Draft Potential pathogenicity of Aeromonas hydrophila complex strains isolated from clinical , food and environmental sources

Complete List of Authors: Albarral, Vicenta; Universitat de Barcelona, Microbiologia i Parasitologia Sanitàries Sanglas, Ariadna; Universitat de Barcelona, Microbiologia i Parasitologia Sanitàries Palau, Montserrat; Universitat de Barcelona, Microbiologia i Parasitologia Sanitàries Miñana-Galbis, David; Universitat de Barcelona, Microbiologia i Parasitologia Sanitàries Fusté, M. Carmen; Universitat de Barcelona, Microbiologia i Parasitologia Sanitàries; Universitat de Barcelona, Institut de Recerca de la Biodiversitat (IRBio)

Gammaproteobacteria (Martin-Carnahan and Joseph 2005).Aeromonas are autochthonous inhabitants of aquatic environments, including chlorinated and polluted waters, although they can also be isolated from a wide variety of environmental and clinical sources.They cause infections in vertebrates and invertebrates, such as frogs, birds, various fish species and domestic animals.Aeromonas is considered to be an emerging pathogen in humans (Janda and Abbott 1998), producing intestinal and extra-intestinal diseases.Most of the clinical isolates correspond to A. hydrophila, A. caviae, and A. veronii bv.Sobria, which are described as the causative agents of wound infections, septicaemia and meningitis in immunocompromised people, and diarrhoea and dysenteric infections in the elderly and children (Kirov et al. 2004).
Two main types of enterotoxins have been described in this genus, cytotonic and cytotoxic.Cytotoxic enterotoxins produce severe damage in the small intestine mucous membrane.Aer toxin (also named aerolysin or Act), the main enterotoxin described in Aeromonas, shows haemolytic, cytotoxic and enterotoxic activities, is thermolabile, inhibits phagocytosis and increases the TNF-α and interleukin IL-1β intracellular levels in macrophages.The Aeromonas cytotonic enterotoxins, such as Alt (thermolabile) and Ast (thermostable), do not destroy the intestinal mucous membrane.Instead, they increase the prostaglandin intracellular levels in the enterocytes and activate the adenylate cyclase, thereby increasing cAMP synthesis and triggering a fluid secretion, which causes liquid diarrhoea (Aguilera-Arreola et al. 2007;Chang et al. 2008;Janda and Abbott 2010;Ottaviani et al. 2011).Some D r a f t D r a f t

Bacterial strains
A total of 127 strains belonging to the "Aeromonas hydrophila complex" (26 A. bestiarum, 15 A. dhakensis, 29 A. hydrophila, 11 A. piscicola, 6 A. popoffii and 40 A. salmonicida), isolated from fresh waters, marine bivalve molluscs and clinical samples, were analysed in this study (Table S1).Strains were obtained from different culture collections, our own collection or donated by different researchers.These strains were previously identified in a taxonomic study based on multilocus sequence analysis of cpn60, dnaJ, gyrB and rpoD genes (Fusté et al. 2012;Albarral 2013).Strains were grown on TSA plates with 1% NaCl, at 25-30ºC.Cultures were maintained in TSA (1% NaCl) tubes.For a long-term preservation, strains were kept at -80ºC using glycerol 20% (v/v).

Enzymatic activities
Casein hydrolysis, DNase, hemolysis of sheep blood and starch hydrolysis tests were performed as previously described (Reddy et al. 2007).Gelatinase activity was determined on TSA supplemented with 0.4% gelatin (Pickett et al. 1991;Miñana-Galbis et al. 2004).Elastase production was assessed according to Popoff and Lallier (1984).Lecithinase activity was tested following the recommendations of Esselmann and Liu (1961).All media used contained 1% (w/v) NaCl (Holt et al., 1994).Incubations were performed at 30ºC, with the exception of the psychrotrophic strains, which were grown at 25ºC.

Antibiotic susceptibility tests
Bacterial susceptibility to different antibiotics was determined using the disc diffusion method (Murray and Baron 2003).Antimicrobial agents tested included 8 β-lactams, 4 aminoglycosides and 8 antibiotics of other groups.Antibiotics and the concentrations tested are shown in Table S2.

