Molecular, Genetic, and Biochemical Characterization of OXA-484 Carbapenemase, a Difficult-to-Detect R214G Variant of OXA-181

OXA-244, an R214G variant of OXA-48, is silently spreading worldwide likely because of difficulties in detection using classical screening media. Here, we characterized two clinical isolates of Escherichia coli and Citrobacter youngae that displayed reduced susceptibility to carbapenems but were lacking significant carbapenemase activity as revealed by negative Carba NP test results. However, positive test results were seen for OXA-48-like enzymes by lateral flow immunoassays. WGS revealed the presence of a blaOXA-181-like gene that codes for OXA-484, an R214G variant of OXA-181. BlaOXA-484 gene was located on a 58.4-kb IncP1-like plasmid (pN-OXA-484), that upon transfer into E. coli HB4 with impaired permeability, conferred carbapenem and temocillin resistance (MICs > 32 mg/L). E. coli TOP10 (pTOPO-OXA-484) revealed reduced MICs in most substrates as compared to E. coli TOP10 (pTOPO-OXA-181), especially for imipenem (0.25 mg/L versus 0.75 mg/L) and temocillin (16 mg/L versus 1028 mg/L). Catalytic efficiencies of OXA-484 were reduced as compared to OXA-181 for most ß-lactams including imipenem and temocillin with 27.5- and 21.7-fold reduction, respectively. Molecular modeling confirmed that the salt bridges between R214, D159, and the R1 substituent’s carboxylate group of temocillin were not possible with G214 in OXA-484, explaining the reduced affinity for temocillin. In addition, changes in active site’s water network may explain the decrease in hydrolysis rate of carbapenems. OXA-484 has weak imipenem and temocillin hydrolytic activities, which may lead to silent spread due to underdetection using selective screening media or biochemical imipenem hydrolysis confirmatory tests.

Substitutions at position R214 in the β5-β6 loop are particularly interesting, as they significantly impair carbapenem and temocillin hydrolyzing activity as compared to their parental enzymes [6,7].As temocillin and carbapenems are present in several screening media used for the detection of CPEs, reduced hydrolysis of these molecules can result in difficulties in the detection of these variants [16][17][18].Because of the difficulties of detection, R214 mutants have the potential to silently spread [1,16,19,20].
Here, we investigated the effect of the R214G substitution on OXA-181 hydrolytic activities and compared them to OXA-244 and OXA-232.In addition, we analyzed the genetic environment of bla OXA-484 genes in different Enterobacterales from France, UK, and Germany.Finally, using molecular modeling based on the X-ray structure of OXA-181, we were able to propose an explanation for the reduced carbapenem and temocillin hydrolysis.
Recombinant plasmids were extracted and sequenced with an automatic sequencer (ABI Prism 3100; Applied Biosystems, Les Ulis, France) as previously described [6,7].The nucleotide sequences were analyzed using software available at the National Center for Biotechnology Information website (http://www.ncbi.nlm.nih.gov,accessed on 1 July 2024).

Plasmid Characterization and Conjugation Assays
Plasmid DNA of E. coli 172D10, extracted using the Kieser method [28], was electroporated into the E. coli TOP10 strain.Transformants growing on TSA plates containing 100 µg/mL ampicillin were screened by PCR as previously described [6,7].Plasmids extracted from parental strains and transformants were subsequently analyzed on 0.7% agarose gel stained with ethidium bromide.Filter mating-out assays between E. coli 172D10 and azide resistant E. coli J53 as the recipient were performed as previously described [29].Transconjugants were selected with 100 µg/mL ampicillin and 100 µg/mL azide.

β-Lactamase Purification and Steady-State Kinetic Parameters
An overnight culture of E. coli strain BL21 (DE3) harboring pET41b-OXA-484 was used to inoculate 2 L of LB broth containing 50 mg/L kanamycin.Th expression and purification of OXA-484 were carried out as previously described [29].The protein concentrations were determined by measuring the OD at 280 nm and with the extinction coefficients obtained from the ProtParam tool (https://web.expasy.org/protparam/,accessed on 1 July 2024).

