Data from: Associations among antibiotic and phage resistance phenotypes in natural and clinical Escherichia coli isolates

The spread of antibiotic resistance is driving interest in new approaches to control bacterial pathogens. This includes applying multiple antibiotics strategically, using bacteriophages against antibiotic-resistant bacteria, and combining both types of antibacterial agents. All these approaches rely on or are impacted by associations among resistance phenotypes (where bacteria resistant to one antibacterial agent are also relatively susceptible or resistant to others). Experiments with laboratory strains have shown strong associations between some resistance phenotypes, but we lack a quantitative understanding of associations among antibiotic and phage resistance phenotypes in natural and clinical populations. To address this, we measured resistance to various antibiotics and bacteriophages for 94 natural and clinical Escherichia coli isolates. We found several positive associations between resistance phenotypes across isolates. Associations were on average stronger for antibacterial agents of the same type (antibiotic-antibiotic or phage-phage) than different types (antibiotic-phage). Plasmid profiles and genetic knockouts suggested that such associations can result from both colocalization of resistance genes and pleiotropic effects of individual resistance mechanisms, including one case of antibiotic-phage cross-resistance. Antibiotic resistance was predicted by core genome phylogeny and plasmid profile, but phage resistance was predicted only by core genome phylogeny. Finally, we used observed associations to predict genes involved in a previously uncharacterized phage resistance mechanism, which we verified using experimental evolution. Our data suggest that susceptibility to phages and antibiotics are evolving largely independently, and unlike in experiments with lab strains, negative associations between antibiotic resistance phenotypes in nature are rare. This is relevant for treatment scenarios where bacteria encounter multiple antibacterial agents.抗生素耐药性的扩散正推动学界探索控制细菌性致病菌的全新策略。这些策略包括战略性联合使用多种抗生素、利用噬菌体（bacteriophages）对抗耐药菌，以及将两类抗菌剂联合应用。所有这些策略都依赖于不同耐药表型（resistance phenotypes）之间的关联——即对一种抗菌剂耐药的细菌，对其他抗菌剂也会表现出相对易感或耐药的特性，且受此类关联的影响。实验室菌株实验已证实部分耐药表型间存在强关联，但我们仍缺乏对自然种群与临床种群中抗生素耐药表型和噬菌体耐药表型之间关联的定量认知。为解决这一问题，我们对94株自然来源与临床来源的大肠杆菌（Escherichia coli）分离株进行了抗生素与噬菌体耐药性检测。我们在不同分离株的耐药表型间发现了多组正相关关联。平均而言，同类抗菌剂（抗生素-抗生素或噬菌体-噬菌体）之间的关联强度高于不同类抗菌剂（抗生素-噬菌体）之间的关联。质粒图谱分析与基因敲除实验表明，此类关联可由耐药基因的共定位，以及单个耐药机制的多效性效应所导致，其中包括一例抗生素-噬菌体交叉耐药现象。核心基因组系统发育（core genome phylogeny）与质粒图谱可用于预测抗生素耐药性，但噬菌体耐药性仅能通过核心基因组系统发育进行预测。最后，我们利用观测到的关联，预测了参与此前未被表征的噬菌体耐药机制的基因，并通过实验进化验证了该预测结果。我们的研究数据表明，细菌对噬菌体与抗生素的易感性在很大程度上独立演化；且与实验室菌株实验结果不同，自然种群中抗生素耐药表型间的负相关关联十分罕见。这一发现对于细菌暴露于多种抗菌剂的临床治疗场景具有重要参考价值。