Hidden Resistance Genes Lurk in Susceptible Pseudomonas, Genomic Study Warns

A dangerous secret is hiding inside bacteria that laboratory tests declare to be susceptible. Researchers at a tertiary care hospital in eastern India have shown, through whole-genome sequencing, that a clinical isolate of Pseudomonas aeruginosa which appeared vulnerable to most antibiotics in routine testing actually carried a battery of resistance genes, silently waiting in its genome. The study, published in International Microbiology, examined multidrug-resistant and non-multidrug-resistant isolates side by side and found that while resistant strains showed tight agreement between their genes and their observable behavior, the apparently susceptible isolate harbored a cryptic resistome that standard diagnostics could never have detected.

Pseudomonas aeruginosa is one of medicine’s most formidable opponents. This Gram-negative bacterium accounts for roughly ten to eleven percent of all hospital-acquired infections and can strike the lungs, bloodstream, urinary tract, eyes, and skin, with particular menace for immunocompromised patients. It belongs to the notorious ESKAPE group of pathogens and resists treatment through multiple overlapping strategies: enzymes that destroy antibiotics, altered drug targets, reduced membrane permeability, active efflux pumps, and protective biofilms. The World Health Organization has designated it a priority pathogen, and recent surveillance has documented a steady rise in carbapenem-resistant strains between 2014 and 2024, driven by the organism’s remarkable genetic plasticity and its propensity for horizontal gene transfer.

The core problem the researchers set out to address is a diagnostic blind spot. Traditional phenotypic susceptibility testing, whether by disk diffusion or automated systems, measures what a bacterium does under laboratory conditions, not what it is genetically capable of doing. Molecular panels, meanwhile, only look for pre-established markers and can miss novel or situation-specific determinants. Phenotype-based approaches are slow and open to interpretation, while targeted molecular methods are costly and limited in scope. Neither reliably detects silent resistance determinants or early adaptive changes, which means a patient’s infection may appear treatable on paper while the underlying genome tells a more troubling story.

The investigation began with 1,295 culture-positive Pseudomonas aeruginosa specimens collected at the Central Laboratory of IMS and SUM Hospital in Odisha between January 2023 and December 2024. All isolates underwent antimicrobial susceptibility testing using both the Kirby-Bauer disk diffusion method and the VITEK-2 automated system, following CLSI and EUCAST standards. Multidrug resistance was defined using the internationally recognized Magiorakos criteria. To select representative strains for deep sequencing, the team applied principal component analysis and hierarchical clustering across twelve antibiotics, calculating a resistance burden score for each isolate. From this landscape they chose three isolates spanning the spectrum: two multidrug-resistant strains, designated 0040 and 1609, and one largely susceptible non-MDR isolate, 0727, drawn from diverse specimen sources including urine, pus, and tracheal aspirates, and from both hospital- and community-acquired settings.

Species identity was confirmed by 16S rRNA sequencing, and genomic DNA was sequenced on an Illumina NovaSeq platform using paired-end reads. After quality trimming with Trimmomatic, cleaned reads were mapped to the well-characterized PA14 reference genome with BWA-MEM, achieving greater than 92 percent coverage at 30-fold depth for all three isolates. Variant calling with the GATK HaplotypeCaller applied deliberately stringent filters, retaining only high-confidence homozygous alternate variants with perfect allele frequency, and SnpEff annotated the functional consequences. Notably, none of the isolates carried loss-of-function mutations in the mismatch repair genes mutS, mutL, or uvrD, ruling out a hypermutator state and indicating that the elevated variant loads reflected deep lineage divergence rather than runaway mutation. Resistance genes were identified against the Comprehensive Antibiotic Resistance Database using strict confidence thresholds that admitted only perfect and strict hits.

The genomic arithmetic told a striking story. The multidrug-resistant isolates diverged from PA14 by approximately 69,000 variants, compared with roughly 58,700 for the non-MDR strain, indicating substantial evolutionary separation despite superficially similar phenotypes among the resistant pair. In the MDR isolates, genotype and phenotype agreed closely across five antibiotic classes. Their resistance was anchored by beta-lactamase variants PDC-67 and OXA-396, chromosomal enzymes that become dangerous when regulatory adaptations amplify their output. Crucially, the team detected variants in the regulatory elements ArmR and cprS, which are known to drive overexpression of these beta-lactamases, providing a coherent mechanistic explanation for the observed multidrug resistance.

The non-MDR isolate 0727 told a very different and more unsettling tale. Despite being susceptible to most antibiotic classes in phenotypic testing, it carried a gyrA T83I mutation, a well-known fluoroquinolone resistance-associated change, along with the PDC-1 and OXA-847 beta-lactamase variants. None of these determinants were expressed as measurable resistance. The pattern constitutes what the researchers call a silent resistome: resistance genes present and intact but transcriptionally dormant or functionally masked, perhaps awaiting regulatory shifts, efflux upregulation, or stress-mediated induction to switch on. Similar genotype-phenotype discordance has been documented in Shigella and in Gram-negative uropathogens, reinforcing the emerging consensus that gene presence alone does not equal resistance, and that quinolone resistance in particular typically requires synergistic factors beyond single point mutations.

