Unexplored Genetic Diversity of Multidrug- and Extremely Drug-Resistant Acinetobacter baumannii Isolates from Tertiary Hospitals in Honduras

Isolate collection

Clinical isolates were obtained from patients with A. baumannii infections in five tertiary care hospitals located in two Honduran departments: Cortés and Francisco Morazán. Initial identification of isolates was performed using VITEK 2 (bioMérieux, France). Age, sex, and source of the samples are listed in Supplementary Table S1 (Supplementary Data are available online at www.liebertpub.com/mdr). Antimicrobial susceptibility profiles for all strains were determined using BD EpiCenter™ Microbiology Data Management System V7.22A (BD Diagnostics, France) at Department of Microbiology, Instituto Nacional de Cancerología (Supplementary Table S2). Multidrug resistant (MDR) and extensively drug resistant (XDR) were defined according to the criteria used by Magiorakos et al. ⁸

Plasmid profiles

Plasmid profiles of the isolates were visualized on agarose gels according to the protocol described by Hynes and McGregor. ⁹ Briefly, cell were harvested in mid-logarithmic phase and lysed in the agarose gel well with lysozyme and sodium dodecyl sulfate before electrophoresis. Images were obtained staining the gel with ethidium bromide and photographed in a U.V. transilluminator.

Genome sequencing

DNA was obtained with the Genomic DNA Purification Kit (Thermo Scientific), using the manufacturer's instructions. Sequencing was performed at the Unidad de Secuenciación e Identificación de Polimorfismos, Instituto Nacional de Medicina Genómica (INMEGEN, México). DNA libraries were constructed with a TruSeq DNA Library Preparation Kit v3 and sequenced on an Illumina MiSeq platform with 2 × 300 bp paired-end reads. Genome assemblies were carried out using ABySS 2.0.1 ¹⁰ and VELVET 1.2.10, ¹¹ and the best assembly of each genome was deposited in GenBank and annotated with the NCBI annotation pipeline. The ST assignation of each A. baumannii isolate was obtained from the PubMLST database (https://pubmlst.org). Supplementary Table S3 gives the accession numbers and ST assignation of these isolates. eBURST analysis was performed with goeBURST ¹² and the identification of acquired antibiotic resistance genes was done with ResFinder. ¹³

Selection of homologous groups, model selection, and phylogenetic inferences

For the phylogenetic analysis, we used the genome assemblies of our A. baumannii collection and 41 genomes from NCBI (Supplementary Table S4). To obtain the homologous groups, we first blasted the proteome of each genome against the rest of the proteomes with a cutoff e-value of 1.0e − 30. Homologous protein groups were identified with PanOCT, ¹⁴ requiring that proteins within a homologous group show an aligned region of ≥90% of their lengths and ≥80% identity. For our phylogenetic analysis, we only considered single gene families (SGFs), which have only one gene per genome, and these were aligned with MUSCLE, ¹⁵ specifying 50 iterations. To create a DNA alignment in frame, we used the program TRANALING, ¹⁶ and to detect and discard SGFs with recombination signals, we used PhiPack ¹⁷ setting a p-value cutoff of 0.05. Then, the SGFs that did not show recombination signals were concatenated to form a superalignment. On this alignment, we constructed a maximum likelihood (ML) phylogeny using the generalized time-reversible model (GTR) gamma distributed rate variation among sites (+ G) model, which was the most adequate model according to the jModelTest 2 program. ¹⁸ We ran a nonparametric bootstrap analysis (100 replicates) on the ML phylogeny to establish the support for the clades. To determine the relatedness between the Honduran strains, the number of single-nucleotide polymorphisms (SNPs) was determined from the concatenated alignment in a pairwise manner by MEGA 7. ¹⁹

