Serial proteomic analysis identifies small extracellular vesicle-MASP2 as an early biomarker of chemotherapy response in advanced pancreatic cancer

Abstract

Background: The average survival of advanced pancreatic cancer (APC) is 6-12 months with first-line chemotherapy. Only one-third receive second-line treatment. No early biomarker exists to guide chemotherapy efficacy before tumor progression occurs. Circulating small extracellular vesicles (sEVs) are a potential source of biomarker discovery.

Methods: Longitudinally collected sEVs from chemotherapy treated APC patients at pre-treatment, remission and relapse underwent proteomic profiling by mass spectrometry (MS). GO and KEGG analyses assessed differential protein characteristics, while protein-protein interactions and upstream analyses explored potential mechanisms. A candidate biomarker was validated by ELISA in larger patient cohorts of responders and non-responders. Gene knock-down and overexpression studies and tumor immunohistochemistry (IHC) evaluated potential function and localization.

Results: MS identified 34 proteins unique to remission, 132 unique to treatment resistance, and 9 differential across both phases. Complement cascade alterations best reflected response to treatment. Lectin pathway component MASP2 (Mannose-Binding Lectin-Associated Serine Protease 2) emerged as a predictive biomarker: >20 % decline in sEV-MASP2 levels at month 2 (M2) of chemotherapy predicted response in 72 % of responders, whereas >20 % increase predicted treatment resistance in 73 % of non-responders. sEV-MASP2 at M2 was prognostic for survival (11 vs. 8 months; p = 0.0037), unlike CA 19-9 (11 vs. 12 months) and retained significance when CA 19-9 was unevaluable. Functional data indicated that sEV-MASP2 alterations largely reflect systemic rather than tumor site-specific activity.

Conclusions: Complement pathway activity tracks with chemotherapy response and resistance in PC. Changes in sEV-MASP2 may serve as an early predictive/prognostic biomarker, helping to improve decision making in this lethal malignancy.

Keywords: Biomarker; Extracellular vesicles; MASP2; Pancreatic cancer; Proteomics.

Fig. 1

Mass-spectrometry workflow and identification of differentially expressed sEV proteins in discovery dataset. (A) Changes in CA 19–9 levels in patients (IDs: 33, 41, 47, 70 and 74) across pre-t, BR and treatment TR timepoints, with corresponding representative CT scans. (B) Schematic for experimental setup and replication sets for mass spectrometry. (C) Venn diagram showing overlapping and unique DE sEV-proteins across pre-t, BR and TR timepoints. (D) Volcano plot showing DE sEV-proteins between pre-t and BR, highlighting upregulated (red) and downregulated (blue) proteins (|fold change| ≥ 1.4; Tukey’s HSD p value ≤ 0.05). (E) Heatmap with hierarchical clustering depicting z-score normalized differential expressions of sEV-proteins between pre-t and BR for the five tested patients. (F) Volcano plot showing DE sEV-proteins between BR and TR, highlighting upregulated (red) and downregulated (blue) proteins (|fold change| ≥ 1.4; Tukey’s HSD p value ≤ 0.05). (G) Heatmap with hierarchical clustering depicting z-score normalized differential expressions of sEV-proteins between BR and TR for the five tested patients. (H) Normalized abundance with corresponding p values of the nine overlapping DE sEV-proteins (AOC1, IDE, PLPBP, MASP2, EPHA10, IGKJ4, MBL2, APOL1, and TF) across pre-t, BR and TR in the five APC patients in the dataset. (I) Heatmap of z-score normalized expression profiles for the nine overlapping DE sEV-proteins across the three timepoints in the discovery dataset.

Fig. 2

GO, KEGG and IPA pathway analyses of differential sEV-proteins across treatment phases in the discovery dataset. (A, B) Circos plot depicting the top significantly enriched GO terms for the biological process (BP), molecular function (MF) and cellular component (CC) categories between (A) pre-t to BR, and (B) BR to TR transitions. (C, D) Significantly enriched KEGG pathways between (C) pre-t and BR, and (D) BR and TR. For C-D, x-axis shows pathway significance as –log10 (p value). (E) Heatmap showing the differential sEV-proteins and their expression patterns (log ratios) within the complement cascade for pre-t to BR and BR to TR transitions, as identified through IPA canonical pathway analysis of the discovery dataset. Expression log ratios: green = downregulation, red= upregulation, scaled by intensity. Pathway directions are indicated by z-score, where blue = inhibition, orange = activation, scaled by intensity. (F) Schematic of the complement cascade highlighting key differential proteins from our dataset (fuchsia circles) that regulate the pathway. Color coding: Orange for predicted activation, blue for predicted inhibition (intensity-scaled). Expression measurements are coded as red for upregulation, green for downregulation (intensity-scaled). For the overlapping proteins, indicated expression measurements are based on BR to TR transition.

Fig. 3

GSEA, PPI and upstream regulator analyses of differential sEV-proteins across treatment phases in the discovery dataset. (A, B) GSEA enrichment for the differential proteins between (A) pre-t and BR, and (B) BR and TR. (C-D) Protein-protein interaction (PPI) network of differential sEV-proteins by STRING analysis between (C) pre-t vs BR with the 47 differential sEV-proteins showing a dense network enriched for complement effectors and immune-regulatory nodes, and (D) BR vs TR with 145 differential sEV-proteins forming a broader interconnections encompassing complement, immune and metabolic clusters. For C-D, the nodes represent the proteins, and the edges indicate the interactions between two proteins. (E) Heatmap showing the top upstream regulators of differential protein profiles across pre-t to BR and BR to TR. Pathway regulation directions are indicated by z-scores where blue = inhibition, orange = activation. Highlighted is the IPA-predicted MYC regulation of differential effector sEV-proteins between pre-t–BR and BR–TR transitions.

