Role of bacteriophages in STEC infections: new implications for the design of prophylactic and treatment approaches [version 1; peer review: 2 approved with reservations]

Shiga toxin (Stx) is considered the main virulence factor in Shiga toxinproducing Escherichia coli (STEC) infections. Previously we reported the expression of biologically active Stx by eukaryotic cells in vitro and in vivo following transfection with plasmids encoding Stx under control of the native bacterial promoter. Since stx genes are present in the genome of lysogenic bacteriophages, here we evaluated the relevance of bacteriophages during STEC infection. We used the non-pathogenic E. coli K12 strain carrying a lysogenic 933W mutant bacteriophage in which the stx operon was replaced by a gene encoding the green fluorescent protein (GFP). Tracking GFP expression using an In Vivo Imaging System (IVIS), we detected fluorescence in liver, kidney, and intestine of mice infected with the recombinant E. coli strain after treatment with ciprofloxacin, which induces the lytic replication and release of bacteriophages. In addition, we showed that chitosan, a linear polysaccharide composed of D-glucosamine residues and with a number of commercial and biomedical uses, had strong anti-bacteriophage effects, as demonstrated in vitro and in vivo. These findings bring promising perspectives for the prevention and treatment of hemolytic uremic syndrome (HUS) cases.


Introduction
Infections by Shiga toxin-producing Escherichia coli (STEC) strains are a serious public health concern, resulting in diarrhea, hemorrhagic colitis, and haemolytic uremic syndrome (HUS).
Stx is the main virulence factor in STEC strains. The stx gene is present in the genome of prophages, which are similar to the bacteriophage lambda found in the lysogenic form of various E. coli strains. Previously we reported that the native promoter of the Stxencoding gene can drive expression of the toxin in eukaryotic cells in both in vivo and in vitro conditions 1,2 .
Many questions remain unanswered with regard to the mechanism by which STEC infection causes HUS. In particular, we are interested in understanding how Stx enters the systemic circulation and why only very small numbers of bacteria are sufficient to induce HUS in humans.
Based on our previous observations that the native stx gene promoter is active in host cells, we seek to understand the role that bacteriophages play in the pathogenesis of STEC strains. Recently, it was reported that bacteriophages carrying the stx gene are required for the development of HUS in the murine model 3 . We hypothesise that eukaryotic host cells might be transduced with and/or infected by Stx-encoding bacteriophages, leading to Stx dissemination in vivo to enter the systemic circulation. This would also explain why very small numbers of bacteria are sufficient to develop HUS.
In order to test whether bacteriophages are responsible for the induction of HUS, we used an anti-bacteriophage agent to inactivate them. Chitosan, a linear polysaccharide polymer obtained after the deacetylation of chitin, the structural element in the exoskeleton of crustaceans, possesses strong antimicrobial activity against several pathogenic microorganisms 4 . Its antiviral activity was reported on the bacteriophage c2, which infects Lactococcus strains, and on bacteriophage MS2, which infects E. coli 5 without affecting significantly the growth of the bacterial culture 6 . In order to test our hypothesis, which would make Stx-encoding bacteriophages a new target for preventing and treating STEC infections, we used chitosan as an anti-bacteriophage agent in vitro and in vivo.
Inactivation of bacteriophages was observed in vitro after incubation with chitosan, inhibiting both the infection of, and replication in bacterial cells, and the transduction of eukaryotic cells.
GFP dissemination was significantly reduced in mice treated with chitosan following infection with a non-pathogenic strain carrying a bacteriophage in which the stx gene was replaced by the GFPencoding sequence. Last, preliminary results showed partial protection by chitosan in vivo of mice infected with STEC.
These results contribute to understanding STEC infections, posing implications for a similar scenario to occur in other infections caused by bacteria carrying lysogenic bacteriophages.

Materials and methods
Strains C600ΔTOX:GFP, a lysogenized C600 strain carrying the 933W bacteriophage in which the stx gene was replaced by the gfp sequence, was generously provided by Dr. Alison Weiss 7 . EDL933W, an enterohemorrhagic E. coli (EHEC) strain carrying the wild-type bacteriophage from which C600ΔTOX:GFP was obtained, was generously provided by Dr. Luis Carlos de Souza Ferreira, LDV-USP, Brazil.