The collection

Previous studies have suggested that the worldwide increase in A. baumannii infections is the result of the rapid expansion of a limited number of clones. The prevalence of the ICs and specifically, IC-II in Europe, North America, the Middle East, Asia, and Oceania suggests that this situation could also occur in the rest of the world. To explore which clones are circulating in Honduras and which antibiotic resistance genes they carry, we gathered a small collection of nosocomial isolates (16 isolates in total), from March 2015 to June 2016, sampled from five tertiary care health institutions located in two departments in Honduras. It is important to note that our collection was made without any specific criteria regarding the antibiotic resistance profiles of the isolates. The isolates were obtained from several patients and all were initially identified as A. baumannii as per the automated system VITEK 2. Antimicrobial susceptibility profiles revealed that 75% of the strains were XDR and the rest were MDR. All isolates were nonsusceptible to ampicillin, cefazolin, cefoxitin, ertapenem, and fosfomycin c/G6P. Half of the isolates were nonsusceptible to imipenem and 56.25% to meropenem. However, all were susceptible to colistin (Supplementary Table S2).

The isolates came from different sources (Supplementary Table S1); where the most abundant source was endotracheal tubes (36.4%), followed by blood (18.1%) and wounds (13.6%). To avoid working with the same or very closely related clones, the plasmid profiles of all strains were obtained by in situ lysis gel electrophoresis (Supplementary Fig. S1). This analysis showed that with the exception of two isolates (read below), which had the same profile, the plasmid profiles of the isolates varied widely in terms of the number and size of plasmids. Therefore, almost each isolate possessed a distinctive plasmid profile, suggesting that, in general, they did not come from the same nosocomial outbreak. The exceptions were HEU2 and MCR54, collected in different hospitals located in distinct Honduran departments.

Phylogenetic relationships of the isolates

Then, to evaluate the relationships between the Honduran isolates and isolates obtained from other parts of the world, we sequenced the genome of all members of our collection with an Illumina MiSeq platform. We first used the genome sequences to define their STs under the MLST Bartual scheme. ²⁰ We found that five isolates have new allele combinations and, in consequence, curators of the A. baumannii database (PubMLST) provided us with new ST numbers (Supplementary Table S2). Unfortunately, we did not obtain enough sequence information for isolate MCR6739 to establish its ST. In total, nine STs were found in these isolates and the most common were ST_B447 and ST_B758, each one with three isolates. It is important to point out that isolates from the ST_B447 have been recorded in the Czech Republic, United States of America, and Canada, whereas isolates belonging to ST_B758 have been described in Mexico and in Canada.^7,21 Another isolate (MCR10179) belongs to ST_B229, an ST that has been reported in Mexico, Brazil, and the United States of America (PubMLST database). Furthermore, the eBURST analysis showed that the Honduran isolates belong to nine different clonal complexes, none of them related with the clonal complexes embracing the ICs. The most common STs, ST_B447 and ST_B758, belonged to clonal complexes CC447 and CC636, respectively. These observations suggest the existence of new ICs preferentially circulating in North and South America. On the other hand, our collection also contains five isolates with new allele combinations, suggesting that they could be Honduran endemic clones not previously described. We also constructed a phylogenetic tree to establish more accurately the relationships between these isolates. The ML phylogenetic tree (Fig. 1) showed that the Honduran isolates are highly diverse as they are located, forming small clusters, on different parts of the tree. In line with this, we found that the average difference between any two Honduran strains was 1,456 SNPs. Furthermore, from this tree, it is clear that the Honduran isolates do not have a close association with isolates from the ICs. These results are congruent with eBURST analysis, considering that Honduran isolates sharing the same ST were located within the same clade. It is important to point out that despite the small sample studied here, these results indicate an unexpected diversity of A. baumannii in Honduras.

OPEN IN VIEWER

FIG. 1.

