Genomic Epidemiology of Global Carbapenemase-Producing Enterobacter spp., 2008–2014

We performed whole-genome sequencing on 170 clinical carbapenemase-producing Enterobacter spp. isolates collected globally during 2008–2014. The most common carbapenemase was VIM, followed by New Delhi metallo-β-lactamase (NDM), Klebsiella pneumoniae carbapenemase, oxacillin 48, and IMP. The isolates were of predominantly 2 species (E. xiangfangensis and E. hormaechei subsp. steigerwaltii) and 4 global clones (sequence type [ST] 114, ST93, ST90, and ST78) with different clades within ST114 and ST90. Particular genetic structures surrounding carbapenemase genes were circulating locally in various institutions within the same or between different STs in Greece, Guatemala, Italy, Spain, Serbia, and Vietnam. We found a common NDM genetic structure (NDM-GE-U.S.), previously described on pNDM-U.S. from Klebsiella pneumoniae ATCC BAA-214, in 14 different clones obtained from 6 countries spanning 4 continents. Our study highlights the importance of surveillance programs using whole-genome sequencing in providing insight into the molecular epidemiology of carbapenemase-producing Enterobacter spp.

We performed whole-genome sequencing on 170 clinical carbapenemase-producing Enterobacter spp. isolates collected globally during 2008-2014. The most common carbapenemase was VIM, followed by New Delhi metallo-βlactamase (NDM), Klebsiella pneumoniae carbapenemase, oxacillin 48, and IMP. The isolates were of predominantly 2 species (E. xiangfangensis and E. hormaechei subsp. steigerwaltii) and 4 global clones (sequence type [ST] 114, ST93, ST90, and ST78) with different clades within ST114 and ST90. Particular genetic structures surrounding carbapenemase genes were circulating locally in various institutions within the same or between different STs in Greece, Guatemala, Italy, Spain, Serbia, and Vietnam. We found a common NDM genetic structure (NDM-GE-U.S.), previously described on pNDM-U.S. from Klebsiella pneumoniae ATCC BAA-214, in 14 different clones obtained from 6 countries spanning 4 continents. Our study highlights the importance of surveillance programs using whole-genome sequencing in providing insight into the molecular epidemiology of carbapenemase-producing Enterobacter spp.
T he emergence of carbapenem resistance is a major public health concern because these agents are regarded as one of the last effective therapies available for treating serious infections caused by Enterobacteriaceae (1). Carbapenemases are important causes of carbapenem resistance because they can be transferred between members of the Enterobacteriaceae. The most common carbapenemases among clinical Enterobacteriaceae are the Klebsiella pneumoniae carbapenemases (KPCs) (Amber class A), IMPs, VIMs, New Delhi metalloβ-lactamase (NDMs) (class B or metallo-β-lactamases), and oxacillin (OXA) 48-like (class D) enzymes (2).
Recent surveillance studies have shown that Enterobacter spp. are often the second or third most common Enterobacteriaceae species associated with carbapenemases (3,4). Typically, KPCs are common among Enterobacter spp. from the United States and South America (5). VIMs are often limited to Europe, NDMs to the Indian subcontinent, and OXA-48 to North Africa and the Middle East (5).
Comprehensive global data regarding the different Enterobacter species and molecular epidemiology are currently limited. We designed a study that used short-read whole-genome sequencing to describe the molecular characteristics and international distribution of Enterobacter spp. with different carbapenemases (n = 170) obtained from 2 global surveillance systems during 2008-2014.

Whole-Genome Sequencing
We used the Nextera XT DNA sample preparation kit (Illumina, San Diego, CA, USA) to prepare libraries for sequencing. We multiplexed and sequenced samples on an Illumina NextSeq500 for 300 cycles (151 bp paired-end).

Genomic Analysis
We obtained draft genomes by using SPAdes version 3.10.1 (7). We identified species based on the hsp60 gene sequences (8). We created whole-genome phylogenetic trees, including reference strains for identification of E. cloacae complex (9; online Technical Appendix 1 Table 2).

