Phosphorylation of the β-Galactoside-binding Protein Galectin-3 Modulates Binding to Its Ligands*

The β-galactoside-binding protein galectin-3 has pleiotropic biological functions and has been implicated in cell growth, differentiation, adhesion, RNA processing, apoptosis, and malignant transformation. Galectin-3 may be phosphorylated at N-terminal Ser6, but the role of phosphorylation in determining interactions of this endogenous lectin with its ligands remains to be elucidated. We therefore studied the effect of phosphorylation on binding of galectin-3 to two of its reported ligands, laminin and purified colon cancer mucin. Human recombinant galectin-3 was phosphorylated in vitro by casein kinase I, and separated from the native species by isoelectric focusing for use in solid phase binding assays. Non-phosphorylated galectin-3 bound to laminin and asialomucin in a dose-dependent manner with half-maximal binding at 1.5 μg/ml. Phosphorylation reduced saturation binding to each ligand by >85%. Ligand binding could be fully restored by dephosphorylation with protein phosphatase type 1. Mutation of galectin-3 at Ser6 (Ser to Glu) did not alter galectin ligand binding. Metabolic labeling or separation by isoelectric focusing confirmed the presence of phosphorylated galectin-3 speciesin vivo in the cytosol of human colon cancer cells from which ligand mucin was purified. Phosphorylation significantly reduces the interaction of galectin-3 with its ligands. The process by which phosphorylation modulates protein-carbohydrate interactions has important implications for understanding the biological functions of this protein, and may serve as an “on/off” switch for its sugar binding capabilities.

The ␤-galactoside-binding protein galectin-3 has pleiotropic biological functions and has been implicated in cell growth, differentiation, adhesion, RNA processing, apoptosis, and malignant transformation. Galectin-3 may be phosphorylated at N-terminal Ser 6 , but the role of phosphorylation in determining interactions of this endogenous lectin with its ligands remains to be elucidated. We therefore studied the effect of phosphorylation on binding of galectin-3 to two of its reported ligands, laminin and purified colon cancer mucin. Human recombinant galectin-3 was phosphorylated in vitro by casein kinase I, and separated from the native species by isoelectric focusing for use in solid phase binding assays. Non-phosphorylated galectin-3 bound to laminin and asialomucin in a dose-dependent manner with half-maximal binding at 1.5 g/ml. Phosphorylation reduced saturation binding to each ligand by >85%. Ligand binding could be fully restored by dephosphorylation with protein phosphatase type 1. Mutation of galectin-3 at Ser 6 (Ser to Glu) did not alter galectin ligand binding. Metabolic labeling or separation by isoelectric focusing confirmed the presence of phosphorylated galectin-3 species in vivo in the cytosol of human colon cancer cells from which ligand mucin was purified. Phosphorylation significantly reduces the interaction of galectin-3 with its ligands. The process by which phosphorylation modulates protein-carbohydrate interactions has important implications for understanding the biological functions of this protein, and may serve as an "on/off" switch for its sugar binding capabilities.
Galectin-3 may be phosphorylated at NϪterminal Ser 6 (major) and Ser 12 (minor) (40) and the major acidic residues on both sides of Ser 6 make this a likely substrate for casein kinase I and/or for casein kinase II (26,40,41). Phosphorylated galectin-3 has been demonstrated to be present in the cytosolic and nuclear fractions of 3T3 fibroblasts (26,41), and in cultured polarized canine epithelial (Madin-Darby canine kidney) cells (40). It has been suggested that phosphorylation may modulate the intracellular function and translocation of galectinϪ3, but the functional role of phosphorylation in determining interactions of this endogenous lectin with its ligands remains to be determined (40).
The present study employs large-scale separation of galectin-3 and its phosphorylated species to examine the effect of phosphorylation on the binding of galectin-3 to two of its reported ligands, laminin (38) and purified colon cancer mucin (39). Phosphorylation reduced saturation binding to each ligand by greater than 85%, and ligand binding could be fully restored by dephosphorylation with protein phosphatase type 1. Metabolic labeling confirmed the presence of phosphorylated galectin-3 in vivo in the cytosol of human colon cancer cells from which ligand mucin was purified.
