Molecular and microbiological report of a hospital outbreak of NDM-1-carrying Enterobacteriaceae in Mexico

Objectives To characterize the microbiological, molecular and epidemiological data of an outbreak of carbapenem-resistant Enterobacteriaceae (CRE) in a tertiary-care hospital in Mexico. Methods From September 2014 to July 2015, all CRE clinical isolates recovered during an outbreak in the Hospital Civil "Fray Antonio Alcalde" in Jalisco, Mexico were screened for antimicrobial susceptibility, carbapenemase production, carbapenemase-encoding genes, and plasmid profiles. Horizontal transfer of imipenem resistance; and clonal diversity by pulsed-field gel electrophoresis (PFGE) and multilocus sequence typing (MLST); as well as biofilm production and the presence of 14 virulence genes were analyzed in selected isolates. Results Fifty-two carbapenem-resistant isolates corresponding to 5 species were detected, i.e., Klebsiella pneumoniae (n = 46), Enterobacter cloacae (n = 3), Escherichia coli (n = 1), Providencia rettgeri (n = 1) and Citrobacter freundii (n = 1) with carbapenemase encoding genes blaNDM-1 (n = 48), blaVIM (n = 3), blaIMP (n = 1) and blaKPC (n = 1) detected in these isolates. The blaNDM-1 gene was detected in plasmids from 130- to 170-kb in K. pneumoniae (n = 46); E. cloacae (n = 3), E. coli (n = 1) and P. rettgeri (n = 1). The transfer of plasmids harboring the blaNDM-1 gene was obtained in eight transconjugants. One plasmid restriction pattern was detected, with the blaNDM-1 identified in different restriction fragments. Predominant clone A of K. pneumoniae isolates archived 28/46 (60%) isolates and belongs to ST392. Besides, ST307, ST309, ST846, ST2399, and ST2400 were detected for K. pneumoniae; as well as E. cloacae ST182 and E. coli ST10. The fimA and uge genes were more likely to be identified in K. pneumoniae carbapenem-susceptible isolates (p = <0.001) and biofilm production was more liable to be observed in carbapenem-resistant isolates (p = <0.05). Conclusions Four Enterobacteriaceae species harboring the blaNDM-1 gene were detected in a nosocomial outbreak in Mexico; horizontal transfer and strain transmission were demonstrated for the blaNDM-1 gene. Given the variation in the size of the plasmid harboring blaNDM-1, complex rearrangements must also be occurring.


Introduction
Nosocomial infections caused by carbapenem-resistant Enterobacteriaceae (CRE) are of particular concern since they can spread rapidly worldwide, and few treatment options remain available for these diseases [1]. Several carbapenem resistance mechanisms have been described in bacteria, and one of the most important in Gram-negative species is the production of carbapenemase enzymes [1,2]. A high diversity of carbapenemases has been reported in Enterobacteriaceae, including the Ambler class A bla KPC , class B metallo-β-lactamases, bla VIM , bla IMP , and bla NDM , and class D carbapenemase bla OXA-48 type [3]. The bla NDM-1 gene is located most frequently on large conjugative plasmids of several incompatibility groups. These plasmids also harbor genes conferring resistance to almost all antibiotics that are used to treat enterobacterial infections [2,3].
In Mexico, the first report of bla NDM-1 was identified in Providencia rettgeri [4], and it was subsequently identified in a Klebsiella pneumoniae pediatric isolate [5]. Furthermore, horizontal transfer and clonal dissemination have been reported in this country, in an outbreak caused by K. pneumoniae, Escherichia coli and Enterobacter cloacae strains harboring a 101-kb IncFII plasmid carrying the bla NDM-1 gene [6].
The analysis of sixteen NDM-1-producing enterobacterial isolates from eight countries showed that the spread of the bla NDM-1 gene is not related to specific clones, specific plasmids, or a single genetic structure [3,7]. The rapid and successful spread of carbapenem-resistant NDM-1-positive organisms may be associated with other antibiotic resistance mechanisms; however, the coexistence of multidrug resistance and virulence mechanisms has also been proposed [8].
Several virulence factors have been described in K. pneumoniae, including adhesins, capsular serotype, iron-scavenging systems, lipopolysaccharide, and biofilm production [9]. The aim of this study was to characterize the epidemiological, microbiological and molecular data of an outbreak of CRE in a tertiary-care hospital in Mexico.

Hospital setting and recognition of outbreak
In September 2014, resistance to carbapenem was detected in five clinical isolates of Enterobacteriaceae in the Laboratory of Bacteriology. The infection control department was alerted, and from there all isolated Enterobacteriaceae between September 2014 and July 2015 were collected for analysis. Before September 2014, we had no carbapenem-resistant isolates.

