Recently, we reported that the laminin α5 synthetic peptide A5G27 (RLVSYNGIIFFLK, residues 2,892-2,904) binds to the CD44 receptor of B16-F10 melanoma cells via the glycosaminoglycans on CD44 and inhibits tumor cell migration, invasion, and angiogenesis in a dominant-negative manner. Here, we have identified the potential mechanism of A5G27 activity using WiDr human colorectal carcinoma cells. WiDr cells bound to the laminin A5G27 peptide via the heparin-like and chondroitin sulfate B glycosaminoglycan side chains of CD44. Cell binding to fibroblast growth factor (FGF2) was blocked by laminin peptide A5G27 but not by either a scrambled version of this peptide or by another laminin peptide known to bind cell surface proteoglycans. FGF2 signaling involving tyrosine phosphorylation was also blocked by laminin peptide A5G27 but was not affected by peptide controls. Finally, we have shown that peptide A5G27 directly blocks FGF2 binding to heparin. Peptide A5G27 has sequence homology to a region on FGF2 that binds heparin and the FGF receptor and is important in FGF2 central cavity formation. We conclude that peptide A5G27 inhibits metastasis and angiogenesis by blocking FGF2 binding to the heparan sulfate side chains of CD44 variant 3, thus decreasing FGF2 bioactivity.
Laminins are extracellular matrix glycoproteins that are present in all basement membranes. They are triple-helical molecules composed of one α, one β, and one γ chain. There are five α, three β, and three γ chains identified to date, which assemble to form a family that has at least 15 laminin isoforms. Laminins are biologically active for cell adhesion, migration, angiogenesis, differentiation, tumor growth, and metastasis ( 1). Laminin-1 sites for these activities have been identified by using synthetic peptides ( 2– 6). Some of these peptides regulate the malignant phenotype ( 6– 9). Most of the active peptides are localized in the globular domains and play a critical role in binding to cell surface receptors in a peptide-specific and cell type–specific manner ( 10– 14). Several receptors for these active sequences have been identified. Two peptides in the globular domains interact with syndecans ( 5, 15). One is a laminin α1 chain peptide, AG73 (RKRLQVQLSIRT, residues 2,719-2,730), that promotes cell adhesion and salivary gland cell differentiation by binding to the heparan sulfate glycosaminoglycans on syndecan-1. A laminin α3 chain peptide, A3G75aR (NSFMALYLSKGR, residues 1,412-1,423), promotes cell adhesion via binding to syndecan-2 and syndecan-4.
The laminin α chains are generally large (Mr = 400,000) and contain a COOH-terminal globular domain consisting of five modules, LG1 to LG5. The globular modules on the α chains are of particular interest because of their biological activity. The α5 chain is a component of laminin-10 (α5β1γ1) and laminin-11 (α5β2γ1), which are important in malignancy ( 16). Previously, 113 overlapping synthetic peptides from the α5 chain G-domain were screened for cell attachment activity with HT1080 and PC12 cells and binding sites were identified ( 17). We tested these 113 peptides in various in vivo and in vitro assays with B16-F10 mouse melanoma cells and identified four peptides (A5G27, 73, 81, and 101) that inhibit melanoma cell growth and metastasis ( 18). These four peptides also inhibited melanoma cell angiogenesis, invasion, and migration. One of the four peptides, A5G27 (RLVSYNGIIFFLK, residues 2,892-2,904), bound to the glycosaminoglycan side chains of CD44, a receptor important in tumor metastasis.
