The effect of normal, metaplastic, and neoplastic esophageal extracellular matrix upon macrophage activation

Abstract

Introduction: Macrophages are capable of extreme plasticity and their activation state has been strongly associated with solid tumor growth progression and regression. Although the macrophage response to extracellular matrix (ECM) isolated from normal tissue is reasonably well understood, there is a relative dearth of information regarding their response to ECM isolated from chronically inflamed tissues, pre-neoplastic tissues, and neoplastic tissues. Esophageal adenocarcinoma (EAC) is a type of neoplasia driven by chronic inflammation in the distal esophagus, and the length of the esophagus provides the opportunity to investigate macrophage behavior in the presence of ECM isolated from a range of disease states within the same organ.

Methods: Normal, metaplastic, and neoplastic ECM hydrogels were prepared from decellularized EAC tissue. The hydrogels were evaluated for their nanofibrous structure (SEM), biochemical profile (targeted and global proteomics), and direct effect upon macrophage (THP-1 cell) activation state (qPCR, ELISA, immunolabeling) and indirect effect upon epithelial cell (Het-1A) migration (Boyden chamber).

Results: Nanofibrous ECM hydrogels from the three tissue types could be formed, and normal and neoplastic ECM showed distinctive protein profiles by targeted and global mass spectroscopy. ECM proteins functionally related to cancer and tumorigenesis were identified in the neoplastic esophageal ECM including collagen alpha-1(VIII) chain (COL8A1), lumican, and elastin. Metaplastic and neoplastic esophageal ECM induce distinctive effects upon THP-1 macrophage signaling compared to normal esophageal ECM. These effects include activation of pro-inflammatory IFNγ and TNFα gene expression and anti-inflammatory IL1RN gene expression. Most notably, neoplastic ECM robustly increased macrophage TNFα protein expression. The secretome of macrophages pre-treated with metaplastic and neoplastic ECM increases the migration of normal esophageal epithelial cells, similar behavior to that shown by tumor cells. Metaplastic ECM shows similar but less pronounced effects than neoplastic ECM suggesting the abnormal signals also exist within the pre-cancerous state.

Conclusion: A progressively diseased ECM, as exists within the esophagus exposed to chronic gastric reflux, can provide insights into novel biomarkers of early disease and identify potential therapeutic targets.

Keywords: Macrophage; decellularization; esophageal adenocarcinoma; extracellular matrix; gastroesophageal reflux; metaplasia.

Figure 1.. ECM preparation.

(A) An established rat surgical model of EAC (Levrat) progresses from normal squamous epithelium to metaplastic, villiform, columnar epithelium (Barrett’s esophagus) to neoplastic, glandular cell growth with invasion into the basement membrane (esophageal adenocarcinoma) over a period of 33 weeks. (B) Representative H&E confirmation of diseased regions. Normal, metaplastic, and neoplastic tissue were dissected and pooled by diseased region. Scale bar = 25 µm (C) Decellularization efficacy assessed by absence of nuclei by H&E and DAPI stain (n=3). Scale bar = 250 µm. (D) Representative H&E and DAPI stains of human normal and human neoplastic native tissue, and ECM showing decellularization efficacy (n=3). Scale bar = 250 µm.

Figure 2.. ECM hydrogel ultrastructure.

Normal, metaplastic and neoplastic ECM hydrogels from rat source tissue were prepared for scanning electron microscopy and imaged at 2,000x and 10,000x magnification (n=2, 5 technical replicates). The three types of ECM hydrogels show a porous fibrillar network. Scale bar = 10 µm for 2,000x magnification, 1 µm for 10,000x magnification.

Figure 3.. Global and targeted proteomic analysis of normal and neoplastic ECM.

