3.1. Physicochemical Parameters during Cheesemaking and Ripening
The physicochemical characteristics of the two different groups of cow’s cheese samples are listed in
Table 1.
The pH and aw values were decreasing throughout the cheesemaking, and the salt concentration was increasing in all the studied cheeses. At the beginning and the end of ripening, the physicochemical characteristics of cheese artificially contaminated with
E. coli strains (pH, 4.99 ± 0.14 and 5.11 ± 0.04; aw, 0.96 ± 0.01 and 0.87 ± 0.01; NaCl, 2.60 ± 0.63% and 2.98 ± 0.37%) did not differ significantly (
p > 0.05) from those found in control cheeses (pH, 5.19 ± 0.02 and 5.10 ± 0.09; aw
, 0.97 ± 0.01 and 0.88 ± 0.01; NaCl, 1.90 ± 0.42 and 3.25 ± 0.17). Therefore, both groups of hard cheeses were homologous according to these intrinsic factors and were within the typical range in similar commercial cheeses in Spain [
32,
33]. When studying the bacterial effectiveness of the whole process to obtain matured cheese, not only must the target microorganisms be considered, but also the interactive effects of a number of factors, some of them intrinsic parameters such as the ones determined herein. Even though
E. coli belongs to the Enterobacteriaceae family and possesses important spoilage capabilities, intentional addition of this bacterium to the milk destined to cheesemaking makes no difference in terms of physicochemical characteristics of cheese. A few research works have addressed that survival of
E. coli strains can be enhanced by cross-protection when subjected to combinations of stresses such as acid, salt and other factors [
34]. It can be hypothesised that a synergistic effect of these different hurdle agents could be observed in cheese and that the sequence of application of different hurdles is important, but the majority of studies on growth / survival of STEC in cheese were not carried out with non-O157 serotypes [
2].
3.2. Analysis of E. coli Populations in Cheese Samples
Milk used for the manufacture of cheese was negative for both
stx and
eae genes, thus free of STEC or EPEC contamination. Moreover, counts of
E. coli on TBX-plates were ~1 log cfu/mL in the milk and were not above 2.34 log cfu/g (
Figure 1) at their highest concentration (curd formation step) throughout the manufacturing process. Thus, this bacterial indicator was compliant with the hygienic requirements established in the Commission Regulation (EC) No 2073/2005 of 15 November 2005 [
35].
Figure 1 shows the behaviour of bacterial populations during the cheesemaking process and cheese ripening (control and spiked samples). After inoculation of milk with both strains, counts were not significantly different (
p > 0.05), reaching 3.14 ± 0.44 log cfu/mL for STEC MK116C19 and 3.49 ± 0.28 log cfu/mL for aEPEC MK127C9. STEC and aEPEC contamination of cow’s milk, though it is reported as ranging from 2 to 3% and ~6 to 10%, respectively [
8], is not expected to reach the concentrations used in the spiked samples. These initial
E. coli numbers would simulate a possible worst case scenario for food safety taking into consideration that contamination of milk with
E. coli may certainly arise from milking to its final processing depending on how hygienic measures and manufacturing practices are carried out. The source of cheese contamination with foodborne pathogens may not only be raw milk, but also may occur during cheesemaking, aging, or storage from various sources, such as the operators, the plant environment, and the equipment [
12]. It must be also highlighted that counts of STEC in cheese can be higher than usually reported, considering the occurrence of small groups of cells imbibed in the cheese matrix [
27]. Moreover, low temperatures used during cheesemaking exert little inhibitory effect, in view of the fact that STEC strains appear to develop survival mechanisms for growing under chilled temperatures as was reported by Vidovic et al. [
36].
