EGFR-mediated carcinoma cell metastasis mediated by integrin αvβ5 depends on activation of c-Src and cleavage of MUC1

Abstract

Receptor tyrosine kinases and integrins play an essential role in tumor cell invasion and metastasis. We previously showed that EGF and other growth factors induce human carcinoma cell invasion and metastasis mediated by integrin αvβ5 that is prevented by Src blockade. MUC1, a transmembrane glycoprotein, is expressed in most epithelial tumors as a heterodimer consisting of an extracellular and a transmembrane subunit. The MUC1 cytoplasmic domain of the transmembrane subunit (MUC1.CD) translocates to the nucleus where it promotes the transcription of a metastatic gene signature associated with epithelial to mesenchymal transition. Here, we demonstrate a requirement for MUC1 in carcinoma cell metastasis dependent on EGFR and Src without affecting primary tumor growth. EGF stimulates Src-dependent MUC1 cleavage and nuclear localization leading to the expression of genes linked to metastasis. Moreover, expression of MUC1.CD results in its nuclear localization and is sufficient for transcription of the metastatic gene signature and tumor cell metastasis. These results demonstrate that EGFR and Src activity contribute to carcinoma cell invasion and metastasis mediated by integrin αvβ5 in part by promoting proteolytic cleavage of MUC1 and highlight the ability of MUC1.CD to promote metastasis in a context-dependent manner. Our findings may have implications for the use and future design of targeted therapies in cancers known to express EGFR, Src, or MUC1.

Figure 1. MUC1 is required for EGF-dependent tumor cell metastasis.

(a) FG human pancreatic carcinoma cells expressing a control shRNA or MUC1 shRNA and stimulated with or without a 15 minute treatment of EGF were inoculated on to the chorioallantoic membrane of 10 day-old embryonated chicken eggs and assessed for spontaneous pulmonary metastasis (left) and primary tumor formation (right) after 10 days. Cells were washed with PBS prior to inoculation to remove EGF. Metastasis quantified by Q-PCR for human Alu sequence and chicken GAPDH was normalized to a standard curve. Each point represents a separate egg, n≥6 eggs per group. Immunoblot detecting MUC1 expression (far right). P<0.0001 comparing metastasis for cells expressing control shRNA with or without EGF treatment; P = 0.08 comparing metastasis for cells expressing MUC1 shRNA with or without EGF treatment; P<0.0001 comparing metastasis for cells expressing control shRNA or MUC1 shRNA with EGF treatment; P = 0.6 comparing primary tumor mass across groups. (b) Migration assays on a vitronectin (left) or a collagen (right) substrate comparing FG cells expressing a control shRNA or MUC1 shRNA with or without a 15 minute pre-treatment of EGF. Cells were washed with PBS prior to inoculation on Boyden chambers to remove EGF. P = 0.002 comparing migration on vitronectin for cells expressing control shRNA with or without EGF treatment; P = 0.8 comparing migration on vitronectin for cells expressing MUC1 shRNA with or without EGF treatment; P = 0.002 comparing migration on vitronectin for cells expressing control shRNA or MUC1 shRNA with EGF treatment; P = 0.5 comparing migration on collagen across groups. Results are expressed as mean ± s.e.m. of three replicates. Similar findings were observed in 3 independent experiments.

Figure 2. Nuclear localization of MUC1 is EGF-dependent.

(a) Representative images of immunofluorescence of MUC1.FL fused to GFP (MUC1.FL.GFP; green) with or without a 15 minute pre-treatment with EGF; nuclei are counter-stained with TO-PRO-3 (blue). Schematic (left) illustrates the MUC1.FL.GFP protein product including ectodomain (white), cytoplasmic domain (black) and GFP (hatched). Quantification of nuclear MUC1 (bar graph) is expressed as a percentage of total detectable MUC1. Scale bar represents 10 µm. * P<0.0001 compared to unstimulated cells. (b) Immunoblot detecting MUC1 cytoplasmic domain showing enrichment of MUC1 cytoplasmic domain with EGF treatment in the nuclear fraction from FG cells. Fraction purity and loading were determined by immunoblotting for PARP (Nuclear, Nuc) and GAPDH (Cytoplasmic, Cyto). Line graph shows quantification of MUC1 in each fraction by densitometry. (c) Quantitative RT-PCR of FG cells treated for 15 minutes with EGF compared to untreated control. Peak expression changes over a 24 h period are reported. Values have been normalized to β-actin. Results are expressed as mean ± s.e.m. Similar findings were observed in 3 independent experiments.

Figure 3. EGF enhances nuclear localization of MUC1 by regulating its cleavage.

