Dimeric Galectin-8 Induces Phosphatidylserine Exposure in Leukocytes through Polylactosamine Recognition by the C-terminal Domain*

Human galectins have distinct and overlapping biological roles in immunological homeostasis. However, the underlying differences among galectins in glycan binding specificity regulating these functions are unclear. Galectin-8 (Gal-8), a tandem repeat galectin, has two distinct carbohydrate recognition domains (CRDs) that may cross-link cell surface counter receptors. Here we report that each Gal-8 CRD has differential glycan binding specificity and that cell signaling activity resides in the C-terminal CRD. Full-length Gal-8 and recombinant individual domains (Gal-8N and Gal-8C) bound to human HL60 cells, but only full-length Gal-8 signaled phosphatidylserine (PS) exposure in cells, which occurred independently of apoptosis. Although desialylation of cells did not alter Gal-8 binding, it enhanced cellular sensitivity to Gal-8-induced PS exposure. By contrast, HL60 cell desialylation increased binding by Gal-8C but reduced Gal-8N binding. Enzymatic reduction in surface poly-N-acetyllactosamine (polyLacNAc) glycans in HL60 cells reduced cell surface binding by Gal-8C but did not alter Gal-8N binding. Cross-linking and light scattering studies showed that Gal-8 is dimeric, and studies on individual subunits indicate that dimerization occurs through the Gal-8N domain. Mutations of individual domains within full-length Gal-8 showed that signaling activity toward HL60 cells resides in the C-terminal domain. In glycan microarray analyses, each CRD of Gal-8 showed different binding, with Gal-8N recognizing sulfated and sialylated glycans and Gal-8C recognizing blood group antigens and polyLacNAc glycans. These results demonstrate that Gal-8 dimerization promotes functional bivalency of each CRD, which allows Gal-8 to signal PS exposure in leukocytes entirely through C-terminal domain recognition of polyLacNAc glycans.

glycosylation might differentially impact cellular sensitivity toward various galectin family members.
Galectin-8 is a tandem repeat galectin of ϳ36 kDa with two carbohydrate-recognition domains (CRDs). It is expressed in a wide variety of organs and cells, including endothelial cells and human thymocytes (26,27), and its expression is up-regulated in several cancers (28). Similar to Gal-1, Gal-2, and Gal-3, Gal-8 signals key responses in different leukocyte populations (26, 29 -31). However, unlike Gal-1, Gal-2, and Gal-3, which form homo-oligomeric structures required for cellular signaling (6), the oligomeric structure of Gal-8 and the contributions of each CRD required for cell signaling toward leukocytes are not well understood. Gal-8 is currently thought to exist as a monomer with each CRD joined by a common linker region, providing functional bivalency (32)(33)(34). However, the two separate CRDs of Gal-8 share little sequence similarity (34), consistent with preliminary findings that each domain likely recognizes distinct ligands (29,35). As previous studies suggest that galectins cross-link homotypic receptors (6,36), the mechanism whereby Gal-8 may signal similar responses remains enigmatic.
Here we describe the signaling activity and binding specificity of Gal-8 toward cell surface glycans and chemically defined glycans on a glycan microarray. Our results show each CRD of Gal-8 recognizes distinct glycans. More importantly, we found that Gal-8 exists as a dimer, thus functionally expressing four CRDs, allowing functional bivalency at each separate domain. Although each domain recognizes cell surface glycans, only the C-terminal domain of Gal-8 recognizes polyLacNAc cell surface glycans (-3Gal␤1-4GlcNAc␤1-) n and induces preaparesis in HL60 cells. These results challenge the current paradigm concerning the mechanisms of tandem repeat galectin signaling and strongly suggest complex biological roles for this subfamily of tandem repeat galectins.
