How adhesion/growth-regulatory galectins-1 and -3 attain cell specificity: Case study defining their target on neuroblastoma cells (SK-N-MC) and marked affinity regulation by affecting microdomain organization of the membrane
Cell surface glycans constitute versatile biochemical signals for functional trans interactions with endogenous lectins (1, 2). Together with glycoproteins glycolipids serve to present carbohydrate chains with substituted β-galactoside termini (3, 4). Due to their spatial accessibility these epitopes are in principle well positioned as contact sites for binding partners, e.g. from the family of adhesion/growth-regulatory galectins (5). Given the abundance of β-galactosides on the cell surface, an intriguing question concerns these lectins' inherent specificity of ligand selection. Suggestive evidence on human SK-N-MC neuroblastoma cells supports the concept for a direct and functional interaction between the pentasaccharide of ganglioside GM1 and galectins, e.g. the homodimeric (crosslinking) galectin-1, with assumed functional consequence on cell growth (6–9). This tumor model is thus suited to put the hypothesis to the experimental test that a galectin can have a cell-type-specific glycolipid ligand, by blocking its synthesis.
Moreover, since the increase of ganglioside presentation is caused by a microdomain-associated cell surface ganglioside sialidase (10), we can next answer the question whether this topology of ganglioside presentation is crucial for lectin affinity. Alternatively, ligand presence might suffice for lectin binding, as probed by perturbing the integrity of microdomains without altering overall ganglioside content. Our experiments using two inhibitors of ceramide glucosyltransferase and 2-hydroxypropyl-β-glyclodextrin and the macrolide filipin III reveal marked dependence of galectin binding on ganglioside GM1 organized in physiological microdomains. In other words, galectin-1 has specificity for a distinct β-galactoside-containing glycoconjugate and senses the topology of its presentation.
Human galectins obtained by recombinant production were purified by affinity chromatography on lactosylated Sepharose 4B resin, megaprimer PCR-derived His-tagged W68Y/E71Q mutants of galectin-1 by chromatography on Ni-CAM™ HC resin (Sigma, Munich, Germany), and wild-type/mutant proteins were subjected to quality controls by gel electrophoresis and filtration as described (7, 11). Radioiodination was carried out under activity-preserving conditions to yield specific radioactivity of 198 KBq/μg for galectin-1 and 221 KBq/μg for galectin-3 using carrier-free Na125I and Iodobeads, the radioiodinated proteins were then tested for carbohydrate-dependent quantitative cell binding as described (6).
Treatment of Cells and Binding Assays
Neuroblastoma cells (strain SK-N-MC) were cultured in Eagle's minimal essential medium supplemented with 10% fetal calf serum. To inhibit glycolipid synthesis, specific inhibitors of glucosyltransferase-catalyzed biosynthesis of glucosylceramide, N-butyldeoxynojirimycin (NB-DNJ; Sigma-Aldrich, Taufkirchen, Germany) or threo-1-phenyl-2-decanoylamino-3-morpholino-1-propanol (PDMP; Sigma-Aldrich) were applied (12, 13). To that effect cells were cultured in the presence of 50 μM NB-DNJ or 10 μM PDMP for five days. Then the cells were subcultured in 96-well plates (for binding assays) or in 75 cm2 flasks (for measurement of glycolipid-bound neuraminic acid) and grown for additional 5 days in the presence of the inhibitors. For testing the effects of perturbation of the integrity of glycolipid microdomains cells were grown to 90% confluency for 5 days. Then the cells were treated for cholesterol depletion with 5 mM hydroxypropyl-β-cyclodextrin (HPBCD; Sigma-Aldrich) or 3 μg/mL filipin III (Sigma-Aldrich) (14) for 1 h at 37 °C in serum-free medium before binding assays or measurements of glycolipid-bound neuraminic acid were conducted. Carbohydrate-dependent binding of 125I-labeled galectins to the cultured cells was assessed exactly as described previously (6).
Determination of Lipid-bound Neuraminic Acid
Cells were harvested by trypsinization, homogenized by ultrasonication, and a membrane fraction isolated by ultracentrifugation (430000 g, 14 min, 2 °C). Membranes were extracted with chloroform/methanol 2:1 (v/v) and the extracts dried in a stream of nitrogen. After its release by treatment with 200 μL 0.1 M H2SO4 for 1 h at 96 °C neuraminic acid was measured by the thiobarbituric acid assay (15).