Here, we have now determined if other tumor cells recognize this peptide and begun to define the signaling pathway and mechanism involved. We used the cell line WiDr, which is a human colon cancer cell line because it contains defined CD44 variants ( 19). CD44 variants are relatively well studied in colorectal cancer and the level of expression has been used as a marker for disease progression ( 20). First, we confirmed the presence of CD44 and CD44 variants on the surface of WiDr cells with FACScan analysis and immunoprecipitation. In attachment assays, WiDr cells bound to the laminin A5G27 peptide via the heparin-like and chondroitin sulfate B side chains of CD44. CD44 splice variants have been identified in many cancer cell lines, and, in particular, CD44v3 and CD44v6 play an important role in tumor growth and metastasis ( 20– 22). CD44v3 contains Ser-Gly repeats that support covalent attachment of heparan sulfate glycosaminoglycans, and CD44v3-heparan sulfate binds a number of heparin-binding growth factors, including members of the fibroblast growth factor (FGF) family ( 23, 24). FGF2 binds to CD44, augmenting the ability to stimulate FGF2 activities and playing an important role in tumor growth and angiogenesis. FGF2 specifically increases tumor and endothelial cell proliferation, migration, and survival, resulting in increased tumor growth and metastasis ( 25– 29). FGF2 signaling is complex and involves FGF receptors that become tyrosine phospohorylated when activated by the ligand with initiation of protein kinase C–dependent signaling and activation of mitogen-activated protein kinase (MAPK) pathways ( 30, 31). Levels of FGF2 in serum and in urine correlate with malignancy and FGF-transfected tumor cells grow faster in vivo ( 28, 29, 32– 36). Because the glycosaminoglycan chains of CD44 are important for FGF binding, we analyzed WiDr attachment to FGF2. We found that laminin peptide A5G27 inhibits WiDr cell attachment to FGF2 as well as FGF2-induced migration and invasion. This peptide does not block attachment to vascular endothelial growth factor (VEGF121). Furthermore, A5G27-treated cells showed a reduction of FGF2-induced tyrosine phosphorylation. Using heparin affinity chromatography, we found that peptide A5G27 inhibited FGF2 binding to heparin directly. Finally, there is considerable sequence homology (69%) between A5G27 and a sequence in FGF2 that binds heparin and the FGF receptor and is important in FGF2 central cavity formation ( 37). We conclude that the laminin α5 chain peptide A5G27 inhibits the binding between CD44v3-heparan sulfate and FGF2, thus decreasing FGF2-induced activity.
Materials and Methods
Preparation of peptides. All peptides were synthesized by a 9-fluorenylmethoxycarbonyl–based solid-phase method and prepared with a COOH-terminal amide as previously described ( 17, 33). In this study, we used peptides A5G27 (RLVSYNGIIFFLK), the scrambled A5G27 peptide A5G27S (VLFGFLKIYSRIN; ref. 18), C16 (KAFDITYVRLKF; ref. 9), and AG73 (RKRLQVQLSIRT; refs. 6, 14).
Culture of WiDr colorectal carcinoma cells. WiDr colorectal carcinoma cells were obtained from American Type Culture Collection (Rockville, MD) and were cultured in RPMI 1640 (Life Technologies, Grand Island, NY), containing 10% fetal bovine serum (HyClone, Logan, UT), 100 units/mL penicillin, 100 μg/mL streptomycin, and nonessential amino acids solution (Life Technologies, Rockville, MD), at 37°C, 5% CO2.
Flow cytometric analysis. WiDr cells were suspended in RPMI 1640 containing 0.1% bovine serum albumin (BSA) and then were pretreated with heparitinase and chondroitinase ABC (Seikagaku, Falmouth, MA) to remove cell surface glycosaminoglycans ( 6). After digestion, 3 × 105 cells were incubated with either CD44 (Cymbus Biotechnology Ltd., Chandlers Ford, United Kingdom), CD44v3, or CD44v6 (Chemicon, Temecula, CA) antibodies (10 μg/mL) in 100 μL of PBS with 0.1% BSA for 1 hour at 4°C. Cells were washed, resuspended in 100 μL of PBS with 0.1% BSA, and incubated with Cy2-labeled secondary antibodies (10 μg/mL; Jackson Immunoresearch Laboratory, West Grove, PA) for 1 hour at 4°C. At the end of this incubation period, the cells were washed twice and the expression of CD44, CD44v3, and CD44v6 on the cells was analyzed using a FACScan (Becton Dickinson, Mountain View, CA). This experiment was repeated twice.
