OXA-484, an OXA-48-Type Carbapenem-Hydrolyzing Class D β-Lactamase From Escherichia coli

OXA-48-like carbapenemases are among the most frequent carbapenemases in Gram-negative Enterobacterales worldwide with the highest prevalence in the Middle East, North Africa and Europe. Here, we investigated the so far uncharacterized carbapenemase OXA-484 from a clinical E. coli isolate belonging to the high-risk clone ST410 regarding antibiotic resistance pattern, horizontal gene transfer (HGT) and genetic support. OXA-484 differs by the amino acid substitution 214G compared to the most closely related variants OXA-181 (214R) and OXA-232 (214S). The blaOXA–484 was carried on a self-transmissible 51.5 kb IncX3 plasmid (pOXA-484) showing high sequence similarity with plasmids harboring blaOXA–181. Intraspecies and intergenus HGT of pOXA-484 to different recipients occurred at low frequencies of 1.4 × 10–7 to 2.1 × 10–6. OXA-484 increased MICs of temocillin and carbapenems similar to OXA-232 and OXA-244, but lower compared with OXA-48 and OXA-181. Hence, OXA-484 combines properties of OXA-181-like plasmid support and transferability as well as β-lactamase activity of OXA-232.


INTRODUCTION
Carbapenem resistance in Enterobacterales has dramatically increased worldwide in recent years and poses a major threat to public health. Antibiotic treatment options for infections caused by carbapenem-resistant Enterobacterales (CPE) are severely limited, with often very few or even no antibiotic agents remaining effective. Resistance to carbapenems is primarily caused by carbapenemases, i.e., bacterial enzymes that hydrolyze carbapenems and most other β-lactam antibiotics. OXA-48 is one of the most prevalent carbapenemases worldwide with the highest prevalence in the Middle East, North Africa and Europe (Cantón et al., 2012). OXA-48 enzymes lead to carbapenem resistance and are most often found in Escherichia coli and Klebsiella pneumoniae (Pitout et al., 2019).
The rapid dissemination of OXA-48 has been attributed to efficient horizontal gene transfer (HGT) of plasmids harboring bla OXA−48 and a low fitness burden by carriage of these plasmids (Hamprecht et al., 2019). At least 35 variants of OXA-48, commonly referred to as OXA-48-like β-lactamases (e.g., , have been reported . Even though these variants differ only by single or few amino acid substitutions, a high degree of diversity is observed regarding the β-lactam hydrolysis pattern, genetic support of OXA-48-like encoding genes and HGT (Potron et al., 2011(Potron et al., , 2013. The gene encoding OXA-48, bla OXA−48 , is mostly present on an epidemic 63.6 kb IncL plasmid and bracketed by two identical insertion sequences (IS), IS1999, forming the composite transposon Tn1999 (Poirel et al., 2012). In contrast, bla OXA−181 is primarily found on a 51.5 kb IncX3 plasmid and has been associated with ISEcp1 in a Tn2013 transposon structure, whereas bla OXA−232 is commonly located on a non-conjugative 6.1 kb ColE-type plasmid (Qin et al., 2018;Li et al., 2019). In case of bla OXA−244 , the gene is mostly identified on the chromosome of E. coli and has been increasingly identified in the European Union in recent years (European Centre for Disease Prevention and Control, 2020; Kremer et al., 2020;Chudejova et al., 2021).
OXA-484, a variant of the OXA-48-like family, has been previously reported from a collection of OXA-48-like carbapenemase in the United Kingdom, but so far only the sequence is known (Findlay et al., 2017). Here we report on the genetic background, antibiotic resistance phenotype and HGT of this carbapenemase.
Horizontal Gene Transfer of bla OXA−48−like Harboring Plasmids Transconjugation of bla OXA−48−like harboring plasmids was conducted by liquid mating as previously described (Gruber et al., 2014). Briefly, a mixture of donors and recipients in brain heart infusion broth were incubated over night at 37 • C and subsequently plated on chromogenic agar plates containing 100 mg/L sodium azide and 10 mg/L amoxicillinclavulanic acid to select for transconjugants (Tc). Clinical E. coli and K. pneumoniae isolates encoding bla OXA−48 , bla OXA−181 , bla OXA−232 , or bla OXA−484 were employed as donors and sodium azide-resistant E. coli J53 and K. quasipneumoniae subsp. quasipneumoniae PRZ as recipients (Hamprecht et al., 2019). Plasmid DNA from clinical isolates carrying bla OXA−232 was extracted using the PureYield TM Plasmid Midiprep System (Promega, Walldorf, Germany). Transformation of J53 and PRZ using bla OXA−232 encoding plasmids was done by electroporation and transformants (Tf) were selected as described above. Presence of OXA-48-like carbapenemases in Tc and Tf was verified by disk diffusion antibiotic testing and the CARBA 5 lateral flow test. All E. coli J53 Tc were analyzed using long-read genome sequencing and disk diffusion testing as recommended by EUCAST. Transconjugation frequency was determined by dividing the numbers of Tc colonies by the number of acceptor colonies.

