Klebsiella aerogenes adapts within one host through plasmid loss and tissue-specific virulence

In a striking demonstration of how quickly bacteria can reshape themselves inside a single human body, researchers in Mexico have documented the real-time microevolution of Klebsiella aerogenes, an opportunistic hospital pathogen, in a patient who died of severe viral pneumonia complicated by abdominal sepsis. Over just fifteen days, three genetically clonal bacterial isolates recovered from the same 68-year-old woman revealed a remarkable set of adaptations: the complete loss of a plasmid carrying a carbapenem-resistance gene, an unexpected persistence of drug resistance through a chromosomal porin mutation, and a tissue-specific surge in virulence gene expression that made the final isolate dramatically more lethal in laboratory models. The study, published in International Microbiology, offers one of the most detailed longitudinal portraits yet of intra-host bacterial adaptation during a fatal viral co-infection.

The case began at the Hospital General “Dr. Manuel Gea González” in Mexico City, where the patient was admitted to the intensive care unit with severe COVID-19 pneumonia and concurrent abdominal sepsis. Clinicians recovered three Klebsiella isolates during her nineteen-day hospitalization: isolate 18279 from serosanguineous pleural fluid on day 4, isolate 18280 from purulent peritoneal drainage on day 14, and isolate 18281 from pleural fluid on day 19, the day of her death. Routine automated identification by MALDI-TOF mass spectrometry classified the organisms only generically as Klebsiella species, a diagnostic gap the authors say is increasingly consequential following the taxonomic reclassification of Enterobacter aerogenes into the genus Klebsiella. Only whole-genome sequencing and Average Nucleotide Identity analysis, which returned values above 98 percent against reference genomes, definitively confirmed all three isolates as Klebsiella aerogenes.

Genetic fingerprinting using ERIC-PCR and pulsed-field gel electrophoresis showed identical electrophoretic profiles across all three isolates, confirming a strict clonal relationship and establishing a clean longitudinal framework for studying evolution within a single host. All three belonged to Sequence Type 93, Clonal Complex 3, capsular type KL1, a lineage that international surveillance data identify as one of the dominant clonal groups driving human K. aerogenes infections across North America, Asia and Europe. Whole-genome assemblies achieved 100 percent completeness, and comparative analysis revealed a pangenome of 5,025 genes, partitioned into a stable 5,000-gene core and just 25 accessory shell genes whose presence and absence fluctuated noticeably across the three time points, providing an early genomic signal that this clonal population was not standing still inside its host.

The most surprising genomic event was the disappearance of a 6.1 kilobase ColKP3-type plasmid, designated AC129, which carried the blaOXA-232 carbapenemase gene. Isolates 18279 and 18280 harbored the plasmid and its resistance gene, but both the gene and the replicon were entirely missing from the final isolate, 18281. To rule out an assembly artifact, the researchers mapped raw sequencing reads from 18281 back onto the AC129 reference sequence. The result was unambiguous: the 18281 reads yielded an average coverage depth of only 0.98-fold, in stark contrast to the 2,151-fold depth observed for plasmid-bearing isolate 18280, definitively validating a true in vivo plasmid curing event that occurred while the patient was actively receiving complex antimicrobial therapy, including carbapenems, caspofungin and a later switch to tigecycline.

What makes this loss so intriguing is that carbapenem resistance did not disappear with the plasmid. All three isolates remained firmly resistant to imipenem and meropenem, prompting the team to hunt for chromosomal resistance mechanisms. Their attention settled on the outer membrane porin gene ompK36. While classic high-level carbapenem resistance in Klebsiella pneumoniae typically involves a stereotypic Gly115-Asp116 di-amino acid insertion in extracellular loop 3, that insertion was absent here. Instead, computational topology modeling with the PRED-TMBB2 server confirmed a stable 16-stranded transmembrane beta-barrel harboring a distinct modification: a 13-amino acid elongation between residues 160 and 214 that produces a significant spatial shift in extracellular loop 4. This structural elongation, the authors argue, physically constricts the pore and restricts the influx of bulky hydrophilic beta-lactam molecules, sustaining carbapenem resistance independently of any plasmid-borne enzyme and rendering the lost blaOXA-232 plasmid functionally redundant.