Transduction of eukaryotic cells
Baby BHK-21 cells (Syrian hamster kidney fibroblasts from the American Type Culture Collection) cells were grown on 12-well plates (Nunc) in complete medium (10% fetal bovine serum in DMEM medium, Gibco, USA) for use in the transduction assay.
C600ΔTOX:GFP was generously provided by Dr Alison Weiss 7 . This is a non-pathogenic phage resulting from purified 933W bacteriophage in which stx gene was replaced by gfp sequence (φ ΔTOX:GFP). Phages at a multiplicity of infection (M.O.I) equal to 1 were added to BHK-21 cells cultured the day before on 12 wells plate (Nunc). BHK-21 cells were counted with a Neubauer camera, and bacteriophage titer was measured by the titration assay as described below. Transduction of BHK-21 cells was enhanced by centrifugation at 1000 × g for 10 minutes at room temperature as previously reported 1 . After incubation at 37°C for 3 hours, the phage-containing medium was removed. Cells were washed twice with phosphate buffered saline (PBS) and then incubated in complete DMEM medium (Gibco, USA). Twenty four hours posttransduction, cells were washed with PBS, harvested and centrifuged at 2655 × g for 15 minutes. DNA was harvested from pellets after incubation for 5 minutes at 98°C in lysis solution (Tris pH8 50mM, SDS 2%, Triton-X100 5%) and the harvested DNA was used for PCR. Primers: Up-R 5′CCGCTCGAGACTAGTGCAAAAGC-GAGCCTGGTAAATAAATATG3′; Up-D 5′GGAATTCCATAT-GCTCGTTGAGGCATATGAAAATCAGAC3′. The reaction was run in a Eppendorf Termocycler at an initial 92°C for 120 seconds and then at 92°C for 20 seconds and 60°C for 20 seconds and 72°C for 120 seconds for 35 cycles using primers giving a fragment of 1310 bp on the upstream region of gfp gene into the bacteriophage genome.

Bacteriophage induction
The C600ΔTOX:GFP strain was grown in Luria Broth (LB) plus 10 mM CaCl 2 and chloramphenicol (Sigma) (15 μg/ml final concentration) overnight (ON) at 37°C under agitation. The ON culture was diluted to OD 600nm = 0.1 in LB plus 10 mM CaCl 2 and chloramphenicol (Sigma) (15 μg/ml final concentration). Induction was carried out by adding ciprofloxacin to a final concentration of 40 ng/ml 8 . Bacteria were incubated for 6 hours at 37°C under agitation. Cultures were then centrifuged at 5000 rpm for 15 minutes. The bacteriophage-containing supernatant was filtered with 0.2 μm filters and kept at 4°C until the titration assay was performed.
Titration assay E. coli strain Y1090 (ATCC 37197) was grown in LB plus ampicillin ON at 37°C under agitation. The ON culture was diluted 1:100 in LB plus ampicillin and incubated for 2 hours at 37°C under agitation. At the end of the incubation, 500 μl samples of E. coli Y1090 were incubated with 5, 50 and 100 μl of a suspension containing bacteriophages for 30 minutes at room temperature. At the end of this incubation, 3 ml of Top Agar (Tryptone 1%; NaCl 0.5%; GPF dissemination assay (IVIS) Two-month old BALB/c mice were used to infect orally with C600φΔTOX:GFP. Bacteriophage induction in vivo was performed with ciprofloxacin as described immediately below. After 2 hours of induction with ciprofloxacin, 100 μl of chitosan solution at a concentration of 5 mg/ml was administered orally to the mice and GFP dissemination by IVIS was analyzed.
In Vivo Imaging System (IVIS) This time, we used two-month old BALB/c mice. An ON culture of C600:φΔTOX-GFP was used to infect them. The ON culture was centrifuged at 14000 rpm for 15 minutes at 4°C. The pellet was washed with PBS and centrifuged again at 14000 rpm for 15 minutes at 4°C. The pellet was resuspended in a solution of 20% sucrose to have a concentration of 1 × 10 9 CFU/mouse. Mice were inoculated orally with strain C600:φΔTOX-GFP and in vivo bacteriophage excision was induced following the procedures described by Zhang and collaborators 8 . The mice were sacrificed with CO 2 inhalation 24 hours after bacterial inoculation. Blood, spleens, kidneys, lungs, brains, intestines, hearts and livers were harvested by surgical removal and kept in PBS solution and evaluated for GFP expression using the IVIS system. To determine the effects of chitosan in vivo, the mice received 100 μl of a chitosan solution at a concentration of 5 mg/ml.

