Cleavage of galectin-3 by matrix metalloproteases induces angiogenesis in breast cancer



Galectin-3 cleavage is related to progression of human breast and prostate cancer and is partly responsible for tumor growth, angiogenesis and apoptosis resistance in mouse models. A functional polymorphism in galectin-3 gene, determining its susceptibility to cleavage by matrix metalloproteinases (MMPs)-2/-9 is related to racial disparity in breast cancer incidence in Asian and Caucasian women. The purpose of our study is to evaluate (i) if cleavage of galectin-3 could be related to angiogenesis during the progression of human breast cancer, (ii) the role of cleaved galectin-3 in induction of angiogenesis and (iii) determination of the galectin-3 domain responsible for induction of angiogenic response. Galectin-3 null breast cancer cells BT-459 were transfected with either cleavable full-length galectin-3 or its fragmented peptides. Chemotaxis, chemoinvasion, heterotypic aggregation, epithelial-endothelial cell interactions and angiogenesis were compared to noncleavable galectin-3. BT-549-H64 cells harboring cleavable galectin-3 exhibited increased chemotaxis, invasion and interactions with endothelial cells resulting in angiogenesis and 3D morphogenesis compared to BT-549-P64 cells harboring noncleavable galectin-3. BT-549-H64 cells induced increased migration and phosphorylation of focal adhesion kinase in migrating endothelial cells. Endothelial cells cocultured with BT-549 cells transfected with galectin-3 peptides indicate that amino acids 1–62 and 33–250 stimulate migration and morphogenesis of endothelial cells. Immunohistochemical analysis of blood vessel density and galectin-3 cleavage in a breast cancer progression tissue array support the in vitro findings. We conclude that the cleavage of the N terminus of galectin-3 followed by its release in the tumor microenvironment in part leads to breast cancer angiogenesis and progression.

Galectin-3 is associated with interactions between cell–cell and cell–matrix mediated by its carbohydrate-recognition/binding properties.1–3 Its unique chimeric structure enables it to interact with a plethora of ligands and modulate diverse functions like cell growth, adhesion, migration, invasion, angiogenesis, immune functions, apoptosis and endocytosis.4 Galectin-3 consists of 3 structural domains (Fig. 1aI): (a′) NH2-terminal domain containing a serine phosphorylation site, important in regulating its cellular signaling5; (b′) collagen-α-like sequence cleavable by matrix metalloproteinases (MMPs)6, 7; and (c′) COOH-terminal containing a single carbohydrate-recognition domain (CRD) and the NWGR anti-death motif characteristic of the Bcl-2 family.8

Figure 1.

Cell morphology, expression, chemotaxis chemoinvasion and tumorigenic properties of BT-549-H/P64 cells. (aI) Schematic presentation of galectin-3 domains and H/P64 mutation around major MMP cleavage site; (aII) immunofluorescence of actin filaments 24 hr after seeding. (a′) Vector control, (b′) BT-549-H64 and (c′) BT-549-P64; scale bar 200 μm. (b) Western blot of cellular expression of galectin-3 in (I) total cell lysates; (II) the nuclear fraction; (III) conditioned medium; (IV) cell surface by flow cytometric analysis; (cI) chemotaxis toward 50 μg/mL collagen IV using Boyden chamber; (cII) chemoinvasion through Matrigel toward collagen IV. (a′) Vector control; (b′) BT-549-H64; (c′) BT-549-P64; (dI) Reduction of tumor growth in BT-549 cells transfected with noncleavable BT-549-P64 and vector control compared to BT-549-H64; (dII) Angiogenesis in BT-549-H64 (a′, b′) and G33P64 (a″, b″) xenografts using anti-CD34 (a′, a″) and anti-VEGF (b′, b″) antibodies. The IgG antibody control showed no reactivity (not shown). Arrows indicate positive staining. Scale bar 100 μm *p > 0.001.

Cleavage sites of human recombinant galectin-3 by human MMPs were identified between G32-A33 and A62-Y63 amino acids resulting in ∼27 and ∼22 kDa peptides, respectively.6, 9, 10 Cleavage of galectin-3 can be used as a surrogate diagnostic marker of in vivo MMP activity.6 However, the functional significance of this cleavage has remained obscure. It was reported that cleavage drastically improves binding affinity of galectin-3 to laminin11 and endothelial cells.10 However, compared to intact galectin-3, the cleaved ∼22 kDa product shows diminished self-association and ability to hemagglutinate red blood cells.11 It was postulated that cleavage might result in structural alteration of the carbohydrate-recognition domain culminating in higher affinity to glycans and reduction of protein dimerization, thereby abrogating biological properties dependent on such associations.11 To analyze the biological significance of galectin-3 cleavage by MMPs, we constructed mutations at and around the MMP cleavage sites. Substitution of amino acids A33 with G and H64 with P confer resistance to cleavage by MMP-2 and -9.6