Human recombinant galectin-3 and the Ser 6 3 Glu (SG-6) mutant were produced as described previously (36). Asialomucin from human colon cancer cell line LS LIM6 was prepared as described previously (39). Laminin from the basement membrane of the Engelbreth-Holm-Swarm sarcoma laminin, phosphatase inhibitors, and all other reagents, unless specified, were purchased from Sigma.
Cell Culture and Metabolic Labeling-Colon cancer cell lines LS174T, LSLIM6, and HM7 were cultured as described previously (22). For 32 P labeling in vivo, cells were grown in 162-cm 2 flasks (Costar, Cambridge, MA) to confluence. Metabolic labeling and cell lysis was performed according to the procedure described by Huflejt et al. (40) Immunoprecipitation-32 P-Labeled galectin-3 was immunoprecipitated from labeled cell lysates with TIB166 antibodies as described previously (40). As controls, isotype matched normal rat myeloma immunoglobulins were used (Zymed Laboratories Inc., South San Francisco, CA). The immunoprecipitates were boiled for 5 min in reduced sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) 1 sample buffer and separated on a 10% polyacrylamide gel (22). Proteins were identified by either Coomassie Blue staining or immunoblots with R1 antibodies followed by drying and autoradiography.
Casein Kinase I Assay-The casein kinase IϪcatalyzed incorporation of 32 P into human recombinant galectin-3 was carried out as described (40) with the following modification. The reaction mixture contained 50 mM Tris, pH 7.5, 140 mM KCl, 10 mM MgCl 2 , 5 mM dithiothreitol, 0.2 mM [␥-32 P]ATP to a final specific activity of 20 mCi/mmol, 500 units of rat recombinant casein kinase I, and 20 -40 g of affinity purified recombinant galectinϪ3 in a final volume of 50 l. The reaction was carried out for 30 min at 30°C and terminated by addition of reduced SDS-PAGE sample buffer. The sample was boiled for 5 min and separated on a 10% polyacrylamide SDS gel and either visualized by Coomassie Blue staining or transferred to a nitrocellulose membrane for immunoblotting with TIB166 antibodies followed by autoradiography.
Protein Analysis, SDS-PAGE, Isoelectric Focusing, and Immunoblots-Protein concentration was determined using the BioϪRad Protein assay kit (Bio-Rad) and the Coomassie Plus protein assay (Pierce, Rockford, IL). For protein separation the mini-protein unit (Bio-Rad) was used under standard conditions. For isoelectric focusing the IEF Ready Gel, pH 3-10, and IEF cathode and anode buffers (Bio-Rad) were used according to the manufacturer's instructions for nondenatured protein samples. Western analysis was performed as described previously (22). After separation, protein was transferred to nitrocellulose membranes in 0.7% acetic acid, blotted with anti-galectin-3 mAb TIB166, and visualized using an enhanced chemiluminescent detection system (Roche Molecular Biochemicals, Indianapolis, IN). Radiolabeled protein was detected by autoradiography. The relative amount of phosphorylated and nonphosphorylated galectin-3 was estimated by densitometric scoring using a digital imaging system (Alpha Innotech, San Leandro, CA). The radioactivity incorporated into the phosphorylated galectin-3 excised protein band was determined by using a ␤-counter (Packard, Downer Grove, IL).
Purification of Phosphorylated Galectin-3-For purification of phosphorylated galectin-3 (galectin-3-P), the casein kinase I phosphorylation reaction was scaled up. Affinity purified galectin-3 (280 g) was phosphorylated as above, except that 3500 units of casein kinase I and 34 mCi/mmol [ 32 P]ATP were used in a final volume of 350 l. The reaction was carried out for 20 h at 30°C, and terminated by addition of ␤Ϫmercaptoethanol (5 mM final) and glycerol (10% final). The reaction mixture was immediately separated on an IEF gel pH 3-10.
