Journal list menu

Volume 274, Issue 11 p. 2715-2727
Free Access

Membrane trafficking of CD98 and its ligand galectin 3 in BeWo cells − implication for placental cell fusion

Paola Dalton

Paola Dalton

Department of Physiology, Anatomy and Genetics, University of Oxford, UK

Search for more papers by this author
Helen C. Christian

Helen C. Christian

Department of Physiology, Anatomy and Genetics, University of Oxford, UK

Search for more papers by this author
Christopher W. G. Redman

Christopher W. G. Redman

Nuffield Department of Obstetrics and Gynaecology, John Radcliffe Hospital, Oxford, UK

Search for more papers by this author
Ian L. Sargent

Ian L. Sargent

Nuffield Department of Obstetrics and Gynaecology, John Radcliffe Hospital, Oxford, UK

Search for more papers by this author
C. A. R. Boyd

C. A. R. Boyd

Department of Physiology, Anatomy and Genetics, University of Oxford, UK

Search for more papers by this author
First published: 20 April 2007
Citations: 52
P. Dalton, Department of Physiology, Anatomy and Genetics, University of Oxford, Oxford OX1 3QX, UK
Fax: +44 186 527 2420
Tel: +44 795 286 8502
E-mail: [email protected]

Abstract

CD98 heavy chain (CD98hc), expressed at high levels in developing human trophoblasts, is an integral membrane protein with multiple N-linked glycosylation sites and known to be important for cell fusion, adhesion, and amino acid transport. Western blotting and flow cytometry were used to study the effect of brefeldin A, an inhibitor of protein translocation through the Golgi, on CD98hc in the human placental trophoblast cell line BeWo. Although brefeldin A treatment caused increased cell surface expression of CD98hc, a novel partially glycosylated form of the protein was found and, concomitantly, cell fusion was reduced. Western blotting showed that CD98 and galectin 3, a proposed ligand for the glycosylated extracellular domain of CD98hc, co-immunoprecipitated, and double-label immuno-electron microscopy confirmed that CD98hc associated with galectin 3. Furthermore, cell fusion was reduced (specifically) by the disaccharide lactose, a known ligand for the carbohydrate recognition domain of galectin 3, suggesting that the association was functional. Taken together, the data suggest that N-glycosylation of CD98 and subsequent interaction with galectin 3 is critical for aspects of placental cell biology, and provides a rationale for the observation that, in the mouse, truncation of the CD98hc extracellular domain leads to early embryonic lethality [Tsumura H, Suzuki N, Saito H, Kawano M, Otake S, Kozuka Y, Komada H, Tsurudome M & Ito Y (2003) Biochem Biophys Res Commun308, 847–851].

Abbreviations

  • BFA
  • brefeldin A
  • CRD
  • carbohydrate recognition domain
  • EM
  • electron microscopy
  • ER
  • endoplasmic reticulum
  • FACS
  • fluorescence activated cell sorting
  • Lac
  • lactose
  • PFA
  • paraformaldehyde
  • CD98, a multifunctional membrane protein originally discovered on the surface of activated T cells [1], is now known to be present in many cell types and all malignant cell lines [2]. The CD98 antigen (also known as FRP-1 and 4F2) is a dimeric structure consisting of a type 2 heavily glycosylated integral membrane protein of around 80 kDa (heavy chain) covalently attached to a nonglycosylated integral membrane protein of 40 kDa (light chain); there are six possible light chains, which are expressed differentially according to the tissue of origin [3,4]. The heavy and light chains are linked by a single extracellular disulfide bond. In this heterodimeric form, the CD98 protein is an amino acid transporter transferring specific groups of amino acids across the plasma membrane, the group and the mechanism depending on the properties of the specific light chain. Transfection studies in mammalian cells have indicated that whereas CD98hc can be expressed on the plasma membrane on its own, trafficking of the light chain to the cell surface is possible only in the heterodimeric form and apparently independently of disulfide linkage [5].

    Although roles for CD98 in cellular differentiation, adhesion, growth, apoptosis and amino acid transport have been reported, plausible mechanisms underlying most of these functions are only starting to emerge, and formation of activated complexes with other proteins, in particular β1-integrin, galectin 3 and CD147, has been proposed by various investigators [6–9].

    CD98 expression is also necessary for virus-induced cell fusion and for osteoclast formation [10–12] and, importantly, it is found in cytotrophoblasts and on the plasma membrane of the syncytiotrophoblast of the human placenta [13,14]. Furthermore, manipulation of CD98 expression by antisense oligoneucleotide and small interfering RNA affects both amino acid transport and cell fusion in BeWo cells [15–17]. More recently, we have shown that CD98 involvement in the process of cell fusion that is necessary for syncytiotrophoblast formation is a distinct function from its role in amino acid transport. Indeed, by crosslinking CD98hc with monoclonal antibodies to CD98, we have shown increased surface expression of this molecule and increased fusion of BeWo cells (a well-established choriocarcinoma cell line that can undergo fusion and morphologic differentiation similar to the formation of syncytiotrophoblast by the cytotrophoblasts in the placenta). In contrast, LAT1 (one of the six known light chains) surface expression and amino acid transport were disrupted [18].

