Protozoan Predation of Escherichia coli O157:H7 Is Unaffected by the Carriage of Shiga Toxin-Encoding Bacteriophages

Escherichia coli O157:H7 is a food-borne bacterium that causes hemorrhagic diarrhea and hemolytic uremic syndrome in humans. While cattle are a known source of E. coli O157:H7 exposure resulting in human infection, environmental reservoirs may also be important sources of infection for both cattle and humans. Bacteriophage-encoded Shiga toxins (Stx) carried by E. coli O157:H7 may provide a selective advantage for survival of these bacteria in the environment, possibly through their toxic effects on grazing protozoa. To determine Stx effects on protozoan grazing, we co-cultured Paramecium caudatum, a common ciliate protozoon in cattle water sources, with multiple strains of Shiga-toxigenic E. coli O157:H7 and non-Shiga toxigenic cattle commensal E. coli. Over three days at ambient laboratory temperature, P. caudatum consistently reduced both E. coli O157:H7 and non-Shiga toxigenic E. coli populations by 1–3 log cfu. Furthermore, a wild-type strain of Shiga-toxigenic E. coli O157:H7 (EDL933) and isogenic mutants lacking the A subunit of Stx 2a, the entire Stx 2a-encoding bacteriophage, and/or the entire Stx 1-encoding bacteriophage were grazed with similar efficacy by both P. caudatum and Tetrahymena pyriformis (another ciliate protozoon). Therefore, our data provided no evidence of a protective effect of either Stx or the products of other bacteriophage genes on protozoan predation of E. coli. Further research is necessary to determine if the grazing activity of naturally-occurring protozoa in cattle water troughs can serve to decrease cattle exposure to E. coli O157:H7 and other Shiga-toxigenic E. coli.


Introduction
Diseases caused by the bacterium Escherichia coli O157:H7 are major public health concerns worldwide, with the highest numbers of reported cases in the United States, United Kingdom, Argentina, Canada and Japan [1,2]. Many human infections are associated with the consumption of contaminated food and water, or through direct contact with cattle or their surrounding environments [3,4]. E. coli O157:H7 virulence factors include the locus of enterocyte and effacement (LEE) pathogenicity island and one or more Shiga toxins (Stx), primarily Stx1, Stx2a, and Stx2c. The LEE pathogenicity island is responsible for attaching and effacing lesions seen in both humans and cattle. Stx are encoded by lambdoid lysogenic bacteriophages and are released from lysed bacteria. In humans, Stx targets endothelial cells in the colon contributing to hemorrhagic colitis [5], and renal glomerular endothelial cells in some cases resulting in the life-threatening disease, hemolytic uremic syndrome (HUS) [6,7].
The high prevalence of carriage of Shiga toxin-encoding bacteriophages within the E. coli O157:H7 lineage raises a possibility that these prophages confer a selective advantage to the bacterium; however, the nature of this selective advantage is currently unknown. While Stx play a central role in E. coli O157:H7 pathogenesis in humans, the infection is likely too rare to provide the bacterium a significant selective advantage. In cattle, E. coli O157:H7 colonize the rectoanal junction (RAJ) resulting in transient and intermittent fecal shedding at levels as high as 10 6 colony forming units (CFU) per gram of feces [8,9]. Unlike humans, cattle lack vascular receptors for Stx [10] and therefore are not generally susceptible to disease associated with E. coli O157:H7 [10][11][12]. Furthermore, deletion of Stx genes does not change the ability of E. coli O157:H7 to colonize the bovine RAJ [13]. Given the lack of an observed selective role for Stx in mammalian hosts, it has been suggested that Stx may alternatively offer a selective advantage to E. coli O157:H7 in the environment by protection against predation (grazing) by free-living protozoa [14][15][16][17].
Previously, some investigators have observed that Stx-encoding bacteriophages protect E. coli O157:H7 from grazing by laboratory strains of Tetrahymena pyriformis [14] and T. thermophila [15,17], even killing the latter [15,17]. Other investigators have observed that E. coli O157:H7 is readily grazed by colpodean and oligohymenophorean ciliates [18,19] isolated from dairy farm waste-water and by unidentified protozoa collected from cattle water troughs [20]. The goal of this study was to determine the roles of Stx and their encoding bacteriophages in protection against protozoan grazing. We evaluated the ability of protozoa to graze diverse genotypes of Shiga-toxigenic E. coli O157:H7 over three days using cattle water trough isolates of the protozoon, Paramecium caudatum. We also evaluated the ability of protozoa to graze E. coli O157:H7 strain EDL933 and isogenic mutants with deletion of the Stx2a subunit A-encoding nucleotide sequence, deletion of the entire Stx 2a-encoding bacteriophage sequence, and/or deletion of the entire Stx1-encoding bacteriophage sequence, using both P. caudatum and the laboratory strain of T. pyriformis previously described as susceptible to Stx [14].

