Abstract
Enterococcus faecalis is not only a prevalent species among dairy microbial community but also a well-documented opportunistic pathogen. Food safety should exclude the possibility of consumer exposure to its virulence traits through consumption of dairy products. In this study, an integrated approach based on both phenotypic and genotypic methods was applied to investigate the incidence of antibiotic resistance and pathogenicity potential in 40 E. faecalis isolated from 10 Italian raw milk cheeses over a 13-year period (1997–2009). Among the 14 tested antibiotics, resistance to tetracycline, rifampicin, chloramphenicol, and erythromycin was observed, whereas vancomycin-resistant enterococci were not found. A high incidence (90% of strains) of the tet(M) gene emerged, whereas tet(K), tet(S), tet(L), int, and ermB genes were occasionally amplified (12.5%, 10%, 7.5%, 2.5% and 30%, respectively). No strain was positive for vancomycin-resistant determinants. Among the seven virulence determinants considered, the asa1, gelE, esp, and efaA genes were harbored. No other gene encoding for either different virulence factors (cylA, hyl, and ace) or amino acid decarboxylase activity (hdc, tdc, and odc) was detected. Consequently, E. faecalis isolated from raw milk cheeses does not represent a substantial reservoir of antimicrobial resistance and virulence factors if compared with clinical strains. However, this species occasionally harbors detrimental traits; thus, the possibility that it could be a route for transmission of pathogenic genes through dairy products should never be disregarded.
Introduction
E
Anyway, besides dominating among fermentation microbiota such as in dairy products (Castro et al., 2016; Mormile et al., 2016; Kirmaci, 2017), E. faecalis is the most frequently incriminated enterococcal species in nosocomial infections (O'Driscoll and Crank, 2015). In the gut, this species has proved to adopt a variably virulent phenotype under harsh environments (e.g., lack of nutrients), which can negatively modulate the intestinal microbiota, inducing virulence activation (Krezalek and Alverdy, 2018). Enterococci are not considered “Generally Recognized As Safe” (Araújo and Ferreira, 2013) and not recommended for the Qualified Presumption of Safety list proposed by the European Food Safety Authority (EFSA BIOHAZ Panel, 2016), even if no cases of infection due to the consumption of a food product containing enterococci have been registered so far (Chajęcka-Wierzchowska et al., 2017).
They are also among the most antibiotic-resistant bacteria, and multiresistance (resistance to at least three different classes of antibiotics) contributes to their emergence as opportunistic pathogens (Santagati et al., 2012). Strains showing novel resistance mechanisms are consistently isolated (Lebreton et al., 2014). What is more, in light of the presence at high concentration in animal feces, E. faecalis is an ideal indicator of antibiotic resistance (AR) in Gram-positive microorganisms (EFSA et al., 2018). AR displayed by foodborne E. faecalis represents a public health hazard since exposure through food could lead to acquisition by humans of AR bacteria or bacteria-borne AR genes (EFSA et al., 2008). Besides, the prevalence of AR in commensal enterococci provides information on the selective pressure exerted by the antimicrobial agents on the intestinal bacterial community in food-producing animals (EFSA, 2018). AR is really one of the biggest global emergences (Abat et al., 2018), and E. faecalis is one of the most important active players because of its intrinsic or acquired resistance plus dissemination of resistance determinants within and beyond the genus.
The present work aimed at discovering through phenotypic and genotypic methods the AR, virulence markers, and genes encoding for amino acid decarboxylase activity in E. faecalis strains isolated through the technological process of Italian raw milk cheeses.
Materials and Methods
Bacterial strains
Forty E. faecalis strains belonging to the Institute of Sciences of Food Production - Italian National Research Council (CNR - ISPA) collection were considered. Strains had been isolated over a 13-year period (1997–2009) from different samples (milk, curd, and cheese) of 10 Italian goat and cow raw milk cheeses. They had been previously identified by species-specific PCR assay, and partial 16S rRNA sequencing and their genetic relatedness were assessed by two DNA fingerprinting techniques: random amplified polymorphic DNA (RAPD)-PCR and repetitive extragenic palindromic (rep)-PCR (Silvetti et al., 2014). They were propagated aerobically at 37°C using M17 broth (Scharlau Microbiology, Barcelona, Spain).
