Vaborbactam increases meropenem susceptibility in Pseudomonas aeruginosa clinical isolates displaying MexXY and AmpC upregulation

ABSTRACT To evaluate the resistance mechanisms among Pseudomonas aeruginosa clinical isolates exhibiting meropenem (MEM) MIC values higher than meropenem-vaborbactam (MEV). P. aeruginosa clinical isolates collected in US hospitals from 2014 to 2019 were susceptibility tested. Whole-genome and transcriptome sequencing were performed. Results were analyzed for strain typing, acquired β-lactamases, and mutations in chromosomal genes; gene expression was measured for known β-lactam resistance contributors. Results were compared to a control group of 10 P. aeruginosa isolates displaying MIC values at 8 mg/L for meropenem ± vaborbactam (MEM = MEV). Out of 88 isolates displaying MEM > MEV, 33 (37.5%) isolates had reproducibly lower MIC values for meropenem-vaborbactam compared to meropenem when retested. The expression of mexX, mexY, mexZ, and ampC was significantly greater among a higher percentage of the MEM > MEV isolates. Furthermore, the association of mexXY and ampC overexpression was detected in 17/33 MEM > MEV isolates and only 1/10 MEM = MEV isolate. In addition, the Pseudomonas-derived cephalosporinase amino acid substitution R79Q was detected among 33.3% of the isolates displaying MEM > MEV, and none of the isolates displayed MEM = MEV. Other resistance mechanisms were not observed or were equally observed in both groups. In rare cases, vaborbactam plays a role in lowering the meropenem MIC values in P. aeruginosa clinical isolates likely due to the inhibition of the AmpC gene that was overexpressed in the presence of upregulation of MexXY with or without alterations in the AmpC gene. IMPORTANCE Pseudomonas aeruginosa isolates are intrinsically resistant to multiple antimicrobial agents and meropenem is an important therapeutic option to treat infections caused by this organism. Meropenem-vaborbactam activity is similar to that of meropenem alone against P. aeruginosa isolates. Isolates belonging to this species that display lower meropenem-vaborbactam compared to meropenem are rare. We initiated this study to understand the resistance mechanisms that could lead to lower meropenem-vaborbactam MIC values when compared to meropenem alone. We documented that isolates displaying lower meropenem-vaborbactam exhibited overexpression of MexXY and AmpC. In addition, isolates displaying the R79Q PDC (AmpC) mutation were more likely to display lower meropenem-vaborbactam when compared to isolates displaying the same MIC values for these agents.

(AmpC) from P. aeruginosa when compared to imipenem and other antipseudomonal agents (4).Moreover, meropenem, unlike imipenem, is not as affected by OprD deficiency.Despite its stability against described mechanisms, meropenem can be extruded from the cell by the RND-family efflux systems MexAB-OprM and MexXY (5).These mechanisms alone might not lead to meropenem resistant MIC values, but in combination with other intrinsic and acquired resistance mechanisms, these mecha nisms confer meropenem MIC values in the resistant categories.
The interplay of resistance mechanisms has been pointed out as the main cause of carbapenem resistance among P. aeruginosa clinical isolates.In a study evaluating 33 carbapenem-resistant P. aeruginosa isolates from eight hospitals in New York, the isolates had combinations of decreased OprD, increased AmpC expression (39.4%), or OprD decrease and overexpression of efflux (60.6%) (6).In an evaluation of carbapenemnonsusceptible P. aeruginosa isolates from 14 European countries, the overexpression of AmpC, MexAB-OprM, and MexXY was noted among 37.2%, 20.1%, and 35.7% of the isolates, respectively, while OprD loss or decrease was noted among 94.9% of the isolates (7) indicating that most isolates had multiple resistance mechanisms.In a Spanish multicenter study, AmpC overexpression was prevalent among meropenem-resistant isolates; however, around 30% of the isolates also exhibited overexpression of MexAB-OprM or MexXY (8).
Meropenem-vaborbactam was approved in 2017 by the US FDA for the treatment of complicated urinary tract infections, including pyelonephritis caused by Enterobac terales isolates in adult patients.This β-lactam/β-lactamase inhibitor combination was approved by the European Medicines Agency in 2018 for the same indication, with an additional approval for P. aeruginosa.Meropenem-vaborbactam was approved with a prolonged infusion dose and granted a P. aeruginosa breakpoint of ≤8 mg/L for susceptibility by the EUCAST (9).
The activity of meropenem-vaborbactam against P. aeruginosa is described as the same as meropenem alone (10) as vaborbactam does not offer protection for decreased permeability or increased extrusion of meropenem from the cell.However, during the post-approval surveillance for meropenem-vaborbactam, 88/17,180 (0.5%) P. aeruginosa clinical isolates collected in US hospitals from 2014 to 2019 displayed meropenem-vabor bactam MIC values lower than the results for meropenem.In this study, we evaluated the resistance mechanisms of these isolates using 10 isolates with the same MIC values for meropenem and meropenem-vaborbactam as control isolates.

