Reverse engineering antibiotic sensitivity in a multidrug-resistant Pseudomonas aeruginosa isolate

Abstract

Antibiotic resistance is a pervasive and growing clinical problem. We describe an evaluation of a reverse engineering approach for identifying cellular mechanisms and genes that could be manipulated to increase antibiotic sensitivity in a resistant Pseudomonas aeruginosa isolate. We began by chemically mutating a broadly resistant isolate of P. aeruginosa and screening for mutants with increased sensitivity to the aminoglycoside amikacin, followed by performing whole-genome transcriptional profiling of the mutant and wild-type strains to characterize the global changes occurring as a result of the mutations. We then performed a series of assays to characterize the mechanisms involved in the increased sensitivity of the mutant strains. We report four primary results: (i) mutations that increase sensitivity occur at a high frequency (10(-2)) relative to the frequency of those that increase resistance (10(-5) to 10(-10)) and occur at a frequency 10(4) higher than the frequency of a single point mutation; (ii) transcriptional profiles were altered in sensitive mutants, resulting in overall expression patterns more similar to those of the sensitive laboratory strain PAO1 than those of the parental resistant strain; (iii) genes found from transcriptional profiling had the more dramatic changes in expression-encoded functions related to cellular membrane permeability and aminoglycoside modification, both of which are known aminoglycoside resistance mechanisms; and finally, (iv) even though we did not identify the specific sites of mutation, several different follow-up MIC assays suggested that the mutations responsible for increased sensitivity differed between sensitive mutants.

FIG. 1.

(a) Hierarchical clustering (b) and principal component analysis of transcriptional profiles of strain B1, strain PAO1, and sensitive mutants M5 and M31. Replicates are labeled 1 and 2.

FIG. 2.

Differentially expressed genes (log fold difference [base 2] ≥ 1; P ≤ 0.05) between sensitive mutants and strain B1. The pie charts show the distribution of the functional categories of differentially expressed genes between (a) B1 and M5 and (b) B1 and M31. Following the name of each functional category are the number of genes in that category and the percentage of differentially expressed genes for which those genes account.

FIG. 3.

Amikacin MICs for untreated cells and spheroplasts of strain PAO1, strain B1, and mutants M5 and M31. In all cases, the standard deviation was 0 (n = 4).