Antibiotic susceptibility testing
Sensitivity to 14 antibiotics representative of the major classes of critically important antibiotics (Table 1) was tested by the disk diffusion method. Saline suspensions with concentration of ∼106 cells/mL were produced from overnight cultured enterococci by measuring optical density at 625 nm (Infinite F200 PRO; Tecan, Männedorf, Switzerland). The inoculum concentration was confirmed by plating on M17 agar (Scharlau Microbiology) plates, incubating at 37°C for 72 h. One hundred microliters of these suspensions were inoculated in Mueller–Hinton agar (Biolife Italiana, Milan, Italy) plates, and the antibiotic disks provided by Oxoid Ltd. (Basingstoke, United Kingdom) were placed onto them. After 18 h (24 h for vancomycin) of incubation at 37°C, each strain was classified as sensitive, intermediate, or resistant based on the inhibition zone diameters in accordance with the Clinical and Laboratory Standards Institute guidelines (Clinical and Laboratory Standards Institute, 2017).
Patterns of Antibiotic Resistance and Their Determinants of the Enterococcus faecalis Strains
In table, the active antibiotics are not reported (ampicillin 10 μg, levofloxacin 5 μg, mupirocin 200 μg, nitrofurantoin 300 μg, penicillin G 10 U).
I, intermediate resistance; R, resistant; C, chloramphenicol 30 μg; CIP, ciprofloxacin 5 μg; E, erythromycin 15 μg; OX, oxacillin 1 μg; QD, quinupristin–dalfopristin 15 μg; RD, rifampicin 30 μg; S, streptomycin 10 μg; TE, tetracycline 30 μg; VA, vancomycin 30 μg.
DNA extraction
Total DNA was extracted from the overnight cultures by the Microlysis kit (Aurogene s.r.l., Rome, Italy) following the manufacturer's instructions.
Detection of tetracycline, vancomycin, and erythromycin genes
The presence of five genes most commonly involved in resistance to tetracycline [tet(K), tet(L), tet(M), tet(O), and tet(S)] was investigated by PCR according to Ng et al. (2001). As positive control, Enterococcus faecalis VC124 [tet(M) and int] and Streptococcus thermophilus VC59 [tet(L) and tet(S)] from the CNR ISPA collection were used (Morandi et al., 2015). The presence of the integrase gene (int), connected with the Tn916-Tn1545 transposon family of tet(M)-carrying bacteria, was explored (Choi and Woo, 2015). The strains were also tested for the presence of genes responsible for resistance to vancomycin (Lemcke and Bülteb, 2000; Depardieu et al. 2004) and erythromycin (Comunian et al. 2010).
Detection of virulence and amino acid decarboxylase genes
Genomic DNA was used to detect the incidence of genes encoding for the following virulence factors and amino acid decarboxylating enzymes: cytolysin activator (cylA), aggregation substance (asa1), gelatinase (gelE), hyaluronidase (hyl), enterococcal surface protein (esp), adhesin of collagen protein (ace), endocarditis antigen (efaA), histidine decarboxylase (hdc), tyrosine decarboxylase (tdc), and ornithine decarboxylase (odc). PCR was performed as previously described by Martín-Platero et al. (2009), except for the gene efaA, whose amplification was conducted with the annealing temperature used by Creti et al. (2004).
Hemolytic activity
Hemolysin production was assessed by streaking fresh cultured enterococcal strains onto the surface of brain heart infusion agar (Scharlau Microbiology) supplemented with 5% (v/v) defibrinated sheep blood (Oxoid Ltd.) and incubating at 37°C under both aerobic and anaerobic conditions. After 48 h, the presence or absence of clear zones around bacterial growth was recorded as reported by Al Atya et al. (2015).
