Characterization of Vegetative Bacillus cereus and Bacillus subtilis Strains Isolated from Processed Cheese Products in an Italian Dairy Plant

Processed cheese is a commercial product characterized by high microbiological stability and extended shelf life obtained through the application of severe heat treatment. However, spore-forming bacteria can survive through thermal processes. Among them, microorganisms belonging to Bacillus genus have been reported. In this study, we examined the microbiological population of the first hours’ production of processed cheeses in an Italian dairy plant during two seasons, between June and October 2020. Matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS) was used to identify bacteria colonies, allowing the isolation of Bacillus cereus and Bacillus subtilis strains. These results were further confirmed by amplification and sequencing of 16 rRNA bacterial region. A multi-locus sequence type (MLST) analysis was performed to assess the genetic similarity among a selection of isolates. The fourteen B. cereus strains showed two sequence types: ST-32 was observed in only one strain and the ST-371 in the remaining thirteen isolates. On the contrary, all twenty-one B. subtlis strains, included in the study, showed a new allelic profile for the pycA gene, resulting in a new sequence type: ST-249. For B. cereus strains, analysis of toxin genes was performed. All isolates were positive for nheABC, entFM, and cytK, while hblABCD, bceT, and ces were not detected. Moreover, the biofilm-forming ability of B. cereus and B. subtilis strains was assessed, and all selected isolates proved to be biofilm formers (most of them were stronger producers). Considering the genetical similarity between isolates, jointly with the capacity to produce biofilm, the presence of a recurring Bacillus population could be hypothesized.


Introduction
Processed cheeses are produced by melting into a homogeneous molten blend, one or more cheeses with other ingredients (cream, butter, whey) and usually emulsifying salts [1,2]. The mixture is then blended, heated, and portioned, according to the commercial use. The temperature and the time of treatments varies depending on the formulation, the desired product texture, and shelf-life characteristics. Processing may be carried out in batch cookers in which temperature-time treatment varies (e.g., 70-95 • C for 4-15 min), by obtaining pasteurized processed cheese or in continuous cookers in which the blend is typically heated with direct steam injection to ∼140 • C for ∼10 s, or by manufacturing sterilized high-moisture processed cheese [1].
The thermal processes allow the inactivation of most vegetative forms, but bacterial spores can survive [3]. Spore-forming bacteria are ubiquitous, founded mainly in the soil, but they are also able to colonize the gastrointestinal tract of insects and animals [4]. They belong to the Firmicutes phylum and, among them, Bacilli and Clostridia are the most relevant in the dairy industry [5]. Due to their ubiquity in nature, it is a real challenge A dairy cheese processing plant located in northern Italy served as a pilot site for this research. Hard cheeses, mainly cheddar, are used in addition to milk, butter, emulsifying salts, and other ingredients to produce processed cheeses, as shown in Figure 1.
Thirty-two batches of processed cheeses were analyzed during eight weeks, from June to October 2020. Sixteen batches were collected in summer and the other sixteen in autumn. For each batch, we examined the first 300 g produced within the first hours of processing, since the characteristics of the production line direct towards a possible greater contamination of the product during the initial phases of production.

Isolation and Enumeration of Bacterial Colonies
Twenty-five grams of cheese samples were suspended with 225 mL of a sterile physiological saline solution and peptone (Oxoid, Basingstoke, UK) using the Stomacher ® 400 Circulator (Seward Limited, Worthing, UK) for 1 min at a medium speed. The homogenate was serially diluted, and the count of vegetative cells was performed.
For enumeration and colonies identification of aerobic bacteria each dilution was spread-plated in duplicate onto a plate count agar (Oxoid, Basingstoke, UK), incubated at 30 • C and 37 • C for 24 h/48 h.
For presumptive anaerobic microorganisms, iron sulfite agar (Oxoid, Basingstoke, UK) was used, and the plates, in duplicate, were incubated under anaerobic conditions at 37 • C and 50 • C for 24 h/48 h. . Flow chart of processed cheese production.