To place these clinical isolates in global context, the team constructed a phylogenetic tree from 454 genomes, combining their three isolates, the PA14 reference, and 450 publicly available assemblies selected for quality and recency. Pairwise distances computed with Mash revealed that the two MDR isolates clustered tightly in a single clade with short terminal branches, indicating recent common ancestry and minimal divergence between them, embedded among publicly available clinical genomes. The closest neighbor to this pair was a 2021 clinical isolate from South Korea. In contrast, isolate 0727 occupied a distinct terminal branch in a separate clade, grouping most closely with diverse lineages that included a 2007 cystic fibrosis isolate from Denmark, a pattern consistent with weaker antibiotic selection and greater phylogenetic spread among susceptible strains. The findings echo prior work showing that hospital-derived resistant strains often form compact phylogenetic groups under shared selective pressure, raising the possibility of common transmission pathways that would require formal epidemiological tracing to confirm.

The study also surfaced an intriguing metabolic signal. The MDR isolates produced only diffuse fluorescent green pigmentation in broth culture, while the non-MDR strain displayed a stronger greenish-blue pigment gradient, differences the authors link cautiously to the metabolic cost of sustaining resistance. Maintaining efflux pumps and drug-destroying enzymes is energetically expensive, and previous research has shown that multidrug-resistant strains often suppress costly secondary metabolites such as pyocyanin to conserve resources for survival. The authors stress that these pigment observations are visual only and require quantitative biochemical confirmation, but they add a compelling dimension to the picture of resistance as an adaptation with measurable physiological trade-offs.

The clinical implications are sobering. A phenotypically susceptible isolate can carry latent determinants with the theoretical potential to become active under antibiotic pressure, meaning that a seemingly safe treatment choice could, in principle, select for emergence of resistance from within the infecting population. The authors argue that genomic data should be integrated into surveillance alongside phenotypic testing, and they call for transcriptomic studies to reveal how silent resistomes are regulated and when they might be roused. They acknowledge the limitations of sequencing only three isolates from a single institution and emphasize that larger, longitudinally sampled collections with functional validation are needed. Still, their strategy of using multivariate phenotypic screening to select a small, representative subset for whole-genome sequencing offers a practical template for resource-limited settings, where sequencing every isolate remains out of reach. As antimicrobial resistance accelerates worldwide, this study makes clear that what diagnostic labs cannot see in a Petri dish may already be written in the bacterial genome, waiting for its moment.

Beyond the immediate findings, the study illustrates how reference-guided sequencing workflows can be adapted to routine clinical laboratories. By mapping reads against the well-characterized PA14 reference rather than attempting full de novo assembly, the researchers kept computational demands modest while still resolving tens of thousands of variants, an approach that smaller hospital laboratories in low- and middle-income settings could realistically adopt as sequencing costs continue to fall.

The reliance on the Comprehensive Antibiotic Resistance Database also underscores a broader shift in resistance detection. Curated databases with defined confidence levels allow laboratories to distinguish high-confidence resistance determinants from ambiguous hits, reducing the false alarms that have historically plagued in silico resistance prediction. This standardization is becoming increasingly important as genomic surveillance programs expand globally and as public health agencies move toward harmonized interpretation of resistance genotypes.

The phylogenetic placement of the isolates carries practical weight as well. Tight clustering of the multidrug-resistant pair, alongside a clinical isolate from South Korea, fits a pattern increasingly documented worldwide in which resistant hospital strains spread across borders through patient movement, medical tourism, or shared equipment lineages. Distinguishing such imported or transmitted clones from locally evolved ones is central to infection control, and the Mash-based distance approach used here offers a rapid screen for that purpose.

Finally, the silent resistome concept reframes how susceptibility results should be read. A susceptible report reflects conditions in a growth medium at a single moment, not the full adaptive potential encoded in the genome. As regulatory shifts, efflux changes, or selective pressure during therapy can awaken dormant determinants, longitudinal monitoring of patients and institutional surveillance that pairs phenotypic testing with periodic sequencing may become essential to anticipate resistance before it emerges at the bedside.

Subject of Research: Genotype-phenotype concordance and cryptic resistomes in clinical Pseudomonas aeruginosa characterized by comparative whole-genome sequencing.

Article Title: Comparative genomics reveals genotype-phenotype concordance and cryptic resistomes in clinical Pseudomonas aeruginosa

Article References: Comparative genomics reveals genotype-phenotype concordance and cryptic resistomes in clinical Pseudomonas aeruginosa. (n.d.). https://doi.org/10.1007/s10123-026-00892-3

Image Credits: AI Generated

DOI: 10.1007/s10123-026-00892-3

Keywords: Pseudomonas aeruginosa, antimicrobial resistance, whole-genome sequencing, multidrug resistance, silent resistome, genotype-phenotype concordance, beta-lactamase, phylogenetics, hospital-acquired infection, genomic surveillance, ESKAPE pathogens, gyrA mutation