Phenotypic and Genotypic Characterization of Enterotoxigenic Escherichia coli Clinical Isolates from Northern Colombia, South America

Enterotoxigenic Escherichia coli (ETEC) are major causes of childhood diarrhea in low and middle income countries including Colombia, South America. To understand the diversity of ETEC strains in the region, clinical isolates obtained from northern Colombia children were evaluated for multiple locus sequencing typing, serotyping, classical and nonclassical virulence genes, and antibiotic susceptibility. Among 40 ETEC clinical isolates evaluated, 21 (52.5%) were positive for LT gene, 13 (32.5%) for ST gene, and 6 (15%) for both ST and LT. The most prevalent colonization surface antigens (CS) were CS21 and CFA/I identified in 21 (50%) and 13 (32.5%) isolates, respectively. The eatA, irp2, and fyuA were the most common nonclassical virulence genes present in more than 60% of the isolates. Ampicillin resistance (80% of the strains) was the most frequent phenotype among ETEC strains followed by trimethoprim-sulfamethoxazole resistance (52.5%). Based on multiple locus sequencing typing (MLST), we recognize that 6 clonal groups of ETEC clinical isolates circulate in Colombia. ETEC clinical isolates from children in northern Colombia are highly diverse, yet some isolates circulating in the community belong to well-defined clonal groups that share a unique set of virulence factors, serotypes, and MLST sequence types.


Introduction
Enterotoxigenic Escherichia coli (ETEC) are important enteric pathogens worldwide, especially affecting children in developing countries [1,2]. ETEC strains are responsible for ∼400 million diarrheal cases annually in children less than 5 years of age, resulting in 300,000 to 500,000 deaths, and they are the most common causes of traveller's diarrhea, accounting for 50% of all traveler's diarrhea episodes [3,4]. ETEC strains belong to a highly diverse group of strains with respect to enterotoxin type, colonization surface antigens (CSs), serotypes, and ancestral lineages [5][6][7][8].
ETEC strains are defined by the presence of plasmidencoded heat-labile toxin (LT) and/or the heat-stable toxins (ST) [9,10]. ST, a guanylin homologue expressed in intestinal cells, is a heterogeneous peptide with two major subtypes STa, present predominantly in human ETEC isolates, and STb, present predominantly in animal ETECs [11]. Both subtypes induce diarrhea in piglets [12]. STa is further subdivided in two variants, STh and STp (from their initial detection among pigs) that have been reported in ETEC clinical isolates from different parts of the world [11,13]. ETEC strains also express plasmid-or chromosomally encoded colonization surface antigens (CSs). These heterogeneous pili or nonpili surface structures are believed to promote small intestine ETEC colonization and they are currently considered important vaccine targets. Twenty-two different CSs have been identified among human ETEC of diverse geographic origins [14,15]. ETEC isolates may produce one or more CSs, while some isolates do not express any or do not produce recognizable CSs [2].

BioMed Research International
LT and ST toxin types and CSs profiles from clinical ETEC isolates vary from one geographic region to another [2,6,16,17].
ETEC also expresses a variety of nonclassical virulence factors that may be essential for pathogenicity and promising vaccine targets. Among nonfimbrial adhesins/invasins, Tia is a 25 kD outer membrane protein that interacts with host cell surface proteoglycans and by itself is sufficient to promote bacterial adherence and epithelial cell invasion when cloned into laboratory E. coli strains [18,19]. The labile enterotoxin output gene (leoA), encoding a cytoplasmic protein with GTPase activity, is required for maximal LT secretion. Both Tia and LeoA are encoded in a 46-Kb pathogenicity island (Tia-PAI) [20,21]. The TibA protein encodes a glycosylated autotransporter that mediates adhesion to surface epithelial cells, autoaggregation, and biofilm formation [22,23]. The etpBAC locus encodes three proteins: EtpA, a 170 kDA secreted glycoprotein, EtpB a transport pore, and EtpC, a putative glycosyltransferase required both for optimal secretion and glycosylation of EtpA. The EtpA glycoprotein appears to act as a molecular bridge, binding the exposed regions of FliC at the flagellar tip and host surface structures [24,25]. EatA, a serine protease autotransporter of the Enterobacteriaceae (SPATE) family, was shown to increase ETEC virulence in an animal model, by degrading mucin and facilitating LT release [25,26]. Finally, the irp2 and fyuA genes, located in the high-pathogenicity island (HPI), encode a yersiniabactin-like iron scavenging system [27].
ETEC is the leading cause of diarrhea in children less than 5 years of age in Colombia, South America [28,29], yet no information is available of the phenotypes and genotypes associated with these strains. The objectives of this study were to identify the most common genotypes associated with Colombian ETEC clinical isolates with respect to enterotoxins, CSs, nonclassical virulence genes, and genomic profiles and to determine the most common ETEC O:H serotypes and antimicrobial susceptibility patterns. Recognizing the most frequent circulating strains including the most common potential antigens may help prioritize ETEC diarrhea prevention measures including vaccine development research strategies.
We found that ETEC isolates were positive for LT, ST, or both ST/LT genes, the most prevalent CSs were CS21 and CFA/I, and the most common nonclassical virulence genes were eatA, irp2, and fyuA. Based on MLST, serotyping, and virulence genotype, Colombian ETEC clinical isolates showed broad genetic diversity, yet 6 distinctive clonal groups were identified.