DNA Extraction
For DNA extraction, isolated conies were inoculated in brain heart infusion (BHI) broth (Oxoid, Basingstoke, UK) and incubated overnight at 30 °C. DNA extraction was performed using the DNeasy Blood and Tissue Kit (Qiagen Ltd., Manchester, UK), according to the manufacturer's instructions. The lysis step was carried out overnight at 37 °C with enzymatic lysis buffer (20 mM Tris·Cl, pH 8.0, 2 mM sodium EDTA, 1.2% Triton ® X-100) and a solution of freshly prepared lysozyme (20 mg/mL). DNA samples were eluted in 50 μL of 10 mM Tris·Cl pH 8.0 and stored at −20 °C until used. DNA yield and purity were determined using Nano drop 2000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA).

Isolated Colonies Identification by Matrix Assisted Laser Desorption Ionization-Time of Flight Mass Spectrometry (MALDI-TOF MS) and 16S rRNA Sequencing
The identification of single colonies grown on plate count agar was performed using matrix-assisted laser desorption-ionization time-of-flight mass spectrometry (MALDI-TOF MS) analysis in a Microflex LT MALDI Biotyper mass spectrometer (Bruker Daltonics, Hamburg, Germany).
For each plate, at least five colonies were selected according the morphotypes. Briefly, colonies were directly spotted on the MALDI plate and treated with 1 μL of formic acid and then overlaid with 1 μL of saturated α-cyano-4-hydroxycinnamic acid (HCCA) matrix solution which was then allowed to dry at room temperature. Each colony was tested in duplicate. The acquired mass spectra were imported into the MALDI Biotyper software (version 3.1) (Bruker Daltonics, Hamburg, Germany) and compared to a collection of different spectra stored in a database. The identification of an organism is expressed as a score based on pattern matching between samples and reference spectra. . Flow chart of processed cheese production.

DNA Extraction
For DNA extraction, isolated conies were inoculated in brain heart infusion (BHI) broth (Oxoid, Basingstoke, UK) and incubated overnight at 30 • C. DNA extraction was performed using the DNeasy Blood and Tissue Kit (Qiagen Ltd., Manchester, UK), according to the manufacturer's instructions. The lysis step was carried out overnight at 37 • C with enzymatic lysis buffer (20 mM Tris·Cl, pH 8.0, 2 mM sodium EDTA, 1.2% Triton ® X-100) and a solution of freshly prepared lysozyme (20 mg/mL). DNA samples were eluted in 50 µL of 10 mM Tris·Cl pH 8.0 and stored at −20 • C until used. DNA yield and purity were determined using Nano drop 2000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA).