Detection of virulence genes
The presence of genes encoding exotoxins, namely aerolysin Aer, heat-labile enterotoxin Alt, heatstable enterotoxin Ast, and the T3SS inner membrane component (ascV) was determined by D r a f t 6 polymerase chain reaction (PCR).Genomic DNA extraction was carried out using a commercial kit (Realpure ® genomic DNA extraction kit, Durviz).Table 1 shows the primers used for the PCR.
PCR amplification for aer was carried out in a total volume of 50 µl containing 400 ng of genomic DNA as a template, PCR buffer (500 mM KCl, 150 mM Tris/HCl pH 8.0, 15 mM MgCl 2 ), 0.3 mM dNTPs, 2.5 U AmpliTaq Gold ® DNA polymerase and 1 µM of each primer (aer-f-d and aer-r-d).The reaction mixtures were subjected to the following thermal cycling program: denaturizing at 95°C for 5 min, 35 cycles of 94°C for 1 min, 56°C for 1 min and 72 °C for 1 min, and a final extension step at 72°C for 10 min.In the cases where the gene was not detected, a two-stage nested PCR analysis was performed in order to increase its sensitivity and specificity.In the first-round of PCR, a fragment containing 1161 pb was amplified using the F-Ah-a and aer-r-d primers, with the same conditions and temperature cycles as mentioned above.For the second-round of PCR, primers aer-f-d and aer-r-d, which targeted an internal fragment of the first-round PCR product, were employed using 50 ng of PCR product as a template.
PCR for the alt gene was performed in a final volume of 50 µl containing 200 ng of genomic DNA as a template, PCR buffer (500 mM KCl, 150 mM Tris/HCl pH 8.0, 15 mM MgCl 2 ), 0.8 mM dNTPs, 1.25 U AmpliTaq Gold ® DNA polymerase and 0.2 µM of each primer (altF2 and altR2).The reaction mixtures were subjected to the following thermal cycling program: denaturizing at 95 °C for 5 min, 35 cycles of 94°C for 45 s, 52°C for 1 min and 72°C for 1 min, and a final extension step at 72°C for 10 min.
When the gene was not detected, amplification was performed using another pair of primers (altF1 and altR1) with the same thermal cycling program.
PCR amplification of the ast gene was carried out in a total volume of 50 µl containing 450 ng of genomic DNA as a template, PCR buffer (500 mM KCl, 150 mM Tris/HCl, pH 8.0, 15 mM MgCl 2 ), 0.4 mM dNTPs, 1.25 U AmpliTaq Gold ® DNA polymerase and 0.8 µM of each primer.The reaction mixtures were subjected to the following thermal cycling program: denaturizing at 95°C for 5 min, 35 cycles of 94°C for 1 min, 56°C for 1 min and 72°C for 1 min, and a final extension step at 72 °C for 10 min.

D r a f t
A PCR reaction for ascV was conducted as described by Martino et al. (2011) in the 59 strains selected for their adherence capacity and cytopathic effect properties.An alternative PCR was performed to detect ascV in negative strains with a new primer (ascV-R+).The reactions were carried out in a final volume of 50 µl containing 250 ng of genomic DNA as a template, PCR buffer (500 mM KCl, 150 mM Tris/HCl pH 8.0, 15 mM MgCl 2 ), 1 mM dNTPs, 2.5 U AmpliTaq Gold ® DNA polymerase and 1 µM of each primer (ascV-F and ascV-R+).The reaction mixtures were subjected to the following thermal cycling program: denaturizing at 95 °C for 5 min, 35 cycles of 95°C, 54°C and 72°C for 1 min, and a final extension step at 72°C for 5 min.