Statistical analysis
Cluster analysis was performed considering 16 variables, including AR phenotypes (chloramphenicol, erythromycin, quinupristin–dalfopristin, rifampicin, streptomycin, and tetracycline), AR genes [tet(K), tet(L), tet(M), tet(S), and int], and virulence factors (cylA, asa1, gelE, esp, and efaA) of 40 strains. Binary 0/1 matrix was created based on the absence or presence of phenotypic and genotypic parameters (in column) of the enterococcal strains (in rows). Clustering of strains was performed using average linkage clustering with the BioNumerics 5.0 software package (Applied Maths, Kortrijk, Belgium) employing the Jaccard coefficient. RAPD banding patterns were analyzed with BioNumerics version 5.0 and clustered by unweighted-pair group method with arithmetic mean analysis (Granato et al., 2018).
Results
Antibiotic susceptibility
Table 1 lists the AR profiles of the selected antibiotics for the E. faecalis strains. All strains were resistant to oxacillin, whereas susceptibility was detected to ampicillin, levofloxacin, mupirocin, nitrofurantoin, and penicillin G in all tested strains. Vancomycin-resistant enterococci were not found. Twenty strains were able to grow in the presence of streptomycin, and low incidence of resistance to quinupristin–dalfopristin and tetracycline was observed (n = 14, 35% strains and n = 11, 27.5% strains, respectively). Rifampicin resistance was found in three strains, whereas resistance to chloramphenicol was detected in lower percentages (n = 2, 5% strains). Only one strain was erythromycin-resistant. Intermediate resistance was detected in some cases, above all for ciprofloxacin, erythromycin, and streptomycin. Although multiresistance phenotypes were frequently identified, the most common phenotype was single resistance toward oxacillin, which was harbored by 13 strains. Twelve, 9, and 4 strains were resistant to 2, 3, and 4 antibiotics, respectively (Table 1). One strain possessed resistance to five antibiotics (chloramphenicol, oxacillin, quinupristin–dalfopristin, streptomycin, and tetracycline), whereas another one strain possessed resistance to six antibiotics (chloramphenicol, erythromycin, oxacillin, rifampicin, streptomycin, and tetracycline). The different patterns were uniformly distributed within the 40 strains of different dairy and geographical origins, and no correlation with origin was revealed (Table 2). Examining the relationship between the isolation year and the AR profile, no noticeable difference appeared between the oldest and the most recent strains (Table 1). Molecular analysis of tetracycline- and erythromycin-resistant genes was conducted. All the strains possessing a phenotypic tetracycline resistance contained at least one tet gene (Table 1). A high incidence of the tet(M) gene was found (36 of the 40 strains), whereas 9 strains carried the tet(K) or tet(S) genes. The presence of the tet(L) gene was revealed in three strains, whereas the tet(O) gene was never identified. One strain harbored the int gene. Strains possessing multiple tetracycline-resistant genes were detected. Twelve strains, including the only one phenotypically resistant to erythromycin, were positive for the amplification of the ermB gene. On the contrary, no vancomycin-resistant genes were amplified.
Hemolysis and Virulence Determinants of the Enterococcal Strains
−, Hemolysis or gene absence; +, hemolysis or gene presence.
Virulence factors
A very low incidence of hemolytic activity was revealed (Table 2). Only 2 strains were β-hemolytic, and the other 38 strains presented γ- or α-hemolysin production. The γ-hemolytic was the most detected phenotype, occurring in 28 strains. The β-hemolytic activity was more pronounced when incubation was conducted under anaerobic conditions. The occurrence of hemolytic phenotypes did not present any particular correlation with the strains origin (Table 2). PCR amplification showed positive results for four of seven virulence markers: gelE, asa1, esp, and efaA (Table 2). None of the enterococcal strains carried the cylA gene, which is required for the expression of cytolysin activator and strictly connected with hemolysis. Moreover, no amplification occurred for either hyl or ace genes. With regard to their distribution within E. faecalis strains, the asa1 gene was the most common factor, found in 31 strains, whereas the gelE, efaA, and esp genes were present in 29, 23, and 9 strains, respectively. All strains carried at least one of the seven virulence determinants (Table 2), and multiple virulence genes were frequently revealed. In addition, none of the enterococcal strains harbored amino acid decarboxylase genes for the production of histamine, tyramine, and putrescine.
