Extended-spectrum β-lactamases , transferable quinolone resistance , and virulotyping in extra-intestinal E . coli in Uruguay

Introduction: To characterize extended-spectrum β-lactamases (ESBLs) and plasmid-mediated quinolone resistance (PMQR) genes in Escherichia coli isolates obtained from extra-intestinal samples in three Uruguayan hospitals. Methodology: Fifty-five ESBL-producing E. coli isolates were studied. Virulence genes, ESBLs, and PMQR genes were detected by polymerase chain reaction. ESBL-producing isolates were compared by pulsed-field gel electrophoresis. Multi-locus sequence typing was also performed on 13 selected isolates. Results: Thirty-seven isolates harbored blaCTX-M-15 (67.3%), eight blaCTX-M-2 (14.6%), five blaCTX-M-14 (9.1%), three carried both blaCTX-M-2 and blaCTX-M-14, one blaCTX-M-9, and one blaCTX-M-8. Among the CTX-M-15 producers, 92% belonged to sequence types ST131 and ST405, and carried aac(6’)Ib-cr as well. Isolates harboring blaCTX-M-2, blaCTX-M-14, blaCTX-M-9, or blaCTX-M-8 were found to be genetically unrelated. Conclusions: The successful dissemination of CTX-M-15-producing E.coli isolates seems to be linked to the spreading of high-risk clones and horizontal gene transfer. A trade-off between carrying more antibiotic resistance and less virulence-related genes could partially account for the evolutionary advantages featured by successful clones.


Introduction
Escherichia coli is a versatile pathogen, responsible for intestinal and extra-intestinal infections.Urinary tract infections (UTIs), sepsis, bacteremia, and meningitis are among the latter.The ability of E. coli to cause extra intestinal infections relies on a number of virulence factors that include adhesins, tissue-damaging effectors, and factors conferring resistance to the bactericidal activity of serum, among others [1,2].
Identification of extra-intestinal pathogenic E. coli strains (ExPEC) can be achieved by detecting at least two of the following genes: papA, papC, sfa/foc, afa/dra, iutA, or kpsM II [3].The occurrence of these genes is associated with strains belonging to phylogenetic groups B2 and D (associated with ExPEC), and less frequently with groups A and B1, corresponding to commensal E. coli strains [4].
The emergence of successful clones such as E. coli ST131 or clonal group A (CGA), both distributed worldwide, is a clear example of coexistence between antibiotic multiresistance and virulence factors [5][6][7].
Co-resistance in E. coli to oxyiminocephalosporins (especially mediated by ESBLs) and quinolones is an increasing event worldwide, particularly in Latin America [8].Yet, the available information concerning the dissemination of different clones and/or the presence of virulence factors is scarce in South America [9,10].
In the present work, we describe the ESBLs present in E. coli strains obtained from extra-intestinal samples from three hospitals in Uruguay.

Bacterial isolates
Fifty-five ESBL-producing extra-intestinal E. coli isolates were obtained from three Uruguayan hospitals located in two different cities, between 1 March 2010 and 28 February 2011.Each isolate was designated with letters E, C, or HP according to the hospital of origin.Thirty-six isolates were recovered from urine, seven from blood, five from skin lesions, two from abscesses, three from respiratory samples, and two from peritoneal fluid samples.

Characterization of ESBL and PMQR genes
Isolates with positive ESBL-screening results were further analyzed by polymerase chain reaction (PCR) for the presence of bla CTX-M , bla TEM , bla PER-2 , and bla SHV genes using specific primers [11]; PCR products were sequenced on both strands using the same primers.

Multi-locus sequence typing (MLST), determination of CGA, and PFGE
MLST was performed on selected isolates, representative of the most frequent pulsetypes, by gene amplification and sequencing of seven housekeeping genes (adk, fumC, gyrB, icd, mdh, purA, and recA), according to the protocol and primers specified at the E. coli MLST web site (http://mlst.ucc.ie/mlst/dbs/Ecoli).
XbaI PFGE analysis was performed as previously described [17].Additionally, CGA was determined for isolates characterized by MLST and belonging to phylogenetic group D by single-nucleotide polymorphisms (SNPs) analysis of fumC, as described by Johnson et al. [18].PFGE profiles were analyzed with BioNumerics fingerprinting software (Applied Maths, St-Martens-Latem, Belgium).Dendrograms were generated using the unweighted pair-group method (UPGMA) using arithmetic averages, based on the Dice similarity coefficient, with a 2.0% band position tolerance.PFGE profiles sharing > 85% similarity were considered to be genetically related [19].
Conjugative plasmid incompatibility groups (Inc) and addiction systems were determined by PCR-based replicon-typing according to Carattoli et al. and Mnif et al., respectively, using genomic DNA obtained from transconjugants as a template [21,22].

