We have previously reported that silencing of galectin-4 expression in polarized HT-29 cells perturbed apical biosynthetic trafficking and resulted in a phenotype similar to the inhibitor of glycosylation, 1-benzyl-2-acetamido-2-deoxy-β-d-galactopyranoside (GalNAcα-O-bn). We now present evidence of a lipid raft-based galectin-4-dependent mechanism of apical delivery of glycoproteins in these cells. First, galectin-4 recruits the apical glycoproteins in detergent-resistant membranes (DRMs) because these glycoproteins were depleted in DRMs isolated from galectin-4-knockdown (KD) HT-29 5M12 cells. DRM-associated glycoproteins were identified as ligands for galectin-4. Structural analysis showed that DRMs were markedly enriched in a series of complex N-glycans in comparison to detergent-soluble membranes. Second, in galectin-4-KD cells, the apical glycoproteins still exit the Golgi but accumulated inside the cells, showing that their recruitment within lipid rafts and their apical trafficking required the delivery of galectin-4 at a post-Golgi level. This lectin that is synthesized on free cytoplasmic ribosomes is externalized from HT-29 cells mostly in the apical medium and follows an apical endocytic–recycling pathway that is required for the apical biosynthetic pathway. Together, our data show that the pattern of N-glycosylation of glycoproteins serves as a recognition signal for endocytosed galectin-4, which drives the raft-dependent apical pathway of glycoproteins in enterocyte-like HT-29 cells.
A characteristic feature of enterocyte polarization is the formation of distinct apical and basolateral membranes separated by tight junctions. The apical membrane faces the lumen and consists of a brush border of microvilli containing enzymes and transporters for intestinal digestion. This asymmetric structure originates from polarized vesicular trafficking to apical or basolateral membrane domains.
Polarized protein targeting in epithelial cells is based on apical and basolateral sorting signals, which guide the incorporation of proteins into distinct vesicles that then move to the apical or basolateral membrane. Signals for basolateral targeting are encoded by short amino acid sequences containing tyrosine or dileucine motifs localized in the cytoplasmic tail of proteins (1). These sequences are recognized by heterotrimeric adaptor complexes, which then mediate the routing to the basolateral membrane (2). The mechanism for apical delivery appears more variegated than basolateral targeting as various apical sorting signals were found in extracellular, transmembrane or cytoplasmic domains of respective proteins (3). A segregation of proteins for apical transport was proposed to occur in trans Golgi network (TGN) by the recruitment of proteins into detergent-resistant membranes (DRMs), which are membrane microdomains enriched in cholesterol and sphingolipids (4–6). However, recycling endosomes (REs) also appear to play an important role (7). Oligomerization or lectin-mediated cross-linking of raft-associated proteins has been recently proposed as a directing mechanism for apical delivery (8,9). Of note, cross-linking of glycoproteins is a hallmark of galectin activity. Among mammalian lectins, the galectins function to form clusters, arrays or lattices through multivalent protein–carbohydrate interactions (10–12). Galectins recognize the basic disaccharide unit Galβ1-3/4GlcNAc; however, the affinity is enhanced by branching of N-glycans and/or repeating of N-acetyllactosamine units (13).
Our previous work showed the crucial role of glycosylation in the secretion of mucins by goblet cell-like HT-29 cells and in the targeting of membrane proteins to the brush border of enterocytic HT-29 cells. Treatment of polarized HT-29 colon carcinoma cells by an inhibitor of glycosylation 1-benzyl-2-acetamido-2-deoxy-β-D galactopyranoside (GalNAcα-O-bn) induced a failure to deliver membrane or secreted apical proteins and their abnormal storage inside the cells. Initially, this drug was considered to inhibit exclusively the elongation of N-acetylgalactosamine O-linked to a serine or threonine residue in a competitive manner. However, this inhibitor is intensely converted into the disaccharide GalGalNAcα-O-bn inside the HT-29 cells and inhibits the biosynthesis of galactosylceramides and sulfated glycosphingolipids because of the mobilization of uridine diphosphate-Gal for the production of metabolites of GalNAcα-O-bn (14,15). We thus looked for a putative lectin receptor in apical delivery using proteomics approach of DRMs and we identified an endogenous lectin, that is, galectin-4. This galectin is mostly expressed in the gastrointestinal tract of mammals (16). Galectin-4 is a monomer that contains two carbohydrate-binding sites connected by a linker peptide (16). It shows a high affinity for -3Galβ1-3GalNAc pyranoside and sulfated glycosphingolipids (15,17,18). Interestingly, knocking down galectin-4 expression in HT-29 5M12 cells resulted in a phenotype similar to that of GalNAcalpha-O-bn-treated cells, that is, the failure of raft-dependent apical delivery of proteins and their abnormal localization inside the cells (15). Interestingly, the nonraft-dependent apical delivery was later shown to require another galectin, that is, galectin-3 in MadinDarby canine kidney (MDCK) cells and mouse enterocytes (19,20).
