Genome Sequencing of Four Multidrug-Resistant Enterobacter aerogenes Isolates from Hospitalized Patients in Brazil

1 Laboratorio de Quimica e Funcao de Proteinas e Peptideos, Centro de Biociencias e Biotecnologia, Universidade Estadual do Norte Fluminense Darcy Ribeiro, Campos dos Goytacazes, Brazil, National Center for Biotechnology Information, National Library of Medicine, National Institutes of Health, Bethesda, MD, USA, 3 Laboratorio de Bacteriologia, Unidade Laboratorio de Analises Clinicas, Hospital de Clinicas, Universidade Federal do Parana, Curitiba, Brazil, 4 Faculdades e Instituto de Pesquisa Pele Pequeno Principe, Curitiba, Brazil

Hospital outbreaks due to E. aerogenes have been reported in Europe since the mid-1990s and have been related to an epidemic extended-spectrum beta-lactamase (ESBL) clone carrying the bla TEM-24 gene (Bosi et al., 1999;Galdbart et al., 2000;Dumarche et al., 2002;Salso et al., 2003). Constitutive AmpC a (beta-lactamase) overexpression is the major cephalosporin resistance mechanism in Enterobacter spp., happening more often than the acquisition of ampC genes through the activity of mobile genetic elements (Perez-Perez and Hanson, 2002). Further, the increased expression of ESBLs led to the adoption of carbapenems to treat E. aerogenes infections (Perez-Perez and Hanson, 2002;Davin-Regli and Pagés, 2015).
Carbapenems have been considered the antibiotic of choice for treating patients infected with ESBL-producing Enterobacteriaceae (Vardakas et al., 2012). However, emergence of carbapenem-resistant E. aerogenes isolates during carbapenem therapy of hospitalized patients (Chen et al., 2008), cases of sepsis due to carbapenem-resistant E. aerogenes after liver transplantation (Chen et al., 2009) and hospital disseminations of carbapenemase-producing E. aerogenes have been recently reported in several countries (Lavigne et al., 2013;Kuai et al., 2014;Qin et al., 2014;Pulcrano et al., 2016). Acquisition and expression of carbapenemases constitute the primary mechanism underlying the development of carbapenem resistance (Rapp and Urban, 2012). Nevertheless, loss of function mutations in porin genes and increased expression of efflux pumps or their regulators have also been associated with carbapenem resistance profiles (Pradel and Pages, 2002;Yigit et al., 2002;Bornet et al., 2003).
Broad-spectrum antimicrobial-resistant E. aerogenes isolates, some resistant to carbapenems (Qin et al., 2014) and last-line therapeutic options such as colistin (Diene et al., 2013), have been responsible for outbreaks in the United States of America (Wong et al., 2010), China (Qin et al., 2014), Japan (Goshi et al., 2002), France (Diene et al., 2013), Fiji (Narayan et al., 2009) and Brazil (Tuon et al., 2015). However, few reports related to E. aerogenes epidemiology, pathogenesis, and molecular characterization have been conducted in Brazil. Recently, five panresistant E. aerogenes isolates were reported in a Brazilian teaching hospital, resulting in a high mortality rate (37.5%) among 16 infected patients (Tuon et al., 2015). We have observed high prevalence (>20%) of ESBL-producing Enterobacteriaceae spp., in particular K. pneumoniae and E. aerogenes, in our hospital since 2003 (Nogueira Kda et al., 2014. Previous molecular characterization studies conducted over 5 years in our hospital showed high prevalence of bla CTX-M2 , -M15 , -M59 , bla SHV-2 and bla TEM genes in Enterobacter spp. isolates (Nogueira Kda et al., 2014. The presence of bla PER-2 was also detected in a few isolates (Nogueira Kda et al., 2014. Given the severity of E. aerogenes infections and the urgent need to better understand the genetic basis of multidrug resistance, here we report the whole-genome sequencing and resistance gene repertoire of four multidrugresistant E. aerogenes isolated from hospitalized patients in Brazil.