The initial levels of the two tested strains steadily increased until the curd formation and a similar trend was observed on indigenous
E. coli counts (
stx -/
eae -) in control cheese. All studied bacterial populations showed the highest counts during the curd formation after an increase of more than 1 log cfu/mL, in accordance with previously reported studies [
22,
23,
24]. Thus, this initial multiplication is attributed to favourable growth conditions for
E. coli at the beginning stages of cheesemaking, principally temperature (close to 37 °C), aw (around 0.99), and pH (greater than 5). Moreover, the significant increase in the studied pathogenic strains in the transition from milk (liquid) to curd (solid) can be mainly caused by bacterial concentration, owing to the fact that the curd behaves as a physical barrier trapping the bacteria and expelling the whey, concentrating up to ten times [
24]. Thereafter, both autochthonous bacteria and the tested
E. coli strains decreased their level throughout cheese aging.
3.3. Fate of aEPEC and STEC Strains in Raw Milk Cheese
A different behaviour between the diarrhoeagenic
E. coli strains was noted after cheese pressing, with STEC counts decreasing significantly faster than aEPEC strain counts (
p < 0.05). From that point, and during the cheese ripening, low pH, ranging from 4.99–5.15 (
Table 1 and
Figure 2), could be the factor explaining STEC behaviour. Oh et al. [
37] found that, among several studied parameters in Cheddar cheese extracts, the low pH mainly affected the behaviour of STEC strains. It has been reported that although the optimum pH for STEC multiplication is around 7, it is capable of growing in a pH range between 4.5 and 9. Thus, these bacterial group find suitable pH for its multiplication in milk (pH 6–7) and also in foodstuffs of moderate acidity, such as some types of cheeses [
15].
During the cheese maturation, the decrease in STEC and aEPEC populations was significant (
p < 0.05) from day 7 and day 15, respectively, whereas indigenous
E. coli population (
stx-/
eae-) on control cheese was below detection limit (<0.40 log cfu/g) after 30 d of ripening (
Figure 1). The low pH was partly responsible for the reduction in aEPEC numbers. Nevertheless, our data suggest that a significant decrease in aw and the increase in salt concentration at the same time (
Figure 2) acted as microbial hurdles in the cheese and could play an important role as well. It is assumed that the minimum aw required for
E. coli multiplication is from 0.94 to 0.95, whereby it would be a limiting factor for the growth of this pathogen during the cheese ripening. Process control during ripening is equally important as controlling the cheesemaking process, as cheeses exposed to surface contamination when ripening can undergo changes in their physicochemical characteristics that may promote greater survival or even growth of
E. coli [
2]. Moreover, it has been reported that several factors during the manufacturing and cheese ripening, including NaCl concentration, acidity, storage temperature, and ripening point, may cause a synergistic effect on the growth of pathogenic
E. coli [
21,
23].
The lactic acid bacteria (LAB) counts were around 8 log cfu/g during cheesemaking (
Figure 1 and
Figure 2), increasing their numbers in the curd and slightly decreasing throughout the ripening up to
ca. 7 log cfu per gram. A similar trend was observed in the remaining investigated bacterial parameters. During the cheesemaking process of many varieties of cheeses, pH values drop to 5.2 caused by metabolic activity of LAB, being even sharper in some types [
38]. LAB are mainly responsible for pH dropping, which, in accordance with our data, regularly declined until the beginning of ripening (
Table 1). They also produce other substances by their natural activity that can inhibit bacterial growth, playing a key role in the control of pathogenic bacteria in cheese [
39]. Natural microflora of raw milk as well as LAB used as starter cultures during cheesemaking can exert an antagonistic influence against STEC and other bacterial pathogens [
40]. In our study, reduction in diarrhoeagenic
E. coli numbers may be also related to LAB numbers for the reason that large initial LAB populations (up to ~8 log cfu/g) appear to be required to produce adverse effects on
E. coli, particularly belonging to O157:H7 serotype [
21]. However, there seems to be no compelling reason to argue for a strong antagonistic activity, by means of competition or antibiosis. Scientific literature reports both that
E. coli was affected by indigenous microflora and not to play a significant role in its growth or survival [
23].