Whole Genome Sequencing
Whole genome sequencing was carried out for all clinical isolates using short-read technology (MiSeq or NovaSeq platform, Illumina, San Diego, United States) and long-read technology (MinION platform, Oxford Nanopore Technologies, Oxford, United Kingdom). DNA was extracted from isolates using the DNeasy UltraClean Microbial Kit (Qiagen, Hilden, Germany). For Illumina sequencing, a v3 reagent kit was applied generating either 150 or 250 bp paired-end reads. Library preparation for Nanopore sequencing was done using the SQK-RBK004 rapid barcoding kit. Sequencing was performed on a MinION MK1B sequencer utilizing a R9.4.1 flow cell. Raw signal data was base called and demultiplexed using the high accuracy base calling model of guppy basecaller version 4.0.11.

Bioinformatic Analysis
Raw data was filtered using trimmomatic for short reads and NanoFilt for long reads resulting in datasets of reads with an average genome coverage of at least 100-fold for short-reads and at least 30-fold for long-reads (Bolger et al., 2014;de Coster et al., 2018). De novo hybrid assembly was conducted using Unicycler version 0.4.8 utilizing the bold assembly mode (Wick et al., 2017). The obtained assemblies were annotated using Prokka version 1.14.6 (Seemann, 2014). Sequence types were determined using the software mlst 1 v2.19.0 (Carattoli et al., 2014). ABRicate 2 v1.0.1 was applied using the databases PlasmidFinder and NCBI AMRFinderPlus for identification of plasmid incompatibility groups and antibiotic resistance genes, respectively, using thresholds of 100% gene coverage and ≥98% nucleotide sequence identity (Feldgarden et al., 2019). ISfinder was employed for annotation of insertion sequences (Siguier et al., 2006). Plasmid sequences were aligned using MAFFT 1.4.0 and visualized using Geneious 3 11.1.5 (Katoh and Standley, 2013).

Statistical Analysis
For the comparison of transconjugation frequencies, continuous variables were assessed by Mann-Whitney U test. A P-value of < 0.05 was considered significant.

Identification of a Clinical Isolate
Harboring bla OXA−484 The E. coli isolate JS316 (EC-JS316) was obtained from a rectal swab of a patient admitted to the University Hospital Frankfurt in Germany. The patient was hospitalized due to acute dengue fever following a stay in India. The isolate EC-JS316 showed resistance to ertapenem with a MIC of 1 mg/L but was susceptible to imipenem (MIC 0.25 mg/L) and meropenem (MIC 0.125 mg/L) ( Table 2). Furthermore, the isolate was resistant to piperacillin-tazobactam, cefotaxime, fluoroquinolones, gentamicin, tobramycin and trimethoprim-sulfamethoxazole, but remained susceptible to amikacin, fosfomycin, tigecycline and colistin (Supplementary Table 2

Genome Analysis of E. coli EC-JS316
The genome of EC-JS316 was sequenced and assembled using a combined short-and long-read sequencing approach, resulting in the nucleotide sequences of a circular chromosome and five circular plasmids (Supplementary Table 3). The chromosome had a size of 4,718,403 bp coding for 4,398 predicted proteins and a GC-content of 50.8%. The ST of EC-JS316 was 410, which has been associated with international high-risk clones and acquisition of ESBL as well as different carbapenemases including OXA-181 (Roer et al., 2018;Patiño-Navarrete et al., 2020).

