Characterization of 2 Klebsiella pneumoniae carbapenemase–producing Enterobacterales isolated from canine rectal swabs

Abstract

Globally, carbapenemase-producing Enterobacterales (CPE) cause life-threatening, hospital-acquired infections in people, and have been reported recently among veterinary patients. Organisms that produce a Klebsiella pneumoniae carbapenemase (KPC) are one of the most common CPE isolated from people but have been reported only rarely in animals. We characterized 2 KPC-producing Enterobacterales isolated from companion animal rectal swabs during the response to an outbreak caused by a strain of bla_NDM-5 Escherichia coli. Both isolates were characterized by whole-genome sequencing (WGS) and analysis. The first isolate (case A) was from an immunosuppressed 6-y-old Yorkshire Terrier and was identified as E. coli (ST372) with a bla_KPC-18 gene and an IncFII plasmid. The second isolate (case B) was from a 3-y-old Labrador Retriever with acute diarrhea and was identified as Citrobacter koseri with a bla_KPC-2 gene, multiple plasmids (ColRNAI, pKPC-CAV1193), and a putative enterotoxin gene (senB). Further research is needed to determine what role animals might play in the epidemiology of CPE in communities. It is imperative that all CPE isolated from companion animals be fully characterized by WGS and the associated case examined. All veterinary isolates should be sequenced and shared for surveillance, monitoring, and investigation purposes.

Carbapenemase-producing Enterobacterales (CPE) have been recognized as one of the most important threats to human healthcare given their propensity to cause life-threatening hospital-acquired infections. ⁹ CPE are able to colonize the gastrointestinal tract of people and animals and can go undetected, which makes them a challenge for infection prevention. ⁹ The main classes of carbapenemase enzymes are Klebsiella pneumoniae carbapenemase (KPC), New Delhi metallo-beta-lactamase (NDM), Verona integron-encoded metallo-beta-lactamase (VIM), imipenemase (IMP), and OXA-48/Oxa-48–like. ⁹ Reports of CPE have increased in companion animal veterinary medicine^7,8,18,21,22; descriptions of CPE outbreaks in veterinary settings demonstrate the importance of surveillance and isolate characterization.^7,21

Klebsiella pneumoniae carbapenemase (KPC) enzymes, the first of which (KPC-1) was reported in 2001 in North Carolina, are Ambler class A beta-lactamases (serine beta lactamase), and at the time of discovery were unique from previously described carbapenemase enzymes (class B or metallo-beta-lactamases). ¹⁹ Outbreaks caused by KPC-producing bacteria in human hospitals were initially identified in the northeastern United States, and have since occurred throughout the United States and globally. ¹⁹ KPC is the most common carbapenemase identified in the United States; the metallo-beta-lactamases are more common internationally. ¹⁹ Clinical isolates that produce a KPC can present a diagnostic challenge given the variable imipenem or meropenem minimum inhibitory concentrations (MIC) observed among isolates. Additional phenotypic testing such as the modified Hodge test or modified carbapenem inactivation method (mCIM) are required to confirm production of a carbapenemase.^6,19 Chromogenic media have been shown to have high sensitivity for the detection of KPC-producing organisms from human rectal screening specimens compared to MacConkey agar supplemented with antimicrobials.^19,26 KPC enzymes have been reported in clinical isolates from companion animals such as a dog in Brazil with a UTI caused by a K. pneumoniae harboring a bla_KPC-2 gene, and 2 isolates of Enterobacter cloacae complex with bla_KPC-4 in Ohio.^8,22 We report here 2 isolates of KPC-producing Enterobacterales identified at a veterinary teaching hospital (VTH) during the response to a CPE outbreak caused by a strain of E. coli harboring the bla_NDM-5 gene. We performed whole-genome sequencing (WGS) to characterize the 2 isolates of KPC-producing Enterobacterales.