The galectin-3-P protein band was visualized by autoradiography. The non-phosphorylated galectin-3 was identified on a control lane by staining with IEF staining solution (Bio-Rad). Both galectin-3-P and nonphosphorylated galectin-3 bands were excised and eluted with 50 mM ammonium bicarbonate, pH 7.6, and 5 mM EDTA at 37°C for 2 h. The separated proteins were concentrated on a Speed-Vac to 100 l and dialyzed against 50 mM Tris, pH 7.0, at 4°C for 12 h.
Solid-phase Binding Assay-The individual wells of a 96-well polystyrene microtiter plate (Costar) were coated with either Engelbreth-Holm-Swarm sarcoma laminin or asialomucin at 1 g/100 l of phosphate-buffered saline/well at 4°C overnight. The nonspecific proteinbinding sites were saturated with 1% blocking buffer made in phosphate-buffered saline for 1 h at room temperature. Serial dilutions of gel eluted galectin-3, galectin-3-P, SG-6 mutant galectin-3, and dephosphorylated galectin-3 samples in 0.5% blocking buffer were then added to the coated wells and incubated for 1 h at 37°C. Control experiments contained 0.3 M lactose. The wells were next incubated with rabbit anti-human galectin-3 antibodies (R1), diluted 1:3000 in 0.5% blocking buffer for 30 min at 37°C, followed by incubation with biotinylated goat anti-rabbit IgG F(ab) 2 fragment (1:4000) for 30 min at 37°C. In the last incubation step, ABC reagent (Vectastain ABC Kit, Vector, Burlingame, CA) was added for 30 min at 24°C. The proteincoated wells were developed with the ABTS substrate and developed color monitored at 405 nm using a plate reader (Bio-Rad).

Phosphorylation of Galectin-3 and Purification of Galectin-
3-P-Galectin-3 has been previously reported to be an in vitro substrate for casein kinase I (40). Rat recombinant casein kinase I was therefore used to phosphorylate affinity purified human galectin-3. Phosphorylated galectin-3 was characterized by a slight decrease in electrophoretic mobility on 10% SDS-polyacrylamide gels, while retaining immunological recognition by both monoclonal and polyclonal antibodies to galectin-3 (Fig. 1). Upon isoelectric focusing, two major bands with distinct pI values resolved at pI 8.2 and pI 7.6 with 32 P radioactivity associated with the more acidic band (Fig. 2). Immunoblot analysis confirmed that the two bands were immunoreactive with antibody to galectin-3. This demonstrated the feasibility of separating galectin-3 into its more acidic phosphorylated form and its nonphosphorylated form for further experiments. Phosphogalectin was then purified using a "scaled up" phosphorylated reaction described under "Experimental Procedures." After separation by isoelectric focusing, bands representing phosphorylated and nonphosphorylated galectin-3 species were excised and eluted by diffusion in ammonium bicarbonate. Products were then analyzed by Western analysis. Galectin-3 and galectin-3-P were purified to homogeneity from the reaction mixture (Fig. 3A) and all radioactivity was found to be associated with galectin-3-P (Fig. 3B).
Phosphorylation Alters Binding of Galectin-3 to Its Ligands-A better understanding of the function of galectin-3 in normal and neoplastic epithelium will require determination of the distinct role of phosphorylation in ligand-lectin interactions (40). We therefore examined the effect of phosphorylation of galectin-3 on binding to two of its characterized ligands, laminin and colon cancer mucin. Ligand binding was quantitated for native protein, galectin-3-P and after dephosphorylation of galectin-3-P with the serine-threonine phosphatase protein phosphatase type 1 (Fig. 4). Galectin-3 bound both ligands in a dose-dependent manner with half-maximal binding at 1.5 g/ml (Fig. 5). This binding was sugar dependent as it was abrogated by 0.3 M lactose. Galectin-3-P exhibited markedly reduced binding, with saturation reaching only 15% at galectin-3 saturation binding levels for both ligands. Dephosphorylation of galectin-3-P fully restored galectin-3-ligand binding, confirming the role of phosphorylation in diminishing binding of galectin-3 to its ligands.
