In vivo selection of carbapenem resistance during persistent Klebsiella pneumoniae sequence type 395 bloodstream infection due to OmpK36 deletion

Abstract

Non-carbapenemase-producing carbapenem-resistant Enterobacterales (non-CP CRE) may be associated with a grave outcome. The common underlying mechanism is beta-lactamases and mutations in outer membrane porins. We report a case of a deep-seated infection caused by Klebsiella pneumoniae ST395 not amenable to source control, involving recurrent bloodstream infection, resulting in in vivo selection of carbapenem resistance under therapy. Three consecutive K. pneumoniae blood isolates were studied using short- and long-read sequencing. The genomes were subject to resistome and virulome, phylogenetic, and plasmid analyses. ompK36 porins were analyzed at the nucleotide and amino acid levels. Genomes were compared to 297 public ST395 K. pneumoniae genomes using cgMLST, resistome, and porin analyses and the EuSCAPE project. Relevant ompK36 and micF sequences were extracted and analyzed as above. The three sequential K. pneumoniae blood isolates belonged to the same clone. Subsequent CR isolates revealed a new large deletion of the ompK36 gene also involving the upstream region (deletion of micF). Comparison with public ST395 genomes revealed the study isolates belonged to clade B, representing a separate clone. N-terminal large ompK36 truncations were uncommon in both public data sets. In vivo selection of non-CP CRE K. pneumoniae could have substantial clinical implications. Such selection should be scrutinized through repeated cultures and frequent susceptibility testing during antimicrobial treatment, especially in the context of persistent or recurrent bloodstream infections and when adequate source control cannot be achieved. The occurrence of an unusually large deletion involving the ompK36 locus and upstream micF should be further studied.

Keywords: Klebsiella; carbapenem; in vivo; porin loss; resistance; selection.

Fig 1

Mid-sagittal abdominal computed tomography showing discitis osteomyelitis complex post-L3 kyphoplasty. Imaging demonstrates an abnormality centered on L2/L3 with a loss of intervertebral disc space height and bony destruction, and associated sclerotic change of the L3 superior endplate and the inferior L2 vertebral endplate, which extends to the L4 superior endplate. Soft tissue (green arrow) extends posteriorly from L2/L3 intervertebral space into the central canal, resulting in moderate to severe canal stenosis. Vertebral body and intervertebral disc space height at other levels appear maintained.

Fig 2

Sequence alignment of the annotated ompK36 region of study isolates generated using MUMmer and visualized using pyGenomeViz. The ompK36 gene (blue) is complete in the parent strain and partially deleted in subsequent isolates. The upstream micF gene (green) appears intact in the parent strain (HMC-Kp-1) and is deleted in subsequent isolates (HMC-Kp-2 and HMC-Kp-3).

Fig 3

Alignment of the annotated plasmid AB192 sequence (70 Kb, IncR type) from the three study isolates (three middle plasmids) found to be identical. AB192 appears to have a very high similarity to two publicly available plasmid sequences also derived from K. pneumoniae ST395 [upper and lower plasmids; CP052366.1 (>98% identity and 91.3% coverage) and CP052333.1 (>98% identity and 96% coverage)]. Both plasmids are related to South Korean K. pneumoniae ST395 genome sequences (GCF_012965685.1_ASM1296568v1 and GCF_012971465.1_ASM1297146v1), which appeared to be non-carbapenemase-producing and have an intact ompK36. Antimicrobial resistance genes are colored red. The gray scale denotes direct alignment and the red an inverted alignment.

Fig 4

Minimum spanning tree showing an ad hoc cgMLST analysis of the three sequential K. pneumoniae ST395 blood isolates (red rings) and 297 public ST395 genomes. The scheme consisted of 4,223 loci. Node size is proportional to the number of isolates, and numbers denote differing alleles. The four clades (A–D) are represented by different colors.