Analysis of class 2 integrons as a marker for multidrug resistance among Gram negative bacilli

Abstract Class 1 and 2 integrons are considered the paradigm of multidrug resistant (MDR) integrons. Although class 1 integrons have been found statistically associated to Enterobacteriaceae MDR isolates, this type of study has not been conducted for class 2 integrons. Escherichia coli and 3 species that were found that harbored more than 20% of class 2 integrons in clinical isolates, were selected to determine the role of intI2 as MDR marker. A total of 234 MDR/191 susceptible non-epidemiologically related isolates were analyzed. Seventy-four intI2 genes were found by PCR and sequencing. An intI2 relationship with MDR phenotypes in Acinetobacter baumannii and Enterobacter cloacae was found. No statistical association was identified with MDR E. coli and Helicobacter pylori isolates. In other words, the likelihood of finding intI2 is the same in susceptible and in MDR E. coli and H. pylori strains, suggesting a particular affinity between the mobile element Tn7 and some species. The use of intI2 as MDR marker was species-dependent, with fluctuating epidemiology at geographical and temporal gradients. The use of intI2 as MDR marker is advisable in A. baumannii, a species that can reach high frequencies of this genetic element.


Introduction
The integron/cassettes system is a successful double component genetic mechanism associated to Lateral Genetic Transfer. The integron is composed by three elements: a gene encoding a site-specific recombinase (intI), a recombination site (attI), and a promoter (P C ) that enables the expression of the gene cassettes inserted in the variable region of the integron [1]. A gene cassette usually contains a single gene flanked downstream by a palindromic short sequence, the attC site, which is a specific recombination site for functional integron integrases. Most gene cassettes have unknown function. However, in nosocomial Gram-negative isolates, more than 130 antimicrobial resistance gene cassettes (ARGC) conferring resistance to several clinical relevant antibiotic families have been identified [2][3][4][5][6][7]. Although class 1 and class 2 integrons are the most frequent integrons among clinical samples [2][3][4][5][6][7][8], some genetic differences have been identified between both classes. Class 1 integrons are usually found in the 20% up to 80% of Enterobacteriaceae and Pseudomonas aeruginosa clinical isolates around the world [2][3][4][5][6][7]. They have been found with many arrays of ARGCs within the variable region [2][3][4]. Furthermore, more than 30 alleles have been described for intI1 [9]. In contrast, class 2 integrons usually possess the typical ARGC array of dfrA1-sat2-aadA1-ybeA-ybfA-ybfB-ybgA. Most intI2 have been found harboring the same allele with a stop codon at position 179 resulting in a nonfunctional IntI2 [8,10]. This phenomenon could be an explanation for the few arrays of ARGC documented in the literature. Geographical distribution of class 2 integrons varies substantially between countries [8,11]. For example, class 2 integrons were found in the 50% of Acinetobacter baumannii strains isolated in Argentina from 1982 to 2007, while no positive intI2 isolate has been found in A. baumannii strains from 29 hospitals in United Kingdom [8,11]. Recent studies identified the emergence of class 2 integrons with novel ARGC arrays in Enterobacteriaceae and A. baumannii strains [5,8,12], suggesting that these species are active reservoirs.
A previous analysis of class 1 integrons in 867 non-repeated isolates (619 strains resistant to at least two antimicrobial agents, and 248 completely susceptible or resistant to only one antimicrobial agent) comprising 8 species of Enterobacteriaceae from 23 European hospitals showed a significant relation between multidrug resistance (MDR) and the intI1 gene, independent of species or origin [3]. To our knowledge, a similar study on the association of class 2 integrons with the MDR phenotype of Gram-negative bacilli has not been yet conducted. There are previous reports on intI2 dissemination performed with MDR strains, but they did not investigate what happens simultaneously in several susceptible bacterial populations [6,8,[13][14][15][16][17]. Since the use of MDR markers can be advisable during outbreaks and/or for epidemiological surveillance [18], our goal was to evaluate if intI2 could be used to early detect MDR isolates.
In previous studies, we have analyzed the frequencies of intI2 among 9 Enterobaceriaceae species, Helicobacter pylori, P. aeruginosa and A. baumannii [8,15,19]. As an epidemiological difference with class 1 integrons, we have found that only 3 species, A. baumannii, Enterobacter cloacae and H. pylori, harbored class 2 integrons in more than 20% of the isolates. In particular, a wide dispersion of class 2 integrons was found in A. baumannii strains, isolated from Argentinian clinical settings during 25 years (1982-2007) [15]. We hypothesize that, in this scenario, the intI2 will behave as intI1 for MDR phenotypes in species where it was frequent (>20%). The aim of this study was to analyze the relationship between the presence of the intI2 within the MDR and the susceptible phenotypes in clinical isolates of the mentioned 3 species. We also include E. coli isolates, since it has been already shown that this species harbors class 2 integrons in several geographical regions [8,16,20,21].

