Combinatorial treatment with statins and niclosamide prevents CRC dissemination by unhinging the MACC1-β-catenin-S100A4 axis of metastasis

Abstract

Colorectal cancer (CRC) is the second-most common malignant disease worldwide, and metastasis is the main culprit of CRC-related death. Metachronous metastases remain to be an unpredictable, unpreventable, and fatal complication, and tracing the molecular chain of events that lead to metastasis would provide mechanistically linked biomarkers for the maintenance of remission in CRC patients after curative treatment. We hypothesized, that Metastasis-associated in colorectal cancer-1 (MACC1) induces a secretory phenotype to enforce metastasis in a paracrine manner, and found, that the cell-free culture medium of MACC1-expressing CRC cells induces migration. Stable isotope labeling by amino acids in cell culture mass spectrometry (SILAC-MS) of the medium revealed, that S100A4 is significantly enriched in the MACC1-specific secretome. Remarkably, both biomarkers correlate in expression data of independent cohorts as well as within CRC tumor sections. Furthermore, combined elevated transcript levels of the metastasis genes MACC1 and S100A4 in primary tumors and in blood plasma robustly identifies CRC patients at high risk for poor metastasis-free (MFS) and overall survival (OS). Mechanistically, MACC1 strengthens the interaction of β-catenin with TCF4, thus inducing S100A4 synthesis transcriptionally, resulting in elevated secretion to enforce cell motility and metastasis. In cell motility assays, S100A4 was indispensable for MACC1-induced migration, as shown via knock-out and pharmacological inhibition of S100A4. The direct transcriptional and functional relationship of MACC1 and S100A4 was probed by combined targeting with repositioned drugs. In fact, the MACC1-β-catenin-S100A4 axis by statins (MACC1) and niclosamide (S100A4) synergized in inhibiting cancer cell motility in vitro and metastasis in vivo. The MACC1-β-catenin-S100A4 signaling axis is causal for CRC metastasis. Selectively repositioned drugs synergize in restricting MACC1/S100A4-driven metastasis with cross-entity potential.

Fig. 1. The MACC1 secretome induces CRC cell migration via S100A4.

Culture medium of MACC1-overexpressing cells stimulated migration of SW480 cells (A), and in HCT116, HT-29 and LS174T (B). The same culture supernatants rescued migration of SW620/shMACC1 that was diminished after depletion of MACC1 (C). Baseline protein (left) and mRNA (right) expression of MACC1 and S100A4 in the human colon carcinoma cell lines SW480, SW620, HCT116, HT-29 and LS174T (D). In a SILAC analysis of cell culture medium of SW480/vector and SW480/MACC1 cells, S100 proteins were secreted de novo in MACC1-overexpressing cells (E). Western blot from cell culture supernatant (sample volume proportional to respective cell count at time of medium harvest) confirmed increased presence of soluble S100A4 (sS100A4) in the secretome of SW480/MACC1 cells (F). Numeric results shown means ± SEM of 3 independent experiments, test for significance with ANOVA and Tukey correction for multiple testing.

Fig. 2. MACC1 and S100A4 robustly identify high risk CRC patients.

IHC of MACC1 and S100A4 in each 2 tumors of non-metastasized and metachronously metastasized primary tumors confirms overexpression of both biomarkers in CRC that yielded metachronous metastases well after surgical removal of the primary tumor (A). MACC1 and S100A4 expressions correlate in CRC tumors. Gene expression levels were analyzed in three additional cohorts of 44, 54 and 117 CRC tumors, respectively. Co-expression was examined with Spearman correlation, and we found a positive correlation of MACC1 and S100A4 expression levels in all datasets (B–D). Kaplan–Meier analysis for MFS and OS of patients based on the MACC1-S100A4 combination panel, test for significance with log rank (Mantel-Cox) test. Combined overexpression of MACC1 and S100A4 was associated with dismal MFS and OS in primary CRC tumors (E), and high levels of MACC1 and S100A4 mRNA transcripts detected in liquid biopsies (preoperative blood samples) predicted poor OS (F). Significant intergroup differences are indicated with asterisks, where applicable.

Fig. 3. MACC1 promotes S100A4 in vitro and in vivo.

