GFRα1 released by nerves enhances cancer cell perineural invasion through GDNF-RET signaling

Abstract

The ability of cancer cells to invade along nerves is associated with aggressive disease and diminished patient survival rates. Perineural invasion (PNI) may be mediated by nerve secretion of glial cell line-derived neurotrophic factor (GDNF) attracting cancer cell migration through activation of cell surface Ret proto-oncogene (RET) receptors. GDNF family receptor (GFR)α1 acts as coreceptor with RET, with both required for response to GDNF. We demonstrate that GFRα1 released by nerves enhances PNI, even in the absence of cancer cell GFRα1 expression. Cancer cell migration toward GDNF, RET phosphorylation, and MAPK pathway activity are increased with exposure to soluble GFRα1 in a dose-dependent fashion. Dorsal root ganglia (DRG) release soluble GFRα1, which potentiates RET activation and cancer cell migration. In vitro DRG coculture assays of PNI show diminished PNI with DRG from GFRα1(+/-) mice compared with GFRα1(+/+) mice. An in vivo murine model of PNI demonstrates that cancer cells lacking GFRα1 maintain an ability to invade nerves and impair nerve function, whereas those lacking RET lose this ability. A tissue microarray of human pancreatic ductal adenocarcinomas demonstrates wide variance of cancer cell GFRα1 expression, suggesting an alternate source of GFRα1 in PNI. These findings collectively demonstrate that GFRα1 released by nerves enhances PNI through GDNF-RET signaling and that GFRα1 expression by cancer cells enhances but is not required for PNI. These results advance a mechanistic understanding of PNI and implicate the nerve itself as a key facilitator of this adverse cancer cell behavior.

Fig. 1.

Generated cell lines were characterized. (A and B) Expression of RET and GFRα1 mRNA by generated MiaPaCa-2 and 3T3 cell lines. (C) Immunofluorescence microscopy for RET (red), GFRα1 (green), and DAPI (blue) in generated MiaPaCa-2 and 3T3 cell lines. (D) Boyden chamber migration assays using glial cell line-derived neurotrophic factor (GDNF) as an attractant over 24 h for MiaPaCa-2 derived cell lines. (E) MiaPaCa-2 siRNA control, siRNA RET, and siRNA GFRa1 cell proliferation over 24 h. (F) Boyden chamber migration assays using GDNF, GDNF plus soluble GFRα1-Fc fusion protein, or excised murine DRG as an attractant over 24 h for 3T3-derived cell lines. (G) 3T3-RET-GFRα1, 3T3-RET, and 3T3 cell proliferation are similar over 24 h.

Fig. 2.

Soluble GFRα1 promotes cell migration toward GDNF and RET activation. (A–C) MiaPaCa-2, MiaPaCa-2 shGFRα1, and 3T3-RET cell line Boyden chamber migration assays over 24 h using varying concentrations of GDNF and soluble GFRα1 as attractants. (D–G) Western blots of proteins isolated from MiaPaCa-2, 3T3-RET, MiaPaCa-2–shGFRα1, and MiaPaCa-2–shRET cells undergoing Boyden chamber migration assays at conditions identical to A–C.

Fig. 3.

DRG release soluble GFRα1 that promotes cell migration and RET activation. (A) Conditioned media from excised murine DRG following 0, 4, 6, or 8 d of exposure were concentrated and underwent Western blotting to detect released GFRα1. Total protein measured by Ponceau staining was used as a loading control. (B) 3T3-RET cells were cocultured with DRG from 0 to 12 h, and protein was isolated for Western blotting. (C and D) 3T3-RET cells underwent Boyden chamber migration assays over 24 h using conditioned media from DRG as an attractant with the addition of varying concentrations of an anti-GFRα1 antibody; Western blots were performed at parallel conditions. (E and F) MiaPaCa-2 cells underwent Boyden chamber migration assays over 24 h using conditioned media from DRG as an attractant with the addition of varying concentrations of an anti-GFRα1 antibody; Western blots were performed at parallel conditions.

Fig. 4.

GFRα1 potentiates PNI in a nerve–cancer coculture in vitro assay. (A) Coculture of murine DRG with MiaPaCa-2 cells in Matrigel permits assessment of the degree of PNI. By day 6, MiaPaCa-2 cells exposed to IgG as a control condition exhibit robust invasion, extending along neurites from the DRG. (B) DRG–MiaPaCa-2 coculture assay of PNI, when exposed to anti-GFRα1 antibody, demonstrates diminished PNI at day 6 compared with control (A). (C) The degree of PNI may be quantified in the DRG–MiaPaCa-2 coculture assay. Areas where MiaPaCa-2 cells are in direct contact with DRG neurites were demarcated, and the area was calculated using MetaMorph software. (D) The mean total area of invasion is compared between control IgG and anti-GFRα1 antibody exposed DRG–MiaPaCa-2 coculture assays (^†P < 0.05; t test).

