Synergistic Impacts of Organic Acids and pH on Growth of Pseudomonas aeruginosa: A Comparison of Parametric and Bayesian Non-parametric Methods to Model Growth
Synergistic Impacts of Organic Acids and pH on Growth of Pseudomonas aeruginosa: A Comparison of Parametric and Bayesian Non-parametric Methods to Model Growth
复制标题
DOI：
10.3389/fmicb.2018.03196
复制
发表时间：
2019-01-08
影响因子：
5.2
通讯作者：
Lund, Peter A.
中科院分区：
文献类型：
--
作者：
Bushell, Francesca M. L.;Tunner, Peter D.;Lund, Peter A.
不同的弱有机酸对于局部治疗那些对标准治疗无效的机会性病原体感染的伤口具有巨大的潜力。这些酸长期以来一直在食品工业中用作抑菌化合物，并且在某些情况下已经用于临床。不同有机酸的影响随 pH 值、浓度和所使用的具体有机酸的不同而变化，但迄今为止，还没有任何针对机会性病原体的研究以受控和系统的方式检验了这些关键变量之间的详细相互作用。因此，我们综合评估了几种不同的弱有机酸对机会病原体铜绿假单胞菌生长的影响。我们使用半自动读板机生成了两种不同菌株（模型实验室菌株 PAO1 和来自医院烧伤病房的临床分离菌株 PA1054）在不同浓度和 pH 值的一系列有机酸中的生长概况，并具有总共 162,960 个数据点的高水平复制。然后，我们比较了两种不同的建模方法来解释这个时间分辨数据集：参数逻辑回归（有或没有包含滞后阶段的组件）与非参数高斯过程（GP）回归。由于 GP 没有对生长的性质做出事先假设，因此在生长不遵循标准 s 形函数形式的情况下，这种方法被证明是优越的，这在细菌在压力下生长时很常见。乙酸、丙酸和丁酸都比其他测试的酸对生长更不利，尽管 PA1054 在非胁迫条件下比 PAO1 生长得更好，但随着胁迫水平的增加，这种差异很大程度上消失了。正如对有机酸行为方式的了解所预期的那样，它们的效果在与低 pH 值结合时显着增强，其中丙酸的相互作用最大。我们的方法适合于描述压力源之间的组合相互作用，特别是在它们对增长的影响导致逻辑增长模型不适合的情况下。
Different weak organic acids have significant potential as topical treatments for wounds infected by opportunistic pathogens that are recalcitrant to standard treatments. These acids have long been used as bacteriostatic compounds in the food industry, and in some cases are already being used in the clinic. The effects of different organic acids vary with pH, concentration, and the specific organic acid used, but no studies to date on any opportunistic pathogens have examined the detailed interactions between these key variables in a controlled and systematic way. We have therefore comprehensively evaluated the effects of several different weak organic acids on growth of the opportunistic pathogen Pseudomonas aeruginosa. We used a semi-automated plate reader to generate growth profiles for two different strains (model laboratory strain PAO1 and clinical isolate PA1054 from a hospital burns unit) in a range of organic acids at different concentrations and pH, with a high level of replication for a total of 162,960 data points. We then compared two different modeling approaches for the interpretation of this time-resolved dataset: parametric logistic regression (with or without a component to include lag phase) vs. non-parametric Gaussian process (GP) regression. Because GP makes no prior assumptions about the nature of the growth, this method proved to be superior in cases where growth did not follow a standard sigmoid functional form, as is common when bacteria grow under stress. Acetic, propionic and butyric acids were all more detrimental to growth than the other acids tested, and although PA1054 grew better than PAO1 under non-stress conditions, this difference largely disappeared as the levels of stress increased. As expected from knowledge of how organic acids behave, their effect was significantly enhanced in combination with low pH, with this interaction being greatest in the case of propionic acid. Our approach lends itself to the characterization of combinatorial interactions between stressors, especially in cases where their impacts on growth render logistic growth models unsuitable.