Ethics statement
The local ethics committee (Comité de Ética en Investigación del Antiguo Hospital Civil de Guadalajara "Fray Antonio Alcalde," Jalisco, Mexico) approved this study with reference number 003/16. Informed consent was waived by the Ethics Committee because no intervention was involved and no patients identifying information was included.

Clinical isolates and patients
Enterobacteriaceae species were identified by Matrix-Assisted Laser Desorption Ionization-Time-of-Flight Mass Spectrometry (MALDI-TOF MS) using the Bruker Biotyper (Bruker Daltonics, Germany) as described previously [10].
Drug susceptibility was performed for all isolates using the VITEK automated system and confirmed by the broth microdilution method. Guidelines of the European Committee on Antimicrobial Susceptibility Testing (EUCAST) version 6.0 were used for colistin and tigecycline [11]. Guidelines of the CLSI were used for amikacin, gentamicin, ertapenem, imipenem, meropenem, ceftriaxone, trimethoprim/sulfamethoxazole, aztreonam, ampicillin, ciprofloxacin, fosfomycin, chloramphenicol [12]. Multidrug-Resistance (MDR) was defined as non-susceptibility to one or more agents of at least three different antibiotic classes [13].
We recovered demographic and clinical data from patients infected with a carbapenem resistant-carbapenemase producer isolates. From each enterobacterial species, one isolate per patient was analyzed.

Plasmid analysis
Plasmid profiles were obtained from all carbapenem-resistant isolates according to the method described by Kieser [20]. In isolates that presented different plasmid profile according to Kieser method, the S1 nuclease assay was performed [21].
Horizontal transfer of carbapenem resistance by bacterial conjugation with E. coli J53-2 as the recipient strain was performed by liquid and solid-phase mating as described [22,23] in isolates that presented a different plasmid profile according to S1 nuclease assay. Transconjugants were selected on Luria-Bertani (LB) agar supplemented with rifampin (100 μg/ml) plus imipenem (2 μg/ml) when the conjugation was unsuccessful the assay was performed in LB agar supplemented with rifampin (100 μg/ml) plus cefoxitin (30 μg/ml). Enzymatic digestion with HinIII (Invitrogen, California, USA) was performed in transconjugants with only one plasmid present. The incompatibility groups were detected by PCR replicon typing in these isolates [24].
Additionally, Southern hybridization with a non-radioactive probe (ECL direct nucleic acid labeling and detection system; GE Healthcare, Piscataway, NJ) of the bla NDM-1 gene was performed in transconjugants and bla NDM-1 positive isolates.

Clonal diversity studies
Clonal diversity was performed by pulsed-field gel electrophoresis (PFGE) and Multilocus sequence typing (MLST) analysis of selected isolates. For PFGE, chromosomal DNA was prepared using the methodology described by Kaufmann [25] with some modifications. Chromosomal DNA from the isolates was digested with 10 U of XbaI (Takara Bio Inc., Shiga, Japan) with following conditions: temperature of 14˚C, the voltage of 6 V/cm, run time of 23 h, and switch time of 1-30 s. PFGE patterns were analyzed visually, and when the restriction patterns presented 100% similarity, the isolates were classified as a clone. When two or three difference in the restriction pattern were detected the isolates were considered as subtypes as suggested by Tenover et al. [26].

Virulence factors
Detection of genes encoding virulence factors and determination of biofilm formation were conducted only for the K. pneumoniae isolates. For a comparison of virulence factors of carbapenem-resistant (bla NDM-1 positive) and carbapenem-susceptible isolates, a group of twentythree carbapenem-susceptible isolates was randomly selected for analysis. The susceptible isolates were obtained in the same period of the carbapenem-resistant isolates (September 2014-June 2015) and were collected from similar specimens and hospital wards to carbapenem-resistant isolates.
Furthermore, semi-quantitative determination of biofilm formation was performed in these isolates (both carbapenem-resistant and carbapenem susceptible) by crystal violet staining as previously described by Bandeira et al., with modifications described by Burmølle et al. [9,31]. The biofilm index (OD 595 /OD 600 ) was used to normalize the amount of biofilm formed to the total cell content of each sample tested. The biofilm production was classified using the biofilm index as non-adherent (<0.90), weakly adherent (>0.90-<1.20) and strongly adherent (>1.20) The cut-off values were defined according to a comparison in the classification with others methodologies previously reported [32]. Staphylococcus aureus ATCC 29213 (a high biofilm producer) and E. coli ATCC 25922 (a low biofilm producer) were used as quality control organisms.
The bla VIM gene was detected in 3/52 (5.6%) isolates belonging to K. pneumoniae (n = 2) and C. freundii (n = 1). The bla KPC gene was detected in one isolate of K. pneumoniae, and the P. rettgeri isolate harbored two carbapenemase genes (bla NDM-1 and bla IMP ). The bla OXA-48 gene was not detected in any of the isolates.
The first carbapenem-resistance isolate was an NDM-1-producing K. pneumoniae (Fig 1). Isolates harboring the bla NDM-1 gene were detected throughout the study.

