CS21 positive multidrug-resistant ETEC clinical isolates from children with diarrhea are associated with self-aggregation, and adherence

Background: Enterotoxigenic Escherichia coli (ETEC) colonize the human intestinal mucosa using pili and non-pili colonization factors (CFs). CS21 (also designated Longus) is one of the most prevalent CFs encoded by a 14 kb lng DNA cluster located in a virulence plasmid of ETEC; yet limited information is available on the prevalence of CS21 positive ETEC isolates in different countries. The aim of this study was to evaluate the prevalence of CS21 among ETEC clinical isolates from Mexican and Bangladeshi children under 5 years old with diarrhea and to determine the phenotypic and genotypic features of these isolates. Methods: ETEC clinical isolates positive to lngA gene were characterized by genotype, multidrug-resistance, self-aggregation, biofilm formation, and adherence to HT-29 cell line. Results: A collection of 303 E. coli clinical isolates were analyzed, the 81.51% (247/303) were identified as ETEC, 30.76% (76/247) were st+/lt+, and 25.10% (62/247) were positive for the lngA gene. Among the lngA+ ETECs identified, 50% of isolates (31/62) were positive for LngA protein. The most frequent serotype was O128ac:H12 found in 19.35% (12/62) of lngA+ ETEC studied. Multidrug-resistance (MDR) lngA+ ETEC isolates was identified in 65% (39/60), self-aggregation in 48.38% (30/62), and biofilm formation in 83.87% (52/62). ETEC lngA+ isolates were able to adhere to HT-29 cells at different levels. Two lngA isogenic mutants were constructed in the ETEC E9034A and ETEC73332 clinical isolate, showing a 77% and 98% reduction in adherence, respectively with respect to the wild type. Conclusion: ETEC isolates that have the lngA gene showed features associated with self-aggregation, and adherence to HT-29 cells, important characteristics in the human gut colonization process and pathogenesis.


