Proteomic analysis of formalin-fixed paraffin-embedded tissue by MALDI imaging mass spectrometry

Abstract

Archived formalin-fixed paraffin-embedded (FFPE) tissue collections represent a valuable informational resource for proteomic studies. Multiple FFPE core biopsies can be assembled in a single block to form tissue microarrays (TMAs). We describe a protocol for analyzing protein in FFPE-TMAs using matrix-assisted laser desorption/ionization (MALDI) imaging mass spectrometry (IMS). The workflow incorporates an antigen retrieval step following deparaffinization, in situ trypsin digestion, matrix application and then mass spectrometry signal acquisition. The direct analysis of FFPE-TMA tissue using IMS allows direct analysis of multiple tissue samples in a single experiment without extraction and purification of proteins. The advantages of high speed and throughput, easy sample handling and excellent reproducibility make this technology a favorable approach for the proteomic analysis of clinical research cohorts with large sample numbers. For example, TMA analysis of 300 FFPE cores would typically require 6 h of total time through data acquisition, not including data analysis.

Figure 1

Workflow for investigation of peptide content in FFPE-TMA. Flowchart showing the procedure for in situ proteomics on FFPE-TMA samples by imaging mass spectrometry.

Figure 2

Trypsin and matrix application. (a) Representative acoustic droplet ejection of picoliter volumes of reagent solution from the reagent source reservoir to a MALDI target plate holder. (b) TMA section scanned after 30 cycles of trypsin (top) and matrix (bottom), respectively.

Figure 3

Protein identification workflow. Representative workflow for protein identification by MS/MS. (a) Tissue section is microspotted with trypsin and matrix for MALDI MS/MS analysis. (b) The mass spectrum produced is acquired, and a parent ion of m/z 1,460.9 is selected for fragmentation. (c) Magnification of the parent ion peak at m/z 1,460.9 shown in panel b. The peak at m/z 1,460.9 is the monoisotopic peak for the intact protonated peptide. The three peaks at higher mass, each separated by 1 Da, result from incorporation of one or more of the less abundant isotopes into the molecule, with 13C at a natural abundance of 1.11% being the main contributor. (d) MS/MS spectrum of the 1,460.9 parent ion and analysis of the fragment ions produced. This figure is reproduced, with permission, from reference .

Figure 4

Portrait spotting setup. (a) Description of loading the source reagent reservoir of Portrait 630 instrument. Reagent solution, trypsin or matrix, is slowly pipetted into the source reservoir between the baffle and the reservoir wall. The source is filled with a volume ranging from 500 to 600 µl. (b) Target plate holder view for calibration and sample slide insertion. With the round corner at the holder on the right, place the calibration slide below the ITO slide. (c) Enlarged view of calibration patterns for ejector parameter and X/Y offset calibration, respectively. (d) Enlarged view of the TMA section after trypsin and matrix spotting.

Figure 5

IMS correlation with pathological diagnosis. (a–d) Representation of the statistical classification of four TMA cores correlating with the pathological diagnosis. Pathologist diagnosis is shown in panels a–d; four TMA cores have undergone H&E staining, and pathological analysis using light microscopy. Adenocarcinoma (a,c) and squamous cell carcinoma (b,d) were marked in each biopsy for the cancerous region (red line) and the noncancerous region (green line). (e–h) The marked cancer regions were then co-registered with serial TMA sections analyzed by MALDI IMS and ClinProTools for classification, enabling the spot position of each spectrum to be linked to that same precise histological region in the TMA. Adenocarcinoma is shown in red and squamous cell carcinoma in green. For the statistical classification, mass spectra from the cancer regions of each biopsy were first grouped to create an average spectrum representative of both the adenocarcinoma and squamous cell carcinoma. The peaks present in these average spectra were compared through statistical analysis to identify a subset of peaks that were significantly different between each group to be used as the class identifiers. A support vector machine (SVM) algorithm was used to build a classification model. The SVM model was run against all spectra in the data set and the outcome of each classification was visualized using the class imaging function in FlexImaging software. This model classified two biopsies as squamous cell carcinoma (f,h, peptide images in green) and the other two biopsies as adenocarcinoma (e,g, peptide images in red), in agreement with the pathology diagnosis. This figure is reproduced, with permission, from reference .

Figure 6

MALDI images of three tryptic peptides originating from heat shock protein β-1 detected in two of three squamous cell carcinoma core biopsies analyzed in a TMA. This demonstrates the protein heterogeneity that can be measured in biopsies of the same nominal disease classification. This figure is reproduced, with permission, from reference .