    The macrocyclic lactone brefeldin A (BFA) is a metabolite of the fungus Eupenicillium brefeldianum and has antiviral, antibacterial and antifungal activities. Most importantly, though, it specifically and reversibly blocks protein transport from the endoplasmic reticulum (ER) to the Golgi apparatus in many cell types and species. Distinct morphologic changes accompany several specific and reversible effects on cellular protein traffic; however, the precise effects of BFA vary among cell types. Because of its numerous and reversible effects on protein transport and processing, BFA has become an important tool for cell biologists [19,20]. We decided to employ this drug to perturb the protein trafficking and function of CD98 and galectin 3, which has been proposed as an endogenous crosslinker for CD98 [21–23]. Galectin 3 was originally found by Ho and Springer as a surface marker called Mac2, which is present on the cell surface of inflammatory macrophages [24]. Galectins belong to a β-galactoside-binding family of proteins defined by their conserved peptide sequence elements, which are crucial for the carbohydrate-binding activity of those lectins. Fourteen galectins (galectin 1–14) have been found in mammals so far, and are also known in birds, amphibians, fish, nematodes, Drosophila, sponges, and fungi. A common feature of all galectins is the strong modulation of their expression during development. Galectin 3 is expressed widely in epithelial and immune cells, and its expression is correlated with cancer aggressiveness and metastasis. It is reported to be involved in various biological phenomena, including cell growth, adhesion, differentiation, angiogenesis, and apoptosis (indeed, it is the only antiapoptotic galectin family member). Galectin 3 is composed of one carbohydrate recognition domain (CRD), consisting of 130 amino acids, and of an additional non-CRD domain, which is involved in the oligomerization of galectin 3. The oligomerization results in the formation of a galectin 3 molecule that possesses multivalent CRDs. Oligomerization enables galectin 3 to mediate crosslinking of its ligands. In order to crosslink surface ligands to exert its activities, galectin 3, which is mainly intracellular, has to be released extracellularly; however, this protein contains no hydrophobic sequences that may function as signal sequences or transmembrane domains, and is secreted by unknown mechanisms [21,25] (although alternative spliced forms of galectin 3 that contain transmembrane domains have been detected in chicken osteoblast-like cells and in intestine [26]). Finally, there is evidence for galectin 3 as a factor in RNA splicing, based on the localization of the protein in the nucleus [27].

    In this article, we further examine to what extent the functions of CD98, other than amino acid transport, are independent of dimerization with the light chain LAT1, and whether interaction with galectin 3 is necessary to facilitate fusion. The properties of and putative relationship between these two molecules are discussed in the context of cellular distribution and cellular fusion.

    Results

    Expression of CD98 increases over time almost linearly after forskolin treatment

    Fusion of BeWo cells is enhanced by forskolin treatment, which, by activating adenylyl cyclase, results in an increase in intracellular cAMP concentration. We have previously shown, using single-color flow cytometry [fluorescence activated cell sorting (FACS)], a significant increase of CD98 expression on intact BeWo cells after forskolin treatment for 24 h [18]. Here, we determined the levels of expression of CD98 by western blotting in cell extracts from BeWo cells cultured with or without forskolin for 12 h, 24 h, 36 h, or 48 h (Fig. 1A,B): CD98 expression in cells cultured in the presence of the vehicle (dimethylsulfoxide, control cells) was not substantially different from that in cells treated with 100 µm forskolin after 12 h of culture. After 24 h, whereas there was no increase in control cells, the addition of forskolin produced a 20% increase in CD98 expression. This stimulation increased almost linearly up to 48 h, at which time there was approximately 35% more CD98 in BeWo cells cultured in forskolin-containing medium than in control cells. The mean CD98 expression of the two types of culture was significantly different, with a P-value of 0.032 (two-tailed paired t-test).

    Details are in the caption following the image

    The expression of CD98 increases over time almost linearly after forskolin treatment. Western blotting on a 4–12% Bis/Tris NuPage gel run under reducing conditions with Mops running buffer: BeWo cells at 50–60% confluence were treated with dimethylsulfoxide (vehicle control, – F) or 100 µm forskolin (+ F) for the indicated times at 37 °C. (A) Immunoblot after incubation with rabbit anti-(human CD98) (Santa Cruz, 1 : 200), horseradish peroxidase-conjucated goat anti-rabbit IgG and 3,3′-diaminobenzidine (DAB); the data shown are representative of two independent experiments performed in triplicate. (B) Absorbance of the 80 kDa band quantified by densitometry. The data are the means of two independent experiments performed in triplicate ± SEM. The mean CD98 expression of the two types of culture was significantly different, with a P-value of 0.032 (two-tailed paired t-test).

