Role of blaTEM and OmpC in the piperacillin-tazobactam resistance evolution by E. coli in patients with complicated intra-abdominal infection

Contents lists available at ScienceDirect Journal of Infection journal homepage: www.elsevier.com/locate/jinf Role of bla TEM and OmpC in the piperacillin-tazobactam resistance evolution by E. coli in patients with complicated intra-abdominal infection Lydia Gálvez-Benítez a,b,1 , José Manuel Ortiz de la Rosa a,b,1 , Angel Rodriguez-Villodres a,b,c , Carlos S. Casimiro-Soriguer a,b , Irene Molina-Panadero d , Rocío Alvarez-Marin a,b,c , Rémy A. Bonnin e,f,g , Thierry Naas e,f,g , Jerónimo Pachón b,h , José Miguel Cisneros a,b,c , José Antonio Lepe a,b,c , Younes Smani b,d,i,⁎ a Clinical Unit of Infectious Diseases, Microbiology and Parasitology, Virgen del Rocío University Hospital, Seville, Spain b Institute of Biomedicine of Seville (IBiS), Virgen del Rocío University Hospital/CSIC/University of Seville, Seville, Spain c Centro de Investigación Biomédica en Red de Enfermedades Infecciosas, Instituto de Salud Carlos III (CIBERINFEC, ISCIII), Madrid, Spain d Centro Andaluz de Biología del Desarrollo, Universidad Pablo de Olavide/Consejo Superior de Investigaciones Científicas/Junta de Andalucía, Sevilla, Spain e Team ReSIST, INSERM U1184, Université Paris-Saclay, CEA, Inserm, Immunologie des Maladies Virales, Auto-Immunes, Hématologiques et Bactériennes (IMVA-HB/ IDMIT), 94270 Le Kremlin Bicêtre, France f Service de Bactériologie-Hygiène, Hôpital Bicêtre, AP-HP, 94270 Le Kremlin-Bicêtre, France g Associated French National Reference Center for Antibiotic Resistance: Carbapenemase-Producing Enterobacteriales, 94270 Le Kremlin-Bicêtre, France h Department of Medicine, School of Medicine, University of Seville, Seville, Spain i Departamento de Biología Molecular e Ingeniería Bioquímica, Universidad Pablo de Olavide, Sevilla, Spain article info Article history: Accepted 9 July 2023 Available online 11 July 2023 Keywords: Escherichia coli ß-lactamase OmpC Piperacillin-tazobactam Resistance Complicated intra-abdominal infection summary Piperacillin-tazobactam resistance (P/T-R) is increasingly reported among Escherichia coli isolates. Although in vitro experiments have suggested that bla TEM gene plays a key role in the P/T-R acquisition, no clinical in vivo study has yet confirmed the role of bla TEM or other genes. Therefore, we aimed to identify the mechanisms underlying P/T-R by following up patients with E. coli complicated intra-abdominal infections (cIAI) who experienced P/T treatment failure. Four pairs of strains, clonally related from four patients, were isolated both before and after treatment with P/T dosed at 4 g/0.5 g intravenously. The P/T MIC was tested using broth microdilution, and β-lactamase activity was determined in these isolates. Whole-genome sequencing (WGS) was performed to decipher the role of bla TEM and other genes associated with P/T-R. Changes in the outer membrane protein (OMP) profile were analyzed using SDS-PAGE, and bla TEM and ompC transcription levels were measured by RT-qPCR. In addition, in vitro competition fitness was performed between each pairs of strains (P/T-susceptible vs. P/T-resistant). We found a higher copy number of bla TEM gene in P/T-R isolates, generated by three different genetic events: (1) IS26-mediated duplication of the bla TEM gene, (2) generation of a small multicopy plasmid (ColE-like) carrying bla TEM , and (3) adaptive evolution via reduction of plasmid size, leading to a higher plasmid copy number. Moreover, two P/T-R strains showed reduced expression of OmpC. This study describes the mechanisms involved in the acquisition of P/T-R by E. coli in patients with cIAI. The understanding of P/T-R evolution is crucial for effectively treating infected patients and preventing the spread of resistant microorganisms. © 2023 The Author(s). Published by Elsevier Ltd on behalf of The British Infection Association. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). Introduction The ß-lactamases are the main cause of ß-lactam resistance in Gram-negative bacteria 1 , whose mechanisms of action are based on the hydrolysis of the ß-lactam ring. Combinations of ß-lactams/ßlactamase inhibitors (BL/BLI) were introduced to address ß-lacta- mase activity. 1,2 Some of the most commonly used combinations in Journal of Infection 87 (2023) 220–229 https://doi.org/10.1016/j.jinf.2023.07.005 0163-4453/© 2023 The Author(s). Published by Elsevier Ltd on behalf of The British Infection Association. This is an open access article under the CC BY license (http:// creativecommons.org/licenses/by/4.0/). ]]]] ]]]]]] ⁎ Correspondence to: Andalusian Center of Developmental Biology, CSIC, Department of Molecular Biology and Biochemical Engineering, University of Pablo de Olavide, Carreter de Utrera Km 1, 41013 Seville, Spain. E-mail address: [email protected] (Y. Smani). 1 Both authors contributed equally. clinical practice include ampicillin/sulbactam (A/S), amoxicillin/clavulanic acid (A/C), and piperacillin/tazobactam (P/T), as well as the recently introduced ceftazidime/avibactam (CZA) and ceftolozane/ tazobactam (C/T). P/T is a broad-spectrum BL/BLI commonly used for the empiric treatment of severe infections, including complicated intra-abdom- inal infection (cIAI) and febrile neutropenia. 1 The use of P/T has increased in recent years, as well as the number of P/T-resistant Escherichia coli isolates, limiting its therapeutic use. 3–6 bla TEM enzymes have been described to lead the P/T resistance acquisition in E. coli, either through an overproduction of bla TEM-1 or an evolution of inhibitor-resistant TEM variants (IRT), or an evolution of bla TEM variants with increased hydrolytic capabilities. 