Fig. 5.

DRG from GFRα1^+/− mice attract less cancer cell migration and less PNI compared with DRG from wild-type GFRα1^+/+ mice. (A) Protein isolated from lysed DRG from GFRα1^+/− and GFRα1^+/+ mice underwent Western blotting for GFRα1. (B) The migration of MiaPaCa-2 shGFRα1 in Boyden chamber assays was quantified using GFRα1^+/− DRG or GFRα1^+/+ DRG as an attractant (^†P < 0.001; t test). (C) DRG coculture assays were performed using MiaPaCa-2 shGFRα1 cells. Greater PNI was noted with GFRα1^+/+ DRG compared with GFRα1^+/− DRG. (D) The average area of invasion by MiaPaCa-2 shGFRα1 cells in DRG assays using GFRα1^+/− DRG compared with GFRα1^+/+ DRG (^††P < 0.05; t test).

Fig. 6.

A murine model of sciatic nerve PNI demonstrates that cancer cells lacking GFRα1 still maintain the ability to invade nerves. (A) Sciatic nerve tumors were grown in mice after implantation of Mia shControl, Mia shGFRα1, and Mia shRET cells (n = 4 per group). Tumors grew at varying rates. Tumor volumes at the time of sacrifice are demonstrated; comparisons were made between groups at these different time points to standardize tumor volume. (B) Sciatic nerve-invasion length was measured at the time of animal sacrifice (week 6 for Mia shControl, week 7 for Mia shGFRα1, week 9 for Mia shRET). (C) The sciatic nerve index (hind paw span) is a measure of sciatic nerve function and was measured at week 1 and at weeks 6, 7, and 9 (^†P < 0.01; t test). (D) Nerve function scores (a measure of hind limb function) were measured weekly for each group.

Fig. 7.

Gross, MRI, and surgical images of sciatic nerve PNI by cancer cells lacking GFRα1 maintaining an ability to invade nerves. (A–D) Representative mice bearing no tumor, Mia shControl tumor, Mia shGFRα1 tumor, or Mia shRET tumor, respectively, in the right sciatic nerve at the indicated time points. The Mia shControl and Mia shGFRα1 mice demonstrate right hind limb paralysis, whereas the Mia shRET and normal mice have intact hind limb function. (E–H) Representative T2-weighted MRI images of mice. The Mia shControl and Mia shGFRα1 tumors demonstrate thickened sciatic nerves (arrows), whereas the Mia shRET tumor and non–tumor-bearing animal show normal caliber sciatic nerves (arrows). (I–L) Representative sciatic nerves at sacrifice with tumor exposure. The Mia shControl and Mia shGFRα1 tumors (asterisks) both demonstrate thickened and invaded sciatic nerves, with the Mia shGFRα1 nerves less severely infiltrated. In contrast, Mia shRET tumor (asterisk) and non–tumor-bearing animal show normal caliber sciatic nerves.

Fig. 8.

Histologic images of proximal sciatic nerve PNI by cancer cells lacking GFRα1 maintain an ability to invade nerves. (A) H&E-stained normal proximal sciatic nerve 6 wk after injection with PBS shows normal proximal sciatic nerve histology and caliber. (Scale bar, 500 µm.) (B) H&E-stained proximal sciatic nerve 6 wk after injection with Mia shControl cells demonstrates extensive PNI with expansion of the nerve by infiltrating tumor cells. (Scale bar, 500 µm.) (C) H&E-stained proximal sciatic nerve 7 wk after injection with Mia shGFRα1 cells demonstrates proximal sciatic PNI, although to a lesser severity compared with the shControl cells. (Scale bar, 500 µm.) (D) H&E-stained proximal sciatic nerve 9 wk after injection with Mia shRET cells demonstrates no significant PNI. (Scale bar, 500 µm.) (E) Immunofluorescence microscopy was performed on excised Mia shControl, shRET, and shGFRα1 primary sciatic nerve tumors stained for RET and GFRα1 expression (upper images) and p-RET expression (lower images) with DAPI nuclear staining. (Scale bar, 100 µm.)

Fig. 9.

A tissue microarray of 141 surgically excised human pancreatic ductal adenocarcinomas, in which PNI is nearly ubiquitous, was assessed by immunohistochemistry for cancer cell GFRα1 expression. (A–C) Representative sections of GFRα1 0 (A), 1+ (B), and 2+ (C) specimens are shown. (D) Wide variance in expression was noted, with 52 cases staining 0 (negative), 67 cases stained 1+ (mild to moderately positive), and 22 cases stained 2+ (strongly positive).