Figure 2 shows that STEC population in inoculated cheese dropped below initial level (3.14 ± 0.44 log cfu/mL) after 30 days of ripening (−0.18 log (Nt/No) cfu/g). During the same period, the aEPEC strain kept its numbers above 3.49 log cfu/g (initial level of inoculum) and required 45 days of ripening to reduce the initial increase observed before maturation (−0.43 log (Nt/No) cfu/g). Regulation (EC) No 853/2004 of the European Parliament and of the Council of 29 April 2004 [
41] introduced a ripening period of at least 60 days as a safety measure in cheeses manufactured from raw milk not complying with the microbiological criteria, which are established for total bacterial and somatic cells counts. Several studies, focused on pathogenic
E. coli strains at different milk concentrations (10
1–10
3 cfu/mL) for cheese manufacturing, observed a significant reduction during 60 days of ripening, but none reported a complete elimination of this pathogen [
21,
22,
23,
24]. Our findings confirm these observations as, after 60 days of cheese ripening, the studied pathogenic
E. coli populations decreased by 1.32 log cfu/g for STEC and by 0.59 log cfu/g for aEPEC compared to the corresponding bacterial levels inoculated in raw milk (
Figure 2). These reduction ratios indicate that the 60-day aging requirement per se would be both ineffective to control completely aEPEC and uncertain for inactivating STEC. It is also noteworthy that both strains were still detectable after 90 days of ripening (
Figure 1).
As far as we know, this is the first report showing the fate of an atypical EPEC strain during manufacturing of raw milk cheese. Our results point out that aEPEC O25:H2 (MK127C9 strain) adapted better than the STEC strain to the ripening conditions and reached higher numbers of survivals (2.11 ± 0.23 log cfu/g) after 90 days of maturation, showing an average reduction of less than 1.5 log cfu/g throughout the whole cheesemaking process. By contrast, STEC O140:H32 (MK116C9 strain) was more sensitive to ripening conditions but was also detected at the end of cheese maturation, determining a total reduction in counts lower than 2.5 log cycles. These facts suggest that when these pathogens, principally aEPEC, are in the raw milk, they could survive standard cheesemaking and remain viable in cheese beyond 60–90 days, as reported elsewhere for serotype O157:H7 [
22,
42]. The consideration of the serotype, beside the strain and type of cheese, is necessary to assess the bacterial effect of cheesemaking and ripening process in view of the fact that it was reported important differences on growth and persistence among STEC serotypes [
22,
23].
The main mechanisms involved in the survival of bacteria during cheesemaking are the acidic, osmotic, and heat shock stress responses, which can act individually or in combination. One of the key regulators of the general stress response in
E. coli is the activation of the
rpoS gene, which has been reported to differ among strains [
38], thus being a possible explanation of the differences observed in our study. An additional explanation to the lower survival of STEC MK116C9 comparing with aEPEC MK127C9 may be due to the induction of Stx- encoding prophages. The different stresses produced during the cheesemaking and ripening could induce the lytic cycle [
23,
24] and, consequently, foster the reduction of STEC population. DNA of viral origin is a highly frequent element of the bacterial genome, which can be represented for fully functional prophages. These genetic elements can carry genes that influence the virulence of the bacterial host, such as Shiga toxin genes, or their metabolic activities. They can be activated spontaneously, merely as a result of randomness in gene expression or from induction of the host SOS response as it was reviewed by Nanda et al. [
43]. This review set the focus on the triggering of spontaneous activity of prophages and the physiological consequences of this process on microbial populations. When prophage- carrying strains were grown in mixed populations, the spread of viral DNA was set, with a portion of bacteria being killed due to lytic development of prophages and survivors undergoing lysogenic conversion. Thus, the presence of lysogenic bacteriophages may add a potential drawback for survival of the bacteria, which is well used by bacterial competitors. The prophage- encoded Shiga toxin is a major virulence factor in Stx-producing
Escherichia coli. Toxin production and phage production are linked and occur after induction of the RecA-dependent SOS response. This induction could be promoted by food- related stress. Fang et al. [
44] observed that the effect of stressors, such as lactic acid or sodium chloride, reduced bacterial counts by 1 to 2 log cfu/mL in some foods.