Immunoprecipitation. Immunoprecipitation of the biotinylated cell surface membrane proteins isolated from the peptide affinity chromatography columns, prepared and run as previously described ( 18), was done using CD44, CD44v3, and CD44v6 antibodies as specified by the manufacturers. WiDr cell pellets were collected and incubated for 1 hour on ice with 1 mL 0.5% NP40 in PBS (NP-PBS) with complete protease inhibitor mixture (Boehringer Mannheim, Indianapolis, IN). The supernatants were collected after centrifugation for 15 minutes at 11,000 × g at 4°C. Protein G beads were incubated twice, 2 hours each, with NP-PBS and with the unlabeled WiDr cell protein extract to prevent nonspecific binding. Beads were then incubated with either control rat IgG, CD44, CD44v3, or CD44v6 antibodies for 2 hours in NP-PBS containing 0.1% BSA. Then, the precipitated 2.0 mol/L NaCl eluate from the peptide affinity column that was digested with chondroitinase ABC and heparitinase ( 18) was incubated with the beads in the presence of 0.1% BSA and 0.5% NP40 for 12 hours. The beads were washed five times with radioimmunoprecipitation assay buffer [0.15 mol/L NaCl, 1% NP40, 0.5% deoxycholate, 0.1% SDS, 50 mmol/L Tris (pH 8.0)] and once in Tris buffer. Immunoprecipitated proteins were removed from the beads by boiling in sample buffer (Invitrogen, Carlsbad, CA) and then separated by SDS-PAGE. The filter was blocked with 5% skim milk, washed thrice for 10 minutes each in TBS-T (0.1% Tween 20), and then incubated with a species-appropriate horseradish peroxidase–conjugated secondary antibody (Santa Cruz Biotechnology, Santa Cruz, CA) for 1 hour at room temperature. The filters were visualized by enhanced chemiluminescence. These experiments were done at least thrice.
Attachment assay using peptide-coated plates. Tumor cell attachment was assayed in U-bottomed 96-well plates coated overnight with synthetic peptides (0-5 μg). Wells were washed with PBS, blocked with 1% BSA in PBS, and washed again with PBS. Cells, detached by 0.02% EDTA in PBS and resuspended in RPMI 1640 containing 0.1% BSA, were added (5 × 104/100 μL) to each well and incubated for 1 hour at 37°C, 5% CO2. The attached cells were stained with 50 μL of 0.2% crystal violet aqueous solution in 20% methanol for 10 minutes. Dishes were extensively washed and bound dye was solubilized in 2% SDS and quantitated at 600 nm. Assays were done in triplicate at least thrice.
Inhibition of cell attachment by glycosaminoglycans and by A5G27. Plates were prepared as described above. The wells were then preincubated with 5 μg of the glycosaminoglycans in 25 μL of RPMI 1640 containing 0.1% BSA for 30 minutes at 37°C. Heparin, heparan sulfate (purity >99%), chondroitin sulfates A (purity >70%), B (purity >99%), and C (purity >90%), and hyaluronic acid (Sigma, St. Louis, MO) were added to the wells. Tumor cells (5 × 104) were added in 25 μL of RPMI 1640 containing 0.1% BSA, resulting in a final volume of 50 μL. Attachment was measured as described above. In some assays, plates were coated with either 15 ng of recombinant human FGF2 or VEGF121 (R&D Systems, Minneapolis, MN) in 50 μL of H2O and were prepared as described above. Cells (5 × 104 in 100 μL of 0.1% BSA in PBS with peptide; 0-100 μg/mL) were added to each well and incubated for 1 hour at 37°C, 5% CO2. The attached cells were stained and quantitated as described above. Assays were done in triplicate at least thrice.
Cell attachment after removal of cell surface glycosaminoglycans. U-bottomed, 96-well plates were coated with either 500 ng/well of peptide or 15 ng/well FGF2. Two different batches of FGF2 and peptides A13 (negative control that binds integrins; ref. 3) and AG73 (positive control that is known to bind via glycosaminoglycan chains; ref. 6) were used. The wells were blocked and washed as described above. Cells were released from the dish with trypsin and 5 × 104 cells/100 μL of RPMI-0.1% BSA were treated with either 0.05 units/mL heparitinase or 0.02 units/mL each of heparinase and chondroitinase ABC on a rotator at 37°C for 30 minutes. After enzyme treatment, the cells were added to the wells and incubated at 37°C for 40 minutes. Attachment was assessed as described above.