3,4,6–9 Other ß-lacta- mases such as bla OXA-1 or bla ampC , as well as loss of porins are also frequent causes of P/T resistance. 10–14 In E. coli, BL/BLI resistance is a gradual and unidirectional process that extends first from A/S, then A/C and finally P/T. We have previously described a new concept on antimicrobial resistance, called extended-spectrum BL/BLI resistance (ESRI), which is defined as the acquisition of high-level BL/BLI resistance (resistance to A/S, A/C and P/T) from a low- (resistance to A/S with susceptibility to A/C and P/T) or moderate-level (resistance to A/S and A/C and susceptibility to P/T) resistance phenotype, in which the presence of bla TEM plays an important role. 7,15 Exposure to subinhibitory concentrations of P/T has been shown to trigger ESRI isolates in vitro by two different mechanisms: the first based on duplication of bla TEM through the action of Insertion Sequence 26 (IS26); and the other based on transposition of the bla TEM gene into a multicopy plasmid (such as ColE1-like plasmids). 7 Two other studies have demonstrated that the bla TEM-1 gene, located on a 10 kb tandem repeated genomic resistance module, was associated with P/T resistance when the E. coli 907355 strain was grown in vitro in the presence of subinhibitory concentrations of P/T, 4 and the IS26-associated amplification of bla TEM1b gene was found to be associated with P/T resistance in the E. coli EC78 strain. 5 It is important to note that both studies only analyzed individual E. coli strains that were resistant to P/T. Here, we have identified four pairs of E. coli isolates, recovered from four different patients over two different times, prior and after P/T-treatment, which showed the acquisition of P/T resistance after its administration. Our aim was to determine, for the first time, whether treatment with P/T in patients with cIAI could select for resistance to P/T in E. coli, using genomic sequencing with long-read and short-read techniques, along with molecular and microbiological assays. Additionally, we sought to characterize the bacterial resistance mechanisms that contribute to the development of this BL/BLI resistance. Understanding how bacteria acquire resistance to antibiotics is crucial for effective treatment of infected patients and for preventing the spread of resistant microorganisms. Materials and methods Study design, patients and ethics This is a prospective observational cohort study from February 1st, 2020 to December 31th, 2021. Adult patients (≥18 years) with cIAI and bloodstream infection (BSI) caused by E. coli at the Virgen del Rocío University Hospital (Seville, Spain) were included in the study. The role of E. coli isolated from intra-abdominal and blood samples, as colonization or infection, was evaluated according to previously defined criteria. 16 The study was approved by the Ethics Committee of Virgen Macarena and Virgen del Rocío University Hospitals (approval number C.I. 0088-N-20). Written informed consent was signed by all patients before the inclusion in the study. Bacterial isolates Four isolates pairs (AEC-11/AEC-51, AEC-15/AEC-24, AEC-99/AEC- 106 and PT3/PT4) from four patients were isolated in the pre- and post-therapy with P/T dosed 4 g/0.5 g intravenously every 8 h in extended infusion. E. coli isolates were identified by MALDI-TOF (Bruker Daltonik, GmbH, Germany). In addition, E. coli MG1655 and its ompC mutant (ΔompC) strains were used. 17 Finally, E. coli ATCC 25922 strain was used as a standard control in all the experiments. Resistance profile The A/C and P/T antimicrobial susceptibility profiles of the four isolates pairs were initially tested by broth microdilution using the MicroScan WalkAway NMDR-1 panels. P/T MICs for these isolates and for the MG1655 and ΔompC strains were subsequently confirmed by broth microdilution reference method and interpreted according to European Committee on Antimicrobial Susceptibility Testing (EUCAST) guidelines. 18 Additionally, the P/T MIC was determined using an approved gradient diffusion assay (E-test) (Liofilchem, Italy) following the protocol published recently by Rafael Canton research group. 19 Antimicrobial selection pressure The diluted bacterial inoculum (10 5 CFU/mL) of the different isolates and strains were incubated with sub-inhibitory concentrations of P/T (Merck, Spain) corresponding to dilutions 1-fold below the MICs at 37 ºC for 24 h. Positive bacterial growth was re-adjusted to 10 5 cfu/mL for P/T MIC determinations and for further incubation with a 2-fold increased concentration of P/T concentration of 256/ 4 mg/L or a P/T concentration that did not allow bacterial growth. Removal of antimicrobial selection pressure In order to evaluate the stability of the resistance, daily passages of the P/T-resistant isolates (AEC-51, AEC-24, AEC-106 and PT4) on sheep blood agar plates (ThermoFisher, Spain) were made for 15 days and the P/T MICs were determined on days 5, 10 and 15 for each isolate. Whole-genome sequencing and bioinformatic analysis Whole-genome sequencing (WGS) was performed with two different techniques: short-read Illumina sequencing and long-read Oxford Nanopore technology to obtain a consensus sequence. Total DNA was extracted from colonies using (i) the Invitrogen PureLink™ Genomic DNA Mini Kit following the manufacturer’s instructions (ThemoFisher, France) and (ii) the Magcore® extractor system H16 with a Magcore Genomic DNA Large Volume Whole Blood Kit (RBC Bioscience Corp., Taiwan). A DNA library was prepared either (i) using a NEB Ultra II FS DNA library prep kit for Illumina (NEB, France) and then run on the MiSeq sequencer (Illumina, France) to generate paired-end 150 pb reads and (ii) by using