    CD98 surface expression increases after cell treatment with BFA

    In this series of experiments, BeWo cells were cultured in six-well plates in the presence of the vehicle or 100 µm forskolin for 24 h. At 20 h, 22 h, 23 h, and 23.5 h (for a total of 4 h, 2 h, 1 h, and 30 min, respectively), BFA was added to half of the wells to a final concentration of 5 µg·mL−1, and the cells were returned to culture for the remaining period. Single-color FACS, while confirming that forskolin stimulation significantly enhanced CD98 surface expression as compared to control cells (dimethylsulfoxide), clearly showed, contrary to expectation, a time-dependent increase of CD98 surface expression on intact BeWo cells after BFA treatment for both control and forskolin-incubated cells (Fig. 2A). However, this was not due to an increase in the amount of CD98, as total (surface plus cytoplasm) CD98 expression did not significantly change (Fig. 2B), suggesting that BFA treatment had increased CD98 trafficking to the cell surface.

    Details are in the caption following the image

    CD98 surface expression increases after BFA treatment. BeWo cells at 50–60% confluence were incubated in medium containing dimethylsulfoxide (vehicle control) or 100 µm forskolin (Forsk) for 24 h at 37 °C. BFA was added for the indicated times before harvesting, and CD98 was detected by single-color flow cytometry. Cells were labeled with goat anti-(human CD98) and rabbit anti-(goat IgG) conjugated with fluorescein isothiocyanate. (A) Labeling of surface antigens on intact BeWo cells. (B) Labeling of surface and intracellular antigens after cell permeabilization. n = number of cell samples.

    Detection of partially glycosylated/ unglycosylated CD98 after cell treatment with BFA and tunicamycin

    We then used SDS/PAGE and western blotting to look at CD98 expression in cell lysates from BeWo cells cultured for 24 h as above but with or without BFA (5 µg·mL−1) only for the last 4 h. We speculated whether BFA, known to produce distinct morphologic/structural effects at the ER–Golgi level, could cause alteration not only of CD98 trafficking to the plasma membrane but also of its structure.

    Interestingly, after incubation of the blots with rabbit anti-(human CD98), we observed an extra band that ran lower than the normal CD98 band of ∼ 80 kDa (reduced blots) or ∼ 110–120 kDa (nonreduced blots) and had an approximate molecular mass of ∼ 64 kDa or ∼ 80 kDa, depending on whether the gels had been run under reducing or nonreducing conditions. The extra band was present only in the samples treated with BFA in either control or forskolin-stimulated cells (Fig. 3A,B), and presumably corresponds to partially glycosylated CD98 proteins that failed to complete the complex process of N-glycosylation in the ER–Golgi apparatus. This is consistent with the results obtained when we treated BeWo cells for 24 h with tunicamycin, an antibiotic that inhibits the first steps of N-linked glycosylation and blocks the formation of new N-glycosidic protein–carbohydrate linkages. Under reducing conditions, in the absence of forskolin, an extra band of lower molecular mass (∼ 53 kDa) was detected in these lysates (3, 2). After forskolin treatment, known to stimulate CD98 expression, in addition to the 53 kDa band, a tight band running at approximately 49 kDa was clearly identifiable. This, we suggest, corresponds to the fully unglycosylated CD98 molecule, and is compatible with the theoretical 30 kDa mobility shift that is predicted based on the four potential extracellular N-glycosylation sites. The number of bands seen in 3, 2 must reflect the total population of immunoreactive CD98 molecules after 24 h of tunicamycin treatment; these molecules normally will only be present transiently, and thus the duration of glycosidase inhibition will determine the precise pattern observed.

    Details are in the caption following the image

    Detection of partially glycosylated and unglycosylated CD98 after cell treatment with BFA and tunicamycin. Western blotting under reducing (A) or non reducing (B) conditions on a 10% Bis/Tris NuPage gel: BeWo cells at 50–60% confluence were treated with dimethylsulfoxide (vehicle control, – Forskolin) or with 100 µm forskolin (+ Forskolin) for 24 h at 37 °C. Immunoblots were incubated with rabbit anti-(human CD98) (1 : 200), horseradish peroxidase-conjucated goat anti-rabbit IgG and DAB. A novel band (arrow) of ∼ 64 kDa (A1) or ∼ 80 kDa (B) was present in whole cell lysates from BFA-treated cells in both control and forskolin-stimulated cells, presumably a partially glycosylated form of CD98. A band of lower molecular mass (∼ 53 kDa) was present in both control and forskolin-stimulated cell lysates after tunicamycin treatment (A2), with an additional band of ∼ 49 kDa after forskolin treatment. Single representative blots from two experiments run in duplicate.

    Cell fusion decreases after pulse treatment with BFA for 4 h

    To investigate the relationship between the changes observed in CD98 expression and structure after BFA treatment of BeWo cells with functional alterations, we used two-color FACS to quantify cellular fusion.