Sequence Analysis of pOXA-484
The plasmid pOXA-484 consisted of a conserved IncX3 backbone region of ∼31 kb length encoding 21 genes for plasmid replication, mobilization and stabilization as well as 19 hypothetical proteins (Figure 1). The second ∼20 kb region of the plasmid contained the two antibiotic resistance genes bla OXA−484 and qnrS1 inside a mosaic region of 12 insertion sequences, 6 genes coding for hypothetical proteins, and 4 disrupted genes (Figure 1). Two IS26 sequences form a functional composite transposon including the gene bla OXA−484 (Qin et al., 2018). The bla OXA−484 was flanked by a truncated ISEcp1 element and two AT-rich direct target repeats (ATCTT), which has been associated with mobilization of bla OXA−181 from ColE2 to IncX3 plasmids (Liu et al., 2015). For sequence comparison of pOXA-484 to plasmids encoding related OXA-48-like variants, we conducted whole genome   (Qin et al., 2018;Li et al., 2019).

Horizontal Gene Transfer of pOXA-484
To evaluate the HGT efficiency of pOXA-484, we compared the transconjugation frequencies of plasmid pOXA-484 carrying bla OXA−484 to the frequencies of plasmids encoding bla OXA−48 , bla OXA−181 , and bla OXA−232 , using the recipient strains E. coli J53 and K. quasipneumoniae subsp. quasipneumoniae PRZ. Four representative clinical isolates carrying bla OXA−48 or bla OXA−181 were chosen from a collection of clinical isolates based on species (E. coli and K. pneumoniae) and different STs. Since OXA-232 is rarely found in E. coli, only two clinical K. pneumoniae isolates harboring bla OXA−232 were selected (Table 1 and  Supplementary Table 4).
E. coli J53 Tc were further investigated by long-read genome sequencing and disk diffusion which verified the presence of the respective bla OXA−48−like as well as qnrS1 in bla OXA−181 and the bla OXA−484 Tc, respectively, as anticipated (Supplementary Tables 5, 6). For bla OXA−181 and bla OXA−484 Tc only the bla OXA−48 -like harboring IncX3 plasmids were transferred. Notably, in three out of four bla OXA−48 Tc additional plasmids were identified: a 61.1 kb IncFII plasmid in Tc-EC-2700, a 5.2 kb Col440II plasmid in Tc-EC-1402 and a 3.5 kb Col(pHAD28) as well as a 10.7 kb Col440II plasmid in Tc-EC-1673. However, these plasmids did not harbor any antibiotic resistance genes nor was an impact on HGT frequency or antibiotic resistance phenotype observed when comparing all bla OXA−48 Tc to each other (Figure 3 and Supplementary Table 6).

Beta-Lactam Resistance Mediated by OXA-484
To analyze the resistance phenotype caused by OXA-484, we evaluated MICs of Tc or Tf harboring the natural plasmids as well as Tf carrying pTOPO expression vectors encoding the genes for OXA-484, OXA-181, OXA-232, OXA-244, and OXA-48, respectively. Since bla OXA−244 is generally localized on the chromosome, only MICs of Tf carrying pTOPO expression vector encoding bla OXA−244 were analyzed.
The presence of pOXA-48-like plasmids resulted in increased MICs for penicillins and carbapenems compared to J53 and PRZ parental strains, whereas only minor differences were observed for cephalosporins and aztreonam ( Table 2). Carbapenem MICs of pOXA-484 were either unchanged or slightly lower compared to other OXA-48 variants. In contrast, differences in MICs of at least 2-fold between the OXA-48-like variants were detected for the antibiotics piperacillin ± tazobactam and temocillin. Tc with plasmids encoding bla OXA−48 and bla OXA−181 presented higher MICs for temocillin compared to recipients harboring bla OXA−484 and bla OXA−232 but lower MICs for piperacillin ± tazobactam in case of J53.
In J53 and PRZ Tf carrying the high copy pTOPO expression vector, the lower MICs for temocillin caused by OXA-484 and OXA-232 compared to OXA-181 and OXA-48 were also observed. Transformants carrying TOPO encoding bla OXA−244 showed almost identical MICs for all antibiotics tested compared to Tf pTOPO OXA-484. Furthermore, pTOPO OXA-48 Tf displayed higher carbapenem MICs compared to the other Tf.
Taken together, the β-lactam resistance patterns of Tc and Tf harboring bla OXA−484 revealed similar MICs to OXA-232 and OXA-244. Compared to OXA-181 and OXA-48, piperacillin ± tazobactam MICs were higher for Tc pOXA-484 J53, whereas temocillin MICs were lower for both Tc and Tf harboring either the natural plasmid or pTOPO expression vector.

