Molecular Epidemiology of Global Carbapenemase-Producing Citrobacter spp. (2015–2017)

Citrobacter spp. are gaining recognition as important causes of hospital-acquired infections in humans. Among Citrobacter spp., carbapenemase-producing strains are cause of utmost concern to health care services globally due to their ability to resist therapy with virtually any beta-lactam antibiotic. ABSTRACT The emergence of carbapenem resistance is a significant public health concern. The rate of infections caused by carbapenemase-producing Citrobacter spp., particularly C. freundii, is increasing. Concomitantly, comprehensive global genomic data on carbapenemase-producing Citrobacter spp. are scarce. We used short read whole-genome sequencing to describe the molecular epidemiology and international distribution of eighty-six carbapenemase-producing Citrobacter spp. obtained from two surveillance programs (2015 to 17). The common carbapenemases were KPC-2 (26%), VIM-1 (17%), IMP-4 (14%) and NDM-1 (10%). C. freundii and C. portucalensis were the principal species. C. freundii consisted of multiple clones obtained mainly from Colombia (with KPC-2), the United States (with KPC-2, -3), and Italy (with VIM-1). Two dominant C. freundii clones were identified: ST98 was linked with blaIMP-8 from Taiwan and blaKPC-2 from the United States, and ST22 was linked with blaKPC-2 from Colombia and blaVIM-1 from Italy. C. portucalensis consisted mainly of two clones: ST493 with blaIMP-4 which was limited to Australia, and ST545 with blaVIM-31 which was limited to Turkey. Class I integron (In916) with blaVIM-1 was circulating between multiple sequence types (STs) in Italy, Poland, and Portugal. In73 with blaIMP-8 was circulating between various STs in Taiwan, while In809 with blaIMP-4 was circulating between different STs in Australia. The global carbapenemase-producing Citrobacter spp. population is dominated by diverse STs with different characteristics and varied geographical distribution and thus requires continued monitoring. Ongoing genomic surveillance should use methodologies able to distinguish between C. freundii and C. portucalensis. IMPORTANCE Citrobacter spp. are gaining recognition as important causes of hospital-acquired infections in humans. Among Citrobacter spp., carbapenemase-producing strains are cause of utmost concern to health care services globally due to their ability to resist therapy with virtually any beta-lactam antibiotic. Here, we described the molecular characteristics of a global collection of carbapenemase-producing Citrobacter spp. C. freundii and C. portucalensis were the most common species among Citrobacter spp. with carbapenemases from this survey. Importantly, C. portucalensis was misidentified as C. freundii when using Vitek 2.0/MALDI-TOF MS (matrix-assisted laser desorption/ionization–time of flight mass spectrometry) phenotypic identification, which has important implications for future surveys. Among C. freundii, we identified two dominant clones: ST98 with blaIMP-8 from Taiwan and blaKPC-2 from the United States, and ST22 with blaKPC-2 from Colombia and blaVIM-1 from Italy. As for C. portucalensis, the dominant clones consisted of ST493 with blaIMP-4 from Australia and ST545 with blaVIM-31 from Turkey.

IMPORTANCE Citrobacter spp. are gaining recognition as important causes of hospitalacquired infections in humans. Among Citrobacter spp., carbapenemase-producing strains are cause of utmost concern to health care services globally due to their ability to resist therapy with virtually any beta-lactam antibiotic. Here, we described the molecular characteristics of a global collection of carbapenemase-producing Citrobacter spp. C. freundii and C. portucalensis were the most common species among Citrobacter spp. with carbapenemases from this survey. Importantly, C. portucalensis was misidentified as C. freundii when using Vitek 2.0/ MALDI-TOF MS (matrix-assisted laser desorption/ionization-time of flight mass spectrometry) phenotypic identification, which has important implications for future surveys. Among C. freundii, we identified two dominant clones: ST98 with bla IMP-8 from Taiwan and bla KPC-2 from the United States, and ST22 with bla KPC-2 from Colombia and bla VIM-1 from Italy. As for C. portucalensis, the dominant clones consisted of ST493 with bla IMP-4 from Australia and ST545 with bla VIM-31 from Turkey.
T he continuous rise in infections caused by carbapenemase-producing Enterobacterales (CPE) is a cause of utmost concern to health care services globally (1). Infections caused by CPE are linked to high mortality rates in humans because these isolates are resistant to virtually all beta-lactam antibiotics and often harbor genetic determinants that confer resistance against other drug classes such as aminoglycosides and fluoroquinolones (2). CPEs often contain various plasmid-borne resistance genes, including KPCs, NDMs, and OXA-48like carbapenemases (3).
Among the CPE, Klebsiella pneumoniae and Escherichia coli represent the main clinical load of infections (3). Nevertheless, Citrobacter spp. are gaining recognition as important causes of nosocomial infections (4). Recent evidence has suggested that the rate of infections caused by carbapenemase-producing Citrobacter spp., particularly C. freundii, is increasing (5,6). In general, carbapenemase-producing Citrobacter spp. populations are genetically diverse (6), but a few sequence types (STs) such as ST22 and ST19 have recently emerged and demonstrated the potential to become dominant clones in health care settings (5).