(a) Immunoblot detecting MUC1 cytoplasmic domain showing increased levels of cleavage product with EGF treatment in whole-cell lysates from FG cells. (b) Representative images of immunofluorescence of MUC1.FL and MUC1.CD fused to GFP (MUC1.FL.GFP and MUC1.CD.GFP, respectively; green); nuclei are counter-stained with TO-PRO-3 (blue). Schematics (left) illustrate MUC1.FL.GFP and MUC1.CD.GFP protein products including ectodomain (white), cytoplasmic domain (black) and GFP (hatched). Quantification of nuclear MUC1 (bar graph) is expressed as a percentage of total detectable MUC1. Scale bar represents 10 µm. * P<0.0001 compared to cells expressing MUC1.FL.GFP. (c) Quantitative RT-PCR of FG cells treated for 15 minutes with EGF (white) or expressing MUC1.CD (black) compared to untreated or vector controls, respectively. For cells treated with EGF, peak expression changes over a 24 h period are reported. Values have been normalized to β-actin. Results are expressed as mean ± s.e.m. Similar findings were observed in 3 independent experiments.

Figure 4. MUC1 cytoplasmic domain is necessary and sufficient for migration and metastasis.

(a) Migration assays on a vitronectin (left) or a collagen (right) substrate comparing FG cells co-expressing MUC1 shRNA and shRNA-resistant full-length MUC1 (MUC1.FL) or cytoplasmic domain-deleted MUC1 (MUC1.CT3) with or without a 15 minute pre-treatment of EGF. Cells were washed with PBS prior to inoculation on Boyden chambers to remove EGF. Schematic (top) illustration of MUC1.FL and MUC1.CT3 protein products including ectodomain (white) and cytoplasmic domain (black). Immunoblot detecting MUC1 expression with Cell Signaling Technology anti-MUC1 clone VU4H5 (bottom). P<0.0001 comparing migration on vitronectin for cells expressing MUC1.FL with or without EGF treatment; P = 0.3 comparing migration on vitronectin for cells expressing MUC1.CT3 with or without EGF treatment; P = 0.3 comparing migration on collagen across groups. Results are expressed as mean ± s.e.m. of three replicates. Similar findings were observed in 3 independent experiments. (b) Migration assays on a vitronectin (left) or a collagen (right) substrate comparing FG cells expressing vector control or MUC1 cytoplasmic domain (MUC1.CD). Schematic (top) illustration of MUC1.FL and MUC1.CD protein products including ectodomain (white) and cytoplasmic domain (black). Immunoblot detecting MUC1 expression (far right). P = 0.0003 comparing migration on vitronectin; P = 0.2 comparing migration on collagen. Results are expressed as mean ± s.e.m. of three replicates. Similar findings were observed in 3 independent experiments. (c) Assessment of spontaneous pulmonary metastasis (left) and primary tumor formation (right) for FG cells expressing vector control, MUC1.CT3, or MUC1.CD in the chick CAM model. Metastasis quantified by Q-PCR for human Alu sequence and chicken GAPDH was normalized to a standard curve. Each point represents a separate egg, n≥6 eggs per group. * P<0.005.

Figure 5. Src activity is necessary and sufficient for MUC1 cleavage, and MUC1 is required for Src-dependent migration.

(a) Immunoblot detecting MUC1 cytoplasmic domain showing pre-treatment with a Src inhibitor (bosutinib, 500 nM) blocks EGF-dependent MUC1 cleavage in whole-cell lysates from FG cells. (b) Immunoblot detecting MUC1 cytoplasmic domain showing increased levels of cleavage product in FG cells expressing constitutively active Src compared to vector control. (c) Immunoblot detecting MUC1 cytoplasmic domain showing enrichment of MUC1 cytoplasmic domain in the nuclear fraction from FG cells expressing constitutively active Src compared to vector control. Fraction purity and loading were determined by immunoblotting for PARP (Nuclear, Nuc) and GAPDH (Cytoplasmic, Cyto). (d) Migration assays on a vitronectin (left) or a collagen (right) substrate in a Boyden chamber comparing FG cells co-expressing constitutively active Src and either a control siRNA or one of two different MUC1 siRNAs. Immunoblot detecting MUC1 expression (far right). P<0.0001 comparing migration on vitronectin for cells expressing vector control or active Src; P<0.0001 comparing migration on vitronectin for cells expressing control siRNA or either MUC1 siRNA; P = 0.2 comparing migration on collagen across groups. (e) Migration assays on a vitronectin (left) or a collagen (right) substrate in a Boyden chamber comparing FG cells expressing either vector control or MUC1.CD with or without a 15 minute pre-treatment of EGF in the presence or absence of a Src inhibitor (bosutinib, 500 nM). P = 0.7 comparing migration on vitronectin for cells expressing MUC1.CD with or without Src inhibitor treatment. Similar findings were observed in 3 independent experiments.