a R9.4 flow cell and the 1D native barcoding genomic DNA kit (SQK-LSK109) on a MinION sequencer (Oxford Nanopore Technologies, UK). De novo assembly of Illumina reads was performed using CLC Genomics Workbench 21.0 (QIAGEN, France). Basecalling was performed in a GPU cluster with four Tesla v100 using guppy 5.0.16 in high accuracy mode to obtain the Nanopore reads. Then and in order to perform a genomic analysis, nanopore long reads were assembled using flye 2.9. The acquired antimicrobial resistance genes were identified using ResFinder v.4.1 (Center for Genomic Epidemiology, CGE, https://cge. food.dtu.dk/services/ResFinder/). The genome was annotated using the Rapid Annotations using Subsystems Technology (RAST) server tool. 7 The ISfinder database (https://isfinder.biotoul.fr/) was used to L. Gálvez-Benítez, J.M.O. de la Rosa, A. Rodriguez-Villodres et al. Journal of Infection 87 (2023) 220–229 221 identify insertion sequence (IS) elements. The plasmid content was analyzed using PlasmidFinder (https://cge.cbs.dtu.dk/services/ PlasmidFinder/) and manually searched by homology. Additionally, MLST tool was used to determine the clonality of the different pairs of isolates from each patient. 7 The bla TEM gene copy number was estimated using the coverage ratio of the bla TEM-1b gene (or its genetic context) and the mean of three chromosomally encoded genes (replication initiator protein dnaA, the ß-subunit of RNA polymerase rpoB and the malate dehydrogenase mdh). To obtain the consensus sequence, reads were filtered with fastp and assembled with flye (version 2.9). Later, the assembly was corrected with medaka (version 1.7.2). Finally, the consensus sequence produced by medaka was corrected by pilon (version 1.20.1). BLAST analysis and sequence comparison using BRIG Consensus sequences of the plasmid harboring bla TEM gene were blasted by pairs (P/T-susceptible vs P/T-resistant) using the BLAST Ring Image Generator (BRIG) software (version 0.95-dist). The blastn algorithm was used to generate a circular BLAST between the different plasmids. Quantification of ß-lactamase activity The ß-lactamase activity was determined using the ß-lactamase Activity Kit (BioVision Inc, Milpitas, CA, USA). 7 Briefly, the bacterial culture pellets were weighed and resuspended in ß-lactamase assay buffer. Samples were sonicated and centrifuged. Five microliters of each sample were incubated with nitrocefin, which is a chromogenic cephalosporin substrate frequently used to find beta-lactamase enzymes. The optical density (OD) at 490 nm was immediately measured in kinetic mode by the CLARIOstar reader (BMG LABTECH, France). RNA extraction and RT-qPCR Bacterial RNA was purified using the RNeasy Mini Kit (Qiagen, Netherlands) from overnight cultures. Contaminating DNA was removed from the RNA samples and the reverse transcription step was carried out using the QuantiTect Reverse Transcription kit (Qiagen, Spain), according to the manufacturer’s instructions. The primer3 v.0.4.0 software (http://bioinfo.ut.ee/primer3-0.4.0/) was used to design the primers. We used specific primers to amplify the bla TEM and ompC genes and the housekeeping rrsG gene 7,20 (Table S1). The quantitative real-time PCR assay was carried out with SYBR Premix Ex Taq (Takara, Japan) in a MxPro 3005p system (Stratagene, CA, USA). Gene expression was relatively quantified using the Comparative CT (∆∆Ct) Method (Applied Biosystems Guide, USA). Growth rate measurements Growth rate was measured as the maximum increase in OD over time during exponential growth. OD measurements were performed in 96-well plates containing 200 µL Luria Bertani (LB) or Mueller Hinton (MH) (Merck, Spain) broth/well by the Clariostart plate reader. OD at 600 nm was measured with 20 min intervals for 12 h and incubated at 37ºC with shaking for 5 s every measure point. The doubling time (DT) was determined by the exponential curve fitting function to generate an equation in the form y = Ae Bx , where A and B are numbers and x is the time between doubling of y (A 600 nm ). The equation simplifies to DT = ln2/B. Colonies were pre-cultured during 2–3 h with shaking at 37ºC to reach the exponential phase before the inoculation of the final plate. In vitro competition experiments The in vitro competition index (CI) was determined from growth cultures, consisting of AEC-11 or AEC-51, and a mixed inoculum of equivalent CFUs from AEC-11 and AEC-51 as previously described. 20 Briefly, bacteria with a titer of 5.5 log CFU/mL were grown in LB broth. At 0, 4, 8 and 24 h, 10 µL aliquots were taken and AEC-11 and AEC-51 CFUs were determined by plating serial log dilutions on LB agar (Sigma, Spain) or LB agar plus P/T (64/4 µg/mL). The AEC-51 in competition with AEC-11 was determined after plating the mixed inoculum on LB agar plus P/T to select only the AEC-51 colonies. The AEC-11 in competition with AEC-51 was determined after plating the mixed inoculum on LB agar. The obtained log CFU/mL was subtracted from the log CFU/mL of AEC-51 in competition. The CI was defined as the number of AEC-51 CFUs recovered per number of AEC-11 CFUs recovered at 4, 8 and 24 h, divided by the number of AEC-51 CFUs inoculated per number of AEC-11 CFUs inoculated at 4, 8 and 24 h, respectively. The same protocol was used with the rest of the paired isolates (AEC-15 vs AEC-24, AEC-99 vs AEC-106 and PT3 vs PT4). Analysis of outer membrane proteins by SDS-PAGE Bacterial cells were grown in LB to the logarithmic phase and lysed by sonication. Outer membrane proteins (OMPs) were extracted with sodium lauroylsarcosinate (Merck, Spain) and recovered by ultracentrifugation as described previously. 