Ectopic MACC1 in HCT116 increased the activity of a S100A4 promoter-driven luciferase reporter. Concomitantly, S100A4 was increased in mRNA and protein levels (A top). Knockout of MACC1 in SW620 was followed by reduced S100A4-promoter-driven luciferase as well as decreased S100A4 mRNA and protein expression (A bottom). IHC of S100A4 in tumors of Apc^Min and vil-MACC1/Apc^Min mice confirms overexpression of S100A4 in MACC1-overexpressing littermates (B). In tumors of vil-MACC1/Apc^Min that overexpress human MACC1 (hMACC1, C), mouse-intrinsic S100a4 (mS100a4) was upregulated at RNA level (D).

Fig. 4. MACC1 employs S100A4 through β-catenin signaling to drive cancer cell motility.

MACC1-specific cell motility depends on S100A4. Overexpression of MACC1 in HCT116 cells induced transwell migration, but not in S100A4-KO counterparts (A). MACC1-induced cell migration in HCT116 and SW480 cells, and this effect was reverted by niclosamide, a transcriptional inhibitor of S100A4 (B, C). MACC1 increased S100A4 in presence of DMSO, and three independent β-catenin inhibitors (niclosamide, FH535 and LF3) largely reversed this upregulation on mRNA and protein level to expression levels of HCT116/vector cells (D). Numeric results shown as means ± SEM of 3 independent experiments, test for significance with Student’s t-test, or ANOVA and Tukey correction for multiple testing.

Fig. 5. MACC1 interacts with β-catenin and induces its transcriptional activity via post-translational modification.

In a mass-spectrometry-based analysis of the MACC1 interactome in SW620 cells, several peptides interacting with MACC1 mapped to β-catenin, suggesting a direct PPI (A). Co-IP experiments on whole-cell lysates, cytoplasmic and nuclear protein of SW620 confirmed direct interaction between MACC1 and β-catenin (B). This finding was recapitulated by co-incubating recombinant MACC1 and β-catenin proteins in cell-free Co-IP assays (C). Impact of MACC1 overexpression or MACC1 knock-out is shown on TCF-reporter activity and mRNA expression of β-catenin and the β-catenin/TCF target genes Cyclin-D1, MMP7 and S100A4 in HCT116 and SW620 cells. In HCT116 cells MACC1 overexpression increases TCF reporter activity and TCF target gene expression, while β-catenin gene expression itself is unaffected. In SW620 cells MACC1 knockout decreases TCF reporter activity and TCF target gene expression, whereas β-catenin gene expression is also unaffected (D). In a DigiWest experiment, MACC1-overexpressing HCT116 cells demonstrated increased phosphorylation of p-Ser-552-β-catenin, confirmed in semiquantitative western blots, while total β-catenin protein was not altered (E). Comparative immunoprecipitation of β-catenin protein showed increased binding of TCF4 in HCT116/MACC1 cells, also shown by densitometry (F). Boxplots show means ± SEM of 3 independent experiments, test for significance with Student’s t-test.

Fig. 6. Statins and niclosamide synergize in suppression of CRC cell wound healing and metastasis.

HCT116 human colorectal cancer cells with endogenous MACC1 and S100A4 expression were treated with three concentrations (1.25, 2.5 and 5 µM) of atorvastatin and niclosamide (0.25, 0.5 and 1 µM) alone and in combinations thereof (A). After drug application in vitro wound healing was monitored for 48 h using the IncuCyte live cell imaging system. Atorvastatin and niclosamide were able to reduce wound closure compared to control cells. Combining a statin with niclosamide increased this effect synergistically. Results are shown as means ± SEM of at least 3 experiments. Treating xenografted SCID beige mice (n = 60, 10 animals per group) with human equivalent doses of either atorvastatin (3.25 mg/kg, 20 mg per patient per day), fluvastatin (3.25 mg/kg, 20 mg per patient per day) or niclosamide (250 mg/day, 1.5 g per patient per day) alone reduced metastasis formation in the liver. Confirming the in vitro results, the combination of the tested drugs, each statin with niclosamide, was superior to single drug treatment (B, D). While visualization of xenografted cells by luminescence did not show significant reductions just before mouse killing (C), human cell load in the murine liver was quantifiably reduced in human satellite DNA specific qPCR (D). Numeric results are shown as mean ± SEM, and level of significance indicated as per ANOVA and Dunnett’s correction for multiple testing.