Virulent Synergistic Effect between Enterococcus faecalis and Escherichia coli Assayed by Using the Caenorhabditis elegans Model

Background The role of enterococci in the pathogenesis of polymicrobial infections is still debated. The purpose of this study was to evaluate the effect of virulent enterococci in the presence or absence of Escherichia coli strains in the in vivo Caenorhabditis elegans model. Principal Findings This study demonstrated that there was a synergistic effect on virulence when an association of enterococci and E. coli (LT50 = 1.6 days±0.1 according to the tested strains and death of nematodes in 4 days±0.5) was tested in comparison with enterococci alone (LT50 = 4.6 days±0.1 and death in 10.4 days±0.6) or E. coli alone (LT50 = 2.1±0.9 and deaths 6.6±0.6) (p<0.001). In addition, there was a relation between the virulence of E. faecalis strains alone and the virulence potential of the association with E. coli strains. Finally, in the presence of avirulent E. coli strains, enterococci have no effect (LT50 = 4.3±0.5 and deaths in 10.8±0.8), independently of the level of their own virulence, demonstrating that the ‘enterococci effect’ only occurred in the presence of virulent E. coli strains. Conclusion This study allows a better understanding of a bacterial cooperation. Moreover, it could help to optimize the antibiotic regimen during polymicrobial infections.


Introduction
The role played by enterococci in the pathogenesis of intraabdominal infections is still debated because enterococci are frequently isolated from the polymicrobial flora [1][2][3]. In order to get insight into the role of enterococci in polymicrobial infections, Carbon et al. developed animal models and showed that enterococci might have a synergistic effect on the pathogenic traits of associated microorganisms, particularly Escherichia coli [4][5][6]. In the present study, we used another in vivo model, namely the Caenorhabditis elegans model, which was previously used to explore virulence of various micro-organisms (Serratia marcescens, Salmonella enterica, Pseudomonas aeruginosa, Burkholderia cepacia, E. coli, Staphylococcus aureus and Streptococcus pyogenes [7][8][9][10][11][12][13]). Garsin et al. also used this model to investigate Enterococcus faecalis related virulence factors [14,15]. To our knowledge, the present study is the first one in which the C. elegans model was used to assess the virulence of associated microorganisms such as E. coli and E. faecalis.

Results
The virulence factor-encoding genes detected in the different strains of E. faecalis are listed in Table 1.
When E. coli and E. faecalis strains were associated, a virulent synergistic effect was observed, irrespective at the specific uropathogenic E. coli and E. feacalis strains associated ( Fig. 1C  and 1D). The LT50s obtained with the strain associations varied between 1.5 and 1.7 days and were significantly shorter than the LT50s observed with E. coli strains alone (p,0.001). To demonstrate the role of enterocci in the increase in virulence, the Enterococcus spp. strain was replaced by the S. mitis strain (which harboured ply gene). In the nematode killing model, E. faecalis and S. mitis strains presented a similar virulence (LT50 NSTRMIT 5.4 days vs LT50 NEF26 4.6 days; (p: not significant)) (Fig. 1A). The association of E. coli NECS19923 and the S. mitis strain did not increase the virulence observed in the nematode model (LT50 NECS19923+NSTRMIT 2.2 days) in comparison with the virulence obtained with E. coli alone (LT50 NECS19923 2.1 days) (Fig. 1C). Conversely, the virulence of the association of the S. mitis strain with E. faecalis (LT50 NEF26+NSTRMIT 3.6 days) was significantly higher than the virulence observed with either S. mitis alone (LT50 NSTRMIT 5.4 days) and enterocci alone (LT50 NEF26 4.6 days) (p,0.001) (Fig. 1C). All these findings strongly suggested the essential role of E. faecalis strains in the synergistic effect. However this synergistic effect of the association between E. coli and E. faecalis was only detected in presence of virulent E. coli strains. This was suggested by the LT50s observed when E. faecalis strains were associated with the avirulent E. coli OP50 strain (LT50 NEF26 4.5 days vs LT50 NEF26+OP50 4.3 and LT100 NEF26 10.4 days vs LT100 NEF26+OP50 10.8; p: not significant).
Moreover, as indicated in Fig. 2, a correlation was observed between the value of the LT50 of the E. faecalis strains tested alone and the value of the LT50 of the association of E. coli and E. faecalis strains.
To verify the ingestion of E. coli, S. mitis and E. faecalis strains and the proliferation in the C. elegans intestine, L4 C. elegans were fed on a mixed lawn of E. faecalis and E. coli, E. faecalis and S. mitis or S. mitis and E. coli at a same ratio 1:1. At 72 h, the nematodes were washed to remove bacteria from their surfaces and then disrupted to recover bacteria from inside their digestive. The number of E. faecalis, S. mitis and E. coli CFU within the nematode gut varied between 10 5 and 10 6 bacteria per worm between each combination without statistical difference ( Table 2).