21 The OMP profiles were determined by SDS-PAGE using 10% SDS gels and 6 μg of OMP, followed by Simply Blue SafeStain gel (ThermoFisher, Spain). For peptide mass fingerprinting, a Simply Blue SafeStain band representing an unidentified protein with molecular weight 40 kDa was excised from the SDS-PAGE gel and sent for MALDI-TOF-TOF (MS-MS/MS) analysis. Data obtained from peptide MS fingerprinting were matched against the National Centre for Biotechnology Information (NCBI) database (http://www.ncbi.nlm.nih.gov) using the Mascot program. Statistical analysis Group data are presented as means ± standard errors of the means (SEM). The student t-test was used to determine differences between means using the GraphPad Prism 9. A p-value < 0.05 was considered significant. Results and discussion Demographics and clinical data analysis Demographics and clinical features of the four patients included in this study are summarized in Fig. 1 and Table 1. Patient #1 was an 63-year-old male who had been diagnosed with a neuroendocrine tumor of the pancreas and was undergoing surgery. He was carrying an abdominal drainage catheter due to an infection of the surgical site. He was admitted due to an increase in flow through the drainage, and a diagnosis of intra-abdominal abscess was made. Samples were taken and multisusceptible E. coli (AEC-11) was isolated. Treatment with P/T dosed 4 g/0.5 g every 8 h iv. in extended infusion was started. After 12 days, treatment was simplified to A/C, and he was discharged from the hospital. Treatment with A/C was continued at home. However, 73 days later, he was readmitted for removal of the drainage catheter and sampling, where P/T-resistant E. coli (AEC-51) was isolated. Patient #2 was an 66-year-old woman diagnosed with colorectal adenocarcinoma who underwent emergency surgery due to intestinal obstruction. On the 8th day after admission, she was diagnosed with a postsurgical intra-abdominal abscess which was drained, and samples were taken. Multisusceptible E. coli (AEC-15) L. Gálvez-Benítez, J.M.O. de la Rosa, A. Rodriguez-Villodres et al. Journal of Infection 87 (2023) 220–229 222 was isolated, and she started treatment with P/T on the 3rd day since admission, dosed 4 g/0.5 g every 8 h iv. in extended infusion for 19 days. Seventeen days after the start of treatment, a new sample was taken from the abscess, where P/T-resistant E. coli (AEC-24) was isolated. Patient #3 was a 72-year-old man diagnosed with ampulloma who was admitted with symptoms suggestive of cholangitis. On the 13th day of admission, a post-surgical intra-abdominal abscess was drained, and E. coli P/T-susceptible (AEC-99) was recovered. He began treatment with P/T dosed 4 g/0.5 g every 8 h iv. in extended infusion on the 1st day of admission, continuing for 14 days. Ten days after the start of treatment, a new sample was taken due to fever onset, and E. coli P/T-resistant (AEC-106) was recovered. Patient #4 was an 63-year-old man diagnosed with myelodysplastic syndrome with excess blasts. He was admitted for febrile neutropenia with an abdomino-pelvic abscess focus. Blood cultures were taken on admission, and multisusceptible E. coli (PT3) was isolated. Treatment was started with P/T, dosed 4 g/0.5 g every 8 h iv. in extended infusion, which was maintained for 8 days. On the 7th day, a new blood culture was taken, and P/T-resistant E. coli (PT4) was isolated. MLST analysis The MLST analysis performed with the assembled Whole Genome sequences revealed the same ST type for each pairs of isolates from the same patient, being ST162, ST48, ST73, and ST88 recovered from the patient #1, #2, #3, and #4, respectively (Table 1). Analysis of P/T resistance in E. coli isolates For each patient, a pair of E. coli isolates with the same ST were recovered from each single patient (AEC-11 and AEC-51 from patient #1; AEC-15 and AEC-24 from patient #2; AEC-99 and AEC-106 from patient #3, PT3 and PT4 from patient #4) from two separate infection episodes, suggesting within-patient evolution to P/T resistance. All these isolates were carbapenem and 3rd to 4th generation cephalosporins susceptible and none produced an extended-spectrum ß-lactamase (Table S2). The broth microdilution assay demonstrated that the MIC for P/T of the second isolate was 32-fold higher than that of the first isolate in all four cases, prior to intravenous therapy with P/T (Fig. 2A). Almost similar or even higher fold changes were observed with the E-test assay (Table S3). Since E. coli harboring bla TEM acquired resistance to P/T after incrementally increasing the selection pressure (i.e. concentration) of P/T, 5,7 we hypothesized that P/T-susceptible isolates (AEC-11, AEC- 15, AEC-99 and PT3) became resistant to P/T after exposure to P/T. In the in vitro selection pressure, the new P/T MIC of these isolates increased 32-fold with respect to the initial MIC; these isolates thus became resistant to P/T (Table S4). Of note, the P/T resistance observed for AEC-51, AEC-24, AEC-106 and PT4 remained stable when the antibiotic pressure was removed from the medium. Indeed, after growth in medium free of P/T for 15 days, none had returned to their initial phenotype, except for one isolate (PT4) that decreased its P/T MIC from 256 to 32 mg/L (which is still considered as resistant by EUCAST) (Fig. 2B). This latter is in line with our previous in vitro study, in which different E. coli isolates Fig. 1. Cartoon of the four patients hospitalization, treatments and samples obtention. L. Gálvez-Benítez, J.M.O. de la Rosa, A. Rodriguez-Villodres et al. Journal of Infection 87 (2023) 220–229 223 acquiring P/T resistance by selective pressure kept their resistance to P/T in medium free of P/T for 15 days. 7 Genomic analysis of the region surrounding the bla TEM-1 gene in E. coli To better