Recently, we reported that cleavage of galectin-3 contributes significantly to the tumorigenic potential of the galectin-3 null human breast carcinoma cell line BT-549. The cell clones transfected with cleavage resistant galectin-3 P64 or G33/P64 showed reduced tumor growth in nude mice accompanied by increased sensitivity to apoptosis and reduced angiogenesis compared to cleavable galectin-3 H64.6 Cleavage of galectin-3 could also be observed during progression of human breast12 and prostate13 cancers. An allelic variation in the galectin-3 gene resulting in H64 or P64 amino acids was noted earlier and regarded as a nontumor specific variation.14 In light of the above data, we analyzed the genotype distribution of the H64 or P64 allele in disease free and breast cancer patients. The H/H64 allele exists in disease free Caucasian and Asian women at a frequency of 12% and 5%, respectively, versus 37% and 82%, respectively, in breast cancer patients suggesting that H/H allele is associated with increased breast cancer risk and explains in part the noted disparity in breast cancer incidence in these 2 races.12 Because galectin-3 is not a classic oncogene, but rather a cancer associated gene product, it is reasonable to assume that it exerts its function after getting cleaved by MMPs during the progression stage, i.e., after cancer initiation and promotion by other genes/factors, leading to the malignant phenotype. As the animal studies indicated an increased neoangiogenesis in BT-549 xenografts harboring cleavable galectin-3 compared to the xenografts harboring noncleavable galectin-3,6 we questioned if galectin-3 cleavage could be related to angiogenesis during human breast cancer progression. In this report, we have examined the functional differences between cleavable and noncleavable galectin-3 on tumor phenotypes with special emphasis on angiogenesis-related functions. Angiogenic switch appears to occur at early stages of tumor progression, as indicated by studies on preinvasive lesions of prostatic and breast cancer15, 16 and can be triggered by various signals, genetic mutations, hypoxia/metabolic stress, mechanical stress and the immune/inflammatory response in addition to a variety of angiogenic factors. In our study, we have investigated if cleavage of galectin-3 and its cleavage products could be related to angiogenesis in human breast cancer and have examined its effects on endothelial cell migration, invasion, morphogenesis and cellular interactions resulting in focal adhesion complex formation. Although 2 cleavage sites for MMP-2/-9 have been recognized in galectin-3, the focus of this manuscript is on the role of the naturally occurring single nucleotide polymorphism (SNP) resulting in H64 or P64 galectin-3 with the goal of establishing this cleavage as a therapeutic target. The results suggest that cleavage of galectin-3 resulting in release of its N-terminal is in part responsible for the migration and interactions of endothelial cells within the tumor microenvironment leading to angiogenesis.

Material and Methods

Cell lines, antibodies and reagents

The human breast cancer cell lines MCF10A and MCF10AT1 cells were maintained in Dulbecco's Minimal Essential Medium (DMEM)/F-12 medium (Invitrogen Corporation, Carlsbad, CA) with 0.1 μg/mL cholera toxin, 10 μg/mL insulin, 0.5 μg/mL hydrocortisone, 0.02 μg/mL epidermal growth factor and 5% horse serum (Sigma-Aldrich, St. Louis, MO). MCF-7 cells were maintained in DMEM/F12 medium supplemented with 10 μg/mL insulin and 5% fetal bovine serum (FBS) (Invitrogen). T47D, SK-Br-3 and MDA-MB-231 cells were maintained in DMEM (Invitrogen) containing 10% FBS, essential and nonessential amino acids (Invitrogen), vitamins and antibiotics (Mediatech Cellgro, Herdon, VA). Galectin-3 null BT-549 cells were transfected with galectin-3 cDNA containing A or C191 resulting in H or P64 amino acid and the stable cell clones, namely, 11-9-1-4 (BT-549-H64) and M64 (BT-549-P64) as described6 and were maintained as above. SUM-159 and SUM-149 cells were maintained in Ham's F-12 medium supplemented with 5% FBS, 5 μg/mL insulin and 1 μg/mL hydrocortisone (Sigma-Aldrich). Bovine adrenal medullary endothelial cells (BAMEC) were maintained in Earle's Minimal Essential Medium (Invitrogen) containing 10% heat-inactivated FBS (Hyclone, South Logan, UT), 2 mM glutamine (Mediatech) and antibiotics. All cell lines were grown in a 5% CO2 incubator at 37°C to near confluence and detached from the monolayer with 0.25% trypsin and 2 mM EDTA for 1–2 min at 37°C. The use of cell lines was approved by the Human Investigation Committee, Wayne State University, Detroit, MI. A monoclonal antibody (mAb) specific for intact galectin-3 was isolated from the hybridoma TIB166 clone (ATCC, Manassas, VA). A custom made antipeptide monospecific polyclonal antibody (pAb) (anti-hL31) was prepared against the whole molecule (Genemed, S. San Francisco, CA), which recognized the intact as well as fragmented galectin-3.6, 9

Cell morphology and galectin-3 expression

To view cell shape and distribution of actin filaments, BT-549-H64/P64 cells were fixed with 3.5% paraformaldehyde and stained with 1:1500 dilution of rhodamine-conjugated phalloidin (Sigma-Aldrich). Visualization and documentation were accomplished with an Olympus BX40 microscope supporting a Sony DXC-979MD 3CCCD video camera and stored with the M5+ microcomputer imaging device (Imaging Research, Brock University, St. Catherine, ONT, Canada). Expression of galectin-3 was studied in cell lysates, nuclear fraction and conditioned media by Western blot analysis. The cell surface expression was studied on TIB166 (1:100 dilution) stained cells using fluorescence activated cell sorter (FACScalibur and Cell Quest, Beckton Dickinson, San Jose, CA) as described.17

Tumor growth in nude mice

Cells (2 × 106) suspended in Matrigel were injected into NCR nu/nu mice, obtained from Taconic into the mammary fat pad region s.c on both sides in 2 groups of 6 mice each. Tumor growth was measured weekly and the tumor volume was measured using the following formula: volume = length × width × width/2. The xenografts were harvested after 35 or 56 days as described. The tumors were fixed with 10% buffered formalin, embedded in paraffin and processed for immunohistochemistry, as described.10