Binding of Galectin to Aminoalkyl Glycosides Immobilized on N-Hydroxysuccinimidyl-activated Glass Surface-Glycan microarrays were prepared as described previously (38,39). For galectin recognition of glycans on the printed glycan microarray, Gal-8, Gal-8 domains, or Gal-8 mutants were incubated in a solution of PBS containing 0.005% Tween 20 and 14 mM 2-mercaptoethanol for 1 h at 25°C. The slide was immersed in PBS containing 0.005% Tween 20, drained, and then overlaid with fluorescein isothiocyanate-streptavidin. After 1 h at room temperature in a dark humid chamber, the slide was washed by successive immersion in PBS/0.01% Tween 20 (three times) and water/0.1% Tween 20 (twice). The slide was briefly rinsed with distilled water and dried under microfiltered air. An image of bound fluorescence was obtained using a microarray scanner (Scan Array Express, PerkinElmer Lifer Sciences). The integrated spot intensities were determined using Metamorph software (Universal Imaging).
Enzymatic Deglycosylation-Prior to enzymatic deglycosylation, HL60 cells were fixed by washing three times in PBS at 4°C, followed by resuspension in 2% paraformaldehyde buffered in PBS (pH 7.4) at 4°C. Cells were allowed to fix overnight on a shaker at 4°C. Following fixation, cells were washed three times in PBS and two times in the appropriate buffer as recommended by the manufacturer. For treatment of cells with neuraminidase, cells were washed in PBS followed by incubation with 100 milliunits of Arthrobacter ureafaciens neuraminidase for 1 h at 37°C. For treatment of cells with bovine testes ␤-galactosidase, cells treated with A. ureafaciens neuraminidase were washed in PBS (pH 5.0) followed by incubation with 100 milliunits of bovine testes ␤-galactosidase for 12 h at 37°C. For treatment of cell with Escherichia freundii endo-␤-galactosidase, cells were washed in 50 mM sodium acetate, pH 5.8, and incubated with 200 milliunits of E. freundii endo-␤-galactosidase (Seikagaku Kogyo) for 24 h at 37°C. Following enzymatic deglycosylation, cells were washed in PBS. Buffer control treatments lacking enzymes were used for each individual condition.
Galectin Binding to Cells-For lectin binding, cells were washed twice in PBS at 4°C and incubated with biotinylated Gal-8, Gal-8 domains, or Gal-8 mutants or the indicated plant lectins (LEA and RCA-I, Vector Labs) at a concentration of between 5 and 10 g/ml at 4°C for 1 h. As controls, cells were incubated with 50 mM lactose along with the galectins. Cells were washed three times and incubated with Alexa Fluor 488 streptavidin or Alexa Fluor 633 streptavidin (Molecular Probes) at 4°C for 1 h. Cells were washed twice, followed by resuspension in 400 l of PBS for analysis by flow cytometry using a FACSCalibur flow cytometer (BD Biosciences). The bars in each graph represent the percent change in binding when compared with the binding of control buffer treated cells from each enzymatic pair. Error bars in each graph represent standard deviation of duplicate analysis.
Chemical Cross-linking of Gal-8 and Gal-8 CRDs-Gal-8, Gal-8N, or Gal-8C (5 M each) were incubated with a 50-fold molar excess bis(sulfo-succinimidyl) suberate (BS 3 ) according to the manufacturer's instructions for 30 min at room temperature. Unreacted BS 3 was quenched with 1 M Tris-HCl. Following quenching, samples were boiled in SDS containing 14 mM 2-ME, followed by SDS-PAGE. For protein visualization, gels were incubated in a 50% methanol/10% acetic acid solution overnight, followed by incubation in 50% ethanol for 1 h. Gels were pretreated with 0.04 g of sodium thiosulfate in 200 ml of deionized water (dH 2 O) for 1 min. Pretreated gels were washed with deionized water (three times), followed by impregnation with 0.4 g of silver nitrate in 200 ml of dH 2 O containing 150 l of 37% HCHO for 20 min. Following impregnation, gels were washed with dH 2 O (three times), and protein was detected following incubation with 12 g of Na 2 CO 3 in 200 ml of dH 2 O, plus 100 l of 37% HCHO and 4 ml of pretreatment solution. For co-incubation with Gal-8C and Gal-8N, Alexa-labeled Gal-8C was incubated alone or with Gal-8N followed by cross-linking as outlined above and subjected to SDS-PAGE. As a control, Alexa-labeled Gal-1 was cross-linked as outlined above. Protein was detected by silver stain as outlined above or by detection of fluorescence using a Fluorochem imaging system.