understand and describe the underlying mechanisms responsible for the significative increase in P/T MICs, the four pairs of isolates were sequenced by Oxford Nanopore technologies (long read) and corrected using Illumina short reads, allowing complete assembly of the plasmids and of the genomes. Analysis of the different pairs of isolates revealed the following results: (i) in the AEC-11 isolate the mean chromosomal coverage (MCC) was 12.06x, and the bla TEM-1 coverage was 9.07x, representing a single copy per cell. Meanwhile, in AEC-51 isolate, the MCC was 12.21x, and the bla TEM-1 coverage was 233.75x representing 25 copies per cell. Analysis of the coverage of the region surrounding the bla TEM-1 gene in AEC-51 isolate revealed five copies of IS26 bracketing the bla TEM-1 gene, generated through a series of  15 kb DNA fragment translocation events by IS26. The repeated fragment is composed by resistance genes (bla TEM-1 , tet (B), dfrA, and ant(3′)-la), and MGEs (IS4, IS26, and IntI) (Fig. 3A). The latter finding aligns with a study by Hubbard et al., in which they identified, in a pair of clonal E. coli isolates obtained from a single patient across two distinct infection episodes, an increase in bla TEM-1b gene amplification when a translocable unit containing IS26 was reintroduced into the chromosome, resulting in the formation of a tandem array of translocable units. 6 Indeed, resistance to antibiotics by increase of resistance gene copy numbers has already been described among Enterobacterales in vitro and in vivo. 7,22 (ii) in the AEC-15 isolate the MCC was 51.44x, and the bla TEM-1 gene coverage was 4.63x, representing a single copy per cell. Meanwhile, in AEC-24 isolate, the MCC was 48.14x, and the bla TEM-1 coverage was 108.67x representing 25 copies per cell. Similarly, in the AEC-99 isolate the MCC was 5.39x, and the bla TEM-1 coverage was 2.51x, representing a single copy per cell. Meanwhile, in AEC-106 isolate, the MCC was 2.56x, and the bla TEM-1 coverage was 45.17x representing 35 copies per cell. Genetic context analysis revealed that plasmids of AEC-24 and AEC-106 isolates underwent similar modifications with slight differences. While a single copy of bla TEM-1 gene and no single nucleotide polymorphisms (SNPs) were identified in the two pairs of E. coli, a consistent deletion of approximately 60 kb and 30 kb was observed in the pAEC-24 and the pAEC-106 plasmids respectively (Fig. 3C and 3D). In pAEC-24, the consistently deleted region encoded SOS system inhibitor genes, plasmid partition system, methyltransferase genes, and the main conjugation machinery. Interestingly, deletions was surrounded by IS26, suggesting that intramolecular transposition and recombination lead by the insertion sequence allowed for the observed loop-out dynamics (Fig. 3C). On the other hand, pAEC-106 lost regions encoding transporter system, iron related genes, and two resistance determinants. In this case, deletions was surrounded by IS1, IS4, and IS26, which could be the responsible of the reorganization of the plasmid backbone (Fig. 4D). ISs and mainly IS26 is often found on bacterial plasmid associated with antibiotic resistance genes, and believed to be involved in the plasmid reorganization through intramolecular replicative transposition events. 23–26 Hence, we showed that the reported adaptative evolution mechanism (deletions of costly plasmid genes) in pAEC-24 and pAEC-106 is also used as a resistance mechanism by increasing plasmid copy number in order to overcome antibiotics, even when this change entails an increased fitness cost. Table 1 Demographics, clinical, epidemiological and microbiological data of the patients and the Escherichia coli isolates. Patient Sex Age Source Dose regimen Treatment duration (days) Isolates MLST Resistance phenotype Resistance genotype #1 Male 63 Pancreatic abscess P/T dosed 4 g/0.5 g IV (extendedinfusion 4 h) a 12 AEC-11 162 AMP, TIC, PIP, CIP, LEV, TMP-SXT bla TEM-1 , tet(B), sul2, dfrA17, aadA5, aph(6)- ld, aph(3″)-lb AEC-51 162 AMP, AMC, TIC, TCC, PIP, P/T, CIP, LEV, TMP-SXT bla TEM-1 , tet(B), sul2, dfrA17, aadA5, aph(6)- ld, aph(3″)-lb #2 Female 66 Post-surgical intra-abdominal abscess P/T dosed 4 g/0.5 g IV (extendedinfusion 4 h) a 18 AEC-15 48 AMP, TIC, PIP bla TEM-1 , tet(A) AEC-24 48 AMP, AMC, TIC, TCC, PIP, P/T bla TEM-1 , tet(A) #3 Male 72 Post-surgical intra-abdominal abscess P/T dosed 4 g/0.5 g IV (extendedinfusion 4 h) a 14 AEC-99 73 AMP, AMC, TIC, PIP, TMP-SXT bla TEM-1 , tet(B), sul2, dfrA8, aph(6)-ld, aph (3″)-lb AEC-106 73 AMP, AMC, TIC, TCC, PIP, P/T, TMP-SXT bla TEM-1 , sul2, dfrA8, aph(6)-ld, aph(3″)-lb #4 Male 63 Blood (Perianal abscess) P/T dosed 4 g/0.5 g IV (extendedinfusion 4 h) a 8 PT3 88 AMP, AMC, TIC, PIP, TMP-SXT bla TEM-1 , tet(A), sul2, dfrA5, aph(6)-ld, aph (3″)-lb PT4 88 AMP, AMC, TIC, TCC, PIP, P/T, TMP-SXT bla TEM-1 , tet(A), sul2, dfrA5, aph(6)-ld, aph (3″)-lb AMP, ampicillin; AMC, amoxicillin/clavulanic acid; TIC, ticarcillin; TCC, ticarcillin/clavulanic acid; PIP, piperacillin; P/T, piperacillin/tazobactam; CIP, ciprofloxacin; LEV, levofloxacin; TMP-SXT, trimethoprim/sulfamethoxazole. a extended infusion every 8 h. The sources are the same for the first and second isolates of each patient. L. Gálvez-Benítez, J.M.O. de la Rosa, A. Rodriguez-Villodres et al. Journal of Infection 87 (2023) 220–229 224 (i) in the PT3 isolate the MCC was 26.24x, and the bla TEM-1 coverage was 21.66x, representing a single copy per cell. In PT4 isolate, the MCC was 28.75x, and the bla TEM-1 coverage was 517.89x representing 18 copies per cell. Genetic context analysis of PT4 isolate reveals that two copies of bla TEM-1 gene localized in the pPT4-IncFIB plasmid (150 kb), which showed 100% identity with the pPT3-IncFIB plasmid (150 kb), and in the pSmPT4 plasmid (8 kb), which was generated through the action of the IS26 and harbor resistance genes such as bla TEM-1 , dfrA, aph(6′)-lc and aph (3′)-l, and two origin of replication (repA and repC) (Fig. 3B). The reduced size of the pSmPT4 plasmid is commonly associated to a increased copy number, leading to a major expression and production of bla TEM-1 with the consequent elevated MIC of P/T. Similar adaptative evolution in conditions selecting for plasmid carriage was observed in E. coli after an gradual pressure by P/T in vitro. 