Indirect immunoflourescence

The cells were trypsinized and seeded at a density of 20,000 cells/chamber on a 8 chamber slide. After 24 hr, the cells were fixed, permeabilized and processed for indirect immunoflourescence as described18 using anti-FLAG (1:30), anti-hL31 (1:100), anti-β1,-β3 integrin (1:50), anti-FAK (1:50), anti-pFAK (1:50) (Transduction Laboratories, San Jose, CA) antibodies. Flourescein isothiocyanate (FITC) labeled goat anti-rabbit IgG (Sigma) at 1:50 dilution or Texas Red conjugated goat anti mouse IgG (Sigma) at 1:1,500 dilution were used as secondary antibodies. Counterstaining was performed with DAPI, and the cells were observed under an Olympus 1X71 microscope supporting a Hamamatsu 1394 ORCA-ERA video camera and stored using Slidebook digital Microscopy Software (Intelligent Imaging Innovations). The primary antibody was omitted in the controls.

Western blot analysis and gelatin zymography

The cells were suspended in a sample buffer containing 62.5 mM Tris, 10% glycerol, 1% SDS, 1% 2-mercaptoethanol and 1% bromophenol blue at a density of 5,000 cells/μL. To collect the conditioned media, 2 × 106 cells were seeded in a 35-mm Petri dish and the medium was replaced with serum free medium. Conditioned media were collected after 24 hr, concentrated using 5K NMWL Ultrafree centrifugal filter (Millipore Corporation, Billerica, MA), protein concentration was estimated using Bio-Rad Protein Assay reagent (Bio-Rad Laboratories, Hercules, CA) and equivalent amounts (20–40 μg) were subjected to SDS-PAGE using a 12.5% acrylamide gel and Western blot analysis with a 1:500 dilution of anti-galectin-3 mAb or 1:1,000 pAb. Following washes the blots were reacted with 1:5,000 dilutions of IR Dye 680 or IR Dye 800 conjugated corresponding secondary antibody mix (Molecular Probes, Eugene, OR) and scanned by Odyssey infra red imaging system (LI-COR Biosciences, Lincoln, NE) to locate the respective proteins.

The activity of gelatinolytic enzymes (MMP-2 and -9) in the conditioned media (10 μg protein/lane) were detected by electrophoresis on 8% polyacrylamide gels containing 1 mg/mL gelatin as described.19 Briefly, the gel was washed with 2.5% Triton X-100 to remove SDS and incubated with 50 mM Tris HCl pH 8.0 and 5 mM CaCl2 overnight. The gel was stained with 1% Coomassie Brilliant blue and destained with 10% methanol 5% glacial acetic acid. Clear bands in the blue gel indicate gelatinolytic activity.

Chemotaxis and chemoinvasion

Chemotaxis of BAMEC or galectin-3 transfected cell clones of BT-549 cells was performed using a Boyden chamber (Neuroprobe, Cabin John, MD) as described20 using 50 μg/mL of collagen IV (Sigma), recombinant H/P64 galectin-3 (10 μg/mL) or conditioned media from transfected clones (at physiological concentrations) as chemoattractants. Chemoinvasion was studied using a BD BioCoat Matrigel invasion chamber (BD Biosciences Discovery Labware, Bedford, MA) according to the manufacturer's instructions. Briefly, the Matrigel coated chambers were rehydrated and 100 μg/mL collagen IV was added to the wells of a 24-well plate. The Matrigel coated chambers containing the cells were then transferred to the 24-well plates and incubated for 22 hr. Chemoinvasion was measured by removing the noninvading cells with a cotton swab, fixing and staining the invading cells by 100% methanol and toluidine blue, respectively. The cells were counted and photographed using the Zeiss Axiovert 35 microscope as described.

PCR amplification and genotype analysis of various breast cancer cell lines

Genomic DNA was extracted from cells grown to 80% confluence using AccuPrep® Genomic DNA extraction kit (Bioneer, Alameda, CA). The following primer pair was designed to amplify exon 3 of human galectin-3 in breast cancer cell lines



The target sequence was amplified in an Eppendorf® Thermal Cycler using Fast Cycling PCR Master kit. The resulting ∼350 bp DNA fragment was purified using QIAquick® PCR Purification Kit (Qiagen, Valencia, CA), and sequenced using the sense primer at Applied Genomics Technology Center, a Core facility at Wayne State University and Karmanos Cancer Institute.