INTRODUCTION
Bacterial adherence is the first and most important step in bacterial infection. Diverse pathogenic mechanisms allow the bacteria to adhere to host tissues and aggregate at the infection site with the subsequent colonization and dissemination to other anatomical sites within the host (Nataro and Kaper, 1998;Craig et al., 2004). Enterotoxigenic Escherichia coli (ETEC), a leading traveler's diarrhea etiologic agent, also cause 300,000-500,000 deaths annually in children under the age of 5 years living in developing countries (WHO, 2006). Profuse watery diarrhea is the disease's hallmark, and similarly to cholera, it is mediated by the secretion of a cholera-like heat-labile (LT) and/or heat-stable toxins (ST) (Nataro and Kaper, 1998). These enterotoxins are believed to be encoded by large virulence plasmids and co-regulated with pili and non-pili surface proteins (Brinton, 1965;Beachey, 1981). Colonization of the small bowel mucosa by ETEC isolates is mediated by a broad variety of fimbrial and non-fimbrial surface structures called colonization factors (CFs). CFs includes colonization factor antigens (CFAs), coli surface antigens (CSs), and putative colonization factors (PCFs) (Gaastra and Svennerholm, 1996;Nataro and Kaper, 1998). More than 25 CFs have been described in human ETEC isolates (Qadri et al., 2005) and they are strongly associated with ETEC infection. Evidence of the role of CFs in pathogenesis include a study reporting that pili-less ETECs rarely induce diarrhea (Ahren and Svennerholm, 1985). CFs are highly prevalent in different geographic regions of the world and recognized by specific receptors not yet described (Gaastra and Svennerholm, 1996;Nataro and Kaper, 1998;Isidean et al., 2011). Epidemiological studies have shown that CFs produced by ETEC have immunogenic properties, and also have a protective immunity, which can be achieved through multiple infections by the host (Ahren and Svennerholm, 1985;Nataro and Kaper, 1998).
Human ETEC isolates express CS21 (Longus), a type IV pilus (T4P), with >20 μm in length. CS21 is expressed at 37 • C when grown on blood agar plates and Pleuropneumoniae-Like Organisms (PPLO) agar (Giron et al., 1994;Mazariego-Espinosa et al., 2010). CS21 is composed of a single repeating structural protein of 22-kDa called LngA. The lngA gene, encoding the LngA major subunit, is contained in a large virulence plasmid of 90 kb (Giron et al., 1994). The N-terminal amino acid sequence of the LngA protein shares homology with the CofA (CFA/III) pilin subunit ETEC, TcpA of V. cholerae, and BfpA of enteropathogenic E. coli (Taylor et al., 1987;Girón et al., 1991Girón et al., , 1997Giron et al., 1994). The gene cluster organization and amino acid sequence identity, suggest that these genetically diverse T4P may have a common ancestor (Taniguchi et al., 2001).
CS21 plays an important role in the adherence to intestinal epithelial cells and T84 cells) and display twitching motility activity; likewise, it was shown that CS21 assembly is also influenced by nutritional growth conditions that phenotypically affect the expression of LngA (Mazariego-Espinosa et al., 2010). In addition, CS21 is associated with bacterial selfaggregation, protection against environmental stress, biofilm formation, and adherence to primary intestinal epithelial cells and recently (Guevara et al., 2013) demonstrated the role of CS21 in the pathogenesis of ETEC in vivo using a neonatal mice challenge infection model (Clavijo et al., 2010;Guevara et al., 2013). Human-ETEC isolates encoding CFAs isolates from different geographic regions (Argentina, Bangladesh, Chile, Brazil, Egypt, and Mexico), have shown a wide distribution of the lngA gene (Girón et al., 1995a;Gutierrez-Cazarez et al., 2000;Pichel et al., 2002). In addition to CS21, ETEC has another T4P called CFA/III for which cofA is the major subunit, both are paralogous in nature and belong to evolutionarily distinct types of fimbriae (Gomez-Duarte et al., 2007); however, CFA/III has lower prevalence than CS21 (Honda et al., 1984(Honda et al., , 1985Isidean et al., 2011). The aim of this work was to genotypically and phenotypically characterize a collection of clinical lngA + ETEC isolates through PCR, resistance profiles, self-aggregation properties, biofilm formation, and adherence to HT-29 cells to further understand if these attributes are related to the pathogenesis of CS21 positive clinical isolates from children with diarrhea.

BACTERIAL ISOLATES
Previously, clinical isolates identified as ETEC by serotyping were obtained from an E. coli culture collection repository at the Facultad de Medicina from Universidad Nacional Autónoma de México (UNAM) for the last three decades. All clinical ETEC were isolated from stool samples of children under 5 years old with diarrhea from México and Bangladesh (Table S1). Isolates were kept at −70 • C in Brain Heart Infusion broth (BHI, Difco, New Jersey, USA) with 15% glycerol, until their use. The ETEC isolates were cultured on 5% sheep blood agar plates (BBL, Franklin Lakes, NJ) at 37 • C and on MacConkey agar (Difco, New Jersey, USA).

IDENTIFICATION OF VIRULENCE GENES BY PCR
Genomic DNA was purified using a commercial Wizard kit (Promega, USA) from a bacterial culture grown overnight in Luria Bertani broth (LB, Difco, New Jersey, USA) at 37 • C. PCR assays were performed using a commercial PCR kit (Promega, USA) with the following reaction mixtures: 5 μl of reaction buffer [MgCl 2 (50 mM); Taq polymerase (0.l U), and dNTPs (2 mM)], 0.5 μl of the forward primer (10 μM), 0.5 μl of the reverse primer (10 μM), 3 μl of water, and 1 μl of DNA (100 ng) from each isolate. PCR reactions were conducted in a thermal cycler (Applied Biosystems GeneAmp PCR system 9700, NewYork, USA), with the specific melting temperature for each primer (Table 1). PCR products were separated on a 1.5% agarose gel in TAE (Tris-Acetate-EDTA) at 100 v, stained with ethidium bromide (5 μg/ml), and visualized by UV transilluminator (Bio-Imaging Systems, AccesoLab, México, D.F., México). To identify the st, lt, lngA, and cs3 and cfaI genes in clinical isolates, the genomic DNAs of ETEC E9034A (O8:H9, st + , lt + , lngA + , and cs3 + genes) and ETEC H10467 (cfaI + gene) were used as positive controls; whereas, the genomic DNA of E. coli HB101 was used as negative control. Based on PCR results, ETEC isolates positive for the lngA gene were selected for further characterization.

