cheese during production and determine Clostridium species in defective cheese [72]. Butyric acid fermentation, the late-blowing defect in cheese caused by the outgrowth of clostridial spores present in raw milk, could result in significant product loss, especially in the production of semi-hard types of cheese like Gouda cheese, but also in Grana and Gruyère cheese. A method of PCR amplification of a part of the 16S rRNA gene, in combination with hybridization with species-specific DNA probes, was developed to allow the specific detection of clostridial sequences in DNAs extracted from cheese. The sensitivity was increased by using nested PCR. It was concluded that only C. tyrobutyricum strains can cause butyric acid fermentation in cheese [73]. Other investigators studied late-blowing – severe cheese spoilage caused by clostridia at fifty-three cheese samples of different origins, using culture-dependent (MPN, enrichment, molecular identification of isolates) and culture-independent methods. Using culture-dependent methods, clostridial isolates were collected from several cheese samples, while fewer positive results were obtained with the samples when using the MPN procedure than when using real-time PCR or PCR-DGGE. The culture-independent techniques complemented the traditional microbiological methods rather than replacing them entirely [74].

The Staphylococcus genus

Borelli et al. [75] studied the Staphylococcus spp. population dynamics during the ripening of Canastra Minas cheese at three farms in Minas Gerais, Brazil. The presence of coagulase (coa), thermonuclease (nuc), and enterotoxins sea, seb, sec, and sed genes was examined in Staphylococcus strains isolated over the sixty-day ripening period. Additionally, staphylococcal enterotoxins A, C, and D were investigated. All coagulase-positive isolates tested positive for the coa gene. There was no correlation between biochemical test results and PCR results for coagulase-negative strains. Coagulase and thermonuclease genes coexisted in 41.3% of Staphylococcus spp. None of the Staphylococcus strains tested positive for enterotoxins SEA, SEB, SEC, and SED, nor were enterotoxins A, C, and D detected in any cheese samples [75]. Silva et al. found that the physico-chemical characteristics of Minas Frescal cheese (MFC) favor the growth of Staphylococcus spp. and allow the production of enterotoxins by specific strains. They characterized the physical-chemical aspects (pH, storage temperature, and salt content) and the presence of Staphylococcus spp. in MFC samples. Staphylococcal isolates were obtained and subjected to PCR assays to identify them as Staphylococcus aureus (nuc) and to detect staphylococcal enterotoxin-related genes (sea, seb, sec, sed, see). Selected isolates were identified as S. aureus, but none presented classical enterotoxin-related genes. According to the study, the best temperature to avoid staphylococcal growth was 7.5 °C and storage at temperatures lower than 7.5 °C could have prevented staphylococcal growth and the potential production of enterotoxins [76]. Amplified Ribosomal-DNA Restriction Analysis (ARDRA) was implied to differentiate the genus Staphylococcus of Red-Smear types of cheese. 2.4 kbp PCR-product was amplified. Species-specific restriction patterns were found by using restriction enzymes HindIII and XmnI. ARDRA results agreed with the results of partial sequencing of the 16S rDNA. This study emphasized that ARDRA could fully replace the biochemical identification with ID32 Staph (BioMerieux). The genomic restriction samples of cheese-related S. equorum strains by SmaI and SacI gave unique strain-specific restriction patterns of the investigated cheese [77].

Proteobacteria

The name Proteobacteria was first proposed by Stackebrandt et al. in 1988 [78]. Still, the grouping of bacteria was done by Woese in 1987 with the informal name of “purple bacteria and their relatives” [79]. Proteobacteria, the largest bacterial phylum, are Gram-negative and have a lipopolysaccharide in their outer membrane as reviewed by Rizzatti et al. [80]. Based on phylogenetic analysis of the 16S rRNA gene, the Proteobacteria phylum is divided into six classes: Alphaproteobacteria, Betaproteobacteria, GamMAProteo­bacteria, Deltaproteobacteria, Epsilonproteobacteria, and Zetaproteobacteria [80]. Proteobacteria members are often found in cheese and generally impact the products negatively [81]. However, they could actively participate in ripening, especially in artisanal and homemade cheese prepared without starter cultures [82,83].

