Serum biomarkers of Burkholderia mallei infection elucidated by proteomic imaging of skin and lung abscesses

Abstract

Background: The bacterium Burkholderia mallei is the etiological agent of glanders, a highly contagious, often fatal zoonotic infectious disease that is also a biodefense concern. Clinical laboratory assays that analyze blood or other biological fluids are the highest priority because these specimens can be collected with minimal risk to the patient. However, progress in developing sensitive assays for monitoring B. mallei infection is hampered by a shortage of useful biomarkers.

Results: Reasoning that there should be a strong correlation between the proteomes of infected tissues and circulating serum, we employed imaging mass spectrometry (IMS) of thin-sectioned tissues from Chlorocebus aethiops (African green) monkeys infected with B. mallei to localize host and pathogen proteins that were associated with abscesses. Using laser-capture microdissection of specific regions identified by IMS and histology within the tissue sections, a more extensive proteomic analysis was performed by a technique that combined the physical separation capabilities of liquid chromatography (LC) with the sensitive mass analysis capabilities of mass spectrometry (LC-MS/MS). By examining standard formalin-fixed, paraffin-embedded tissue sections, this strategy resulted in the identification of several proteins that were associated with lung and skin abscesses, including the host protein calprotectin and the pathogen protein GroEL. Elevated levels of calprotectin detected by ELISA and antibody responses to GroEL, measured by a microarray of the bacterial proteome, were subsequently detected in the sera of C. aethiops, Macaca mulatta, and Macaca fascicularis primates infected with B. mallei.

Conclusions: Our results demonstrate that a combination of multidimensional MS analysis of traditional histology specimens with high-content protein microarrays can be used to discover lead pairs of host-pathogen biomarkers of infection that are identifiable in biological fluids.

Keywords: Biomarker; Burkholderia mallei; Burkholderia pseudomallei; Calprotectin; FFPE; Formalin-fixed paraffin embedded tissue; Glanders; GroEL; Imaging mass spectrometry; LC-MS/MS; Laser capture microdissection; Melioidosis; Protein microarray.

Figure 1

Overview of the proteomics strategy for biomarker discovery. Abscesses of infection were microscopically identified in thin-sectioned tissues (formalin-fixed, embedded in paraffin) by histology (H&E stained) and localization of bacteria by specific antibody (IHC). The tissue sections were next examined by imaging mass spectrometry (IMS) to identify analyte masses that were localized to the selected regions of interest. Using laser-capture microdissection of select regions of the tissue sections identified by IMS and histology, a more extensive proteomic analysis could then be performed by a technique that combines the physical separation capabilities of liquid chromatography (LC) with the sensitive mass analysis capabilities of mass spectrometry (LC-MS/MS). Finally, the LC-MS/MS data was compared to masses observed by IMS for highest confidence in biomarker identification.

Figure 2

Histopathology and imaging of B. mallei infection in the lung of a Chlorocebus aethiops monkey. Tissues were collected at the time of death. (A) Histological image (H&E stain) of infected lung tissue. (B) Immunohistochemistry using B. mallei-specific antibody to visualize bacteria. (C) Collection of ion intensity maps, spatial resolution of 75 μm, localized to site of infection. Signal intensity images are presented as a blue (lowest) to red (highest). (D) Mass spectrum extracted from the inflammatory abscess (R1-Blue) overlaid with the mass spectrum from two background regions (R2-Red and R3-Green) normalized to total ion current. The ion 1325.6 m/z is more abundant in the abscess compared to the background.

Figure 3

Laser-capture tissue microdissection. (A) Diseased lung tissue was formalin-fixed, embedded in paraffin and stained with Mayer’s hematoxylin. (B) Microscopic image (2X magnification) indicating the region prior to microdissection (green circle). (C) Microdissected target area of the tissue section captured on an adhesive cap. (D) Tissue section showing the removed microdissected area.

Figure 4

Identification of the bacterial protein GroEL from infected Chlorocebus aethiops lung tissue. (A) Total protein coverage attained for GroEL from LC-MS/MS data generated from the LCM tryptic digestion. The peptide sequence YVASGMNPMDLK is boxed in red. (B) CID MSMS spectra of 1356.6, which matched GroEL peptide YVASGMNPMDLK, with a mass deviation of −1.8 ppm. (C) Fragment b and y ions that matched predicted b and y ions are highlighted in blue and red, respectively. Graphed below is the observed mass deviation from predicted mass for each matched fragment ion.

Figure 5

Recombinant B. mallei GroEL analyzed by MALDI TOF-TOF. (A) Total protein coverage of GroEL was 46 percent. YVASGMNPMDLK was observed by MALDI MSMS without methionine oxidation. (B and C) CID MSMS spectra of YVASGMNPMDLK with a mass of 1325.6 m/z. Fragment b and y ions that matched predicted b and y ions are highlighted in blue and red respectively.

Figure 6

Histopathology comparison of the 1325.6 IMS in both diseased lung (A) and skin (B) from individual Chlorocebus aethiops monkeys. Digital microscopic images obtained from immunohistochemical identification of B. mallei in diseased lung and skin sections were overlaid with IMS images for ion 1325.6. Digital images of healthy control tissue sections of lung and skin were used as negative controls.

Figure 7

Differentially detected host and pathogen proteins of B. mallei infection. (A) Heatmap of IgG interactions from Chlorocebus aethiops sera (individuals in columns) with microarrayed B. mallei proteins (in rows). (B) Antibody responses (±SD) to GroEL in comparison to controls were significantly elevated (p ≤ 0.05) in serum collected after infection with B. mallei (middle panel) from four Chlorocebus aethiops monkeys that succumbed to infection, on days 5 (ID 1), 11 (ID 2), 13 (ID 3), 14 (ID 4); and in serum collected on 28 from one monkey that recovered from infection (ID 5). There were no measurable antibody responses from any of the monkeys to the control proteins y1030 and y1025 from the closely related bacterium Yersinia pestis (lower panel). (C) Venn diagram representing the number of qualitative host protein changes observed in the lung (35), the skin (125), and overlapping (15). The proteins overlapping in the Venn diagram are presented in Table 1.