Isolated Colonies Identification by Matrix Assisted Laser Desorption Ionization-Time of Flight Mass Spectrometry (MALDI-TOF MS) and 16S rRNA Sequencing
The identification of single colonies grown on plate count agar was performed using matrix-assisted laser desorption-ionization time-of-flight mass spectrometry (MALDI-TOF MS) analysis in a Microflex LT MALDI Biotyper mass spectrometer (Bruker Daltonics, Hamburg, Germany).
For each plate, at least five colonies were selected according the morphotypes. Briefly, colonies were directly spotted on the MALDI plate and treated with 1 µL of formic acid and then overlaid with 1 µL of saturated α-cyano-4-hydroxycinnamic acid (HCCA) matrix solution which was then allowed to dry at room temperature. Each colony was tested in duplicate. The acquired mass spectra were imported into the MALDI Biotyper software (version 3.1) (Bruker Daltonics, Hamburg, Germany) and compared to a collection of different spectra stored in a database. The identification of an organism is expressed as a score based on pattern matching between samples and reference spectra. The degree of matching is expressed as a logarithmic score with values ≥ 2.00 to indicate high confidence species identification. Scores of 1.70-1.99 and <1.70 indicate, respectively, low confidence and no organism identification [23]. Only isolates with the highest score value of mass spectra were kept frozen at −20 • C, and considered for further analysis.
As a complementary approach to MALDI-TOF MS, sequencing of the 16S rRNA gene was performed to confirm strains identification. Bacterial DNA was amplified by PCR using 16S rRNA gene targeting primers 8F (5 -TGGAGAGTTTGATCCTGGCTCAG-3 ) and 533R (5 -TACCGCGGCTGCTGCTGGCAC-3 ) (Sigma Aldrich, Darmstadt, Germany) [24]. PCR reactions were performed in a volume of 50 µL, consisting of 1 U of recombinant Taq DNA polymerase (Thermo Fisher Scientific, Waltham, MA, USA), 0.2 mM each of dNTPs (Thermo Fisher Scientific, Waltham, MA, USA), 2 mM MgCl 2 , 25 pmol of each primer, and 50 ng of DNA. PCR conditions were 10 min at 94 • C followed by 30 cycles at 94 • C for 30 s, 60 • C for 30 s, and 72 • C for 45 s, and a final extension at 72 • C for 10 min. The amplification products were purified with an ExoSAP kit (Thermo Fisher Scientific, Waltham, MA, USA) and then sequenced with SeqStudio Genetic Analyzer (Applied Biosystems, Bedford, MA, USA) using the BigDye Terminator v 1.1 cycle sequencing kit (Thermo Fisher Scientific, Waltham, MA, USA). Subsequently, sequences were aligned in the GenBank database using the web-based basic local alignment tool BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi/, accessed on 4 March 2021) to identify species with highest similarity and they were aligned against each other using a Clustal W program [25].
PCR amplifications were performed using Taq DNA polymerase recombinant (Thermo Fisher Scientific, Waltham, MA, USA), in a volume of 50 µL with the following cycling profile: a denaturation step at 94 • C for 10 min, followed by 35 cycles at 94 • C for 1 min, 54-60 • C for 1 min and 72 • C for 45 s, and final extension at 72 • C for 10 min. PCR products were purified using an ExoSAP Kit protocol (Thermo Fisher Scientific, Waltham, MA, USA) and sequenced with BigDye Terminator 1.1 (Thermo Fisher Scientific, Waltham, MA, USA). The samples were analyzed by capillary electrophoresis using SeqStudio genetic analyzer (Applied Biosystems, Bedford, MA, USA). DNA sequences obtained after amplifications were aligned and compared to sequences available on the pubMLST database (https://www.pubmlst.org/, accessed on 4 March 2021). Finally, an allele number was assigned to each gene and the combination of the seven alleles defined the sequence type (ST), which is used to categorize the target strains.

Analysis of Toxin Genes
Isolated colonies were tested for the presence of enterotoxin genes. Five different enterotoxins (HBL, NHE, CytK, BceT, and entFM) and one emetic toxin were examined [27]. The set of primers used, and the amplification conditions were performed, as previously described [28].
DSM4282 and DSM4312 were used as reference strains for diarrhoeagenic and emetic gene sets. The PCR products were analyzed in 2% agarose gel. The oligonucleotides were validated in simplex, and all PCR experiments were performed twice for each strain.

