Peanut lectin stimulates proliferation of colon cancer cells by interaction with glycosylated CD44v6 isoforms and consequential activation of c-Met and MAPK: functional implications for disease-associated glycosylation changes

By: Ravinder Singh, Sreedhar Subramanian, Jonathan M. Rhodes, Barry J. Campbell
Published: 29 March 2006

Abstract

Peanut agglutinin lectin (PNA) binds the Thomsen–Friedenreich (TF) oncofetal carbohydrate antigen (galactoseβ1-3N-acetylgalactosamineα) that shows increased expression in colon cancer, adenomas, and inflammatory bowel disease. PNA is mitogenic, both in vitro and in vivo, for colon epithelial cells.

In these cells, PNA binds predominantly to cell-surface TF antigen expressed by high molecular weight isoforms of the transmembrane glycoprotein CD44 that are generated in inflamed and neoplastic colonic epithelia by altered RNA splicing.

Our aim was to identify the signaling mechanism underlying the proliferative response to PNA. This was investigated in HT29, T84, and Caco2 colon cancer cells. Parallel lectin and immunoblotting of PNA affinity-purified HT29 cell membrane extracts showed PNA binding to high molecular weight CD44v6 isoforms.

Within 5 min, PNA (25 µg/mL) caused a 6-fold increase in phosphorylation of hepatocyte growth factor receptor c-Met, known to co-associate with CD44v6. This was followed by the downstream activation of p44/p42 mitogen-activated protein kinase (MAPK) over 15–20 min.

The presence of 100 µg/mL asialofetuin, a TF antigen-expressing glycoprotein, blocked both PNA-induced c-Met and MAPK activation. A similar PNA-induced c-Met and MAPK phosphorylation was also seen in T84 cells that express CD44v6 but not in Caco2 cells that lack CD44v6. PNA-induced cell proliferation was completely blocked by 1 µM PD98059, an inhibitor of MAPK activation (p < 0.0001).

The expression of TF antigen by CD44 isoforms in colonic epithelial cells allows lectin-induced mitogenesis that is mediated by phosphorylation of c-Met and MAPK. It provides a mechanism by which dietary, microbial, or endogenous galactose-binding lectins could affect epithelial proliferation in the cancerous and precancerous colon.

Results

PNA binds CD44v6, which is associated with the receptor tyrosine kinase c-Met, in colon cancer cells

Immunoblot analysis of PNA-agarose affinity-purified cell-surface membrane proteins from HT29 cells using anti-CD44v6 antibodies revealed two PNA-reactive CD44v6 isoforms (Figure 1A).

CD44v6 immunoprecipitated from HT29 cell lysate was subjected to immunoblot analysis using either anti-CD44v6 or anti-c-Met antibodies (Figure 1B and C). This confirmed that CD44v6 and c-Met co-immunoprecipitate and therefore suggests that they may be physically associated.

PNA activates c-Met in HT29 colon cancer cells

Immunoblot analysis of HT29 cells treated with 25 µg/mL PNA, using a phospho-specific Met antibody that recognizes a peptide containing phospho-Tyr (1349) that provides a docking site on activated Met for downstream factors, revealed rapid phosphorylation of c-Met in response to PNA (Figure 2A).

Densitometric analysis revealed significant increases in c-Met phosphorylation of 2.8 ± 0.17 (mean ± SD) and 6.3 ± 0.68-fold at 2.5 and 5 min, respectively (p < 0.001 analysis of variance [ANOVA] versus control; n = 3), returning to near baseline levels after 30 min (Figure 2B). PNA (25 µg/mL; 5 min) failed to activate c-Met in the presence of 100 µg/mL asialofetuin (Figure 2C).

In response to treatment with 100 ng/mL HGF/SF, used as a positive control for Met receptor activation, significant phosphorylation of c-Met was seen, from 5 to 15 min, in both HT29 cells and Caco2 cells. This response started to fall at 30 min in Caco2 cells and to a lesser degree in HT29 cells (Figure 2D).

PNA activates MAPK, and inhibition of the MEK1/2-MAPK signaling pathway blocks PNA-induced cell proliferation

Immunoblotting of PNA-treated HT29, using specific anti-phospho p44/42 MAPK antibodies, demonstrated the activation of both p42 and p44 MAP (Figure 3A). After 5 min, addition of 25 µg/mL PNA to HT29 cells initiated significant p44 MAPK phosphorylation (2.2-fold) reaching a peak at 20 min (3.1 ± 0.07-fold; mean ± SD, p < 0.001 ANOVA; n = 3).