Discussion
Both the clinical significance of enterococci in peritonitis and the use of empiric anti-enterococcal coverage have been the subject of a long-lasting and ongoing debate [16][17][18][19]. In a literature review recently published by Harbarth and Uckay, was presented evidence arguing in favour of using empirical therapy with enterococcal coverage for intra-abdominal infections among certain patients considered as high-risk [19].
The pathogenic role of enterococci has been studied in animal models. Onderdonk et al. showed in a model of peritonitis in rat that enterococci inoculated alone induced neither early death nor later abscesses [20]. They demonstrated a synergic effect between enterococci and anaerobes but not between enterocci and E. coli. Later, Carbon et al. underlined several important points by using another model of experimental intra-abdominal infection. They studied the pathophysiological role of enterococci in a non fatal model of peritonitis in rats by implanting a gelatin capsule containing E. coli and Bacteroides fragilis with or without increasing concentrations of E. faecalis or heat-inactivated enterococci [6]. Effects were measured after several hours and several days following inoculation. Thus, they observed that the highest concentration of enterococci was correlated with an increase in E. coli and proinflammatory cytokines (tumor necrosis factor and interleukin) concentrations [21]. They suggested that enterococci might limit phagocytosis and intracellular death of other pathogens, particularly E. coli. In another study, the same authors evaluated the effects of E. faecalis strains containing various combinations of virulent factors in mouse and rat models of peritonitis. However, no conclusion was provided because of variable findings [4].
Currently, ten of enterococcal virulence factors are known. Their exact role in pathogenic mechanisms remains dubious and in several publications no conclusions were possible or were contradictory. In this study we screened for the six most studied virulence factors. Aggregation substance, encoded by asa1, is a pheromone-inducible protein that increases (i) bacterial adherence to renal tubular cells and heart endocardial cells, (ii) internalization by intestinal cells, and (iii) the valvular vegetation mass in an animal model of endocarditis. Gelatinase, encoded by gelE, is an extracellular zinc endopeptidase that hydrolyzes collagen, gelatin, and small peptides and that has been shown to exacerbate endocarditis in an animal model [22]. The production of cytolysin (encoded by cylA) has also been shown to significantly worsen the severity of endocarditis and endophthalmitis in animal models and is a virulence factor that affects C. elegans killing [14,22]. The enterococcal surface protein, encoded by esp, is associated with increased virulence, colonization and persistence in the urinary tract, and biofilm formation. Hyaluronidase, encoded by hyl, contributes to invasion of the nasopharynx and pneumococcal pneumonia [22]. Hemolysin, encoded by hlyB, is a cytotoxic protein that contributes to the destruction of red blood cells. Vergis et al. showed that the presence of gelatinase, hemolysin and enterococcal surface protein singly or in combination was not associated with mortality among patients with bacteremia due to Table 1. Detection of virulence factor (VF)-encoding genes (asa1, gelE, cylA, esp and hyl) and production of hemolysin (hlyB), gelatinase and aggregation substance among E. faecalis strains.