Phylogenetic group and virulotyping
Classification of isolates into phylogenetic groups was determined by PCR as previously described [4], based on the presence or absence of three DNA fragments: chuA-, yjaA, and Tspe4.C2.Additionally, the sub-grouping scheme proposed by Branger et al. [23] was used.Briefly, the absence of the three DNA fragments corresponds to subgroup A 0 , whereas the occurrence of only yjaA corresponds to A 1 ; on the other hand, B2 2 corresponds to the occurrence of chuA and yjaA, and B2 3 is defined by the presence of the three DNA fragments.Finally, subgroup D 1 features the presence of only chuA, whereas subgroup D 2 features chuA and Tspe4.C2.
Detection of virulence-related genes was performed in two stages.First, all isolates underwent PCR virulence screening (VS) as previously described by Johnson et al [3].

Results
The 55 isolates under study were phenotypically identified as ESBL-producers; however, the proportion of isolates resistant to cefotaxime or to ceftazidime was dissimilar.
All of the isolates were susceptible to carbapenems (data not shown).
Fifty of the ESBL-producing isolates were also resistant to ciprofloxacin, 35 of which harbored aac(6')Ib-cr (33 carried bla CTX-M-15 and two carried bla CTX-M-14 ).Additionally, qnrA was detected in a single isolate, along with bla CTX-M-9 .Furthermore, qnrB was detected along with bla CTX-M-8 ; however, in both cases, MIC values to ciprofloxacin were within the susceptibility range defined by the CLSI (Figure 1).No qnrC, qnrD, qnrS, or qepA genes were detected.

Clonal relationship
PFGE assays were performed on all ESBLproducing isolates; four of them were untypeable (three CTX-M-2 producers and one CTX-M-15 producer).
PFGE analysis of bla CTX-M-15-bearing isolates indicated that they were clustered in five pulsetypes; two major pulsetypes, named A and B, accounted for 20 and 13 isolates, respectively, whereas pulsetypes C and D comprised two isolates each, and pulsetype E included a single isolate (Figure 2).
MLST analysis showed that all of the isolates in pulsetype A belonged to sequence type ST131, whereas isolates in pulsetype B belonged to ST405.
Isolates belonging to ST131 were detected in the three hospitals included in this study and were recovered from various sources (urine, blood, surgical wound infections, and peritoneal fluid).

Determination of clonal group A
Phylogenetic group D included 21 isolates, 13 of which belonged to pulsetype B and ST405.Furthermore, another 6 isolates underwent MLST studies.Sequence analyses of fumC aimed at the detection of CGA showed that only a single isolate (HP21), belonging to ST69 and carrying bla CTX-M- 9 /qnrA, displayed the SNP C288T (Table 2).