7 The latter study described the transposition event of a transposon (Tn2) from a plasmid IncI1 to a ColE1-like plasmid, in contradiction to our study, where the pSmPT4 was the product of the circularization of a plasmidic fragment induced by the action of the IS26. 7 Impact of the P/T resistance on bacterial fitness Previous studies on plasmid evolution have described that the deletions of large DNA fragments increased the plasmid stability in the host through reducing the fitness cost of the original plasmid. 27,28 In order to test the effect on the fitness of the P/T- susceptible and -resistant bacteria, growth rates were determined. Surprisingly, growth rate assays revealed increased doubling times (DT) for the resistant isolates AEC-24 (DT = 149 min) and AEC-106 (DT = 171 min) in comparison with their paired isolates AEC-15 (DT = 95.5 min) and AEC-99 (DT = 104.5 min), respectively (Fig. 4A). This increase in doubling time is probably related to their reduced fitness. For the two other pairs of isolates (AEC-11/AEC-51 and PT3/PT4), the growth rates were identical, suggesting no fitness cost in those types of resistance mechanisms (Fig. 4A). To confirm this observation, in vitro competition experiments between the four pairs of isolates have been performed. The competition index between AEC-51 and AEC-11, and PT4 and PT3 did not show significant decrease of AEC-51 and PT4 growth from 1 to 0.55 and 0.65 at 4 h of culture, respectively. Meanwhile, the competition Fig. 2. P/T resistance and stability in Escherichia coli isolates. (A) MIC of P/T in the four pairs isolates of E. coli. (B) Stability of P/T MIC during 15 days in the absence of P/T. Fig. 3. BLAST atlas of the bla TEM-1 gene harboring plasmids. (A) pAEC-11 plasmid (IncFIB) compared to the pAEC-51 plasmid (IncFIB). (B) pPT3 plasmid (IncFIB) compared to the pPT4 (IncFIB) and pSmPT4 plasmids. (C) pAEC-15 plasmid (IncFII) compared to the pAEC-24 plasmid (IncFII). (D) pAEC-99 plasmid (IncFII/FIA/FIB) compared to the pAEC-106 plasmid (IncFII/FIA/FIB). GC content is represented on the distance scale (in kbp) on the inner map. Resistance genes (red), Mobile Genetic Elements [MGEs] (blue), replication genes (orange), conjugative transfer genes (green) and other genes (black) are also indicated in the external ring. The maps were drawn using BLAST Ring Image Generator (BRIG) (http://sourceforge.net/projects/brig/). L. Gálvez-Benítez, J.M.O. de la Rosa, A. Rodriguez-Villodres et al. Journal of Infection 87 (2023) 220–229 225 index between AEC-24 and AEC-15, and AEC-105 and AEC-99 showed a significant decrease of AEC-24 and AEC-105 growth from 1 to 0.29 and 0.007, respectively, at 4 h of culture. Finally, at 24 h of culture all the resistant isolates have reduced their competition index significantly, from 1 to 0.006–0.099 (Fig. 4B). These data indicate that, despite the IS26-based duplication and generation of smaller plasmids (ColE-like) in P/T-R isolates, their fitness cost remained unaffected. Hubbard et al. reported similar findings, where no significant difference in fitness cost was observed in E. coli harboring IS26-based duplications associated with P/T resistance. 6 However, Adler et al. identified a fitness cost associated with IS26- mediated amplification of other antibiotic resistance genes. 29 While the absence of fitness cost could result in the persistence of the P/T-R phenotype even in the absence of P/T pressure (Fig. 2B), the two isolates (AEC24 and AEC106) exhibited a reduction in plasmid size and fitness cost, resulting in a higher plasmid copy number. However, despite these changes, they still maintained the P/T-R phenotype. Further investigations are needed to understand why AEC24 and AEC106 exhibit persistence of P/T resistance. bla TEM-1 is involved in the acquired resistant to P/T Since isolates of E. coli carrying bla TEM enzymes could acquire P/T resistance by expressing higher β-lactamase activity, 7 we suggest that the increase in P/T MIC values of the study isolates would be due to higher β-lactamase activities after hyperproduction of bla TEM enzymes. Indeed, Fig. 5A shows that AEC-51, AEC-24, AEC-106 and PT4 isolates resistant to P/T showed higher β-lactamase activity than their paired susceptible E. coli isolates. Noteworthy, in the absence of selective pressure, these isolates maintained higher β-lactamase activity, which may explain the long-term stability of P/T resistance in the absence of this antibiotic observed in this and in another study. 7 The hyperproduction of bla TEM-1 in the isolates of this study has been previously reported and is widely associated with P/T resistance. 4–7,15,30 However, other β-lactamases such as the inhibitor variant of bla TEM-1 , bla OXA-1 , and bla CTX-M-15 hyperproduction have also been linked to P/T resistance. 