Discussion
Besides being naturally present in many artisanal cheeses produced in Southern Europe, where it even reaches up to 108 colony-forming units per gram of cheese, E. faecalis is a well-documented opportunistic pathogen owing to unique virulence traits, including resistance to routinely prescribed antibiotics.
These traits can be shared as genetic elements through conjugation, mainly as pheromone-responsive conjugation system. The transmission of genetic information can occur in the gastrointestinal tract (Choi and Woo, 2015; Fuka et al., 2017). Enterococcal intrinsic resistance against cephalosporins, clindamycin, trimethoprim–sulfamethoxazole, fluoroquinolones, low concentrations of aminoglycosides, and β-lactams has been reported. In addition, strains can develop resistance to various antimicrobials (tetracycline, chloramphenicol, rifampin, glycopeptides, quinolones, macrolides–lincosamides–streptogramins, high levels of aminoglycosides, and β-lactams) by mutation or acquisition of genetic material carried by transposons or plasmids (EFSA, 2008; Frazzon et al., 2010; Depardieu and Courvalin, 2017). In this study, we deliberately selected E. faecalis strains isolated in a period (late 1990s and 2000s) when higher rates of antibiotics were used. Thus, the risk assessment would cover the potential AR under the worst foreseeable conditions. Since the use of antibiotics in animal husbandry has reduced over the past years, the safety of our strains corroborates the safety of strains of most recent origin. All the examined strains were sensitive to antibiotics of clinical importance, such as penicillins (ampicillin and penicillin G) and vancomycin. This is a positive result, as these antimicrobials are often the preferred treatment of multiple-resistant infections, and what is more, the occurrence of vancomycin resistance is reported to be steadily increasing worldwide within E. faecalis species (Freitas et al., 2013). Previous researches highlighted that the occurrence of AR among dairy E. faecalis strains varies in relation to origin. E. faecalis isolated from Tolminc cheese showed a very low AR (Čanžek Majhenič et al., 2005). On the contrary, a widespread resistance to chloramphenicol, erythromycin, and tetracycline was reported by Templer and Baumgartner (2007) in Appenzeller and Schabziger cheeses. Jamet et al. (2012) revealed that AR is common in enterococci isolated from French cheese, with tetracycline, erythromycin, and chloramphenicol as the most prevalent ARs detected, emphasizing the origin of tetracycline- and erythromycin-resistant genes from the environment and farms. Reduced antibiotic utilization at the farm level may be reached through the implementation of the Hazard Analysis and Critical Control Point procedures that, together with the application of good hygienic and manufacturing practices and standard sanitation operating procedures, are the basic conditions for an effective food safety management system in the dairy industry (Cusato et al., 2013). The absence of vancomycin-resistant strains recorded in this study is in agreement with the findings of other researchers, who reported low or no incidence of vancomycin resistance in European cheeses, suggesting that this acquired resistance is still restricted to the clinical setting (Morandi et al., 2006; Jamet et al., 2012). Among β-lactams, only oxacillin was inactive against all the strains. Although natural intrinsic resistance to semisynthetic penicillins (i.e., oxacillin) is reported (Klare et al., 2003), Gajan et al. (2013) found susceptibility to oxacillin among E. faecalis strains (7%) isolated from hospitalized patients. Also, ciprofloxacin, levofloxacin, and nitrofurantoin were active agents. These results are encouraging since ciprofloxacin belongs to fluoroquinolones, which usually exhibit a weak activity against enterococci, whereas nitrofurantoin is commonly used in treatment of urinary infections caused by vancomycin-resistant enterococci. Ineffectiveness of the streptogramin combination quinupristin–dalfopristin against E. faecalis was previously described (Fair and Tor, 2014), but this antibiotic inhibited most strains used in this trial. In addition, it is worth mentioning that all the strains showed a mupirocin inhibition halo superior to 18 mm, thus demonstrating for the first time the susceptibility to this antimicrobial in strains of food origin. The susceptibility of E. faecalis species to this antibiotic is of relevant medical interest, due to its topical use in the hospital setting against methicillin-susceptible and methicillin-resistant Staphylococcus aureus (Ristagno et al., 2018). Mupirocin-resistant E. faecalis strains could represent a health threat for the possible transfer of resistance to