Based on these results, we have now investigated the lipid raft-based galectin-4-dependent mechanism of glycoprotein delivery to the brush border of enterocyte-like HT-29 cells.
Having previously identified galectin-4 and its high-affinity ligands, that is, glycosphingolipids, as major components of the DRM fraction detergent insoluble at 37°C in HT-29 5M12 cells (15), we investigated whether this lectin is involved in DRM formation.
Depletion of galectin-4 inhibits the recruitment of apical glycoproteins within DRMs
To determine whether knocking down galectin-4 expression induced changes in the protein pattern associated with lipid rafts, we analyzed the protein pattern of the DRM fraction detergent insoluble at 37°C in control and galectin-4-knockdown (KD) cells (Figure 1A). The specificity of isolated DRMs was controlled for the absence of basolateral glycoproteins using the glycoprotein marker gp525 that was previously shown to be exclusively localized in the detergent-soluble fraction of HT-29 5M12 cells (21). DRMs of control and galectin-4-KD cells were analyzed using two-dimensional (2-D) electrophoresis and matrix-assisted laser desorption/ionization–time-of-flight (MALDI-TOF) mass spectrometry (MS). The amount of each identified protein in the DRM fractions was determined using the software SameSpot (Progenesis). The identity of proteins and their relative ratio in the DRM fractions of control and galectin-4-KD cells are shown in Table 1. Several proteins were found at a higher level in the DRM fraction of galectin-4-KD HT-29 5M12 cells: the ubiquitous markers of DRMs (flotillin-1 and prohibitin, ratios of 1.5- and 1.7-fold, respectively), the ionic pump (vacuolar-type H+-adenosine triphosphatase, subunit A, ratio of 2.9) and β-actin (ratio of 3.1-fold). In addition, some proteins of the cytoskeleton family, that were not detected in the DRM fraction of control cells, appeared in the DRM fraction of galectin-4-KD cells (tropomyosin and proteins of the actin-related protein complex Arp 2/3). Conversely, the two glycoproteins identified in the DRMs of control cells, that is, the transmembrane glycoprotein dipeptidylpeptidase-IV (DPP-IV) and the glycosyl-phosphatidylinositol (GPI)-anchored complement regulatory protein (CD59), were, respectively, strongly decreased (by 10-fold) or not detected in the DRMs of galectin-4-KD cells. Thus, galectin-4 depletion induces two types of alterations in the protein pattern of DRMs: (i) a loss of glycoproteins and (ii) an enrichment in proteins associated with the actin network.
Table 1. Identification of proteins contained in the DRM fraction insoluble at 37°C isolated from control and galectin-4-KD HT-29 5M12 cells
The relative ratio of proteins in galectin-4-KD cells in comparison to control cells was determined using the software SameSpot (Progenesis).
Markers of DRMs
Apical membrane glycoproteins
Vacuolar-type H+-adenosine triphosphatase, subunit A
Vacuolar-type H+-adenosine triphosphatase, subunit A
Membrane cytoskeleton-associated proteins
Actin-related protein 2/3 complex
To confirm the decrease in the amount of glycoproteins in the DRMs of galectin-4-KD cells, independent analyses were undertaken by western blotting. In a representative western blot from two separate experiments, Figure 1B shows that the level of glycoproteins was either decreased [DPP-IV; carcinoembryonic antigen (CEA)] or not detected [CD59; non-specific cross-reacting antigen (NCA)] in the DRM fraction of galectin-4-KD cells. In contrast, the raft marker flotillin-1 was increased in the DRM fraction of galectin-4-KD cells. Altogether, these data suggested a direct implication of galectin-4 in the recruitment of apical glycoproteins within DRMs through a lectin-type interaction.
