High-Sensitive Spatial Proteomics for Pancreatic Cancer Progression Analysis

Abstract

Pancreatic cancer remains as one of the most challenging malignancies to diagnose and treat due to the late development of symptoms and limited early diagnostic options. Intraductal papillary mucinous neoplasms (IPMNs) are non-invasive precursors to invasive pancreatic ductal adenocarcinoma (PDAC)and an understanding of the changes in patterns of protein expression that accompany the progression from normal ductal (ND) cell, to IPMN to PDAC may provide avenues for improved earlier detection. In this study, we present an optimized spatial tissue proteomics workflow, termed SP-Max (Spatial Proteomics Optimized for Maximum Sensitivity and Reproducibility in Minimal Sample), designed to maximize protein recovery and quantification from limited laser micro dissected (LMD) samples. Our workflow enabled the identification of more than 6,000 proteins and the quantification of over 5,200 protein groups from FFPE tissue contours of pancreatic tissues. Comparative analyses across ND, IPMN, and PDAC revealed critical molecular differences in protein pathways and potential markers of progression. SP-Max provides a systematic, reproducible approach that significantly enhances our ability to study precancerous lesions and cancer progression in pancreatic tissues at unprecedented resolution.

Keywords: FFPE Tissue; High-sensitivity Mass Spectrometry; Intraductal Papillary Mucinous Neoplasm; Pancreatic Ductal Adenocarcinoma; Spatial Proteomics.

Figure 1.. Workflow of LMD-based FFPE tissue preparation for high-sensitivity spatial proteomics.

Regions of interest (ROIs) were annotated on hematoxylin and eosin (H&E)-stained FFPE tissue sections to differentiate between normal duct (ND) and distinct disease stages. Laser microdissection (LMD) was utilized to precisely isolate these ROIs, with careful slide and cap inspection to ensure proper sample collection from LMD. The SP-Max workflow includes optimized steps to minimize sample loss, particularly through the application of a temperature gradient (+20°C) during the de-cross linking and enzymatic digestion process, enhancing protein recovery. Dissected tissues underwent sonication for tissue lysis and were subjected to controlled heating (80°C for de-cross linking followed by 37°C for digestion). Resulting peptides were loaded on Evotips and analyzed via LC-MS/MS using a data-independent acquisition (DIA) approach. Subsequent data analysis provided high-resolution proteomic patterns across tissue regions, enabling comprehensive spatial proteomics.

Figure 2.. Tissue Staining, areas and thickness for reliable spatial proteomic analysis

(A) Tissue slides stained with Hematoxylin, Toluidine Blue O (TBO), or left unstained (NS). (B) Number of protein groups identified from a 420 μm × 420 μm tissue area with 5 μm thickness using different staining or non-staining methods. (C) Estimated equivalent cell counts based on sample volume (μm³) for various tissue areas and thicknesses, assuming a single cell is 20 μm diameter and volume of approximately 4,000 μm³. (D) Number of identified protein groups across different tissue sample sizes and amounts, varying in area and thickness. (E) Heatmap illustrating correlation between biological replicates and between scaled sample sizes, measured across three different tissue thicknesses (5 μm, 10 μm, and 20 μm).

Figure 3.

Characterization of pancreatic cancer and IPMN proteome. (A) Protein group counts identified across individual samples from PDAC (red), IPMN (green), and ND (blue) tissue sections. A total of 10 cases were analyzed, with 17 marked tissue features corresponding to specific disease conditions. At least three biological replicates were captured from each feature. (B) Cell-type-specific signature proteins derived from a single-cell transcriptome atlas of the human pancreas, highlighting key markers in ductal, acinar, beta, mesenchymal, and other cell types. (C) Principal component analysis (PCA) showing distinct clustering of ND, IPMN, and PDAC tissue samples. PDAC and IPMN, representing cancerous and precancerous tissues of pancreas, respectively, cluster closely, while ND forms a separate group. (D) Heatmap of 1,649 significantly regulated proteins identified by ANOVA (FDR < 0.01, S0 = 0.1), clustered into five distinct expression patterns (labeled A–E). The right panel shows expression profiles of each cluster across ND, IPMN, and PDAC LMD-tissue lesions.

Figure 4.

Biological process enrichment and pathway analysis of protein clusters in pancreatic cancer progression. (A) Biological processes (BPs) enriched in proteins from Cluster E in Figure 3D, which are upregulated in both IPMN and PDAC. A total of 138 proteins with a ≥2-fold change compared to normal duct (ND) were analyzed. (B) Biological processes enriched in proteins from Cluster B in Figure 3D, which are downregulated in IPMN and PDAC. A total of 70 proteins with a ≥2-fold change compared to ND were analyzed. (C) Single-sample gene set enrichment analysis (ssGSEA) of all proteins from Figure 3D, demonstrating the enrichment of Reactome pathways for individual tissue samples. The enrichment scores for each sample are provided in Supplementary Table S6.

Figure 5.

Altered proteins in IPMNs. (A) Bubble plot depicting Reactome pathway enrichment of Cluster C proteins. The size of each bubble represents the number of proteins contributing to the pathway, while the color indicates statistical significance (FDR). (B) Quantitative differences in expression levels between ND, IPMN, and PDAC for the nine proteins enriched in the O-linked glycosylation of mucins pathway. Each protein demonstrates elevated level of proteins in IPMN compared to ND, followed by downregulation in PDAC.