The transferability of vanA and vanB glycopeptide resistance determinants with a defined plasmid (n = 9) or chromosomal (n = 4) location between Enterococcus faecium strains of human and animal origins was compared using filter mating (in vitro) and germ-free mice (in vivo) as experimental models. Moreover, the stability of exconjugants in vivo in the absence of antibiotic selection was examined.

Higher transfer rates were observed in vivo for four of six vanA and five of six vanB donor strains. For plasmid-encoded resistance, several log higher transfer frequencies were observed in vivo for some strains. Moreover, the in vivo model supported transfer of plasmid-encoded vanB (1 × 10⁻⁷ exconjugants/donor) when repeated in vitro experiments were negative (estimated < 1 × 10⁻⁹ exconjugants/donor). Readily detectable transfer of plasmid-located vanA and vanB as well as large chromosomal (>200 kb) vanB elements was observed after 24 h. The number of plasmid-mediated vanA exconjugants generally decreased markedly after 3 days. However, exconjugants containing a plasmid harbouring the vanA transposon Tn1546 linked to the post-segregational killing system ω-ε-ζ persisted stably in vivo in the absence of glycopeptides for more than 20 days.

The overall results support the notion that the in vitro model underestimates the transfer potential. Rapid transfer of vanA plasmids from poultry- and pig-derived strains to human faecal E. faecium shows that even transiently colonizing strains may provide a significant reservoir for transfer of resistance genes to the permanent commensal flora. Newly acquired resistance genes may be stabilized and persist in new populations in the absence of antibiotic selection.