Molecular Modeling
The effect of the R214G mutation in OXA-181, resulting in OXA-484, was evaluated by molecular modeling.The Dunbrack rotamer library (swapaa command), which is part of the UCSF Chimera software v 1.17.3 [31,32] and which predicts the conformation of the amino acid sidechain based on the global energy minimum of the protein, was used to generate in silico the R214G substitution based on the OXA-181 structure (PDB code 5OE0).Interatomic clashes were identified based on VDW radii [33] using UCSF Chimera software [31,32].Corina 3.60 (Molecular Networks GmbH, Erlangen, Germany) was used to generate three-dimensional structures of the β-lactam ligands.Molecular docking calculations were performed using Gold (Cambridge Crystallographic Data Centre, Cambridge, UK) [34] and the GoldScore scoring function.The binding site, defined as a 20 Å radius sphere, was centered on the OG oxygen atom of Ser70.All other parameters had default values.The receptor-ligand complex images were produced using UCSF Chimera [32].

Nucleic Acid Sequences
Genomes of E. coli 172D10 and C. youngae 173G7 were deposited under GenBank accession numbers JBAGCG000000000 and JBAGCF000000000, respectively (bioproject PRJNA1075029).Genomes of the three UK isolates had been submitted to NCBI as part of the bioproject PRJNA788733 (biosamples SAMN24019010, SAMN24297355, SAMN24020527).The sequence of the E. coli EC-JS316 was retrieved from GenBank under accession number CP058621.

Antimicrobial Susceptibility Testing and Carbapenemase Confirmation Tests
Bacterial identification of E. coli 172D10 and C. youngae 173G7 isolates was confirmed by MALDI-TOF upon reception at the F-NRC.Disk diffusion antibiograms revealed that both isolates were resistant to amoxycillin, ticarcillin, piperacillin, amoxicillin/clavulanic acid, ticarcillin/clavulanic acid, piperacillin/tazobactam, and temocillin, of reduced susceptibility to carbapenems, and remained susceptible to all other ß-lactams tested, including cefiderocol and ceftazidime/avibactam.In addition, both isolates were susceptible to all the other tested antibiotic families, e.g., aminoglycosides, fluoroquinolones, chloramphenicol, tigecyclines, and colistin (confirmed by MIC testing [MIC = 0.25 mg/L]), except to rifampin and to a sulfonamide/trimethoprim association.
Biochemical carbapenemase confirmatory tests based on imipenem hydrolysis (homemade Carba NP [22]) gave repeatedly negative results with E. coli 172D10 and inconsistent results with C. youngae 173G7, although the resistance phenotype was compatible with the production of an OXA-48-like carbapenemase [1].The LFIA NG-Test CARBA 5 (NG Biotech) confirmed the presence of an OXA-48-like enzyme in both isolates [23] (Table 1).Finally, as these OXA-48-like producing isolates displayed reduced susceptibility to temocillin and carbapenems, the ability to grow on standard chromogenic and selective media, such as ChromID ® CARBA SMART, was tested.Using an inoculum of 10 3 CFU, both isolates failed to grow on either sides of the plate, suggesting an OXA-244-like variant with lower carbapenem and temocillin hydrolytic activities [16,18].PCR/sequencing re-sults revealed the presence of bla OXA-484 gene, a single nucleotide derivative of bla OXA-181 gene, resulting in a single amino acid substitution, R214G.
Overall, MIC values for ampicillin and cephalothin were not affected, while those of carbapenems (meropenem and ertapenem) varied in the same way as for imipenem.MIC values of E. coli HB4 expressing the natural plasmid pN-OXA-484 revealed increased MICs (>32 mg/L for carbapenems) (Table 1), suggesting that these enzymes may confer carbapenem resistance when expressed in a bacterium with impaired outer membrane permeability.

