Long-read metagenomic characterization of the resistome and mobilome in heavy-metal-contaminated soils

Introduction Heavy-metal contamination is widely proposed to co-select for antibiotic resistance by placing metal- and antibiotic-resistance genes on the same mobile elements, yet the evidence for this rests largely on short-read co-occurrence rather than on direct observation of physical linkage. Methods We applied Nanopore long-read metagenomics to four soils from the Ridder polymetallic mining and metallurgical district of East Kazakhstan, where zinc, cadmium and lead exceed a single local reference site by up to two orders of magnitude. Physical linkage between antibiotic- and metal-resistance genes was tested at the contig, read, genome and community levels against chance expectation. Results Ensemble binning of two complementary assemblies produced 261 metagenome-assembled genomes, 97% of them lacking a GTDB species assignment and dominated by Actinomycetota, together with 73 closed circular elements including complete phages and plasmids. Screening this catalogue recovered 991 antibiotic-resistance and 395 metal-resistance loci. In these metal-enriched soils, the two classes were rarely physically linked. Only four contigs carried both, against 28.9 expected under a degree-preserving null model, a 0.14-fold depletion that was robust to catalogue thresholds; co-carriage on single reads was likewise rare. At the genome level the classes were independent, and in 22 of 24 dual-carrier genomes they occupied different contigs. The mobilome nonetheless carried metal resistance preferentially: contigs called as plasmids were enriched for metal genes relative to the catalogue base rate, and a 142-kb plasmid predicted to be conjugative encoded copper, cadmium/zinc, mercury and silver efflux systems, a complete type-IV secretion system and a CBASS anti-phage defense system, but no antibiotic-resistance gene. No genome belonged to an ESKAPE or WHO-priority pathogen genus. Discussion These soils are therefore a large environmental resistance reservoir, but no direct clinical-risk linkage was demonstrated in the present dataset. Community co-occurrence should not be read as evidence of physical linkage. Co-selection through cross-resistance or co-regulation was not excluded.