Strains Used in This
Study. Thirty-two ETEC clinical isolates from children less than 5 years of age with diarrhea and 8 ETECs obtained from healthy children from two studies were used. Seven of these ETEC strains came from children with diarrhea from a prevalence study previously described in two Caribbean cities in Colombia [29]. The remaining ETEC strains were obtained from case-control studies of children less than 5 years of age also in Cartagena, Colombia [28]. Twenty-five strains were from cases and 8 strains from healthy controls (Table S1). All ETEC clinical isolates identified from the two epidemiological studies mentioned above were included in the present study. These isolates were identified by multiplex PCR using ST primers (ST.F-5 GCTAAACCAGTA(G/A)GGTCTTCAAAA3 and ST.R-5 CCCGGTACA(G/A)GCAGGATTACAACA 3 ) and LT primers (LT.F-5 GCACACGGAGCTCCTCAGTC-3 and LT.R-5 TCCTTCATCCTTTCAATGGCTTT 3 ) [29]. Reference ETEC and non-ETEC strains were used as positive and negative controls for all assays and they are described in Table 1. ETEC strains carrying different CS genes, used as controls for PCR detection assays, were kindly provided by Dr. Steven Savarino from the Naval Medical Research Center.

DNA Techniques.
Unless otherwise specified, standard methods were used for plasmid isolation, genomic DNA isolation, and agarose electrophoresis DNA separation [30]. E. coli clinical isolates were processed for isolation of genomic DNA as previously described [31]. In brief, overnight liquid cultures were centrifuged, and the pellet was resuspended in water, boiled for 10 min, and centrifuged again. The supernatant containing a crude DNA extract was used as a DNA template for PCR assays.

Multilocus Sequence Typing (MLST).
Genetic diversity of ETEC strains was analysed by multilocus sequence typing (MLST) by using the University College Cork E. coli MLST scheme (http://mlst.warwick.ac.uk), which is based on sequencing of internal regions of 7 housekeeping genes adk, fumC, gyrB, icd, mdh, purA, and recA [34]. Phylogenetic trees were constructed using the Phylogeny.fr software available online at http://www.phylogeny.fr/version2 cgi/index.cgi by the cluster W method [35]. Forty seven-gene DNA sequence concatamers for each ETEC strain as well as 4 E. coli control sequence concatamers from ancestral groups A, B1, B2, and D were assembled and aligned using the ClustalW program. The phylogenetic tree using the PhyML program was constructed using bootstrapping procedure (100 straps) [35,36]. Some ETEC strains were assigned to clonal groups. A clonal group was defined as a group of more than one ETEC strains that do not seem to share ancestral origin with other ETEC and that have at least 2 strains with identical MLST DNA sequence.

Serotyping.
Serotyping was performed at the E. coli Reference Center, Pennsylvania State University, according to standard methods for determining the O antigen [37]. H typing was performed using a fliC PCR-RFLP method [38].

Colombian ETEC Clinical Isolates Carry LT, ST, and LT/ST Enterotoxins. ETEC clinical isolates from children
with diarrhea and with no diarrhea previously described were evaluated for the presence of LT and/or ST enterotoxin genes (See Table S1) [28,29]. As shown in Table 2, LT-containing ETEC strains were the most frequently detected group (52.5%) followed by ST-containing ETEC strains (32.5%) and ETEC-LT/ST strains (15%). All ETEC-ST strains tested were positive for the STh variant. No STp variants were identified.