Genomic Features of OXA-484 Producers
The genomes of E. coli 172D10 and C. youngae 173G7 were determined.Only contigs bigger than 500-bp were retained for further analysis.The genomes were estimated to be 5,462,546 bp and 4,744,353 bp in size, respectively, with a mean sequencing coverage of over 130×.E. coli 172D10 and C. youngae 173G7 belonged to ST69 and ST491, respectively.
Acquired resistance genes and chromosomal point mutations involved in resistance were further sought.Two acquired β-lactamases, bla TEM-1 , and bla OXA-484 genes were identified in E. coli 172D10, while in C. youngae 173G7 ST491, only bla OXA-484 was present in addition to the chromosome-encoded bla CMY-157 gene (Table 2).In addition, an aadA1 gene, conferring resistance to spectinomycin and streptomycin, two sulfonamide resistance genes, sul1 and sul2, tet(A), mdf (A), and two dihydrofolate reductase genes, dfrA14 and dfrA17, were also present in E. coli 172D10, while in C. youngae ST491 aadA1, aph(6)-Id, sul1, and three macrolide resistance genes were identified: ere(A), mdf (A), and mph(A).No mutations were identified in topoisomerase genes known to confer resistance to fluoroquinolones in Gram-negative rods.Besides the bla OXA-484 gene, only a few shared AMR determinants were observed with the isolates from UK and Germany (Table 2).VirulenceFinder identified the nlpl gene encoding the lipoprotein NlpI precursor responsible for epithelial cell adhesion and invasion in Uropathogenic E. coli (UPEC) isolates [35].

Genetic Support and Environment of bla OXA-484 Gene
The direct transfer of the β-lactam resistance marker E. coli 172D10 into E. coli J53 by mating-out experiments revealed the transfer of the 58.4-kb plasmid (Figure 1), confirming the self-transferable nature of plasmid pN-OXA-484.Similarly, electroporation of this plasmid into E. coli TOP10 yielded transformants resistant to β-lactams and reduced susceptibility to carbapenems (Table 1).

Genetic Support and Environment of blaOXA-484 Gene
The direct transfer of the β-lactam resistance marker E. coli 172D10 into E. coli J53 by mating-out experiments revealed the transfer of the 58.4-kb plasmid (Figure 1), confirming the self-transferable nature of plasmid pN-OXA-484.Similarly, electroporation of this plasmid into E. coli TOP10 yielded transformants resistant to β-lactams and reduced susceptibility to carbapenems (Table 1).According to PlasmidFinder, seven different plasmid replication origins belonging to the incompatibility groups IncB, IncN, IncP1-like, IncX4, col156, Col8282, and p0111 were identified in E. coli 172D10 and an IncP1-like one in C youngae 173G7 (Table 2).Plasmid reconstruction revealed that the blaOXA-484 gene was carried by an IncP1-type plasmid of ca.58.4-kb in both isolates.This plasmid was different in size and in nature from previ-   According to PlasmidFinder, seven different plasmid replication origins belonging to the incompatibility groups IncB, IncN, IncP1-like, IncX4, col156, Col8282, and p0111 were identified in E. coli 172D10 and an IncP1-like one in C youngae 173G7 (Table 2).Plasmid reconstruction revealed that the bla OXA-484 gene was carried by an IncP1-type plasmid of ca.58.4-kb in both isolates.This plasmid was different in size and in nature from previous plasmids described to contain bla OXA-484 genes, such as the IncX3 plasmid identified in E. coli ST410 from Germany, UK, and Switzerland (Figures 2 and 3) [8][9][10][11][12][13][14][15].The only part that was shared between these plasmids was the ISEcp1-based composite transposon carrying bla OXA-484 gene (Figure 2B).This structure was inserted into the IncP1 backbone, as revealed by a target site duplication on both sides.In addition, in both isolates, the open reading frame of ISEcp1 was disrupted by the insertion of another insertion sequence, ISKpn12, in contrast to what was observed for the IncX3 plasmids (Figure 3).
The kcat/Km of OXA-484 for the hydrolysis of penicillin G, ampicillin, and cephalothin were like those of OXA-181 (Table 3), suggesting that the R214G substitution does alter the hydrolysis of these substrates.Interestingly, the hydrolysis of penicillin G was 61-fold lower with OXA-244.Oxyimino-cephalosporins such as cefotaxime were weakly hydrolyzed, but the kcat/Km could not be determined precisely, as the Km value was >1000 mM, determined experimentally.For ceftazidime, a bulkier oxyimino-cephalosporin, no hydrolysis was observed even with 1.92 µM purified enzyme and up to 500 µM substrate, as shown for OXA-48 [3,4].
Regarding temocillin, the Km was 3-fold higher than for OXA-181, showing that OXA-484 has a lower affinity for temocillin.Moreover, kcat for temocillin of OXA-484 was 7.5fold lower than that of OXA-181, thus resulting in a 21.7-fold lower catalytic efficiency for temocillin compared to OXA-181.
The k cat /K m of OXA-484 for the hydrolysis of penicillin G, ampicillin, and cephalothin were like those of OXA-181 (Table 3), suggesting that the R214G substitution does alter the hydrolysis of these substrates.Interestingly, the hydrolysis of penicillin G was 61-fold lower with OXA-244.Oxyimino-cephalosporins such as cefotaxime were weakly hydrolyzed, but the k cat /K m could not be determined precisely, as the K m value was >1000 mM, determined experimentally.For ceftazidime, a bulkier oxyimino-cephalosporin, no hydrolysis was observed even with 1.92 µM purified enzyme and up to 500 µM substrate, as shown for OXA-48 [3,4].
Compared to OXA-181, OXA-484 had k cat /K m values 3.3-, 7.5-, and 27.5-fold lower for ertapenem, meropenem, and imipenem hydrolysis, respectively (Table 3).The reduced catalytic efficiency for imipenem of OXA-484 is mainly due to a 75-fold reduction in the turnover number.As observed for OXA-232, the k cat /K m for ertapenem of OXA-484 was 33.3-fold lower than that for imipenem but was not affected as compared to the value for OXA-181 or OXA-48 (0.6 s −1 .mM−1 vs. 2 and 1 s −1 .mM−1 , respectively).
Overall, the hydrolysis profiles were consistent with those of OXA-244.The catalytic efficiencies, k cat /K m , were concordant with the MIC results and confirm that the hydrolysis of imipenem depends on the nature of the residue in position 214, irrespective of the backbone of the enzyme.