As part of the response to a previously described outbreak of bla_NDM-5 Escherichia coli at a VTH, dogs were screened for CPE colonization between 2019 and 2020. ⁷ Fecal specimens or rectal swabs were submitted to the microbiology laboratory as part of a new biosecurity surveillance strategy within the hospital, based on public health recommendations to conduct point-prevalence testing. Two different sampling strategies were used to show the absence of in-hospital transmission: 1) a point-prevalence survey was conducted once a week and all hospitalized patients were tested, and 2) paired admission/discharge testing was initiated. Rectal swabs were plated to chromogenic agar (Chromid Carba agar plates; bioMérieux) and incubated at 37°C for 18–24 h. Presumptive CPE colonies were confirmed to produce a carbapenemase by mCIM. ⁶ Organism identification and antimicrobial susceptibility testing were performed on the Vitek 2 (bioMérieux) with the GNID and GN98 cards.

Short-read, paired-end WGS was performed on the 2 confirmed carbapenem-resistant Enterobacterales (CRE) isolates at 2 different reference laboratories. The first isolate (PPP13-8) was sequenced in partnership with the Veterinary Laboratory Investigation and Response Network (Vet-LIRN) according to methodology described previously. ³ The second isolate (50629-19) was sequenced using a fee-for-service commercial laboratory (NextSeq 2000 platform, Nextera library prep chemistry; Illumina). In both cases, raw sequence files were uploaded to the NCBI pathogen detection pipeline for assembly by SKESA assembler, AMRFinderPlus, and phylogenetic analysis.^5,10,24 Phylogenetic analysis within the pipeline identifies isolates within 30 single-nucleotide variations (SNVs) by the maximum compatibility algorithm. ⁵ Assembled libraries were downloaded from the corresponding NCBI accession and additional analysis performed with the multilocus sequence typing (MLST), PlasmidFinder, and VirulenceFinder tools.^2,13,17

For case A, the organism (isolate PPP13-8; NCBI biosample SAMN13262693) was identified as E. coli. The isolate was from a 6-y-old female Yorkshire Terrier diagnosed with meningoencephalitis of unknown origin at another referral specialty hospital where the animal had been hospitalized for 3 d and immunosuppressive therapy was started. Per patient records, no antimicrobial had been prescribed for at least 6 mo prior to admission. At the time of fecal specimen collection for a point-prevalence survey, the dog had been hospitalized overnight at the VTH for prolonged infusion of the immunosuppressive chemotherapeutic drug cytarabine (300 mg/mm²). Concurrently, the dog was treated with oral immunosuppressive therapy of cyclosporine (5.8 mg/kg) and a tapered dose of prednisolone (0.7 mg/kg q12h). The E. coli isolate was typed by MLST as ST372. PlasmidFinder identified an IncFII plasmid that harbored the antimicrobial resistance gene bla_KPC-18 in addition to bla_EC (a serine beta lactamase) and acrF (multidrug export protein). VirulenceFinder 2.0 results identified several virulence factor genes including the urovirulence factor usp and siderophore receptors ireA and iroN (Suppl. Tables 1, 2). There were no isolates within 30 SNVs in the NCBI pathogen detection database by phylogenetic analysis with the maximum compatibility algorithm.

For case B (isolate 50629-19, NCBI biosample SAMN19094653), the organism was identified as Citrobacter koseri. The isolate was cultured from a fecal specimen collected on admission from a 3-y-old, castrated male Labrador Retriever presented to the VTH Emergency Service because of severe acute enteritis. The recent medical history of the dog included bilateral tibial plateau leveling osteotomy 6 mo earlier at our institution (VTH, University of Pennsylvania, Philadelphia, PA, USA), at which time the patient was administered perioperative cefazolin (20 mg/kg 20 min prior to surgery and every subsequent 90 min), and a more recent case of superficial bacterial folliculitis that was treated by a primary care veterinarian with clindamycin (unknown dose). A fecal sample was submitted to culture for Salmonella spp. and Campylobacter spp., and both cultures were negative. The patient was hospitalized for supportive care for ~3 d, and metronidazole was administered initially intravenously (10 mg/kg q12h) and subsequently transitioned to an oral formulation (12.5 mg/kg q12h for 7 d). Diarrhea resolved within 48 h of admission. WGS analysis identified the antimicrobial resistance genes bla_KPC-2, bla_TEM-1, and bla_CKO. There is no established MLST scheme for C. koseri and no isolates within 30 SNVs in the NCBI pathogen detection database by phylogenetic analysis with the maximum compatibility algorithm. AMRFinderPlus detected bla_CKO and bla_TEM-1 genes. PlasmidFinder detected 2 plasmids (ColRNAI, pKPC-CAV1193). VirulenceFinder 2.0 (E. coli database) detected a gene with 98.89% homology to the senB gene that encodes enterotoxin TieB (Suppl. Tables 1, 3). This may suggest a role in the pathogenesis of the enteritis.