A study examined the diversity and dynamics of predominant microbial communities in artisan Gouda-type cheese samples produced under varied conditions. Twenty-two cheese types, differing in milk source, treatment, production environment, and brining conditions, were analyzed using PCR-DGGE using total DNA and DNA isolated from culturable fractions. GamMAProteobacteria were among the observed microorganisms. The combined PCR-DGGE approach using both total DNA and culturable fractions proved effective in assessing the impact of technological and environmental factors on microbiota diversity and dynamics in Gouda-type cheese. [84]. Ozturkoglu et al. determined that Proteobacteria classes are predominant during the early ripening of traditional Turkish Divle Cave cheese [85]. The microbial diversity of the surface of a commercial red-smear cheese, Livarot cheese, sold on the retail market, was investigated using culture-dependent and independent approaches. Regarding the bacteria present, 40 bacteria from the cheese surface were collected, dereplicated using SSCP analysis and identified using rRNA gene sequencing. Fluorescence in situ hybridization (FISH) analysis was also used to study the cheese microbial diversity with class-level and specific rRNA-targeted probes for bacteria, and it was confirmed that Gammaproteobacteria were important microorganisms in this cheese [86]. An inordinately high number of samples from Mozzarella and whey cheese products of Italian and German production was investigated because of publications about the so-called blue Mozzarella event in 2010. Nogarol et al. investigated a Pseudomonas fluorescens strain since it was supposed that the microorganism might have caused the so-called blue Mozzarella. Molecular characterization of 181 isolated Pseudomonas fluorescens strains was conducted using a newly optimized pulsed-field gel electrophoresis protocol. The investigators proved a possible cross-contamination among cheese samples from the two countries [87]. The evolution of the population of Enterobacteriaceae in one of the traditional Spanish cheese, San Simón cheese, during the manufacture and ripening processes and its interrelation with the changes in some physicochemical parameters was investigated. This study was based on Enterobacteriaceae and coliform counts on selective media in samples of milk, curd, and both inner and surface zones of cheese during various ripening stages across five batches of artisan cheese. Counts were consistently higher by approximately one log unit in the inner portions than the surface. Enterobacteriaceae concentration increased during the initial week of ripening, gradually declining after but not completely disappearing by the end of ripening. [88]. The role of Enterobacteriaceae strains from dairy sources in interacting with caseins during cheese manufacturing and ripening was investigated. These strains exhibited active proteolytic systems affecting all casein fractions under cheese-making and ripening conditions. Similar effects on caseins were observed among strains within the same genus. Using specific methods, Enterobacter, Escherichia, Hafnia, and Serratia strains were isolated from fresh raw milk cheese varieties. Capillary electrophoresis was employed to determine residual caseins in cheese samples individually inoculated with ten Enterobacteriaceae strains. [89].

Actinobacteria

Actinobacteria are filamentous bacteria belonging to the phylum Actinobacteria and order Actinomycetales. They are Gram-positive and free-living saprophytes [90]. Actinobacteria is one of the dominant groups of microorganisms that produce industrially important secondary metabolites. A wide range of antibiotics, enzymes, herbicides, vitamins, pigments, larvicides, phytohormones, and surfactants are obtained from them [91]. Members of the phylum have adopted different lifestyles, and they include pathogens (Coryne­bacterium, Mycobacterium, Nocardia, Propionibacterium, and Tropheryma), soil inhabitants (Micromonospora and Streptomyces species), plant commensals (Frankia spp.), and gastrointestinal commensals (Bifidobacterium spp.) [92]. Their clinical significance is controversial because a straightforward combination of phenotypic and molecular methods to characterize Actinobacteria at the species level are still lacking [93].

Several aerobic species have been reported to be present at the surface of smear-ripened types of cheese for decades and have been used as ripening cultures in the dairy industry. One example is the traditional Turkish Divle Cave cheese, whose microbial diversity was evaluated in three independent batches. Using molecular techniques, 23 bacterial species were identified in the interior and the cheese’s outer part on days 60 and 120. Actinobacteria were predominant during the later stages of cheese ripening. They included pigment-producer coryneform bacteria such as Arthrobacter sp. and Brevibacterium sp. [85]. However, pathogenic species have also been reported within different kinds of cheese. Mycobacterium avium subsp. paratuberculosis (MAP) is a critical animal pathogen with a worldwide distribution that may contribute to human Crohn’s disease. Because of that, the behaviour of MAP in Lighvan cheese was investigated with particular reference to the strains of MAP, inoculum load, and storage time. One laboratory and two native strains of MAP were inoculated to cheese milk. The behaviour of MAP throughout the manufacture, ripening, and storage stage of Lighvan cheese was tracked using propidium monoazide (PMA) qPCR and culture examination. PMA-qPCR showed parallel outcomes in comparison with the culture assay. MAP was not affected during the ripening, and during the storage period, MAP population decreased depending on the strain in all cheese batches. MAP persisted for two months longer in cheese batches with higher initial inoculum [94].