Determining Gal-8 Binding Affinity toward HL60 Cells-Cell binding experiments were performed as outlined previously (20). HL60 cells were biotinylated with NHS-LC-Sulfo-Biotin (Pierce) according to the manufacturer's protocol. Biotinylated cells were fixed in 2% paraformaldehyde buffered in PBS, pH 7.4, at 4°C, followed by washing three times in PBS. Cells were incubated in streptavidin-coated 96 microtiter wells (Pierce) at 50 l per well. Cells were incubated with the indicated concentrations of Alexa Fluor 488 Gal-8 followed by washing three times and detection of binding using a PerkinElmer Life SciencesVictor 2 fluorometer. Analysis of binding isotherms and curve fittings was accomplished using SigmaPlot software (SYSTAT Software Inc.) Measuring Galectin-induced PS Exposure-For Annexin-V staining, cells were treated for 1 h with 100 milliunits of A. ureafaciens neuraminidase or buffer control (RMPI 1640 media or Hanks' balanced salt solution). Cells were then washed in complete RPMI, followed by resuspension in complete RMPI at 10 6 cells/ml. Cells were treated with the indicated concentrations of Gal-8, Gal-8 domains, or Gal-8 mutants at 37°C and 5% CO 2 for the time indicated followed by disengagement with 50 mM lactose and staining with Annexin-V (CalTag/Invitrogen, Carlsbad, CA) as outlined previously (2). Galectin binding toward cells treated with A. ureafaciens neuraminidase was performed as outlined above.

Gal-8 Induces PS Exposure in HL60
Cells-To elucidate the binding specificity of Gal-8 toward leukocyte cell surface glycans, we first examined the ability of Gal-8 to induce PS exposure in HL60 cells. We have successfully used this approach to define the cell surface binding specificity for Gal-1, Gal-2, and Gal-3 (40). Gal-8 induced robust PS exposure in HL60 cells (Fig.  1, B and E). Importantly, inclusion of thiodigalactoside (TDG) inhibited Gal-8-induced PS exposure (Fig. 1, C and E), whereas sucrose had no effect (Fig. 1, D and E), indicating that Gal-8induced PS exposure required recognition of cell surface glycans. Gal-8 also induced PS exposure in a dose-dependent manner ( Fig. 1F) with similar kinetics as observed with other galectins (2, 40) (Fig. 1G). This PS exposure occurred independently of apoptosis, because there was no enhancement of cell or DNA fragmentation (Fig. 2, A-E), and cell viability was unaltered by Gal-8 treatment over 72 h (Fig. 2F).
Desialylation of HL60 Cells Does Not Alter Gal-8 Binding Yet Enhances Gal-8-induced PS Exposure-Because HL60 cells have functional leukocyte glycan ligands for Gal-8, we examined the glycan binding requirements of Gal-8 for functional receptors. Gal-8 bound to HL60 cells (Fig. 3A), and binding was inhibited by TDG but not by sucrose (Fig. 3A), which demonstrates that binding and signaling by Gal-8 require recognition of surface glycan ligands (Fig. 1E). A recent study demonstrated that sialylation results in differential effects on the binding and signaling of other galectin family members (40). Thus, we examined the effects of desialylation on Gal-8 recognition of cell surface glycans. Treatment of cells with neuraminidase did not significantly alter Gal-8 binding (Fig. 3, B and H), although RCA, a plant lectin, which recognizes terminal galactose residues (41,42) that would be exposed after removal of sialic acid, showed a significant increase in binding (Fig. 3, C and H). We next determined whether desialylation of cells by treatment with neuraminidase affected cellular sensitivity to Gal-8-induced PS exposure. Interestingly, desialylated HL60 cells were significantly more sensitive toward Gal-8-induced PS exposure (Fig. 3, D-G and I). Taken together, these results demonstrate that removal of cell surface sialic acid does not appreciably alter Gal-8 binding, but it does enhance cellular sensitivity to Gal-8induced PS exposure.