Screening of Biofilm Forming Strains Using Micro-Method Assay
Eight isolates were tested for biofilm forming ability and were selected as representative of: (i) species identified; (ii) collection's season; and (iii) genetic profile. The experiments were performed according to the protocol previously described by Stepanovic et al. [29].
Subsequently, 200 µL of inocula accurately diluted were added to 96-well polystyrene microplates (Sarstedt, Nümbrecht, Germany), in triplicate for each isolate, while the negative control wells contained the uninoculated broth. The microplates were incubated at 30 • C for 24 h in static conditions. After the incubation, the medium was discarded and each well was carefully washed three times with 300 µL of PBS sterile phosphate buffer saline solution (PBS, pH 7.3, Oxoid, Basingstoke UK) to remove unattached cells. Biofilms were heat-fixed at 60 • C for 1 h and then stained with 150 µL of 2% w/v crystal violet solution (Chem-lab, Zedelgem, Belgium) for 15 min. Unbound crystal violet was discarded by washing each well 3 times with distilled water and dried at 37 • C for 15 min. To quantify the biofilm formation, 150 µL of 95% ethanol solution (Honeywell, Charlotte, NC, USA) were added to each well, and the absorbance was measured at 595 nm on a microplate reader (iMark plate reader, Sydney, NSW, Australia). A mean of OD values was calculated for each strain (ODs) while the cut-off OD (ODc) was calculated as three standard deviations above the mean OD of the negative control [21]. The strains were classified as weak (ODc < ODs ≤ 2 × ODc), moderate (2 × ODc < ODs ≤ 4 × ODc), strong (4 × ODc < ODs), or no (ODs ≤ ODc) biofilm producers.

Statistical Analysis
The study of temporal trend of microbial counts was evaluated by the Wilcoxon test to compare the total bacterial median values of 16 batches during two seasons (n = 32). For biofilm forming ability, data were analyzed performing a one-way analysis of variance (ANOVA) followed by a Tukey's multiple comparison test. Statistical analyses and graphing were conducted with GraphPad Prism version 8.4.3 (GraphPad Software, San Diego, CA, USA). Differences were considered statistically significant when p values were less than 0.05.

Enumeration of Vegetative Bacterial Contamination and Identification of Isolates
Only viable microbiota was quantified and studied. No vegetative anaerobic bacteria were identified during the entire considered period (a number < 10 CFU/mL was detected). Cultures in PCA showed total bacterial load of vegetative cells varied over time, with a highest density observed in summer (range of 2.0-2.4 Log CFU/mL), whereas lowest levels occurred in autumn (range of 1.3-1.6 Log CFU/mL) with statical significance differences assessed during the two seasons (p < 0.001) ( Figure 2).
During the entire collection period, a total of 103 colonies were identified by MALDI-TOF with high confidence score. Results showed the presence of microorganisms of Bacillus genus, in particular B. subtilis and B. cereus (Table 1). A prevalence of B. subtilis was evident in all 4 weeks of autumn season and in 2/4 weeks of summer period, while during week I (summer), the number of B. cereus was higher compared to B. subtilis and, in the week III (summer), a balance of both species was highlighted (Table 1).
Moreover, to support MALDI identification, an independent molecular analysis was performed. The sequencing of 16S rRNA was assessed on thirty-five isolates, including both species identified and also representative of the collection weeks (Supplementary  Table S1). Results showed a sequence's homology >99.9% both for B. subtilis and for B. cereus, which were in accordance with MALDI-TOF MS identification.
As a result, all thirty-five isolates, fourteen B. cereus, and twenty-one B. subtilis were subjected to further characterizations (Supplementary Table S1).  During the entire collection period, a total of 103 colonies were identified by MALDI-TOF with high confidence score. Results showed the presence of microorganisms of Bacillus genus, in particular B. subtilis and B. cereus (Table 1). A prevalence of B. subtilis was evident in all 4 weeks of autumn season and in 2/4 weeks of summer period, while during week I (summer), the number of B. cereus was higher compared to B. subtilis and, in the week III (summer), a balance of both species was highlighted (Table 1). Moreover, to support MALDI identification, an independent molecular analysis was performed. The sequencing of 16S rRNA was assessed on thirty-five isolates, including both species identified and also representative of the collection weeks (Supplementary

