Abstract: 
In natural environments, antibiotics are often present at sub-inhibitory concentrations. Antibiotic resistance genes (ARGs) may have initially evolved for environmental adaptation rather than resistance to clinically relevant antibiotics. The assembly of aresistome, which is the collection of all ARGs, within a developing microbial community remains poorly understood. Germfree models with low initial bacterial loads can unveil bacterial colonization and resistance development under an antibiotic pressure.Our studies aimed to address the following questions: (1) What are the key ARGs in soil and plant resistomes,and what are the core bacterial families, whether hosting ARGs or not? (2) How do initial bacterial loads and/or an antibiotic treatment influencethe evolution of soil’s and plant’s microbiomes and resistomes?(3) How does the plant resistome affect the antibiotic resistance of the foodborne pathogenSalmonella Typhimurium? (4) How does the resistome develop and look like in the pristine cave, which represents an environment with less human interventions in planetary history?
Soils with varying initial bacterial loads were initially exposed to a non-germfree environment and treated with different concentrations of tetracycline (TET). Both germfreesoil and TET treatments altered bacterial community and resistome profiles compared with untreated natural soil. Multidrug resistance (MDR)genes, especially multidrug efflux pumps, along with their primary host Burkholderiaceae, were dominant in the development of soil resistomes, rather than enhancing TET-related ARGs. Next, we inoculatedS.Typhimurium onto lettuce with germfree or natural leaves and grew them in germfree and natural soil,respectively, under various TET treatments. Only germfree soil affected the bacterial community and resistome profiles, with leaves not serving as a site forSalmonellaresistance development. MDR genes,particularly multidrug efflux pumps, and their primary host Burkholderiaceae, remained the key factors in resistome development in lettuce roots and soils. Lastly, we profiled bacterial communities and resistomes in 47publicly available datasets of 14 pristine cave environments. Microhabitats (sediments, microbial mats, water, biofilms, and minerals) in different environmental conditions led to distinct microbiome profiles. MDR genes, especially multidrug efflux pumps, and their bacterial hosts were highly prevalent and abundant. Variations in multidrug efflux pumps primarily accounted for significant differences in resistomes between microhabitats. This study represents a pioneering investigation into resistome assemblies within germfree soil and lettuce models, underscoring the significance of germfree models for resistome research. The findings emphasize the need to explore the inherent multidrug efflux pumps with their versatile capabilities, given that TET selections did not necessarily select for clinically relevant ARGs,and also may explain that the prevalence of mobile efflux pumps has been on the rise recently. Lastly, the analysis of resistomes in unspoiled cave environments reinforces the notion that resistomes originally emerged as a consequence of bacterial adaptation to their surroundings.