Fig. 4

sEV-MASP2 and serum CA 19–9 alterations and survival outcomes in APC cohorts. (A, C) Changes in serum sEV-MASP2 concentrations (pg/mL) of the R patient group (n = 25) tracked across (A) pre-t, BR, and TR, and (C) pre-t, month 2 (M2), and TR. (B,D) Histograms illustrating the groupwise mean alterations in MASP2 concentrations within the R group (n = 25) through (B) pre-t, BR, and TR, and (D) pre-t, M2, and TR timepoints. (E) Changes in serum sEV-MASP2 concentrations (pg/mL) of the NR patient group (n = 15) at pre-treatment (pre-t) and post-treatment (post-t) timepoints. (F) Histograms illustrating the groupwise mean alterations in MASP2 concentrations within the NR group (n = 15) at pre-t and post-t timepoints. (G, H) Waterfall plot showing percentage changes in (G) serum sEV-MASP2 levels, and (H) serum CA 19–9 levels in R and NR cohorts at month 2 compared to pre-t (baseline). For G&H, Patterned bars represent cases with a minimum 50 % reduction in serum CA 19–9 levels from baseline (baseline CA 19–9 reported as >10,000 U/mL); # denotes instances where CA 19–9 value at baseline or month 2 were unavailable. Chemotherapeutic drug (FOLFIRINOX or Gemcitabine/nab-paclitaxel) and PC stage (III or IV) are indicated for each patient. (I-K) Kaplan-Meier (K-M) plot for overall survival (OS) in PC patients. (I) Compares a > 20 % rise (blue) vs. a > 20 % drop (red) in serum sEV-MASP2 levels at month 2 vs. pre-t (n = 60). (J) Compares a >20 % rise (blue) vs. >20 % drop in serum CA 19–9 levels at month 2 compared to pre-t in CA19–9 eligible patients within the APC cohort evaluated for sEV-MASP2 (n = 42). (K) Compares a >20 % rise (blue) vs. >20 % drop in serum CA 19–9 levels at month 2 compared to pre-t in all qualified PC patients (n = 54) with known CA 19–9 changes. (L) ROC curve from a nominal logistic regression model assessing the prognostic efficacy of sEV-MASP2, using a ± 20 % change to predict 6 months OS in PC patients. Statistical significance was evaluated using the Chi-Square test (=14.58, p = 0.0001). (M) K-M plot analyzing survival in PC patients with high bilirubin (≥ 1.3) or low serum CA 19–9 (< 37 U/mL) at pre-t (n = 23), for >20 % rise (blue) vs. > 20 % drop (red) in serum sEV-MASP2 levels at 2 months compared to pre-t. For panels I-K and M, median survival times, hazard ratio (HR), and log-rank test significance are presented to highlight the differences in the survival outcomes between the groups compared.

Fig. 5

Functional assays and IHC assessments of MASP2 in PC cells and tissues. (A) Western blot analysis for MASP2 across PC cell lines (Panc-10.05, Panc-1, Capan2, BxPC3, SW1990, HPAF1, MIA PaCa2, and CFPAC1), and pancreatic epithelial cell line HPDE. (B, C) Western blot for MASP2 in CFPAC1 cell line transfected with (B) control vector (CTRL) or MASP2 OE plasmid, and (C) scrambled control (siCT) or siMASP2. For panels A-C, relative protein levels were normalized to GAPDH. (D) Wound-healing assays were conducted using CTRL, MASP2 OE, siCT, or siMASP2 transfected CFPAC1 cells. Representative images for areas occupied by cells at 0 h and 48 h post-wounding. Scale bars = 1000 µm. (E) Quantitative analysis of wound closure, based on five independent experiments. (F) Representative images from the EdU proliferation assay showing the growth of CFPAC1 cells transfected with CTRL, MASP2 OE, siCT or siMASP248 h post-transfection. GFP staining (green) indicates proliferating cells in the S phase of mitosis, while nuclear DAPI staining (blue) represents all cells present. (G) Quantitative analysis of the percentage of EdU-positive cells in each group, demonstrating the proportion of cells undergoing DNA synthesis. Data presented as Mean ± SEM, n = 5. (H) Abundance and distribution of MASP2 measured by IHC staining in primary (n = 8) and liver metastatic (n = 8) PDAC biopsies, normal pancreas (n = 4) and normal liver (n = 4). IHC panels display representative images for H&E (top panel), negative control (Negative CTRL) (middle panel), and MASP2 (bottom panel) staining in the tissues. Black arrows indicate areas of positive staining in primary and liver metastatic PDAC tissues. Scale bars = 20 µm. (I) Quantitative evaluation of MASP2 expression in acinar and ductal cells from normal pancreas, and tumor cells from primary and liver metastatic PDAC tissues. (J) Quantitative assessment of MASP2 expression in stromal cells in normal pancreas, and in primary and liver metastatic PDAC tissues. Data presented as Mean ± SEM. (K) Box plot depicting MASP2 levels in PDAC tumors (n = 137) vs. paired normal pancreatic tissues (n = 74) from the CPTAC PDAC dataset. z-values represent standard deviations from the median across samples. Log2 spectral count ratios were first normalized within individual sample profiles and then normalized across all samples.