Molecular Characterization by MLST Analysis
Basing on sequence alignment of multiple housekeeping genes, the fourteen strains of Bacillus cereus grouped in two different sequence types (Table 2A). The most common ST group was ST-371, which was detected in thirteen isolates, while in one isolate (BC_2), ST-32 was identified (Table 2A).
For Bacillus subtilis, six of seven allelic profiles matched with the existing types. A new allelic profile was identified in the pycA gene, and it was deposited and registered as "108" (https://pubmlst.org/bigsdb?db=pubmlst_bsubtilis_seqdef&page=alleleInfo& locus=pycA&allele_id=108, accessed on 4 March 2021). This new allele showed four SNPs of difference from the more similar allele "107". Moreover, the presence of the new allele, and the combination of the other ones, produced a new sequence type, which was submitted and registered as "ST-249" (https://pubmlst.org/bigsdb?page=profileInfo&db= pubmlst_bsubtilis_seqdef&scheme_id=1&profile_id=249, accessed on 11 March 2021). All isolated strains of B. subtilis showed the same sequence type (Table 2B).

Prevalence of Toxigenic Determinants in Bacillus cereus Isolates
The distributions of the diarrheal enterotoxin genes and emetic toxin cerulide genes in the fourteen Bacillus cereus isolates are shown in Table 3. Table 3. Toxin profile of Bacillus cereus isolates included in the study. Results +/− indicate presence or absence of genes examined.

HBL Complex NHE Complex Other Enterotoxins Emetic Strains
hblA hblC hblD nheA nheB nheC entFM bceT cytK ces In the present analysis, the presence of hemolysin hbl was not detected. All isolates were positive for all three genes coding for non-hemolysin enterotoxin nhe. Positivity for entFM and cytK was also detected in all analyzed bacterial isolates. In contrast, the bceT gene and emetic strains were not identified (Table 3).

Biofilm Forming Potential
Eight isolates were tested for biofilm formation, according to the criteria, as previously reported (see Section 2.7). For B. cereus, the strain tested and isolated with ST-32 was collected in summer (BC_2). Among another three sets of strains with ST-371, one derived from summer (BC_14) and the other two were from autumn (BC_36 and BC_44) (Figure 3). For B. subtilis, two isolates collected during summer (BS_8 and BS_23) and two in autumn (BS_42 and BS_54), all with the same sequence type (ST-249) were included (Figure 3). Eight isolates were tested for biofilm formation, according to the criteria, as previously reported (see Section 2.7). For B. cereus, the strain tested and isolated with ST-32 was collected in summer (BC_2). Among another three sets of strains with ST-371, one derived from summer (BC_14) and the other two were from autumn (BC_36 and BC_44) ( Figure  3). For B. subtilis, two isolates collected during summer (BS_8 and BS_23) and two in autumn (BS_42 and BS_54), all with the same sequence type (ST-249) were included ( Figure  3). Results showed that all isolates produced biofilms: five (5/8) were classified as strong producers (BC_2, BC_14, BS_23, BS_42, and BS_54), one moderate (BC_44), and two (2/8) weak (BC_36 and BS_8) biofilm producers, with statistical significance difference observed ( Figure 3).

Discussion
Although many scientific reports are focused on dairy products, few studies have attempted to characterize processed cheeses. An understanding of the microbiological quality of processed cheese is essential for answering questions related to the food quality and safety of these dairy products. Thus, the main objective of our study was to describe the residual microbiological population at the end of processing of processed cheeses in a dairy plant located in northern Italy.
Isolated bacterial colonies grown in PCA were identified by MALDI-TOF MS. In recent years, the use of mass spectrometry for microbial identification rapidly increased. Conventional culture methods, although remaining fundamental, are often time-consuming since the preparation of one or more subcultures for a species level identification could be necessary. In this contest, MALDI-TOF represents a suitable method for the identification of culturable microorganisms through the analysis of their protein profiles [30], with the advantage to be a fast and cost-effective technique. Moreover, microbial identification was also confirmed with an independent molecular method based on the sequence of 16s rRNA regions, with a complete concordance among the two techniques. Our results showed a homogenous bacterial population belonging to Bacillus genus, as both B. subtilis and B. cereus species were identified. Results showed that all isolates produced biofilms: five (5/8) were classified as strong producers (BC_2, BC_14, BS_23, BS_42, and BS_54), one moderate (BC_44), and two (2/8) weak (BC_36 and BS_8) biofilm producers, with statistical significance difference observed ( Figure 3).