Maximum likelihood phylogenetic tree. The tree was constructed on the concatenated alignment of all the single gene families not showing recombination signals. For each isolate the sequence type (Bartual scheme) is shown next to its name. Honduran isolates are indicated in bold letters. Isolates belonging to the International Complex II and I are marked with the two big brackets. For sake of clarity, the bootstrap values are shown just for some nodes; the rest of the nodes are marked with plus signs for those nodes with 70 or higher bootstrap support values, whereas asterisks denote nodes with <70 bootstrap support.

Antibiotic resistance genes

The second major goal of our study was to identify the antibiotic resistance genes present in the Honduran isolates and for that, we analyzed the draft genomes using the ResFinder database. A summary of these results is described in Table 1. The most common resistance genes were those related with beta-lactam antibiotics. As expected, all the isolates from our collection have two intrinsic encoded beta-lactamases: blaADC-25 and blaOXA-51-like alleles (blaOXA-64, blaOXA-65, blaOXA-68, blaOXA-69, blaOXA-132, or blaOXA-180). In two isolates, the blaADC-25 gene is closely associated with an ISAba1 element (HEU2 and MCR54) and in other with an ISAba27 element (MCR26739). To the best of our knowledge, an association between ISAba27 and a blaADC allele has not been reported before. Four isolates HEU55808, HEU5, MCR6056, and MCR10179, obtained from two hospitals, contain a chromosomal encoded blaOXA-23 gene in close proximity with an ISAba1 element, suggesting that this element influences blaOXA-23 expression and potentially increases the minimal inhibitory concentrations for carbapenems.

Two different isolates obtained from the same hospital possess a blaNDM-1, a gene encoding a class B metallo-lactamase (MCR9238 and MCR10172, see Table 1). This gene is usually carried in plasmids; however, considering that we have only draft genomes of these isolates, it was not possible to determine whether blaNDM-1 is located on a plasmid; nonetheless, in both cases, this gene is linked to a Tn125-like element. We also found three isolates (HEU2, MCR54, and IHSS3526) harboring a blaCTX-M15 gene that encodes an extended-spectrum beta-lactamase, and in the HEU2 isolate, a carbenicillin-hydrolyzing beta-lactamase allele (blaCARB-8) was found; notably, to the best of our knowledge, this has not previously been reported in A. baumannii. Finally, we identified in isolate MCR6739 a gene linked to sulbactam resistance (blaTEM-1B).

In addition, we identified several genes related to aminoglycoside resistance and the most common were the strAB genes, which are involved in streptomycin resistance; these genes were found in 5 of the 16 isolates studied here. Another gene involved in aminoglycoside resistance was armA, a 16S-methyltransferase gene that confers resistance to 4,6-disubstituted deoxystreptamines. This gene was found in isolates HEU2 and MCR54. Even more, we found three types of aminoglycoside modifying enzymes: first, aadA1 and aadA2 both encoding aminoglycoside adenyltransferases involved in streptomycin and spectinomycin resistance. Second, two genes encoding O-phosphotransferases: aph(3′)-Ic and aph(3′)-Via. Both genes linked to kanamycin, neomycin, and puromycin resistance, and the last one also to amikacin and gentamicin B resistance. Finally, three genes encoding N-acetyltransferases, involved in different types of aminoglycoside resistance, were found: aac(3)-Ia (gentamicin resistance), aac(3)-IId (gentamicin and tobramycin resistance), and aac(3)-IIa.

Besides the genes involved in resistance to beta-lactam and aminoglycoside antibiotics, we identified genes linked to resistance to other antibiotic families: in three isolates, we found the macrolide resistance genes msrE and mphE. Seven isolates possess a sul2 gene and three possess a sul1 gene, both related to sulfonamide resistance. Four isolates have genes related to tetracycline resistance; two possess a tetB gene and the others two possess tetA or tetG. A couple of isolates have a chloramphenicol resistance-related gene, catA1, and one of these also has a florphenicol resistance gene (floR). Finally, in three isolates, we identified genes involved in trimethoprim resistance: dfrA1 in one isolate and dfrA12 in the other two.