Molecular Modeling
An in silico study was performed to identify the structural determinants that could explain the experimentally determined differences between the hydrolytic profiles of OXA-484 in comparison with OXA-181.The OXA-181 structure (PDB code 5OE0) was used to generate a model of OXA-484 by introducing in silico the mutation R214G.
The salt bridge between D159 and R214 that was present in OXA-48 and OXA-181 was lost in OXA-484, similarly to OXA-244 [7].The loss of this salt bridge resulted in significant changes in the shape of the active site at its periphery (Figure 4).In addition, as previously shown for OXA-48, R214 also established a favorable ionic interaction with the carboxylate group of the R1 substituent of temocillin, and this interaction was lost in OXA-244, which has a glycine residue in position 214 [6,7].Here we show, through molecular docking calculations, that this interaction was also absent in the complex of temocillin with OXA-484, which has an OXA-181 backbone and a glycine in position 214 (Figure 4).
Overall, the hydrolysis profiles were consistent with those of OXA-244.The catalytic efficiencies, kcat/Km, were concordant with the MIC results and confirm that the hydrolysis of imipenem depends on the nature of the residue in position 214, irrespective of the backbone of the enzyme.

Molecular Modeling
An in silico study was performed to identify the structural determinants that could explain the experimentally determined differences between the hydrolytic profiles of OXA-484 in comparison with OXA-181.The OXA-181 structure (PDB code 5OE0) was used to generate a model of OXA-484 by introducing in silico the mutation R214G.
The salt bridge between D159 and R214 that was present in OXA-48 and OXA-181 was lost in OXA-484, similarly to OXA-244 [7].The loss of this salt bridge resulted in significant changes in the shape of the active site at its periphery (Figure 4).In addition, as previously shown for OXA-48, R214 also established a favorable ionic interaction with the carboxylate group of the R1 substituent of temocillin, and this interaction was lost in OXA-244, which has a glycine residue in position 214 [6,7].Here we show, through molecular docking calculations, that this interaction was also absent in the complex of temocillin with OXA-484, which has an OXA-181 backbone and a glycine in position 214 (Figure 4).