Bacterial isolates
A total of 425 Gram-negative clinical strains, including 195 E. coli, 44 E. cloacae, 137 A. baumannii, and 49 H. pylori isolates, that display susceptible or MDR phenotype (see below), were used. Isolates were obtained from 4 regions of Argentina, separated by more than 1000 km. Five out of 7 hospitals that provided strains for this study, have never been included in previous studies from our laboratory ( Table 1). All isolates met the criteria of nosocomial infection. Until used, strains were frozen at −80 °C in Brain Heart Infusion (BHI) (Difco Laboratories, Detroit, USA), supplemented with 20% (v/v) glycerol.

Detection of intI2
Polymerase chain reaction (PCR) amplification using total DNA was performed using specific primers as previously described [24,25]. DNA products were analyzed by conventional agarose gel electrophoresis and confirmed by sequencing. Multiple sequence alignment analysis, using the CLUSTAL O (1.1.0) tool, was performed with sequences of the intI2 from an isolate and sequences available in the GenBank accession No. AJ001816.

Detection of dfrA1
PCR amplification was used to detect the dfrA1 gene cassette as previously described [8].

Statistical analysis
Frequencies of inti2 in both susceptible and MDR strains were compared [3]. In order to determine if there were significant differences in occurrence of intI2 between isolates with and without MDR strains, two statistical methods were applied, using STATISTICA 8.0 package. Assuming that occurrence of MDR followed a stepwise function with susceptibility adopting a 0-value and multidrug resistance a 1-value, we applied Gamma statistics [26], which is a non-parametric correlation test that it is especially suitable for 0-1 data. A 2 × 2 Fisher exact test was used as additional test for E. coli because the sample size allows us to perform it [26].

Class 2 integrons distribution among the studied isolates
The intI2 was detected by PCR with specific primers in 74/425 of total tested isolates. Taking into account only the MDR isolates, slight changes in the epidemiology of intI2-positive isolates of E. coli have been found (8.3 vs 9.2%) compared to our previous study [8]. In contrast, the MDR isolates of H. pylori, E. cloacae and A. baumannii showed a decrease in the frequency of intI2-positive isolates (37.5 vs 6.6%, 33.0 vs 12.0% and 50.0 vs 43.2%, respectively) [8,19].

intI2 association to MDR phenotype
Two scenarios were found concerning the association of intI2 positive-strains with MDR phenotype. Positive amplifications of E. cloacae and A. baumannii for intI2 were detected only in MDR isolates (Table 1). IntI2 was present in both susceptible and MDR isolates (Table 1) of E. coli and H. pylori. However, there were non-significant differences in occurrence of intI2 between isolates with and without MDR phenotype in this two species. Gamma correlations were non-significant (p > 0.05) for E. coli (b = 0.0019) and H. pylori (b = −0.404), and non-significant differences (p > 0.05) were found when the frequencies of occurrence of the intI2 were compared between susceptible and MDR E. coli isolates using a Fisher Exact test (χ 2 = 0.0001).

Analysis of SXT resistance in E. coli
We conducted an additional analysis with the aim of disentangling the responses of E. coli to SXT related to the presence of intI2. We were especially interested in this antibiotic because we found by PCR with specific primers [8] that the dfrA1 gene cassette was located in first place in the variable region of class 2 integrons in 17 out of our 18 intI2-positive E. coli isolates. Fisher exact test was applied to a 2 × 2 matrix composed by the absolute frequencies of isolates with and without intI2, resistant and susceptible to SXT. A striking presence of intI2 in SXT resistant isolates compared with susceptible isolates was observed. Therefore, response to SXT phenotype alone was different to MDR phenotype in E. coli in regards to the presence of the intI2.