DISCUSSION
This study is the first characterization of the carbapenemase OXA-484, a variant of the growing family of OXA-48-like enzymes. The dissemination of OXA-48-like carbapenemases has been attributed to highly efficient HGT to different species, low fitness burden of plasmids encoding bla OXA−48−like , association with epidemiological successful lineages and difficult to detect carbapenemase phenotypes (Hamprecht et al., 2019;Pitout et al., 2019). The plasmid pOXA-484 from isolate EC-JS316 was transferable to E. coli and K. quasipneumoniae subsp. quasipneumoniae, revealing efficient intraspecies and intergenus transfer (Figure 3). The transconjugation frequencies of this plasmid are very similar to those of the widely distributed IncX3 plasmid harboring bla OXA−181 . Sequence analysis of pOXA-484 revealed the close relationship to IncX3 plasmids containing bla OXA−181 suggesting that pOXA-484 might has evolved from pOXA-181 (Figure 2). Sequence analysis of IncX3 plasmids harboring bla OXA−181 has revealed the potential mobilization of bla OXA−181 by action of an ISEcp1 from ColE-plasmids to IncX3 plasmids (Liu et al., 2015). The highly similar sequences between the IncX3 plasmids harboring bla OXA−484 and bla OXA−181 including the ISEcp1 elements and two AT-rich direct target repeats bracketing bla OXA−484 suggest the same mechanism of mobilization for bla OXA−484 . The spread of OXA-181 and OXA-232 has additionally been associated with specific highrisk clones like ST101, ST307 and ST15 for K. pneumoniae and ST38 and ST410 for E. coli (Cubero et al., 2015;Li et al., 2019;Pitout et al., 2019;Chudejova et al., 2021). Notably, the E. coli isolate JS316 belongs to the high-risk ST410, which has been identified as a worldwide distributed extraintestinal pathogenic E. coli lineage causing nosocomial outbreaks (Roer et al., 2018;Patiño-Navarrete et al., 2020).
Substitution of arginine at amino acid 214 in OXA-48 to glycine in OXA-244 (R214G), which is also found in OXA-484, led to a reduced hydrolysis activity and MICs for temocillin and carbapenems (Hoyos-Mallecot et al., 2017). Reduced temocillin hydrolysis will most likely affect the sensitivity of temocillinbased screening plates and tests (Hopkins et al., 2019). Recently, increasing numbers of isolates expressing OXA-244 in the European Union have been linked with difficult detection because of only moderately elevated MICs for carbapenems of these isolates (European Centre for Disease Prevention and Control, 2020; Kremer et al., 2020). Likewise, the isolate EC-JS316 harboring bla OXA−484 did not grow on selective agar plates used for detection of CPE, suggesting that OXA-484 might be more easily missed and hence be more prevalent than currently known.
IncX3 plasmids have no significant negative effect on fitness of Enterobacterales isolates, which has also been shown for highly successful plasmids of the IncL group harboring bla OXA−48 and which is a predictor for efficient dissemination (Kassis-Chikhani et al., 2013;Hamprecht et al., 2019). These properties of the IncX3 plasmids might promote the further distribution of OXA-484, as already shown for other carbapenemases like NDM or KPC as well as CTX-M type ESBLs (Kassis-Chikhani et al., 2013;Wu et al., 2019).

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: https://www.ncbi.nlm. nih.gov/bioproject/PRJNA644256 and https://www.ncbi.nlm. nih.gov/bioproject/PRJNA644257.

ETHICS STATEMENT
All bacterial strains were isolated as part of routine microbiological diagnostics and stored in an anonymized database. According to the Ethics Committee of the Hospital of Johann Wolfgang Goethe-University, Frankfurt am Main, no informed consent or ethical approval of the study is necessary.

AUTHOR CONTRIBUTIONS
JSo and SG designed the study, analyzed the data, and drafted the manuscript. JSo, FK, KG, FW, MT, and SR-C performed the experiments. JSo carried out the sequencing and bioinformatic analyses. JSa, AH, and VK edited the manuscript. All authors reviewed the manuscript.

ACKNOWLEDGMENTS
We thank the team of curators of the Institute Pasteur MLST and whole genome MLST databases for curating the data and making them publicly available. We are indebted to Tobias Herther for outstanding technical assistance.