The presence of Mycobacterium avium subsp. paratuberculosis in retail cheese types from Greece and the Czech Republic was studied, and it showed that 31.7% and 3.6% of the samples reacted positively by PCR and culture, respectively. Although at a low level for viable cells of Mycobacterium avium subsp. paratuberculosis, consuming these cheese varieties would have resulted in human exposure to M. avium subsp. paratuberculosis [95]. Sampling was conducted in artisanal kinds of cheese produced without commercial starter culture from raw sheep or goat milk on small-scale farms for the presence of Mycobacterium avium subsp. paratuberculosis. Samples were tested by qPCR and culturing. A presence of an average of 56.57% and 66.6% was found in cheese samples and farms [96]. A study was also carried out to detect Mycobacterium avium subsp. paratuberculosis in cheese curds obtained from pasteurized milk in Wisconsin and Minnesota’s northern and southern regions. The cheese curd samples were tested for MAP by PCR prescreen and culturing on selective media. MAP could not be cultured, but 1% and 5% of the samples were PCR-positive with different primer sets [97].

Metagenomic studies of cheese

Principle of the metagenomic studies

Metagenomics is a newly created technique born at the beginning of the twenty-first century and is used to determine the microbial population in an environment. When there are difficulties presented by traditional techniques in transferring all the microorganisms present in an environment to the laboratories, the use of next-generation sequencing (NGS) is justified [98]. The rise of metagenomics offers a leap forward for understanding the genetic diversity of microorganisms in many complex environments by providing a platform to identify potentially unlimited numbers of known and novel microorganisms. Next-generation sequencing (NGS) platforms can be divided, depending on the type of readings, as follows: platforms with short read sequencing technologies or second-generation technologies (AB Solid, Illumina, 454 Roche and Ion Torrent), and platforms with single-molecule real-time long read sequencing technology, or third generation technology (Pacific Bioscience and Oxford Nanopore) [99]. The short-read technologies involve massive sequencing of short (250–800 bp), clonally amplified DNA molecules sequenced in parallel [100]. Many computational tools are designed and dedicated to short-read data mining. Still, they include longer running times and difficulties with de novo assembly, haplotype phasing, and identifying transcript isoforms and structural variants [101]. For example, the Illumina platform involves PCR (Polymerase chain reaction), which is carried out by attaching DNA fragments to primers immobilized on a solid surface. The nucleotides added in DNA polymerization are chemically modified and fluorescently labelled and can emit a signal that measures total internal reflection fluorescence (TIRF) [102]. The development of the 454 sequencing platform includes higher throughput, simplifying all in vitro sample preparation, and the miniaturization of sequencing chemistries, enabling massively parallel sequencing reactions to be conducted at a scale and cost [103]. The Ion Torrent PGM detects the protons released as nucleotides are incorporated during synthesis [104]. In long-read technologies sequences >10 kb are generated directly from native DNA [105]. In Pacific Bioscence platform, a DNA polymerase is immobilized at the bottom of a well while DNA is shifted until the entire molecule has been replicated. A sensor called ZMW or zeromode waveguide is used [106]. It was reported that the top 5% of reads could be greater than 135 kb in length [104], and the technology does not require DNA amplification and thus avoids AT and GC-rich regions amplification difficulty associated with PCR [101]. Oxford Nanopore Technology (ONT) sequencing platform can generate reads greater than 1 Mb [107]. It includes better phasing of polymorphic genes, detection and proper characterization of structural rearrangements, data collection in real-time, and faster turnaround times [101]. Still, the data is subject to signal-to-noise constraints that result in higher error rates (2–15%) relative to that seen with short-read technologies [108].

Over recent decades, using high-throughput sequencing (HTS) techniques much knowledge has been accumulated about microbial communities’ composition, diversity, and structure [109]. HTS techniques have been preferred, significantly boosting resolution power and sensitivity level. The advantages of HTS technologies are undoubted, but resolution power, discrepancy between active and inactive cells, robust analytic pipelines, cost, and time reduction for integrated approaches are problems in these technologies [110]. The widespread use of omics platforms and bioinformatic tools enabled the investigation of the cheese-associated microbial community in both phylogenetic and functional terms [111].