Galectin-4 binds to the glycoproteins of DRMs
To determine whether galectin-4 was able to interact with the DRM-associated glycoproteins, delipidated DRMs of HT-29 5M12 cells were subjected to affinity chromatography on a galectin-4–Sepharose 4B column, and elution of bound glycoproteins was carried out with 3 m lactose. The flow-through and bound fractions were analyzed by western blot. In a representative western blot from two separate experiments, Figure 1C shows that all these glycoproteins were primarily detected in the fraction retained on the galectin-4 column. The nonglycosylated protein flotillin was used as negative control. These results showed that galectin-4 directly interacted with DRM-associated glycoproteins and that the cognate sugar lactose eluted the glycoproteins, pointing to a lectin–glycan interaction.
The experimental setting of a solid-phase assay was further used to corroborate that galectin-4 binds to DRM-associated glycoproteins and that the detected binding was carbohydrate dependent. Galectin-4 was shown to bind to immobilized delipidated DRMs in a concentration-dependent manner, and the binding was competitively inhibited by a mixture of lactose and the glycoprotein asialofetuin, its N-glycans being strong galectin ligands (data not shown).
Glycans of DRMs are enriched in complex N-glycans
To gain insight on the glycan motifs involved in the DRM association of apical glycoproteins, we undertook investigations on the structures of N- and O-glycans in the DRM fraction versus the membrane fraction remaining soluble in the presence of nonionic detergent at 4°C. After reduction, carboxyamidomethylation and trypsin digestion of these fractions, the N-glycans were released by digestion with N-glycosidase F (PNGase F), purified on a Sep-Pack C18, desialylated, permethylated and analyzed by MALDI-TOF-MS and nano-electrospray ionization-quadrupole-time of flight-MS/MS. The permethylation derivatization increases the sensitivity of detection and leads to predictable fragmentation. The PNGase F-released N-glycans were also characterized by linkage analysis. The structural assignments of the N-glycans were based on molecular weight, MS–MS and linkage data (Figure 2; Supplementary Results: monosaccharide composition (Table S1), structure (Figure S1) and linkage analysis (Table S2) of each released N-glycans).
Based on the MALDI-MS data and currently accepted models of eukaryotic N-glycan biosynthesis, the major N-glycans may be conveniently grouped into the following classes: (i) high-mannose-type structures having from five to eight mannose residues and no core fucosylation, (ii) complex-type structures containing N-acetyllactosamine (Galβ1,4GlcNAc) on the two nonreducing mannose and (iii) hybrid-type structures with five mannose residues and an N-acetyllactosamine on only one nonreducing mannose.
Striking differences were observed between DRMs and detergent-soluble membranes (Figure 2). The spectrum of detergent-soluble membranes was dominated by signals consistent with high-mannose glycans (Man5GlcNAc2–Man9GlcNAc2), whereas complex N-glycans were much more abundant in DRMs. The major complex N-glycans in DRMs consisted of a series of biantennary (mass-to-charge ratio m/z: 1621, 2070, 2244 and 2418), triantennary (m/z: 2693) and tetraantennary (m/z: 2968, 3142 and 3187) N-glycans with the N-acetyllactosamine Galβ1,4GlcNAc sequence. Several glycans were substituted by fucose on the core and/or the antennae (m/z: 2244, 2418, 2693, 3142 and 3387), and bisected structures (with a branched GlcNAc residue on the Man3GlcNAc2-Asn core) were also identified (m/z: 3387). Then, reductive elimination on de-N-glycosylated peptides from both fractions was carried out to determine the major O-glycan structures. After Dowex purification and borate removal, the O-glycans were permethylated, purified on a C18 Sep-Pak cartridge and analyzed by MALDI-TOF-MS. Two major molecular ions were observed equivalently in both detergent-insoluble and detergent-soluble fractions (m/z: 895 and 1256). These major O-glycans are consistent with sialylated and bisialylated T-antigen structures (Neu5Acα2-3Gal β1-3GalNAc-ol and Neu5Acα2-3Galβ1-3(Neu5Ac α2-6)GalNAcol).
Thus, the characteristic of DRMs in comparison to the soluble membrane fraction is their considerable enrichment in complex structures with branched N-acetyllactosamine. To look whether the N-acetyllactosamine sequence was intensified at the apical surface, confocal microscopy analysis was performed using Datura stramonium lectin (DSL), which binds well to branched N-glycans with N-acetyllactosamine and its repeats. Results showed an intense labeling of the apical membrane in control HT-29 5M12 cells, whereas the basolateral labeling was much fainter (Figure 3A). In galectin-4-KD cells, the apical labeling was strongly decreased while a cytoplasmic labeling appeared. The cytoplasmic labeling in galectin-4-KD cells is consistent with the intracellular accumulation of apical glycoproteins in galectin-4-KD cells.