ANTIBIOTIC SUSCEPTIBILITY
Antibiotic susceptibility test was determined for ETEC isolates positive was used as quality control.

SELF-AGGREGATION ASSAY
The bacterial self-aggregation phenotype was analyzed using flatbottom 24-well tissue culture plates (Corning, NY, USA); the assay consists of bacterial clumping during growth in liquid media. Bacteria grown overnight in Terrific broth (TB) (Amresco LLC, Ohio, USA) at 37 • C were subcultured using a 1:100 dilution and incubated at 37 • C in TB for 4 h. Self-aggregates were visualized directly on a bacterial suspension by an inverted light microscope (Olympus, Center Valley, PA) (Clavijo et al., 2010). Triplicates of the assays were performed at three different times. ETEC isolates: E9034A and lngA − clinical isolate 114246 (background: O132:H25, st + , lt + , ecpA + ) were used as positive and negative controls, respectively. To establish the role of lngA in selfaggregation, a comparison between isogenic mutants E9034A vs. E9034 lngA::km and 73332 vs. 73332 lngA::km was made.

BIOFILM FORMATION
ETEC clinical isolates lngA positive were quantitatively analyzed according to the protocol described by Saldaña et al. (2009). In brief, 24-well plates containing 1 ml of PPLO were inoculated with 10 μl (1.5 × 10 8 bacteria/ml) of bacterial suspensions and incubated at 37 • C for 24 h. Biofilms on the surface of the wells were slowly washed three times with phosphate buffer saline 1x (PBS) (pH 7.4) and fixed with 2% formalin at 4 • C overnight. Wells with fixed biofilms were decanted, washed three times with PBS and stained with 1 ml of 1% crystal violet for 20 min. The excess of crystal violet was removed and the plates washed with water twice. Subsequently, 1 ml of 70% methanol was added for biofilm quantification by measuring optical density at 600 nm. Assays were made in Triplicates and repeated three consecutive www.frontiersin.org December 2014 | Volume 5 | Article 709 | 3 times. Enteroaggregative E. coli 042 and ETEC lngA − clinical isolate 114246 were used as positive and negative controls, respectively. ETEC lngA − absorbance was used to classify the biofilm formation into four categories as described by Saxena et al. (2014): non-biofilm, weakly, moderately, or strongly biofilm formers based upon the ODs of bacterial films (Saxena et al., 2014). The cut-off OD 600 nm for a 24 wells-plates is defined as two standard deviations above the mean OD 600 nm of the negative control.
To establish the role of lngA in biofilm formation, a comparison between isogenic mutants E9034A vs. E9034 lngA::km and 73332 vs. 73332 lngA::km was performed.