Discussion
Although many scientific reports are focused on dairy products, few studies have attempted to characterize processed cheeses. An understanding of the microbiological quality of processed cheese is essential for answering questions related to the food quality and safety of these dairy products. Thus, the main objective of our study was to describe the residual microbiological population at the end of processing of processed cheeses in a dairy plant located in northern Italy.
Isolated bacterial colonies grown in PCA were identified by MALDI-TOF MS. In recent years, the use of mass spectrometry for microbial identification rapidly increased. Conventional culture methods, although remaining fundamental, are often time-consuming since the preparation of one or more subcultures for a species level identification could be necessary. In this contest, MALDI-TOF represents a suitable method for the identification of culturable microorganisms through the analysis of their protein profiles [30], with the advantage to be a fast and cost-effective technique. Moreover, microbial identification was also confirmed with an independent molecular method based on the sequence of 16s rRNA regions, with a complete concordance among the two techniques. Our results showed a homogenous bacterial population belonging to Bacillus genus, as both B. subtilis and B. cereus species were identified.
The presence of spore-forming bacteria, in particular Bacillus spp., has been previously reported in processed cheeses [3]. Several studies demonstrated the ability of spore-forming bacteria to survive after severe heat treatment [31][32][33].
In a study focused on dairy processing and food products, authors founded B. cereus as the most frequent species identified, followed by B. subtilis and B. licheniformis [34]. An incidence of B. cereus around 4.0-5.0 log CFU/mL was also reported in pasteurized and sterilized milk [16,35]. A comparable degree of contamination was highlighted in artisanal Mexican cheeses, in which the incidence of B. cereus was about 30% [36]. A greater prevalence (35-46%) was reported in raw milk and Ghanaian traditional milk products, but with a minor total count (2.0-3.2 log CFU/mL), while the research of B. cereus prevalence in soil samples was considerably higher, i.e., around 70% [33]. In an Italian dairy plant, the presence of B. cereus was monitored on different areas of processing and in the final product [37]. The microorganism was founded in environmental samples, while no detection was assessed in dairy product [37]. In a study by Liu et al. [38], liquid milk and milk powder showed a high prevalence of B. cereus around 44% and 26%, respectively. Lower percentages of B. cereus were founded in raw milk (26%), pasteurized milk (12%), cheese (10%), UHT milk (8%), and powdered infant formula (7%) [31].
In our study, only two Bacillus species were identified. This evidence could be related to the technology of the production process. Anyhow, despite species-specific differences in inactivation temperatures observed among Bacillus genus, i.e., in the case of B. cereus and B. subtilis, heat resistance temperatures were similar [39].
This extraordinary spore resistance could generate problems for the food industry, as the spores survive many of the currently applied food preservation treatments and may cause spoilage [3,40]. It has been reported that Bacillus spp., especially B. cereus, produced various extracellular enzymes which can alter the organoleptic quality of milk and dairy products [41]. Further studies will be necessary to assess the ability of our isolates to produce proteases and lipases.
In consideration of high diffusion of Bacillus spp. in the environment and in food, various molecular typing methods were used to evaluate the genetic relationship between isolates. Among them, a random amplified polymorphic DNA polymerase chain reaction (RAPD_PCR) was proposed as a tool to explore the genetic diversity in B. subtilis [42] and B. cereus [43], using short random primers to amplify multiple DNA fragments. This kind of method requires time-consuming and laborious preparation. It may also be subjected to great variability and, therefore, might be difficult to perform for large-scale analyses. Other genotyping methods are based on sequencing of specific housekeeping genes, such as panC [44] and rpoB [45], to distinguish very closely related ecological populations.