Adhesion capacity and cytopathic effect
According to the previous results obtained with the assayed virulence factors (enzyme activities, profiles of antibiotic sensitivity and detection of aer, alt and ast genes), we selected 59 strains from a total of 127 (Table S1) and determined their adherence capacity and cytopathic effect on the cellular line Caco-2.In this selection we included strains with a variable range of virulence.
Strains were cultured on double concentrate tryptic soy broth (TSB) without dextrose and incubated at 30⁰C overnight.The strains were then washed twice with Dulbecco's Phosphate Buffered Saline (PBS-D) (Invitrogen 14190-094).Bacterial cell growth was estimated by viable counting and optical density, until reaching a biomass of 1x10 8 CFU ml -1 (OD = 0.35 at 580 nm).Our aim was to achieve a Caco-2:bacterial cell ratio of 1:10 (Couto et al. 2007).
A cell monolayer was grown in 75 cm 2 flasks (Cultek 430641) containing DMEM (Dulbecco's Modified Eagle Medium high glucose, glutaMax) (Invitrogen 31966-021) supplemented with 10% FBS (heat inactivated fetal bovine serum) (Invitrogen 10500-064) and 1% penicillin-streptomycin (Invitrogen 15140-122) and incubated at 37⁰C and 5% CO 2 in a Nuaire incubator (NU-425-400E).The cell culture medium was changed every 48h (Freshney 2005).The Caco-2 monolayer was used at a semiconfluence of 80%.Thus, 1 ml of a Caco-2 cell suspension at a concentration of 3x10 5 cells ml -1 was grown on 13 mm diameter glass coverslips placed in 24-well tissue culture plates (Cultek 3524) in D r a f t 8 DMEM supplemented with 10% FBS and 1% penicillin-streptomycin and incubated in a 5% CO 2 atmosphere at 37⁰C.For the adhesion assays, the monolayers were used 24h after seeding.The monolayers were washed twice with PBS-D, inoculated in triplicate with 1ml of bacterial suspension containing 1x10 7 UFC ml -1 , and incubated for 90 min at 37⁰C and 5% CO 2 (Guimarães et al. 2002;Couto et al. 2007).After two washes with PBS-D to remove non adherent bacteria on the cell monolayers, cell cultures were fixed with methanol for 2 min and stained with Giemsa solution (Merck) (1:10 with tap water) and washed twice with water.The coverslips were removed from the wells, mounted on glass slides, and examined by optical microscopy.

Results and discussion
The majority of published studies on virulence factors in the genus Aeromonas are based on the detection of strains isolated from diverse backgrounds (Kingombe et al. 1999;Chacón et al. 2003;Kingombe et al. 2004;Ottaviani et al. 2011), specific sources such as treated and untreated water for human consumption (Scoaris et al. 2008;Carvalho et al. 2012), food (Chang et al. 2008), fish and seafood (Nam and Joh 2007;Martino et al. 2011;Sreedharan et al. 2013), diarrheal stools (Kannan et al. 2001), etc.In this work we determined the presence of virulence factors in a set of 127 AHC strains of different origins (environmental, food and clinical) to verify if there were any differences among them and to determine their potential pathogenicity.
The determination of lipolytic activity is controversial, because different lipases can be detected depending on the medium used.According to Merino et al. (1999), Aeromonas can produce two types of phospholipases: phospholipase A1, which is cytotoxic and hemolytic, and whose activity is determined in tributyrin agar, and phospholipase C or lecithinase, whose activity is detected using egg yolk agar.Phospholipase C displays cytotoxic and hemolytic activities in fish but not in sheep blood.
We detected lecithinase activity in 89.06% of the AHC strains (84.6-100%, depending on the species), all the A. popoffii strains being positive (Table 2 and S1).In contrast, Scoaris et al. (2008) detected this activity in only one out of five A. hydrophila strains, also in egg yolk agar.However, studies that detect lipolytic activity using tributyrin agar (Soler et al. 2002;Castro-Escarpulli et al. 2003;Chacón et al. 2003) or TSA supplemented with Tween 80 (Beaz-Hidalgo et al. 2009;Carvalho et al. 2012) obtained very similar results to ours, reporting in all cases values higher than 90%.
Another virulence factor analysed in this work was DNase activity, which is usually determined for taxonomic purposes (Chacón et al. 2003).The high incidence of this activity detected in our study, 93.0% (88.5-100%, depending on the species), agrees with the results published by other authors: D r a f t 10 for example, Chacón et al. (2003) reported that 87% of A. hydrophila, A. bestiarum, A. salmonicida and A. popoffii strains tested positive, with the DNase-encoding gene detected in 91%.Some tables of key phenotypic tests for the differentiation of Aeromonas species include elastase activity (Miñana-Galbis et al. 2010;Figueras et al. 2011), but this is the first time that this test has been used in a study of virulence factors in this genus.Elastase activity is essential for the pathogenicity of A. hydrophila and is associated with its virulence in fish (Pemberton et al. 1997;Cascón et al. 2000;Janda and Abbott 2010).In our strains the results were variable: all the A. popoffii strains tested negative, while 54.5% -93.3% of the other AHC species were positive.Hemolytic toxin production is considered to be evidence of the pathogenic potential of Aeromonas.βhemolytic activity is usually due to aerolysins but also to enzymes with lipase activity.Discrete βhemolysis was found in 83.6% of the strains (72.4-100%, depending on the species), while αhemolysis was determined in only 2.34%.All the A. dhakensis and A. popoffii and most of the A. piscicola, A. bestiarum and A. salmonicida strains were hemolytic, while A. hydrophila showed the lowest hemolytic activity (Table 2).In contrast with our results, Huys et al. (1997) and Soler et al. (2002) described this activity in only 15% of the A. popoffii strains tested.Such discrepancies may be due to the blood type, media, and the incubation temperature used for detection.The results obtained with the other AHC species coincide with published data (Huys et al. 2002;Chacón et al. 2003;Martínez-Murcia et al. 2005;Martin-Carnahan and Joseph 2005;Martínez-Murcia et al. 2008;Beaz-Hidalgo et al. 2009), since more than 85% of the strains were β-hemolytic.