Sample Collection and Identification
E. aerogenes isolates C10, D2, D3, and E9 were obtained between 2006 and 2012 from patients hospitalized in wards or intensive care units at the Hospital de Clínicas of the Universidade Federal do Paraná (Curitiba, Brazil). The main selection criterion for genome sequencing was the MDR phenotype, particularly in carbapenem resistant isolates. The negative laboratory tests for carbapenemases were also taken into account, as divergent enzymes or alternative resistance mechanisms could be relevant to the observed MDR phenotypes. C10 and D2 samples were isolated from different body sites of the same patient. Isolates were grown in selective medium with an ertapenem disk (10 ug) and stored at −80 • C in trypticase soy broth containing glycerol 15%. Identification of isolates was performed using Vitek R 2 Compact (BioMérieux S.A., Marcy l'Etoile, France) and by mass spectrometry using Microflex LT instrument (Bruker Daltonics, Bremen, Germany). This study was carried out in accordance with the Brazilian legislation and was approved by the Institutional Ethics Review Board of the Hospital de Clínicas, Universidade Federal do Paraná (IRB#: 2656.263/2011-11). Our study involved only bacterial isolates and no human specimens were analyzed or stored. Further, we used no patient information other than the anatomical sites from where the isolates were collected. Therefore, the same Ethics Review Board exempted us from obtaining informed consent forms.

Resistance Profile Analysis
Antimicrobial Susceptibility Testing Isolates were tested by agar dilution against 15 antibiotics according to the Clinical and Laboratory Standard Institute guidelines (CLSI, 2015a). Minimal inhibitory concentration (MIC) was interpreted as recommended by CLSI standards (CLSI, 2015b). Polymyxin, tigecycline and fosfomycin breakpoints were interpreted using EUCAST standards (Eucast, 2016). Modified Hodge test (MHT), double-disk synergy and hydrolysis assay were performed to determine the carbapenem resistance phenotypes, as previously described (Carvalhaes et al., 2010;Eucast, 2013).

Genome Sequencing, Assembly, and Annotation
Genomic DNA was extracted using DNeasy 96 Blood & Tissue Kit (QIAGEN Silicon Valley, Redwood City, USA). DNA quality was assessed using a Bioanalyzer 2100 system (Agilent Technologies, Santa Clara, USA). DNA quantification was performed using Qubit (Thermo Fisher Scientific Inc., Waltham, USA). Illumina sequencing libraries with an average fragment size of 550 bp were prepared using Illumina TruSeq DNA PCR-free LT Kit (Illumina Inc., San Diego, USA). Wholegenome sequencing of paired-end (PE) libraries was performed using a HiSeq 2500 instrument in RAPID run mode (Illumina Inc., San Diego, USA) at the Life Sciences Core Facilities of the State University of Campinas (São Paulo, Brazil). Qualitybased trimming and filtering was performed using Trimmomatic version 0.32 (Bolger et al., 2014). PE reads were assembled de novo using Velvet version 1.2.10 (Zerbino and Birney, 2008) and contigs were scaffolded using SSPACE version 3.0 (Boetzer et al., 2011). Gene predictions and annotations were performed using NCBI Prokaryotic Genome Automatic Annotation Pipeline (PGAAP; Angiuoli et al., 2008).

Identification of Antibiotic Resistance Genes
Antibiotic resistance-related genes were predicted using the ResFinder database version 2.1 (Zankari et al., 2012) with the following parameters: "all databases" were used for antimicrobial configuration, type of reads as "assembled genomes/contigs" and thresholds of 98 identity and 80% coverage between sequences. This dataset of resistance genes was complemented with BLASTp searches against the ARDB (Antibiotic Resistance Genes Database) version 1.1 (Liu and Pop, 2009) using "resistance gene complete" database, 40% identity and e-value of 0.0001.

Resistance Profiles
All isolates showed MDR profile and had increased MIC for at least one carbapenem. Information regarding collection date and site, clinical setting, PFGE profile and antimicrobial resistance profiles of each isolate are available in Table 1. Among the four analyzed samples, C10 and D2 were isolated from different body sites of the same patient within a short period of time (a month) and belong to the same PFGE profile. These genomes allow one to analyze the possible genome plasticity between the isolates. D3 and E9 samples were isolated from two patients with an interval of collection date greater than 5 years. D3 and E9 were also interesting because of their sensitivity to meropenem and resistance to ertapenem and imipenem. Surprisingly, E9 showed resistance to carbapenems but not to 3rd (ceftazidime and cefotaxime) and 4th generation (cefepime) cephalosporins ( Table 1). All isolates possessed bla AmpC and bla TEM , as detected by PCR. The gene bla CTX-M2 was found in all isolates except E9. Phenotypic tests (i.e., Modified Hodge test and double-disk synergy) to detect carbapenemases were positive for C10, D2, and E9. However, no class A, B, and D carbapenemase encoding genes were detected by PCR. All isolates tested negative in carbapenem hydrolysis assays.