Phylogenetic Analysis
We created a recombination-free, core single-nucleotide polymorphism (SNP)-based phylogenetic tree and identified SNPs by mapping the reads or aligning the genomes against E. xiangfangensis type strain LMG27195 (9) using the RedDog pipeline (https://github.com/katholt/RedDog). We removed recombination sites according to Gubbins (12) and removed prophages identified by PHAST (13). We included core SNPs and sites that were present in all genomes to create a maximum-likelihood tree using RAxML with the general time-reversible plus gamma substitution model (14). We visualized the tree by using iTOL version 3 (15).
To identify clades within certain sequence types (STs), we used a phylogeny-free population genetics approach of core SNPs, conducting hierarchal clustering analysis with the Bayesian Analysis of Population Structure program (16). We included all 1,048 available Enterobacter spp. genomes in the NCBI Reference Sequence Database (http://www.ncbi.nlm.nih.gov/refseq) as of June 20, 2017. An in silico MLST analysis identified 282 STs from 950 typeable genomes. We included a total of 201 genomes of STs 78, 90, 93, 105, 108, 114, and 171 for the clustering analysis (online Technical Appendix 1 Table 3). For each E. hormaechei subspecies or E. xiangfangensis, the hierarchal Bayesian Analysis of Population Structure clustering analysis (16) was conducted with 3 nested levels with a priori upper bound of the number of clusters between one fourth to one half of the total number of isolates. We defined clades by using the second level of clustering.

Global Distribution of Carbapenemases among Enterobacter spp.
We included a total of 170 carbapenemase-producing Enterobacter strains in the study. The VIMs (VIM-1, 4, 5, 23, and 31; n = 51 [46 were only positive for VIM, and 5 co-produced OXA-48]) were the most common carbapenemase among this collection, followed by NDMs (NDM-1, 6, and 7; n = 43 [41 were positive only for NDM, 1 also co-produced OXA-48. and 1 co-produced KPC-2]); KPCs (KPC-2, 3, 4, and 5; n = 38 [37 were only positive for KPC, and 1 co-produced NDM]); OXA-48 (n = 31 [25 were only positive for OXA-48, 5 co-produced VIM, and 1 co-produced NDM]); and IMPs (IMP-1, 4, 8, 13, and 14; n = 14). Enterobacter spp. with bla VIM were mostly limited to Europe; isolates with bla NDM were present predominantly in the Balkans, India, and Vietnam; isolates with bla KPC were mainly found in the United States and South America; isolates with bla OXA-48 were largely present in North Africa and the Middle East; and isolates with bla IMP occurred mostly in the Philippines, Taiwan, and Australia. The global distribution of isolates from this study was similar to what had previously been reported for other members of Enterobacteriaceae, especially Klebsiella spp. with carbapenemases (5,17).

E. aerogenes Distant from E. cloacae Complex
We identified 10 isolates as E. aerogenes. These results are described in online Technical Appendix 2.

E. xiangfangensis Identified as the Most Common Species
The E. cloacae complex (n = 160) from our study was obtained from intraabdominal (n = 69), urine (n = 56), skin and soft tissue (n = 19), blood (n = 2), and respiratory specimens (n = 14). We  Table 1). E. xiangfangensis was frequent in the Balkans (e.g., Croatia, Romania, and Serbia), whereas E. hormaechei subsp. steigerwaltii was mostly prevalent in Greece and Vietnam (online Technical Appendix 2 Table  1). This overrepresentation was attributable to the presence of particular STs among these species (online Technical Appendix 2 Table 2).  (Figure 1). The remaining species did not contain a dominant ST, and we found new STs among E. cloacae cluster IV (ST832 and ST834) and E. cloacae subsp. cloacae (ST835, ST836, and ST837).
The minor STs, including ST105 and ST108 (both with 6 isolates), were distinguished on the basis of their molecular epidemiology. ST105 from E. xiangfangensis belonged to a single clade and was only present in Croatia, where it contained bla VIM-1 . All the E. hormaechei subsp. oharae isolates belonged to ST108, which was divided into 5 clades; isolates from 2 of the clades (ST108C and ST108D) were from this collection, whereas isolates representing the other clades were from different studies (23; Figure 4). Clade 108C (n = 4) was present in Spain with bla VIM-1 (n = 2) and China with bla IMP-1 (n = 2), and ST108D (n = 2) was found in Australia (with bla IMP-4 ) and Israel (with bla OXA-48 ).

