Tumor-associated macrophage-derived GDNF promotes gastric cancer liver metastasis via a GFRA1-modulated autophagy flux

Abstract

Purpose: Liver metastasis, a lethal malignancy of gastric cancer (GC) patients, execrably impairs their prognosis. As yet, however, few studies have been designed to identify the driving molecules during its formation, except screening evidence pausing before their functions or mechanisms. Here, we aimed to survey a key driving event within the invasive margin of liver metastases.

Methods: A metastatic GC tissue microarray was used for exploring malignant events during liver-metastasis formation, followed by assessing the expression patterns of glial cell-derived neurotrophic factor (GDNF) and GDNF family receptor alpha 1 (GFRA1). Their oncogenic functions were determined by both loss- and gain-of-function studies in vitro and in vivo, and validated by rescue experiments. Multiple cell biological studies were performed to identify the underlying mechanisms.

Results: In the invasive margin, GFRA1 was identified as a pivotal molecule involved in cellular survival during liver metastasis formation, and we found that its oncogenic role depends on tumor associated macrophage (TAM)-derived GDNF. In addition, we found that the GDNF-GFRA1 axis protects tumor cells from apoptosis under metabolic stress via regulating lysosomal functions and autophagy flux, and participates in the regulation of cytosolic calcium ion signalling in a RET-independent and non-canonical way.

Conclusion: From our data we conclude that TAMs, homing around metastatic nests, induce the autophagy flux of GC cells and promote the development of liver metastasis via GDNF-GFRA1 signalling. This is expected to improve the comprehension of metastatic pathogenesis and to provide a novel direction of research and translational strategies for the treatment of metastatic GC patients.

Keywords: Autophagy; GFRA1.; Gastric cancer; Liver metastasis; Tumor associated macrophage.

Fig. 1

Oncogenic roles of GFRA1 in vivo and in vitro. (A) Protocol of GFRA1-silencing and liver-metastasis modelling. AGS cells were stably transfected with shGFRA1 and then injected into the spleen of nude mice for liver-metastasis formation. (B) Representative images of whole livers from LM mouse models. (C) Representative images of TUNEL (green) signals of above specimens derived from metastatic lesions. (D) Representative IHC images of Ki67-staining of above metastatic AGS tumors. (E) Histograms of TUNEL-positive cells and Ki67-positive cells of metastatic tumors as indicated. (F) Apoptotic curves of AGS (left panel) and HGC-27 (right panel) cells transfected with shGFRA1 or scramble plasmids, cultured in FBS-free medium over five days. (G) Apoptotic curves of AGS (left panel) and HGC-27 (right panel) cells in the presence of recombinant protein rGDNF (20 ng/ml), specific agonists BT18 (50 μM) or BT13 (50 μM), and pre-transfected with shGFRA1 or scramble plasmids. (H) Representative flow cytometric apoptosis assays of AGS (left panel) and HGC-27 (right panel) cells transfected with above plasmids in FBS-free medium. Scale bar: respectively labelled in every module, “ns” not significant, *p < 0.05, **p < 0.01

Fig. 2

Expression profiles of GDNF and GFRA1 in metastatic GC specimens. (A) Relative GFRA1, GDNF and RET mRNA levels in tumor tissues from single primary and metastatic GC patients (n = 10, respectively). (B) Protein levels of GFRA1, GDNF and RET in tumor tissues from single primary and metastatic GC patients. SGC: single gastric cancer, PGC: primary GC and LM: liver metastasis. (C-E) Correlation analysis of protein levels in a tumor microarray by IHC scores (n = 69, in total). Heatmap of GFRA1 levels between primary gastric and metastatic liver lesion shown in C, heatmap of GDNF levels between primary gastric and metastatic liver lesion shown in C, and correlation between GFRA1 and GDNF levels in metastatic lesions shown in E. (F) Representative immunofluorescent images of GDNF (green) in metastatic liver lesions from three GC cases. ANL: adjacent normal liver. (G) Relative CD206 and ACTA2 mRNA levels in tissues from different regions (n = 10), which encode CD206 and SMA, respectively. IM: invasive margin tissues and AL: adjacent liver tissues. (H) Double-staining of GDNF and CD206 in metastatic lesions from three GC cases. (I) Double-staining of GFRA1 and CD206 in metastatic lesions from three GC cases. (J-K) Correlation analysis of protein levels ina tumor microarray by IHC scores (n = 69, in total). Heatmap of GDNF levels and CD206 positive cells in metastatic tissues shown in K, and heatmap of GFRA1 levels and CD206 positive cells shown in L. Scale bar: respectively labelled in every module, “ns” not significant, *p < 0.05, **p < 0.01