Demographic and clinical data
The 52 isolates were recovered from 51 patients, with one of them was co-infected with two bla NDM-1 species (E. cloacae and E. coli). The mean age was 44 (0-81) years; 66.7% were male, and the mean length of stay (LOS) before positive culture was 26 (range 3-76) days. Clinical characteristics and outcome of the patients are shown in S1 Table. The most frequent cause of hospitalization was brain injury (20% of patients), followed by lower respiratory tract infection (15%), chronic renal failure, and burn injuries (7.5% each).
From the CRE isolates, 51/52 (98%) were classified as MDR and the single non-MDR isolate corresponded to C. freundii positive for bla VIM carbapenemase. The C. freundii isolate was susceptible to the three carbapenems tested by microdilution but resistant to meropenem and ertapenem using the VITEK system. The other K. pneumoniae isolates that harbored bla VIM presented only resistance to imipenem.
Considering the E. cloacae isolates, all were resistant to all antibiotics tested except for tigecycline (one strain resistant).
Regarding only the NDM-1-producing K. pneumoniae, a high drug resistance was detected. The minimum inhibitory concentration (MIC) range, MIC 90 , and MIC 50 , as well as the percentage of resistant and susceptible isolates to each of the antimicrobial agents tested, are indicated in Table 1.

Plasmid pattern, hybridization and transfer of carbapenem resistance
The plasmid profiles of all the 52 CRE clinical isolates harbored between one and five plasmids with sizes of 40 to170 kb. The plasmid profile of NDM-harboring Enterobacteriaceae isolates (n = 48) was heterogeneous with 15 different sizes of plasmids according to Kieser method and eight different sizes by S1 nuclease assay.
Characteristics of 15 representative isolates of Enterobacteriaceae harboring bla NDM-1 are shown in Table 2. Thirteen isolates (K. pneumoniae, n = 11; and E. cloacae, n = 2) were successful in transfer the resistance with eight transconjugants receiving only one plasmid, (from 130 to 150 kb) and contained the bla NDM-1 gene. Enzymatic digestion revealed one unique Classification of resistance and susceptibility to amikacin, gentamicin, ertapenem, imipenem, meropenem, ceftriaxone, trimethoprim/sulfamethoxazole, aztreonam, ampicillin, ciprofloxacin, fosfomycin, chloramphenicol was based on CLSI interpretive criteria. Classification of resistance and susceptibility to colistin and tigecycline was based on EUCAST interpretative criteria.
restriction pattern, with the bla NDM-1 gene present on a >1.5 kb fragment in seven transconjugants and a smaller 1.0 Kb fragment in one transconjugant (Fig 2). The IncFIIk and IncFIIy incompatibility groups were identified in the eight transconjugants and the E. coli isolate that presented a single plasmid, in 88.9% (8/9) and 22.2% (2/9) of the plasmids, respectively. The Southern hybridization experiments showed the presence of the bla NDM-1 gene in four different plasmids, with sizes from 130 to 170 kb. One bla NDM-1 gene copy was identified in    (19.5%, 9/46) with two or three differences in the restriction pattern in comparison with clone A restriction pattern.
Regarding E. cloacae isolates, they were classified as closely related with only two different bands in the restriction pattern and are considered as subtypes. The Clone A and A1 presented one and two isolates, respectively.

Virulence factors
Virulence factors analyzed in both carbapenem-resistant and carbapenem susceptible isolates showed that the presence of fimA and uge genes was more likely to be detected in carbapenemsusceptible isolates (P = <0.001) ( Table 3).

