Biofilm formation capacity and presence of virulence factors among commensal Enterococcus spp. from wild birds

Enterococci are opportunistic pathogens that can form biofilms during infections and many virulence determinants are involved in this process. Although the virulence factors are often analysed in Enterococcus spp. from humans and food animals, little is known about gut enterococcal isolates from wild birds. Therefore, the determination of virulence factors among enterococci isolated from wild birds may provide new information about a possible source of infection for humans and animals or vice versa via the environment. We analysed different phenotypic and genotypic traits in enterococci from wild birds related to potential virulence in humans and animals and to evaluate biofilm formation and its relationship to virulence genes. The E. faecalis isolates were characterised by greater frequency of biofilm formation in BHI than E. faecium. There was a correlation between hydrophobicity and biofilm formation in BHI broth in E. faecalis. None of the isolates was haemolytic. The presence of some adhesion and gelatinase genes was detected in biofilm-positive isolates. The enterococcal pathogenic factors (esp, hyl, and cyl operon genes) did not seem to be necessary or sufficient for production of biofilm by analysed bacteria. Enterococcus species isolated from wild birds should be considered as a possible source of some virulence determinants.

gelatinase (encoded by gelE), serine protease (encoded by sprE), and hyaluronidase (encoded by hyl) are the most frequent virulence factors affecting host cells by their toxic or destructive effects 3 .
Enterococci are Gram-positive elements of the gut microbiota and significant agents of opportunistic infections in animals and humans 4,5 . Enterococcus spp. from avian species associated with diseases include mainly Enterococcus faecalis, E. faecium, E. durans, E. hirae and E. cecorum [6][7][8] . The collection of Enterococcus isolates from wild birds analysed in this study, especially E. faecalis and E. faecium, are resistant to multiple antibiotics and represent different epidemic clones (e.g. CC17, CC81, CC116) 9,10 . The same clones have previously been found in hospitalised patients in Poland and other European countries, as well as globally in unhospitalised patients, livestock, pets, food and wastewater [11][12][13] . For bacteria to be pathogenic, in addition to antibiotic resistance they must also possess virulence factors 14 . In addition, enterococci are naturally able to acquire, accumulate, and transmit extrachromosomal elements encoding virulence traits 15 . Thus, wild birds may be considered a source of enterococci that are potentially pathogenic as well as resistant to antibiotics. Transmission of such enterococcal strains from wild birds to humans and animals is possible via environment, such as: water polluted with faeces, soil, dietary patterns in wild animals/birds, meat or direct contact during hunting in humans 1 . Taking into account migration of birds, the same etiological agent can occur in distant places.
Monitoring faecal Enterococcus spp. as indicator bacteria in different populations, including wild birds makes it feasible to compare the prevalence of virulence in determined STs/CC and to detect the transfer of potentially pathogenic strains from animals to humans and vice versa. However, a few reports of the prevalence of virulence factors in antibiotic resistant enterococci from wild animals have been published [16][17][18][19] .
The objective of this work is to analyse different phenotypic and genotypic traits in enterococcal isolates from the gut microbiota of different wild birds related to potential virulence in humans and animals and to evaluate in vitro biofilm formation and its relationship to presence of virulence genes.

Results
Haemolytic and gelatinase activities. None of the isolates was haemolytic on 5% horse blood agar.
Hydrophobicity and biofilm formation. The in vitro hydrophobicity test revealed twenty strains with hydrophobicity higher than 50%; eighteen of them represented E. faecalis and the other two strains belonged to the species E. faecium and E. casseliflavus. The E. faecalis strains showed significantly higher hydrophobicity than E. faecium (P < 0.001; Chi-square test), i.e. as many as 33.3% of the E. faecalis isolates displayed hydrophobicity at the level of 100% and another 33.3% exhibited %H in the range of 50-70%. Detailed data are shown in Table S1.