To identify insertions, deletions, or rearrangements, the assembly of 50629-19 (GCA_018478165.1) and annotated pKPC-CAV1193 plasmid (CP013325) were compared using SVIM-asm (v.1.0.2). ¹¹ First, minimap2 (v.2.17) was used in asm5 mode to map the isolate assembly to the plasmid of interest. Then the subsequent bam file was sorted and indexed using Samtools (v.1.10), and SVIM-asm was run in haploid mode to call structural variants.^15,16 Two large deletions (4,878 and 1,066 bp) were detected at positions 4,969 and 29,480, respectively. The deletions were manually cross-referenced with the annotated plasmid. The larger deletion was flanked by 2 resolvase pseudogenes and contained DNA invertase and transposase genes and 1 gene encoding an unidentified hypothetical protein. The smaller deletion included a gene encoding a transposase and a pseudogene of the conjugal transfer protein. The reference plasmid was originally identified in an isolate of K. pneumoniae from a ventilated airway sample of a woman in Virginia, USA. ²³ BLAST analysis of the plasmid revealed homologous plasmids identified in a variety of bacteria from the order Enterobacterales including E. coli, Enterobacter spp., and Citrobacter freundii from human clinical specimens and human hospital environmental sources and water run-off.^1,4 The effects of the deletions on the plasmid are unclear, and further investigation would be required to determine potential roles in a Citrobacter or animal niche.

Risk factors for colonization of animals with CPE have not been well-characterized, but several risk factors previously identified in people were present in these animal cases. These include previous in-patient hospitalization (cases A, B), recent antimicrobial administration (case B), and immunocompromised health status including immunosuppression (case A). ²⁵

The organism isolated from case A is an important finding because ST372 is a lineage of extraintestinal pathogenic E. coli (ExPEC), which has been identified as the most prevalent cause of urinary tract infection (UTI) in dogs. ¹⁴ There is one additional report from France of an E. coli ST372 isolate from a dog with an abdominal infection with that isolate harboring an OXA-48 gene. ¹⁸ Acquisition of carbapenemase genes by pathogenic lineages of E. coli in companion animals should be of concern to the veterinary profession because of the clinical implications of these extensively drug-resistant organisms. In addition, the veterinary profession must be made aware of the potential public health impact of these organisms.

CPE are considered opportunistic pathogens that may cause extraintestinal infections such as ascending UTI or contaminated wound infection, but more often colonize the gastrointestinal tract with no apparent clinical signs. ⁹ However, in case B, the CPE isolate may have contributed to the enteritis of unknown origin. WGS analysis of the isolate identified a gene homologous to the senB gene that produces the putative enterotoxin TieB. ²⁰ A report described the isolation of a bla_KPC-2 C. koseri from a human patient with diarrhea, but a definitive link between the organism, gene, and clinical syndrome was not fully elucidated. ¹²

Use of carbapenems is off-label in veterinary medicine in the United States, and very little is known of their frequency of use by the veterinary profession. The 2 organisms described in our study point to the potential diversity of CPE that may be spreading silently in companion animals. It is critically important to build 1) primary surveillance systems in veterinary diagnostic laboratories to detect CPE in companion animals, and 2) an outbreak response capacity to prevent the continued spread of these extensively drug-resistant organisms. The emergence of CPE in pets must be rapidly contained before they become more prevalent. Veterinary laboratories, or government partners, must take responsibility for the characterization, reporting, and investigation of every case of CPE. Performance of WGS with subsequent sharing of data in public databases should be considered the gold standard method to characterize these organisms to build capacity for the epidemiologic investigation of cases and outbreaks of CPE in animals and people.