Therefore, DRMs and the brush border contain a high density of complex N-glycans, and consequently a high density of the N-acetyllactosamine sequence, which is the basic unit recognized by galectins (13).
Inhibition of the processing of N-glycans shifts the delivery of apical glycoproteins to the basolateral membrane
To confirm the role of complex N-glycans in apical delivery, we have analyzed the trafficking of CEA in the presence of 1-deoxymannojirimycin, an inhibitor of α-mannosidase I that inhibits the trimming of Man9GlcNAc2 to Man8GlcNAc2 and consequently the N-glycan processing into hybrid and complex N-glycans (22). Thus, in the presence of this inhibitor, glycoproteins carry high-mannose glycans. HT-29 5M12 cells were treated by 1-deoxymannojirimycin for 24 h and then processed for confocal microscopy (Figure 3B). Interestingly, CEA was mainly found at the basolateral membrane in cells treated by 1-deoxymannojirimycin. Similar observations were obtained for DPP-IV (data not shown). These data confirmed that the processing of high-mannose glycans into complex glycans was determinant for the galectin-4-dependent delivery to the apical membrane and further showed that the apical glycoproteins with high-mannose glycans were delivered to the basolateral surface.
Galectin-4 is internalized and recycled back to the apical surface
Our data showing the role of galectin-4 in raft association and apical delivery of glycoproteins implied that galectin-4 entered the biosynthetic pathway. Yet, galectins do not have a signal sequence for transport into the endoplasmic reticulum and are localized in the cytoplasm. However, members of the galectin family are known to be secreted through a nonclassical secretory pathway (23). We thus examined whether HT-29 5M12 cells secreted galectin-4. Because galectin-4 was not detected by western blot analysis of native culture medium of HT-29 5M12, the apical and basolateral culture media were concentrated by 20-fold before western blot analysis and signal quantification. Galectin-4 was found mostly secreted in the apical medium [Figure 4; mean (SD) apical/basolateral ratio of two experiments, 3.3 (0.6)]. Flotillin-1 was used as negative control and was not detected in the basolateral and apical media (data not shown). Therefore, a step of internalization from the apical medium could allow galectin-4 to enter the biosynthetic pathway. To test this hypothesis, we used fluorescein isothiocyanate (FITC)-labeled recombinant galectin-4 that was added to the apical medium at 4°C for 30 min. Cells were then shifted to 37°C for various periods and processed for confocal microscopy.
After the addition of galectin-4-FITC to the apical medium, the labeled galectin-4 bound to the brush border of microvilli, and after 2-min incubation at 37°C, galectin-4-FITC showed colocalization with the early endosome marker (EEA1), indicating an internalization through the endocytic pathway (Figure 5). After 10-min incubation at 37°C, galectin-4-FITC was mostly localized inside the cells, but no colocalization was seen with the TGN marker ST3Gal I. After 20-min incubation at 37°C, galectin-4-FITC showed an extensive colocalization with the apical recycling endosome (ARE) marker Rab11a. This colocalization began to be detected after 12-min incubation (data not shown). In contrast, no colocalization was detected with the late endosome marker mannose-6-phosphate receptor (M6PR) and the lysosomal marker Lamp2. After 3-h incubation at 37°C, galectin-4-FITC was localized at the subapical and apical surface, and colocalization between galectin-4-FITC and apical markers was observed. The recycling of recombinant galectin-4 at the extracellular surface was further confirmed by immunostaining with an antibody against galectin-4 on nonpermeabilized cells (data not shown). Thus, galectin-4 is endocytosed and then recycled back to the apical surface of the cells.
To determine whether the internalization of galectin-4 from the apical surface had a role in the apical trafficking of glycoproteins, we used GalNAcα-O-bn. Indeed, our previous work showed that GalNAcα-O-bn treatment of HT-29 5M12 cells inhibits the apical delivery of glycoproteins in a similar way as it occurs in galectin-4-KD cells, suggesting an impact of GalNAcα-O-bn on the availability of galectin-4 in apical trafficking. To determine whether this was the case, we analyzed binding and uptake of galectin-4-FITC in GalNAcα-O-bn-treated cells. Binding and internalization of galectin-4-FITC were not observed in GalNAcα-O-bn-treated HT-29 5M12 cells as shown by the comparison of the three-dimensional representations of control and GalNAcα-O-bn-treated cells after internalization of galectin-4-FITC (Movie S1 and Movie S2, respectively). Altogether, these data showed that the effect of GalNAcα-O-bn treatment on intracellular trafficking was because of the lack of endocytic–recycling pathway of galectin-4.
