Phenotypic and genotypic analysis of bio-serotypes of Yersinia enterocolitica from various sources in Brazil

Introduction: Yersinia enterocolitica is a well-known foodborne pathogen widely distributed in nature with high public health relevance, especially in Europe. Methodology: This study aimed to analyze the pathogenic potential of Y. enterocolitica isolated strains from human, animal, food, and environmental sources and from different regions of Brazil by detecting virulence genes inv, ail, ystA, and virF through polymerase chain reaction (PCR), phenotypic tests, and antimicrobial susceptibility analysis. Pulsed-field gel electrophoresis (PFGE) was used for the assessment of phylogenetic diversity. Results: All virulence genes were detected in 11/60 (18%) strains of serotype O:3, biotype 4 isolated from human and animal sources. Ten human strains (4/O:3) presented three chromosomal virulence genes, and nine strains of biotype 1A presented the inv gene. Six (10%) strains were resistant to sulfamethoxazole-trimethoprim, seven (12%) to tetracycline, and one (2%) to amikacin, all of which are used to treat yersiniosis. AMP-CEF-SXT was the predominant resistance profile. PFGE analysis revealed 36 unique pulsotypes, grouped into nine clusters (A to I) with similarity ≥ 85%, generating a diversity discriminatory index of 0.957. Cluster A comprised all bio-serotype 4/O:3 strains isolated from animal and humans sources. Conclusions: This study shows the existence of strains with the same genotypic profiles, bearing all virulence genes, from human and animal sources, circulating among several Brazilian states. This supports the hypothesis that swine is likely to serve as a main element in Y. enterocolitica transmission to humans in Brazil, and it could become a potential threat to public health as in Europe.


Introduction
Yersinia enterocolitica was of clinical and epidemiological interest in the 1970s, evidenced by the significant increase of studies published around the world [1].It is widely distributed in nature in aquatic and animal reservoirs, with swine serving as a major reservoir for human pathogenic strains [2].Yersiniosis is a zoonotic bacterial disease with high public health relevance, especially in Europe due to its high levels of occurrence, where it is the third most common bacterial enteric disease [3][4].
In Brazil, several studies on Y. enterocolitica and other Yersinia species show its isolation from human, animal, food, and environmental sources; occurrences are most frequently reported in sporadic cases [5].Brazil does not have official data about the incidence of this pathogen.The sporadic cases have been reported in only a few studies that have been published in scientific journals around the world.
Regarding clinical aspects, Y. enterocolitica O:3 may cause a variety of gastrointestinal problems, such as acute diarrhea, terminal ileitis, and mesenteric lymphadenitis.The pathogenicity and virulence mechanisms are still complex, and various chromosomal and extra-chromosomal factors have already been described [2].The ail, inv, and yst genes, located in the chromosome of pathogenic Y. enterocolitica strains, are the most frequently used chromosomal targets [6].The inv (invasive gene) and ail (attachment invasion locus) genes are responsible for the production of an invasin and an outer membrane protein called Ail, respectively, which enable the bacteria translocation through the intestinal epithelium, cell fixation, and subsequent invasion.The role of the ystA gene (yersinia stable toxin) in the production of thermostable enterotoxin, causing diarrhea, has also been emphasized [7].
Another virulence factor in Y. enterocolitica O:3 is a plasmid called pYV (plasmid Yersinia virulence) [8].It allows the microorganism to survive and multiply in lymphatic tissues and encodes the production of several proteins called Yops (Yersinia outer proteins).They play a major role in yersinial virulence, whose effects on macrophages and polymorphonuclear leukocytes include the inhibition of phagocytosis and respiratory burst [2].In plasmids, the locus virF contains information on Yops transcription [9].
In the last decades, several different DNA-based methods for epidemiological typing have been used in epidemiological outbreak studies of Y. enterocolitica [5].PFGE is the gold standard for bacteria and is used both for studies of hospital outbreaks and for the comparison of bacterial populations involving microorganisms of different countries or regions [10].
The aim of this study was to analyze the virulence genes, antimicrobial susceptibility behavior, and genotypes of Y. enterocolitica Brazilian strains isolated from humans and animals, which could act as a reservoir and a transmission route.

Bacterial strains and bacteriological analysis
A total of 60 Y. enterocolitica strains were randomly selected.The isolate sources included 31 strains from tonsillar and tongue surfaces of healthy swine; 10 from slaughterhouses in Campinas city isolated between 2007 and 2008; 5 strains from food isolated in São Paulo state (SP) (2008-2011); and 14 isolates from humans with various clinical circumstances coming from Bahia (BA), Minas Gerais (MG), Paraná (PR), Rio de Janeiro (RJ), and Santa Catarina states (SC) between 2005 and 2011.
The strains belonged to the Listeria collection (CLIST), Laboratory of Bacterial Zoonosis at Oswaldo Cruz Institute, Fiocruz, Rio de Janeiro.