    We used 3,3′-dioctadecyloxacarbocyanine perchlorate (DiO) cell-labeling solution, a lipophilic tracer that is weakly fluorescent in water but highly fluorescent and quite photostable when incorporated into membranes, and Mitotracker deep red 633, a cell-permeant mitochondrion-selective dye, to uniformly label suspended BeWo cells as previously described [18]. Briefly, flow cytometry analysis of a 50 : 50 mixed cell population from cells stained either with DiO or Mitotracker red and then cultured together allows us to quantify cellular fusion/stable aggregation, which is represented by double-positive cells. To better evaluate the effects of BFA treatment on cell fusion in this group of experiments, BFA was added for 4 h in the middle of a 24–26 h culture to cells incubated with or without 100 µm forskolin. The medium was then changed back to dimethylsulfoxide or forskolin alone for the remaining culture time. We note that, inevitably, the magnitude of the effect will be reduced by the diverse cellular stages (proliferation, aggregation, fusion, etc.) of the BeWo cell population in the window of the pulse of BFA. Two-color FACS analysis showed a decrease in cellular fusion after pulse BFA treatment as compared to both groups of BFA-untreated cells (P = 0.048, one-way ANOVA) (Fig. 4). However, when we measured CD98 surface expression in control or forskolin-stimulated cells, we found that this was still increased in the presence of BFA (data not shown).

    Details are in the caption following the image

    Cell fusion decreases after pulse treatment with BFA for 4 h. Double-color flow cytometry assays for detection of fused (double-positive) cells. BeWo cells were prestained with DiO (maximum emission 501 nm) or Mitotracker deep red 633 (maximum emission 665 nm) dye. Single-color cells or a 50 : 50 mixture of both cells were then cultured for 12–14 h in medium containing dimethylsulfoxide or 100 µm forskolin (Forsk) at 37 °C, and BFA (final concentration 5 µg·mL−1) was then added to the medium for 4 h. After that, the medium was replaced with fresh medium containing dimethylsulfoxide or 100 µm forskolin, and cells were cultured for a further 8–10 h. The graph shows data normalized to dimethylsulfoxide control (dimethylsulfoxide = 10%); n = number of cell samples. Statistical analysis: one-way anova, P-value 0.048.

    Galectin 3 and CD98 co-immunoprecipitate

    We have previously postulated a role for galectin 3, an S-type lectin containing a carbohydrate-binding domain, as a physiological ligand of CD98 in vivo[18].

    Figure 5A shows the primary intracellular distribution of galectin 3 in the cytoplasm and nucleus of BeWo cells, determined by indirect immunofluorescence. To investigate the possible ligand-binding role of galectin 3 in relation to CD98, BeWo cell lysates were incubated with a goat polyclonal antibody against CD98 or a goat polyclonal or mouse monoclonal antibody against galectin 3. Original whole cell lysates and immunoprecipitates were then subjected to SDS/PAGE and western blotting (Fig. 5B), and cut into single strips as described in Experimental procedures. Whole lysates of BeWo cells (lanes 2 and 7) were probed with rabbit anti-(human CD98) (lane 2) or with goat anti-(human galectin 3) (lane 7). Goat anti-(human CD98) immunoprecipitates (lanes 3–6) were probed with rabbit anti-(human CD98) (lane 3), goat anti-(human galectin 3) (lane 4), mouse anti-(phosphatidylinositol 3-kinase) (lane 5, as irrelevant control antibody), and rabbit IgG (lane 6, as negative control). Goat anti-(human galectin 3) immunoprecipitates (lanes 8, 9 and 11) were probed with mouse anti-(human galectin 3) (lane 8), rabbit anti-(human CD98) (lane 9), and rabbit IgG (lane 11). Mouse anti-(human galectin 3) immunoprecipitate (lane 10) was probed with rabbit anti-(human CD98). The results clearly demonstrate galectin 3 and CD98 co-immunoprecipitation in BeWo cell extracts. A band equivalent to the molecular mass of galectin 3 monomers (∼ 28–30 kDa) was present in CD98 immunoprecipitates (with an additional band of higher molecular mass, probably corresponding to galectin 3 dimers). A band equivalent to the molecular mass of CD98 (∼ 80 kDa) was present in the reverse immunoprecipitation experiment, whether or not the immunoprecipitates were prepared using the goat or the mouse anti-(human galectin 3).

    Details are in the caption following the image

    Galectin 3 is detected in BeWo cells and co-immunoprecipitates with CD98. (A) Immunofluorescence: galectin 3 (fluorescein isothiocyanate) primary distribution in the cytoplasm and nucleus of BeWo cells; nuclei are stained with DAPI. (B) Co-immunoprecipitation: western blotting of a 10% Bis/Tris NuPage gel run under reducing conditions with Mes running buffer. BeWo cell original total lysate (lanes 2 and 7) was probed with rabbit anti-(human CD98) (lane 2) or with goat anti-(human galectin 3) (lane 7). Goat anti-(human CD98) immunoprecipitates (lanes 3–6) were probed with rabbit anti-(human CD98) (lane 3), goat anti-(human galectin 3) (lane 4), mouse anti-(human PI3Kinase) (lane 5, irrelevant control antibody) and rabbit IgG (lane 6). Goat anti-(human galectin 3) immunoprecipitates (lanes 8, 9 and 11) were probed with mouse anti-(human galectin 3) (lane 8), rabbit anti-(human CD98) (lane 9) and rabbit IgG (lane 11). Mouse anti-(human galectin 3) immunoprecipitate (lane 10) was probed with rabbit anti-(human CD98). Arrows indicate CD98 immunoreactivity (upper arrows) or galectin 3 immunoreactivity (lower arrows).

