An important mechanism for generation of new genes is by duplication-divergence of existing genes. Duplicationdivergence includes several different submodels, such as subfunctionalization where after accumulation of neutral mutations the original function is distributed between two partially functional and complementary genes, and neofunctionalization where a new function evolves in one of the duplicated copies while the old function is maintained in another copy. The likelihood of these mechanisms depends on the longevity of the duplicated state, which in turn depends on the fitness cost and genetic stability of the duplications. Here, we determined the fitness cost and stability of defined gene duplications/amplifications on a low copy number plasmid. Our experimental results show that the costs of carrying extra gene copies are substantial and that each additional kilo base pairs of DNA reduces fitness by approximately 0.15%. Furthermore, gene amplifications are highly unstable and rapidly segregate to lower copy numbers in absence of selection. Mathematical modeling shows that the fitness costs and instability strongly reduces the likelihood of both sub-and neofunctionalization, but that these effects can be offset by positive selection for novel beneficial functions.

Plasmid-encoded β-lactamases, including non-ESBL enzymes, have a strong influence on the frequency and resistance level of spontaneous carbapenem-resistant mutants. The fitness cost associated with the loss of OmpC/OmpF in E. coli most likely reduces the survivability of porin mutants and could explain why they have not emerged as a clinical problem in this species.

Our results show that antibiotic resistance evolution can occur via several parallel pathways and that new mechanisms may appear after the most common pathways (i.e. β-lactamases and loss of porins) have been eliminated. These findings suggest that strategies to target the most commonly observed resistance mechanisms might be hampered by the appearance of previously unknown parallel pathways to resistance.

The gut is a hot spot for transfer of antibiotic resistance genes from ingested exogenous bacteria to the indigenous microbiota. The objective of this study was to determine the fate of two nearly identical blaCMY-2-harboring plasmids introduced into the human fecal microbiota by two Escherichia coli strains isolated from a human and from poultry meat. The chromosome and the CMY-2-encoding plasmid of both strains were labeled with distinct fluorescent markers (mCherry and green fluorescent protein [GFP]), allowing fluorescence-activated cell sorting (FACS)-based tracking of the strain and the resident bacteria that have acquired its plasmid. Each strain was introduced into an established in vitro gut model (CoMiniGut) inoculated with individual feces from ten healthy volunteers. Fecal samples collected 2, 6, and 24 h after strain inoculation were analyzed by FACS and plate counts. Although the human strain survived better than the poultry meat strain, both strains transferred their plasmids to the fecal microbiota at concentrations as low as 102 CFU/ml. Strain survival and plasmid transfer varied significantly depending on inoculum concentration and individual fecal microbiota. Identification of transconjugants by 16S rRNA gene sequencing and matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) revealed that the plasmids were predominantly acquired by Enterobacteriaceae species, such as E. coli and Hafnia alvei. Our experimental data demonstrate that exogenous E. coli of human or animal origin can readily transfer CMY-2-encoding IncI1 plasmids to the human fecal microbiota. Small amounts of the exogenous strain are sufficient to ensure plasmid transfer if the strain is able to survive the gastric environment.

IncI1 plasmids play a central role in the transfer of antimicrobial resistance genes among Enterobacteriaceae in animals and humans. Knowledge on the dynamics of IncI1 plasmid transfer is limited, mainly due to lack of culture-independent methods that can quantify donor strain survival and plasmid transfer in complex microbial communities. The aim of this study was to develop a culture-independent method to study the dynamics of IncI1 plasmids transfer by fluorescence-activated cell sorting. We genetically modified three wild-type Escherichia coli of animal (n = 2) and human (n = 1) origin carrying bla or bla on two epidemic IncI1 plasmids (pST12 and pST7). Non-coding regions on the chromosome and on the IncI1 plasmid of each strain were tagged with mCherry (red) and GFPmut3 (green) fluorescent proteins, respectively, using lambda recombineering. A gene cassette expressing mCherry and lacI was inserted into the chromosome, whereas the plasmid was marked with a GFPmut3 cassette with LacI repressible promoter. Therefore, gfpmut3 was repressed in donor strains but expressed in recipient strains acquiring the plasmids. We demonstrated that genetic engineering of the strains did not affect the growth rate and plasmid transfer-ability in filter and broth matings. A proof-of-concept experiment using the CoMiniGut, an in vitro model of the colon, proved the validity of our method for studying the survival of wild-type E. coli and horizontal transfer of IncI1 plasmids under different pH and oxygen conditions. The dual-labeling method by fluorescent proteins is useful to determine persistence of exogenous E. coli and transfer dynamics of IncI1 plasmids in microbial communities.