BACTERIAL ADHERENCE TO HT-29 CELLS
Colon adenocarcinoma HT-29 cells (ATCC HTB-38) were employed for the adherence assays, as described by Mazariego-Espinosa et al. (2010). Clinical ETEC lngA positive isolates were grown in PPLO overnight at 37 • C. Cell monolayers contained in 24-well polystyrene plates (Corning, NY, USA), were culture in Dulbecco's minimal Eagle medium (DMEM) (Invitrogen, Carlsbad, CA) supplemented with 10% FBS (Fetal Bovine Serum) (Gibco, USA) at 37 • C under 5% CO 2 until reaching an 80% confluence. The monolayers were infected with 10 μl of each bacterial suspension to an OD 600 nm of 0.5 in the McFarland scale and incubated for 4 h at 37 • C and 5% CO 2 . Infected monolayers were washed three times with 1x PBS (pH 7.4) to remove unbound bacteria and incubated of 0.1% Triton-100x (Amresco Bioscience, Sweden) for 5 min; serial dilutions were plated on LB, incubated at 37 • C for 24 h, and colony forming units (CFU) counted to determine CFU/ml. Adhesion assays were performed in triplicate and on three different days. Adherence values correspond to averages of triplicate assay and the corresponding standard deviation. ETEC isolates adherence was classified with cut-off CFU/ml for Enteroaggregative E. coli 042 is defined as two standard deviations above the mean of the positive control.

CONSTRUCTION OF AN lngA ISOGENIC MUTANT
The lambda red system was used to construct an isogenic mutant of the lngA gene in clinical isolate 73332 and the E9034A as described by Datsenko and Wanner (2000), Mazariego-Espinosa et al. (2010), and Mazariego-Espinosa et al. (2010). Briefly, the kanamycin resistance (km r ) cassette from plasmid pKD4 was amplified using the following primers lngA F (5 ttgagtttacctgagcagtacaggtacttagaattttgtctgcagtcaaggtgtagggctggagctgct tc 3 ) and lngA R: (5 tgctatccgtgtataaccggacacagaaatttaaagaagaggcaagaaaacatatgaatatcctccttag 3 ). The PCR-assay was performed in a 200 μl reaction mixture containing 10 mM Tris-HCl, 3 mM MgCl 2 , 200 μM deoxynucleoside triphosphates (dNTPs), 2 U of DNA polymerase, 0.25 μM each primer, and 10 μl of template DNA. The PCR products obtained were electroporated into ETEC clinical isolate 73332 and E9034A, harboring the pKD46 plasmid, and transformants were selected for kanamycin resistance.

STATISTICAL ANALYSIS
The results showed are the mean of three experiments performed in triplicate on different days. Statistical analysis was done using the Student's t-test.

IDENTIFICATION OF CS21 AMONG ETEC CLINICAL ISOLATES FROM CHILDREN WITH DIARRHEA
A total of 303 clinical isolates previously identified as ETEC by serotyping, were obtained from Mexican and Bangladeshi children stool and screened by PCR for the following genes: st, lt, lngA, cs1, cs3, cfa/I, and cofA. Eighty one point fifty one percent (247/303) of the isolates amplified at least one gene (

FIGURE 1 | Resistance profiles for ETEC positive isolates for the lngA gene.
Sixty-two ETEC isolates were tested for resistance against different antibiotic categories. * One ETEC isolate was resistant to a third generation cephalosporin, therefore an Extended Spectrum Beta Lactamases (ESLB) assay was performed, and was positive for this test. n = number of positive ETEC lngA isolates.

www.frontiersin.org
December 2014 | Volume 5 | Article 709 | 5 FIGURE 2 | Genotypes found in MDR ETEC isolates positive for the lngA gene. Sixty-five percent (39/60) of ETEC isolates were resistant to at least three antibiotic families. The association of multidrug-resistance to the genotypes was determined. n = number of positive ETEC lngA isolates.
Enteroaggregative E. coli 042 shows as rhomb on gray (positive control).
Enteroaggregative E. coli 042 shows as rhomb on gray (positive control).

FIGURE 6 | Role of Longus in ETEC isolates E9034A and 73332. (A)
Quantitative adherence assay in CFU after 6 h of infection to HT-29 cells. The percentage of each isolate is shown in a bar graph. (B) Growth curve in PPLO medium to corroborate growth rate among wild type isolates and isogenic mutants in the lngA gene. h, hours.
A growth kinetic curve with wild type ETEC and lngA mutants was performed in PPLO media ( Figure 6B) showing that there was no difference in growth among the wild-type and isogenic mutants ETEC isolates.