(A,E) Global mass spectrometry analysis and (B-D) targeted proteomic analysis of human normal and neoplastic ECM (n=3, 5 technical replicates). (A) Principal component analysis (PCA) of human normal and neoplastic ECM and native tissue by global proteomic analysis. (B) Nanomolar concentration of total ECM per gram of tissue for normal and neoplastic tissue by targeted proteomics (n= 3, means ± SEM). * P < 0.05. (C) Partial least squares discriminate analysis (PLSDA) analysis showing compositional differences between normal and neoplastic ECM by targeted proteomics. (D) Hierarchical clustering analysis of 18 ECM and ECM-associated proteins quantified in all neoplastic and normal ECM samples by targeted proteomics, presented as a heat map, with log₂ transformed protein concentration (nmol/g). (E) Volcano plot of differentially expressed proteins in neoplastic ECM to normal ECM (fold change) by global proteomics. Fold change threshold set to >2 or <−2, significance threshold set to P < 0.05, student’s t-test. Labeled proteins indicate ECM and ECM-associated proteins.

Figure 4.. Normal, metaplastic, and neoplastic ECM promote a distinctive M1-like and M2-like signature.

A human mononuclear cell line (THP-1) was activated to naïve macrophages (M0). The M0 macrophages were treated with normal, metaplastic, and neoplastic ECM hydrogels (250 µg/mL) from rat source tissue, positive controls for M1-like activation (LPS/IFNγ) and M2-like activation (IL-4) or negative controls (pepsin, media) for 24 hours and tested for a panel of (A) pro-inflammatory and (B) anti-inflammatory genes and transcription factors. Heterologous urinary bladder matrix (UBM) was included as a heterologous tissue control, as a biomaterial known to promote an M2-like activation. (n=4, technical duplicates, means ± SEM). (C) TNFα concentration of THP-1 cells treated with normal, metaplastic, and neoplastic ECM from human source tissue for 24 hours by ELISA (n=4, technical duplicates). Patient ID table shows the combination of samples used from 7 patients (A-G). The normal and neoplastic ECM was patient-matched (“n1”), distinctive (“n2”), or combined patient-matched (“n3”, “n4”). Only 1 Barrett’s (metaplastic) human tissue sample was available and used for all 4 replicates. (means ± SD). * P < 0.05, ** P < 0.01, *** P < 0.005, **** P < 0.0001.

Figure 5.. Immunolabeling of ECM treated macrophages.

A human mononuclear cell line THP-1 was activated to naïve macrophages (M0). The M0 macrophages were treated with normal, metaplastic, and neoplastic ECM hydrogels (250 µg/mL) from rat source tissue; positive controls for M1-like activation (LPS/IFNγ) and M2-like activation (IL-4); negative controls (pepsin, media); or heterologous urinary bladder matrix (UBM) for 24 hours. (A) Macrophages were fixed and labeled with TNFα (“M1” like) in green, IL1RN (“M2 like”) in orange, pan macrophage marker CD11b in red, and DAPI in blue. Representative images shown. Scale bar = 150 µm. Pepsin and UBM control images are in Figure S4. (B) Images were quantified in CellProfiler. * P < 0.05 (n=3, 5 technical replicates, 3 images/replicate, means ± SEM).

Figure 6.. Metaplastic and neoplastic ECM increased migration of normal esophageal epithelial cells through macrophage paracrine effects.

Macrophages were conditioned with medium, M1-like positive stimulus (LPS/IFNγ), M2-like positive stimulus (IL-4), pepsin (25 µg/mL), UBM (250 µg/mL), or normal, metaplastic, and neoplastic ECM (250 µg/mL) from rat source tissue for 24 hours. Heterologous urinary bladder matrix (UBM) was included as a heterologous tissue control, as a biomaterial known to promote an M2-like activation. The secretome of the conditioned macrophages was collected for 4 hours in serum-free medium and used as a chemoattractant in a Boyden chamber assay using normal esophageal epithelial cells (Het-1A). (A) Representative images of migrated Het-1A cells to each treatment and (B) quantification (n= 3, technical quadruplicates, means ± SEM). * P ≤ 0.05, ** P ≤ 0.01, *** P ≤ 0.001, **** P ≤ 0.0001.