Amplicon-based metagenomic study

In an amplicon-based approach, a PCR step is necessary to select the desired target gene. Taxonomically relevant genes are the common targets of this analysis, with the 16S ribosomal RNA (rRNA) gene being the universal target for bacteria. Reads obtained can be referred to appropriate databases. It is a way to identify the present operational taxonomic units (OTUs). Their relative abundance can also be estimated because the number of reads associated with a given taxon will be proportional to its levels in the sample, as reported by de Filippis et al. [13]. Amplicon-targeted (AT) metagenomics, based on 16S rRNA sequencing, returned a more extensive microbiota diversity at genus and species levels for artisanal starters of several PDO kinds of cheese. It was inappropriate for populations with high strain diversity and other gene targets to be tested in AT approaches [110]. Amplicon-based metagenomic analyses were applied to different types of cheese from different geographical areas to investigate microbial communities (Table 1).

Table 1. Examples of amplicon-based metagenomic studies of different kinds of cheeses worldwide

Examples	Platforms	References
Studies performed on platforms with short reads
Bulgarian Krokmach cheese	Illumina	[45]
Bulgarian Green cheese	Illumina	[112]
Colombian Paipa cheese	Illumina	[113]
Cyprian Halloumi cheese	Illumina	[114]
Greek Halitziacheeses	Illumina	[115]
Cheese from the Uighur Autonomy Region in China	Illumina	[116]
Italian Buffalo Mozzarella cheese	Illumina, Ion Torrent	[117,118]
Turkish Tulum cheese	Illumina, Ion Torrent	[119,120]
French Tomme d’Orchides cheese	Illumina	[121]
Brine-salted continental-type cheese	Roche 454	[122]
Irish artisanal cheeses	Roche 454	[123]
Mexican Bola de Ocosingo cheese	Roche 454	[124]
Mexican artisanal Adobera cheese from Los Altos de Jalisco	Ion Torrent	[120]
Studies performed on platforms with short reads
Russian local Buryat cheeses	PacBio	[125]
Austrian Vorarlberger Bergkäse	PacBio	[21]
Kazakhstan’s traditional artisanal cheeses	PacBio	[126]
Belgian artisanal cheeses	PacBio	[126]
Russian artisanal cheeses from Kalmykia	PacBio	[126]
Slovak traditional ewe’s milk cheese	Oxford Nanopores	[127]
Swiss Brie cheese	Oxford Nanopores	[128]
Swiss Camambert cheese	Oxford Nanopores	[128]

Whole shotgun metagenomic study

In a shotgun-based approach, the whole DNA or RNA is fragmented by enzymatic or mechanical method, and each piece is sequenced. The final result represents the whole genomic potential (metagenome) of the microbial populations present in the sample. Bioinformatics analysis allows the identification of the presence and abundance of specific genes [13]. Shotgun metagenomics (based on total DNA) and metatranscriptomics (based on total RNA) are potent in depicting the diversity and functionality of undefined cultures. Still, they have been rarely applied because of high costs, laboriousness, and their need for bioinformatics skills [110]. Whole shotgun metagenomic analyses were also applied to different types of cheese from different geographical areas to investigate microbial communities (Table 2).

Table 2. Examples of whole-metagenome sequencing studies of different kinds of cheeses worldwide

Examples	Platforms	References
Studies performed on platforms with short reads
Comparative study of eight different cheeses from China, Italy, and Greece	Illumina	[129]
Halitzia cheese from Cyprus	Illumina	[130]
Montenegrin brine cheeses	Illumina	[131]
Greek Kasseri cheese	Illumina	[132]
Greek Kopanisti cheese	Illumina	[133]
Argentinean artisanal cheeses	Ion Torrent	[134]
Artisanal Flemish cheese in Belgium	Ion Torrent	[135]
Studies performed on platforms with short reads
Experimentally produced cheese obtained from dairy ewes fed a dietary hemp seed supplementation	Oxford Nanopores	[136]

Examples of some metagenomic studies of different kinds of cheese

For the first time in 1986, Pace et al. [137] proposed the idea of cloning DNA directly from environmental samples to analyze the complexity of natural microbial populations. The strategy was based on shotgun cloning of 16S rRNA genes using purified DNA from natural samples. The investigation emphasizes that the methodology allowed the recovery and subsequent sequencing of individual rRNA genes, although the DNA originated from a mixed population of microorganisms [138]. In 1998 Handelsman et al. emphasized a new frontier in science, which is the mining for novel chemical compounds from uncultured microorganisms, which comprise more than 99% of the microbial diversity in the bacterial community in cow’s milk artisanal kinds of cheese from Northwestern Argentina [139].