DISCUSSION
ETEC isolates are a frequent cause of diarrhea worldwide; ETEC causes nearly 400 million diarrhea episodes every year affecting children from developing countries as well as travelers (Wenneras and Erling, 2004;Riddle et al., 2006). ETEC colonizes the small bowel epithelium trough the expression of fimbrial and nonfimbrial CFs and producing LT and/or ST enterotoxins causing massive excretion of water and electrolytes (Qadri et al., 2005;Turner et al., 2006;Guttman and Finlay, 2008;Croxen and Finlay, 2010). After screening CFs in ETEC isolates from different geographical origins, CFA/I and Longus are the most prevalent CFs (Giron et al., 1994;Girón et al., 1995a;Gutierrez-Cazarez et al., 2000;Pichel et al., 2002). A strategy to control enterotoxigenic E. coli infections has been the development of a vaccine that triggers protective immunity against the most prevalent CFs (Giron et al., 1994;Girón et al., 1995a;Gutierrez-Cazarez et al., 2000;Pichel et al., 2002). In this study we analyzed a total of 303 clinical isolates obtained from Mexican and Bangladeshi children stools, classified as ETEC by serotyping. The st + /lt + (30.76%) was the most frequent genotype amplified for ETEC isolates analyzed in this study and the frequency was higher than previously reported by Guerra et al. (2014), in which only 15% of ETEC isolates from Colombia were positive for both enterotoxin genes (Guerra et al., 2014). However, a systematic review by Isidean et al. (2011), reported a global distribution of st + /lt + of 33% in ETEC clinical isolates, similar to the findings in this study. The frequency of the lngA gene among ETEC clinical isolates in several countries may vary between 8.5 and 50% (Girón et al., 1995a;Gutierrez-Cazarez et al., 2000;Nishimura et al., 2002;Pichel et al., 2002;Guerra et al., 2014). Meanwhile, in this study the lngA gene was amplified in 25.10% of the isolates. A lower frequency of the lngA + gene associated to st + , st + /lt + , and lt + genes was observed when compared with data described by Nishimura et al. (2002); which reported a frequency of 56%, 31%, and 10%, respectively (Nishimura et al., 2002). Likewise, lngA-positive ETEC isolates (25.10%) also harbor genes coding for other CFs. In other reports, lngA + was most highly associated to isolates coding for CFA/II (Girón et al., 1995b;Nishimura et al., 2002) and CFA/I that has been described as the most prevalent CF in developing countries. However, in this study it appeared in low frequency 6.45%, for cfaII (csI, cs3), and 17.74% for cfaI (Pichel et al., 2002;Qadri et al., 2005). Interestingly, lngA + isolates were found alone in 11.29% of the ETEC isolates. This result suggest that these isolates could have lost the plasmid that code for enterotoxins, leaving only the megaplasmid coding for lngA + ; therefore, lngA is proposed as a molecular marker together with lt and st genes, for identification of the ETEC pathotype. The most prevalent serotypes found in ETEC lngA + isolates were O128ac:H12, O78:H12, and O6:H6. The lngA gene was also described in a considerable number of ETEC serotypes (O2:NT, O6:NM, O6:H2, O6:H16, O6:H48, O8:H2, O8:H6, O8:H21, O8:H29, O20:H2, O25:NM, O25:H42, O49:NM, O128:NM, O139:H28, O148:H28, O153:H2, O153:H45, O158:H10, and O159:H4) (Girón et al., 1995a). These data provide further evidence of the wide distribution of CS21 among ETEC clinical isolates.
Although the treatment of diarrhea caused by ETEC with antibiotic agents is generally restricted to severe and moderate hospitalized cases or immunocompromised patients, there is an increased number of clinical isolates of MDR ETEC, which could be a potential public health problem; for this reason, the study of antibiotic resistance should be carried out in combination with the frequencies of CFs and enterotoxins. We found that 46.15% of MDR ETEC isolates was able to produce the LngA protein. In this study ETEC isolates were MDR and producers of the LngA protein providing the bacteria colonization and pathogenic advantages. CS21, a type IV pilus and a common CF in ETEC, induce microcolony formation, twitching motility, adherence to cells, and self-aggregation (Clavijo et al., 2010;Mazariego-Espinosa et al., 2010). Self-aggregation has been associated to bacterial protection against antibiotics in a similar manner as bacterial biofilms avoid contact with lumen intestinal bactericidal agents such as lactoferrin, lysozyme, and secretory antibodies. In this study, 48.38% of the ETEC isolates were able to self-aggregate and 76.66% of the ETEC isolates produced the LngA protein.