D r a f t 11
Sensitivity to amoxicillin-clavulanic acid, ampicillin and ticarcillin varied among the strains.A. dhakensis showed the highest β-lactam resistance, although 100% were sensitive to aztreonam, 67% to ceftriaxone, 53% to imipenem, and 6.7% to ticarcillin (see Table S2).
Fig. 2 shows strains sensitivity to aminoglycoside antibiotics.All the A. popoffii strains were sensitive to these antibiotics.The majority of the remaining species were sensitive to amikacin and gentamicin (73-100%) and resistant to streptomycin (97.5-100%), while sensitivity to tobramycin was variable.Fig. 3 shows the results obtained with the mixed antibiotic group.Most strains were sensitive to ciprofloxacin, chloramphenicol, fosfomycin, tetracycline and trimethoprim-sulfamethoxazole (87-100%), except A. dhakensis, which showed diverse sensitivity to chloramphenicol, tetracycline and trimethoprim-sulfamethoxazole (53-67%).Only A. salmonicida subsp.achromogenes LMG 14900 T and A. dhakensis LMG 19566 were sensitive to erythromycin.Variable results were obtained with colistin.
Besides helping to select the appropriate antibiotic for infection therapy, the pattern of sensitivity to antibiotics can also be used as a phenotypic trait to identify Aeromonas (Awan et al. 2009;Igbinosa et al. 2012).Of the 25 antibiotics used in this study, ciprofloxacin and fosfomycin were the most effective, with most of the strains analysed being sensitive (97.6% and 98.4%, respectively).On the contrary, the majority of the strains were resistant to ampicillin and erythromycin.The sensitivity pattern shown by A. dhakensis and A. popoffii was different compared with the other AHC species, independently of the origin of the strains (Figs.1-3).These results agree with those published in other studies (Huys et al. 1997;Martínez-Murcia et al. 2008;Beaz-Hidalgo et al. 2009;Janda and Abbott 2010).

Virulence genes
Detection of virulence genes is important to determine the pathogenic potential of the Aeromonas strains, since enteropathogenicity in Aeromonas has been attributed to the production of exoenzymes, exotoxins, and adhesins, although the exact mechanism associated with virulence has not been well established (Pemberton et al. 1997).The virulence gene encoding aerolysin, the most studied in D r a f t or cytolytic enterotoxin.Nowadays, some authors consider that both the act (corresponding to the cytolytic enterotoxin) and aerA gene (corresponding to aerolysin) encode the same protein, due to the high homology of their nucleotide sequences (Buckley and Howard 1999;Chopra and Houston 1999).
Another virulence factor is the T3SS used by Gram negative bacteria to export and deliver effector proteins into the cytosol of host cells (Sierra et al. 2010;Piqué et al. 2015).In this study, the T3SSencoding gene (ascV) was detected in 44% of the strains assayed overall, with variability among the species (Table 2, Table S3): 77% A. hydrophila, 53% A. bestiarum, 33% A. dhakensis, 20% A. salmonicida, 100% A. piscicola (two strains) and 0% A. popoffii (three strains).Similar results (45% Using PCR to detect these four virulence genes has limitations, since this technique can give false negatives, as observed in this work.However, we obtained more positive results than other authors, probably due to applying more than one strategy, such as the use of different pairs of primers, the addition of additives like DMSO to the PCR, or carrying out nested PCRs in cases with negative or equivocal results.