The discordance between Gal-8 binding and cell signaling suggests a more complex relationship between the interaction of Gal-8 with cell surface glycans and subsequent signaling events than we anticipated. Previous studies demonstrated that the ability of Gal-3 to bind cell surface glycans independently of the sialylation status of the cells results from recognition of internal N-acetyllactosamine (LacNAc) motifs within long chain poly-N-acetyllactosamine (poly-LacNAc) (-3Gal␤1-4GlcNAc␤1-) n glycans (40). By contrast, Gal-1 and Gal-2, which also recognize cell surface polyLac-NAc glycans (20,37,40), primarily recognize the terminal LacNAc unit, making modifications of this LacNAc relevant to glycan recognition (40). To examine whether Gal-8 dis-plays similar binding preferences as Gal-3, we first treated cells with ␤-galactosidase, which removes the terminal galactose of terminal LacNAc-containing glycans. Treatment of cells with ␤-galactosidase did not significantly alter Gal-8 binding (Fig. 3J), although RCA binding was significantly reduced (Fig. 3J), demonstrating the accessibility of cell surface glycans to ␤-galactosidase treatment. By contrast, treatment of cells with endo-␤-galactosidase, which specifically cleaves polyLacNAc, significantly reduced Gal-8 binding in a manner comparable to that observed for LEA (Fig. 3K), a plant lectin that strongly binds polyLacNAc glycans (43). These results suggest that Gal-8 recognizes cell surface polyLacNAc residues through internal LacNAc motif recognition rather than through the terminal LacNAc determinant.
Gal-8N and Gal-8C Fail to Induce PS Exposure in HL60 Cells-Unlike Gal-1, Gal-2, and Gal-3, Gal-8 possesses two unique carbohydrate binding domains, suggesting that a more complex interaction with cell surface glycans may occur. To test this, we first examined whether the C-terminal domain (Gal-8C) and/or the N-terminal domain (Gal-8N) can recognize leukocyte counter receptors as individual domains (Fig. 4A). Both CRDs bound HL60 cells ( Fig. 4B and data not shown). Binding by each domain required glycan recognition and was inhibited by TDG but not sucrose ( Fig.  4B and data not shown). Importantly, treatment of cells with neuraminidase significantly reduced cell surface binding by  Gal-8N (Fig. 4, C, J, and L). By contrast, neuraminidase treatment significantly enhanced binding by Gal-8C (Fig. 4, D, K, and L), consistent with the enhanced cellular sensitivity toward Gal-8 induced PS exposure (Fig. 3I). Unexpectedly, although both domains bound HL60 cells, neither domain induced PS exposure (Fig. 4, F, G, and I). Furthermore, neither domain inhibited the ability of full-length Gal-8 to induce PS exposure (Fig. 4I). Because treatment of cells with neuraminidase enhanced Gal-8C binding and cellular sensitivity to Gal-8-induced PS exposure, we explored whether treatment of cells with neuraminidase might enhance cellular sensitivity toward either domain. However, pre-treatment of cells with neuramin-idase failed to alter cellular sensitivity toward either independent domain (data not shown). Furthermore, co-incubation of both Gal-8N and Gal-8C failed to induce PS exposure in HL60 cells (Fig. 5, A-D). Taken together, these results demonstrate that, although each domain recognizes cell surface glycans, neither by itself can induce PS exposure. Although treatment of cells with neuraminidase increased Gal-8C binding, similar to the enhancement of cellular sensitivity to Gal-8-induced PS exposure, treatment with neuraminidase failed to make cells sensitive to Gal-8C.
Gal-8 Exists as a Dimer-Previous studies with other galectins, especially Gal-1, demonstrated a requirement for galectin  dimerization in galectin-induced signaling, most likely due to a need for galectin-induced cross-linking of cell surface receptors in the successful propagation of cellular signaling (6). The current paradigm concerning tandem repeat galectin signaling suggests that the linker region provides the necessary bivalency between the two separate CRDs, consistent with the inability of each independent CRD or co-incubation with both CRDs to signal on their own (Figs. 4I and 5D). However, to explore whether each domain indeed behaves as a monomer, we performed chemical cross-linking studies using BS 3 , a homobifunctional, water-soluble, non-cleavable cross-linker with a diameter of 11.4 Å. Incubation of Gal-8C with or without BS 3 did not trap a homodimeric Gal-8C species (Fig. 5E). By contrast, incubation of Gal-8N with BS 3 , resulted in significant trapping of a dimeric Gal-8N species (Fig. 5E).