Bacterial isolates and susceptibility testing
Among 17,180 P. aeruginosa collected worldwide from 2014 to 2019 as part of the meropenem-vaborbactam surveillance studies (11), 88 isolates displayed MIC values for meropenem-vaborbactam lower than meropenem (MEM > MEV) and meropenemvaborbactam ≤8 mg/L when initially tested.These 88 isolates were submitted to confirmatory susceptibility testing against meropenem ± vaborbactam alongside 20 isolates exhibiting MIC values at 8 mg/L for meropenem and meropenem-vaborbac tam (MEM = MEV) susceptibility tested as potential controls.The control isolates were selected to represent a diverse set of hospitals that participated in the surveillance study and displayed different susceptibility profiles.Species identification was confirmed when needed by matrix-assisted laser desorption ionization-time of flight mass spectrometry using the Bruker Daltonics MALDI Biotyper (Billerica, Massachusetts, USA) following the manufacturer's instructions.
Standards Institute (CLSI) procedures (12).Quality control (QC) testing was performed to ensure proper test conditions.QC strains included Escherichia coli ATCC 25922 and NCTC 13353, Klebsiella pneumoniae ATCC 700603 and ATCC BAA-1705, and P. aeruginosa ATCC 27853.CLSI guidelines were used for the interpretation of susceptibility results (13).Vaborbactam was provided by Melinta Therapeutics.Other agents were acquired from Sigma-Aldrich (Saint Louis, Missouri, USA) or US Pharmacopeia (Rockville, Maryland, USA).

Characterization of β-lactam resistance mechanisms
All 33 P. aeruginosa isolates confirmed to have MEM > MEV and 10 isolates with MEM = MEV selected as control were subjected to whole-genome sequencing (WGS).Genomic libraries were constructed using the Nextera XT protocol and index kit (Illumina, San Diego, California, USA) following the manufacturer's instructions and sequenced on a MiSeq Sequencer (Illumina) or constructed using the Illumina DNA Prep protocol with IDT for Illumina Unique Dual indexes (Illumina) and sequenced on a NextSeq 1000 Sequencer (Illumina).FASTQ format files for each sample set were assembled independ ently using de novo assembler SPAdes 3.9.0(14) with K-values of 21, 33, 55, 77, and 99 and "careful mode" to reduce the number of mismatches.This process produced a FASTA format file of contiguous sequences with the best N50 value.An in-housedesigned software using the target assembled sequences (15) as queries to align against numerous resistance determinants from the NCBI Bacterial Antimicrobial Resistance Reference Gene Database (https://www.ncbi.nlm.nih.gov/bioproject/PRJNA313047) was used to search for β-lactamase genes.Potential matches were generated with the criteria of >94% identity and 40% minimum coverage length (16).
Total RNA was extracted and purified from log phase bacterial cultures that dis played a cell density of optical density (OD) 600 of 0.3 to 0.5 using the RNeasy Mini Kit in the Qiacube workstation (Qiagen, Hilden, Germany) according to the manufactur er's instructions.Residual DNA was eliminated by treatment with RNAse-free DNase (Promega, Madison, Wisconsin, USA).Quantification of total RNA and sample quality was assessed using the RNA 6000 Pico kit on the Agilent 2100 Bioanalyzer (Agilent Technologies, Santa Clara, California, USA) according to the manufacturer's instructions.Only preparations with acceptable RNA integrity numbers (RIN) ≥7 and/or that showed no visual degradation were used for experiments.
A total of up to 1 µg of RNA was subjected to rRNA depletion using Ribo-Zero Plus as part of the Illumina Stranded Total RNA Prep Kit with Ribo-Zero Plus (Illumina) according to the manufacturer's instructions.The resultant rRNA-depleted RNA was then purified using RNA-specific magnetic beads and eluted in 8.5 µL of elution buffer for downstream library preparation.Whole-transcriptome RNA sequencing cDNA library preparation was performed using the Illumina Stranded Total RNA Prep, Ligation (Illumina) with eluted Ribo-Zero-treated RNA samples as input material.Library preparation was performed according to the manufacturer's instructions, beginning with mRNA fragmentation.mRNA fragmentation was accomplished using the entire eluted Ribo-Zero-treated RNA sample (8.5 µL) combined with 8.5 µL of EPH3 (Elute, Prime, Fragment 3HC Mix).Sequencing was completed on a NextSeq 1000 Sequencer using NextSeq 1000 P2 Reagents.An independently prepared replicate of the control reference isolate (P.aeruginosa PAO1) was included with each sequencing run to serve as an internal control.
Differential gene expression was estimated using an RNA-seq pipeline.First, pairedend reads were trimmed, corrected, and filtered using FASTA.Quality-controlled reads were aligned to a P. aeruginosa PAO1 reference assembly (ASM676v1), filtered based on alignment scores, and assigned to loci to calculate per-gene counts using EDGE-pro (17).Counts were normalized across samples using the trimmed mean of M-values normaliza tion, and fold change expression was calculated according to an exact test based on the quantile-adjusted conditional maximum likelihood method using edgeR (18).Synonyms and gene ontology (GO) terms were collected from UniProt to aid in interpretation.
Differences in expression were considered significant if they were ±five-fold of the expression of P. aeruginosa PAO1.

