Suppression of pancreatic cancer liver metastasis by secretion-deficient ITIH5

Abstract

Background: Previously, we identified ITIH5 as a suppressor of pancreatic ductal adenocarcinoma (PDAC) metastasis in experimental models. Expression of ITIH5 correlated with decreased cell motility, invasion and metastasis without significant inhibition of primary tumour growth. Here, we tested whether secretion of ITIH5 is required to suppress liver metastasis and sought to understand the role of ITIH5 in human PDAC.

Methods: We expressed mutant ITIH5 with deletion of the N-terminal secretion sequence (ITIH5Δs) in highly metastatic human PDAC cell lines. We used a human tissue microarray (TMA) to compare ITIH5 levels in uninvolved pancreas, primary and metastatic PDAC.

Results: Secretion-deficient ITIH5Δs was sufficient to suppress liver metastasis. Similar to secreted ITIH5, expression of ITIH5Δs was associated with rounded cell morphology, reduced cell motility and reduction of liver metastasis. Expression of ITIH5 is low in both human primary PDAC and matched metastases.

Conclusions: Metastasis suppression by ITIH5 may be mediated by an intracellular mechanism. In human PDAC, loss of ITIH5 may be an early event and ITIH5-low PDAC cells in primary tumours may be selected for liver metastasis. Further defining the ITIH5-mediated pathway in PDAC could establish future therapeutic exploitation of this biology and reduce morbidity and mortality associated with PDAC metastasis.

Fig. 1. Full-length ITIH5 (ITI) is secreted extracellularly and cleaved in PDAC cells, whereas secretion-deficient ITIH5Δs (Δs) is not secreted extracellularly.

a Organisation of ITIH5 domain structure labelled by amino acid (AA) position for ITI and Δs expression constructs expressed in the highly metastatic and low endogenous ITIH5-expressing S2-007 human PDAC cell line. Domain map adapted from Himmelfarb et al. Key features: cleavage site (CS); domains (grey): secretion signal peptide (SP), vault protein inter-α-trypsin domain (VIT), von Willebrand type A domain (vWA), multicopper oxidase domain (MCOD) and FLAG epitope (black). Labelled locations of ITIH5 and FLAG antibody binding. b Expression of full-length (ITI) and secretion-deficient ITIH5 (Δs) in S2-007 cells and immunoprecipitation (IP) from conditioned media to detect secreted ITIH5. ITIH5 antibody (binds on N-terminal side of cleavage site at AA578-606) and FLAG antibody (binds on the C-terminal side of CS after the end of ITIH5 open-reading frame). c Western blotting of extracellular vesicle (EV) preparations (particle-size range 20–300 nm, mean 155 nm measured using Nanosight^TM particle analysis device) for ITIH5. Only minimal ITIH5 is present in EVs. d, e Western blotting of ITIH5 in cellular fractions. Two separately conducted cellular fractionations are shown per blot where either cytoplasmic fraction was separated from nuclear fraction or membrane/membrane-associated fraction was separated from cytoplasmic fraction. Antibodies used were the N-terminal-detecting ITIH5 (D) or C-terminal-detecting FLAG (E). Extracellularly, large and small fragments of ITIH5 are present. f ITIH5 is detected in the cytoplasm and nucleus using immunocytochemistry. Scale bar is 25 µm. g Locations of predicted ITIH5 nuclear localisation signals as predicted by sequence recognition algorithm. The algorithm is described in detail.^, Na/K, sodium/potassium ATPase; GAPDH, glyceraldehyde-3-phosphate dehydrogenase.

Fig. 2. Expression of ITIH5 changes cell morphology and migratory capacity.

a Expression of full-length ITIH5 and secretion-deficient ITIH5Δs in metastatic human PDAC cell lines (S2-007 and MIAPaCa-2) correlates with rounded cell morphology and tight epithelial-like clustering. Expression of both ITIH5 and ITIH5Δs slowed cell motility in an in vitro wound healing/scratch assay S2-007, 8 h (b) and MIAPaCa-2, 72 h (d). c, e Quantification of the results from wound-healing assay measuring the percent of wound closure in S2-007 (c) and MIAPaCa-2 (e). * and ** S2-007 ITIH5 *P < 0.001 or ITIH5Δs **P < 0.001 compared to control (Con) using Tukey’s test. Error bars show standard error. For MIAPaCa-2 cells, trends were similar although differences were not statistically significant.

Fig. 3. Expression of full-length ITIH5 or secretion-deficient ITIH5Δs suppresses liver metastasis.

a Representative images of livers following intrasplenic injection of human PDAC cells (S2-007) transfected with control vector (Con), ITIH5 (ITI) or ITIH5Δs (Δs). White arrows highlight liver metastases. b Quantification of gross liver metastases from Con, ITI and Δs (median: 16, 10 and 6, respectively). c Quantification of metastasis size (mean length (cm) of the longest dimension of gross liver metastases) from Con, ITI and Δs (median: 2.65, 1.45 and 1.35, respectively) ***P < 0.001. d Quantification of liver metastasis (by liver weight) from Con, ITI and Δs (median: 1.66, 1.56 and 1.51, respectively). *P = 0.028. e Photomicrograph of representative H&E-stained sections of liver with metastases (dashed lines). f Photomicrograph of representative H&E-stained sections of lung from mice in (a, e). Black arrows highlight lung metastases. No lung metastases were identified in the ITIH5-injected mice. g Ratio of the area of liver metastases compared to the total histologic liver area. No significant differences. h Number of lung metastases in Con and Δs mice (ITI group not shown since there were no lung metastases). No significant differences. Error bars show standard error.

Fig. 4. Expression of ITIH5 correlates with reduced liver metastasis in vivo and in human PDAC.

a Representative images of IHC expression and gross liver images of ITIH5 in control (Con), ITIH5 (ITI) and ITIH5Δs (Δs) expressing PDAC liver metastases. b Quantification of photomicrographs of IHC images from in vivo samples shown in (a) (n = 3). The control group showed frequent low-intensity (0+, 1+) staining, whereas ITI and Δs showed high-intensity (2+, 3+) staining. c Representative images of IHC expression for ITIH5 and gross in vivo liver images from tumours expressing full-length ITIH5. Small metastases (small met) expressed higher ITIH5. In contrast, large met had lower (or lost) ITIH5 expression. d Representative image of one slide from TMA containing matched human primary and metastatic PDAC and stained for ITIH5 using IHC. ITIH5 expression is generally low in both primary and metastatic PDAC. Focal areas of nonspecific staining of necrotic tissue are also visible. e Representative high-magnification images of ITIH5 IHC in normal human pancreas, primary tumour and matched liver metastasis. f Quantification of ITIH5 in normal pancreas, primary PDAC and PDAC metastases from TMA samples.