Fig. 3

TMA-endowed malignancy of GFRA1-positive cells in vitro and in vivo. (A) Diagram of co-culture assay between tumor associated macrophages (TAMs) and GC cells. (B-C) Representative images and histograms of co-culture colony formation under gradient concentrations of fetal bovine serum (n = 3, respectively). (D) Apoptotic curves of AGS (left panel) and HGC-27 cells (right panel) cultured in conditioned medium (CM) collected from nc- or siGDNF- TAMs. (E) Mice grouping information (n = 8 for each group). Ov: overexpressing and Clodro: Clodronate. (F) Representative images of luminescence emitted by LM-model mice 4 weeks after intrasplenic injection. (G) Histogram showing the luminescence intensity of model mice (n = 8, respectively). (H) Survival curves of four groups of indicated mice (n = 8, respectively). (I) Representative IHC images showing the F4/80 positive cells in metastatic lesions from model mice. Scale bar: 50 μm. “ns” not significant, *p < 0.05, **p < 0.01, ***p 0.001

Fig. 4

The GDNF-GFRA1 axis modulates the cellular autophagy-lysosome system. (A) Intracellular levels of mTOR, S6K, and the autophagy-related proteins BECN1, LC3, P62 and cleaved caspase 3 under starvation. (B) Representative images of transmission electron microscopy showing the levels of autophagosomes in scramble and shGFRA1 cells. (C) Representative images of AGS cells transfected with GFP-RFP-LC3 plasmids. Corresponding signals are labelled in the right lower corner. (D) Co-localization ratios of yellow vs. red (Y/R) puncta in the GFP-RFP-LC3 fluorescent models (n = 3, respectively). (E) Double staining for LC3 (green) and LysoTracker (red) in AGS cells in which GFRA1 is silenced. (F) Co-localization ratios of yellow vs. green (Y/G) puncta shown in a histogram (n = 3). (G) LysoSensor assay in AGS cells in which GFRA1 is silenced. Positive LysoSensor signal shown in blue and nuclear dye PI shown in red. (H) Histogram showing the mean fluorescent intensity of single cells using a LysoSensor assay (n = 3). (I) GSEA results of GC specimens from the TCGA database, grouped by GFRA1 expression levels. (J) Representative images of a FLUO-4 AM assay in scramble and shGFRA1 AGS cells, reflecting intracellular Ca²⁺ concentrations. (K) Histogram showing cytosolic Ca²⁺ levels calculated by FLUO-4 AM assay. Scale bar: respectively labelled in every module, “ns” not significant, *p < 0.05, ***p < 0.005

Fig. 5

GFRA1-dependent liver metastasis is inhibited by autophagy-targeted treatment. (A) Apoptotic curves of AGS cells (left panel) and HGC-27 cells (right) incubated with 20 ng/ml rGDNF or 100 nM autophagy inhibitor bafilomycin A₁. (B) Flow diagram of LM modelling and inhibitor treatment of mice. 1 mg/kg bafilomycin A₁ was intraperitoneally injected per day till six weeks. (C) Mice grouping information (n = 8 for each group). (D) Representative images of luminescence emitted by LM-model mice, 4 weeks after intrasplenic injection. (E) Histogram showing the luminescence intensity of model mice (n = 8, respectively). (F) Survival curves of these four groups of mice (n = 8, respectively). (G) Representative images of gross livers from LM-bearing mice. (H) Histogram showing the weights of livers and metastatic hepatic tissues, respectively (n = 8). Scale bar: respectively labelled in every module, “ns” not significant, **p < 0.01, ***p < 0.001

Fig. 6

Expression levels of GFRA1 correlate with a poor prognosis of GC patents. (A) Increasing GFRA1 levels along with TNM stages of GC patients in the TCGA database. (B-C) Survival curves based on the TCGA database (B) and the Kaplan–Meier Plotter website (C), analysed using the Kaplan–Meier method. The cut-off value was defined as median. HR(H/L): Hazard ratio of high vs. low expression. (D) Kaplan–Meier plots based on our liver-metastatic tumor microarray (n = 69), split by IHC scores. GFRA1 levels of primary lesions are shown in the left panel, and those in metastatic lesions in the right panel. (E) Schematic diagram of the anti-apoptotic mechanism in invasive margins through GFRA1-regulated autophagosome-lysosome fusion under metabolic stress