Phenotypic tests for detection of virulent plasmid
To assess the presence of the virulent plasmid, the following phenotypic tests were performed: autoagglutination phenotypic tests at 37°C [15], binding of crystal violet [16], absorption of Congo red in the Congo red-magnesium oxalate agar medium (CRMOX) [17], and the activity of pyrazinamidase [18].All phenotypic tests, except for the pyrazinamidase activity test, were performed at 28°C and 37°C.

Detection of virulence genes
The presence of inv, ail, virF, and ystA genes was analyzed through polymerase chain reaction (PCR) with their respective primers (Table 1) using Platinum PCR Super Mix Kit (Invitrogen, Carlsbad, USA).The conditions for performing PCR reactions were followed according to the reference of each gene in Table 1.DNA extraction was performed using a DNA extraction kit (DNeasy Blood & Tissue Kit; Qiagen, Hilden, Germany), following the manufacturer's guidelines.
The plugs were digested with 10U/µL Not I (BioLabs, Ipswich, USA) and the electrophoresis system was CHEF-DR III (Bio-Rad, Hercules, USA).The gel was stained with ethidium bromide solution (0.5 μg/mL) and visualized by UV light.The molecular weight standard used was Salmonella serovar Braenderup H9812.
Data analysis was performed with BioNumerics version 4 (Applied Math, Austin, USA).Banding patterns were compared through the Dice coefficient by using the UPGMA method (unweighted pair group method with averages) to determine the similarity of bands and by adopting Tenover et al.'s [24] criterion in the definition of cluster types.A cut-off ≥ 85% was set forth in formation of groups considered genetically related.However, isolated strains with the 100% band patterns of similarity were considered to be representatives of the same pulsotype of PFGE [25].For the assessment of the PFGE discriminatory capacity, the calculation of the Simpson's index of diversity was used [26].

Results
The data obtained in phenotypic tests, in antimicrobial resistance behavior (AR), and in PCR of virulence genes are listed in Table 2. Eleven positive strains in the phenotypic tests, except for pyrazinamidase activity, presented all virulence genes, including the plasmid gene virF, confirming the presence of the plasmid in nine animal and two human isolate strains, since the strain was only considered as the plasmid bearer when phenotypic tests and the PCR test for detection of virF gene were positive.The strains belonging to biotype 1A were positive for pyrazinamidase activity.Ten strains isolated from human sources (O:3/4) presented only three chromosomal virulence genes and nine strains belonging to biotype 1A presented the inv gene.
The PFGE analysis provided elements for the preparation of a dendrogram, consisting of nine clusters (A to I), allowing the grouping of the isolated strains with ≥ 85% similarity.Within the clusters, 36 isolated strains had pulsotype with 100% similarity, as shown in Figure 1.Cluster A comprised all serotype strains O:3 biotype 4 isolated from animal and human sources.The strains belonging to biotype 1A from food and environmental sources were spread throughout the others clusters (B to I).