Strain
Origin gelE cylA esp hyl asa1 VF score Clumping factor hlyB E. faecalis [23]. In this work, the most virulent E. faecalis strains did not necessarily contain the most number of virulence genes and no virulent 'markers' could be identified, including Cyl, that caused faster killing in the nematode model [14]. This confirms the difficulty in understanding the virulence of enterococci.
In the present study, using for the first time the C. elegans model to investigate microorganism association, we clearly demonstrated a synergistic interaction between E. coli and E. faecalis, two bacteria frequently isolated from peritonitis pus and UTI and also known to independently induce an intestinal infection in this nematode model [9,14]. The LT50s of the associations of E. faecalis and E. coli strains were definitively smaller than the LT50s observed with each microorganism tested alone. Interestingly the higher the level of virulence of the E. faecalis strain, the more virulence was observed in the bacterial association with E. coli. However, the synergic effect did not seem to be influenced by the number of virulence factors of E. faecalis as described previously [24], but seemed to be influenced by the virulence of E. coli. Indeed no synergistic effect was observed between E. faecalis and E. coli OP50, a strain with low virulence potential. Such findings are in accordance with a recent data that have shown an association between the presence of numerous virulence factors, specific O types, B2 phylogenetic group and virulence in the C. elegans and murine models among uropathogenic E. coli strains [25,26]. Finally the determination of the number of bacteria within the C. elegans digestive tract validated the presence and the overgrowth of E. faecalis, E. coli and S. mitis in the intestine of nematodes. This result demonstrates the establishment of a true infection validating our results. As previously demonstrated, E. faecalis not only accumulates in the intestinal lumen but proliferates in the gut, resulting in a long-lasting infection that persists for the entire lifespan of the worms [14].
Many questions concerning the exact pathogenicity of enterococci remain unanswered. However, this model suggests the role of enterococci in increasing virulence in a polymicrobial infection. This is consistent with Carbon's model. This study allows us to better understand bacterial cooperation and it could help to optimize the antibiotic regimen during polymicrobial infections.

Bacterial strains and growth conditions
Eight E. faecalis strains (NEF26, NEF42, NEF889, NEF21895, NEF17649, NEF64, NEF48015, NEF27828) isolated from the blood and urinary tract infections [27] (Table 1) were studied alone and combined with other microorganisms. These microorganisms, which also were studied alone, comprised Streptococcus mitis strain NSTRMIT isolated from the blood and E. coli strains NECS19923, NECS23048 both isolated from UTIs [9]. Two others strains were used as controls in the different experiments namely E. faecalis JH2-2 (control strain for expression of aggregation substance) and the avirulent E. coli OP50 (an international standard food for nematodes without known virulence factors). Bacteria were grown in trypticase soy broth or agar (TSA) or Columbia agar gelose with 5% fresh sheep blood (bioMerieux, Marnes La Coquette, France) at 37uC in 5% CO 2 .

Nematode Killing Assay
The C. elegans infection assay was carried out as described by Kurz et al. [25] except that the Fer-15 mutant line, which has a temperature sensitive fertility defect, were used rather than wild type N2 worms. Fer-15 was provided by the Caenorhabditis Genetics Center, which is funded by the NIH National Center for Research Resources (NCRR). To synchronize the growth of worms, eggs were collected using the hypochlorite method. Overnight cultures in brain heart infusion (BHI) medium of E. coli, S. mitis and E. faecalis strains were harvested by centrifugation, washed once and suspended in phosphate buffered saline solution (PBS) at pH 7.0 at a concentration of 10 5 CFU/ml. BHI agar plates were inoculated with 10 ml of each strain and incubated at 37uC for 8210 h. Plates were allowed to cool to room temperature and seeded with L4  stage worms (20-30 per plate). Plates were then incubated at 25uC and scored each day for live worms under a stereomicroscope (Leica MS5) [9]. At least three replicates repeated 5 times were performed for each selected clone. A worm was considered dead when it no longer responded to touch. Worms that died as a result of becoming stuck to the wall of the plate were excluded from the analysis. Lethal Time 50% (LT50) and death corresponded to time (in hours) required to kill 50% and 100% of the nematode population, respectively.
Measuring the number of bacteria within the C. elegans digestive tract This assay was carried out as described by Garsin et al. [14]. Five C. elegans were picked at 72 h, and the surface bacteria were removed by washing the worms twice in 4 ml drops of M9 medium on a BHI agar plate containing 25 mg/ml gentamicin. The nematodes were placed in a 1.5 ml Eppendorf tube containing 20 ml of M9 medium with 1% Triton X-100 and were mechanically disrupted by using a pestle. The volume was adjusted to 50 ml with M9 medium containing 1% Triton X-100 which was diluted and plated on BHI agar containing 50 mg/ml ampicillin. At least three replicates were performed for each bacterial combination.

Statistical analysis
To compare the entire survival curves in nematode killing assays, we used a Cox regression. In order to perform pairwise comparison between two different strains, we used a log rank test. The analysis was carried out using SPSS 6.1.1 (SPSS Inc., Chicago, IL, USA).