Microbial agents
Paramecium caudatum was isolated from cattle water troughs where, frequently, it is a predominant member of the protozoan flora (unpublished observations). P. caudatum was cultivated in lidded Cellstar 1 24 well polystyrene culture plates with 1 ml of autoclaved trough water at ambient laboratory temperature, and the amplified protozoa were maintained at 4°C. Identification was confirmed by 18S rDNA sequences, as follows: Protozoan DNA was extracted (MagMAX™ Total Nucleic Acid Isolation Kit, AM1840, Applied Biosystems Inc.) per the manufacturer's recommendations for liquid samples. Polymerase chain reaction (PCR) for the 18S rRNA gene was performed on the extracted DNA using primers (Invitrogen, Carlsbad, CA) P-SSU-342f and Medlin B described by Karnati et al [21] (Table 1). Each 25 μl PCR reaction contained 20 pmol of each primer, 1X PCR buffer, 125 μM of deoxynucleotide triphosphates (dNTP), 2mM magnesium chloride, and 2.5 U of Invitrogen™ Platinum Taq polymerase. Thermocycler (Bio-Rad iCycler) settings included an initial denaturation step (94°C, 2 minutes), followed by 35 cycles of [denaturation (94°C, 1 minute), annealing (37°C, 65 seconds), and extension (72°C, 3 minutes)] with a final 6 minute extension step at 72°C followed by a hold at 4°C. The 1,360-bp amplicon was visualized via gel electrophoresis, using a 0.8% agarose gel stained with ethidium bromide (Sigma-Aldrich, St. Louis, MO). The PCR product was cleaned using EXO/SAP with 8U Exonuclease 1 and 1.6 U FastAP, a Thermosensitive Alkaline Phosphatase (Thermo Scientific) at thermocycler settings of 20 minutes at 37°C, 15 minutes at 80°C, followed by a 4°C hold. The cleaned PCR product was submitted to Amplicon Express (Pullman, WA) for forward and reverse 18S sequencing using the same primers used for amplification. Sequences were compared to best fit matches using the NCBI BLAST algorithm (http://blast.ncbi.nlm.nih.gov/Blast.cgi).
Tetrahymena pyriformis strain GL-C, used as a control for bacterial grazing, was obtained from Cornell University (SD00707, Tetrahymena stock center, Ithaca, NY). T. pyriformis was cultivated at ambient laboratory temperature in 25cm 2 polystyrene tissue culture flasks (Corning Glass Works, Corning, NY) in modified Neff medium (Rich Axenic Nutrient Media, https://tetrahymena.vet.cornell.edu/recipes.php) and amplified cultures were maintained at 4°C.
E. coli strains used in this study (Table 2) were maintained as glycerol stocks at -80°C. Overnight cultures of these bacterial stocks were prepared with the appropriate antibiotics (rifampicin 50 μg/ml, nalidixic acid 30 μg/ml, or chloramphenicol 25 μg/ml) according to their antibiotic resistance profile (Table 2) in Lennox broth (LB) (Invitrogen™, Carlsbad, CA) shaken (250 rpm) at 37°C. The overnight cultures were used to streak LB plates with the same selective antibiotics for isolation of colonies. Isolated colonies were subsequently grown as individual  Table 2) were created using the pKD46 and pKD3 plasmids (E. coli Genetic Resource Center, Yale University, New Haven, CT) as described by Datsenko et al. [22] and primers listed in Table 1. EDL933 was cultured in LB at 37°C shaking at 250 rpm to an optical density of 0.6 at 600 nm (OD 600 ). The cells were made electrocompetent by washing three times with ice-cold 10% glycerol. Electrocompetent cells were transformed with the pKD46 plasmid via electroporation, and ampicillin (100μg/ml) was used to select for transformed colonies at 30°C overnight. Transformed colonies were grown in LB with 5mM L-arabinose at 30°C shaking at 200 rpm to an optical density of 0.6 at 600 nm (OD 600 ). Cells were made electrocompetent as described above and transformed with the linear DNA fragment amplified from a pKD3 plasmid flanked by DNA sequences homologous to the knock-out target [22]. Chloramphenicol (25μg/ml) was used to select for transformed colonies at 30°C overnight. Isolated transformed colonies were then grown at 43°C to cure the pKD46 plasmid, and banked in glycerol at -80°C. Prior to banking, EDL933 mutants were confirmed using PCR for the knockout target gene (stx2a A subunit) and by Stx-associated bacteriophage insertion (SBI) genotyping for the Stx 2a-, Stx 1-, and combined Stx 1-& Stx 2a-bacteriophage knock-outs [23]. The loss of Stx production was confirmed by negative results on the Meridian ImmunoCard STAT!EHEC lateral flow devices (Bioscience, Inc., Germany) for Stx 1 and 2 compared with the parent strain.