this pathogen. Tetracycline resistance is often discovered in multidrug-resistant microorganisms and involves different mechanisms, such as efflux pumps proteins [tet(K) and tet(L)] or ribosomal protection proteins [tet(M), tet(O), and tet(S)]. Congruence between tetracycline phenotypic and molecular results was noted, and the tet(M) gene was found in all 11 strains positive to phenotypic resistance. Twenty-five of the 36 strains carrying the tet(M) gene were unable to express resistance phenotypically, suggesting the existence of a silent gene. It could be that the gene expression depends on environmental and cultural conditions as well. This lack of correlation between resistance phenotype and genotype highlights the need of both the approaches to investigate the safety of enterococcal strains. Resistance mediated by the tet(M) gene affording ribosomal protection is often the predominant one, probably not only due to its dissemination by the Tn916-type transposons but also due to its presence in conjugative plasmids and on the chromosome (Garrido et al., 2014). Recently, evidence of the mutual plasmid-mediated transferability of the tet(M) gene between a foodborne Listeria monocytogenes and E. faecalis was demonstrated both in vitro and in a food matrix (Haubert et al., 2018). The present study also provides the demonstration of a remarkable presence (10%) of the tet(S) gene in E. faecalis of food origin in comparison to other authors, who found a lower incidence (1–3%) (Huys et al., 2004; Wilcks et al., 2005). The frequent correlation between the intermediate resistance to erythromycin and the presence of the ermB gene could represent an evidence of the development of the resistance mechanism to this macrolide antibiotic in E. faecalis, probably as a consequence of prolonged exposure to subinhibitory antibiotic levels (Cantón et al., 2013). However, it is important to remember that in vitro AR does not always reflect the in vivo situation (Barbosa et al., 2009).
Pathogenesis derives from a combination of several factors including aggression ability of invading microorganism. Therefore, the presence of virulence factors allowing an infecting strain to colonize and invade host tissue has been carefully investigated (Barbosa et al., 2010). Hemolysin has been recognized to contribute to the severity of enterococcal infections. However, the absence of hemolytic activity in enterococci does not necessarily exclude their virulence (Psoni et al., 2006). This is corroborated by the present study, where genetic elements for virulent traits were commonly detected in non β-hemolytic strains. Due to their easy transferability by means of conjugative plasmids, the presence of genetic determinants encoding for β-hemolysis is considered undesirable in food strains, especially those used as starters in food fermentations (Tuncer et al., 2013). A very low incidence of β-hemolytic production on sheep blood agar plates was observed among E. faecalis strains, thus confirming results obtained in previous works (Barbosa et al., 2010). On the contrary, no equivalence in results was found with other authors' work, which used human or horse red blood cells and reported remarkable presence of β-hemolysis (Creti et al., 2004; Poeta et al., 2006). This divergence in results probably depends on the major resistance of sheep erythrocytes to hemolysin-mediated lysis, as already stated by Semedo et al. (2003). In addition, the positive strains displayed a very weak β-hemolytic activity under aerobiosis, whereas it became more pronounced when incubation was conducted under anaerobic conditions, as also detected by Abriouel et al. (2008). Actually, Day et al. (2003) demonstrated the oxygen regulation of hemolysis. As β-hemolytic phenotype represents the expression of cytolysin operon, a strict correlation exists between hemolysin and cytolysin, a bacterial toxin that exhibits bactericidal properties, causing the rupture of membranes, like those of bacterial cells, erythrocytes, and other mammalian cells (De Vuyst et al., 2003). In the present study, β-hemolytic enterococcal strains did not present the cytolysin activator gene cylA. Consequently, the hemolytic activity of these strains should be due to another cytotoxic component (De Vuyst et al., 2003) or the hemolysin genes should reside on the chromosome instead of pheromone-responsive plasmids (Franz et al., 2001). The detection of virulence determinants was lower when compared with strains of different origins (i.e., clinical). Conversely, enterococci isolated from dairy samples along the production process of traditional raw milk cheeses in Southern Italy expressed a similar virulence gene profile as enterococcal strains involved in human infections (Gaglio et al., 2016).