CSs Were Frequently Found among ETEC Clinical Isolates.
CSs are piliated and nonpiliated structures believed to be involved in ETEC colonization of the human gut. In our study 75% of all clinical isolates were positive for at least one CS and 25% were negative for any CS (Table 2). Among ETECassociated CS described in the literature, 10 different types of CSs were detected among Colombian ETEC isolates. Alone or in association, the most prevalent ETEC CSs were CS21, CFA/I, CS6, and CS5, present in 50.0%, 32.5%, 20%, and 12.5%, respectively. Detection of CS21 was present in 12 (92.3%) out of 13 ETEC strains containing ST, followed by 7 (33.3%) out of 21 ETEC strains containing LT and only 1 (16.6%) out of 6 ETEC strains containing LT and ST toxins. CFA/I was found in 12 (92.3%) out of 13 ETEC strains containing ST strains and only in 1 (4.7%) out of 21 ETEC strains carrying LT. CFA/I was not detected among ETEC containing ST and LT toxins. CS6 was found in 4 (19%) out 21 LT-containing ETEC strains and in 4 (66.6%) out of 6 ST and LT containing ETEC strains. CS5 was only found among 5 (83.3%) out 6 ST/LT-containing ETEC strains. CS1, CS2, CS7, CS12, CS13, and CS17 CSs were only detected among LT-containing ETEC strains. Among CSnegative ETEC, 9 out 21 (42.8%) were LT-containing ETEC strains, and only 1 out of 12 (8.3%) were ST-containing ETEC strains.

Phylogenetic and MLST Results.
To evaluate the genetic relatedness among Colombian ETEC isolates and with ancestral E. coli strains, MLST was conducted and a phylogenetic tree was analyzed. MLST sequences from E. coli pathogens representing ancestral E. coli groups A, B1, B2, and D were included in the analysis as controls. The phylogenetic tree constructed from Colombian ETEC MLST sequences is highly diverse as demonstrated by the extended branching ( Figure 2). Despite genetic diversity, most ETEC strains seem to be derived from E. coli ancestral groups A and B1, as determined by the phylogenetic tree evolutionary relationships using E. coli control strains MLSTs from ancestral groups A, B1, B2, and D. Only a minority of the ETEC strain MLSTs were associated with E. coli strain MLSTs from phylogenic groups B2 and D. Nineteen (47.5%) ETEC strains are clustered into six allelic groups. Based on identical MLST sequences within each cluster, we have designated these clusters as clonal groups 1 to 6. Based on evolutionary relationships using E. coli control strain MLSTs, clonal groups 1, 2, 4, 5, and 6 are associated with ancestral groups A and B1, while clonal group 3 MLST is associated with ancestral group D and B2. Analysis of MLST sequence types (SeqT) found that all 40 ETEC isolates have one designated specific SeqT (Table 3). SeqT is defined as the allelic profile resulting from the seven alleles assigned to each one of the 7 house-keeping loci sequences.