To gain further insight on the role of galectin-4 recycling in apical trafficking, we studied whether the addition of recombinant galectin-4 from the apical side of galectin-4-KD cells can restore the transport of apical glycoproteins after internalization and recycling. Galectin-4-KD cells were thus incubated with recombinant galectin-4 for 24 and 48 h. Cells were then processed for confocal microscopy with an antibody directed against the apical marker DPP-IV, and the cellular distribution of DPP-IV was measured by quantification of the fluorescent signal (Figure 6). Results showed that the addition of exogenous galectin-4 led to a progressive increase in the distribution of DPP-IV at the apical surface, indicating that apical transport of DPP-IV was progressively restored after internalization and recycling of galectin-4. In conclusion, the endocytic–recycling pathway of galectin-4 is required for the apical trafficking of glycoproteins.
In galectin-4-depleted cells, the apical markers exit the Golgi but accumulated inside the cells
Having previously shown that depletion of galectin-4 inhibited the apical delivery of newly synthesized DPP-IV using cell surface biotinylation (15), HT-29 5M12 cells were transfected with a vector construct coding for DPP-IV–green fluorescent protein (GFP) to gain further insight on intracellular trafficking. Control and galectin-4-KD HT-29 5M12 cells stably expressing DPP-IV–GFP were analyzed by confocal microscopy in kinetic experiments after accumulation of proteins in the TGN for 2 h at 19.5°C using antibodies directed against a TGN marker (ST3Gal I) and an apical marker (DPP-IV). Immediately upon shifting from 19.5°C to 37°C, DPP-IV–GFP showed colocalization with the TGN marker ST3Gal I in control and galectin-4-KD cells (Figure 7). However, we can also see DPP-IV outside the Golgi, particularly at the apical membrane of control cells because of the time for residual trafficking of DPP-IV before the TGN block (15). After incubation for 30 min at 37°C, DPP-IV–GFP had exited the Golgi in both cell types. DPP-IV–GFP reached the apical membrane in control cells but not in galectin-4-KD cells where DPP-IV–GFP accumulated inside the cells.
To follow the targeting pathway of DPP-IV–GFP, we have monitored the route of DPP-IV–GFP in living control and galectin-4-KD cells using a confocal imaging approach. After incubation at 19.5°C, images were taken every 5 min for 1 h after shifting to 37°C. For each time-point, apical and intracellular GFP signals were quantified from z planes (Figure 8). The strong increase in the apical signal in control cells clearly showed the arrival of DPP-IV–GFP at the apical surface of control cells. The slight increase in the intracellular signal was representative of the exit of DPP-IV–GFP from the TGN after shifting the cells to 37°C. In contrast, the intracellular signal markedly increased in galectin-4-KD cells up to 40 min after shifting the cells to 37°C, indicating an intracellular accumulation of DPP-IV–GFP. The decrease in the intracellular signal after 40 min is likely representative of the intracellular degradation of accumulated proteins as shown by the partial colocalization of the accumulated cargo with Lamp2 in galectin-4-KD cells (15). The apical signal in galectin-4-KD cells only showed a slight increase, resulting from the low proportion of cells still expressing galectin-4 in the population of galectin-4-KD cells.
Previous studies have shown that the inhibitor of glycosylation GalNAcα-O-bn inhibits the delivery of glycoproteins to the brush border membrane of enterocyte-like HT-29 cells (21). Then, we identified a lectin, galectin-4, as the main effector of GalNAcα-O-bn-mediated effects. Galectin-4 is a member of the galectin family of lectins, which is mostly expressed in the gastrointestinal tract of mammals (16). We identified galectin-4 and its high-affinity ligands, that is, glycosphingolipids, as major components of DRMs in HT-29 cells, and silencing of galectin 4 expression induced a block in apical delivery of glycoproteins as in GalNAcα-O-bn-treated HT-29 cells (15).