For example, using metagenomic analyses, the microbiota of the unique Bulgarian Cherni Vit Green cheese [112] and the microbiota of Greek, goat, artisanal-type and industrial-type Gidotyri cheese variety [140] were determined for the first time. The artisanal cheese types from Belgium (unripened acid-curd kinds of cheese, smear- and mold-ripened soft kinds of cheese, and Gouda-type and Saint-Paulin-type cheese varities) were investigated concerning Listeria monocytogenes growth for the first time by using metagenomics [141].

Amplicon-based metagenomic study

Examples of amplicon-based metagenomic studies performed on platforms with short reads

Regarding the platforms with short-read sequencing technologies, the microbiota of Paipa cheese, a semi-ripened product made from raw cow’s milk in Colombia, was identified through Illumina technology. It was found that Firmicutes was the main phylum found in cheese (relative abundances: 59.2–82.0%), followed by Proteobacteria, Actinobacteria and Bacteroidetes. Lactococcus was the main genus, but other lactic acid bacteria (Enterococcus, Leuconostoc and Streptococcus) were also detected. The investigators concluded that lactic acid bacteria are the main bacterial group in Paipa type of cheese. There were other bacterial groups, including spoilage bacteria, potentially toxin producers, and bacteria potentially pathogenic to humans and/or prone to carry antimicrobial resistance genes, which are also relevant in cheese [113]. The bacterial diversity of Cyprus Halloumi traditional cheese was characterized by implying a high-throughput sequencing. The sequencing runs were performed on a MiSeq Illumina sequencing platform. It was found that the cheese’ microbiome was mainly comprised of lactic acid bacteria (LAB), but halophilic bacteria, such as Marinilactibacillus and Halomonas, pore-forming bacteria and spoilage bacteria were also identified [114]. The same platform was used to study the microbiota of the Greek Halitzia cheese. Three samples were investigated, as follows: raw goat milk without the addition of starters, pasteurized goat milk without the addition of starters and pasteurized milk with the addition of starters. The results showed the lactic acid bacteria were predominant. Coliforms and coagulase-positive staphylococci declined during ripening, and the staphylococci were not detected at the end of ripening. Yeasts were dominant throughout ripening. Thus, the investigators proved that Halitzia cheese produced from raw milk was as safe as pasteurized milk [115]. Zheng et al. revealed the diversity of bacterial and fungal communities at different time points after cheese ripening. They showed the relationship between bacterial and fungal profiles and the chemical components related to cheese flavour. The library was sequenced on an Illumina MiSeq platform with 250 bp paired-end reads. It was shown that bacteria contribute more to flavour than yeast and that fungi play an essential role in the ripening of cheese from the Kazak minority from the Uighur Autonomy Region in China [116]. Marino et al. used an NGS approach to investigate high-moisture Mozzarella cheese produced from cow and buffalo milk. Libraries were sequenced on a 300-bp read MiSeq (Illumina) platform to establish the microbiome of the cheese. A higher microbial diversity was found in the cow cheese, involving more psychrotrophic taxa. Members of Lactobacillus and Streptococcus were present in the buffalo cheese. The investigators concluded that a higher prevalence of psychrophilic species and potential spoilers was observed in samples collected in mass retail and indicated that the diversity of the microbiota would be due to the longer exposure to refrigeration temperatures and the longer time from production to consumption [117]. Dimov investigated the microbiota of Krokmach cheese, which is a traditional, fermented milk product in Bulgaria. By using next-generation sequencing and a library for the Illumina HiSeq 2 x 250 bp paired reads were prepared and were used for the NGS with 30 000 tags per sample. 75% of OTUs were found to be Lactococcus lactis subsp. Lactis, 8.5% of Exiguobacterium sp., as well as the presence of Apilactobacillus kunkeei, Megamonas sp. and Kluyvera georgiana among the dominant and subdominant species and genera [45]. In another study, Dimov et al. used 16S, and ITS2 NGS metagenomics, accomplished on the Illumina HiSeq 2 × 250 bp paired end reads to characterize the microbiota of the Bulgarian green cheese Cherni Vit for the first time. 20 eubacterial and 16 fungal species belonging to the environment were detected, paying attention to the fact that starter cultures were not used in this cheese. Beneficial microbiota has been found, which had the potential to inhibit pathogenic bacterial species as well as to neutralize produced mycotoxins [112]. 16S rRNA amplicon-based sequencing approach using an Illumina MiSeq platform was used to determine the bacterial diversity of Turkish artisanal Tulum cheese made from raw milk. Gezginc et al. found organisms belonging to Firmicutes and Proteobacteria [119]. Ceugniez et al. revealed the fungal microbiota evolution using metagenetics-based Illumina technology targeting the ITS2 domain of 5.8S fungal rDNAs. They studied the artisanal Tomme d’Orchies cheese in France. Yarrowia lipolytica and Galactomyces were found in the cheese [121].