Expression and secretion of galectin-3 fragments in BT-549 cells

Galectin-3 fragments resulting from the 2 MMP cleavage sites were prepared by PCR amplification from a pcDNA3.1 plasmid containing full-length galectin-3. The primer pairs used for various fragments were as follows:







Fragment 33–250: Forward: 5′-GCCCAAGCTTGCAGGGGGCTACCCAGGG-3′


Fragment 63–250: Forward: 5′-GCCCAAGCTTTACCATGGAGCACCAGGA-3′




PCR amplification was performed with an initial denaturation step of 95°C for 2 min followed by 30 cycles of denaturation at 94°C/30 sec, annealing at 60°C/30 sec and extension at 68°C/1 min. The resulting DNA fragments were purified using QIAquick® PCR Purification Kit (Qiagen), digested with Hind III and EcoR1 restriction enzymes (Invitrogen), ligated to predigested p3XFLAG-MYC-CMV expression vector (Sigma-Aldrich) containing N-terminal 3xFLAG and C-terminal c-myc upstream and downstream of the multiple cloning sites, respectively, and a preprotrypsin leader sequence for secretion. Transformation was performed in XL-10 Gold cells. DNA was extracted from the selected colonies, sequenced and transfected into galectin-3 null BT-549 cells using FuGENE HD transfection reagent (Roche Applied Science, Indianapolis, IN) according to the manufacturer's instructions. The stable clones were analyzed for RNA transcription by PCR amplification of 1 μg RNA using the OneStep RT-PCR kit (Qiagen) and the following primer pair:



The expression and secretion of fusion protein was analyzed by Western blot analysis and immunofluorescence as described.

Immunohistochemical analysis

A breast cancer progression tissue array (BR480) consisting of 5 cases of normal tissue, 8 lobular hyperplasia, 4 lobular atypical hyperplasia, 4 lobular carcinoma in situ, 8 ductal carcinoma in situ, 4 comedo-type intraductal carcinoma with early infiltrate, 9 infiltrating ductal carcinoma and 8 infiltrating lobular carcinoma was purchased from US Biomax (Rockville, MD). Four micrometers of serial sections from the tissue array or mouse xenografts were processed for immunohistochemical analysis using anti-TIB166, anti-hL-31, anti-smooth muscle actin (Dako, Carpentaria, CA), anti-pan keratin (Dako), anti-CD34 (Cell Sciences, Canton, MA) and anti-VEGF (Santa Cruz Biotechnology, Santa Cruz, CA) antibodies at 1 μg/mL as described.6 Visualization and documentation were accomplished as described above.

Measurement of blood vessel density and galectin-3 cleavage

The blood vessels in the tissue array were marked using 2 antibodies by differential staining. One section was stained with anti-alpha smooth muscle actin antibody as a marker of smooth muscle cells, pericytes and myoepithelial cells. Another serial section was stained with anti-pan keratin antibody as a marker of myoepithelial and epithelial cells. Each section in the array was photographed in 4 different areas and the number of blood vessels was counted manually. Only those structures that were negative for keratin staining were considered as blood vessels. The average number of blood vessels per section in each category was calculated.

Cleavage of galectin-3 was analyzed using the M5+ microcomputer imaging device by calculating the mean grain density of each section of antibodies labeled tissue array keeping all the parameters (hue, intensity and saturation) constant. The cleavage of galectin-3 in each section was calculated as the percent of total grain density. The average percentages of all the sections from each category were plotted using Microsoft excel.

Interaction of galectin-3 transfected BT-549 cells with endothelial cell

Interactions between tumor and endothelial cells were studied in 3D cocultures on Matrigel as described earlier10, 20 with some modifications. Briefly, 50,000 BAMEC and 50,000 BT-549-H64/P64 cells were prelabeled with viable stain DiO or DiI respectively (Invitrogen), washed twice with complete medium, mixed and seeded on pregelled Matrigel in an 8 chamber slide. After 24 and 48 hr, branching morphogenesis was studied using an Olympus 1X71 microscope. Quantitation of the formation of capillary-like network was performed manually by measuring the lengths of the tubular structures in 3 photographs taken from each chamber and assigning them an arbitrary unit.

Migration of cocultured galectin-3 transfected BT-549 and endothelial cells

BAMEC or BT-549-H64/P64 cells (2.4 × 104) were seeded in each chamber of the cell culture insert (ibidi GmbH). Cells were prelabeled with DiO or DiI as described. After 24 hr, the cell culture insert was removed and the cell migration of the cocultures toward each other for wound healing was observed. In some cultures, MMP-2/MMP-9 inhibitor III (Calbiochem) was added (10 μM) at the time of removal of the insert. The cells were fixed with 3.5% paraformaldehyde at various time points, photographed or stained for FAK, pFAK and β-1,-3-integrins by immunofluorescence techniques as described.

Statistical analysis

Statistical analysis was performed using Microsoft Excel. The p-values were calculated using 2 sample t-test, assuming unequal variances. Values < 0.05 were considered statistically significant.


Effect of H/P64 amino acid substitution on cell morphology and cellular localization of galectin-3

Freshly seeded vector control and BT-549-P64 cells demonstrated a spindle shaped and BT-549-H64 cells a flattened morphology. Immunostaining of these cells with rhodamine-labeled phalloidin indicated that actin filaments were stretched across the width of BT-549-H64 cells within 24 hr of seeding (Fig. 1aIIb′), but not in control or P64 cell clones (Figs. 1aIIa′ and 1aIIc′). Of note, BT-549-H64 cells looked much larger with a well-formed cytoskeleton compared to control and BT-549-P64 cells (Figs. 1aIIa′–1aIIc′). Presence of galectin-3 was detected in the total cell lysates (Fig. 1bI), as well as in the nuclear fraction of both galectin-3 expressing clones (Fig. 1bII). Using flow cytometry, we detected galectin-3 on the cell surface of both the variants (Fig. 1bIV), however, secretion of full-length galectin-3 as well as the ∼27 kDa cleaved polypeptide was detected in the BT-549-H64 clone only (Fig. 1bIII).