RESULTS
Of the 88 (0.5% of the overall isolates) P. aeruginosa isolates initially displaying merope nem MIC values greater than meropenem-vaborbactam and meropenem-vaborbactam ≤8 mg/L, 33 isolates were confirmed to have greater meropenem MIC results upon retesting (phenotype MEM > MEV; Fig. 1A through C).A toltal of 19 isolates displayed meropenem-vaborbactam MIC values one dilution lower than meropenem and MIC values >8 mg/L for both agents (Fig. 1C).In total, 12 and 2 isolates exhibited mer openem-vaborbactam MIC values lower than meropenem by two or three dilutions, respectively (Fig. 1C).
Among the 20 isolates that displayed MIC values at 8 mg/L for meropenem and meropenem-vaborbactam and were initially used as controls for the susceptibility testing, all 20 displayed reproducible MIC values.This MIC was selected because this is the concentration achieved with a 3 h of infusion often used for meropenem and recommended for meropenem-vaborbactam and isolates at this MIC would be considered susceptible to both agents.Ten of these isolates were randomly selected for further evaluation (phenotype MEM = MEV).
MLST results revealed a diversity of sequence types (STs) among the 43 P. aeruginosa evaluated by WGS (33 MEM > MEV and 10 MEM = MEV), and there were no predominant types in each group (Table 1).In addition, high-risk clones described among P. aeruginosa isolates, such as ST111 or ST235, were only detected in two isolates, one from each group.Only three isolates harbored acquired β-lactamase-encoding genes (Table 1).One MEM = MEV carried genes encoding OXA-9 and CARB-2 (also known as PSE-1).One MEM > MEV harbored bla KPC-2 and another carried bla OXA-2 .
The variant of chromosomal AmpC from P. aeruginosa, Pseudomonas-derived cephalosporinase amino acid (PDC), was differently represented in the MEM = MEV phenotype group compared to the MEM > MEV phenotype isolates (Tables 1 and  2).PDC-5 and PDC-1 were only noted among MEM > MEV isolates.PDC-3 was more common among MEM = MEV isolates compared to the MEM > MEV isolates (four isolates and one isolate, respectively).Conversely, PDC-8 and PDC-31 were detected in a greater number of MEM > MEV phenotype isolates (four and five isolates, respectively) when compared to MEM = MEV isolates (one of each).When analyzing discrete amino acid substitutions in the PDC sequences and comparing them to PDC-1, the substitution R79Q was noted in 11 (33.3%)isolates of the MEM > MEV phenotype and was not found among the MEM = MEV isolates (Table 2).This alteration is located near the helix H-2 and was observed among the alleles PDC-5, PDC-5-like, PDC-6, and PDC-71 (19).Alterations in the PDC Ω-loop that have been associated with resistance against novel β-lactam/β-lactamase inhibitor combinations were noted among 13 (39.4%)and 4 (40.0%)isolates from the MEM > MEV and MEM = MEV groups, respectively.
Among the other intrinsic β-lactamases in P. aeruginosa, variations within the sequences of the OXA-50 family member and PIB-1 (PA5542) seemed to be randomly distributed among the two groups (Table 1).
When the expressions of mexX and mexXY were combined as a single result due to the presence of these genes in the same operon, 30 (90.9%) isolates of the MEM >    Unfortunately, due to the small sample size (only 43 isolates), no statistical power was achieved to corroborate the differences noted in the presence/absence of features; however, the analysis of the expression rates for these genes demonstrated statistically significant higher levels of expression among the MEM > MEV group when compared to the MEM = MEV group (Fig. 2).The average expression of mexX, mexY, mexZ, and PDC was, respectively, 78.8/33.2X,46.8/22.9X,9.9/4.3X,and 222.4/119.0X,compared to the baseline for MEM > MEV/MEM = MEV isolates.The other genes analyzed did not display differences among the two groups (Table 2), including mexAB-oprM overexpression detected among 16 (48.5%)and 5 (50.0%) isolates from the MEM > MEV and MEM = MEV groups, respectively.Disruption of OprD was noted among 29/33 (87.9%) of the MEM > MEV isolates and 9/10 (90.0%) of the MEM = MEV isolates (Table 2), and one additional isolate had reduced expression of OprD.