    CD98 and galectin 3 co-localize in the plasma membrane, cytoplasm and nucleus

    We have previously shown CD98 expression and distribution by immuno-electron microscopy (immuno-EM) [18]. In the current study, we performed immuno-EM of galectin 3. We found that galectin 3 was uniformly distributed in the cytoplasm and nucleus, even if it was scarce on the cellular membrane, in both the dimethylsulfoxide-treated and the forskolin-treated groups, although in the latter, sporadic clustering of immunoreactivity was observed (Fig. 6A). To further confirm the close localization of the galectin 3 and CD98 molecules, we used immuno-EM and a standard double gold technique: double immunoreactivity was determined using an appropriate secondary antibody−10 nm gold complex to detect anti-CD98 (smaller-diameter particles) and an appropriate secondary antibody−15 nm gold complex to detect anti-galectin 3 (larger-diameter particles). The electronmicrographs in Fig. 6B,C clearly show co-localization of these two molecules in the plasma membrane, in the cytoplasm and in the nucleus of forskolin-treated BeWo cells.

    Details are in the caption following the image

    Galectin 3 co-localizes with CD98. (A). Immuno-EM: electron micrograph of forskolin-treated cell (× 25 000) showing galectin 3 (arrows); note occasional clustering of gold particles. (B, C) Double labeling immuno-EM: electron micrographs of forskolin-treated cells showing co-localization of galectin 3 and CD98 at the plasma membrane (B,C), in the nucleus (B), and in the cytoplasm (C) (arrows). Sections were sequentially stained using as secondary antibodies anti-(goat IgG)−10 nm gold complex to detect anti-CD98 (smaller-diameter particles) and anti-(rabbit IgG)−15 nm gold complex to detect anti-galectin 3 (larger-diameter particles). Three representative fields; scale bars 200 nm.

    Inhibition of galectin 3 binding to membrane glycoproteins affects cellular fusion

    We then investigated whether the close proximity of CD98 and galectin 3 in several cellular locations was indicative of a functional association.

    Galectin 3, like most members of the galectin family, acts as a receptor for ligands containing poly(N-acetyl-lactosamine) sequences through the C-terminus CRD. We used the high affinity of galectin 3 for lactose (Lac) to inhibit binding between the glycosylated sites of CD98 and galectin 3 CDR domains, and measured its effect on cellular fusion. The cells were labeled either with DiO and Mitotracker Deep Red 633, or with DiO and 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindodicarbocyanine perchlorate (DiD), another lipophilic tracer with markedly red-shifted fluorescence excitation and emission spectra in the same range as Mitotracker Deep Red. We followed the protocol employed for two-color FACS after BFA treatment; here, however, pulsed incubation with BFA for 4 h was substituted by an equal incubation time with 50 mm Lac or, in some experiments, with 50 mm maltose, which has a much lower affinity for galectin 3. Interestingly, we observed a small but significant reduction in cell fusion in the presence of Lac (Fig. 7).

    Details are in the caption following the image

    Inhibition of galectin 3 crosslinking membrane glycoproteins affects cellular fusion. Double-color flow cytometry assays for detection of fused (double-positive) cells. BeWo cells were prestained with DiO (em. 501 nm) or DiD (em. 665 nm) dye. Single-color cells or a 50 : 50 mixture of both cells were then cultured for 12–14 h in medium containing dimethylsulfoxide or 100 µm forskolin (Forsk) at 37 °C, when Lac (final concentration 50 mm) or maltose (Malt, final concentration 50 mm) was added to the medium for 4 h. The medium was then replaced with fresh medium containing dimethylsulfoxide or 100 µm forskolin, and cells were cultured for a further 8–10 h. The graph shows data normalized to dimethylsulfoxide control (dimethylsulfoxide = 10%); n = number of cell samples. Statistical analysis: one-way anova, P-value 0.0148.

    Discussion

    Syncytial fusion is a rare event in cell biology. In humans, we typically find three syncytial tissues: syncytiotrophoblast, striated muscle fibers and chondro-osteoclast. Syncytiotrophoblast forms during implantation, and is then maintained at the villous maternal–fetal interface throughout pregnancy. A useful model of trophoblast syncytialization is the choriocarcinoma cell line BeWo; these cells are able to fuse, and fusion can be also enhanced by forskolin treatment.

    Recently, the use of the fungal metabolite BFA to cause Golgi breakdown showed that part of Golgi glycosylation enzymes recycle to the ER, whereas Golgi matrix proteins are retained in a set of cytoplasmic membranes; this has led to the suggestion that BFA disrupts a dynamic membrane-recycling pathway between the ER and cis/medial Golgi, effectively blocking membrane transport out of but not back to the ER [28]. However, both the dynamic interaction between ER and Golgi and the mechanism of action of BFA are still subjects of intense discussion.