Statistical analysis
Statistical significance between treatments and controls was analyzed using the Prism 5.0 software (GraphPad Software), and the P value is indicated by asterisks in the figures.
All other data correspond to the means ± standard errors of the means (SEM) for individual mice. Statistical differences were determined using the one-way analysis of variance (ANOVA).

Results
Induction of φΔTOX:GFP by ciprofloxacin and anti-phage effects of chitosan Bacteriophage lytic induction was triggered in E. coli C600ΔTOX: GFP using ciprofloxacin 8 . We observed a significant decrease in the optical density of the bacterial culture after addition of the antibiotic and the release of phages into the culture supernatant (Figure 1, panel A and B). The bacteriophage titer was analyzed at different time points and a significant increase was observed after induction ( Figure 1, panel B). The effect of chitosan as an anti-bacteriophage agent was also examined. To this aim, we added chitosan at a final concentration of 5 mg/ml to the bacterial culture 2 or 4 hours post-induction, and we observed the complete inactivation of the φΔTOX:GFP, without measurable toxic effects to the bacterial strain ( Figure 1, panels A and B).

Transduction of mammalian cells with φΔTOX:GFP
We previously reported the capacity of φΔTOX:GFP to transduce macrophages in vitro 1 . To further evaluate the ability of chitosan to inhibit bacteriophage transduction of mammalian cells, BHK cells were transduced for 3 hours with φΔTOX:GFP, φΔTOX:GFP plus chitosan or φΔTOX:GFP treated with DNAse. Addition of DNAse to the bacteriophage sample would preclude any free DNA in the Agar 0.7%) plus CaCl 2 (10 mM final concentration) was added, and plated on LB-Amp agar plates. Plates were incubated at 37°C and lysis plaques were visually counted.
Bacteriophage inactivation assay φΔTOX:GFP was incubated with 5 mg/ml of a chitosan (Sigma 448877) solution in phosphate buffer 10 mM, at pH = 7 for 10 minutes at room temperature, and bacteriophage titers were measured as described in titration assay section.
Chitosan was also used in the bacteriophage induction assay described above. Chitosan was added 2 and 4 hours post-induction and bacteriophage titers were analyzed at 6 hours post-induction.
Mice BALB/c mice were bred in-house at the animal facility of the Microbiology Department of the São Paulo University, Brazil. The experimental protocol of this study followed the ethical principles for animal experimentation adopted by the Brazilian College of Animal Experimentation (COBEA) and was approved by the Ethics Committee on Animal Experiments of the Institute of Biomedical Sciences (Protocol number 106), University of São Paulo, in accordance with the principles set forth in the Guide for the Care and Use of Laboratory Animals (National Institutes of Health, 1985).
Male mice aged 6 weeks (18 to 20 g) were used for the In Vivo Imaging System (IVIS). Immature male and female DBA-2 mice (17-21 days of age, approximately 8-11 g body weight) were used immediately after weaning for the infection assays with EDL933W strain (n = 4). Mice were maintained under a 12-h light-dark cycle at 22 ± 2°C and fed a standard diet and water ad libitum.