The microbiota in a brine-salted, continental-type cheese revealed differences in bacterial diversity during ripening between cheese samples produced early and those produced late in the production day. The study was carried out by high-throughput amplicon sequencing to profile as 16S rRNA amplicons from the V4 region were sequenced on a Roche 454 FLX platform. It was found that cheese samples produced late in the day had a more diverse microbial population than their early equivalents [122]. Here, high-throughput sequencing was employed to reveal the highly diverse bacterial populations present in 62 Irish artisanal kinds of cheese and, in some cases, associated with the cheese rinds. The 16S rRNA V4 amplicons were sequenced and revealed the presence of several genera not previously associated with cheese, including Faecalibacterium, Prevotella, and Helcococcus, and, for the first time, detected the presence of Arthrobacter and Brachybacterium in goats’ milk cheese [123]. The dynamics of the bacterial community of “Bola de Ocosingo” cheese, a Mexican artisanal raw milk cheese, was investigated by high-throughput sequencing on the same platform. Dairy samples were collected from three producers in Chiapas, Mexico, during the dry (March-June) and rainy seasons (August-November). The raw milk containing high bacterial diversity was reduced throughout cheese manufacture. Streptococcus thermophilus, Lactococcus lactis, Lactobacillus helveticus, Lactobacillus delbrueckii and Lactiplantibacillus plantarum from which potential probiotic strains may be obtained, predominated during processing [124].

Sequencing results from five libraries obtained from two mixed goat and cow milk samples, one buffalo Mozzarella cheese, one goat Crescenza cheese and one artisanal cured Ricotta cheese, were able to detect all expected species, using Ion Torrent PGM sequencing platform (Thermo Fisher Scientific Inc.) [118]. The same platform was also used for the study of the microbial diversity of commercial traditional Izmir Tulum (IT) and Izmir Brined Tulum (IBT) cheese samples from Izmir, Turkey, which revealed an abundance of the species S. thermophilus and S. infantarius subsp. infantarius and genera Bifidobacteria and Chryseo­bacterium [120]. The same approach was used for studying the bacterial community of the artisanal Adobera cheese from Los Altos de Jalisco, and Streptococcus spp., Lactococcus spp., and Lactobacillus spp. were found to be dominant in this cheese. It was proven that acidification has a key role in defining the cheese core microbiota by limiting the proliferation of undesirable bacterial genera [142].

Examples of amplicon-based metagenomic studies performed on platforms with long reads

Regarding the single-molecule real-time long read sequencing technology, an investigation of the bacterial microbiota of 7 produced traditional local Buryat kinds cheese was carried out by using РacBio single-molecule, real-time (SMRT) sequencing platform. Sixty-two new LAB strains were identified, and it was noted that the bacterial microbiota differentiated by region [125]. The same platform was used to study the evolution of the rind microbiota (bacteria and fungi) throughout the ripening of Austrian Vorarlberger Bergkäse (VB) was conducted. In this case it was based on amplicon-targeted sequencing of the entire 16S rRNA genes and the fungal ITS. The results showed dynamic changes in the rind microbiota during ripening. That is, Staphylococcus equorum and Candida were dominant in the fresh products, Psychrobacter and Debaryomyces flourished at early ripening times and S. equorum, Brevibacterium, Corynebacterium, Scopu­lariopsis at the latest ripening times [21]. Kazakhstan’s traditional artisanal types of cheese have a long history and are widely consumed. The SMRT platform also explored their microbiota. The bacterial diversity profiles of six traditional artisanal cheese samples were investigated, followed by a comparative analysis of the microbiota composition between the current dataset and cheese samples originating from Belgium, the Russian Republic of Kalmykia (Kalmykia) and Italy. Lactococcus lactis, Lactobacillus helveticus, Lactobacillus delbrueckii and Streptococcus thermophilus accounted for 12.18% of the microbiome. There were differences between the bacterial communities identified in the Kazakhstan kinds of cheese compared to those in Belgium, Kalmykia, and Italy, and it was concluded that the cheese origin influenced the microbiota composition [126].

