Visualization of Mycobacterial Biomarkers and Tuberculosis Drugs in Infected Tissue by MALDI-MS Imaging

Abstract

MALDI mass-spectrometry imaging (MALDI-MSI) is a technique capable of the label-free identification and visualization of analytes in tissue sections. We have previously applied MALDI-MSI to the study of the spatial distribution of tuberculosis (TB) drugs in necrotic lung granulomas characteristic of pulmonary TB disease, revealing heterogeneous and often suboptimal drug distributions. To investigate the impact of differential drug distributions at sites of infection, we sought to image mycobacterial biomarkers to coregister drugs and bacteria in lesion sections. The traditional method of visualizing Mycobacterium tuberculosis inside lesions is acid-fast staining and microscopy. Directly analyzing and visualizing mycobacteria-specific lipid markers by MALDI-MSI provides detailed molecular information on bacterial distributions within granulomas, complementary to high-spatial-resolution staining and microscopy approaches. Moreover, spatial monitoring of molecular changes occurring in bacteria during granuloma development can potentially contribute to a greater understanding of pulmonary-TB pathogenesis. In this study, we developed a MALDI-MSI method to detect and visualize specific glycolipids of mycobacteria within TB lesions. The biomarker signal correlated well with the bacteria visualized by IHC and acid-fast staining. This observation was seen in samples collected from multiple animal models. Although individual bacteria could not be visualized because of the limit of spatial resolution (50 μm), bacterial clusters were clearly detected and heterogeneously distributed throughout lesions. The ability to visualize drugs, metabolites, and bacterial biomarkers by MALDI-MSI enabled direct colocalization of drugs with specific bacterial target populations (identifiable by distinct metabolic markers). Future applications include assessing drug activity in lesions by visualizing drug-mediated lipid changes and other drug-induced mycobacterial metabolic responses.

Figure 1

Identification of mycobacteria-specific phosphatidylinositols (PIs) and phosphatidylinositol mannosides (PIMs) by the MALDI-MS analysis of intact M. tuberculosis (MTB) and MTB-infected rabbit-lung granulomas (scale bar = 2 mm). (A) MALDI-MS spectrum from intact MTB showing the major PIM species and PI-TBSA. (B) Image of H&E-stained rabbit-lung tissue featuring a large caseous granuloma. (C,D) Image of an adjacent ZN-stained whole-tissue section and a magnified region showing bacilli. (E) MALDI-MS spectrum acquired directly from the caseum of the rabbit granuloma shown in (B–D). The PIM profile shown in (A) is conserved. (F) structure of Ac₁PIM₂. (G) MALDI-MS/MS spectrum of Ac₁PIM₂ (m/z 1413). All the expected product ions are observed.

Figure 2

MALDI-MSI of PIM and PI-TBSA distributions within rabbit caseous-granuloma sections. (A) Image of H&E-stained rabbit-lung tissue featuring a large cavitating granuloma. (B) Confocal-microscopy image of MTB-specific antigen 85 within a serial tissue section. A high bacterial signal is observed in the caseum immediately surrounding the cavity void. (C) ZN-stained serial section. (D) High-magnification zoom of the caseum region containing the high bacterial load. (E) High-magnification zoom of the caseum region in which no bacteria were detected. (F) Manually drawn outline of the bacteria-dense caseum region showing similar distribution to that in (B). (G–I) MALDI-MS images of PI-TBSA, Ac₁PIM₂, and AC₂PIM₂, respectively. The distributions of the three biomarker lipids closely correlate to the distribution of antigen 85 in (B). (J) RGB overlay of the three biomarker images shown in G–I. All three markers colocalize.

Figure 3

Differential distribution of specific PIM species within distinct necrotic mouse granulomas. (A) Native MS images for each lipid displayed alongside an overlay of a coregistered H&E scan of the same tissue section after MSI acquisition. PI-TBSA and Ac₁PIM₂ are homogeneously distributed throughout the caseum, whereas Ac₂PIM₂ signal is absent in a band near the outer caseum edge. (B) Magnified regions of the MALDI-MS images and H&E overlays as indicated by the black box in (A). Note the absence of Ac₂PIM₂ signal within the neutrophilic ring situated between the foamy macrophages and necrotic core.

Figure 4

Evaluation of the spatial-resolving capabilities and sensitivity of the MALDI-MSI method (scale bar = 2 mm). (A) Image of H&E-stained mouse-lung tissue featuring multiple type I caseous granulomas. (B) Image of a serial ZN-stained whole-tissue section. (C) MALDI-MS image of the summed ion intensities of PI-TBSA, Ac₁PIM₂, and Ac₂PIM₂. (D) Coregistration and overlay of the summed ion images in (C) with the ZN-stained reference in (B). (E) Zoomed-in image of the granuloma border (red box in (D)) showing the presence of localized MTB clusters. (F) Zoomed-in image of the coregistered image shown in (D). The summed PI signal is observed to colocalize with the MTB clusters. (G) Zoomed-in image of the bacteria-rich necrotic foci in the type II lesions (black box in (D)). (H) Zoomed-in image showing PIM and MTB colocalization within the necrotic foci of the type II lesions.

Figure 5

Colocalization of TB drugs with the MTB biomarker (PI-TBSA). (A) H&E reference and MALDI-MS images of PI-TBSA and rifampicin (RIF) distribution in lung granulomas from a rabbit administered a single drug dose. RIF signal is not detected in the bacteria-dense necrotic core. (B) H&E reference and MALDI-MS images of PI-TBSA and RIF in lung granulomas from a rabbit administered RIF at steady state for 7 days. Two large cavities are present, and the drug is observed throughout the tissue, including in the bacteria-dense caseum surrounding the open cavity. (C) H&E reference and MALDI-MS images of PI-TBSA and moxifloxacin (MXF) from a rabbit administered a single dose. The drug signal is most abundant within the cellular rim with limited penetration into the necrotic core in which the MTB biomarker is located.