Protozoan grazing of Shiga-toxigenic E. coli compared to non-Shiga toxigenic E. coli Protozoa culture suspensions (1 ml) were added to 1.5 ml autoclaved conical tubes with loosened caps (USA Scientific). On day zero, Shiga toxigenic E. coli and non-Shiga toxigenic E. coli with selectable resistance markers ( Table 2) were added to these suspensions to create mixed cultures, with multiplicities of infection (MOI) between 1 x 10 4 and 1 x 10 5 bacterial cfu per protozoan. Colony forming units were counted from serial decimal dilutions of the co-culture suspensions following 0, 24, 48 and 72 hours of co-culture using LB plates with appropriate selective antibiotics. Protozoan densities were determined every 24 hours via light microscopy. For each co-culture experiment, parallel cultures containing the same bacterial inoculum in the absence of protozoa served as negative controls, using sterile modified Neff media for T. pyriformis assays and cattle trough water treated with 10 μg/ml cycloheximide (Sigma-Aldrich, St. Louis, MO) for P. caudatum assays. Three biological replicates were performed for all coculture experiments.
Growth patterns of E. coli O157:H7, cattle commensal E. coli, and EDL933 mutants Bacterial growth curves for each of the bacterial strains (Table 2) were determined using a Bioscreen C MBR (Oy Growth Curves AB Ltd., Piscataway, NJ). For each growth curve, approximately 1 x 10 7 bacteria/ml of individual bacterial strains used throughout the study was added to ten separate wells (five biological replicates) of a sterilized Honeycomb 2 100-well plate (Oy Growth Curves AB Ltd.). Bacterial inocula were suspended in cycloheximide-treated trough water (P. caudatum studies) or modified Neff media (T. pyriformis studies) to mimic the culture conditions of co-culture experiments. The mean OD 420-580nm (wideband) were determined every 2 hours for three days at ambient laboratory temperature (23°C) without shaking to create growth curves (S1 and S2 Figs).

Statistical analysis
Repeated measures ANOVAs were used to evaluate differences in protozoan grazing of Shigatoxigenic and non-Shiga toxigenic E. coli strains over three days of co-culture and in protozoan proliferation over three days with and without added E. coli. P-values <0.05 were considered statistically significant.

P. caudatum grazing of E. coli O157:H7
When co-cultured with P. caudatum for 3 days, viable cell counts of both Shiga-toxigenic E. coli O157:H7 strain EDL933 and cattle commensal Stx-negative E. coli strain #186 were reduced by 1-3 log cfu/ml (Fig 1). No statistically significant differences in the reduction of Shiga-toxigenic versus non-Shiga toxigenic E. coli strains were observed. Fig 2 depicts the ratios of protozoan-grazed Shiga-toxigenic E. coli to grazed non-Shiga toxigenic E. coli. Five diverse strains of Shiga-toxigenic E. coli, each with a different Stx profile (Table 2), were used in these co-culture experiments along with cattle commensal E. coli strain #186. While after the first day there was some variation in the ratios of Shiga-toxigenic:non-Shiga toxigenic bacteria grazed by P. caudatum, by the end of the second and third days, the ratios were near unity for all strains used in this study, with no apparent effect by Stx profile. No decrease in the numbers of P. caudatum were observed for over the three day period; the tendency was instead for increased protozoan numbers compared to control cultures without added bacteria (Fig 3).  approximately 3-logs over three days (Fig 4), with a tendency towards greater predation of the wild-type strain compared to any of the Stx mutants by the end of three days. Co-cultures of the wild-type and mutant strains of EDL933 with T. pyriformis produced very similar reductions of 1-to 3-logs, and no statistically significant differences in the degree of grazing of these strains were observed at any time point. In contrast to P. caudatum, at the end of day 1 of coculture with T. pyriformis there is a trend towards greater predation of mutants compared to wild-type E. coli O157:H7. However, this trend disappeared by the second day. Similar to P. caudatum, T. pyriformis also proliferated over three days when co-cultured with Shiga-toxigenic E. coli (data not shown).