It has been previously demonstrated (40) that galectin-3 is predominantly phosphorylated on Ser 6 . Control phosphorylation reactions were performed with purified galectin-3 mutated at this site (Ser 6 3 Glu) (36). Mutated galectin-3 purified from this source bound its ligands to a similar or increased degree compared with native galectin-3 (Fig. 5). These results confirm that the reduced ability of galectin-3-P to bind its ligands is not simply the result of the purification procedure or the phosphorylation reaction per se.
Galectin-3 Is Phosphorylated in Human Colon Cancer Cells-Colon cancer mucin is an important ligand for galectin-3, and interactions between the endogenous lectin and this ligand is altered by phosphorylation (Fig. 5). While phosphorylated galectin-3 has been demonstrated in cells such as 3T3
fibroblasts (31,41) and cultured polarized canine kidney cells (40), it has not been detected in normal or neoplastic human cells including those from colon cancer. Human colon cancer cells were, therefore, metabolically labeled with 32 P as described under "Experimental Procedures," and cell extracts subjected to immunoprecipitation with anti-galectin-3 mAb TIB166. Phosphorylated galectin-3 was detected in all three colon cancer lines studied, and the amount of phosphorylated galectin-3 correlated with metastatic potential (22) (Fig. 6A). Greater than 90% of phosphorylated galectin-3 was associated with the cytosol, and no galectin-3 could be detected in the nuclear fraction (not shown). In order to further document the expression of phosphorylated galectin-3 in these cells, proteins from total cellular homogenates of human colon cancer cells were separated by isoelectric focusing, transferred to nitrocellulose membranes, and galectin-3 and galectin-3-P visualized by immunoblotting (Fig. 6B). Results from these experiments mirrored those obtained by metabolic labeling. Galectin-3 and phosphorylated galectin-3 were again detected in cellular homogenates from colon cancer cells, and the amount of galectin-3-P varied between cell lines. Galectin-3-P represented approximately 8% of total galectin-3 in colon cancer cell line LS174T (8.3%, 8.4%, n ϭ 2), but greater than 20% of total galectin-3 in cell lines LSLIM6 (26.3%, 29.8%) and HM7 (22.6%, 29.6%).

DISCUSSION
The ␤-galactoside-binding protein galectin-3 plays a central role in a variety of biological functions including cellular recognition and adhesion (20, 23), regulation of apoptosis (17), and pre-mRNA splicing (18). This endogenous lectin is found at elevated levels in a wide variety of neoplastic cell types, and is involved in tumorigenicity and cell metastasis in vivo (18, 19, 22,44 -46), at least in part by promoting adhesive interactions (22). Galectin-3 acts as a receptor for ligands containing polyϪNϪacetyllactosamine (Gal␤4GlcNac␤3Gal␤4GlcNAc) sequences, but how galectin-3-ligand recognition and interactions are regulated remains to be determined. Binding to at least some ligands is associated with positive cooperativity determined by repetitive sequences within the NϪterminal domain (47).
Protein phosphorylation is important in modulating a variety of cell processes including cell growth and differentiation, and plays a major role in signal transduction pathways (48 -50). It has recently become apparent that the phosphorylation or dephosphorylation of certain adhesion molecules may play a role in modulation of cell signaling (47,48). E-selectin, for example, is constitutively phosphorylated in cytokine-activated human endothelial cells and undergoes enzymatically regulated dephosphorylation following leukocyte adhesion (47). Galectin-3 has been shown to be phosphorylated, predominantly at N-terminal Ser 6 , and both phosphorylated and nonphosphorylated forms have been found in 3T3 fibroblasts and polarized canine kidney epithelial (Madin-Darby canine kidney) cells (26,40,41). Phosphorylated galectin-3 has been detected in both the cytoplasm and nucleus of 3T3 fibroblasts, and the ratio of phosphorylated to nonphosphorylated galectin-3 varies with the proliferative status of the cells. The precise functions of phosphorylated galectin-3 and the role of phosphorylation in determining the interactions of galectin-3 with its ligands remains unknown. Phosphorylation does not appear to play a role in regulating cellular localization of the lectin (36).
