The Role of Antibiotic Resistance Genes in the Fitness Cost of Multiresistance Plasmids

Abstract

By providing the bacterial cell with protection against several antibiotics at once, multiresistance plasmids have an evolutionary advantage in situations where antibiotic treatments are common, such as in hospital environments. However, resistance plasmids can also impose fitness costs on the bacterium in the absence of antibiotics, something that may limit their evolutionary success. The underlying mechanisms and the possible contribution of resistance genes to such costs are still largely not understood. Here, we have specifically investigated the contribution of plasmid-borne resistance genes to the reduced fitness of the bacterial cell. The pUUH239.2 plasmid carries 13 genes linked to antibiotic resistance and reduces bacterial fitness by 2.9% per generation. This cost is fully ameliorated by the removal of the resistance cassette. While most of the plasmid-borne resistance genes individually were cost-free, even when overexpressed, two specific gene clusters were responsible for the entire cost of the plasmid: the extended-spectrum-β-lactamase gene bla_CTX-M-15 and the tetracycline resistance determinants tetAR. The bla_CTX-M-15 cost was linked to the signal peptide that exports the β-lactamase into the periplasm, and replacement with an alternative signal peptide abolished the cost. Both the tetracycline pump TetA and its repressor TetR conferred a cost on the host cell, and the reciprocal expression of these genes is likely fine-tuned to balance the respective costs. These findings highlight that the cost of clinical multiresistance plasmids can be largely due to particular resistance genes and their interaction with other cellular systems, while other resistance genes and the plasmid backbone can be cost-free. IMPORTANCE Multiresistance plasmids are one of the main drivers of antibiotic resistance development and spread. Their evolutionary success through the accumulation and mobilization of resistance genes is central to resistance evolution. In this study, we find that the cost of the introduction of a multiresistance plasmid was completely attributable to resistance genes, while the rest of the plasmid backbone is cost-free. The majority of resistance genes on the plasmid had no appreciable cost to the host cell even when overexpressed, indicating that plasmid-borne resistance can be cost-free. In contrast, the widespread genes bla_CTX-M-15 and tetAR were found to confer the whole cost of the plasmid by affecting specific cellular functions. These findings highlight how the evolution of resistance on plasmids is dependent on the amelioration of associated fitness costs and point at a conundrum regarding the high cost of some of the most widespread β-lactamase genes.

Keywords: antibiotic resistance; fitness cost; plasmid-mediated resistance.

FIG 1

Relative growth rates of E. coli MG1655 transconjugants with different clinical plasmids. The value for the parental strain (containing no plasmid) is set to 1, and all growth rates of the plasmid-carrying strains are relative to that. Gray indicates strains carrying IncI plasmids, and black indicates IncF plasmids. Error bars denote standard deviations from 4 biological and 2 technical replicates. Statistical analysis was performed by one-way analysis of variance (ANOVA) adjusted using Dunnett’s test (*, adjusted P value of <0.05; **, adjusted P value of <0.01).

FIG 2

Relative fitness of mutants with altered conjugation efficiency. Error bars denote standard deviations from 16 biological replicates, including dye swaps. Statistical analysis was performed by one-way ANOVA adjusted using Dunnett’s test (***, adjusted P value of <0.001).

FIG 3

(A) Resistance region of pUUH239.2 (41 kb). Black arrows display IS26 elements, dark gray arrows display resistance genes, and nonlabeled arrows demonstrate transposase genes (tnp) or other nonresistance genes. (B) Relative fitness of deletion mutants compared to the native pUUH239.2 plasmid. (C) Relative fitness of targeted deletion mutants compared to the native pUUH239.2 plasmid. Error bars denote standard deviations from at least 20 biological replicates, including dye swaps. Statistical analysis was performed by one-way ANOVA adjusted using Dunnett’s test (*, adjusted P value of <0.05; **, adjusted P value of <0.01; ***, adjusted P value of <0.001).

FIG 4

(A) RT-qPCR data of antibiotic resistance genes from the pUUH239.2-containing strain. The gene expression of the resistance genes is relative to those of the housekeeping genes cysG and hcaT. Error bars denote the standard deviations from three biological and three technical replicates. (B) Violin plot representing the distribution of transcript abundances (in log₁₀ reads per kilobase per million [RPKM]) of chromosomal (dark gray) and plasmid pUUH239.2 (light gray) genes. The dots denote the median values of the distribution. The resistance gene bla_CTX-M-15 is one of the top 10 most-expressed genes. chr, chromosomal. (C) Relative gene expression of the β-lactamase genes bla_CTX-M-15 and bla_CTX-M-14 from clinical plasmids. Error bars denote the standard deviations from three biological and three technical replicates. (D) Effect on the maximum growth rate by the overexpression (by the addition of 0.05% l-arabinose) of resistance genes from pBAD18 expression vectors. All values are relative to the value for the empty vector, which is set to 1. Error bars denote the standard deviations from 4 biological replicates.

FIG 5

(A and B) Relative growth rates of bla_TEM-1 (A), bla_CTX-M-15 (B), as well as hybrid genes with an exchanged signal sequence (SS) when overexpressed from pBAD18. (C) Relative growth rates of cells expressing only the signal sequence of the β-lactamases or the β-lactamases without the signal sequence. Induction was performed with 0.01% l-arabinose, and values for all strains are relative to those for the empty pBAD18 expression vector. Error bars denote standard deviations from 4 biological replicates.

FIG 6

(A) Relative growth rates of strains with pBAD18 vectors containing ybdM or mmuP induced with 0.01% l-arabinose. Error bars denote standard deviations from four biological replicates. (B) Relative fitness of E. coli MG1655 mutants with deletions of ybdM and mmuP compared to the parental strain. WT, wild type. (C) Relative fitness of E. coli MG1655 mutants with deletions of ybdM and mmuP carrying pUUH239.2 compared to the parental strain carrying native pUUH239.2. (D) Relative fitness of E. coli MG1655 mutants with deletions of ybdM and mmuP carrying pUUH239.2::blaCTX-M-15 compared to the parental strain carrying pUUH239.2::blaCTX-M-15. Competition assays were performed with at least 8 biological replicates, including dye swaps. Statistical analysis was performed by one-way ANOVA adjusted using Dunnett’s test (*, adjusted P value of <0.05; **, adjusted P value of <0.01; ***, adjusted P value of <0.001).