Alternatively, long reads can be generated by the Oxford Nanopores Technologies™ sequencing platform. Plany et al. used a mock bacterial community resembling the one common in cheese types from unpasteurized milk to test the potential of different bioinformatic approaches for processing MinION nanopore sequencing data. Natural bacterial communities were also analyzed. The traditional sheep cheese was obtained from Ľuboš Manica BRYSYRT company (Tisovec, Slovakia) in the summer of 2020. The ability of Nanopore sequencing on the MinION platform to provide long sequencing reads allowed the use of almost entire rrn operon sequences as markers, suitable for studies of cheese microbial communities [127]. The ONT platform was also used for the taxonomic classification of the water buffalo milk microbiota by amplifying and sequencing the full-length 16S rRNA genes, which, among the “normal” microbiota species, also allowed the identification at the species level of microbial agents of subclinical mastitis as an additional advantage [128]. The biodiversity of the mycobiota of soft cheese rinds such as Brie or Camembert in the Southern Switzerland Alps was studied by the ONT platform [143].

Whole shotgun metagenomic study

As mentioned above, in addition to amplicon sequencing, more quantitatively reliable results for classifying the microbial population can be obtained by shotgun-based sequencing (Table 3). The main advantage of whole genome shotgun sequencing (WGS) over the amplicon-based methods is the sequencing of broader regions of the genome, which sequences only a single region. That results from an adequate sequencing depth and breadth of coverage of the microbial metagenome the WGS method can generate, even including microorganisms of rare abundance and identification of specific genes in the microbiota [144]. Still, sometimes there is a limitation in the application of WGS due to the high costs, laboriousness and need for bioinformatics skills [110]. Both short- and long-read platforms can be used for whole-metagenome sequencing.

Table 3. Comparison between amplicon and shotgun sequencing

	Amplicon sequencing	Shotgun sequencing
Advantages	- Small sample volumes	- Very high resolution
	- The results are associated with a single kingdom	- Sequencing broad regions of the genome
	- Cheap
Disadvantages	- Smaller taxonomic resolution	- Results for a single kingdom are harder to distinguish
	- Variations in the number of amplicons	- Need of larger sample volumes
		- Expensive

Examples of whole-metagenome sequencing performed on platforms with short reads

Several examples of analyzed microbiomes exist regarding the sequencing technologies platforms with short reads. In many instances, the information acquired also allowed analyses of the metabolomes. For example, in a study performed on the Illumina HiSeq platform, it was revealed there were significant variations in the cheese microbiome and metabolome between samples. In the same study, deep metagenome sequencing identified four novel species (belonging to the Salinicoccus, Kocuria, and Glutamicibacter genera) [129]. Another study performed on the same platform identified a bacterial strain, Enterococcus faecalis PK23, that exhibited enhanced proteolytic activity compared to the other isolates in traditional white-brined cheese Halitzia in Cyprus [130].

Other examples are different traditional Montenegrin brine cheeses where the genomic diversity of different Leuconostoc spp. as a naturally occurring part of the non-starter lactic acid bacteria (NSLAB) microbiota, which accounts for flavour development, was studied in depth again with the Illumina platform [131]. A similar study was performed with Lactobacillus delbrueckii subsp. lactis, which is employed in the production of various types of cheese. The complete genome sequence of L. lactis ACA-DC 178 isolated from Greek Kasseri cheese was sequenced using the Illumina HiSeq 2000 platform [132].

A further advantage of such kinds of studies is that the extraction of whole genomes allows assessment of the role of the different microorganisms in the ripening process. For example, the first complete genome sequence of Loigolactobacillus rennini ACA-DC 565, a strain isolated from a traditional Greek over-ripened Kopanisti cheese called Mana, can be pointed out. Although the species have been associated with cheese spoilage, the strain ACA-DC 565 would contribute to the intense organoleptic characteristics of Mana cheese. Again, the Illumina HiSeq 2000 platform was used [133].