Phosphorylation of p44 MAPK remained 2.2- and 1.8-fold higher than that of control at 45 and 60 min, respectively. Phosphorylation of p42 MAPK showed a similar pattern, again initiated within 5 min (1.9-fold), reaching a peak at 15 min (2.5-fold) and remaining 1.4-fold higher than that of control even at 60 min, p < 0.01 ANOVA (Figure 3B). Again, the presence of 100 µg/mL asialofetuin blocked MAPK activation induced by PNA in HT29 cells (Figure 3C).

In addition, PNA (25 µg/mL) produced a 38 1.5% (mean SD, n = 3) increase in [methyl-3H]-thymidine incorporation into HT29 cells compared with untreated control (expressed as 100 2%; n = 3). PD98059, an inhibitor of MAPK activation, at a concentration of 1 µM, completely blocked PNA-induced proliferation in HT29 cells (p < 0.0001 ANOVA) (Figure 3D).

The PNA-induced c-Met and MAPK activation is CD44v6 isoform dependent

As seen for CD44v6-expressing HT29 cells, similar activation of c-Met and p42/p44 MAPK also occurred in CD44v6-positive T84 colon cancer cells following treatment with 25 µg/mL of PNA for 5 min.

It is worth noting that although lower levels of CD44v6 were detected in T84 cells (Figure 4A), stimulation of T84 cells with PNA resulted in a significant increase in phosphorylation of c‐Met (2.6 ± 0.16-fold increase; n = 3; p < 0.001 unpaired t‐test), similar to the PNA-induced response observed in HT29 cells (2.8 ± 0.21-fold increase; p < 0.001) when they each were compared with their respective untreated controls (Figure 4B).

PNA failed to significantly activate MAPK or c-Met in the CD44v6-negative Caco2 colon cancer cell line (Figure 4A and B).

PNA increases CD44 and p170 c-Met expression in HT29 colon cancer cells

Following 24-h treatment of HT29 colon cancer cells with 15–60 µg/mL PNA, immunoblot analysis demonstrated no significant increase in the expression of c-Met reactivity, although a modest increase was observed in the p170 precursor protein, and not active p145 c-Met, at 60 µg/mL PNA.

A similar modest increase in the expression of CD44 was also seen with 60 µg/mL PNA when compared with untreated controls (data not shown).

Discussion

These studies show that PNA binds to TF oncofetal carbohydrate antigen (galactoseβ1-3N-acetylgalactosamineα) that resides on v6 isoforms of the high molecular weight glycoprotein CD44.

Furthermore, the interaction between the PNA and the TF-expressing CD44v6 splice variants on the cell surface of HT29 and T84 colon cancer cells results in the phosphorylation of associated c-Met receptor and subsequent activation of the p44/p42 MAPK (ERK 1/2) cell signaling pathway.

Continuous exposure to PNA results in a reduction in the amount of phosphorylated c-Met following the initial increase, a phenomenon likely to be because of endocytosis of the c-Met receptor rather than receptor saturation (Hammond et al., 2003).

Caco2 cells, that do not express CD44v6, do not show c-Met or MAPK activation with PNA and are known to show no significant proliferative response to the lectin (Ryder, Smith et al., 1994). The activation of both c-Met and MAPK is PNA–TF antigen interaction dependent as shown by abrogation of response to PNA in the presence of the TF antigen-expressing glycoprotein, asialofetuin.

Binding of PNA to asialofetuin is TF antigen dependent and almost completely abolished by O‐glycanase treatment of asialofetuin (Singh et al., 2001). Thus, the TF antigen expression on CD44v6 has functional significance for the lectin-induced proliferation of colon cancer cells.

The kinetics of MAPK activation and the inhibition of PNA-induced proliferation by PD98059, a specific inhibitor of MAPK kinase (MEK-1), which phosphorylates and activates p44/42 MAPKs, strongly suggest that this is the mechanism for the proliferative effect of PNA on colonocytes.

It is known that CD44v6 is required for c-Met activation by HGF/SF and subsequent MAPK activation in several cell lines, including HT29 cells (Orian-Rousseau et al., 2002).

The demonstration that exposure to PNA results in the activation of c-Met not only confirms the close functional relationship between c-Met and CD44v6 but also suggests an important function for the TF glycan that is selectively expressed by high molecular weight CD44 splice variants (Singh et al., 2001).

