Within patient horizontal gene transfer dynamics of a blaNDM−7 plasmid among four different bacterial species

TL;DR

A patient carried a carbapenem-resistant plasmid across four bacterial species, showing how antibiotic resistance spreads within a single person.

Contribution

This case report demonstrates horizontal gene transfer of a blaNDM−7 plasmid among multiple bacterial species within one patient.

Findings

A patient carried a blaNDM−7 plasmid in Citrobacter freundii, Klebsiella oxytoca, Raoultella planticola, and Serratia marscescens.

The plasmid was horizontally transferred within the patient, not through identical bacterial strain spread.

Mobile genetic elements like the incX3 plasmid play a key role in spreading antimicrobial resistance.

Abstract

We describe the case of a 73-year old male who was found to be a carrier of an blaNDM -producing Citrobacter freundii shortly after admission. During his admission, he developed abdominal abscesses and received multiple courses of piperacillin-tazobactam. In the following months, he was found to carry three other carbapenemase-positive species: Klebsiella oxytoca, Raoultella planticola and Serratia marscescens.

Two of these strains had clustering carbapenem-sensitive isolates cultured before. The species all carried an blaNDM−7 encoding incX3 plasmid, which demonstrated horizontal gene transfer within this patient.

This case report underlines the importance of mobile genetic elements in infection control, as they serve as transmission vehicles for antimicrobial resistance beyond the spread of identical bacterial strains.

The online version contains supplementary material available at 10.1007/s10096-025-05327-x.

Figures

An overview of hospital admissions, relevant cultures and antimicrobial treatment of our patientFig. 2Overview of the tested antibiotic MIC-values and interpretation of the CPE and their wildtypes if available. (EUCAST version 13.1) S: Susceptible, R: Resistant, I: Susceptible with increased exposure. *: Sensititre™ Gram Negative EUMDROXF AST Plate (Thermofisher, United States of America). : ‘NI’ is given when no interpretation was possible based on the EUCAST clinical breakpoints when the culture was taken. *: VITEK® 2 (Biomérieux, France). : ETEST® (Biomérieux, France)

Full text

Case report

We describe the case of a 73-year old male, who was referred to our tertiary referral hospital for a laparoscopic pylorus-resecting pancreatoduodenectomy because of an intraductal papillary mucinous neoplasm. Three days after surgery, the patient developed fever; a CT-scan revealed peritonitis and leakage from the hepaticojejunostomy. Multiple drains were placed and piperacillin-tazobactam 18 g/day was administered. The abdominal fluid was cultured, revealing amongst others Citrobacter freundii, which was confirmed as carbapenemase-producing using the carbapenem-inactivation method (CIM) [1]. The patient improved and piperacillin-tazobactam was stopped 10 days after drainage.

Shortly after, the patient developed fever again following the placement of a percutaneous transhepatic cholangiogram (PTC) drain. Piperacillin-tazobactam was administered again, and blood cultures revealed the carbapenemase-positive C. freundii, in addition to wildtype Raoultella planticola, Proteus mirabilis and Enterococcus faecium. A spill from the PTC-drain was suspected because of the polymicrobial nature of the blood culture and a clinically stable and quickly recovering patient.

In the next few months, the patient improved and deteriorated several times and was admitted to the Intensive care unit (ICU) twice. These clinical deteriorations were mostly caused by new intra-abdominal collections, which were drained repeatedly. Cultures revealed a carbapenemase-producing Klebsiella oxytoca and a carbapenemase-producing R. planticola. Despite the patient’s status as a known carrier of carbapenemase-producing Citrobacter freundii, he consistently was treated with piperacillin-tazobactam and drainage and improved rapidly.

Several months later, the patient developed a fever due to a new abdominal collection. Cultures showed the previously found NDM-producing C. freundii. However, it also contained carbapenemase-producing S. marcescens, which was clonal with a previously susceptible strain. Piperacillin-tazobactam and drainage did not improve the situation of the patient, thus meropenem and colistin were administered. Due to stagnation of symptoms, antibiotic treatment was eventually changed to aztreonam and ceftazidime-avibactam. Unfortunately, the patient succumbed to septic shock.