Discussion
ETEC diarrhea is a leading cause of morbidity and mortality in children less than 5 year of age living in underserved geographic areas of the world and a leading cause of traveler's diarrhea. ETEC is also a leading cause of morbidity in Colombia, a middle income country in Latin America. In this study, we show that northern Colombian ETEC clinical isolates from children less than 5 years of age are a highly diverse group of strains based on MLST, serotyping, and presence of classical and nonclassical virulence factors, yet 6 clonal groups were identified. A limitation of our study is the limited number of ETEC strains tested and also the strains origin is limited to two mayor urban centers in northern Colombia. Accordingly we will confine our analysis and conclusions to northern Colombian ETEC strains. Further studies will be necessary to evaluate ETEC diversity from all Colombian corners to better define ETEC Colombian virulence and colonization gene diversity and strain clonality. LT, the most frequently toxin type found in the Colombian ETEC isolates, was detected in 67.5% of strains, whether alone or in combination with ST [41]. This entorotoxin pattern was also reported in Bangladesh and Peru, where LT-producing ETEC was seen in 52% and 72% of the cases, respectively [42,43]. This is in contrast with other studies from Indonesia and Chile showing that ST-producing ETEC predominated [33,44]. ST enterotoxin variants STp and STh induce disease in humans, and differentiation of STp from STh may help identify differences in the epidemiology of these two strains [45]. In this study, all Colombian ETEC strains positive for ST were only positive for the STh variant; no ETEC strains positive for STp were identified. Low frequency of STp ETEC strains was also reported among Brazilian, Bolivian, and Chilean ETEC clinical isolates suggesting that STh toxin predominates among Latin American ETECs [33,46,47].
CS21 was the most commonly identified CS among Colombian ETEC isolates, followed by CFA/I and CS6. The association of CS21 and CFA/I among ETECs was also remarkable. Similar findings were reported in Chile where CS21 was found in association with CFA/I [33]. In contrast, CFA/I and CS14 predominated among Bolivian ETEC strains [46]. CS21 has also been identified in ETEC strains from Argentina, Brazil, Bolivia, Egypt, and Bangladesh [33,46,[48][49][50][51].
CS21 is long rod-like fimbria that directs adhesion to intestinal epithelial cells and mediate self-aggregation and twitching motility, and it is involved in pathogenesis [52][53][54]. In this study, CS21 was more often associated with LT producer Colombian ETEC strains. ETEC strains expressing CS21 tend to be isolated in higher proportion among pediatric populations [50]. It is likely that the higher proportion of CS21 among Colombian ETECs is due to the fact that all ETEC strains were obtained from children less than 5 years of age. CFA/I, CS6, and CS21 were detected among all ETEC toxins profiles, demonstrating their extended distribution. Similarly,      [2,14]. In 10 (25%) of Colombian ETEC isolates, no CSs were identified. This indicates that these 10 ETECs do not express any known CSs, they contain CSs variants unable to be recognized by conventional PCR, or they contain unknown CSs unable to be recognized with current PCR assays [2,55]. These strains may express novel CSs pending to be identified. Nonclassical virulence genes were detected among Colombian ETEC strains. The eatA, irp2, and fyuA were the most frequent. These genes are known to be associated with the ETEC HPI pathogenicity island [27]. The distribution of nonclassical virulence genes among Colombian ETEC strains is similar to Chilean ETEC strains, except that etpA and etpB are present at lower rates (18%) [33]. The tia and leoA genes were also detected at low frequency, similar to the reported frequency among strains from Bolivia, Chile, Guatemala, Mexico, and India [33,41,56]. In contrast, irp2 and fyuA genes were detected in more than 70% of the strains as previously described among Chilean strains [33].
Colombian ETEC strains have a widely diverse phylogenetic distribution represented in 17 known MLST sequence types. Ample diversity among human ETEC strains from different geographic regions worldwide was reported previously [8]. Colombian ETECs segregated with any ancestral E. coli clonal groups A, B1, B2, and D. This is consistent with the idea that ETEC strains are representative members of distinct ETEC lineages [8,10]. In regard to serotyping, 16 [57].
Despite serogroup diversity, there is an association between serogroups and genetic linkages as demonstrated by the serotype-specific clonal clusters among ETEC Brazilian strains [58]. In our study, 6 clonal groups were identified based on MLST and some of them shared similar serotypes, CSs, and nonclassical virulence genes. The MLST sequence types for each of clonal groups 1 to 6 that had been described before in the E. coli MLST database suggest that these clonal groups may circulate not only in Colombia but also in other geographic regions. The Colombian ETEC clonal group 1 with MLST SeqT2332 was associated with O128 serogroup. This serotype was previously described among ETEC strains in Bangladesh, Brazil, Egypt, and Tunisia [7,[58][59][60]. The SeqT2332 was described in Mexico according to the University College Cork E. coli MLST Database, yet it is not frequently detected among Bolivian or Mexican ETEC strains. SeqT443 is the second most common sequence type among Colombian ETECs. SeqT423 and SeqT443 predominate in Mexican ETEC isolates while SeqT398 predominates in Guatemala and Mexico ETEC strains [61]. While limited studies are available in Latin America on phylogeny of ETEC clinical isolates using MLST schemes, it is suggested based on the available data that ETEC are highly diverse in Latin America. This study has tested a limited number of ETEC strains and the origin of these isolates is also limited to two mayor urban centers in northern Colombia. Accordingly, we confine our analysis and conclusions to Northern Colombian ETEC strains. Further studies are necessary to evaluate ETEC diversity from all Colombian corners to better define Colombian ETEC genotype, phenotype, clonality, and genetic diversity.

Conclusion
In summary, ETEC clinical isolates from northern Colombia are a highly diverse group of intestinal pathogens that possess multiple combinations of classical and nonclassical virulence factors as well as MLST sequence types and serotypes. Despite the genotypic and phenotypic diversity, 6 well-defined clonal groups were identified. These predominant clonal groups have been circulating within the community for several years and they share an almost identical set of classical and nonclassical virulence factors as well as MLST and serotype. Further research in Colombia and other Latin American countries is necessary to identify the most prevalent ETECassociated virulence factors with immunoprotection potential. Promising ETEC vaccine candidates should take into account predominant antigens, antigenic diversity, and geographic variation.