Effect of galectin-3 H/P64 amino acid substitution on chemotaxis and chemoinvasion of breast cancer cells

BT-549-H64 cells exhibited highest migration toward collagen IV, whereas the migration of BT-549-P64 clones was the same as the vector transfected cell clone (Fig. 1cI). Chemoinvasion through Matrigel toward collagen IV was high in BT-549-H64 (Fig. 1cIIb′), whereas other cell clones showed markedly fewer invasive cells (Figs. 1cIIa′ and 1cIIc′). Migrated cell numbers were 1188 ± 200, 46.8 ± 25 and 43 ± 15 in BT-549-H64, BT-549-P64 and vector alone, respectively. The p-values were <0.001 compared to BT-549-H64.

Effect of H/P64 amino acid substitution on tumor growth and angiogenesis

Subcutaneous injections of 1 × 106 cells in the mammary fat pad region of athymic immunocompromised nude mice resulted in a significantly reduced tumorigenicity as well as tumor growth in BT-549-P64 compared to BT-549-H64 cells (Fig. 1dI: tumor take 60 and none respectively; an average of 30 injections in 15 mice). A small tumor mass was seen in the first 2 weeks, but it gradually decreased in size and by 30 days no tumors were visible. Mice bearing BT-549 H64 tumors were sacrificed on day 35 because of the large tumor burden. However, when the mice were injected with another galectin-3 noncleavable clone BT-459 G33/P64 harboring double mutation, 1 tumor grew to a volume of 196 mm3 by day 55, 4 others remained very small (24–40 mm3), while the rest did not grow even after 55 days.6 Immunohistochemical staining of the tumor sections with anti-CD34 antibody showed a reduced density of newly formed blood vessels in BT-549-G33/P64 tumors (Fig. 1dIIa′) compared to BT-549-H64 (Fig. 1dIIa″). The blood vessels in BT-549-H64 tumors were better developed with lumen in some cases, whereas in G33/P64 xenografts, only a few positive precursor cells were seen (arrows). The BT-549-G33/P64 sections also demonstrated a reduced VEGF expression compared to BT-549-H64 (Figs. 1dIIb′ and 1dIIb″). Control tumors did not show blood vessel formation (not shown). These results demonstrated that lack of tumor growth in BT-549-P64 or BT-549-G33/P64 clones could be due to reduced neoangiogenesis in these clones.

Effect of galectin-3 H/P64 amino acid substitution on growth of breast cancer cells

Although BT-549-P64 or BT-549-G33P64 cells had a low tumor growth compared to BT-549-H64 cells, the cell proliferation of BT-549 transfected cell clones was not affected by H/P64 substitution in galectin-3 (Supporting Information). Two independent methods of cell growth were employed: cell proliferation measured by mitochondrial-dependent reduction of 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) (Sigma) to formazan over a period of 10 days (Fig. S1) and anchorage independent growth by soft agar assay measuring colony forming efficiency and colony size (Fig. S2) as described.20

Effect of H/P64 amino acid substitution on extra cellular cleavage of galectin-3 by MMPs

Conditioned media collected from a panel of breast cancer cell lines were immunoblotted with pAb to localize intact and cleaved galectin-3 fragments and gelatin zymography was performed to analyze the presence of gelatinases MMP-2 and -9. Figure 2 shows that galectin-3 was cleaved only in those cell lines that harbor H64 as well as secrete MMP-2 and/or -9. Two cell lines MCF-7 and T47D harboring P64 secreted intact galectin-3, even though MMPs activity was seen in T47D cells. SUM149 and MCF10AT1 cells, which harbor H64 galectin-3 and active MMPs-2/-9, secrete full-length as well as cleaved galectin-3. However, MDA-MB-231 cells, which harbor H64 allele, but do not show gelatinolytic activity, secrete intact and partially cleaved protein. BT-549, SK-Br-3 and SUM159 contain P64 and did not express or secrete galectin-3. These results indicate that presence of H64 and MMP-2 and/or -9 activity is essential for the cleavage at Ala62-Tyr63 site to occur.

Figure 2.

Cleavage of galectin-3 in vivo. Upper panel: Western blots of conditioned media (40 μg total protein/lane) to analyze galectin-3 secretion and cleavage. H64/P64 status of the cell lines is indicated on top. Lower panel: gelatin zymography of conditioned media (10 μg total protein/lane) to monitor MMP-2 and -9 activity.

Galectin-3 cleavage and angiogenesis in breast cancer progression

Utilizing differential immunohistochemical staining with TIB166 mAb and anti-hL-31 pAb, we show that in normal ducts (Figs. 3aa′ and 3aa″) expression of galectin-3 was seen in epithelial cells using TIB166 (staining intact galectin-3) as well as anti-hL-31 (staining intact + cleaved galectin-3) indicating that there is little or no cleavage of galectin-3. Anti-hL-31 showed stronger activity, but was localized in the same cells as TIB166. In hyperplastic ducts, a few epithelial cells exhibited intense signals with anti-hL-31 (arrowheads), but most of the epithelial cells stained positive with both antibodies indicating some cleavage (Figs. 3ab′ and 3ab″). In lobular and ductal carcinoma in situ (Figs. 3ac′,c″ and 3ad′,d″), most of the epithelial cells exhibited positive immunoreactivity with anti-hL-31 (arrowheads) indicating an increased cleavage, whereas only a few cells were positive for intact protein (arrows). In infilterating carcinoma the invasive cell clusters were found to be positive for cleaved galectin-3 (Figs. 3ae′ and 3ae″; arrowheads). Mean grain counts were performed to analyze the percent cleavage in relation to total galectin-3. The average percent values of all the available samples in each stage were plotted. Figure 3b shows a significant increase in galectin-3 cleavage in DCIS and infilterating carcinoma, however, the LCIS did not show an increased cleavage.

