Infection due to ETEC continues to be a major cause of childhood diarrhoea in developing countries, especially in children aged below 5 years. Studies world-wide often report that CFA/I and CS1-CS6 are the most common CFs; these surface antigens and the LT toxin are current targets for different vaccine strategies 
. The finding of an unexpectedly high number of LT/CS17 expressing ETEC isolates in Bolivia prompted us to further investigate whether this was due to a local clonal outbreak or to genetically unrelated isolates. We found by using MLST that at least two dominant MLST sequence types expressing LT and CS17 were present during several years in the La Paz area in Bolivia. However, several other sequence types were present among ETEC isolates expressing LT and CS17. These results indicate that several clones of LT/CS17 strains were circulating in Bolivia and that the high frequency of CS17 isolates was not due to a single clonal outbreak. In fact our results suggest that several sequence types expressing LT/CS17, such as isolates within the sequence type-423 and -443, co-circulate and that new sequence types emerge over time that share this phenotype. Indeed, ongoing studies in Bolivia seem to confirm that LT/CS17 isolates continue to commonly infect children and cause severe diarrhoea in Bolivia. This result is in contrast to findings in other parts of the world where LT-expressing isolates are considered in general to cause milder symptoms and less severe diarrhoea 
Several studies indicate that the chromosomal background, as measured by serogroup and RAPD, appears to correlate with and perhaps even determine toxin and CF profile in ETEC 
. This is in line with findings in the other pathogenic E. coli
EHEC and EPEC, where results indicate that these pathogens have evolved mainly by clonal expansion and that certain virulence factors, despite being plasmid borne, are only found in particular phylogenetic groups 
. However, the results of this study indicate that ETEC isolates with the same chromosomal background (i.e.
MLST sequence type) may express different toxins and belong to different serogroups; we found that the same sequence type may be very variable with respect to toxin, CF and serogroup (). We have also found in this as well as in other studies that the same toxin/CF profile occurs in different MLST sequence types 
. The diversity among ETEC is supported by recent studies in which ETEC toxins were found in all phylogenetic lineages of E. coli
indicating that ETEC virulence genes may be acquired independently of the chromosomal background, although clonal expansion of particular strains occurs 
. Hence whether an isolate is an ETEC seems to be defined entirely by the presence of toxins and to be less dependent on chromosomal background than is the case for other E. coli
We define an ETEC clone as strains that share the same MLST sequence type, serogroup, toxin and CF profile but also identical sequences in their virulence genes. To convincingly determine whether the recovered LT/CS17 isolates from Bolivia were clonal we sequenced parts of the CS17 encoding operon in order to enhance genetic resolution. We found very little diversity in the CS17-fimbriae encoding operon sequence when comparing Bolivian and Bangladeshi isolates, although only a portion of the operon sequence was determined. One syntenic SNP in the coding region of the usher was uncovered. This SNP was detected in ten Bolivian strains but not in the five remaining Bolivian CS17 strains nor in any of the Bangladeshi strains. All strains belonging to sequence type-423, strain 2278, which had a sequence type related to 423, and two other LT/CS17 strains isolated in 2009 had the SNP. These results strongly supports that the sequence type-423 group contains strains that define a highly related genetic cluster. To this end we now consider the MLST sequence type-423 LT/CS17 O8:H9 SNPBol strains as a clone.
One might expect that clones evolve over time and in an attempt to follow if this was the case for the main Bolivian clone, we compared antibiotic resistance patterns as well as clinical symptoms of the children infected with the clone in order to propose a model of the circulation of this clone. When analysing phenotypic (serogroup and antimicrobial resistance patterns) and genetic traits (MLST and virulence gene SNPs) we found that clinical isolates 18 and 132 were indistinguishable. These ETEC were isolated almost four years apart, indicating that indistinguishable strains persist over time. Strains with sequence type-423 were all isolated between early 2002 and late 2005, and the related isolate 2278 was recovered in 2007 which may indicate that strains within this clonal group began drifting, at least in the gyrB gene, in 2006 or 2007. However at this stage we cannot rule out that isolate 2278 represents a new LT/CS17 clone. Isolates 2761 and 2763 were recovered in 2009 and had the Bolivian SNP but were phylogenetically distant from MLST sequence type-423, hence these strains might have acquired the Bolivian CS17 allele through horizontal transfer, perhaps from one of the older sequence type-423 clonal strains.
It is interesting to notice that the antimicrobial resistance patterns were stable with respect to resistance towards ampicillin, ampicillin–sulbactam, erythromycin, penicillinG and oxacillin but isolates seemed to gain or lose resistance genes towards chloramphenicol, tetracycline and trimethoprim more readily. Most isolates were multi-resistant, although the targets of these resistances were variable. This result further enforces the need for restricted use of antibiotics. In conclusion, we found that several LT/CS17 strains persist in Bolivia but that one clone, sequence type-423 O8: H9 SNPBol
, was repeatedly isolated from children. Another study has suggested that pathogenic E. coli
clones may be increasingly resistant towards antibiotics and increasingly virulent over time 
. Although further studies are needed to see if similar mechanisms are found in Bolivian ETEC clones we could see a tendency of increased virulence over time in MLST sequence type-423 O8: H9 SNPBol
(). In conclusion this study is to our knowledge the first study to indicate that several different LT/CS17 clones circulate in Bolivia but that some individual clones such as MLST sequence type-423 O8: H9 SNPBol
are more persistent than others and repeatedly infect children.