Transfer of ESBL genes and plasmid characterization
Twenty-five isolates representative of each pulsetype, as well as the four untypeable isolates, were selected for conjugation assays: seven isolates harboring bla CTX-M-15 , eight bla CTX-M-2 , five bla CTX-M-14 , three isolates carrying both bla CTX-M-2 and bla CTX-M-14 , one carrying bla CTX-M-9 , and one with bla CTX-M-8 (Table 3).Thirteen transconjugants were obtained.All of the bla CTX-M-14 , bla CTX-M-9 , and bla CTX-M-8 genes were encoded in conjugative plasmids; however, only two bla CTX-M-15 alleles (carried by isolates E13 and E20, pulsetypes A and C, respectively) as well as a single bla CTX-M-2 allele (isolate E44, pulsetype K) were encoded in conjugative plasmids.
Isolates belonging to phylogenetic group D 1 were found to be genetically unrelated; five such isolates were CTX-M-2 producers and were typed as ST393 (n = 2), ST38, ST1158, and ST3627.Other D 1 isolates included a CTX-M-9 and QnrA-producer (ST69), as well as another isolate belonging to the new sequence type ST3555, which harbored both qnrB and bla CTX-M- 8 .
Isolates carrying large numbers of virulence genes (between 9 and 10), namely isolates E38, E69, and HP21, belonged to phylogenetic group D 1 ; nevertheless, they were not among the most disseminated strains.
On the other hand, the 14 isolates belonging to phylogenetic group D 2 and ST405 were VS-.
Recent studies have reported susceptibility results and molecular characterization of E. coli isolates in India, albeit with some differences; while coresistance levels to sulfamethoxazole-trimethoprim and to ciprofloxacin were similar (81.6% and 93.9%, respectively), co-resistance to amikacin was higher (57.1%), and resistance to carbapenems was even detected (20.4%) [24].
In the present study, 41/55 isolates (74.5%) corresponded to phylogenetic groups B2 and D; nevertheless, only 28 (50.9%) could be classified as ExPEC based on VS.Additionally, all of the isolates corresponding to phylogenetic group B2 belonged to ST131.Our results are in accordance with those of Brisse et al., who found that CTX-M-production in ExPEC was exceptional in non-ST131 B2 strains [30].Conversely, Roy et al. recently reported that over a third of CTX-M-15-producing E. coli B2 strains did not belong to ST131 [24].These discrepancies highlight the importance of knowing the local epidemiology, in order to determine the circulating clones in each region.
Multiresistant clone B2 3 -ST131 has been previously associated with a high number of virulence genes [27,31]; in this work, the number of virulence genes associated with this clone was between five and eight.Regarding the clone D 2 -ST405, none of the antibiotic-resistant isolates could be classified as ExPEC (based on VS).Conversely, only a single isolate belonging to CGA was detected, which suggests that this group does not constitute a problem, at least within the studied hospitals.This isolate, however, harbored a large number (10) of virulencerelated genes.Accordingly, this ST69 isolate (HP21) along with isolates E38 and E69 (ST393 and ST38, respectively) carried large numbers of virulence genes and were not disseminated.
In this sense, the multiresistant clone B2 3 -ST131 could have reached an optimal balance between virulence and antibiotic resistance, allowing it to competitively colonize, to cause infection, and to disseminate among different hosts.
On the other hand, we recently reported that a group of non-ExPEC E. coli isolates (based on VS) formed intracellular niches in urothelial cells; this finding was linked to the occurrence or recurrent UTIs [32].Taking those findings into consideration, it is possible that the diversity of virulence-related genes in ExPEC is greater than the VS employed during the present study.
With respect to ESBL production, this is the first study to show a clear predominance in our region of CTX-M-15 in E. coli strains (60%), followed by CTX-M-2 (20%).Until recently, and particularly in Uruguay, the most frequently detected ESBL (in varying proportions) was CTX-M-2 [12,33] .CTX-M-15 was first detected in 2006 in a general hospital [13], and in 2009, this ESBL was detected in the pediatric hospital [11].However, until now, there was no information on the sequence types of ESBL-producing E. coli strains.The presence of CTX-M-15-producing pandemic clones such as ST131 and ST405 has already been reported in Colombia [10] and Argentina (ST131) [9]; nevertheless, there is no information about the sequence types associated with other CTX-M enzymes in our continent.
The CTX-M-9 and QnrA1-producing strain was typed as ST69, which has already been associated with ESBL and non-ESBL-producing strains [34,35].In our study, this was the only isolate that belonged to CGA.Strains belonging to CGA have been widely reported around the world, even in South America [7,18]; nevertheless, in general, there is no information regarding sequence types of CGA strains.
CTX-M-8 was first described in Brazil in 2000, but was only recently detected in other South American countries including Uruguay and Argentina [9,11].Additionally, this ESBL seems to be slowly disseminating in Europe, Africa, and Asia as well [37][38][39].
In our country, the dissemination of CTX-M-15 may be the result of two mechanisms: either the dispersion of successful clones such as ST131 and ST405, or the acquisition of conjugative plasmids featuring multiple addiction systems.
On the other hand, the expansion of bla CTX-M-14 could be mainly due to cross-species dissemination of IncI1 plasmids featuring the pndAC addiction system.In this regard, we recently reported the occurrence of bla CTX-M-14 in an isolate of Salmonella enterica serovar Enteritidis, encoded in a similar plasmid [40].Additionally, in our continent, IncI1 plasmids harboring bla CTX-M-14 and bla CTX-M-19 (both belonging to group 4 in the CTX-M family) have already been described in E. coli strains in our country and in Bolivia [14,36].
Furthermore, only one of the 11 isolates encoding bla CTX-M-2 carried such gene in a conjugative plasmid.Incidentally, such plasmid had no addiction systems.
Recently, Bartoloni et al., in a 20-year study, reported a shift in the epidemiology of ESBLs in Bolivia and Perú, describing an increase in the prevalence of CTX-M-15 and CTX-M-14, and a decrease of CTX-M-2 [41].A similar scenario was also described in Argentina [9].

Conclusions
In this study, we propose some elements that could account for this epidemiologic change.For instance, the association of CTX-M-15 with successful clones or with plasmids carrying addiction systems, and/or the dissemination bla CTX-M-14 in conjugative IncI1 plasmids with pndAC, contrasts with the small proportion of bla CTX-M-2 -encoding conjugative plasmids lacking detectable addiction systems.
Additionally, the link between CTX-M-15 (which confers higher resistance levels to ceftazidime) and PMQR such as aac(6')Ib-cr, creates better chances for co-selection of those isolates carrying both resistance mechanisms, in relation to those carrying only CTX-M-2.
The fact that these changes in the epidemiology of ESBLs go hand in hand with changes in genes conferring resistance to other antimicrobials (such as PMQR) highlights the importance of molecular and epidemiological studies.

Figure 1 .
Figure 1.Main features of 55 E. coli strains studied in this work.

Figure 2 .
Figure 2. PFGE patterns and main features of ESBL-producing ExPEC strains.

Table 1 .
Antibiotics resistance profiles of 55 ESBL producers E. coli and distribution of mainly clones.

Table 2 .
Correlation between MLST and fumC SNPs aimed at the detection of CGA in phylogenetic group D E. coli isolates