7,15,30 None of these β-lactamases were found in any of the studied isolates, suggesting that bla TEM-1 may play a significant role in inducing resistance to P/T. In addition, we analysed the transcription of bla TEM in these isolates, and found that the PT-R isolates contained higher levels of transcripts of bla TEM (Fig. 5B). Of note, other ß-lactamases genes were not detected in these isolates. In order to verify whether differential transcription of bla TEM-1 gene was due to different promoter sequences, upstream regions were analyzed. In all cases, the promoter upstream the bla TEM-1 gene was Pa/Pb, which has been shown to be a strong promoter responsible of bla TEM-1 hyperproduction, leading to resistance to Fig. 4. Impact of the P/T resistance on bacterial fitness. (A) Growth of the four pairs isolates of E. coli, and (B) Competition index between each pairs of isolates during 24 h. * P < 0.05 for 4, 8 and 24 h vs 0 h. L. Gálvez-Benítez, J.M.O. de la Rosa, A. Rodriguez-Villodres et al. Journal of Infection 87 (2023) 220–229 226 ß-lactamase inhibitors such as clavulanate and sulbactam 31 and also tazobactam. 6,8 No mutation was evidenced in the Pa/Pb region of any P/T resistant isolates as compared to the P/T susceptible ones. Role of OmpC in the P/T resistance Additional comparative analysis of the WGS revealed mutations in ompC and ompR genes, a regulator of OmpC expression, 32 in the P/T- resistant AEC-51 and AEC-24 isolates as compared with P/T-susceptible AEC-11 and AEC-15, respectively. The mutation in ompC gene consist in an chromosomal-insertion of adenosine duplication at position 51 of the gene (c.51 dupA) generating an frame shift change, and the mutation of ompR gene consist in single nucleotide variation (c 0.136 C > T) in the position 136 of the ompC gene leading to an amino acid substitution (Arg46Cys). Transcription analysis of ompC gene revealed that ompC gene was less transcribed in P/T-resistant AEC-51 and AEC-24 isolates (Fig. 6A). Furthermore, SDS-PAGE analysis of OMPs showed a reduction of the expression of a protein identified as OmpC by MALDITOF-TOF (MS-MS/MS) (Fig. 6B). Notably, after exposure of MG1655 wt and ΔompC strains with 2-fold increased P/T concentration, the MIC of ΔompC strain was increased progressively from 8 to 32-folds as compared with the MG1655 wt strain (Fig. 6C). OmpC constitute the main OMP in E. coli that is necessary for drug transport across cellular membranes. 33 A reduction in the OmpC expression has been associated with the ß-lactam reduced susceptibility/resistance. 34–36 The importance of the OmpC in the P/ T resistance is less clear. 37 Additional studies are need to decipher how P/T regulates the expression of OmpC. Long read sequencing has shown to be crucial in the elucidation of the antimicrobial resistance mechanisms, mainly those involving Fig. 5. bla TEM is involved in the acquired resistance to piperacillin/tazobactam. (A) ß-lactamase activity of the four pairs of E. coli isolates. (B) bla TEM gene transcription in the four pairs of E. coli isolates. * P < 0.05 for AEC-51 and AEC-51D vs AEC-11, AEC-24 and AEC-24D vs AEC-15, AEC-106 and AEC-106D vs AEC-99, PT4 and PT4D vs PT3. # P < 0.05 for AEC-24 vs AEC-24D, AEC-106 vs AEC-106D, PT4 vs PT4D. Deselected (D) indicates an isolate without selective P/T pressure. Fig. 6. Role of OmpC in the P/T resistance. (A) ompC gene transcription (B) OmpC expression in the two pairs of Escherichia coli isolates, and (C) MIC of P/T in MG1655 and its ompC mutant strains. L. Gálvez-Benítez, J.M.O. de la Rosa, A. Rodriguez-Villodres et al. Journal of Infection 87 (2023) 220–229 227 numerous IS elements and repeat sequences. The acquisition of P/T resistance induced by P/T treatment is not attributed to a clonal outbreak of high-risk clones but rather to multiple acquisitions of resistance mechanisms within the circulating population of clinical strains. In this study, we have found three ways in the acquisition of resistance to P/T: (i) IS26-based duplication, as previously shown. 6,7 We found modifications of a pre-existing plasmid by transposition, rather than acquisition of foreign DNA; (ii) generation of a smaller plasmid (ColE-like) harboring the bla TEM-1 gene as previously reported in in vitro studies, 7 and; (iii) adaptative evolution through the reduction of the plasmid size and fitness cost, leading to a higher plasmid copy number. To the best our knowledge, this is the first study describing IS-mediated deletions of important traits of the plasmid as a driving force of bacterial resistance by increasing the plasmid copy number. This kind of genetic event (recombination, transposition or conjugation) occurs frequently, but the genetic events leading to antibiotic resistance, such as, higher production levels of bla TEM , seem to be responsible in clinical failures of P/T treatments. Conclusions We demonstrated that treatment of E. coli cIAI with P/T can result in the development of resistance to this antibiotic. Three different mechanisms of resistance to P/T were found, in all of which the bla TEM-1 gene played a central role and OmpC expression was reduced. The novelty of this study lies in the identification of a new evolutionary mechanism based on adaptive evolution through the reduction of plasmid size and fitness cost, leading to an increased plasmid copy number and the emergence of P/T-R E. coli isolates. This work represents a significant advancement in our understanding of the mechanisms underlying the acquisition of P/T resistance by E. coli in patients with cIAI. This enhanced understanding of P/T resistance evolution is crucial for effectively treating infected patients and preventing the spread of resistant microorganisms. Funding This work was funded by the Instituto de Salud Carlos III, Subdirección General de Redes y Centros de Investigación Cooperativa, Ministerio de Economía, Industria y Competitividad (grant PI19/01009), by Plan Nacional de I+D+i 2013–2016, and by the Instituto de Salud Carlos III, Subdirección General de Redes y Centros de Investigación