Biofilm formation by the tested E. faecalis and E. faecium detected with the microtitre plate method in BHI broth supplemented with 2% glucose was statistically significantly higher than in TSB broth supplemented with 1% glucose (P < 0.001; Chi-square test). Both species differed significantly in biofilm formation in TSB broth supplemented with 1% glucose (P = 0.00886; Chi-square test). However, the E. faecalis and E. faecium species did not differ in biofilm formation in BHI broth supplemented with 2% glucose (P = 0.59083; Fisher's exact test) (Table S1). Among a total of isolates the ability to biofilm formation in BHI was observed in 87% (47 isolates); 70.4% (38 isolates) were classified as moderately or strongly adherent and 16.7% (9 isolates) were weakly adherent. Seven isolates (13%) showed no ability to form biofilm in BHI broth (Table 1).
In the E. faecalis strains, there was a link between biofilm formation in BHI broth supplemented with 2% glucose and hydrophobicity (P = 0.00718; Fisher's exact test). However, there was no correlation between biofilm formation and hydrophobicity in the E. faecium strains in the same conditions (P = 0.83333; Fisher's exact test).
Detection of virulence genes. The prevalence of virulence genes detected in all isolates is shown in Table 2.
No esp, hyl, cylA, cylB, cylM, and cyl L genes were detected in any of the tested isolates.
The efaA fs and efaA fm genes were found in all E. faecalis and E. faecium isolates, respectively. None of these genes was detected in E. hirae, E. casseliflavus, and E. durans.
The presence of the gelE and sprE genes associated with the fsrABC locus was detected in all 27 E. faecalis isolates, 5 E. faecium, and 1 E. hirae. Additionally, fsrA and fsrC were found in two gelE-and sprE-positive E. hirae isolates. Moreover, the ace gene was detected in the fsrABC-positive isolates, except one E. faecalis.
The sex pheromone determinants (cpd, cob, ccf) were widespread in the analysed E. faecalis isolates (100% strains). The ccf gene was also detected in all but one E. faecium. Moreover, the cpd and cob determinants were showed in E. faecium (five and two isolates, respectively) and E. hirae (tree and two isolates, respectively).
The genotypic patterns of the virulence factors detected in the isolates are shown in Table S1.

Discussion
The study was conducted to determine the prevalence of biofilm-forming ability among gut enterococci isolated from wild birds and its correlation with virulence genes. There are no literature reports of biofilm formation and hydrophobicity in enterococci isolated from wild birds. In in vitro conditions, the microplate method was found to be the most common and effective approach for detection of biofilm production 20 . As was noted during our studies, the composition of the medium was important to show the ability of enterococci to in vitro biofilm formation. Other authors also pointed to similar dependencies 20,21 .
Moreover, we observed that the E. faecalis strains produced biofilm in BHI more often than E. faecium. Leuck et al. 22 recently tested the biofilm-forming ability of E. faecalis isolates on polystyrene dishes. The authors reported that many of the clinical isolates showed low-level biofilm formation even when different types of media were used. However, they observed that clinical isolates were able to form biofilms on a relevant tissue surface.   www.nature.com/scientificreports www.nature.com/scientificreports/ Additionally, we found a correlation between hydrophobicity and biofilm formation in only E. faecalis. In addition, the E. faecalis isolates showed significantly higher hydrophobicity than E. faecium. It is known, that bacterial cell surface hydrophobicity is important for the interactions between the bacterium and host epithelial cells 23 . The hydrophobicity of the enterococcal cell surface is increased by the presence of aggregation substances. Strains possessing the agg or asa1 gene may form large cell aggregates during infection 24 . In the present study, the genetic determinants of aggregation substances were most frequently detected in E. faecalis, followed by E. faecium and E. hirae. The aggregation substance is a sex pheromone plasmid-encoded surface protein. Some of E. faecalis strains with all sex pheromone genes (cpd, cob, and ccf) exhibited the presence of the agg and asa1 genes as well. Similarly, Martin and coworkers 25 noted the presence of the agg, cpd, and ccf genes in all E. faecalis isolates. Additionally, production of sex pheromones by E. faecalis may favour acquisition of antibiotic resistance and virulence from other enterococci, resulting in increased virulence. In the present study, a lower proportion of sex pheromone genes were observed in E. faecium in comparison with E. faecalis isolates. However, the sex pheromone genes in E. faecium may reflect sequence divergence, which may explain this result. Importantly, Eaton and Gasson 26 showed that virulence determinants can be transferred from pathogenic strains to non-pathogenic strains (used as starters in food). The authors did not achieve transfer into E. faecium strains, although sex pheromone cross talk between E. faecium and E. faecalis has been demonstrated 27 .
