Genome-Wide Association Study of Nucleotide Variants Associated with Resistance to Nine Antimicrobials in Mycoplasma bovis

Abstract

Antimicrobial resistance (AMR) studies of Mycoplasma bovis have generally focused on specific loci versus using a genome-wide association study (GWAS) approach. A GWAS approach, using two different models, was applied to 194 Mycoplasma bovis genomes. Both a fixed effects linear model (FEM) and a linear mixed model (LMM) identified associations between nucleotide variants (NVs) and antimicrobial susceptibility testing (AST) phenotypes. The AMR phenotypes represented fluoroquinolones, tetracyclines, phenicols, and macrolides. Both models identified known and novel NVs associated (Bonferroni adjusted p < 0.05) with AMR. Fluoroquinolone resistance was associated with multiple NVs, including previously identified mutations in gyrA and parC. NVs in the 30S ribosomal protein 16S were associated with tetracycline resistance, whereas NVs in 5S rRNA, 23S rRNA, and 50S ribosomal proteins were associated with phenicol and macrolide resistance. For all antimicrobial classes, resistance was associated with NVs in genes coding for ABC transporters and other membrane proteins, tRNA-ligases, peptidases, and transposases, suggesting a NV-based multifactorial model of AMR in M. bovis. This study was the largest collection of North American M. bovis isolates used with a GWAS for the sole purpose of identifying novel and non-antimicrobial-target NVs associated with AMR.

Keywords: Mycoplasma bovis; antimicrobial resistance; fluoroquinolone; genome-wide association study; macrolide; phenicol; tetracycline.

Figure 1

Frequency distributions of Mycoplasma bovis isolates (y-axis) by minimum inhibitory concentrations (x-axis, µg/mL) for each of the nine antimicrobials. Each colour represents an antimicrobial class. Red = fluoroquinolones (ENRO = enrofloxacin); green = tetracyclines (CTET = chlortetracycline, OXY = oxytetracycline); blue = phenicols (FFN = florfenicol), and purple = macrolides (GAM = gamithromycin, TIL = tilmicosin; TIP = tildipirosin, TUL = tulathromycin, TYLT = tylosin tartrate).

Figure 2

Manhattan plots of variants for enrofloxacin, chlortetracycline, oxytetracycline, and florfenicol resistance. Variants are colour-coded, with fixed effects linear model (green) and linear mixed model (red). The y-axis represents the level of statistical significance [−log10 (p-value)] with position of the nucleotide variant on the x-axis. Significant NVs identified within coding regions are represented with a dot, while significant NVs within non-coding regions are denoted with a hollow diamond.

Figure 3

Manhattan plots of variants for gamithromycin, tilmicosin, tildipirosin, tulathromycin, and tylosin tartrate resistance. Variants are colour-coded, with fixed effects linear model (green) and linear mixed model (red). The y-axis represents the level of statistical significance [−log10 (p-value)] with position of the nucleotide variant on the x-axis. Significant NVs identified within coding regions are represented with a dot, while significant NVs within non-coding regions are denoted with a hollow diamond.