Adhesion capacity and cytopathic effect
Different authors have performed adherence, cytotoxic and viability assays with Hep-2, Vero, HT29 and Caco-2 cell lines inoculated with Aeromonas strains (Kirov et al. 1995;Thornley et al. 1996;Guimarães et al. 2002;Castro-Escarpulli et al. 2003;Harf-Monteil et al. 2004;Snowden et al. 2006;Couto et al. 2007;Nam and Joh, 2007;Ottaviani et al. 2011).In this study, the adhesion capacity and cytopathic effect were analysed with Caco-2 cells, using 59 strains representative of all AHC species (Table S1), with different pathogenic potential.This cell line was chosen because, being derived from a colon carcinoma, it exhibits morphological and functional characteristics of the small intestinal cells.
Caco-2 cells are recognized as a substitute for human intestinal epithelial cells and are used to study the adhesion of enteric pathogens (Nishikawa et al. 1994).
Adhesion capacity was considered as positive when more than 10 bacteria were found adhered on the surface of each Caco-2 cell (Ottaviani et al. 2011).Most strains (76.3-81.4%)exhibited adherence, with variable prevalence among the AHC species (Table 2): 100% A. dhakensis and A. piscicola, 84% A. hydrophila, 80% A. bestiarum, 65% A. salmonicida and 33.3% A. popoffii strains.It was difficult to compare our results with those of other authors (Thornley et al. 1996;Castro-Escarpulli et al. 2003 andCouto et al. 2007), when the criteria establishing adherence capacity was not defined.However, we found a higher number of strains with positive results than other studies using the same criteria (>10 bacteria per host cell): around 50% of Aeromonas strains were reported to have adhesive ability by Kirov et al. (1995) and Snowden et al. (2006), and 33% by Ottaviani et al. (2007).Some authors differentiate two patterns of adhesion, diffuse and aggregative, depending on the distribution of the bacteria around the cell (Nishikawa et al. 1994;Thornley et al. 1996;Castro-Escarpulli et al. 2003; D r a f t 14 Couto et al. 2007).Cell culture analysis using optical microscopy showed that the same bacterial strain can adhere around Caco-2 cells in both a diffuse and aggregative way, depending on the bacterial cell density, thus questioning the differentiation between these two adherence patterns.
To evaluate the cytopathic effect, the following parameters were taken into consideration: destruction of the cell line monolayer, nuclear pyknosis, cell rounding and presence of vacuoles in the cytoplasm (Martins et al. 2007).An example of these cytopathic activities is shown in Fig. 4 and Fig. S1.
Although several authors observed complete disruption of the cell monolayer after 120 min infection with several Aeromonas strains (Guimarães et al. 2002;Nam and Joh, 2007), it is noteworthy that in our study most Aeromonas strains (66%) caused complete lysis of the host cells in 90 min (Table 2).
For this reason, nuclear pyknosis, but not cell rounding or cytoplasm vacuolization, was observed (Table S3).Although A. salmonicida strains showed a lower percentage of adherence to the cell line (65%) than the majority of the species studied, probably because the temperature used for infection (37ºC) was higher than the optimum growth temperature for some A. salmonicida strains, their cytopathic effect was higher (95% of strains).No differences were observed between strains of different origin, as observed by Couto et al. (2007), in contrast with Ottaviani et al. (2011), who found a higher adherence in environmental compared with clinical and food strains, although these authors used a different cell line (Hep-2).Additionally, with the possible exception of A. hydrophila, in this study no correlation was observed between ascV detection and adherence and cytopathic effect, in contrast with Krzymińska et al. (2012) who showed a strong association between this gene and the cell-contact cytotoxicity.

Conclusions
In our study a very high prevalence of virulence-related enzymatic activities was detected in all the species belonging to the Aeromonas phenotypic group (AHC), except for that of elastase.Virulence genes aerA, alt and ast were detected in all strains of the AHC species, with the exception of four aerA negative A. hydrophila strains.However, ascV was less prevalent (44%), although it was detected in all AHC species, except A. popoffii.Most strains exhibited an adhesion capacity and/or cytopathic effect on Caco-2 cells, except A. popoffii.No differences in virulence factors were found among our strains with respect to the source of isolation (environmental, food and clinical).In conclusion, the Results are expressed as percentages; n, number of strains.

D r a f t 25
Stanier 1943 belongs to the family Aeromonadaceae within the class positive) were obtained by Krzymińska et al. (2012) but a lower prevalence was reported by Martino et al. (2011) (24%) and Castelo-Branco et al. (2015) (17%).These discrepancies could be explained by the number of strains and the Aeromonas species analysed in the different studies.No differences were observed between clinical and environmental strains in the present study in concordance with

Figure 4
Figure 4 Light micrographs of Caco-2 monolayer cells infected with Aeromonas strains.A) uninfected cells; B) A. bestiarum CECT 4227 T showing nuclear pyknosis; C) A. bestiarum H73 showing adhesion to and destruction of the monolayer cell line; D) A. hydrophila CECT 5174 showing rounding and cytoplasmatic vacuoles.

Table 2 .
Enzymatic activities, detection of virulence genes, adhesion capacity and cytopathic effect.