To date, information regarding the molecular epidemiology of carbapenemase-producing Citrobacter spp., including its genetic diversity and genetic mechanisms of carbapenem resistance, is limited to regional or countrywide studies. From an epidemiological standpoint, such assessments are valuable to specific geographical areas, but fall short in providing a broader overview from a global perspective, which includes a description of global dominant STs and genetic mechanisms of carbapenem resistance. Here, we described the molecular characteristics of a global collection of carbapenemase-producing Citrobacter species isolates obtained systematically from two global surveillance programs. We report the geographical distribution of STs and provide an in-depth assessment of molecular mechanisms associated with carbapenem resistance in carbapenemase-producing Citrobacter spp.
Carbapenemase gene flanking regions. Due to the limitations of short-read sequencing (7), analyses of the immediate carbapenemase gene flanking regions and plasmids harboring carbapenemase genes were insufficient for some of the isolates. We were able to characterize the immediate carbapenemase gene flanking regions for 16/29 isolates with bla KPC , 13/13 isolates with bla NDM , 8/8 isolates with bla OXA-48 , and all the class I integrons containing bla IMP (n = 20) and bla VIM (n = 21).
The bla KPC-2 (n = 9) was situated within Tn4401a in a single C. koseri isolate and within Tn4401b in 8 Citrobacter species isolates which belonged to different species and STs. The bla KPC-3 (n = 7) was situated within Tn4401a (n = 3) and Tn4401b (n = 4) that belonged to different Citrobacter species and STs. The bla NDM were located on truncated Tn125 elements and situated downstream of ISAba125 among all the isolates. The bla NDM upstream regions showed significant diversities with various IS family insertions (e.g., IS3, IS5, IS6, IS30 and IS91). All the bla OXA-48 were situated within Tn1999.2.
The VIM-31-positive isolates (n = 5) were obtained from Turkey and consisted of C. portucalensis ST545. The bla VIM-31 gene was situated in the class 1 integron In669 that also contained aacA4. The VIM-4 (n = 1) was present in C. portucalensis ST63 from Hungary and was situated in In1433 which also contained aacA7 and smr2 (Fig. 3).
VIM and IMP metallo-b-lactamases are rare among CPEs, especially within Klebsiella spp. and E. coli (9). CPE isolates with bla IMP are endemic in Japan, Taiwan, and Australia (3, 10) while CPE with bla VIM are mainly found in Italy and Greece (4, 11). Among the Citrobacter spp. from this study, VIMs and IMPs were the 2nd and 3rd most common carbapenemases. VIMs were found in Citrobacter spp. from Turkey and Italy while IMPs were limited to Taiwan and Australia. The Turkish isolates (n = 5) were positive for bla VIM-31 and belonged to a single clone (C. portucalensis ST545) that contained the class I integron In669. Italian isolates (n = 11) were positive for bla VIM-1 , belonged to multiple C. freundii STs, and contained In916. In916 was also found in Citrobacter spp. from Poland (ST98) and Portugal (ST95). This indicated that identical class I integrons with bla VIM-1 are circulating between different STs in Italy, Poland, and Portugal. A similar scenario was previously reported with Enterobacter spp. harboring bla VIM-1 within In916 from Spain, Greece, and Italy (12). The IMP-8 isolates from Taiwan belonged to ST8, ST98, and ST528 that harbored the identical class I integron In73. The Australian IMP-4-positive Citrobacter species isolates belonged to 2 clones, namely, ST396 and ST493, which contained the identical class I integron In809. This in-depth characterization of AMR isolates showed how genomic surveillance using whole-genome sequencing offered an unprecedented level of detail that clarified the underlying differences among CPEs from various countries. The geographical distribution and carbapenemase types of the Citrobacter spp. from this study were different from those of carbapenemase-producing E. coli obtained from the same surveillance programs over identical periods (2). The most common carbapenemases from the E. coli survey were OXA-181 and NDM-5, while VIMs and IMPs were rare. The Citrobacter spp. results from this survey showed some similarities with carbapenemase-producing K. pneumoniae (13) and Enterobacter cloacae complex (12) obtained from the same surveillance programs. The K. pneumoniae with KPC-2 was obtained mainly from Colombia and the isolate with KPC-3 was mainly from the United States. The E. cloacae complex isolates consisted mainly of VIM-1 isolates obtained from Italy and Greece. Citrobacter spp., K. pneumoniae (1), and Enterobacter spp. (14) are typically hospital pathogens, while E. coli is typically a community pathogen (15). This fact could be partly responsible for the different carbapenemase types and geographical distributions among these species.