Group
Analysis of mutations in genes encoding the known resistance contributors AmpR, ArmZ, MexZ, FtsI, DacB, DacC, NalD, and MexR did not reveal distinct amino acid substitutions among the MEM = MEV and MEM > MEV groups (Table 2).

DISCUSSION
This study compared 33 P. aeruginosa isolates with confirmed meropenem MIC values greater than meropenem-vaborbactam MIC values (MEM > MEV) to 10 isolates with an MIC value at 8 mg/L for meropenem and meropenem-vaborbactam (MEM = MEV).The results revealed that a greater fraction of MEM > MEV than MEM = MEV isolates had an increased expression of the chromosomal ampC combined with an elevated expression of mexXY.Furthermore, the expression levels for the genes encoding AmpC (PDC), MexX, and MexY were significantly greater among the MEM > MEV phenotype isolates when compared to the MEM = MEV isolates.The generation of isogenic mutants with these characteristics was not conducted and is a limitation of the current study.
Riera et al. demonstrated that a combined ampC overexpression and oprD deletion increased meropenem MIC values four-to eightfold in a PAO1 background, conferring meropenem MIC values of 2 to 4 mg/L (20).These results are categorized as resistant when applying current breakpoint criteria that were developed using the standard dose of meropenem of 1 g q8h.Other studies evaluating clinical isolates demonstrated that when an oprD deletion is associated with ampC overexpression, a 32-fold increase was noted in the meropenem MIC values, leading to resistance (MIC 16 mg/L) even when applying a cutoff for 2 g q8h with an extended infusion that should cover MIC values up to 4 or 8 mg/L (21,22).The approved standard dose of meropenem-vaborbactam is 2 g meropenem plus 2 g vaborbactam with a 3 h infusion, q8h.
In a study by Masuda et al., the overexpression of mexXY in isolates that already expressed high levels of ampC in a mexAB-oprM mutant background generated a fourfold increase in meropenem MIC values (5).It can be hypothesized that if the number of meropenem molecules inside the cell is constantly reduced by extrusion by this efflux system, then the AmpC (PDC) molecules could slowly hydrolyze meropenem, preventing its activity.In this case, vaborbactam would bind to the PDC molecules and protect meropenem from hydrolysis, which would explain the lower MIC values for meropenemvaborbactam in the MEM > MEV group.
Lastly, this study indicated that when MIC values for meropenem-vaborbactam are lower than the results for meropenem alone, testing should be repeated for confirmation, as these results were not confirmed in 62.5% of the cases.While the disparity between MIC values for meropenem and meropenem-vaborbactam is not well appreciated, clinicians and microbiologists should be aware of such discrepancies.Meropenem-vaborbactam testing should be considered for P. aeruginosa isolates as populations of the genotypes observed here may be enriched in some settings.

TABLE 2
Genetic characterization results for P. aeruginosa isolates displaying the same MIC values for meropenem and meropenem-vaborbactam (MEM = MEV) compared to meropenem-resistant and meropenem-vaborbactam-susceptible isolates (MEM > MEV) (Continued)MEV group and 6 (60.0%) isolates of the MEM = MEV group exhibited overexpression of the MexXY efflux system.Moreover, when MexXY and PDC overexpression were analyzed together, 17 (51.5%) of the isolates in the MEM > MEV group had both genes overex pressed compared to 1 (10.0%)isolate from the MEM = MEV group.

TABLE 2
Genetic characterization results for P. aeruginosa isolates displaying the same MIC values for meropenem and meropenem-vaborbactam (MEM = MEV) compared to meropenem-resistant and meropenem-vaborbactam-susceptible isolates (MEM > MEV) (Continued)