Figure 3.

(a) Galectin-3 cleavage in breast cancer progression tissue array. Representative sections are shown. (a′–e′) Intact galectin-3 using TIB166 mAb. (a″–e″) Intact plus cleaved galectin-3 using anti-hL31 pAb. (a′, a″) Normal breast tissue; (b′, b″) Ductal hyperplasia; (c′, c″) lobular carcinoma in situ; (d′, d″) ductal carcinoma in situ; (e′, e″) infilterating carcinoma. Brown color represents positive staining. Arrows: intact galectin-3, arrowheads: cleaved galectin-3. Scale bar 100 μm. (b) Quantitation of cleavage of galectin-3 by counting grain density in each section and calculating percentage of cleaved galectin-3. Statistical analysis was performed between normal tissue and various progressive stages. *p < 0.005.

Blood vessel density was examined in the tissue array by differential staining using anti-smooth muscle actin and anti-pan keratin antibodies. Anti-CD-34 antibody could not be used on these sections, because it is specific only for newly formed vessels. Smooth muscle actin is localized in the myoepithelial cells around the blood vessels and epithelial ducts (arrows), whereas keratin is present only in the epithelial cells (arrowheads). By differential analysis of serial sections (Fig. 4a), we identified the blood vessels and observed a slight increase in the number of blood vessels from normal (a,a′) to lobular hyperplasia (b,b′) to LCIS (c,c′). A significant increase was seen in the blood vessel density in DCIS (d,d′) and infilterating carcinoma (e,e′) (Figs. 4a and 4b).

Figure 4.

(a) Blood vessel density in breast cancer progression tissue array. (a′–e′) Smooth muscle actin; (a″–e″) pan keratin. (a′, a″) normal breast tissue; (b′, b″) lobular hyperplasia; (c′, c″) lobular carcinoma in situ; (d′, d″) ductal carcinoma in situ; (e′, e″) infiltrating carcinoma. Arrows: blood vessel; arrowheads: epithelial cells. Scale bar 50 μm. (b) Quantitation of blood vessels/section. Statistical analysis was performed between normal tissue and various progressive stages. *p < 0.005.

Interactions of H/P64 galectin-3 clones with endothelial cells

Next, we examined if the BT-549 cell clones harboring H64 or P64 galectin-3 interacted differently with endothelial cells. When BAMEC are seeded on Matrigel, they organize into small clusters of cells (Fig. 5a). Recombinant H64 galectin-3 as well as BT-549-H64 cells had a stimulatory effect on the morphogenesis of endothelial cells into tube-like structures (Figs. 5b and 5d). Whereas the structures formed in the presence of rP64 were less organized and weaker (Fig. 5c) compared to rH64. BT-549-P64 cells mainly formed large aggregates when cocultured with the endothelial cells (Fig. 5e). When 10 μg/mL rH/P64 or unconcentrated secreted H/P64 galectin-3 were used as a chemoattractant for endothelial cells, secreted H64 galectin-3 showed stronger chemotaxis than secreted P64 galectin-3. RH64 showed higher chemotaxis than rP64, but it was less than the secreted proteins (Fig. 5f). Matrigel was the strongest chemoattractant for endothelial cells, possibly because of the presence of a number of chemotactic agents in the extracellular matrix.

Figure 5.

Effect of recombinant and secreted H64/P64 galectin-3 on endothelial cell morphogenesis and chemotaxis. 3D homotypic (ac) and heterotypic (d, e) cultures of endothelial and BT-549-H64/P64 cell clones on Matrigel. (a) BAMEC cells; (b, c) BAMEC in the presence of 10 μg /mL rH64 (b) or rP64 (c) galectin-3, respectively; (d, e) BAMEC in the presence of equal number of BT-549-H64 or -P64 cells, respectively; scale bar 10 μm. (f) Chemotaxis of BAMEC toward unconcentrated conditioned media collected from BT-549-H64 or P64 cell clones and 10 μg/mL rH64 or rP64 using Boyden chamber. The membranes were scanned and arbitrary units were used on y-axis. The results are mean of 6 samples for each treatment. Bars represent standard deviation. *p < 0.001; *′p < 0.005.

Next, we performed wound-healing assay employing cell culture inserts for analyzing the migration of endothelial cells and BT-549-H/P64 cells toward each other in coculture conditions. By prelabeling the cells with different color fluorescent vital dyes, the 2 cell lines (epithelial vs. endothelial) could be distinguished after removal of the cell chamber. The wound was completely healed by 24 hr in BT-549-H64 (Fig. 6aa″), while it took longer (∼48 hr, not shown) in BT-549-P64 (Fig. 6ab″). At 24 hr, many red BT-549-H64 cells could be seen invading into the green endothelial component (Fig. 6a″). Addition of MMP-2/MMP-9 inhibitor III, which also inhibits the activities of MMPs-1, -3, -7 and -13, to the cultures delayed the wound healing in BT-549-H64 and to a lesser extent in BT-549-P64 cocultures (Fig. 6aa″′). To analyze if focal adhesion was involved in the initial migration, some of the chambers were fixed, permeabilized and processed for immunofluorescence. Increased migration of endothelial cells toward BT-549-H64 (Fig. 6ba′, bottom box) was seen as compared to BT-549-P64 (Fig. 6bb′, bottom box). No difference in the expression of β1, β3 integrins and FAK in BT-549-H/P64 or endothelial cells was observed at 8 or 24 hr (not shown), but pFAK was overexpressed in endothelial cells migrating toward BT-549-H64 at 8 hr (Figs. 6ba′″ vs. 6b″). BT-549-P64 cells exhibited pFAK levels comparable to BT-549-H64 at the same time point (Figs. 6bbvs. 6ba″).