C. freundii (60%) and C. portucalensis (23%) were the principal species among Citrobacter spp. with carbapenemases from this survey. C. portucalensis was first reported in 2017 from Portuguese aquatic samples (16) and later from Nigerian leafy vegetables (17). This bacterium is closely related to and often misidentified as C. freundii, especially when using phenotypic identification systems such as Vitek 2.0 and MALDI-TOF MS (18). C. portucalensis with bla CTX-M-15 (19) and with bla NDM-1 (18) were previously reported from Brazil (two environmental isolates) and China (one clinical isolate), respectively. In our study, MALDI-TOF MS identified C. portucalensis as C. freundii. C. portucalensis was mainly found in Australia and Turkey, but also in Egypt, Hungary, Italy, Jordan, and the Philippines. The C. freundii population was linked with various STs obtained from Colombia, the United States, and Italy. The population structure of C. portucalensis consisted mainly of two clones, namely, ST493 with IMP-4 (Australia) and ST545 with VIM-31 (Turkey). Our results indicate that the geographical and underlying molecular epidemiology is different between carbapenemase-producing C. freundii and C. portucalensis isolates. Therefore, future surveys should use methodologies that can distinguish between C. freundii and C. portucalensis isolates. This will emphasize the role of C. portucalensis in the global dissemination of carbapenemases among Citrobacter spp. We recommend that identification systems in clinical laboratories be updated to include the routine identification of C. portucalensis. This will enable the clinical microbiology community to determine the overall prevalence, clinical significance, and geographical distribution of this newly described Citrobacter species.
High-risk AMR clones such as K. pneumoniae ST258 (3) and E. coli ST410 (20) contribute significantly to the global spread of CPE. We observed a high diversity of carbapenemase-producing Citrobacter spp. clones that was represented by 37 STs, including 4 dominant clones representing .40% of the total Citrobacter spp. population. C. freundii ST98 and ST22 were global, whereas C. portucalensis ST493 and ST545 were limited to Australia and Turkey, respectively. The dominant STs were linked to different carbapenemase genes. The global carbapenemase Citrobacter spp. population is dominated by diverse STs with different characteristics and varied geographical distributions. ST22 and ST98 isolates with different carbapenemases (VIMs, KPC-2, NDM-1, OXA-48) were previously reported from Spain (21,22), Tunisia (23) and Germany (24).
We provide pertinent information about the global distribution of carbapenemases and population structure among a large collection of carbapenemase-producing Citrobacter spp. This study has some limitations. We used short-read sequencing to characterize our collection, which limited our ability to fully reconstruct flanking regions and plasmids harboring carbapenemases. Additionally, many countries contributed few isolates, and statistical approaches to deal with clustered data failed to estimate country-adjusted rates of carbapenemases among Citrobacter spp. Countries contributing few isolates may not be fully representative of the population of carbapenemases in Citrobacter spp. from that region. The global distribution and prevalence of carbapenemases will be influenced by the clonal dissemination of isolates during nosocomial outbreaks that might have occurred during the surveillance period.
In summary, carbapenemase-producing Citrobacter spp. are a highly diverse group of bacteria that contain 4 dominant global clones. Carbapenemase-producing Citrobacter spp. are disseminated either clonally and polyclonally depending on the species, ST, and geographical location. We also demonstrated that the newly described C. portucalensis has a global distribution and is likely involved in some of the clonal outbreaks of carbapenemase-producing Citrobacter spp. Our findings greatly contribute to global and local surveillance activities, particularly those in lower-and middle-income countries. Our findings underline the increasing importance of Citrobacter spp. in the global dissemination of carbapenemases.

MATERIALS AND METHODS
Bacterial isolates. We obtained isolates from two global surveillance programs (2015 to 2017), namely, the Merck Study for Monitoring Antimicrobial Resistance Trends (SMART) and the International Network for Optimal Resistance Monitoring (INFORM) programs. The SMART program includes Gram-negative isolates from intra-abdominal, lower respiratory tract, and urinary tract infections obtained from 55 countries, whereas the INFORM program collects isolates from blood, intra-abdominal, lower respiratory tract, skin, soft tissue, and urinary tract infections obtained from 42 countries. Participating countries are listed in the Appendix (Table S1). Medical centers contributed isolates regardless of their antimicrobial resistance profiles.
"Dominant" sequence types were defined as those representing .5% of the total population of Citrobacter spp. with carbapenemases (27).
Statistical analysis. We used frequency tables to summarize presence of antimicrobial resistance genes, integron types, and plasmid replicon types for each dominant ST. Initially, we attempted to fit exact logistic regression models for clustered data considering isolates clustered within countries, but the models failed to converge. We then used Fisher's exact tests to perform pairwise comparisons of each outcome between dominant STs. We adjusted P values for multiple comparisons for each individual outcome using the false discovery rate (36). We considered statistical significance at the 5% level. We used R version 4.1.2 for all analyses.
Ethics statement. Ethics approval for this study was obtained through the University of Calgary Conjoint Health Research Ethics Board (REB17-1010).
Data availability. The sequencing data were deposited in the NCBI database (BioProject ID PRJNA882828).

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 0.1 MB.