Similar results can also be obtained by using the IonTorrent PGM platform. Suárez et al. described the bacterial diversity of three Northwestern Argentinean artisanal cheeses. Streptococceae and Enterococceae were dominant in the artisanal cheese microbiota, and the bacterial taxa Macrococcus caseolyticus and Streptococcus macedonicus were also found. Using the bacteriocin mining software BAGEL3, two ORFs encoding antimicrobial peptides were identified. In the same study, the genomic sequence of Enterococcus faecium CRL 1879 was obtained, which demonstrated its inhibitory capacity against Listeria monocytogenes. It was concluded that the expression of bacteriocin genes in a food matrix revealed the potential of this strain in food preservation [134]. Metagenome shotgun sequencing on the same platform was also used to investigate the taxonomic diversity of the microbial communities of thirty stable old cheese brines from an artisan and large-scale cheese producer in Flanders (Belgium). The results showed that the genera Tetrag­enococcus, Chromohalobacter, and Halanaerobium were the most abundant in both brines [135]. The purpose of this study was to characterize the phenotype and genotype of 11 M. morganii isolated from cheese regarding the BA (biogenic amines) formation. The cheese isolates’ phylogeny, trehalose fermentation ability, and antibiotic resistance were also ex­­plored [124].

The Roche 454 platform was also widely used for whole-metagenome sequencing of cheese microbiomes and extraction of entire genomes. An example is Ligilactobacillus acidipiscis KCTC 13900, which is essential for generating particular flavours and other ripening processes associated with specific cheeses and whose draft genome sequence was obtained [145]. Similarly, in other studies, the genome of Lacticaseibacillus casei DPC6800 was extracted, which is a nonstarter lactic acid bacterium isolated from a semi-hard Dutch cheese [146], Staphylococcus saprophyticus DPC5671 from cheddar cheese was determined [147], Lactiplantibacillus plantarum 19L3 from traditional Slovak cheese Slovenská bryndza cheese [148] etc.

Examples of whole-metagenome sequencing performed on platforms with long reads

Long reads sequencing platforms can also be used for metagenomic studies. Because of the longer read lengths, they allow easier extraction of whole microbial genomes from the metagenomes - which may be considered their main advantage, together with eliminating the biases caused by the PCR step in the library preparation for amplicon-based metagenomics. However, on the other hand, the experimental cost can be pointed out as a disadvantage, as well as the high-performance requirements for the computers on which the bioinformatic analyses are performed. Because the mentioned biases are less presented in the ONT platform, this platform is the first choice for whole-metagenomic analyses of cheeses. Despite the high experimental price, the PacBio platform is also used, mainly because of its meagre sequencing error rate.

An example of such a study in the ONT platform is the investigation of cheese obtained from dairy ewes fed a dietary hemp seed supplementation. The phylum Proteobacteria and Firmicutes were equally abundant in control and experimental raw milk samples. Bacteroidetes were less abundant. Firmicutes were predominant in all cheese samples, with Streptococcaceae being the most abundant family in the experimental group [149].

Conclusion

This review explored and highlighted the importance of metagenomic analysis in the microbiota studies in artisanal regional cheeses, the microbiota’s importance for cheeses’ organoleptic qualities, and the health benefits of probiotic dairy products. Using innovative, modern methods of analysis of the microbiota in cheeses, such as metagenomic studies, would allow us to obtain more accurate and rapid results. It will eventually allow the conduct of proteomic and metabolomic analyses necessary for the complete characterization of different cheeses. Future studies in this area and expansion of the knowledge of the microbiota of artisanal regional cheeses would have an impact on improving the optimization of cheese production processes. They would encourage the intake of probiotic functional foods to improve human health.

Authors’ contributions

Vilma Posheva: Conceptualization, Methodology, Formal analysis, Investigation, Resources, Writing - Original Draft, Visualization. Tsvetana Muleshkova: Formal analysis, Investigation, Writing - Review & Editing. Slavica Josifovska: Conceptualization, Writing - Review & Editing. Stoyan Chakarov: Conceptualization, Writing - Review & Editing. Svetoslav Dimov: Conceptualization, Methodology, Formal analysis, Investigation, Resources, Supervision, Project administration, Funding acquisition.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Data sharing does not apply to this article as no new data were created or analyzed in this study.

Additional information

Funding

This work was supported by the Bulgarian National Science Fund under Grant number КП-06-66/6 from 13.12.2022.