EHEC infection
Immature male and female DBA-2 mice (17-21 days of age, approximately 8-11 g body weight) were used immediately after weaning for the infection assays (n = 4).
E. coli EDL933W (ATCC 43895) was used for infection of mice following the protocol previously reported by Brando and collaborators 8 . Briefly, E. coli EDL933W was grown in Tryptic Soy Broth (TSB, DIFCO, BD) ON at 37°C. The ON culture was centrifuged at 14000 rpm for 15 minutes and the bacterial pellets were washed twice in PBS. Pellets were resuspended to have a final concentration of 3 × 10 12 CFU/100 μl per mouse.
The bacterial suspension was delivered directly into the stomach of mice after 8 hours of food starvation, via a 5-French paediatric feeding tube. After 4 hours of ingesting the bacterial suspension, mice were given food and water. Control animals received 100 μl of sterile PBS. Survival was observed for one week. Both groups were composed by 4 animals.

Effect of chitosan in vivo
To analyze the effect of chitosan in vivo, immature male and female DBA-2 were infected as described previously and treated with 100 μl of a chitosan solution at a concentration of 5 mg/ml, orally administered 2 hours after infection. Survival was observed for one week. action of chitosan on bacteriophages. Bacteriophage DNA was also detected in cells transduced with φΔTOX:GFP treated with DNAse ( Figure 2).

GFP detection in vivo
To demonstrate the in vivo behavior of bacteriophages, mice were infected with the lysogenic E. coli C600ΔTOX:GFP strain, followed by oral administration of ciprofloxacin 1 hour or 2 hours later. In order to evaluate the effect of chitosan in vivo, a group of mice was administered with chitosan 2 hours post-induction and a control group of uninfected mice was evaluated for auto-fluorescence background control in each organ. Twenty four hours after infection, organs were harvested and examined using the IVIS. As shown in Figure 3, GFP was detected in the intestine, liver and, to a lesser extent, kidney of mice orally infected and treated with ciprofloxacin. Remarkably, the addition of chitosan 2 hours after infection caused a sharp decrease in GFP detection in organs of mice orally infected with the E. coli strain C600ΔTOX:GFP (Figure 3, panels A and B), indirectly indicating reduction of bacteriophages in the cells, GFP release and dissemination. Moreover, viable phages were detected via the lysis plaque assay in intestine homogenates and blood samples of infected mice, in which bacteriophages were induced by ciprofloxacin (data not shown).
Effect of chitosan on the mortality of mice orally inoculated with the EDL933W strain In order to evaluate the in vivo effect of chitosan during the infection process, mice were orally challenged with a wild-type EDL933W strain, based on the model described by Brando and collaborators 9 . Another mouse group was also treated with chitosan, administered orally 2 hours post-infection, and survival was followed for one bacterial lysates prior to the transduction of cells. Untreated cells were used as a control. As shown in Figure 2, the bacteriophage DNA was detected by PCR in mammalian cells, showing the capacity of the virus to transduce this cell line. However, when BHK cells were transduced with bacteriophages pre-incubated with chitosan, no phage DNA was detected, confirming the inactivating   week. In this preliminary study, partial protection was observed in mice treated with chitosan, resulting in a delay in the death time ( Figure 4). Mice infected with EDL933W strain died at 72 hours post-infection, and mice infected followed by treatment with one dose of chitosan died at 168 hours after infection.

Discussion
Lambda bacteriophages are used in gene transfer and vaccine deliv-ery because of their capacity to transduce mammalian cells in vivo 10 . Tyler and collaborators recently showed that prophage induction is required for renal disease and lethality in the EHEC mouse model, suggesting that free bacteriophages encoding Stx may play a direct role in the disease 3 .
In previous reports, we have showed that the native phage promoter controlling Stx expression is active in eukaryotic cells as demonstrated both in vitro 1 and in vivo 2 . Based on these results and The fact that only partial protection was observed in vivo using chitosan may be due to its short half-life 11 . Our results may contribute to understand why only small numbers of bacteria are sufficient to induce HUS in humans. If bacteriophages are induced in the gastrointestinal tract, then replicate, infect bacteria in the intestine and transduce host cells, small numbers of bacteria should be enough to produce a Stx concentration sufficient to cause significant damage.
Altogether, these findings suggest a paradigm change on the role of bacteriophages in STEC infections, indicating they may be responsible for the development of disease rather than their bacterial host. Thus, prophylaxis and treatment of human bacterial infections carrying virulence factors on lysogenic bacteriophages may require targeting of the bacteriophages instead of, or as well as, the bacteria and toxins involved.