However, in Caco2 cells, known to express a functioning HGF receptor (Kermorgant, Aparicio et al., 2001; Kermorgant, Dessirier et al., 2001), we demonstrated that HGF/SF activation of c-Met can occur despite the lack of CD44v6.

It is known that HGF/SF is able to induce Met signaling in HepG2 cells that do not express CD44 and also induces Met signaling in fibroblasts derived from CD44 null mice. In these HGF/SF-responsive but CD44-negative cells, it is thought that substitute molecules may compensate for the lack of CD44 and allow Met activation (Orian-Rousseau et al., 2002).

CD44 isoforms are generated by extensive alternative splicing, and additional variability is introduced by post-translational modification (Ponta et al., 2003; Thorne et al., 2004). The expression of isoforms bearing sequences encoded by exons v4–v7 or v6 and v7 have been shown to be sufficient to confer metastatic potential to non-metastatic cells (Gunthert et al., 1991).

Antibodies directed against CD44v6, or CD44v6 antisense, inhibit tumor growth and metastasis of colon cancer cells in vivo and reduce invasiveness of fibrosarcoma cells in vitro (Seiter et al., 1993; Ponta et al., 1998; Reeder et al., 1998).

The mechanism for variant splicing of CD44 in cancerous and inflamed epithelia is not well understood. It may be affected either by intron length (Bell et al., 1998) or by the effect of pro-inflammatory cytokines (Macdonald et al., 2003). The fact that TF expression is specific for some of the high molecular weight CD44 splice variants suggests that these may contain amino acid sequences that are particularly susceptible to O-glycosylation with this disaccharide (Singh et al., 2001).

This is also an important observation, because little had been known about which macromolecules, besides mucins (Campbell et al., 1995), might carry the TF oligosaccharide.

It is known that the stimulation of c-Met via its natural ligand, HGF/SF, results in wide-ranging biological and biochemical effects in the cell that can include scattering, proliferation, enhanced cell motility, invasion, and eventually metastasis (Ma et al., 2003).

The activation of c-Met results in the recruitment of scaffolding proteins such as HGF receptor-bound protein 2 (Grb2) and Grb2-associated binder 1 (Gab1), which activate Shp2 and the Ras-Raf-ERK signaling pathway.

This causes changes in gene expression of cell-cycle regulators, (such as retinoblastoma protein, Cdk6 and p27), extracellular matrix proteinases (such as matrix metalloproteinases and urokinase plasminogen activator), and in alterations of cytoskeletal functions that control migration, invasion, and proliferation (Birchmeier et al., 2003).

In addition, the activation of c-Met in colorectal carcinoma cells leads to constitutive association of tyrosine-phosphorylated β-catenin (Herynk et al., 2003). Overexpression of c-Met protein correlates with poor prognosis in gastrointestinal, hepatocellular, breast, endometrial, and nasopharyngeal carcinomas (Danilkovitch-Miagkova and Zbar, 2002; Ma et al., 2003; Takeuchi et al., 2003).

Recent studies have provided good evidence for functional collaboration between CD44 isoforms and c–Met. Association between CD44v6 and c-Met receptor is essential for the activation of c-Met tyrosine kinase activity by its natural ligand, HGF/SF, and subsequent activation of MAPK signaling (Orian-Rousseau et al., 2002).

CD44 isoforms decorated with heparin sulfate chains can bind the c-Met ligand, HGF/SF, and this interaction promotes signaling through c‐Met (van der Voort et al., 1999). CD44v3 isoforms, which contain a site for heparin sulfate attachment, and c-Met are co-expressed on colorectal tumors and cell lines (Wielenga, van der Voort, Taher et al., 2000).

These studies further support a possible therapeutic role for MEK1/2-MAPK inhibition by tyrosine kinase inhibitors such as imatinib mesylate and gefitinib in cancer (Dancey, 2003; Levitzki, 2003). c-Met is also an important target for cancer therapy, and many efforts are directed toward the identification of inhibitors that are active in vivo (Birchmeier et al., 2003).

This study raises the possibility that other galactose-binding lectins may have similar interactions with cancerous or precancerous colonic epithelial cells via TF-expressing CD44v6 and its association with c-Met.

Such lectins might include members of the galectin family of endogenous galactose-binding lectins whose expression is markedly altered in cancer (Itzkowitz, 1997; Danguy et al., 2002) or lectins expressed by bacteria in the colonic lumen (Rhodes, 1996; Rhodes and Campbell, 2002).