Because co-incubation of Gal-8C and Gal-8N with leukocytes failed to induce PS exposure, we next examined whether Gal-8C and Gal-8N may dimerize. To accomplish this, we fluorescently labeled Gal-8C to enable detection of Gal-8C migration following cross-linking in a mixture of Gal-8N and Gal-8C. Labeled Gal-8C, similar to Gal-8C alone, could not be significantly cross-linked (Fig. 5F). Furthermore, Gal-8C did not prevent cross-linking of Gal-8N when mixed together in parallel experiments (Fig. 5F). Finally, Gal-8C, when co-incubated with Gal-8N, did not enhance the amount of chemically cross-linked dimer (Fig. 5F). These results suggest that, not only does Gal-8C fail to prevent Gal-8N from dimerization, it does not likely form heterodimers with Gal-8N. As a control to ensure that labeling does not preclude chemical trapping, we labeled Gal-1, previously demonstrated to be dimeric (44,45), using the identical protocol. Incubation of BS 3 with Gal-1 resulted in significant trapping of dimer (Fig. 5G). The ability of Gal-8N to dimerize suggested that full-length Gal-8 may also exist as a dimer. However, to rule out the possibility that Gal-8N only dimerizes as an independent domain and does not reflect a potential quaternary structure of the full-length protein, we incubated Gal-8 with BS 3 . Similar to Gal-8N, incubation of Gal-8 with chemical cross-linker trapped significant amount of dimeric species (Fig.  5H). In addition, analysis of Gal-8 using multiangle light scattering demonstrated that Gal-8 and Gal-8N exist as dimers, whereas Gal-8C is a monomer (data not shown). The identification of unmodified monomeric Gal-8C through light scattering studies also ruled out the possibility that fluorescence labeling might somehow prevent oligomerization. Although future studies will examine in detail the monomer-dimer equilibrium, including the extent to which ligands may regulate this equilib- rium, these results demonstrate that Gal-8 exists as a dimer, likely through homodimeric interactions of the N-terminal domain.
Only Gal-8NM Recognizes Cell Surface polyLacNAc Glycans-The ability of Gal-8 to dimerize potentially changes the current paradigm concerning the nature of tandem repeat galectin cross-linking and signaling of functional cellular receptors. To test this, we examined the specificity of each CRD in the context of the full-length protein. We mutated the critical canonical arginine in each CRD to a histidine (R69H in N-terminal domain and R233H in the C-terminal domain), which generated Gal-8NM and Gal-8CM, respectively (Fig. 6A). To confirm that this mutation eliminated glycan recognition by the respective mutated CRD, we tested whether Gal-8C and Gal-8N exhibit differential recognition of cell surface glycans in the context of the full-length protein. To test this, we first examined the effect of neuraminidase treatment on cell surface binding by Gal-8NM and Gal-8CM. Gal-8NM and Gal-8CM recognized cell surface glycans, and TDG, but not sucrose, inhibited recognition, which showed that binding was carbohydrate-dependent ( Fig. 6B and data not shown). Similar to data with the Gal-8N domain alone, treatment of neuraminidase significantly reduced cell surface binding by Gal-8CM (Fig. 6E). By contrast, Gal-8NM showed enhanced binding toward cell surface glycans following removal of sialic acid, consistent with changes observed in binding of the Gal-8C domain alone (Fig. 6E). Interestingly, treatment of cells with endo-␤-galactosidase resulted in a nearly complete reduction in Gal-8NM binding (Fig. 6, C and F), while the same treatment failed to significantly alter Gal-8CM cell surface recognition (Fig. 6, D and F). Consistent with this, treatment with endo-␤-galactosidase resulted in an intermediate decrease in binding of Gal-8 (Fig. 6F), suggesting that the N-terminal domain, in the context of the full-length protein, retains the ability to recognize non-polyLac-NAc-containing cell surface glycans. Importantly, similar binding preferences also occurred following examination of the Gal-8N and Gal-8C domain constructs alone (Fig. 6G), which further demonstrated that this preference relies on the intrinsic binding properties of each individual domain. Taken together, these results demonstrate that specific mutations in each domain of the full-length protein resulted in binding profiles similar to each domain alone. Furthermore, sialylation differentially impacts the binding of each domain and recognition of cell surface polyLacNAc glycans by Gal-8 occurs through the C-terminal domain.