Competing interests
No competing interests were disclosed.

Grant information
This work was supported by PICT 2411 from the Agencia Nacional de Promoción Científica y Tecnológica, Argentina (to L.V.B) and Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP), Brazil (to L.C.S.F). LVB and PDG are members of the Research Career of CONICET (Consejo Nacional de Ciencia y Tecnología).
The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. the reports previously described, we sought to evaluate whether bacteriophages could be considered a target for treating STEC infections. To this aim, we measured GFP by the strain C600ΔTOX: GFP and the mortality of infected mice following bacteriophage induction, and in vivo inactivation upon chitosan treatment positive expression was analyzed. GFP was observed in liver, intestine and kidney by IVIS on mice in which the bacteriophage lytic phase was induced by ciprofloxacin following infection. Of particular relevance was the observation that chitosan exerted a direct inactivation effect on φΔTOX:GFP in vitro and drastically reduced the detection of fluorescence in mice orally infected with the C600ΔTOX:GFP strain. Bacteriophage transduction of mammalian cells was also inhibited after incubation with chitosan.
Our findings indicate that chitosan possesses strong anti-bacteriophage properties in vitro and in vivo. This positively charged polymeric polysaccharide has been reported to inhibit other bacteriophages and probably acts through electrostatic interactions with negatively charged capsid proteins 5 . Based on these effects we propose that chitosan may be a viable alternative for the treatment of STEC infections. Chitosan is already used in food and medicine, and it is harmless to humans, making it a cheap and safe option for this application.

Raúl Raya
Genetics and Molecular Biology, Centro de Referencia Para Lactobacilos (CERELA), San Miguel de Tucumán, Argentina The article written by Amorim et al. describes the anti-phage activity of chitosan on two variants (wild type and a derivative where the stx gene was replaced by the gfp gene) of the temperate Shiga-toxin producing phage EDL933W. The anti-phage activity was evaluated both in vivo and in vitro. The authors suggest that chitosan could be a viable alternative for the treatment of STEC infections.

Major:
Phage Induction/anti-phage effects of chitosan/Figure 1: It seems that chitosan not only sequesters free-phage particles, but also stimulates the growth of uninduced cells (see induced cells treated with chitosan 2 hours post-induction reached higher final OD values). So, does chitosan inhibit the induction process of the temperate phage? Or, does chitosan also adsorb/inactivate ciprofloxacin? Even though in the Materials and Methods a "Bacteriophage inactivation assay" is described, no data is presented. A dose-response curve should be presented, to determine the phage binding (inactivation) capacity of chitosan.

In vivo experiments:
If the authors suggest that Stx phages, rather than bacterial cells, may be responsible for the development of the STEC infections, why they did not use purified phage in the vivo experiments? Does chitosan adsorb/inactivate the Shiga-toxins? If so, may it explain the delayed response observed in Figure 4 ("EDL933W plus chitosan")?

Minors:
Abstract: Provide a reference after "… plasmids encoding Stx under control of the native bacterial ○ promoter." Change "E. coli K12 strain" to "E. coli C600 strain".

Materials and Methods:
Delete "… was generously provided by Dr. Luis Carlos de Souza Ferreira, LDV-USP, Brazil.", since Dr. Ferreira is one of the authors of the manuscript.
○ Check the sentence "…was generously provided by Dr. Alison Weiss"; it is repeated twice in the Materials and Methods, and also in the Acknowledgments.

○
Please, indicate how DNA was harvested.

○
The sentence "Mice were orally challenged with a wild-type EDL933W" is not correct, since there was a direct delivery of bacterial cells into the stomach of mice.