This paper demonstrates the mechanisms by which galectin-4, a lectin synthesized on free cytoplasmic ribosomes, regulates the raft-dependent apical pathway in enterocyte-like HT-29 cells. First, galectin-4 recruits the apical glycoproteins within DRMs. The high density of complex N-glycans with the N-acetyllactosamine sequence Galβ1,4-GlcNAc in the apically destined glycoproteins provides the avidity for the galectin. Second, in galectin-4-depleted cells, these glycoproteins still exit the Golgi but accumulated inside the cells, showing the requirement of galectin-4 at a post-Golgi level for the clustering and apical delivery of glycoproteins. Galectin-4 is externalized from HT-29 cells and follows an apical endocytic–recycling pathway that is required for the biosynthetic pathway to the brush border membrane.
Galectin-4 is directly implicated in the segregation of apically destined glycoproteins within DRMs and in their apical delivery through the raft-dependent pathway. Beyond raft association, raft clustering is now thought to be a key step in apical sorting (3). In HT-29 cells of intestinal phenotype, galectin-4 plays the role of clustering agent for both GPI-anchored and transmembrane glycoproteins as both types of glycoproteins are depleted in the DRMs of galectin-4-KD cells. Because of the presence of two carbohydrate recognition domains (CRDs), galectin-4 is able to cross-link ligands. Our structural analysis of glycans identified a series of complex-type N-glycans that were found in a high proportion in DRMs and in the brush border, although they were detected in considerably reduced amount in other membranes. They consisted of bi-, tri- and tetraantennary structures with the N-acetyllactosamine Galβ1,4-GlcNAc sequence possibly substituted by fucose and/or sialic acid. In a general way, galectins bind to complex N-glycans with affinities proportional to N-acetyllactosamine content, and the number of N-glycosylation sites cooperates with the degree of N-glycan branching in the avidity for galectins (13,24). Galectin-4 binding was shown to be enhanced by clustered ligand presentation, notably by clusters of tri- and tetraantennary complex N-glycans of CEA (18,25,26). Thus, the density of complex N-glycans in apically destined glycoproteins determines their avidity for galectin-4.
The multivalent affinity of galectin-4 for glycoproteins and glycolipids is involved in the formation of lipid rafts and in the biogenesis of the brush border membrane of intestinal cells. Glycosphingolipids, which are strongly enriched in DRMs, are also high-affinity ligands for galectin-4 (15,18). Thus, galectin-4 can bind at the same time to both a glycolipid and a glycoprotein as the two CRDs of galectin-4 differ significantly from each other in specificity for oligosaccharides (27). In addition, galectin-4 is capable of forming dimers, and the presence of four CRDs enhances the cross-linking potential of the lectin (28). The cross-linking function of galectin-4 persists at the apical membrane: in pig enterocytes, Danielsen and Van Deurs identified complexes of galectin-4 with transmembrane enzymes in DRMs of the brush border (29). Furthermore, galectin-4 was found to bind to sulfated glycosphingolipids and CEA in patches on the cell surface of human colon carcinoma LS174T cells (18). Altogether, these data support the notion that galectin-4-mediated clustering is part of the maturation of the apical membrane of intestinal cells. The nondelivery of apical glycoproteins in the absence of galectin-4 suggests that this lectin plays a pivotal role in the quality control of the biogenesis of the brush border membrane of enterocyte-like HT-29 5M12 cells. In the model of enterocytic Caco-2 cells, for which raft-dependent apical trafficking has also been reported (30,31), galectin-4 begins to be expressed after confluence, when the cells begin their enterocytic differentiation with the concomitant presence of an apical brush border and of brush border-associated hydrolases, suggesting that galectin-4 could also play a role in the differentiation process of this cell type (data not shown).