Phenotypic testing revealed that the plasmid-cured late isolate was also the most dangerous. In Galleria mellonella larval infection models, isolate 18281 drove larval survival down to just 21.8 percent by day 5 post-infection, significantly lower than the 49.5 percent survival seen with isolate 18279 and the 61.8 percent seen with 18280. In dexamethasone-immunosuppressed BALB/c mice, the 18281 isolate induced a significantly more severe and progressive body weight loss by days 4 and 5 compared with its clonal predecessors, while causing no morbidity at all in immunocompetent control mice, a result indicating that the enhanced pathogenicity is contingent on host immune status rather than a uniform increase in intrinsic virulence. Tissue burden quantification at 18 hours post-infection told the same story: isolate 18281 achieved average bacterial loads of 3,740 colony-forming units per gram in the intestine and 11,100 CFU/g in the liver, dramatically outperforming both earlier isolates, with Kruskal-Wallis analysis confirming statistically significant differences across strains for both organs.

The mechanistic thread connecting these observations emerged from reverse-transcriptase quantitative PCR performed on bacteria recovered from infected mouse tissues. The virulence genes examined, irp1, a structural gene of the yersiniabactin iron-acquisition system, and clbA, a colibactin genotoxin gene, both reside on the ICEKp10 mobile genetic island, whose sequence was completely identical across all three isolates. Yet their expression diverged dramatically. Within hepatic tissue, irp1 expression climbed in a striking stepwise fashion along the clinical timeline: a modest 1.643-fold increase in isolate 18279, a 7.12-fold induction in isolate 18280, and a robust 15.66-fold hyper-activation in the plasmid-cured isolate 18281. In the intestine, irp1 was heavily repressed in 18279, unchanged in 18280, and upregulated 1.918-fold only in 18281. Conversely, clbA was intensely and consistently downregulated in vivo across both organs and nearly all isolates, dropping to as little as 0.0001-fold of baseline levels in intestinal tissue.

The authors interpret this pattern as a deliberate transcriptional prioritization. Colibactin is a genotoxin associated with DNA double-strand breaks in host cells and long-term genomic instability, including colorectal carcinogenesis, whereas yersiniabactin is a specialized iron-scavenging siderophore that becomes essential in the iron-restricted microenvironment of deep organs such as the liver. The data support a model in which K. aerogenes represses tissue-damaging genotoxin production while overexpressing iron-sequestration machinery to favor persistence during acute systemic dissemination. Because the ICEKp10 island and its promoter regions are sequence-identical across all three strains, the hyper-activation seen in isolate 18281 must be driven by alternative regulatory mechanisms or downstream networks rather than by changes in the virulence genes themselves, highlighting a form of adaptation invisible to standard genomic comparison.

The researchers are careful to frame their findings as a strong biological correlation rather than a proven causal chain. The study is a single-patient case series with a temporal resolution of only three sampling points, so the observed shifts could reflect stochastic sampling variation rather than a directional evolutionary trajectory. Functional genetic validations, such as irp1 or clbA knockout mutants or plasmid AC129 complementation assays, were not performed, and the enhanced murine morbidity was exclusively observed under pharmacological immunosuppression with dexamethasone. In vitro growth kinetics added another layer of nuance: the plasmid-bearing intermediate isolate 18280 actually grew faster and denser in rich media than both its predecessors and the plasmid-cured 18281, arguing against a simple metabolic-burden explanation for the plasmid loss and pointing instead to in vivo curing driven by stochastic replication failures or host-derived physiological pressures, including localized antibiotic concentration profiles.

Taken together, the study delivers a vivid warning about transitional clonal variants in critically ill patients undergoing viral-mediated immune dysregulation. Rather than progressing along a uniform, linear path, intra-host bacterial microevolution appears to operate as a heterogeneous pool of variants undergoing structural plasmid instability, accessory genome remodeling and compartmentalized transcriptional reprogramming at different anatomical sites. The authors, led by Luis Duarte-Zambrano and Ulises Garza-Ramos of Mexico’s National Institute of Public Health, argue that recognizing these dynamic transitional phases is essential for improving diagnostic protocols, which currently fail to resolve K. aerogenes at the species level on automated platforms, and for designing therapeutic strategies against emerging opportunistic pathogens whose behavior inside a single host can shift faster than conventional susceptibility testing can track.