I confirm that I have read this submission and believe that I have an appropriate level of expertise to confirm that it is of an acceptable scientific standard, however I have significant reservations, as outlined above.
Author Response 01 Aug 2014 Leticia Bentancor, Universidad Nacional de Quilmes, Buenos Aires, Argentina Phage Induction/anti-phage effects of chitosan/ Figure 1: The higher final OD values determined on the cells treated with chitosan 2 hours postinduction, versus the OD value determined on un-induced cells, is not statistically significant. However, we tested if chitosan is capable of inactivating ciprofloxacin. Ciprofloxacin and chitosan were incubated for 10 minutes at room temperature with chitosan at 5mg/ml. After pre-incubation, the mix was used to induce bacteriophage excision. To see a more significant effect on bacteriophage excision, the induction was incubated overnight. The OD value measured showed a non-significant difference between non-induced culture and induced culture with the antibiotic pre-incubated with chitosan (the experiment was performed in triplicate).
On the other hand, purified bacteriophages were incubated with chitosan, and bacteriophage inactivation was observed with a lysis plaques assay. Bacteriophages were incubated for 10 minutes at room temperature with chitosan at 5mg/ml. After incubation, bacteriophage inactivation was evaluated. The bacteriophage solution containing a titer of 4x10 3 pfu/ml was 100% inactivated after chitosan incubation. The assay was performed in triplicate. This result showed the capacity of chitosan to inactivate bacteriophage in vitro.
Also, we did a dose-response curve of chitosan. We used 5mg/ml, 2.5mg/ml, and 1mg/ml of chitosan on purified bacteriophage solution. To evaluate it, the bacteriophage solution was incubated for 10 minutes at room temperature with the different doses of Chitosan and the bacteriophage inactivation was evaluated with a lysis plaques assay. Chitosan at 1mg/ml lost the inactivation activity on the bacteriophages. Chitosan at 5mg/ml and 2.5 mg/ml showed a 100% efficiency on bacteriophage inactivation, however, 1mg/ml of chitosan showed a loss of inactivation, showing between 5-10% of bacteriophage inactivation. This experiment was performed in triplicate.

In vivo experiments:
Chitosan was analyzed in vitro and in vivo on fDTOX:GFP. Inactivation of fDTOX:GFP was observed in vitro with a lysis plaques assay. On the other hand, a decrease of GFP was observed in vivo by IVIS. These results shown that chitosan has the capacity to inactive bacteriophages in absence of Shiga-toxins. A direct action of chitosan on Shiga toxin was not evaluated in this work since we do not have purified Stx2 for such experiments. The authors are working on murine infection with fStx2, but the results obtained will be part of a new publication.
The change was made.
Delete the sentence "Two-month old BALB/c mice … and GFP dissemination by IVIS was analyzed". It is redundant.
We have two different mouse models. First, we have the model used to analyze GFP dissemination in which we used two months old mice. Second, we have the model used to analyze protection effect in which we used immature mice. For this reason we clarify the model every time. Let me know if you consider that we need to delete the sentence "Two-month old BALB/c mice … and GFP dissemination by IVIS was analyzed". ○ Results: Figure 1B: Should "Bacteriophage/ml" be "PFU/ml"? Why phage titers are so low?
Bacteriophage/ml was changed to PFU/ml as reviewer suggested. See below. It is true that bacteriophage titers are low. An optimization for bacteriophage purification was done to obtain a higher titer of bacteriophage. The antibiotics used to induce C600DTOX:GFP was selected as an alternative for mitomycin C. The efficiency of bacteriophage induction is strain dependent. Zhang and collaborators reported a titer equal to 1,3x10 5 pfu/ml using ciprofloxacin but they used pathogenic strain E. coli O157:H7. We also observed a higher titer inducing the EDL933W strain, for this reason we suppose that the low titer observed is dependent on the strain used. ○ Figure 1A: Change "hs" for "hours" or "h".
The change was made. The changes were made. ○ Delete "viable" in "viable phages". Were phages transduced or adsorbed to mammalian cells?
We deleted "viable" in "viable phages". In this context, phages purified from tissue were detected by lysis plaque assay.