Discussion
Phenotypic tests for detection of the plasmid were positive when Y. enterocolitica O:3 strains were incubated at 37°C.This is due to the plasmid characteristic related to this temperature [8].In addition to the virulence plasmid, 11 pathogenic Y. enterocolitica O:3 strains had chromosomally encoded virulence genes (Table 2).The obtained results demonstrated the correlation of biotype and serotype (O:3 biotype 4) with pathogenic potential to susceptible hosts [27].
Ten human isolated strains showed only chromosomal virulence genes (Table 2).One of the hypotheses for this is the loss of plasmid, which can be easily lost at temperatures above 30°C, depending on the culture conditions [28].However, it is not known how the plasmid may be lost below 30°C; further research is required [8].Nine biotype 1A strains, both from food and environmental sources, and one from an animal source (Table 2), showed the inv gene.The strains of this biotype have generally been regarded as avirulent.The genes present in such strains match findings in Brazil [29] and worldwide [30], demonstrating the pathogenic potential of the representatives of this biotype.
Antimicrobial susceptibility test revealed the high resistance of Y. enterocolitica to ampicillin (100%) and cephalotin (97%) (Table 3), which has been commonly described in the literature [31].This resistance is probably due to the presence of chromosomal genes blaA and blaB responsible for producing two β-lactamases, A and B [31].Also, six (10%) strains were resistant to sulfamethoxazoletrimethoprim and seven (12%) to tetracycline, which are two antibiotics used in yersiniosis treatment [7,31].One strain showed resistance to amikacin (AMP-CEF-AMK-SFT profile), an aminoglycoside, used in combination with other antibiotics for the treatment of extra-intestinal yersiniosis [2,7].Such resistance has been described in strains isolated from animals in Brazil [32].
AMP-CEF-SFT was the resistance profile most commonly found in 26 strains of serotype O:3 and not serologically typed and no pattern could be established as to the serotype/biotype or origin.The strain YE42 (profile: AMP, CEF, FOX, SFT, TET) showed resistance to three different classes of antibiotics (cephalosporin, sulfonamide, and tetracycline), except for ampicillin and cephalotin (Table 3).Resistance to multiple antimicrobial agents was a rare event [33].
Low resistance to antibiotics was observed in isolated strains showing all virulence factors (Table 2).
Plasmid pYV plays no role in terms of participation in the antimicrobial resistance profile [34].
PFGE analysis revealed nine clusters (A to I) grouping isolated strains with ≥ 85% similarity (Figure 1).In these clusters, 36 pulsotypes were grouped, generating a discriminatory index of diversity of 0.957.
Cluster A comprised 21 pulsotypes (A1 to A21) with 87% similarity among isolated strains, all from bio-serotype 4/O:3 of animal and human sources.The high levels of similarity between the profiles match the results in the literature.They strongly support the hypothesis that pigs play an important role in the epidemiology of human sporadic Y. enterocolitica 4/O:3 infections [35].
The swine isolated strains belong to the same region and the same period of isolation, presenting a great similarity between them, featuring the epidemiological situation as a restrict niche.It is worth mentioning that A11 comprised two strains isolated from humans, one from Santa Catarina and another from Rio de Janeiro in 2008 and 2005, respectively.Based on this result, this pulsotype might circulate in Brazil, even in geographically distinct regions that are swine farming complexes.Additionally, pulsotypes A13 and A15 grouped strains isolated from human and animal sources, supporting the hypothesis.
It should be noted that swine farming products (meat, sausages, etc.) are typically marketed to various locations, whether or not geographically distant, promoting the circulation of Y. enterocolitica bioserotype 4/O:3, with high phenotypic and genotypic similarity in various regions.
Based on this, two hypotheses may be proposed in order to explain the similarity of about 87% within cluster A. The first one is that these strains may be genetically related (same genetic lineage), and this difference may be due to some genetic events (point mutations) such as deletion or insertion of DNA or loss or gain of restriction sites [24].The second hypothesis relates to the great natural similarity found between strains isolated from humans and from swine.Studies conducted in Finland and Germany showed that strains isolated from humans were almost indistinguishable from strains isolated from swine in those regions [35].
The biotype 1A strains isolated from food and the environment had greater diversity between fingerprints of their bands and were spread to the other clusters (B to I).This diversity may be related to different serotypes from this biotype; some of those strains could not be serologically typed [32,36].
Upon PFGE analysis of the strains with all virulence factors, it was observed that some of the samples were grouped in the same pulsotypes, such as A9, comprising YE14 and 15; A11, comprising strains YE 21, 23, 24, and 26; and A17, comprising the samples YE57 and 58.Based on these results, PFGE may be able to group strains bearing virulence genes.However, additional studies should be performed and further developed, provided that these samples are grouped within the same pulsotypes, rather than grouped according to their origins or their isolation source or serotype/biotype.
No direct correlation could be established between the pulsotypes of Y. enterocolitica and antimicrobial resistance patterns, as was observed in previous studies abroad [34].

Conclusions
Based on the results, Y. enterocolitica strains with bio-serotype 4/O:3, isolated from human and animal sources, showing the same genotypic profiles and bearing all virulence factors, circulate among several Brazilian states.These strains could become a potential threat to public health in Brazil as they have in Europe.

Figure 1 .
Figure 1.PFGE dendogram.Dendrogram built from the analysis of the bands, generating nine clusters (A to I) with similarity ≥ 85%, of 60 Y. enterocolitica strains obtained by PFGE.

Table 1 .
Primers of virulence genes inv, ail, ystA and virF and their respective amplicon

Table 2 .
Results of antimicrobial resistance (AR), PCR of virulence genes and phenotypic tests of 60 Y. enterocolitica strains analyzed (continue on next page)

Table 3 .
Resistance profiles of 60 isolated strains of Y. enterocolitica