Discussion
In contrast to previous reports that carriage of Shiga toxin-encoding bacteriophages by E. coli O157:H7 provides protection against grazing protozoa [14][15][16][17], our experiments did not detect a protective effect of Stx expression or by other factors encoded on the Stx-encoding bacteriophages of E. coli O157:H7 against grazing protozoa. In Fig 4, within the first 24 hours, there was a trend for higher viable cell counts of wild-type E. coli O157:H7 compared to the Stx mutants when co-cultured with T. pyriformis, similar to previous reports with T. thermophila [15,17]. However, the transient difference was not statistically significant and the trend was not seen with co-cultured P. caudatum. In fact, by the end of the third day, the trend was opposite for P. caudatum, with more of the wild-type E. coli O157:H7 grazed than the Stx mutants.
Overall, the variation in grazing of wild-type E. coli O157:H7 compared to Stx mutants by P. caudatum or T. pyriformis was not statistically significant at any time point throughout the course of the study and protection against grazing by the Stx genes or prophage was not demonstrated.
There was also no statistically significant difference when comparing the ratios of five different strains of grazed E. coli O157:H7 to cattle commensal E. coli, with the ratios approaching unity for all strains by the second day and remaining near unity by the end of the study. Each strain of E. coli O157:H7 had a different Stx profile, with none of the different Stx profiles providing protection against grazing protozoa. While the reason(s) underlying the differences between our results and some previously published findings [14][15][16][17] is unknown, potentially the different strains of E. coli and protozoa used in this study and the different (longer) time course investigated may have played a role. In contrast, the results of this study are consistent with the findings of others [18][19][20] that at least some environmental protozoa on cattle farms efficiently graze E. coli O157:H7. Further investigation of the conditions under which Stx expression may protect against protozoan predation is required to clarify these observations. Several species of protozoa graze bacteria as a food source. P. caudatum is a free-living ciliate protozoon commonly found in cattle farm water troughs. This species is highly bactivorous as demonstrated in this report, similar in this regard to T. pyriformis, another free-living ciliate that has been frequently used in biomedical research. While other environmental strains of protozoa may have preferences on bacterial predation, no such preferences were observed with either of these species in regard to Shiga-toxigenic E. coli compared to non-Shiga toxigenic E. coli.
The proliferation of both T. pyriformis and P. caudatum in the presence of Shiga-toxigenic E. coli demonstrates the ability of these protozoa to use the bacteria as a food source for growth and replication. In contrast to the reports evaluating Tetrahymena thermophila [15,17] and Acanthamoeba castellanii [16], there was no indication that either T. pyriformis or P. caudatum was killed by Shiga-toxigenic E. coli O157:H7.
The growth and survival of cattle commensal E. coli, wild-type EDL933 and Stx mutants were similar in cycloheximide-treated trough water over three days (S1 Fig). Within the first 24 hours, there was increased optical density of cattle commensal E. coli at a few time points, however, this slight variation was not seen for the remainder of the study (S1 Fig inset). In these studies, neither Stx nor their encoding bacteriophages provided a selective advantage for E. coli O157:H7. In modified Neff media, optical density measurements suggested that the bacteriophage knock-out mutant strains reached stationary phase at lower numbers compared to wildtype E. coli O157:H7 EDL933 (S2 Fig). This observation does not change the conclusion that protozoa graze Shiga-toxigenic E. coli as well as non-Shiga toxigenic E. coli; rather increased grazing of Shiga-toxigenic E. coli would be required if the wild type strains proliferate to higher levels than the mutants.
The use of several isogenic mutants of E. coli O157:H7 and the use of a common environmental protozoon isolated from farms in the inland Northwest (P. caudatum) differentiates the results of this study from previous literature. This report provides strong evidence that the stx genes and Stx-encoding bacteriophages do not protect E. coli O157:H7 against grazing by some of the most frequently occurring strains of environmental protozoa in cattle water troughs. Continued research is necessary to investigate other potential roles of Shiga toxins and their encoding bacteriophages in the environment and in the cattle host. Additionally, since commonly-occurring P. caudatum are excellent grazers of E. coli O157:H7, evaluating the use of these protozoa as a pre-harvest control measure on cattle farms is also warranted.