○
The sentence "Mice were orally challenged with a wild-type EDL933W" is not correct, since there was a direct delivery of bacterial cells into the stomach of mice.
The sentence "Mice were orally challenged with a wild-type EDL933W" was change by "Mice were intragastrically infected with a wild-type EDL933W". 1.
The mouse experiments were performed with too small a number of mice. 2. Figure 1A of growth curves is missing a crucial control. What happens to E. coli C600 under the ciprofloxacin treatment?
3. Figure 1B: The lack of the 4 hr column in chitosan 4h post-induction does not seem logical to me. There should be a ca 6000 PFU/ml column similar to that one in the induced 4hr sample. This discrepancy should be explained. Figure 2: the PCR experiment does not provide evidence of transduction. The definition of transduction is that DNA moves from one cell to another. PCR detects the phage DNA either free in the cell cytoplasm or packed in endocytosed phage particles. Therefore, the authors need to demonstrate that infective phage particles disappear from infected cells. The experiment also does not exclude the possibility that phage particles are just adsorbed on the cell surface.

5.
The experiment reported in figure 3 should also include bacterial counts from the organs as it is very likely that live E. coli bacteria, after a massive dose of 10 13 bacteria per mouse, end up in the organs. Therefore the authors should demonstrate that the GFP response is not from bacteria infected by the GFP-phages.

6.
The Figure 4 experiment was performed with only 4 mice. Such an experiment should not be shown in a publication. In addition, different chitosan doses should be tested here also.

Minor
Introduction, paragraph 3: This statement on the low number of bacteria during infection should be backed up with a reference. 1.

Materials and Methods:
Dr Alison Weiss is thanked twice for same strain. One time should be enough. In addition, in the acknowledgements she is thanked a third time. The bacterial strain designation in the latter is given differently than elsewhere in the text.

Materials and Methods:
In vitro evaluation of the capacity of Bacteriophage 933W to transduce Vero cells. E. coli EDL933W (ATCC 43895) was used to purify fStx2. E. coli EDL933W strain was grown in Luria Broth (LB) overnight (ON) at 37°C under agitation. The ON culture was diluted to OD600nm = 0.1 in LB. Induction was carried out by adding ciprofloxacin to a final concentration of 40 ng/ml 8 in main text . Bacteria were incubated for 6 hours at 37°C under agitation. Cultures were then centrifuged at 5000 rpm for 15 minutes. The bacteriophage-containing supernatant was filtered with 0.2 mm filters, precipitated and purified. Briefly, supernatant was incubated on ice with a PEG-8000/NaCl solution for 30 minutes. After that, the solution containing bacteriophages was centrifugated and washed. The pellet was resuspended in STE buffer (1ml of Tris pH8, 0,2ml of 0,5M EDTA, 2ml of 5M NaCl, water up to 100ml). Phages at a multiplicity of infection (M.O.I) equal to 1 were added to Vero cells. Transduction of Vero cells was enhanced by centrifugation at 1000 x g for 10 min at room temperature. After 24 hours post transduction, cells were examined by microscopy using Nikon Eclipse TE2000 (NIS-Elements imaging software) equipped with a CCD camera. Dilutions of fStx2 were made to demonstrate specificity. Vero cells were incubated with purified Stx2 as positive control. E. coli C600DTOX:GFP is not an invasive bacteria. Also, E. coli O157:H7 is a non invasive strain; for this reason we do not check for bacteria in organs. Bacteria were checked only on lungs samples, just to see if the inoculation was right. Bacteria were not detected in lungs. The dose used was selected after a previous experiment in which we evaluated the sensibility of IVIS in our system. GFP is not the best fluorescent protein for IVIS system; so, we needed to use a high dose of bacteria. As we described in this work, bacteriophages were detected by lysis plaques assay in intestine homogenates and blood samples of infected mice. It is important to do a highlight in the case of intestine sample, as it is very difficult to find E. coli C600DTOX:GFP. First, because the huge amount of bacteria present in the sample, and also, because the bacteria lysis induced by bacteriophage excision.

3.
The experiment was shown as a preliminary result and this work it is a short communication. The model used has some experimental problems for the ages of mice used. The experiment was started using 6 mice per group, but some mice died after inoculation and not for the infection. For this reason, we had shown only 4 mice per group. We repeat the experiment, and again we have the same problem, 4.