    N-linked glycosylation of membrane proteins is acquired as a post-translational modification in the ER, and further processing takes place in the Golgi before the proteins reach the cell surface. CD98 has been previously shown to be involved, among its many other functions, in trophoblast fusion [17,18]. As glycosylation seems to be important for correct protein folding and for ligand–receptor interactions, and because CD98 is an N-glycosylated protein [29,30], in this study we investigated the effect of BFA on the expression and the function of CD98 and its direct and indirect effect on galectin 3, which binds glycosylated proteins through its CDR site [25].

    By analysing BeWo cells at different time points with SDS/PAGE, we showed that, as a consequence of forskolin treatment, there is a time-dependent increase in CD98 protein expression comparable to that of the CD98 mRNA previously observed [31]. Unexpectedly, we found that CD98 surface expression was also increased in a time-dependent manner in BeWo cells treated with BFA. This result implied the existence, for CD98, of an alternative route to the plasma membrane that is independent of the classic secretory pathway through the trans-Golgi apparatus, which is used by most secretory and transmembrane proteins and can be inhibited by BFA. Furthermore, analysis of western blots probed with CD98 antibody, under reducing and non reducing conditions, showed the presence of an additional band in the BFA-treated cell lysates; this band presumably corresponded to a partially glycosylated form of CD98, after breakdown of the Golgi apparatus, in the last 4 h of culture. Taken together, these two findings would suggest that a partially glycosylated form of CD98 is capable of reaching and inserting into the cell membrane via an unknown mechanism of transport that is independent of the ER–trans-Golgi pathway.

    We next investigated whether CD98 glycosylation was necessary for its role in cellular fusion. For this purpose, it was important to add BFA to the culture when the cells were just starting to fuse; previous experiments had indicated that this occurs after 12–14 h. Moreover, we had found that BeWo cells undergo morphologic changes followed by detachment and death if cultured with BFA for over 6–8 h (data not shown). However, BFA effects are reversible if the drug is removed. We decided, therefore, to add BFA for 4 h in the middle of the culture time, to remove it, and then to observe the number of cells that underwent fusion as compared to untreated cells. This could be quantified by two-color FACS of BeWo cells previously labeled with one of two well-separated fluorescent dyes and then calculation of the number of double-fluorescent cells [18]. The results showed that glycosylation of CD98 is important for the fusion of BeWo cells as, although the molecule was still overexpressed on the surface of BFA-treated cells (data not shown), cellular fusion was decreased as compared to untreated cells. However, as anticipated, the magnitude of the observed effect was moderate.

    It has been suggested that galectin 3 is an endogenous crosslinker for CD98 and may promote CD98 dimerization (and consequent integrin activation) [21]. We have shown in BeWo cells, both by immunofluorescence and by immuno-EM, that galectin 3 is expressed in all three cellular compartments. Immunoprecipitation of BeWo cell total lysates with either goat anti-(human CD98) or goat or mouse anti-(human galectin 3) has also shown that CD98 and galectin 3 co-immunoprecipitate. Consequently, we used immuno-EM to confirm the relative positions of galectin 3 and CD98 in the cells. We showed unambiguous co-localization of the two molecules, both intracellularly, in the nucleus and cytoplasm, and at the plasma membrane. The relative abundance of CD98 molecules as compared to that of galectin 3 molecules at the same location supports the hypothesis of dimerization of CD98 molecules through linking with either monomeric or oligomeric forms of galectin 3. In the latter case, galectin 3 would have several CDR sites and be able to interact with many CD98 molecules.

    If there is an association between CD98 and galectin 3, then disturbing it should disrupt cellular fusion. Indeed, by blocking galectin 3 CDR sites with 50 mm Lac, we showed that we could reduce the fusion of BeWo cells.

    Getting proteins to the correct place at the right time is a logistical challenge for any cell. Proteins destined for the classic secretory pathway, such as immunoglobulins, typically contain N-terminal signal peptides that mediate membrane translocation into the lumen of the ER followed by ER–Golgi-dependent transport to the cell surface. On the other hand, a growing number of proteins (angiogenic growth factors, galectins, inflammatory cytokines, viral proteins) lack a signal peptide but are still secreted from the cell. These proteins do not contain modifications such as glycosylation (which happen at the ER–Golgi level), and their secretion is not inhibited by BFA or similar inhibitors of the classic secretory pathway. In recent years, several distinct ‘nonclassic’ secretory pathways have been demonstrated [32].

    As any morphologic and functional modification of the ER–Golgi–trans-Golgi complex would affect the proteins using this pathway, both structurally (incomplete or null secondary modifications) and functionally (as a result of the failure to reach correct cell locations), in this study we used BFA to investigate CD98 function and protein interactions.

    Our data suggest that CD98 can traffic to the plasma membrane via at least two distinct transport mechanisms in BeWo cells, one dependent upon the classic secretory pathway (glycosylated protein), and the other on an alternative route (nonglycosylated protein). Furthermore, we demonstrate that CD98 glycosylation is necessary for cell fusion and that this in turn requires interaction between CD98 and galectin 3. This lectin, like CD98, is present both in the cytotrophoblasts and in the syncytiotrophoblast [33,34]. Hence, crosslinking of these two molecules in vivo could be an essential molecular mechanism to enable syncytiotrophoblast formation. Our findings now need to be investigated in the intact placenta, e.g. by looking for co-localization of these two molecules in normal placental tissue and in primary cell lines.