In the present study we confirmed that galectin-3 is a substrate for casein kinase I, and demonstrated that phosphorylation is associated with an alteration in molecular weight and electrophoretic mobility suggesting the possibility of a conformational shift induced by phosphorylation. We took advantage of the ability to separate phosphorylated galectin-3 from nonphosphorylated protein by isoelectric focusing to separate sufficient amounts of each species for comparative functional studies.
Phosphorylation reduced saturation binding of galectin-3 to laminin and mucin by greater than 85%. This was not a result of protein denaturation during in vitro phosphorylation since galectin-3 mutated at Ser 6 bound in a manner similar to native galectin-3. A growing body of evidence suggests that protein phosphatases may play a role in modulating cell-cell or cellmatrix adhesion (48). We have demonstrated the ability of the protein serine-threonine phosphatase protein phosphatase 1 (PP1) (51) to dephosphorylate galectin-3-P in vitro. Dephosphorylation fully restored binding of galectin-3 to its ligands. It is possible that galectin-3 is constitutively phosphorylated in vivo by casein kinase-I, and that the ratio of phosphorylated to nonphosphorylated protein is regulated by phosphatase activity (40). A similar mechanism has been proposed for regulation of selectin phosphorylation and leukocyte adhesion to endothelia (50).
Galectin-3 expression correlates with neoplastic transformation and tumor progression in the colon (22,44,46,52), and colorectal cancer metastases express higher levels of galectin-3 than the primary tumors from which they arise (44,52). UpϪregulation of galectin-3 in colon cancer cells by stable transfection leads to an increase in spontaneous metastasis and liver colonization, while down-regulation by antisense methodology significantly reduces metastasis (22). This may be due in part to alterations in cellular adhesive interactions. Phosphorylated galectin-3 was detected in the cytosol of the colon cancer cells from which ligand mucin was purified, and was most abundant in cells with high metastatic potential. We have recently demonstrated that mucin produced by colon cancer cells is an important ligand for galectin-3 (39), and that galectin-3 may actually play a role in regulation of mucin synthesis (53). Our data suggest that phosphorylation of galectin-3 may play a role in regulation of these processes.
Phosphorylation of galectin-3 significantly alters the interaction of galectin-3 with its ligands. The process by which phosphorylation acts as an "on-off switch" for protein-carbohydrate interactions is unknown, but has important implications for understanding the biological functions of this protein. Galectin-3 is unique among galectins in that in addition to a typical carbohydrate recognition domain (located at the C-terminal), it has an unrelated non-carbohydrate-binding N-terminal domain including a 12-amino acid leader sequence containing a casein kinase I serine phosphorylation site (36). The xϪray crystal structure of the carbohydrate recognition domain of galectin-3 has been determined (54), and indicated structural differences between galectin-3 and other galectins which may impact carbohydrate binding specificity. The region corresponding to the dimer interface in galectin-1 and galectin-2 does not, for example, appear to serve a similar role in galectin-3, and oligomerization of galectin-3 may instead depend on interactions between the carbohydrate recognition domain and the N-terminal. While the three-dimensional structure of the N terminus of galectin-3 is not yet known, it appears that it is intact galectin-3, but not the carbohydrate recognition domain alone, which shows avidity for multivalent glycoconjugates (55,56), modulates cell adhesion (23), and induces intracellular signals (35). Phosphorylation of Ser 6 at the N terminus of galectin-3 could lead to a conformational change in the protein, altering its ability to participate in multivalent interactions necessary for its biological functions.