β-lactamases, Antimicrobial Resistance Determinants, and Plasmid Analysis
For each of the 170 isolates, we tabulated the study number, GenBank accession number, species, date, country of isolation, ST, and clade. The β-lactamases, antimicrobial resistance determinants, plasmid replicon types, and plasmid STs are shown in online Technical Appendix 1 Table 1 and online Technical Appendix 2.

Genetic Environments Surrounding the Carbapenemase Genes
We were able to successfully characterize the immediate genetic environments surrounding the carbapenemase  Table 3). We have also described the novel structures found in E. aerogenes (online Technical Appendix 2).
The bla KPC were mainly associated with the Tn4401b isoform (including the 4 novel structures), whereas bla OXA-48 was always associated with Tn1999 (including the 4 novel structures). Isolates with NDM contained ISA-ba125 upstream and ble MBL downstream of the bla NDM , and the bla VIM and bla IMP were situated within diverse class I integrons from various countries (online Technical Appendix 2 Table 2).

Greece, and Italy
In237 was present in ST78 (obtained in 2013) and ST90C (obtained in 2014) from the same institution in Greece. In916 was identified in ST78 (obtained in 2010) and ST114B (obtained in 2014) from the same institution in Italy. In624 was harbored in ST78, ST96, and ST108 from the same institution in Spain (all obtained in 2010). In87 was detected in ST98, ST110, and ST141 from 2 different institutions in Greece (obtained in 2010 and 2014). In4873 was identified in ST114B from 2 different institutions in Greece (obtained in 2013) (online Technical Appendix 2 Tables 1, 2). In110 with bla VIM-1 was present in ST105 from Croatia (obtained in 2013) and ST520 from Spain (obtained in 2012).

Global Distribution of a Common NDM-1 Genetic Structure
The most common genetic structure immediately surrounding the bla NDMs (named NDM-GE-U.S.) in our collection was identical to that previously described on a 140.8 kb IncA/C plasmid (pNDM-U.S.; GenBank accession no. CP006661.1) found in K. pneumoniae ATCC BAA-2146 with bla NDM-1 (24). This bacterium was isolated in 2010 from the urine of a US hospital patient who had previously received medical care in India (25). NDM-GE-U.S., a 3,063bp fragment consisting of ΔISAba125-bla NDM-1 -ble MBL -trpF-dsbC, was present in 16 Table 3).