In galectin-4-depleted cells, the apical glycoproteins exited the Golgi but accumulated inside the cells. Therefore, galectin-4 has to meet the newly synthesized apical glycoproteins at a post-Golgi level to form clustered raft-associated glycoproteins and to generate the apical raft carriers. Several studies reported the involvement of endocytic compartments in the biosynthetic traffic of proteins to the apical surface in MDCK cells. Vesicular stomatitis virus glycoprotein (VSV-G) and an apically targeted VSV-G mutant were shown to transiently enter REs before delivery to the plasma membrane (7). The nonraft-associated protein endolyn traverses AREs, whereas the raft-associated marker hemagglutinin passes through an apical endocytic compartment different from the ARE and apical early endosomes and is not clearly identified at present (32). Therefore, endocytosis of galectin-4 might be the way by which the lectin could join the newly synthesized apically destined glycoproteins in HT-29 cells. Although galectins are cytosolic proteins lacking any sequence for transport into the endoplasmic reticulum, they can be externalized through nonconventional secretory pathway, and different vesicular and nonvesicular mechanisms have been proposed (23,33). Galectins may be endocytosed as was shown for the one CRD galectin-3 and the two CRDs galectin-8 (34,35). Furthermore, the intracellular pathway of these galectins after endocytosis was shown to be connected to their fine carbohydrate specificity (35). In HT-29 cells, nonconventional mechanism(s) allows the externalization of galectin-4 from the cytosol, mostly in the apical medium. Exogenous recombinant galectin-4 added to the apical medium is rapidly internalized from the apical cell surface. In the same way, endogenous galectin-4, when secreted in the apical medium, must be rapidly endocytosed, resulting in a low level of galectin-4 detectable in the apical medium. In accordance, several studies reported the rapid cell surface binding and endocytosis of extracellular galectins (26,34–36). After endocytosis in HT-29 cells, galectin-4 is not directed to the degradative endosomal/lysosomal pathway but follows an apical recycling pathway. The apical endocytic–recycling pathway of galectin-4 is required for apical biosynthetic trafficking as shown by the restoration of apical transport in the galectin-4-KD cells after the addition of exogenous galectin-4. These data show that galectin-4 is directly involved in the sorting of nascent glycoproteins for the delivery to the apical membrane.
In conclusion, the glycosylation pattern of apically destined glycoproteins serves as a recognition signal for their delivery to the brush border membrane of enterocytes. The requirement for a suitable N-glycosylation pattern for this apical trafficking may be linked to the function of glycans in the protection of underlying proteins and of epithelial intestinal surfaces.
Materials and Methods
Antibodies and lectins
For western blotting experiments, monoclonal antibodies (mAbs) against CD59 (ab9183), NCA (ab26286) and flotillin-1 (610821) were purchased from Abcam and BD Biosciences. Polyclonal antibodies (pAbs) against galectin-4 (AF1227) and CD26 (28341) were purchased from RD systems and Abcam, respectively.
For confocal microscopy analysis, mAbs against CD26 (BMS143), EEA1 (610457) and Lamp2 were purchased from Bender Medsystems, BD Biosciences and Abcam, respectively. pAbs against Rab11a and M6PR were purchased from Zymed and Abcam, respectively. mAbs against MUC1 (214D4), ST3Gal I (4B10) and CEA (517) were gifts from J. Hilkens (The Netherlands Cancer Institute, Amsterdam, the Netherlands), U. Mandel (School of Dentistry, Copenhagen, Denmark) and A. Le Bivic (IBDM, Marseille, France), respectively. A pAb against human galectin-4 was used [raised as described previously (37)]. FITC-conjugated DSL (FL-1181) was from Vector Laboratories.
Enterocyte-like HT-29 (clone 5M12) cells were cultured as previously described (15). For the analysis of the conditioned media by western blot, control HT-29 5M12 cells were cultured in serum-free medium for 24 h, and apical and basolateral media were collected and concentrated by 20-fold on centricon 10. GalNAc-O-bn was used at the concentration of 2 mm for 48 h. 1-Deoxymannojirimycin was used at the concentration of 1 mm for 24 h in the presence of leupeptin (100 μg/mL).
Isolation of DRMs
Total membrane fraction was obtained as previously described (21). DRMs were isolated from this fraction after 1% Triton-X-100 treatment at 4°C and ultracentrifugation on a discontinuous sucrose gradient (38). They were further treated with 1% Triton-X-100 at 37°C and centrifuged (39).
Western blot analysis
The samples were processed as previously described (15). Detection was carried out by luminescence using the enhanced chemiluminescence western blotting system (Amersham).
2-D gels and MALDI-TOF-MS analyses were performed as previously described using gels processed by silver staining (21).
Human galectin-4 was purified after recombinant expression and checked for purity and carbohydrate-binding activity as described prior to covalent conjugation to divinyl sulfone-activated Sepharose 4B (26). Proteins of the DRMs were precipitated with 10% trichloroacetic acid; delipidated with acetone and methanol; solubilized in 10 mm Tris buffer at pH 7.0 containing 150 mm NaCl, 1% Triton-X-100, 0.5% Nonidet P-40 and protease inhibitors and incubated with the galectin-4–Sepharose 4B column for 2 h at room temperature. The bound proteins were eluted with 3 m lactose.