Discussion
In this study, we characterized the epidemiological, microbiological, and molecular data of an outbreak of CRE in a tertiary-care hospital in western Mexico and detected that the most commonly gene identified was bla NDM-1 in a predominant clone A of K. pneumoniae. The presence of only one carbapenemase type (bla NDM-1 ) involving K. pneumoniae, E. cloacae, and E. coli in four epidemiologically related patients was reported in a tertiary care hospital in Mexico City in 2015 [6], but not comprising the high number of species and isolates reported in this study.
In addition to being detected in K. pneumoniae, E. cloacae, and E. coli, the bla NDM-1 gene has been reported in P. rettgeri isolates [4]. In this species, resistance to carbapenems is rarely described, and when is reported, it is mainly associated with bla NDM-1 [4,33]. In this study, the P. rettgeri isolate presented both bla NDM-1 and the bla IMP gene. The presence of bla IMP in P. rettgeri has been reported only in Japan [34,35], and to the best of our knowledge, the presence of bla NDM-1 and bla IMP in the same isolate has not been reported to date. Unfortunately, the P. rettgeri isolate lost the plasmid carrying bla NDM-1 during experiments, and we were unable to characterize it. Due to the presence of the bla IMP gene, the isolate remained carbapenem-resistant.
Of the several carbapenemases previously described worldwide, the bla KPC has been reported as the predominant carbapenemase gene associated with CRE intrahospital infections [36]. Nevertheless, in the recent years, bla NDM-1 gene has frequently been associated with outbreaks, particularly with strains of K. pneumoniae and E. coli [3].
We detected more than one CRE species and multiple carbapenemases genes in the same hospital in the same period. This diversity has been previously reported, but in countries geographically distant from Mexico (China and Kuwait) [37,38]. Because of this, our report underscores the importance of active surveillance in all enterobacterial species.
The transfer of plasmids of 130 to170 kb carrying bla NDM-1 was demonstrated for K. pneumoniae, and E. cloacae and these experiments partially explain, the high dissemination inter and intra-species observed during the outbreak. In contrast, the inability of conjugation of E. coli and P. rettgeri could explain the lack of dissemination of these species/plasmids during the outbreak. Similar restriction patterns were detected in eight plasmids of similar size (130 to150 kb) with the bla NDM-1 gene present on the same restriction fragment in 7/8 plasmids. These results and the different size of plasmids, strongly suggest rearrangements of plasmids during the short period of this outbreak. Rearrangements of plasmids have been previously reported for plasmids harboring bla VIM gene [39], but to our knowledge, rearrangements in plasmids encoding bla NDM-1 have not been previously described.
In our study, the 19.6% isolates were resistant to tigecycline. The high resistance detected is a point of concern because the SENTRY Antimicrobial Surveillance Program reported in 2016 that only 2.6% of CRE from Latin America presented tigecycline resistance [40], and by these new findings, we may infer that tigecycline resistance is increasing.
Colistin resistance was also detected (4%), and this is now a serious global menace as previous reports from India and UK showed percentages of resistance of 6% and 11%, respectively [41]. This study is the second report of colistin-resistant Enterobacteriaceae in Mexico [42]. The molecular mechanism of resistance to colistin presented in these colistin-resistant isolates was not acquired due to mcr gene was not detected, and additional analysis is ongoing in our group to evaluate the mechanism(s) involved.
E. cloacae ST182 harboring bla NDM-1 has been reported in Mexico and Finland [6,46]; the ST10 of E. coli has only been reported in isolates harboring ESBL genes and not carbapenemase genes [47].
The combination of clonal expansion and horizontal gene transfer demonstrated in this study has been described in Mexico, UK, and Chennai, India [6,41]. In contrast, isolates from Haryana, India, showed an apparent clonal expansion by the demonstration of two types of predominant plasmids [41].
In our study, we evaluated 14 virulence genes in all carbapenem-resistant and in selected carbapenem-susceptible K. pneumoniae isolates, and we found that the frequency of virulence genes was similar to reported when 6 K pneumoniae KPC +, ST258 were analyzed [30].
The virulence gene distribution between both groups was similar except for the fimA and uge genes. The presence of fimA and uge genes was more frequently detected in carbapenemsusceptible K. pneumoniae isolates (P <0.001). The type 1 fimbriae, encoded by fimA, contribute to the invasion of bladder cells and biofilm formation. Despite the low frequency of the fimA gene in the carbapenem-resistant isolates, these strains were capable of high biofilm production.
In this study, we confirmed that biofilm production is higher in the carbapenem-resistant isolates than in the carbapenem-susceptible isolates as previously reported [8], and the presence of biofilm may contribute to drug resistance. However, more studies are required to define the impact of virulence genes. This studies had several limitations. First, there was a lack of information about the travel of patients or healthcare workers to places where NDM-1-producing strains are endemic. However, bla NDM-1 strains been reported in Mexico previously, and the dissemination could be the result of regional transmission. Another significant limitation of our study is the absence of analysis of CRE carriers in the hospital. The above limitations could be resolved in future studies of CRE in this region.
The results obtained in this study indicate that bla NDM-1 was disseminated horizontally among different species in a tertiary care Hospital in Mexico, also with proof of strain spread predominantly of K. pneumoniae ST392. We have provided evidence of plasmid transfer but, given the variation in plasmid sizes, complex rearrangements must also be occurring. In this analysis, the presence of other carbapenemase genes encoding bla VIM , bla KPC and bla IMP were sporadic.
Supporting information S1