Gal-8 Induces PS Exposure Entirely through C-terminal Domain-Previous studies demonstrated that Gal-1, Gal-2, and Gal-3 signal PS exposure in leukocytes through recognition of cell surface polyLacNAc (40). The dependence of Gal-8 on the C-terminal domain for polyLacNAc recognition and the enhanced binding of Gal-8C following treatment of cells with neuraminidase suggests that the C-terminal domain may alone be responsible for signaling. To test this, we treated cells with Gal-8NM or Gal-8CM followed by examination for PS exposure. Gal-8NM, but not Gal-8CM, induced robust PS exposure in HL60 cells (Fig. 7, A-C and G), and signaling was inhibited by TDG but not by sucrose (Fig. 7H). These results also suggested that the increased sensitivity of cells to Gal-8 following treat- ment with neuraminidase may reflect enhanced recognition of the functional receptor by Gal-8C, despite the failure of neuraminidase to alter Gal-8 binding. Consistent with this possibility, treatment of cells with neuraminidase significantly enhanced PS exposure induced by Gal-8NM (Fig. 7, D-F and G) while failing to alter cellular sensitivity to Gal-8CM. Taken together, these results demonstrate that Gal-8 exists as a dimer and induces PS exposure entirely through C-terminal domain recognition of polyLacNAc glycans.
Gal-8 Recognizes Four Primary Classes of Glycans-To further define the glycan recognition of Gal-8 and each CRD within the protein, we examined binding toward a chemically defined glycan microarray. Examination of Gal-8 at concentrations previously used to explore binding specificity (20, 38 -40) resulted in saturated binding of many glycans (Fig. 8, A and D), suggesting that Gal-8 may accommodate many different glycan modifications. Yet these results are not consistent with the spe-cific types of glycans likely recognized by Gal-8 in concentration ranges over which Gal-8 binds to and signals leukocytes. Although Gal-8 induced PS exposure in the micromolar range, the apparent affinity of Gal-8 toward leukocyte counter-ligands is unknown. To examine the binding in more detail, we measured binding of Gal-8 toward HL60 cells. Gal-8 exhibited saturable binding to HL60 cells, with an apparent K d of ϳ0.5 M (Fig. 8B). Thus, we examined Gal-8 binding toward the glycan microarray at submicromolar concentrations. When binding studies of Gal-8 were performed at 0.3 M, we observed four distinct classes of glycans recognized by Gal-8: sulfated glycans, sialylated glycans, polyLac-NAc, and blood group antigens (Fig.  8, C and E).
Each Gal-8 CRD Binds Distinct Classes of Glycans-Because each CRD of Gal-8 appeared to display unique glycan recognition properties toward cell surface glycans, we next sought to determine whether similar preferences occurred following examination of Gal-8N and Gal-8C binding on the glycan microarray. We first examined the individual domains at a higher concentration to determine whether any overlap in binding may occur. Interestingly, Gal-8N (Fig. 9A) and Gal-8C (Fig. 9B) displayed completely distinct binding even at high concentrations. When evaluated at lower concentrations, Gal-8N and the full-length Gal-8 recognized the same sulfated and sialylated glycans with a similar relative affinity (Fig. 9, E and F). By contrast, Gal-8N did not bind to polyLacNAc glycans or blood group antigens. Gal-8C recognized polyLacNAc and blood group antigens, while exhibiting no binding toward sulfated or sialylated glycans (Fig. 9, G and H). Importantly, sialylation significantly inhibited binding of Gal-8C to polyLacNAc structures (data not shown), consistent with enhanced binding of Gal-8C toward cell surface glycans following neuraminidase treatment. In contrast to Gal-8N, however, Gal-8C displayed much weaker binding toward these glycans (Fig. 9, G and H) when compared with the full-length Gal-8.