Discussion
OXA-48-producing Enterobacterales are now endemic in many countries and are increasingly isolated all over the world [1].Along with the current spread of OXA-48, more than 60 variants have been described [5].These variants can be classified into four groups according to their hydrolytic profile: (i) those that have an enzymatic activity similar to OXA-48, such as OXA-181 [36]; (ii) those that have no carbapenemase activity but, instead, a marked hydrolytic activity against ESC, similar to OXA-ESBLs, such as OXA-163, OXA-247, and OXA-405 [37]; (iii) those that possess carbapenemase and ESC hydrolytic activities, such as OXA-517 and OXA-438 [28,29]; and finally, those that exhibit an overall reduced activity towards all ß-lactams including carbapenems as compared to OXA-48, such as OXA-244 and OXA-232 [6,7].The amino acid sequence comparison of OXA-48-variants suggests a link between the primary structure and the function of these enzymes.Indeed, all OXA-48-variants with an OXA-ESBL phenotype (loss of carbapenem hydrolysis and the gain of activity towards ESC) have an amino acid deletion in the β5-β6 loop, and those demonstrating both activities display a 2-amino-acid deletion [29,38].This observation suggests that this loop plays a role in the substrate specificity.The phenotypic and enzymatic study of OXA-232 (which differs from OXA-181 by only one substitution R214S) [6,7] underlined that the residue 214 is crucial for carbapenem and temocillin hydrolysis.Similarly, phenotypic and enzymatic studies of OXA-244 (which differs from OXA-48 by only one substitution, R214G) [6,7] confirmed these results.The impact of the R214G substitution in an OXA-181 backbone was initially partially investigated by Oueslati et al., where enzymatic studies conducted on an in vitro generated mutant of OXA-232 (S214G) suggested that imipenem and ertapenem hydrolysis could be impaired in a similar manner to OXA-244 [6,7].Here, detailed enzymatic activity of the OXA-484 variant with an R214G substitution in an OXA-181 backbone was analyzed.
From a treatment perspective, these enzymes, even though with reduced activity towards carbapenems, in a background of impaired outer membrane permeability, led to high-level carbapenem resistance.Unlike the E. coli ST410 isolates that were all multidrugresistant [13], ST69 E. coli 172D10 was still susceptible to many classes of antibiotics.E. coli ST69 belongs to a dominant uropathogenic lineage frequently associated with AMR [39,40], while C. youngae ST491 is rarely described especially among CPEs [41].
Both 3D structures of OXA-48 and OXA-181 revealed the presence of a salt bridge between R214 and D159 [4,6] that maintains the shape and the network of water molecules within the binding site.Our molecular modeling study revealed that for OXA-48, a R214G substitution prevents any interaction with D159, which presumably increases the flexibility of this part of the binding site, thus confirming the critical role of the R214-D159 interaction in carbapenem and temocillin hydrolysis.This point is of utmost clinical importance and explains why the detection of OXA-244, OXA-232, and now OXA-484-producers remains a challenge for clinical microbiology laboratories [16][17][18].Indeed, these isolates do not grow on ChromID Carba Smart (bioMérieux, Marcy L'Etoile, France), one of the most used media for CPE screening [16,17,42].The ChromID Carba Smart is a biplate containing, on one side, a carbapenem and, on the other, temocillin, two substrates that are only weakly hydrolyzed by R214G variants of OXA-48/181 [16].
The bla OXA-484 sequence was initially reported in 2017 in five Klebsiella pneumoniae isolates from the UK [8].In December 2019, it was described in an E. coli EC-JS316 strain belonging to ST410 obtained from a rectal swab of a patient hospitalized in Germany following a stay in India [9].Bla OXA-484 was carried on an IncX3 plasmid of 51.5kb.Recently, bla OXA-484 -ST410-E.coli isolates originating from India were described in Switzerland with both an IncX3 bla OXA-484 -carrying plasmid and a core genome identical to EC-JS316 (23 ∆SNPs) [13].This IncX3 plasmid is frequently described in E. coli ST410 strains harboring bla OXA-181 and bla OXA-484 genes isolated from patients but also from environmental sources.It has been suggested that an IS26 composite transposon may underpin the mobilization of bla OXA-484 from this IncX3 plasmid to an IncFII plasmid found in Klebsiella in Switzerland and China and in E. coli ST1722 isolates from Switzerland [13].Here, we observed a likely transfer of the transposon carrying bla OXA-484 to other plasmids, increasing the potential to spread to other E. coli strains and bacterial genera, such as C. youngae.Finally, bla OXA_484 gene has also been described on IncF-type plasmids in patient isolates of K. pneumoniae from the UK and environmental K. pneumoniae isolates from South Africa but also on an IncX3 plasmid in K. variicola from China [14].In several cases, a link with India has been observed [8][9][10][11][12][13][14][15], which is not surprising given the high prevalence of bla OXA-181 genes carried by a common IncX3 plasmid in this country [1].These data suggest that bla OXA-484 and bla OXA-181 genes may have evolved from a common ancestor and spread to different hosts and now to different plasmids, thanks to ISEcp1, an element proficient in transposition at high frequency without a preferred insertion site and at the origin of bla CTX-M-type ESBLs spread worldwide [43].