Our finding corroborates the results reported by Martin and coworkers 25 , who demonstrated that all E. faecalis and E. faecium strains from Wood Pigeon (Columba palumbus) carried the efaA fs and efaA fm virulence genes, respectively, while the ace gene was more often found in E. faecalis strains from wild birds in Poland than from Partridges in Portugal 16 . Both efaA fs and ace genes play a role in the pathogenesis of endocarditis, whereas the role of efaA fm is yet unknown.
Some authors demonstrated that the presence of pili genes in enterococcal strains is required for the establishment of the first step of infection 28 . As shown in our study, most of the analysed Enterococcus spp. carried these genes; however, a correlation was only observed between biofilm formation in BHI with 2% glucose medium and the presence of the pil gene.
No esp, hyl, cylA, cylB, cylM, cyl L genes were detected in any of the tested strains in our studies. The absence of the cyl operon was associated with negative β-haemolysis. Similarly, no β-haemolytic activity was found in enterococal strains from wild boars (Sus scrofa) in Spain, although the strains contained different combinations of cyl genes, in contrast to our results 17 . Olsen and coworkers 29 showed that the cylA gene was detected more often in clinical than in commensal poultry isolates, where none of the isolates was haemolytic. As indicated previously, cytolysin activity requires the presence of the whole cyl operon (cyl L L L S ABM) 26,30 . This was also confirmed in the study conducted by Silva and coworkers 31 . Cytolysin exerts activity against a broad spectrum of cell types including a wide range of Gram-positive bacteria, eukaryotic cells such as human, bovine, and horse erythrocytes, retinal cells, polymorphonuclear leukocytes, and human intestinal epithelial cells 32 .
As in our results, Silva and coworkers 16 did not detect any esp genes in enterococci isolated from Partridge (Alectoris rufa), whereas this gene was described seven years later in six vanA-positive E. faecium isolated from the same bird species 33 . Additionally, the esp gene was found in two vanA/B2-positive E. faecalis strains 34 . Interestingly, the esp gene in E. faecalis is located on a large genetic component (150 kb) characterised by all features of the pathogenicity island, whose presence is characteristic for multiresistant isolates, including vancomycin-resistant 35 . Indeed, the esp gene is present predominantly in strains associated with infections and hospital outbreaks 36,37 .
In contrast to the isolates from the wild birds analysed in our study, the hyl gene was detected in two and one E. faecium strains isolated from Wild Boars and Partridges, respectively 16,17 . Moreover, the hyl gene was found in five vanA-positive E. faecium strains from wild partridges 33 and three vanA/vanB2 E. faecalis strains from two Cattle Egrets and one Common Ringed Plover 34 . Gram-positive genera capable of elaborating hyaluronidase are able to cause infections initiated at a mucosal or skin surface of either humans or animals 38 .
