Resistance Gene Carriage Predicts Growth of Natural and Clinical Escherichia coli Isolates in the Absence of Antibiotics

摘要

Bacterial pathogens that carry antibiotic resistance alleles sometimes pay a cost in the form of impaired growth in antibiotic-free conditions. This cost of resistance is expected to be a key parameter for understanding how resistance spreads and persists in pathogen populations. Analysis of individual resistance alleles from laboratory evolution and natural isolates has shown they are typically costly, but these costs are highly variable and influenced by genetic variation at other loci. It therefore remains unclear how strongly resistance is linked to impaired antibiotic-free growth in bacteria from natural and clinical scenarios, where resistance alleles are likely to coincide with other types of genetic variation. To investigate this, we measured the growth of 92 natural and clinical Escherichia coli isolates across three antibiotic-free environments. We then tested whether variation of antibiotic-free growth among isolates was predicted by their resistance to 10 antibiotics, while accounting for the phylogenetic structure of the data. We found that isolates with similar resistance profiles had similar antibiotic-free growth profiles, but it was not simply that higher average resistance was associated with impaired growth. Next, we used whole-genome sequences to identify antibiotic resistance genes and found that isolates carrying a greater number of resistance gene types grew relatively poorly in antibiotic-free conditions, even when the resistance genes they carried were different. This suggests that the resistance of bacterial pathogens is linked to growth costs in nature, but it is the total genetic burden and multivariate resistance phenotype that predict these costs, rather than individual alleles or mean resistance across antibiotics.IMPORTANCE Managing the spread of antibiotic resistance in bacterial pathogens is a major challenge for global public health. Central to this challenge is understanding whether resistance is linked to impaired bacterial growth in the absence of antibiotics, because this determines whether resistance declines when bacteria are no longer exposed to antibiotics. We studied 92 isolates of the key bacterial pathogen Escherichia coli ; these isolates varied in both their antibiotic resistance genes and other parts of the genome. Taking this approach, rather than focusing on individual genetic changes associated with resistance as in much previous work, revealed that growth without antibiotics was linked to the number of specialized resistance genes carried and the combination of antibiotics to which isolates were resistant but was not linked to average antibiotic resistance. This approach provides new insights into the genetic factors driving the long-term persistence of antibiotic-resistant bacteria, which is important for future efforts to predict and manage resistance. KEYWORDS: Escherichia coli , fitness, antibiotic resistanceINTRODUCTIONThe idea that resistance alleles often have negative effects on pathogen fitness is a key concept for the evolution of resistance to antibiotics and other stressors (1). Numerous studies have taken individual resistance alleles originating from both laboratory-evolved resistant mutants (2, 3) and natural isolates (4) and compared their growth in vitro to genotypes lacking the resistance allele. Meta-analyses have confirmed that costly resistance is the predominant result across such studies (5, 6). Although costs are usually measured in the laboratory, there is good evidence that such data are correlated with the fitness effects of the same alleles in nature (7) and in animal models (8) and with the frequency of segregating resistance mutations in clinical isolates (2). As such, the costs of resistance alleles are considered critical determinants of the long-term spread of resistance, influencing the concentrations that select for resistance (9) and persistence in the absence of selecting antibiotics (1, 10).Despite the abundant evidence that individual resistance alleles tend to be costly, this does not necessarily translate to an overall association between resistance and growth in the absence of antibiotics. For example, there is large variation in costs of resistance among different alleles (2, 11), including some with no costs (12), so that selection in nature may be biased toward resistance alleles with relatively small costs (2, 13). The cost of the same resistance mechanism can also vary across different genetic backgrounds (14,–16). An important example of this is compensatory evolution, where resistant bacteria gain further mutations that ameliorate costs (17,–19). Therefore, in some cases we may observe much weaker associations between carriage of resistance alleles (and therefore antibiotic resistance phenotypes) and growth in the absence of selecting antibiotics than we would predict based on fitness costs estimated for individual alleles. Moreover, bacteria can carry multiple resistance alleles simultaneously, and it rem