Conclusions
The importation of OXA-48-like producers from endemic areas to low-prevalence countries is a major concern, especially with difficult-to-detect mechanisms, such as those mediated by OXA-244 and OXA-484.This may lead to uncontrolled spread and the occurrence of outbreaks, as is already the case for OXA-244 [1,20].Several countries have now reported OXA-484-producers from all the continents, suggesting active dispersion, likely because of under-detection and high mobility through transposition to various plasmid backbones.Therefore, infection prevention and control programs must be rigorously maintained and adapted to mitigate any future clinical impact due to these peculiar CPEs.
Detection of these enzymes is challenging, as they do not grow on all currently available screening media [16,17,42].The mSuperCARBA™ medium has been shown highly efficient in the detection of OXA-244 producers, unlike the ChromID ® CARBA SMART medium [16,17,42].The high prevalence of ESBLs among OXA-244 producers allowed for the detection of 77-78% of them using ESBL-specific screening media [16,17,42].In our collection of isolates, three out of six expressed an ESBL and, thus, could also be isolated on these media.Molecular methods used for the detection of CPE carriers directly from rectal swabs may detect OXA-244 and OXA-484 producers [16,44], but subsequent plating on selective media may not yield any growth.Thus, microbiologists, in these situations, may consider using an ESBL-screening media or a different selective media to grow OXA-244, OXA-232, or OXA-484 producers [16][17][18].Finally, CPE confirmatory tests should not be based only on biochemical assays such as the Carba NP, ß-Carba (Bio-Rad, Marne La Coquette, France), or MBT STAR ® -Carba (Bruker), especially in countries with high prevalence of these enzymes, but should rely on LFIA, such as the NG-Test CARBA 5 (NG Biotech) or molecular assays such as GenXpert (Cepheid, Sunnyvale, CA, USA) [16,23,44].Carbapenem inactivation methods could be an interesting alternative, as the rCIM and CIM tests allowed detection of 7/8 (87.5%) and 6 out of 8 (75%) OXA-244 producers, respectively [45].

Figure 4 .
Figure 4. Crystal structure of OXA-181 (A,C, PDB code 5OE0) and in silico model of OXA-484 (B,D) in cartoon and surface representations, with the docking conformations of temocillin in stick representation.OXA-181, OXA-484, and temocillin are colored in orange, olive, and green, respectively.The surface of Ser70 is colored in magenta.Hydrogen bonds are represented as springs colored in black.

Figure 4 .
Figure 4. Crystal structure of OXA-181 (A,C, PDB code 5OE0) and in silico model of OXA-484 (B,D) in cartoon and surface representations, with the docking conformations of temocillin in stick representation.OXA-181, OXA-484, and temocillin are colored in orange, olive, and green, respectively.The surface of Ser70 is colored in magenta.Hydrogen bonds are represented as springs colored in black.

Table 1 .
Antimicrobial susceptibility and confirmatory test results.
a : values obtained with broth microdilution method; b : values without plasmid; c : Neg: negative test results, Pos: positive test result, Pos/Neg: inconsistent test results; d : OXA: when the OXA-48-like band light up; Neg: when no band was visible after 15 min migration.

Table 2 .
Heatmap displaying the detected plasmid Inc-types, resistance genes, and the MLST of Enterobacterales harboring the bla OXA-484 gene.