    The results reported in this article fit unexpectedly with a recent study on CD98hc knockout mice that suggested an essential role for CD98 in early mouse development. Embryonic lethality was found when the transgene encoding the molecule was truncated at the extracellular domain, leaving intact both the intracellular and the transmembrane parts of the molecule [35]. Our work emphasizes the way in which the external domain of CD98 may play a critical role in trophoblast cell biology.

    Experimental procedures

    Primary antibodies

    Rabbit anti-(human galectin 3) was obtained from Chemicon Europe Ltd (Chandlers Ford, UK). Goat anti-(human CD98) (C-20), rabbit anti-(human CD98) (H-300), normal goat IgG and normal rabbit IgG (isotype-matched controls), goat anti-(human galectin 3) (D-20) and mouse anti-(human galectin 3) (B-2) were obtained from Santa Cruz Biotechnology Inc. (Santa Cruz, CA, USA). Mouse anti-(human PI3Kinase p85α) was obtained from Serotec (Kidlington, UK).

    Secondary antibodies

    Horseradish peroxidase-conjugated goat anti-(rabbit IgG), horseradish peroxidase-conjugated rabbit anti-(mouse IgG) and fluorescein isothiocyanate-conjugated swine anti-(rabbit IgG) were obtained from Dako (Glostrup, Denmark). Horseradish peroxidase-conjugated donkey anti-(goat IgG) and protein A/G plus agarose were obtained from Santa Cruz Biotechnology Inc. Fluorescein isothiocyanate-conjugated rabbit anti-(goat IgG) was obtained from Sigma (Gillingham, UK). Rabbit anti-(goat IgG)−10 nm gold complex to detect anti-CD98 and goat anti-(rabbit IgG)−15 nm gold complex to detect anti-galectin 3 were obtained from British Biocell (Cardiff, UK).

    Cell culture

    BeWo cells were cultured at 37 °C as monolayers in F-12K Nutrient Mixture (Kaighn's modification) supplemented with 10% fetal bovine serum, 2 mm l-glutamine (all Gibco, Paisley, UK), 100 U·mL−1 penicillin and 100 U·mL−1 streptomycin (Sigma) in a humidified atmosphere of 5% CO2 and 95% air. Confluent cells were harvested by trypsinization with trypsin/EDTA in HBSS without Ca2+ and Mg2+ (Gibco), resuspended in fresh medium, and plated in six-well culture plates (BD Falcon, Oxford, UK). When the cells reached 65–70% confluence, forskolin (Sigma) or vehicle (dimethylsulfoxide) was added in fresh medium at a final concentration of 100 µm for 24 h, unless otherwise indicated. In some wells, BFA (Sigma) (final concentration 5 µg·mL−1), tunicamycin (Sigma) (final concentration 10 µg·mL−1), 50 mm Lac or 50 mm maltose were added as indicated in the different experiments. For the two-color FACS experiments, before plating, viable cells were counted by the trypan blue (Sigma) method, resuspended in serum-free medium, and stained with either 10 µL of vybrant DiO or 5 µL of vybrant DiD cell labeling solutions (1 mm) (Molecular Probes, Invitrogen, Paisley, UK) per 106 cells·mL−1 cells for 30 min, or with MitoTracker Deep Red633 (Molecular Probes) at a concentration of 25 nm per 106 cells·mL−1 cells for 15 min; labeling was carried out at 37 °C in the dark with gentle shaking. After extensive washing with warm serum-free medium, each group of stained cells was resupended in complete growth medium and plated either on its own or in a 50 : 50 mixture (DiO-labeled and Mitotracker Red-labeled or DiO-labeled and DiD-labeled cells) in six-well culture plates.

    SDS/PAGE and western blotting

    Confluent cultures from six-well plates were washed with ice-cold Ca2+-free and Mg2+-free Dulbecco's phosphate-buffered saline (D-NaCl/Pi) (Gibco) and then lysed at 4 °C in ice-cold modified RIPA buffer containing 50 mm Tris/HCl (pH 7.4), 1% NP-40, 0.25% sodium deoxycholate, 150 mm NaCl, 1 mm EDTA, and 10 µL of protease inhibitor mixture (Sigma), for 15 min on a rocker. Samples were sonicated three times for 30 s, and clarified by centrifugation at 10 000 g for 15 min at 4 °C (Beckman GS15-R, rotor F402, Beckman Coulter Ltd., High Wycombe, UK). Supernatants (10 µg of protein) were retained, solubilized in NuPAGE sample buffer (Invitrogen), with or without reducing agent, warmed for 10 min at 75 °C, and then run on 10% Novex Bis/Tris NuPAGE gels (Invitrogen). The proteins were transferred to nitrocellulose membranes, blocked using 5% (w/v) nonfat dry milk in 0.01 m NaCl/Pi (Sigma) with 0.05% (v/v) Tween 20 for 1 h at room temperature, and then incubated with rabbit anti-(human CD98) (H-300, 1 : 200) overnight at 4 °C. Horseradish peroxidase-conjugated goat anti-(rabbit IgG) was used for secondary labeling. Immunoreactive bands were identified by SIGMAFAST 3,3′-diaminobenzidine tablets (Sigma) according to the manufacturer's instructions.