It was reported in our study that all E. faecalis strains that exhibited gelatinase activity harboured the gelE, sprE, and fsrABC genes, which is in agreement with results reported by other authors 17,18 . Additionally, only one E. faecium and one E. hirae strains had gelatinase activity and harboured gelE, sprE, and all fsr operon genes. A discrepancy between the presence of gelE 19 including sprE and fsrABC genes 18 and production of gelatinase in enterococci was also observed which coincide with our results. Gelatinase (gelE) is co-transcribed with serine protease (sprE) and regulated by the quorum-sensing fsr locus. It can also cleave sex pheromones, which are known to be potent chemo-attractants 39 and might therefore modulate the host response.
Many authors indicate the presence of numerous genes and virulence factors in both pathogenic and opportunistic bacteria 28,[40][41][42] . However, they do not specify which of them may have of greatest importance for pathomechanism of infections, because it is a complex and multi-stage process and depends of many factors, including bacterial virulence as well as the conditions of the host and habitat and the presence of another components of microbiota. Similarly, it is difficult to do so in the case of enterococci isolated during our studies from the gut microbiota of wild birds. However, based on our findings, commensal enterococci from the wild birds had some virulence determinants and could be a source of potential pathogenic strains for humans and animals, especially that some of them were determined as antibiotic resistant epidemic clones. This hypothesis can be confirmed by the recent results of investigations of virulence factors in vancomycin-resistant enterococci from wild birds obtained by Ben Yahia and coworkers 34 , showing that vanA/vanB2 E. faecalis strains can also harbour the important virulence determinants.

Conclusion
In conclusion, the data presented in this study can help to elucidate the prevalence of virulence factors in enterococcal isolates from wild birds in Poland and indicate that Enterococcus species should be considered as a possible source for virulence determinants. None of the analysed genes should be considered definitive markers of Scientific RepoRtS | (2019) 9:11204 | https://doi.org/10.1038/s41598-019-47602-w www.nature.com/scientificreports www.nature.com/scientificreports/ pathogenicity in the tested bacteria. Moreover, the results of this study showed that the presence of pathogenic factors such as the esp, hyl, and cyl operon genes did not seem to be necessary or sufficient for the production of biofilm by enterococci in the analysed conditions. However, the presence of some adhesion and gelatinase genes has been detected in biofilm-positive isolates. It appears that many environmental conditions, e.g. the medium composition, and genetic factors may be associated with the pathomechanism and production of biofilm by enterococci. Therefore, the environment, e.g. organs outside the gastrointestinal tract where the bacteria live, affects their surface activity and intercellular interactions. In some cases, our strains also possessed silent virulence genes.

Methods
Collection of strains. The collection of 54 Enterococcus isolates (E. faecalis, 27 isolates; E. faecium, 18 isolates; E. hirae, 5 isolates; E. durans, 2 isolates; and E. casseliflavus, 2 isolates) from cloacal swabs of 52 free-living birds representing 25 species was studied (Table S1). The swabs for bacteriological analysis were collected from birds after their delivery to the Centre for Rehabilitation of Wild Birds, University of Life Sciences in Lublin, which receives injured or weak birds. The cloacal samples from the birds were collected by a veterinarian as part of his work and based on the authorization to collect biological material for research purposes by the rpoA gene sequencing, as previously described 43 . In addition, antibiotic resistance and genetic diversity of analysedstrains were determined previously 9,10 . The isolates were stored at −80 °C in Brain Heart Infusion Broth (Oxoid, Basingstoke, Hampshire, UK) with 20% sterile glycerol for further analysis.

Haemolysin and gelatinase activity screening. Haemolysis was evaluated by plating the strains on
Columbia Agar Base (OXOID, Basingstoke, Hampshire, UK) supplemented with 5% defibrinated horse blood (Pro Animali Company, Wroclaw, Poland). The plates were incubated at 37 °C for 24 h in aerobic conditions. A positive result was indicated by the formation of haemolytic (clear) zones around the colonies. E. faecalis ATCC29212 (LGC Standards, Łomianki, Poland) was used as a positive control.