Cooperativa, Ministerio de Economía, Industria y Competitividad, Spanish Network for Research in Infectious Diseases (grant RD16/0016/0009), cofinanced by the European Development Regional Fund “A way to make Europe”/”Investing in your future”. A.R.V, R.A-M, J.A.L. and J.M.C. were supported (grant CB21/13/00006) by CIBERINFEC - Consorcio Centro de Investigación Biomédica en Red, Instituto de Salud Carlos III, Ministerio de Ciencia e Innovación and Unión Europea – NextGenerationEU. L.G.B. was partly financed by a grant from the Spanish Society of Infectious Diseases and Clinical Microbiology (SEIMC) and is supported by the Subprograma Rio Hortega, Instituto Carlos III, Subdirección General de Redes y Centros de Investigación Cooperativa, Ministerio de Ciencia, Innovación y Universidades, Spain (CM22/00196). J.M.O.R. is supported by the Subprograma Sara Borrell, Instituto de Salud Carlos III, Subdirección General de Redes y Centros de Investigación Cooperativa, Ministerio de Ciencia, Innovación y Universidades, Spain (CD21/00098). A.R.V. is supported by the Subprograma Juan Rodés, Instituto de Salud Carlos III, Subdirección General de Redes y Centros de Investigación Cooperativa, Ministerio de Ciencia, Innovación y Universidades, Spain (JR20/00023), and Y.S. has received a Miguel Servet Tipo II contract from the Instituto de Salud Carlos III, Subdirección General de Redes y Centros de Investigación Cooperativa, Ministerio de Economía y Competitividad, Spain (grant CPII20/00018). Funding for open access publishing: Universidad Pablo de Olavide Data availability Whole-genome sequences were deposited in GenBank under the BioProject accession number PRJNA949846. Declaration of Competing Interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgments We like to thank Andrea Vila Domínguez for technical help and Sanna Koskiniemi for the gift of the MG1655 and its ompC mutant strains. Appendix A. Supplementary material Supplementary data associated with this article can be found in the online version at doi:10.1016/j.jinf.2023.07.005. References 1. Bush K, Bradford PA. ß-lactams and ß-lactamase inhibitors: an overview. Cold Spring Harb Perspect Med 2016;6:a025247. https://doi.org/10.1101/cshperspect.a025247 2. Drawz SM, Bonomo RA. Three decades of ß-lactamase inhibitors. Clin Microbiol Rev 2010;23:160–201. 3. Shubert C, Slaughter J, Creely D, van Belkum A, Gayral JP, Dune WM, et al. Population analysis of Escherichia coli isolates with discordant resistance levels by piperacillin-tazobactam broth microdilution and agar dilution testing. Antimicrob Agents Chemother 2014;58(3):1779–81. https://doi.org/10.1128/CMR.00037-09 4. Schechter LM, Creely DP, Garner CD, Shortridge D, Nguyen H, Chen L, et al. Extensive gene amplification as a mechanism for piperacillin-tazobactam resistance in Escherichia coli. mBio 2018;9. https://doi.org/10.1128/mBio.00583-18. e00583-18. 5. Hansen KH, Andreasen MR, Pedersen MS, Westh H, Jelsbak L, Schønning K. Resistance to piperacillin/tazobactam in Escherichia coli resulting from extensive IS- 26-associated gene amplification of blaTEM-1. J Antimicrob Chemother 2019;74:3179–83. https://doi.org/10.1093/jac/dkz349 6. Hubbard ATM, Mason J, Roberts P, Parry CM, Corless C, van Aartsen J, et al. Piperacillin/tazobactam resistance in a clinical isolate of Escherichia coli due to IS26- mediated amplification of bla TEM-1b . Nat Commun 2020;11:4915. https://doi.org/10. 1038/s41467-020-18668-2 7. Rodriguez-Villodres A, Gil-Marqués ML, Alvarez-Marín R, Bonnin RA, Pachón- Ibáñez ME, Aguilar-Guisado M, et al. Extended-spectrum resistant to ß-lactams/ßlactamase inhibitors (ESRI) evolved from low-level resistant Escherichia coli. J Antimicrob Chemother 2020;75:77–85. https://doi.org/10.1093/jac/dkz393 8. Zhou K, Tao Y, Han L, Ni Y, Sun J. Piperacillin-tazobactam (T.Z.P.) resistance in Escherichia coli due to hyperproduction of TEM-1 ß-lactamase mediated by the promoter Pa/Pb. Front Microbiol 2019;10:833. https://doi.org/10.3389/fmicb.2019. 00833 9. Chiaibi EB, Sirot D, Paul G, Labia R. Inhibitor-resistant T.E.M. ß-lactamases: phenotypic, genetic and biochemical characteristics. J Antimicrob Chemother 1999;43:447–58. https://doi.org/10.1093/jac/43.4.447 10. Henderson A, Paterson DL, Chatfield MD, Tambyah PA, Lye DC, De PP, et al. Association between minimum inhibitory concentration, beta-lactamase genes and mortality for patients treated with piperacillin/tazobactam or meropenem from the MERINO Study. Clin Infect Dis 2021;73(11):e3842–50. https://doi.org/10.1093/cid/ ciaa1479 11. Livermore DM, Day M, Cleary P, Hopkins KL, Toleman MA, Wareham DW, et al. OXA-1 ß-lactamase and non-susceptibilty to peinicillin/ß-lactamase inhibitor combinations among ESBL-producing Escherichia coli. J Antimicrob Chemother 2019;74(2):326–33. https://doi.org/10.1093/jac/dky453 12. Bethel CR, Hujer AM, Helfand MS, Bonomo RA. Exploring the effectiveness of tazobactam against ceftazidime resistant Escherichia coli: insights from the comparison between susceptibility testing and beta-lactamase inhibition. FEMS Microbiol Lett 2004;234(1):99–103. https://doi.org/10.1016/j.femsle.2004.03.025 13. Vimont S, Aubert D, Mazoit JX, Poirel L, Nordmann P. Broad-spectrum ß-lactams for treating experimental peritonitis in mice due to Escherichia coli producing plasmidencoded cephalosporinases. J Antimicrob Chemother 2007;60(5):1045–50. https:// doi.org/10.1093/jac/dkm317 14. Pagès JM, James CE, Winterhalter M. The porin and the permeating antibiotic: a selective diffusion barrier in Gram-negative bacteria. Nat Rev Microbiol 2009;6:893–903. https://doi.org/10.1038/nrmicro1994 L. Gálvez-Benítez, J.M.O. de la Rosa, A. Rodriguez-Villodres et al. Journal of Infection 87 (2023) 220–229 228