Carbohydrate-dependent binding of galectin-4 to DRM-associated glycoproteins was determined using a solid-phase assay as described (40,41). DRMs were delipidated, coated on microtiterplate wells (using different amounts of DRMs) and incubated with biotinylated human galectin-4 (5 μg/mL) as probe and then with streptavidin–horseradish peroxidase conjugate as indicator. Asialofetuin was used as positive control. Competitive inhibition of galectin-4 binding to DRMs was performed by coincubation of the labeled galectin-4 with a mixture of 75 mm lactose and 1 mg/mL asialofetuin.
Release, preparation and MS analysis of glycans
Several preparations of DRMs were pooled to obtain a sufficient amount of proteins for structural analysis (300 μg). The proteins/glycoproteins were reduced and carboxyamidomethylated followed by sequential tryptic and peptide N-glycosidase F digestion and Sep-Pak purification. Putative O-glycopeptides remaining after PNGase F digestion of the tryptic glycopeptides were subjected to reductive elimination. Neuraminidase treatment, permethylation of the freeze-dried glycans, MALDI-TOF-MS and electrospray MS/MS analyses of permethylated glycans were performed as described elsewhere (42).
Internalization of galectin-4
HT-29 5M12 cells were cultured on culture-treated Transwell polyester membrane inserts for 10 days. Galectin-4-FITC prepared under full protection of the carbohydrate-binding activity as ascertained by cell assays (43) was added at a concentration of 1.2 μm in the serum-free apical medium, and cells were incubated for 30 min at 4°C. Cells were then shifted to 37°C, washed to eliminate residual galectin-4-FITC, further incubated at 37°C and processed for confocal microscopy. Internalization of galectin-4-FITC was also performed after treatment of confluent cells with 2 mm GalNAcα-O-bn for 48 h in serum-free medium. To determine whether the addition of recombinant galectin-4 could restore the apical transport in galectin-4-depleted cells, galectin-4-KD HT-29 5M12 cells were cultured on membrane inserts for 10 days and recombinant galectin-4 was added for 24 and 48 h. Cells were then processed for confocal microscopy.
Transfection of DPP-IV–GFP
Subconfluent cultures (∼60% confluence) of HT-29 clone 5M12 cells were transfected with 2 μg of the pECD26GFP-C1 vector using Effectene according to manufacturer’s instructions. After 48 h, the culture medium was replaced by the selection medium (DMEM supplemented with 10% inactivated fetal bovine serum and 600 μg/mL of geneticin G418 sulfate). Clones were selected for expansion and further characterized by confocal microscopy.
Inhibition of galectin-4 gene expression by retroviral-mediated RNA interference
Galectin-4-KD HT-29 5M12 DPP-IV–GFP cells were generated by using a retrovirus-mediated RNA interference system as described previously (15).
Cells were processed as previously described (15) using pAbs against Rab11a, M6PR and galectin-4 and mAbs against EEA1, ST3Gal I, DPP-IV, Lamp2, MUC1 or FITC-conjugated DSL (40 μg/mL). Laser microscopy analyses were performed using a Leica instrument (model TCS-NT or model Sp2), Leica Geosystems, France. Files of microphotographs were processed with Adobe Photoshop 5.0. In galectin-4-KD cells treated by recombinant galectin-4, the apical transport was analyzed by quantification of the distribution of the apical marker DPP-IV.
Time-lapse images were collected using a confocal microscopy system equipped with a thermostatic chamber (Leica instrument, model Sp2). Cells stably expressing DPP-IV–GFP were cultured on coverslips and mounted in a perfused open and closed chamber (POC). After accumulation of the proteins in the TGN for 2 h at 19.5°C in DMEM containing 25 mm HEPES buffer and 20 μg/mL cycloheximide, the POC was then warmed at 37°C on the microscope stage and images were taken every 5 min for 1 h with a ×100 objective.
We thank Dr Kai Simons for critical reading of this manuscript. We thank Dr Didier Lefranc for the quantification of 2-D gels, Didier Pointu for the movies and Odile Moreau-Hannedouche and Hélène Fontayne for technical assistance. This work was supported by Institut National de la Santé et de la Recherche Médicale (INSERM), Centre National de la Recherche Scientifique (CNRS) and Ministère de la Recherche et de l’Enseignement Supérieur. The Mass Spectrometry Facility used in this study was funded by the European Community (FEDER), the Région Nord-Pas de Calais (France) and the Université des Sciences et Technologies de Lille I. Confocal microscopy and time-lapse analysis were performed in the IFR114 IMPRT and the IBL/IRI, respectively.