Discussion
The most common carbapenemase among Enterobacter spp. from our study was VIM, followed by NDM, KPC, OXA-48, and IMP. Carbapenemase-producing Enterobacter spp. was dominated by 2 global species, namely E. xiangfangensis with 1 major clone (ST114) and E. hormaechei subsp. steigerwaltii with 2 major clones (ST90 and ST93). ST114 and ST90 were divided into different clades; some of the clades (e.g., 90C and 114B) were located in certain geographic regions affiliated with specific carbapenemases, whereas other clades (114A and 90B) were distributed globally in association with different types of carbapenemases.
The taxonomy of E. cloacae complex is confusing, and uncertainty still remains about what species make up this complex. In the early 2000s, Hoffmann and Roggenkamp (8) sequenced hsp60 and established 12 genetic clusters (I to XII) in E. cloacae complex. In 2005, the same authors further defined the taxonomy of E. cloacae complex and named cluster VII as E. hormaechei subsp. hormaechei, cluster VI as E. hormaechei subsp. oharae, and cluster VIII as E. hormaechei subsp. steigerwaltii (27). In 2014, Gu et al. (28) described a novel Enterobacter species obtained from sourdough in China named E. xiangfangensis, which clustered closest to E. hormaechei.
The first study that described the global distribution of E. cloacae clones was undertaken by Izdebski et al (18), who performed MLST on 173 cephalosporin-resistant E. cloacae isolates obtained from Israel and several countries in Europe. MLST identified 88 STs among this collection, with ST78, ST114, ST108, and ST66 being the most common and widespread clones. A ST78 isolate was positive for KPC-2, and a ST114 isolate was positive for VIM-1 (18). With the exception of this study from Izedebski et al (18), limited information is available regarding the global distribution of ST93, ST90, ST78, ST105, and ST108 and consists mainly of sporadic reports (29)(30)(31)(32).
Chavda et al. (9) characterized 74 carbapenem-resistant Enterobacter spp. (more than half of the isolates were obtained from New Jersey, USA), and most possessed different bla KPC s, whereas only 2 isolates had bla NDM-1 . E. xiangfangensis also dominated, and ST171 was the most common clone. ST171 was rare in our collection (n = 4) but did show genetic and geographic diversity. ST171 was divided into 3 clades: 171A, 171B, and 171C (online Technical Appendix 2 Figure). Clades 171B and 171C are associated with bla KPC from the United States and United Kingdom (online Technical Appendix 2 Figure). Clade 171B (n = 2) contained bla KPC-2 from Colombia and bla NDM-1 from Guatemala. Clade 171A (n = 1) with bla NDM-1 was obtained from South Africa, and clade 171C with bla KPC-3 was obtained from the United States.
We noted interesting associations and geographic distribution between genetic structures surrounding carbapenemase genes and clades, clones, and species. First, identical genetic structures were situated in various STs within the same or different institutions of the same country (e.g., NDM-GE01 with bla NDM-1 in Vietnam; In87 and In237 with bla VIM-1 in Greece; In916 with bla VIM-1 in Italy; In624 with bla VIM-1 in Spain; and NDM-GE03 with bla NDM-1 in Guatemala). Second, identical genetic structure was present in different STs (ST105 and ST520), from different countries (e.g., In110 with bla VIM-1 in Croatia and Spain). Third, different genetic structures were present in the same STs and clades obtained from different countries (e.g., ST78 with In237 from Greece, ST78 with In916 from Italy, ST78 with In624 from Spain, ST114A with NDM-GE02 from Serbia, and ST114A with pNDM-U.S. from Romania). Last, an identical genetic structure (NDM-GE-U.S.) was found in different global species, STs, and clades.
These associations demonstrate that certain mobile genetic elements with carbapenemase genes have the ability to move between clones and clades of Enterobacter spp. on a global scale. This ability is highlighted by ST78 with bla VIM-1 within different integrons (In237, In916, and In624) that circulate between various countries (Greece, Italy, and Spain). As some STs are introduced into different countries, they apparently acquire the local genetic elements prevalent in that country. Of special concern is the description of a common NDM genetic structure, named NDM-GE-U.S., previously found on pNDM-U.S. and first described in a K. pneumoniae from the United States (24). NDM-GE-U.S. was present in different species, clones, and clades obtained from 6 countries spanning 4 continents. Sequence similarity analysis suggested that it was present on different types of plasmids (pK518_NDM1 and pNDM-HN380) among Enterobacter spp. with bla NDM .
Our results support the current understanding that the carbapenem resistance pandemic is the consequence of circulating clones and the spread of mobile genetic elements. We found that certain clones and clades (ST78, ST90C, ST96, ST114A, ST114C, and ST141) containing particular genetic structures (In87, In624, In916, In237, NDM-GE01, NDM-GE02, and NDM-GE03) and carbapenemases were circulating locally within the same or between different institutions in certain countries (Greece, Guatemala, Italy, Spain, Serbia, and Vietnam). Other global clones and clades (ST90B, ST93, and ST108) contained various genetic structures and carbapenemases.
A limitation of this study was that plasmids harboring carbapenemases were not reconstructed because of the limitations of short-read sequencing (33). The characterization of plasmids is vital to fully comprehend the molecular epidemiology of Enterobacter spp. with carbapenemases, and a follow-up study using long-read sequencing is currently under way. In the meantime, our study highlights the importance of surveillance programs using whole-genome sequencing to provide insight into the characteristics and global distribution of clones and clades as well as their association with mobile genetic elements surrounding the different carbapenemase genes.

About the Author
Dr. Peirano is a research associate at Calgary Laboratory Services and the University of Calgary. Her main research interests revolve around the detection and molecular epidemiology of antimicrobial drug resistance mechanisms among gram-negative bacteria.