To determine whether the mutations in each CRD effectively prevented glycan recognition by the respective mutated CRD, we first analyzed each mutant on the glycan microarray at higher concentrations. Gal-8 binding, following mutation of the C-terminal domain (Gal-8CM), pro- duced identical binding with similar relative affinity toward respective glycans as Gal-8N ( Fig. 9, C, E, and F), which shows that this mutation precluded Gal-8C domain from recognizing glycan. Similarly, Gal-8 possessing the analogous mutation in the N-terminal domain (Gal-8NM) produced identical specificity as the Gal-8C domain (Fig. 9D). However, in contrast to Gal-8C domain, Gal-8NM exhibited higher binding toward these respective glycans (Fig. 9, G and H), which suggested that, although the C-terminal domain may not independently dimerize, it likely behaves as a functionally bivalent CRD in the context of the full-length dimeric protein.

DISCUSSION
Using a combined approach of cell surface binding, cell signaling, and glycan recognition by glycan microarray analyses, our results demonstrate that Gal-8 signals PS exposure by dimeric binding through C-terminal CRD recognition of cell surface polyLacNAc glycans. These results provide new information about the mechanism of Gal-8 signaling and strongly suggest that differential recognition of polyLacNAc glycans may underscore key differences in the biological activities of galectin family members.
The dimeric state of Gal-8 enables functional bivalency at each independent CRD and provides an explanation for differences observed between binding and signaling of Gal-8 before and after treatment of cells with neuraminidase. Although neither domain alone induced PS exposure, the N-terminal domain, but not the C-terminal domain, can dimerize, suggesting that cross-linking of functional cell surface receptors must rely on recognition by a functionally bivalent C-terminal domain. Consistent with this, only Gal-8NM, which contains the active C-terminal domain within the context of the full-length Gal-8, induced PS exposure in HL60 cells. The ability of Gal-8N to dimerize may also partially explain the similar relative affinity for glycans on the microarray between Gal-8N, Gal-8CM, and Gal-8. By contrast, Gal-8C, which does not dimerize, showed significantly lower binding than Gal-8. Similarly, Patnick et al. (35) demonstrated that Gal-8N bound much better to CHO cell surface glycans than Gal-8C. In our study, only Gal-8NM appeared to possess similar binding toward polyLacNAc glycans as Gal-8, suggesting that dimerization may not only provide functional bivalency with two C-terminal domains but also may enhance the affinity of binding to polyLacNAc glycans to successfully induce PS exposure.
The inability of Gal-8C to induce PS exposure, in contrast to Gal-8NM, suggests a general requirement for dimerization-induced cross-linking of functional counter receptors in galectin signaling. Previous studies demonstrated that mutations that inhibit Gal-1 dimerization also prevent signaling, although monomeric Gal-1 still appears to bind similar receptors (6). Importantly, monomeric Gal-1 has lower affinity for glycan ligands than dimeric Gal-1 (20), similar to what we observed for Gal-8C. Removal of the N-terminal domain of Gal-3, responsible for mediating Gal-3 oligomerization (46 -52), also diminishes Gal-3-induced signaling (36). In this way, regulation of monomer-dimer equilibrium appears to be an important mechanism of regulating galectin activity.
The unique structural organization of Gal-8 may reflect a general architecture of tandem repeat galectins. Consistent with this, a recent study on the N-terminal domain of Gal-9 (Gal-9N) demonstrated that it is also a dimer, as evidenced by studies of the crystal structure and solution-based experiments (53). Although we found that only the C-terminal domain of Gal-8 binds the functional signaling receptors on HL60 cells, the unique binding properties of each separate domain corroborates previously studies (29,35,54) and suggests that different cells may possess unique sensitivity to the potential signaling effects of each individual domain (28,55). Furthermore, because the N-terminal domain of Gal-8 intrinsically dimerizes, cleavage of the linker region between Gal-8N and Gal-8C may allow regulatory circuits to dissect potential signaling pathways initiated by each separate domain. Recent results demonstrate that the Gal-8 linker region displays sensitivity to cleavage by thrombin (56), providing at least one possible regulatory pathway capable of separating the potential functional consequences downstream of these two domains. By contrast, the N-terminal domain may also facilitate interactions with extracellular matrix components, which often contain highly sulfated glycans similar to those recognized by the N-terminal domain. This would allow the C-terminal domain to freely signal leukocyte responses. In this way, Gal-8 may possess two separate binding domains analogous to many chemokines, with one domain responsible for signals leukocytes and the other allowing traction on the extracellular matrix (57). Interestingly, only the C-terminal domain of Gal-8 mediates cellular adhesion in vitro (31). Several galectins exhibit chemotaxis activity and recognize extracellular matrix components (31, 45, 58 -61), which may reflect a certain level of convergent evolution between these two families.
