Pangenome Analysis of Proteus mirabilis Reveals Lineage-Specific Antimicrobial Resistance Profiles and Discordant...

Urinary tract infections (UTIs) impose a substantial health care burden with increasing antimicrobial resistance and treatment failure rates. Proteus mirabilis is a challenging UTI pathogen due to intrinsic resistances coupled with formation of crystalline biofilms. We combined pangenome analysis, antimicrobial resistance gene (AMR) predication, and antimicrobial susceptibility testing (AST) to determine whether highly multidrug-resistant (MDR) isolates of P. mirabilis arise from distinct lineages and explored the clinical utility of multilocus sequence typing (MLST). The pangenome of 1,001 P. mirabilis genomes from human urine revealed an open conformation driven by strain diversity and the accessory genome. A total of 213 sequence types (STs) were identified and only 7% had ≥11 genomes, highlighting strain diversity. 93% of the P. mirabilis genomes harbored resistance genes for ≥2 antibiotic subclasses, and 25% were predicted to be resistant to >6 subclasses, confirming a high MDR burden. By focusing on the 15 most prevalent STs, we observed that AMR subclasses were largely lineage-specific. However, isolates with very high resistance gene counts (>20) were distributed across multiple STs, indicating that extreme resistance gene accumulation is not restricted to ST. Comprehensive AST of 27 P. mirabilis clinical isolates further revealed variable genotype-phenotype concordance, indicating unresolved mechanisms of resistance not captured in current AMR prediction databases. In summary, our study emphasizes the critical need to expand genomic surveillance of P. mirabilis to fully understand the complex landscape of AMR in this species and translate these insights into improved diagnostic and therapeutic strategies.

Proteus mirabilis is a clinically-challenging cause of urinary tract infections due to multidrug resistance and its ability to form crystalline biofilms that provide further antibiotic protection. In this study, we sought to determine how well sequence typing and antimicrobial resistance gene prediction correlate with laboratory-based antimicrobial susceptibility testing. By analyzing more than 1,000 P. mirabilis genomes from human urine samples, we found that some resistance patterns were sequence type-specific. However, the genome structure of this species suggests frequent horizontal gene transfer, and the most highly-resistant strains did not cluster by lineage. Importantly, many isolates that appeared "susceptible" based on their genomes were in fact resistant upon laboratory testing, revealing hidden or uncharacterized resistance mechanisms. These findings show that current gene-based prediction tools can miss clinically relevant resistance, underscoring the need for further study to guide effective treatment of P. mirabilis infections.