Figure 6.

Cell migration using wound healing assay. (a) BAMEC and BT-549-H/P64 cells were prelabeled with DiO and DiI, respectively, and seeded in each chamber of cell culture insert. After 24 hr, the inserts were removed and cell migration was studied. (a′–a″′) Migration of BAMEC and BT-549-H64; (b′–b′″) migration of BAMEC and BT-549-P64; (a′, b′) 0 hr, (a″, b″) 24 hr, (a′″, b′″) 24 hr in the presence of 10 μM MMP-2/MMP-9 inhibitor III. Scale bar 20 μm. (b) Expression of pFAK in migrating cells. The wound healing chambers were fixed at 8 hr with 3.5% paraformaldehyde, permeabilized with −20°C methanol acetone (1:1), stained with anti-pFAK antibodies and counterstained with DAPI. (a′) BT-549-H64 (at top) and BAMEC (at bottom). Scale bar 50 μm; (a″, a′″) higher magnifications of boxed areas in BT-549-H64 and BAMEC, respectively; scale bar 200 μm. (b′) BT-549-P64 (at top) and BAMEC (at bottom); scale bar 50 μm; (b″, b″′) higher magnifications of boxed areas in BT-549-P64 and BAMEC, respectively; scale bar 200 μm. Red grains represent positive pFAK staining.

Effect of galectin-3 fragments on cell migration and chemotaxis of endothelial cells

Next, we examined which fragment of cleaved galectin-3 plays a role in endothelial cell migration and morphogenesis into organized structures. We constructed peptides of amino acid sequences resulting from cleavages at A62-Y63 and G32-A33 sites (Fig. 7a). To get peptides in their natural folding states; we expressed them as secreted fusion proteins with 3xFLAG, myc and preprotrypsin tags in BT-549 cells. PCR amplification of mRNA showed that the fragments were transcribed in the clones (Figs. 7ba′ and 7bb′). Next, we confirmed the presence of fusion proteins by dual labeling with anti-galectin-3 and anti-FLAG antibodies by immunofluorescence (not shown). Western blot analysis of conditioned media established the secretion of the fragments as the fusion protein. However, the amounts of the protein secreted by clones harboring the smaller fragments (1–32, 1–62 and 33–62) were much less (Fig. 7bc′) compared to the clones harboring larger fragments (Fig. 7bd′). Next, unconcentrated conditioned media were used to analyze chemotaxis of endothelial cells. Conditioned media collected from cells harboring fragments 1–62 and 33–250 induced maximum migration (at the levels of Matrigel control) of endothelial cells (Fig. 7c). Compared to the full-length intact protein (1–250), these migrations were significantly higher. When 3D cocultures of BT-549 cells and endothelial cells were used to study morphogenesis, clones containing the same 2 fragments (i.e., 1–62 and 33–250) showed 85.5 and 78 units of tubes, respectively, compared to 47, 45, 54 and 40 units by 1–32, 33–62, 1–250 and 63–250 clones, respectively (Fig. 6d). BAMEC showed 49.8 units by themselves in 3D cultures on Matrigel (not shown).

Figure 7.

Induction of chemotaxis and morphogenesis of endothelial cells by galectin-3 fragments. (a) Schematic representation of galectin-3 cleavage fragments; (b: a′, b′) mRNA synthesis in various cell clones by RT-PCR. (c′, d′) Western blot analysis of conditioned media from the clones harboring various fragments as fusion proteins. (a′, c′) Lanes 1: 1–32, 2: 33–62, 3: 1–62; (b′, d′) Lanes 1: 63–250, 2: 33–250, 3: 1–250. (c) Migration of endothelial cells using Boyden chamber toward unconcentrated conditioned media collected from clones expressing various galectin-3 fragments. Matrigel (MG) was used as a positive control. *p < 0.001. (d) 3D cocultures of BAMEC and BT-549 cells transfected with various galectin-3 fragments. Total tube forming units were calculated by measuring tube length between each cell cluster from 2 higher magnification pictures taken for each chamber. (a′) 1–32; (b′) 33–62; (c′) 1–62; (d′) 63–250; (e′) 33–250; (f′) 1–250; scale bar 20 μm.