Migration and invasion assays. The migration assay was done using a 48-well microchemotaxis chamber (Neuro Probe, Inc., Cabin John, MD; ref. 38). Polyvinylpyrrolidone-free polycarbonate membranes with 10 μm pore size (Osmonics, Inc., Livermore, CA) were coated with 50 μg/mL of rat tail type I collagen (BD Biosciences, Bedford, MA) in RPMI 1640 that contained 25 mmol/L HEPES for 2 hours at 37°C and dried at least 2 hours before use. The lower wells of the chamber were loaded with 100 μg/mL of peptide and/or 100 ng/mL of FGF2 in RPMI 1640 with 1% BSA and 25 mmol/L HEPES. The coated membrane was then placed over the lower wells. WiDr cells (5 × 104) suspended in 50 μL binding medium (RPMI 1640 with 1% BSA and 25 mmol/L HEPES) were placed in the upper wells. The chamber was incubated for 8 hours at 37°C in 5% CO2. The membranes were fixed and stained using Diff-Quik (VWR, Bridgeport, NJ). The number of the cells on the undersurface of the membrane was counted under a microscope at ×32 magnification. The invasion assay was done in a similar way, except that the upper side of the membranes was coated with Matrigel (BD Biosciences), diluted 1:20 with water, for 2 hours at room temperature. Each experiment was done in triplicate thrice.
Heparin binding. Ten microliters of heparin-acrylic beads (Sigma) were incubated with or without peptides in 40 μL PBS for 1 hour at 37°C. Then, 25 ng of FGF2 was added and the beads were incubated for 2 hours at 37°C. The beads were washed five times with PBS and bound FGF2 was removed from the beads by boiling in sample buffer and then separated by SDS-PAGE. Western blotting was done with anti-FGF2 antibody (R&D Systems). The filters were visualized by enhanced chemiluminescence and quantitated by NIH image. This experiment was done at least thrice.
Determination of protein phosphorylation. Six-well dishes containing confluent WiDr cells were incubated in serum-free RPMI 1640 for 6 hours. The cells were incubated in the presence of 100 μg/mL of A5G27, A5G27S, 50 ng/mL of FGF2, and FGF2 together with either 100 μg/mL of A5G27 or A5G27S for 30 minutes or 2 hours. Media were removed and cells were solubilized in 150 μL of 2% SDS. Proteins were separated on SDS-PAGE before Western blotting with either anti-p44/42, phospho-p44/42, anti-phospho-p38 MAPK (Cell Signaling Technology, Beverly, MA), or antiphosphotyrosine (Upstate, Waltham, MA) antibodies. The results were quantitated by NIH image. This experiment was repeated thrice.
Fluorescence-activated cell sorting analysis with CD44 variant-specific antibodies shows WiDr cell surface expression of human CD44, CD44v3, and CD44v6. We have already reported that the laminin α5 G-domain peptide A5G27 bound CD44, a cell surface proteoglycan known to be important in malignancy ( 18). Human CD44 is alternatively spliced, and variants 3 and 6 are present on many cancer cell lines and are increased in metastatic cell lines and in tumor biopsy specimens ( 22). Here, we have used the cell line WiDr, a human colon cancer cell line, because it contains CD44 variants ( 19). FACScan analysis was done to determine if CD44v3 and CD44v6 were present on these cells. Cells were pretreated with both heparitinase and chondroitinase ABC to digest cell surface glycosaminoglycans to expose the core protein epitopes. WiDr cells stained with CD44, CD44v3, and CD44v6 antibodies ( Fig. 1A ).
CD44 and CD44 variants bind to an A5G27 peptide column. We further confirmed the identity of WiDr cell surface ligands for A5G27. The A5G27-bound biotinylated membrane proteins were prepared from a peptide affinity column as previously described ( 18) and digested with heparitinase and chondroitinase ABC. Then, the CD44 variants were immunoprecipitated with antibodies against CD44 ( Fig. 1B, lane 2), CD44v3 ( Fig. 1B, lane 3), CD44v6 ( Fig. 1B, lane 4), and control rat IgG ( Fig. 1B, lane 1). CD44, CD44v3, and CD44v6 antibodies immunoprecipitated ∼50 to 55 kDa bands ( Fig. 1B, arrow). We used a CD44 antibody that was raised against whole CD44 containing glycosaminoglycan chains. Only this CD44 antibody binds both the core protein and the glycosaminoglycan chains. Lane 2 showed a ∼28 kDa band that was likely the digested glycosaminoglycan chains. The CD44v3 and CD44v6 antibodies were raised against either the human v3 or v6 recombinant core protein. No material was observed in the control (scrambled peptide) column eluates (data not shown) as expected based on data with B16F10 cells ( 18). We conclude that the WiDr surface CD44, CD44v3, and CD44v6 bind to laminin peptide A5G27.