Diversity of intI2 alleles
A putative functional IntI2 protein was found in one E. coli isolate, the MDR E106 strain (GenBank accession No. JN987180). The intI2 from E106 did not possess the common internal stop codon at position 179 [8,10]. Our partial DNA sequence of the intI2 (GenBank accession no. JN987180) has 4 nucleotide changes compared to the intI2 sequence that harbors the internal stop codon, being identical to part of the intI2 from GenBank accession no. EU780012. The remaining intI2 (34/35), showed an internal stop codon at position 179, as it was previously described worldwide [8,10].

Discussion
Class 1 and class 2 integrons are usually detected among MDR clinical isolates, so they are usually referred as MDR integrons [27]. However, several differences among both classes of integrons have been described, such as geographical distribution, level of spread among pathogenic species, and diversity of ARGC in variable regions [2,8,10,11,28]. We report further differences between the two classes of integrons, both in regard to the epidemiology of class 2 integrons over time (in 3 out of 4 species tested), and in the behavior of the intI2 as genetic marker for MDR phenotypes. While frequencies of class 1 integrons remain stable over time [2,7], our study confirmed previous works which suggested that class 2 integrons showed a fluctuating epidemiology [14,19]. We observed a slight increase of intI2 presence in E. coli isolates, and a decrease in the frequency of intI2-positive isolates in H. pylori, E. cloacae and A. baumannii. In particular, the change in the epidemiology of intI2-positive isolates of A. baumannii in Buenos Aires City has been explained by the progressive replacement of the prevalent A. baumannii lineages belonging to CC113 B /CC79 P and CC103 B /CC15 P [29], usually harboring class 2 integrons in their genomes, by new emerging clonal complexes as the widespread CC109 B /CC1 P and the world-wide emerging CC110/ST25 [30]. A. baumannii strains isolated in 2005-2009 from 6 hospitals from Rosario, Argentina, reached 68% of intI2-positive [14], evidencing alternating decreases and increases of this genetic element over time in the same geographical region.
We showed that class 1 and class 2 integrons differed in their behavior as MDR markers. Until now, the role of the intI2 as a MDR marker among Gram negative bacilli has not been established. In Argentina, several bacterial species exhibited low frequency of class 2 integrons [8]. For that reason the study was restricted to E. coli, H. pylori, E. cloacae and A. baumannii strains isolated since 1982 to 2012. We found that the likelihood of finding an intI2 in E. coli and H. pylori isolates was independent of the phenotype of susceptibility or MDR. Instead, there was an evident association of this element with MDR E. cloacae and A. baumannii isolates, taking into account that intI2 was never found in susceptible strains. Since A. baumannii can reach high frequencies (50% and 68% from two independent studies) of class 2 integrons in our geographical region [8,14], the intI2 may be useful as MDR marker in this species.
E. coli showed a particular behavior concerning the presence of class 2 integrons. On one hand, a significant relationship between SXT resistant isolates and the presence of the intI2 was found, probably due to the presence of the dfrA1 gene cassette in the variable region of class 2 integrons, and a putative selection exerted over intI2-positive isolates. On the other hand, as described above, there are similar probabilities of encountering intI2 in susceptible or in MDR E. coli strains. Both features show the different behavior of the two components of the integron/gene cassettes system: antimicrobial resistance gene cassettes (dfrA1 in this case) are selected under antibiotic pressure, while the integron integrase genes are not selected alike by antibiotics. Taking all results as a whole, it is likely that the finding of the intI2 in susceptible or in MDR E. coli and H. pylori isolates is not due to the antimicrobial pressure exerted in the nosocomial or community environment, but to be related to a particular affinity between the mobile element Tn7 and the respective genomes.
We identified one different intI2 allele in an E. coli isolate among the 74 intI2-positive strains. The emergence of this allele is in concordance with previous findings from our geographical region, since at least three intI2 alleles were described in our neighboring country Uruguay [8,12,13].

Conclusion
We found that the use of intI2 as MDR marker is species dependent, a main difference with intI1 gene [3]. Among gram-negative bacilli, our results showed a clear association between the intI2 and MDR A. baumannii or E. cloacae clinical isolates. Although phenotypic susceptibility tests are routinely used for clinical samples, MDR markers based on PCR can provide early data that can be essential for patient therapy. The intI2 gene can be helpful as MDR marker in A. baumannii or E. cloacae outbreaks, and also for epidemiological surveillance.