The ability of Gal-8 to signal leukocytes entirely through glycan recognition by the C-terminal domain provides insight into the functional glycans required for Gal-8-induced PS exposure in leukocytes. We found that the C-terminal domain primarily binds to blood group antigens and polyLacNAc glycans on the glycan microarray. However, HL60 cells do not express blood group antigens, which suggested that Gal-8 signals exclusively through cell surface recognition of polyLacNAc glycans. Indeed, treatment of cells with endo-␤-galactosidase significantly reduced Gal-8C cell surface recognition. Specific elongation of polyLacNAc glycans on selective glycoproteins may direct galectin binding toward functional cell surface receptors. Consistent with this, previous studies showed that only a few glycoproteins in HL60 cells express significant amounts of polyLacNAc (62).
Although Gal-1, Gal-2, Gal-3, and Gal-8C display specificity toward polyLacNAc cell surface glycans, their fine binding preferences for polyLacNAc differ significantly. Previous results demonstrate that Gal-3 recognizes polyLacNAc glycans independently of the polyLacNAc sialylation state (40,63). In contrast, sialylation modulates binding of polyLac-NAc glycans by Gal-1 and Gal-2 (40,63), similar to Gal-8C. However, the mechanism whereby differential sialylation of cell surface ligands may affect Gal-8 binding and signaling appears to differ from Gal-1 and Gal-2. Although desialylation significantly enhanced Gal-8C recognition, similar treatment of cell surface glycans nearly eliminated binding by Gal-8N. As a result, Gal-8N may behave more like a member of the Siglec family of mammalian lectins, which require terminal sialic acid for ligand binding (64). In this way, differences in the sialylation state of functional ligands may alter cellular sensitivity to the potential signaling abilities of either Gal-8N or Gal-8C in the context of the full-length protein.
The unique preferences of Gal-1, Gal-2, Gal-3, and Gal-8 for polyLacNAc may partially reflect differences in quaternary organization and suggest a model of how each of these galectins interacts with leukocytes (Fig. 10). In contrast to the relatively rigid homodimeric structures of Gal-1 and Gal-2 (44,45,65,66), Gal-3 exists as a flexible oligomer through N-terminal domain interactions (46 -49) (Fig. 10). Similar flexibility may exist between the dimeric Gal-8N domain and linkers that attach the C-terminal domain (Fig.  10). Because Gal-1 and Gal-2 exist as rigid homodimers with glycans bound in opposite orientations (20,40), the prefer-ence for polyLacNAc exhibited by these two proteins likely reflects the conformational flexibility provided by polyLacNAc in exposing the terminal LacNAc, allowing concomitant docking of two terminal LacNAc motifs in each domain (Fig. 10). Importantly, the preference of Gal-1 and Gal-2 for polyLacNAc only occurs following polyLacNAc immobilization, because neither demonstrate preference for polyLacNAc in solution-based assays where this type of conformational flexibility loses relevance (20,40). Furthermore, monomeric Gal-1 fails to share polyLacNAc preference (20), further suggesting that Gal-1 and Gal-2 exhibit polyLacNAc preference due to similarities in quaternary structure. By contrast, the quaternary organization of Gal-3 and Gal-8, which may be more flexible, may underscore the ability of these galectins to bind internal LacNAc motifs within poly-LacNAc (Fig. 10). It is also possible that galectin interactions, as depicted in Fig. 10, may involve higher order cross-linking and lattice formation, as proposed for Gal-1 and Gal-3 (67,68).
Taken together, these studies provide new insights into Gal-8 quaternary structures, cell surface interactions, and signaling through each CRD that may generally reflect signaling mechanisms of other tandem repeat galectins. It will be important in the future to examine in detail the domain or domains responsible for dimerization, and the oligomeric nature of glycan recognition and signaling through the CRDs of the tandem repeat galectins.