Soluble heparin, heparan sulfate, and chondroitin sulfate B block WiDr attachment to peptide A5G27. Because CD44, CD44v3, and CD44v6 are present on the surface of WiDr cells, we analyzed the role of glycosaminoglycans in cell attachment to peptide A5G27. Using peptide-coated dishes, WiDr cells attached to A5G27 in a dose-dependent manner (data not shown). The scrambled peptide (A5G27S) did not support cell attachment. WiDr cells bound to A5G27 via the glycosaminoglycan side chains on CD44. Using soluble glycosaminoglycans in the attachment assay, heparin, heparan sulfate, and chondroitin sulfate B strongly inhibited cell attachment to A5G27 by 80% or more (data not shown). These data show that WiDr cells bind to the laminin A5G27 peptide via the chondroitin sulfate B and heparin-like glycosaminoglycan side chains of CD44 similar to the B16F10 cells previously studied with this peptide ( 18).
A5G27 inhibits binding of WiDr to FGF2 but not to vascular endothelial growth factor121. CD44v3 contains Ser-Gly repeats that support covalent attachment of heparan sulfate proteoglycans ( 22). CD44v3 binds a number of heparin-binding growth factors, including members of the FGF family ( 23, 24). Because A5G27 bound CD44v3 and A5G27-WiDr attachment was inhibited by heparan sulfate, we next investigated whether A5G27 could inhibit the binding between FGF2 and CD44v3-heparan sulfate using an attachment assay. U-bottomed, 96-well plates were coated overnight with 15 ng of FGF2. WiDr cells (5 × 104/100 μL), in the presence of peptide, were added and incubated for 1 hour at 37°C, 5% CO2. A5G27 (10 μg/mL) significantly inhibited cell attachment to FGF2 ( Fig. 2A ) in a dose-dependent manner ( Fig. 2B). The control scrambled peptide A5G27S was inactive. In addition, peptide C16, which binds integrins ( 9), and peptide AG73, which binds heparin ( 6), were inactive. Furthermore, as a control, we tested the effects of these peptides on adhesion to VEGF121. We found that none of these peptides significantly blocked adhesion to VEGF121 ( Fig. 2A). The data suggest that A5G27 specifically inhibited the binding between FGF2 and the heparan sulfate glycosaminoglycan chains on CD44.
Removal of cell surface glycosaminoglycans reduces attachment to FGF2. We next tested whether removal of the cell surface glycosaminoglycans would affect cell attachment to FGF2. Cells were pretreated with either heparitinase or a mixture of heparinase and chondroitinase ABC and then their attachment was assessed. Attachment to FGF2 was blocked by 36% with the combination of glycosidases. Attachment to peptide controls gave the expected results with attachment to AG73, a peptide known to have a proteoglycan receptor ( 6), blocked by 46%, and attachment to peptide C16, which is known to bind integrins, not blocked ( 9). These data confirm that attachment to FGF2 is mediated, in part, by a cell surface proteoglycan.
A5G27 inhibits FGF2 binding to heparin. The ability of peptide A5G27 to block heparin binding to FGF2 was studied. Heparin-acrylic beads were pretreated with either 5 or 10 μg of A5G27, A5G27S, AG73, or no peptide and then incubated with recombinant FGF2. After incubation, the bound FGF2 was removed by boiling in SDS sample buffer and characterized by Western blot analysis ( Fig. 3A ). Peptide A5G27 directly blocked FGF2 binding to heparin at both concentrations tested with a 54% inhibition at 5 μg and an 81% inhibition at 10 μg ( Fig. 3B). The scrambled peptide, A5G27S, and another heparin-binding laminin peptide, AG73, did not block this binding. These data show that peptide A5G27 binds to heparin and masks the specific FGF2 binding site on heparin, which decreases FGF2 binding.
A5G27 inhibited FGF2-induced migration and invasion with WiDr cells. Because A5G27 blocked cell attachment to FGF2, we examined whether FGF2-induced migration and invasion were also inhibited by A5G27. A5G27 alone had no effect on WiDr cell migration, but it did block FGF2-induced migration ( Fig. 4A ). Peptide A5G27 alone inhibited cell invasion and it also blocked FGF2-induced invasion ( Fig. 4B) in a dose-dependent manner (not shown). The scrambled (A5G27S) peptide was inactive in both the cell migration and invasion assays. Thus, peptide A5G27 can block FGF2-induced cell migration and invasion.
