LAP-MALDI MS Profiling and Identification of Potential Biomarkers for the Detection of Bovine Tuberculosis

Abstract

Detecting bovine tuberculosis (bTB) primarily relies on the tuberculin skin test, requiring two separate animal handling events with a period of incubation time (normally 3 days) between them. Here, we present the use of liquid atmospheric pressure (LAP)-MALDI for the identification of bTB infection, employing a three-class prediction model that was obtained by supervised linear discriminant analysis (LDA) and tested with bovine mastitis samples as disease-positive controls. Noninvasive collection of nasal swabs was used to collect samples, which were subsequently subjected to a short (<4 h) sample preparation method. Cross-validation of the three-class LDA model from the processed nasal swabs provided a sensitivity of 75.0% and specificity of 90.1%, with an overall classification accuracy of 85.7%. These values are comparable to those for the skin test, showing that LAP-MALDI MS has the potential to provide an alternative single-visit diagnostic platform that can detect bTB within the same day of sampling.

Keywords: S100-A12; bovine tuberculosis; diagnostics; mass spectrometry; matrix-assisted laser desorption/ionization (MALDI).

Figure 1

Visualization of PCA of samples from healthy animals to determine any bias due to their geographical location. Each geographical location is represented by a different color.

Figure 2

(A) Visualization of the linear discriminant analysis (LDA) for discrimination between healthy, mastitis, and bTB samples. (B) Confusion matrix detailing the assignments based upon the LDA model in panel A using “20% out” cross-validation.

Figure 3

(A) Visualization of the linear discriminant analysis (LDA) for the discrimination between samples from healthy and samples from diseased (bTB and mastitis) animals. (B) Confusion matrix detailing the identification of samples from healthy and diseased animals based upon the LDA model in panel A using “20% out” cross-validation.

Figure 4

(A) Loading plot for the first principal component of the principal component analysis (PCA), indicating the peaks most responsible for discrimination. Peaks labeled with an asterisk belong to the ions of S100-A12 as identified by MS/MS analysis of the [M + 10H]¹⁰⁺ ions at the m/z value of approximately 1056. (B) Relative mean ion signal intensities of the overall strongest S100-A12 ion signal ([M + 12H]¹²⁺; m/z 880) for each sample class. The error bars provide the standard errors.