Secreted galectin-3 plays an important role in the progression of breast cancer and is significantly elevated in the sera of breast cancer patients compared to healthy controls.21 It was suggested that galectin-3 could lead to increased invasive potential of tumor cells by inducing interactions with their stromal counterparts as it was localized in cells proximal to the stroma in comedo-DCIS and in invasive cell clusters along with the surrounding stroma.10, 12 Here, we show that the presence of cleavable galectin-3 may be instrumental in acquiring an invasive phenotype by breast carcinoma cells, as the cells transfected with noncleavable galectin-3 species were significantly less invasive compared to their cleavable counterparts. We detected a lack of gelatinolytic activity and partial cleavage of galectin-3 at G32-A33 site in MDA-MB-231 cells. Lack of gelatinolytic activity in MDA-MB-231 cells was also reported by others.22 Using BT-549 cell clones harboring galectin-3, we reported that secreted galectin-3 binds to endothelial cells and induces morphogenesis.23 In another study, NG2-galectin-3-α3β1 integrin complex formation was reported to be responsible for the stimulation of corneal angiogenesis in vivo provoked by NG2, a transmembrane chondroitin sulfate proteoglycan.24 The significance of galectin-3 for the stabilization of the epithelial and endothelial interactive network using a 3D coculture system of in vitro angiogenesis was also demonstrated.10 Moreover, induction of tumor growth and angiogenesis was reported in clones of human prostate cancer cell line LnCaP expressing cytoplasmic galectin-3.25 However, in none of the above studies, was the role of intact or cleaved galectin-3 examined. In an attempt to further our understanding of the significance of galectin-3 cleavage during cancer progression, we show here that functions regulated by endogenous galectin-3, like tumor cell proliferation and anchorage independent growth, were unaffected, whereas functions regulated by secreted galectin-3, like chemotaxis, chemoinvasion and angiogenesis, were affected by the H64/P64 status. Dynamic changes in actin filaments like stress fibers, lamellipodia, filopodia and membrane ruffles are involved in a wide variety of cellular processes including cell migration, invasion, cell cycle control, cellular structure and cell signaling.26, 27 We have now provided direct evidence to show that H64 galectin-3 is responsible for rapid attachment of the cells and organization of actin filaments.

Next, we expanded our study to examine the relationship of galectin-3 cleavage to angiogenesis in human breast cancer. The results showed a correlation between the blood vessel density and galectin-3 cleavage with progressive stages of breast cancer. Cleavage of galectin-3 could be seen in most of the epithelial cells in DCIS/LCIS and in invasive cell clusters in the infiltrating carcinoma. Figure 3a demonstrates representative sections showing cleavage, whereas Figure 3b shows average numbers of all the sections in tissue array. A decreased cleavage as seen in LCIS in Figure 3b may be due to the presence of P64 containing sections. The blood vessel density increased significantly in DCIS and invasive stages of breast cancer suggesting a possible relationship between the 2 phenomena. This could be explained by a higher interaction/affinity of endothelial cells with H64 galectin-3. We reported earlier that galectin-3 induces endothelial cell morphogenesis and angiogenesis.23 Here, we show that while rH64 galectin-3 or the BT-549-H64 cells induce branching morphogenesis of endothelial cells, rP64 galectin-3 or BT-549-P64 cells failed to do so. Endothelial cell migration is a prerequisite for angiogenesis. Cell migration requires cytoskeleton reorganization involving phosphorylation of cytoskeleton associated tyrosine kinases and formation or removal of focal adhesion complexes that are sites of cell substrate contact28, 29 and where the traction force necessary for cellular movement is generated.30 FAK, the main component of focal adhesion complex is phosphorylated in response to signals inducing endothelial cell migration.31 Focal adhesion complex formation is a dynamic multifactorial process and levels of pFAK change with the assembly or disassembly of complexes32 and is regulated by PKC or small GTPases.33, 34 It was also reported that galectin-3 mediated activation of integrins recruits tyrosine-phosphorylated caveolin-1, thereby stabilizing FAK in focal adhesion domains, promoting focal adhesion disassembly and formation and stimulating cellular displacement and motility.34, 35 Our novel coculture wound healing studies using cell culture inserts show that both tumor and endothelial cells migrate toward each other. The migration was associated with upregulation of pFAK in endothelial cells migrating toward BT-549-H64versus BT-549-P64 cells.

Many proteins seem to have a dual or sometimes opposite effects on tumor angiogenesis after proteolytic cleavage. For example, cleavage products of collagen XVIII (endostatin), brain angiogenesis inhibitor 1 (vasculostatin) and prolactin were shown to have antiangiogenic activity,36–38 while cleaved fragments of hyaluronan stimulate angiogenesis.39 To understand the functional significance of the cleaved fragments of galectin-3 in relation to angiogenic response, we synthesized fragments of all expected sizes by PCR, cloned them in 3xFLAG/myc plasmid containing a secretion sequence and transfected them in galectin-3 null BT-549 cells and used them for in vitro angiogenesis-related studies. Both 1–62 and 33–250 fragments show better capillary-like morphogenesis than rest of the fragments.

In summary, the results indicate that part of the galectin-3 regulated functions during cancer progression are dependent on the cleavage of secreted protein, e.g., functions like chemotaxis, chemoinvasion, heterotypic aggregation, epithelial-endothelial cell interactions and angiogenesis. Cleavage of galectin-3 exposes the region encompassing amino acids 33–62 moiety so as to increase its chemotactic properties toward endothelial cells, upregulation of pFAK in migrating endothelial cells and onset of angiogenesis. Thus, we may conclude that the cleavage of galectin-3 by MMPs plays an important role during breast cancer progression and could be a novel therapeutic target.


We thank Dr. P.V. Malathy Shekhar (WSU) for her constructive discussions during the course of this work, Dr. Dipak. Banerjee (Univ. Peurto Rico) for the gift of BAMEC and Ms. Vivian Powell for secretarial assistance. Financial support by National Institute of Health (to A.R.) and the Karmanos Cancer Institute's Strategic Research Grant (to A.R. and P.N.-M.) is thankfully acknowledged.