Protein tyrosine phosphorylation: ∼30 kDa bands are reduced by peptide A5G27 in the presence and absence of FGF2. Because A5G27 inhibited FGF2 binding to WiDr, we examined whether A5G27 could inhibit the tyrosine phosphorylation induced by FGF2. WiDr cells were incubated in the presence of 100 μg/mL of A5G27 ( Fig. 5A, lane 4 ), A5G27S ( Fig. 5A, lane 6), 50 ng/mL of FGF2 ( Fig. 5A, lane 2), FGF2 together with either A5G27 ( Fig. 5A, lane 3) or A5G27S ( Fig. 5A, lane 5), and medium alone ( Fig. 5A, lane 1) for 30 minutes. First, we tested whether FGF2 would induce MAPK activation with WiDr cells. Western blotting with anti-p44/42 and phospho-p44/42 MAPK antibodies did not show any effect of FGF2 alone or in combination with peptide A5G27 on MAPK activation with WiDr cells ( Fig. 5A). These data suggest that such phosphorylation may be cell type specific. Next, we stained with an antiphosphotyrosine antibody. The samples were collected from cells after treatment with FGF2 for 30 minutes ( Fig. 5B) and 2 hours ( Fig. 5C). Several bands were observed on Western blots after 30 minutes of FGF2 treatment but only the ∼30 kDa bands showed FGF2-induced phosphorylation ( Fig. 5B, lanes 1 and 2). The phosphorylation of the ∼30 kDa bands was reduced by ∼80% when the A5G27 peptide was added ( Fig. 5B, lanes 3 and 4). The scrambled peptide A5G27S did not block protein tyrosine phosphorylation ( Fig. 5B, lanes 5 and 6). After a 2-hour incubation with FGF2, Western blotting showed that FGF2-induced phosphorylation had returned to the control level ( Fig. 5C, lane 2); however, A5G27 still inhibited the phosphorylation but to a lesser extent (∼45%; Fig. 5C, lanes 3 and 4). The identity of the ∼30 kDa bands is not known. At the 30-minute time point, a weak phosphorylated band at ∼40 kDa showed increased phosphorylation in the presence of the laminin peptide A5G27 independent of the presence of FGF2 ( Fig. 5B, arrow). Further testing with phospho-p38 antibody ( Fig. 5D) showed that laminin peptide A5G27 increased p38 phosphorylation 1.8-fold in the presence and absence of FGF2. These data show that laminin peptide A5G27 regulates protein phosphorylation in both a FGF2-dependent and FGF2-independent manner.
Laminin promotes the malignant phenotype and increases tumor growth in vivo ( 1). Various active sites for malignancy have been identified at the synthetic peptide level ( 6– 9). Previously, we identified four α5 chain G-domain active peptides that inhibited tumor metastasis and growth in vivo ( 18). One of the four peptides, A5G27, blocked lung colonization, tumor growth, tumor apoptosis and angiogenesis, and FGF2-induced angiogenesis in the chick chorioallantoic membrane assay ( 18). Peptide A5G27 also blocked tumor cell invasion and migration. Peptide A5G27 bound to the glycosaminoglycan side chains of CD44. CD44 pre-mRNA is encoded by 20 exons ( 25, 39), 10 of which can be regulated by alternative splicing. CD44 standard, the smallest CD44 isoform, is ubiquitously expressed in vertebrates, in developing organisms, and in adult organisms, whereas the larger variant isoforms are expressed in only a few epithelial tissues, mainly in proliferating cells and in several cancers.
CD44 has at least 20 isoforms, with the larger variant isoforms playing an important role in malignancy. In particular, CD44v3 expression levels correlate with malignancy ( 40). CD44v3 contains a number of covalently attached heparan sulfate side chains that bind heparin-binding growth factors ( 23, 24, 39, 40). The interaction of CD44v3-heparan sulfate and heparin-binding growth factors promotes cell proliferation, migration, angiogenesis, and survival. Several studies suggest that CD44v3 plays an important role in cancer metastasis and that its functional site is the heparan sulfate side chain that binds the heparin-binding growth factors.