Over a period of five months, 47 Enterobacterales isolates were identified from 30 clinical cultures (of which 15 CIM positive) and 21 (of which 13 CIM positive) screening cultures of this patient using MALDI-TOF MS (Biomérieux, France) and later verified with WGS (supplemental methods). In total, four different carbapenemase-producing Enterobacterales (C. freundii,* K. oxytoca*,* R. planticola and S. marcescens)* were cultured with two of them having clustering wildtype strains. (Fig. 1) WGS and bioinformatics are described in the supplementary methods and elsewhere [2, 3]. Antimicrobial susceptibility testing is shown in Fig. 2. All C. freundii isolates cultured, belonged to sequence type 22. These isolates encoded for the following AMR genes: aac(3)-IId-like, aac(6’)-Ib-cr-like, aadA1, aadA16-like, blaCMY−48, blaNDM−7, blaTEM−1B, mph(A), QnrB6, ARR-3, sul1, sul2, tet(D), dfrA1, dfrA27 and the plasmid replicons Col440I and IncX3. The IncX3 plasmid encoded the blaNDM−7 gene and a bleomycin resistance bleMBL and was identified as conjugative, as indicated by mob suite typing. When comparing the C. freundii blaNDM−7 plasmid to internationally reported sequences, one identical plasmid (PV022885.1) in a K. pneumoniae from a Canadian study [4], and manyother highly similar plasmids with more than 99,9% sequence identity, query coverage and identical sequence orientation were identified via the web interface of BLAST, being present in E. coli, Enterobacter species, Klebsiella species, Citrobacter species, Serratia marcescens and Morganella morganii isolates (as of 22-08−2025). WGS demonstrated that the carbapenemase-producing K. oxytoca ST35, R. planticola and S. marcescens found in later cultures also carried a blaNDM−7 on an IncX3 plasmid similar to the initial C. freundii. Compared to the C. freundii resistance plasmid, K. oxytoca had 1 SNPs, S. marcescens had 1 SNPs and 1 indel and R. planticola had no SNPs or deletions. However, the blaNDM−7 plasmid of R. planticola, showed an insertion of two blaTEM−1 and IS3000 family insertion sequences at the start of the plasmid (Supplementary Figs. 1 and 2). Only the S. marcescens plasmid had a 100% exact match with a different plasmid in the NCBI database, namely LC807797.1, a K. pneumoniae isolate from food or water samples from food markets in Dhaka, Bangladesh [5]. Furthermore, WGS showed that both the wildtype and bla_NDM−7_positive strains of the S. marcescens in our patient clustered together and differed only 1 and 2 SNPs as inferred by split k-mer analysis (SKA).Fig. 1. An overview of hospital admissions, relevant cultures and antimicrobial treatment of our patientFig. 2Overview of the tested antibiotic MIC-values and interpretation of the CPE and their wildtypes if available. (EUCAST version 13.1) S: Susceptible, R: Resistant, I: Susceptible with increased exposure. *: Sensititre™ Gram Negative EUMDROXF AST Plate (Thermofisher, United States of America). **: ‘NI’ is given when no interpretation was possible based on the EUCAST clinical breakpoints when the culture was taken. ***: VITEK® 2 (Biomérieux, France). ****: ETEST® (Biomérieux, France)

Discussion

While the concept of horizontal gene transfer has been suggested before [6–9], in vivo evidence in clinical practice with previous wildtype strains is scarce [10] and therefore our report is of additional value. The strength of our case-report lies in the longitudinal sampling and detailed molecular diagnostics of resistance acquisition within a single patient over five months. Specifically, we document the transition from wild type phenotypes to CPE in vivo, in a Dutch patient without further external exposure, supported by whole-genome sequencing and plasmid typing.

The prevalence of CRE/CPE during the admission period was 0.25% for Enterobacterales other than E. coli, K. pneumoniae and E. cloacae in the Netherlands. The most frequently identified carbapenemase encoding genes were blaOXA-48, blaOX-48-like, blaNDM-1 and blaNDM-5. Notably, blaNDM-7 was only found 16 times out of a total 579 CPE isolates submitted to the RIVM according to theNethMap2023 [11] and confirmed by our colleagues of the RIVM.[personal communication C. Jamin].

During the first months of admission, the patient was repeatedly treated with piperacillin-tazobactam, despite the fact that the isolates were not susceptible. These antibiotics were chosen because of the limited availability of new broader spectrum antibiotics in our low CPE setting and because the patient improved under therapy, likely due to source control. However, the use of broad spectrum that did not treat the organisms may have resulted in cell stress, inducing horizontal gene transfer and subsequent selection of CPE in vivo. Also, the described gene insertion and duplication in the R. planticola with bla_TEM−1_were previously described by the exposure to piperacillin-tazobactam treatment [12]. The use of antibiotics should be evaluated when the identified pathogens are not susceptible to prescribed treatment, highlighting the need to prioritize source control.

Horizontal transfer of mobile genetic elements can facilitate the emergence of resistant strains that can lead to outbreaks. This presents a significant challenge in clinical settings since the presence of multiple species can obscure the outbreak and because therapeutic options become limited leading to increased morbidity, mortality and treatment costs [13–15].

Supplementary methods

WGS and bioinformatics

As part of the national surveillance for CPE, the CPE isolates were sequenced at the national public health institute RIVM, using Maxwell DNA isolation and Illumina DNA prep before sequencing on a P1 2 × 150 flowcell on a NextSeq 2000. For long read sequencing the same Maxwell DNA extraction method was used, and sequencing libraries were made using the Rapid barcoding kit (V14) and sequenced on R10.4.1 minION flowcells on a GridION device. Non-CPE isolates were sequenced at the medical microbiology department of the Maastricht university medical centre. Whole genome sequencing was performed by DNA isolation using MasterPure complete DNA and RNA extraction kit, following supplemented protocol. Short-read sequencing libraries were prepared using NexteraXT kit (Illumina), freely choosing combination of sequencing adapters. Sequencing was performed using a V2 2 × 250 flowcell on the MiSeq instrument (Illumina). Hybrid de novo assembly was performed using unicycler. Species determination by WGS was done by querying the genomes on Type Strain Genome Server (TyGS) Genetic similarity among isolates based on SNPs were determined using SKA on the raw reads [16]. Plasmid sequences were annotated as conjugative, mobilizable or inmobile by Mob suite [17]. Plasmid similarities and genotyping was inferred using Plasmidsimilarity, which uses k-mer (k = 31) Jaccard similarity and ABRicate for AMR gene and plasmid replicon typing. Coding sequences were annotated using BAKTA [18]. All bioinformatic tools were ran using default settings, unless specified otherwise.All bioinformatic tools were ran using default settings, unless specified otherwise.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary file 1 (DOCX 344 KB)