Plasmid-resolved genomics and machine learning-assisted profiling reveal Enterobacter soli as an overlooked host of a blaNDM-1-bearing IncX3 plasmid

Background Plasmid-mediated carbapenem resistance is a major public health concern, yet the role of under-characterized Enterobacter species in carrying clinically important resistance plasmids remains insufficiently defined. In particular, plasmid-resolved genomic information on bla NDM-1 -positive Enterobacter soli is scarce, and it remains unclear whether clinical E. soli can harbor conserved bla NDM-1 -bearing IncX3 plasmids related to those circulating among other Enterobacterales hosts. Methods In this study, we investigated carbapenem-resistant E. soli strain HP81 using complete genome sequencing, antimicrobial susceptibility testing, resistome analysis, plasmid characterization, comparative genomics, and machine learning-assisted plasmid profiling. A public dataset of 1,064 IncX3 plasmids was further used to evaluate whether the key plasmid of strain HP81 conformed to a broader bla NDM -positive IncX3 plasmid profile. Results Strain HP81 was identified as a multidrug-resistant E. soli isolate belonging to sequence type 3367 (ST3367) and carried four plasmids. The key plasmid, pHP81-3, was a 54,652-base-pair IncX3 plasmid carrying bla NDM-1 , ble , bla SHV-12 , multiple mobile genetic elements, and transfer-associated genes. Comparative analysis showed that pHP81-3 was closely related to bla NDM -bearing IncX3 plasmids from other Enterobacterales hosts and contained a conserved bla NDM-1 – ble–trpF–dsbD–cutA resistance module with variable flanking regions. Machine-learning-assisted profiling assigned pHP81-3 to a bla NDM -positive IncX3 plasmid profile, and a without- ble sensitivity model indicated that this classification was not solely dependent on the bla NDM -associated ble feature. Conclusion These findings provide plasmid-resolved evidence that clinical E. soli can serve as an overlooked host of a conserved IncX3- bla NDM resistance plasmid. By combining complete plasmid characterization, comparative genomics, and machine learning-assisted profiling of public IncX3 plasmids, this study expands current understanding of the host range and genomic context of IncX3- bla NDM plasmids and provides a quantitative framework for profiling clinically important bla NDM -positive IncX3 plasmid signatures.