The safety analysis of dairy E. faecalis of Italian origin was completely assessed through the evaluation of potential biogenic amines production. All strains were negative to the main amino acid decarboxylase genes. A discrepancy with other results was observed since a significant ability to produce tyramine has been frequently conferred to enterococci and mostly to E. faecalis isolated from cheese (Burdychova and Komprda, 2007; Martín-Platero et al., 2009), suggesting their undesirable occurrence in cheese during ripening. In fact, the decarboxylase activity on tyrosine represents a particularly negative trait of dairy-borne enterococci, causing tyramine accumulation in cheese, and possible toxicological consequences on human health, such as headache or hypersensitive reactions, due to consumption of high amounts (>100–800 mg/kg of food) of this amine (De Las Rivas et al., 2005; Psoni et al., 2006). Formation of histamine and putrescine from histidine and ornithine among E. faecalis strains has been previously reported as uncommon by other authors (Martín-Platero et al., 2009; Ladero et al., 2012).
Finally, it is interesting to examine these results considering the biodiversity within the E. faecalis strains. In a previous work (Silvetti et al., 2014), the same strains were subjected to RAPD-PCR and rep-PCR to assess exhaustively their genetic diversity. In the present study, strains that were clustered together with a high similarity level (i.e., ID143A and ID160 or AA8 and VS368) showed different patterns of AR and virulence determinants (Fig. 1). Vice versa, strains grouped together according to their virulence traits were genetically distinct, except for a few strains (e.g., AA8 and AA11 or VS368 and VS382B). Given the virulence profiles, a high degree of heterogeneity was found. The cluster analysis of AR and virulence phenotypes and genotypes emphasized that strains genetically very similar were different from a pathogenic point of view, corroborating the hypothesis that environmental conditions select the phenotypic and genotypic bacterial profile.
Unweighted pair-group method with arithmetic mean-based dendrogram derived from the combined random amplified polymorphic DNA-PCR profiles
Conclusions
This study allows to gain better knowledge of the safety aspects of E. faecalis isolated from raw milk cheeses. Previous researches highlighted that the AR and virulence profiles of enterococcal strains from food of animal origin are very similar to what has been described for enterococci isolated from nosocomial infections. On the contrary, the occurrence of virulence determinants and AR noted in this work is of little importance in contrast to those shown by other strains from different sources (i.e., clinical). Our results, therefore, suggest that dairy E. faecalis does not represent a major potential reservoir for the spread of these detrimental traits. However, strains of dairy origin occasionally harbor detrimental traits as revealed by the high phenotypic and genotypic heterogeneity detected in this research; thus, the possibility that this species could be a route for the transmission of pathogenic genes through dairy products should never be disregarded. Moreover, the evaluation of enterococci safety through in vitro expression of virulence traits does not always reflect the real hazard connected with these microorganisms because of the presence of silent genes, which could potentially be activated in relation to particular environmental conditions, thus transforming these bacteria into pathogens or enhancing their pathogenicity.
Footnotes
Disclosure Statement
No competing financial interests exist.