Biofilms are matrix-enclosed communities of bacteria that show increased antibiotic resistance and the ability to evade the immune system (MacFarlane and Dillon, 2007). They can cause recalcitrant infections which cannot be cured with classical antibiotic therapy (Forier et al., 2014). Biofilm production and adherence mechanisms of ETEC isolates are associated with CFs and multiple genes coding for CFs located on a plasmid (CFA/I and CFA/II), as well as on its chromosome (CS2). In our study, 83.87% of the ETEC isolates tested were able to form biofilm at 24 h. Liaqat and Sakellaris (2012), described that all ETEC isolates analyzed were able to form biofilm after 60 h (Liaqat and Sakellaris, 2012). Outstandingly, most of the ETEC isolates were weak and moderate biofilm producers and a relationship was not found.
Many pathogenic bacteria have the ability to adhere to specific host tissues prior to colonization. Thus, bacterial adhesion is the initial step that precedes colonization and in some cases invasion of the epithelial surfaces (Pizarro-Cerda and Cossart, 2006). All ETEC isolates were able to adhere to HT-29 cells, and the isolates that produce the LngA protein were associated with high adherence values. Mazariego-Espinosa et al. (2010), demonstrated that CS21 produced by ETEC E9034A is involved in the phenomenon of adherence as a requirement for mediating the interaction between the bacteria and cultured epithelial cells, specifically intestinal cells (Mazariego-Espinosa et al., 2010). In this study, we demonstrate the role of CS21 for adhesion, using the E9034A prototype strain that has been widely used in other studies and a clinical isolate 73332. The aim was to analyze through quantitative studies the potential of CS21 as an adhesin in clinical isolates of ETEC during the colonization to their host.
The growth curve kinetic was performed to guarantee that the adherence phenotype was not affected by bacterial growth. Isogenic mutants in the lngA gene showed a reduction of adherence from 77% to 98.8%. However, Mazariego-Espinosa et al. (2010) showed a reduction of 32% in adherence levels in the E9034A lngA when compared to ETEC E9034A (Mazariego-Espinosa et al., 2010). The difference observed in adherence levels in both mutants, could be attributed to the generation of a second mutant in the lngA gene employed in this study from a highly adherent colony of the E9034A wild-type strain; which, was previously selected from multiple subcultures in HT-29 cells in different days.
Antibiotic categories such as sulfonamides and beta-lactam containing compounds can be effective prophylaxis against traveler's diarrhea by deficiency of effective vaccine treatments (Brown et al., 2009) this can be explained by the different geographical origins of the Mexican and Bangladeshi isolates. However, the emergence of antibiotic resistant isolates and the genes encoding resistance proteins are known to reside on horizontal transmissible elements such as conjugative plasmids and integrons promoting the spread of resistance (Levy, 2002). The MDR ETEC isolates able to produce LngA showed features associated with self-aggregation, and adherence to intestinal cell lines. These features are important in the colonization process. Taking into consideration the frequencies of occurrence of the widely spread CFs such as CFA/I and CS21 in other geographic regions, the frequencies of occurrence of the lngA pilin gene found here or described by others confirm the importance of the CS21 pilus as a widely distributed antigen among ETEC isolates.