Binding of Galectin-1 to Inhibitor-treated Cells
Exposure of the neuroblastoma cells to the two inhibitors of glucosylceramide synthesis led to a notable reduction of extent of lipid-bound neuraminic acid (Fig. 1). Monitoring cell binding of radioiodinated galectin-1 will reveal any consequences of diminishing ganglioside content. Irrespective of the specific inhibitor used the extent of cell binding by the lectin was markedly reduced (Fig. 2). When mutant proteins with impaired carbohydrate-binding activity, i.e. at the sites W68Y/E71Q, were tested, their complete lack of reactivity to cells excluded noncarbohydrate-dependent cell association (data not shown). Additionally, we performed experiments with galectin-3, which acts as competitive inhibitor for galectin-1 in this cell system (7). Its level of cell association dropped in line with reduction in ganglioside presence, exemplarily shown for NB-DNJ in Fig. 3. Of note, the galectin-reactive ganglioside GM1 is a marker for microdomains (3). Its selection as primary target on the cell surface may thus solely depend on its presence or alternatively be made feasible by this special topological mode of glycan presentation. To answer this question, two different reagents to disrupt the integrity of microdomains were applied.
Binding of Galectin-1 to Cells with Impaired Microdomain Organization
To avoid invalid interpretation we first excluded any effect of cell treatment on the cellular level of ganglioside presence (Fig. 1). Corroborating these data, cell binding experiments led to an only slightly reduced level of saturation with both reagents (Fig. 4). But the amount of galectin-1 to reach saturation differed drastically. As consequence, the dissociation constant increased by a factor of 10–12 from 685 nM in controls to 6.67 μM (HPBCD) and 9.09 μM (filipin III), respectively. Considering the lower degree of crosslinking activity of chimeric galectin-3, a comparatively smaller effect than for homodimeric galectin-1 was expected, if ordered ligand presentation matters. Indeed, the KD increased by a factor of 5–6 in this case, Fig. 3 presenting data on the effects of HPBCD on galectin-3 binding.
Glycosphingolipids are known to intimately associate with membrane glycoproteins, such as distinct growth factor receptors or integrins (3). In contrast, their status as binding partners for tissue lectins is much less explored, as documented by a recent compilation of known ligands for galectins-1 and -3 (16). In fact, looking at known glycoconjugates as lectin targets onthe cell surface, a major role appears to be played by glycoproteins, e.g. carcinoembryonic antigen, laminin, and lysosome-associated membrane glycoproteins as ligands for galectins in human KM12 colon cancer cells or the glycoprotein CD7 as clinically functional galectin-1 counter-receptor of leukemic T cells in Sézary syndrome (17, 18). Our report, applying two inhibitors of glucosylceramide synthesis, describes the ability of ganglioside to serve as initial contact site for galectins-1 and -3. Previous work showing galectin-1 to be a major receptor for ganglioside GM1 (6) points to this as the likely functional ganglioside. A compensation of the inhibitor-dependent ganglioside deficit by glycoprotein glycans is not operative, as has also been seen in vivo for galectin-1 binding to T cells of mice lacking GM2/GD2 synthase (19). In the latter study, Ca2+ influx occurred after crosslinking as also seen for two other ganglioside GM1-specific probes, i.e. cholera toxin B-submit and pentameric anti-GM1 IgM, acting on NG108-15 or N18 neuroblastoma cells (20, 21). Of note, GM1 is also implicated in galectin-1 internalization and sorting into Golgi via raft mediated endocytosis and clathrin-coated vesicles, the latter also operative in GM1-deficient rat C6 glioma cells (22). Due to the intracellular presence of both galectins and ganglioside GM1, e.g. in nuclei and the nuclear envelope (4, 23), these results tempt to suggest a respective interaction also at sites different from the cell surface.
Since this ganglioside is a major constituent of microdomains, these observations prompted us to address the issues whether and to what extent this mode of topological ligand presentation in situ has a bearing on affinity. The ensuing testing of two reagents affecting microdomain integrity but not ganglioside content provided a clear answer. Diminished affinity for two galectins and grading of the reduction between the homodimeric galectin-1 and galectin-3, with its less potent ability for crosslinking (24), reveal physiological ligand clustering as molecular switch for affinity. In hindsight, these results offer an explanation for the low capacity of the proteolytically truncated form of galectin-3 to bind to the SK-N-MC neuroblastoma cells and the low affinity of galectin-1 to this ganglioside in a chromatographic system (7, 25). In practical terms, they can also have implications for considerations to identify cellular ligands based on processing cell extracts after homogenization by lectin affinity chromatography or blotting. Overall, our presented data reveal the strategic coordination of selectivity for a glycan and this glycan's topological presentation to attain specific high-affinity binding.