In this study, to explore the phylogenetic relationships between isolates, we used MLST analysis which is based on direct nucleotide sequencing of seven housekeeping genes with the advantages of being definitive and standardizable [26]. Results showed a high genetic similarity among the isolates. In the case of the fourteen Bacillus cereus included in the analysis, thirteen of them showed the sequence type 371, while in only one strain (BC_2), the ST-32 was detected. The ST-371 has been previously isolated in microorganisms derived from eggs [46] and, to our knowledge, there are no other identifications in dairy products. On the contrary, ST-32 has been described as a contaminant in milk [47] and other sources of an unspecified origin (https://www.pubmlst.org/, accessed on 4 March 2021). All twenty-one B. subtilis isolates showed a common sequence-type, ST-249, which is reported for the first time in this study.
This genetic homogeneity was also assessed by examining the presence of toxin genes for B. cereus. All the fourteen isolates showed the same toxigenic profile. Previous studies reported that nearly all strains of Bacillus cereus possess the genes of at least one of the diarrheal enterotoxins [27,48,49]. In our analysis, all isolates were positive for the three genes coding for non-hemolysin enterotoxin nhe, several authors reported that B. cereus strains isolated from different food sources possess the nhe genes [50][51][52]. Positivity for entFM and cytK was also detected in all analyzed isolates. An analogous result for the cytK gene was reported by Tirloni et al. [37] while a high, but not total, prevalence of both genes was described by Owusu-Kwarteng et al. [33]. In contrast, the hbl, bceT gene and emetic strains were not identified, and the results were in accordance with recent studies which underline the low frequency of emetic B. cereus strains in food products [33,53]. Anyhow, the assessment of genetic positivity of toxin genes in these B. cereus isolates is not necessarily correlate to toxins production, and, therefore, it cannot be associated to a concrete health risk [54].
The strong genetic similarity highlighted in the isolates could be related to a contamination of raw materials, to other external sources, or to the presence of a bacterial population in the processing lines.
Moreover, a role in microbial persistence could be linked to the presence of high percentages of fat in processed cheeses, which could be represented as a protective factor. In a study which monitors the manufacture and aging of cheddar cheese, characterized by different fat percentage, authors observed that the microbial load decreased faster in the group with a reduced percentage of fat than in the control group in which the fat content was higher [55]. A recent study carried out in peanut oil demonstrated that this matrix with high fat content protected bacteria from thermal inactivation by desiccation of bacterial cells with a decrease in water activity [56]. In this sense, fat may provide a microenvironment which may protect bacteria from antimicrobials.
Another hypothesis of recurring contamination could be related to the ability of Bacillus spp. microorganisms to form biofilm [21]. As support, we showed that tested isolates (n = 8) form biofilm: 5/8 classified as strong, 1/8 moderate and 2/8 weak producers. It well known that biofilms could become a reservoir of bacteria much more difficult to reach by common cleaning in place practices. Most of these cleaning systems are designed and optimized to remove milk residues, with variable efficiency in eliminating surfaceadherent bacteria [57]. This limitation may lead to the accumulation of microorganisms on the equipment surfaces, resulting in biofilm formation [58], therefore constituting a source of continuous recontamination of the products.

Conclusions
In this study, a recurrent contamination of microorganism belonged to Bacillus spp., specifically B. cereus and B. subtilis, was highlighted in a dairy plant. Molecular characterization by MLST showed a high genetic similarity between isolates of the same species. Moreover, a study of toxigenic genes showed the same genetic profile in all B. cereus isolates. These results could be related to the presence of a microbial population inside the dairy plant. This hypothesis may be reinforced by the study of biofilm-forming ability, which showed that most of the isolates are characterized by a robust biofilm formation. Biofilms can be formed in different sites of processing lines, contributing to a recurrent contamination of the product. However, since it cannot be excluded, an initial high bacterial count in raw materials, further studies will be necessary to examine this hypothesis, jointly with the sampling of environmental superficies. In conclusion, the presence of Bacillus spp. in the final product, even if in low counts, represent a warming signal which could compromise the suitability of processed cheeses during the shelf life.