Genomic Features
We obtained between 16,841,714 and 25,138,390 150 bp PE reads per library. After genome assembly, 5,833,521 bp were assembled    predict 5,363, 5,380, 5,067, and 5,129 protein-coding sequences in each of the genomes listed above, respectively. Genomic features of the four sequenced genomes are summarized in Table 1.

Antibiotic Resistance Genes
A total of 18 enzymes related to antibiotic resistance were identified using ResFinder, ARDB and PGAAP ( Table 2). All isolates harbor genes related to: (i) aminoglycoside resistance (genes aacA4 and aadA); (ii) beta-lactam resistance, including genes belonging to class A beta-lactamases (TEM family), class B beta-lactamases (Ribonuclease Z), class C beta-lactamases (CMY/LAT/MOX/ACT/MIR/FOX family) and class D betalactamases (OXA-9); (iii) bacitracin resistance (gene bacA), and (iv) sulphonamide resistance (gene sul1; Table 2). Genes sul2 and rmtD were only identified in E. aerogenes D3. The gene sul2 has been implicated on sulphonamide resistance for inducing high expression levels of the enzyme dihydropteroate synthase (Sköld, 2001), while rmtD has been related to aminoglycoside resistance and this variant was identified for the first time in South America in a P. aeruginosa isolate in 2005 (Doi et al., 2007). Interestingly, E. aerogenes D3 was isolated in 2006, indicating that this variant has spread amongst Enterobacteriaceae in Brazil since its first report (Doi et al., 2007). Although the four isolates showed carbapenem-resistance, no carbapenemase gene was identified using molecular detection or in silico analysis. Hence, it is likely that these isolates employ alternative mechanisms to counter carbapenem effects. Various multidrug efflux transporters were found in the genomes described here (Table 2). They belong to four superfamilies: the major facilitator superfamily (MFS), multidrug and toxic compound extrusion (MATE), ATP-binding cassette (ABC) and resistance-nodulation-cell division (RND). RND type of transporters has been often associated with multidrug resistance of Gram-negative bacteria (Nikaido, 1998). In particular, the RND type genes forming the AcrA-AcrB-TolC efflux pump were found in multiple copies in our isolates ( Table 2). Experimental evolution studies of E. aerogenes under successive imipenem exposure reported alterations in membrane permeability with complete loss of porins (e.g., Omp35 and Omp36) and overexpression of AcrAB-TolC efflux pumps (Bornet et al., 2003;Thiolas et al., 2005;Lavigne et al., 2012). As a result of efflux pump expression, the E. aerogenes isolates showed resistance to carbapenems and other antibiotics, especially fluoroquinolones (Bornet et al., 2003;Thiolas et al., 2005;Lavigne et al., 2012). Given the multiple copies of genes encoding efflux pumps in our isolates, it is possible that an increased expression of AcrAB-TolC efflux pumps could contribute to the observed carbapenemresistant profiles.
E. aerogenes is an emergent nosocomial pathogen with a diversity of mechanisms to circumvent antimicrobial activity. Here we reported the phenotypic screens, genome sequencing, and prediction of putative resistance gene repertoires of four multidrug-resistant E. aerogenes isolated between 2006 and 2012.
The data reported here may help understand the biochemistry, evolution, and epidemiology of this important pathogen. The material provided in this work may be used in future comparative genomics and molecular epidemiology studies aiming to clarify the resistance profiles and dynamics of multidrug-resistant Enterobacteriaceae species.

DATA ACCESS
The genome sequence of E. aerogenes C10, E. aerogenes D2, E. aerogenes D3 and E. aerogenes E9 have been deposited in DDBJ/EMBL/GenBank under the accession numbers LUTZ00000000, LSOH00000000, LUTT00000000, and LULD00000000, respectively. Data are available in FASTA, annotated GenBank flat file and ASN.1 formats. The respective genome versions described in this paper are LUTZ01000000, LSOH01000000, LUTT01000000, and LULD01000000. Sequencing reads (fastq format) of each isolate were deposited in Sequence Read Archive (SRA) under the accession numbers SRP083774 (E. aerogenes C10), SRP083784 (E. aerogenes D2), SRP083785 (E. aerogenes D3), and SRP083786 (E. aerogenes E9). Users can download the data for research purposes, citing the present manuscript as original reference.

AUTHOR CONTRIBUTIONS
AG, NV, JP, LD, and TV conceived the idea and designed the study. JP performed the sample collections and wet lab experiments. AG and NV carried out the genome analysis. AG, NV, JP, LD, and TV interpreted the data and wrote the manuscript. All authors have read and approved the final version of this manuscript.