Here, we have characterized the interaction of laminin peptide A5G27 with CD44 and determined a possible mechanism by which this peptide inhibits angiogenesis and metastasis. We found that CD44v3 and CD44v6 were expressed on the surface of WiDr cells by FACScan analysis and immunoprecipitation. WiDr cell attachment to A5G27 was inhibited by heparan sulfate, heparin, and chondroitin sulfate B. These results suggest that A5G27 might bind CD44v3-heparan sulfate and block growth factors from binding. FGF2 is one of the heparin-binding growth factors that binds CD44v3-heparan sulfate. We found that A5G27 inhibited WiDr cell attachment, migration, and invasion to FGF2. Interestingly, A5G27 alone did not affect WiDr cell migration but invasion was inhibited by A5G27 alone. Matrigel, which contains growth factors ( 41), was used to coat the membranes for the invasion assay and it is likely that A5G27 blocked the activity of FGF2 or other growth factors present in the Matrigel that contribute to cell invasion. We also found that glycosidase treatment of the cells to remove cell surface proteoglycans reduced cell binding to FGF2. Finally, we found that the laminin peptide A5G27 inhibited FGF2 binding to heparin directly. A minimal pentasaccharide sequence in heparin/heparan sulfate has been found to bind FGF2 ( 42). We found that peptide A5G27 has sequence homology to one of the regions on FGF2 that binds heparin ( 37). Nine of the 13 amino acids are identical or highly conserved with several of the others being similar in charge ( Fig. 6 ). This region of homology with FGF2 is one of the two major sites for heparin binding and contains key amino acids for FGF receptor binding and for FGF2 central cavity formation in addition to the heparin-binding sequences. A key amino acid for FGF2 central cavitation, leucine at residue 23 on FGF2, is identical in A5G27. The lysine and asparagines at residues 26 and 27 on FGF2 are important for heparin binding. The asparagine is conserved between A5G27 and FGF2. The FGF receptor binding site at amino acids 24 and 25 are not homologous between A5G27 and FGF2. We speculate that A5G27 competes with this site on FGF2 for binding to heparan sulfate on CD44. In preliminary studies, we found that the corresponding homologous FGF2 peptide has cell attachment activity but it is weaker than peptide A5G27. Further peptide analysis of this region will be required to confirm this role. It will also be important to determine if this sequence in laminin is physiologically relevant in vivo as a site that competes for FGF2 binding to heparin-like glycosaminoglycan chains. Laminin has multiple heparin-binding sites that were generally thought to be important in the structural interactions of this protein with perlecan, the proteoglycan present in basement membrane ( 1). Both laminin and FGF2 bind to CD44 and it is now possible to speculate that some of the sites on laminin that bind heparin have important biological functions. In this regard, laminin is highly protease sensitive and many of the active peptides identified on laminin have been found to bind heparin ( 1, 5, 6, 10, 14, 15). This peptide, A5G27, is the first to be able to promote cell attachment but have an inhibitory effect on FGF2 activity. It may have a physiologic role in development or in tumor growth where proteases involved in remodeling could degrade laminin and release active fragments.
As a consequence of the reduced binding to CD44v3-heparan sulfate, A5G27-treated cells showed a reduction in FGF2-induced protein phosphorylation. The FGF2-induced phosphorylation of ∼30 kDa proteins was reduced by laminin peptide A5G27. The identity of these bands is not known. Antibodies to caveolin-1, p-caveolin, Cip1, and p27 failed to react with these bands. 6 Furthermore, we had the unexpected finding that laminin peptide A5G27, in the presence or absence of FGF2, increased p38 phosphorylation. These data suggest that this peptide acts by multiple pathways to regulate cell signaling.
Our results suggest that A5G27 binds directly to heparan sulfate and inhibits the interaction of CD44v3-heparan sulfate with FGF2. Thus, laminin peptide A5G27 blocks FGF2-induced activity. Hyaluronic acid is also known to bind to CD44 ( 43), but this binding was not blocked by peptide A5G27. CD44v3 is one of the larger CD44 isoforms that are expressed in only a few proliferating epithelial tissues and in several cancers ( 39). Many reports show that the larger CD44 isoforms play important roles in malignancy. Anticancer agents that target the larger CD44 isoforms have not been developed but may have therapeutic value. A5G27 might be an important new anticancer agent that functionally blocks the larger CD44 isoforms and functionally blocks FGF2 activity.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
↵6 Unpublished observations.
- Received January 31, 2005.
- Revision received August 18, 2005.
- Accepted September 14, 2005.
- ©2005 American Association for Cancer Research.