Gelatinase production was detected by inoculating the Enterococcus strains onto Trypticase Soy Agar (OXOID, Basingstoke, Hampshire, United Kingdom) containing 3% gelatine (Avantor Performance Materials, Gliwice, Poland). The appearance of a clear halo around the colonies after incubation at 37 °C for 24 h in aerobic conditions followed by refrigeration at 4 °C for 30 min was considered to be a positive indication of gelatinase production. E. faecalis ATCC29212 (LGC Standards, Łomianki, Poland) was used as a positive control.
Hydrophobicity and biofilm assays. Cell surface hydrophobicity was determined using the method developed by Dec and coworkers 44 . Each isolate was subcultured on Columbia Agar Base (OXOID, Hampshire, UK) supplemented with 5% defibrinated horse blood (Pro Animali Company, Wroclaw, Poland) at 37 °C. 24-h cultures were harvested and suspended in 5 ml of 0.9% NaCl to an optical density (OD 600 ) of 0.8-1.0 (A 0 ). Then, xylene (1.7 ml) was added to glass test tubes and the mixtures were vortexed vigorously for 90 s. After phase separation (ca. 15 min.), the optical density of the aqueous phase (A) was measured again and compared with the initial value. The percentage of cell surface hydrophobicity (%H) of the strain adhering to xylene was calculated using the equation: %H = [(A 0 − A)/A 0 ] × 100. Strains with hydrophobicity equal or higher than 50% were considered hydrophobic.
Biofilm assays were conducted based on a method described by Stepanović and coworkers 20 using Tryptic Soy Broth (TSB) (Oxoid, Hampshire, UK) supplemented with 1% glucose and Brain Heart Infusion (BHI) (Oxoid, Hampshire, UK) supplemented with 2% glucose. Each isolate was subcultured on Columbia Agar Base (OXOID, Hampshire, UK) supplemented with 5% defibrinated horse blood (Pro Animali Company, Wroclaw, Poland) at 37 °C. After verification of the purity of the strain, a few colonies with identical morphology are suspended in physiological saline. Then, the turbidity of the bacterial suspension was adjusted to match turbidity comparable to that of the 0.5 McFarland standard (~10 8 CFU/ml) using a densitometer Biosan DEN-1 (Biogenet, Józefów-Otwock, Poland). Then, for each strain tested, 20 µl of bacterial suspensions were transferred to four wells of two separate sterile flat-bottomed 96-well polystyrene microtitre plates containing 180 µl of TSB supplemented with 1% glucose and 180 µl of BHI supplemented with 2% glucose, respectively. For the negative control, 200 µl of broths (TSB and BHI, both with glucose) were dispensed into eight vertical wells per plate. The plates were incubated under stationary aerobic conditions at 37 °C. After incubation for 24 hours, the broths were carefully removed. The wells were gently washed three times with phosphate-buffered saline (PBS, pH 7.2). Following every washing step, the wells were emptied by flicking the plates. Prior to biofilm staining, the plates were left at room temperature for drying in an inverted position overnight. The adherent biofilm layer formed in each microtitre plate well was stained with 200 µl of 0.1% crystal violet solution in water for 15 min at room temperature. After staining, the stain was aspirated with a pipette and excess stain was rinsed off by placing the microtitre plate under running tap water. Washing was continued until the washings were free of the stain. After the microplates were dried at room temperature, the dye bound to the cells was resolubilised with 200 µl of 96% ethanol per well for 30 min without shaking. The optical density (OD) of the resolubilised crystal violet was then measured at 570 nm (OD 570 ) using a microplate reader (Bio-Rad, Model 680). Each assay was performed in quadruplicate on three occasions for 12 readings for each strain. Wells containing uninoculated medium served as negative controls to determine the background optical density. After subtracting the mean background OD 570 readings, the 12 optical density readings per strain were averaged to obtain the mean OD 570 reading for each strain. Based on the bacterial biofilm, the isolates were classified into four categories: non-biofilm producers, weak, moderate, or