    Co-immunoprecipitation

    Whole cell lysates were prepared as described above. Aliquots (10 µg of protein) were retained and solubilized in NuPAGE sample buffer (Invitrogen) for analysis by western blotting. The remaining lysates were precleared with 1 µg of goat IgG and 10 µL of protein A/G plus agarose [for goat anti-(human CD98) and goat anti-(human galectin 3) immunoprecipitates] or with 10 µL of protein A/G plus agarose [for mouse anti-(human galectin 3)] for 10 min at 4 °C; the agarose beads were removed by centrifugation at 4000 g (Beckman GS15-R, rotor F4202), and the cleared lysates were incubated overnight with 2 µg of goat anti-(human CD98) or goat anti-(human galectin 3) or mouse anti-(human galectin 3) at 4 °C. Immune complexes were captured using 20 µL of protein A/G agarose beads for 2 h at 4 °C, and then washed three times with lysis buffer. Following elution with NuPAGE buffer, samples were boiled for 5 min to dissociate beads from the immunocomplexes, and centrifuged at 100 000 g (Eppendorf 5415C, Eppendorf UK Ltd., Cambridge, UK); associated proteins in the supernatants were resolved on a 10% Novex Bis/Tris NuPAGE gel (Invitrogen) under reducing condition with Mes running buffer (Invitrogen). The proteins were transferred to a nitrocellulose membrane, and this was cut into strips corresponding to single protein lanes (revealed after reversibly staining with Ponceau red). Single strips were blocked using either 5% (w/v) nonfat dry milk in 0.01 m NaCl/Pi (Sigma) with 0.05% (v/v) Tween 20 (for strips to be probed with anti-CD98), or 0.01 m NaCl/Pi (Sigma) with 0.05% (v/v) Tween 20 plus 10% normal serum of the host species of the secondary antibody for 1 h at room temperature, and then incubated with either rabbit anti-(human CD98) (H-300, 1 : 200), goat anti-(human galectin 3) (1 : 100), mouse anti-(human galectin 3) (1 : 100), mouse anti-(human PI3 kinase) (1 : 100) as irrelevant control antibody, normal rabbit IgG or normal goat IgG (1 : 100) overnight at 4 °C. Horseradish peroxidase-conjugated goat anti-(rabbit IgG), horseradish peroxidase-conjugated donkey anti-(goat IgG) or horseradish peroxidase-conjugated rabbit anti-(mouse IgG) were used for secondary labeling. Immunoreactive bands were identified by SIGMAFAST 3,3′-diaminobenzidine tablets (Sigma) according to the manufacturer's instructions.

    Flow cytometry − surface staining on intact cells

    Cells from six-well plates were detached with trypsin/EDTA (Gibco). Aliquots of 1 × 106 cells were washed in NaCl/Pi and resuspended in 250 µL of FACS buffer (NaCl/Pi, 1% fetal bovine serum, 0.1% NaN3) with goat anti-(human CD98) (C-20, 1 : 20) or isotype control IgG or no primary antibody. Cells were incubated for 45 min on ice, and then washed three times with FACS buffer. Samples were then incubated with fluorescein isothiocyanate-conjugated rabbit anti-(goat IgG) (1 : 50) for 45 min on ice and washed three times. Samples were finally resuspended in FACS buffer and 2% paraformaldehyde (PFA), and the number of events was analyzed by flow cytometry using a FACSCalibur (BD Biosciences, Oxford, UK) flow cytometer and cell quest software and/or an EPICS Altra (Beckman Coulter Ltd., High Wycombe, UK) flow cytometer and expo32 software.

    Flow cytometry − surface and intracellular staining

    Cell suspensions were fixed in 2% PFA for 20 min at room temperature, washed once in NaCl/Pi, permeabilized with 1% saponin in FACS buffer for 15 min at room temperature, and then stained following the surface staining protocol. After the final wash, samples were fixed again in 2% PFA before analysis.

    Immunofluorescence

    Cells (1 × 103) were plated onto chamber wells (Lab-Tek, Fisher Scientific UK Ltd., Loughborough, UK), grown for 24 h, washed with NaCl/Pi, and fixed with 2% PFA and rinsed. Nonspecific binding sites were blocked with blocking buffer (NaCl/Pi, 0.05% Tween 20, 10% fetal bovine serum, 10% goat serum) for 20 min at room temperature. Cells were then incubated with rabbit anti-(human galectin 3), 1 : 1000 in diluting buffer (NaCl/Pi, 0.05% Tween 20, 1% fetal bovine serum, 1% goat serum) for 1 h at room temperature, washed three times for 5 min, and then incubated with fluorescein isothiocyanate-conjugated rabbit anti-(goat IgG). After three more washes, chambers were removed, and slides mounted with ProLong Gold antifade reagent with 4′,6-diamidino-2-phenylindole (DAPI) (Molecular Probes). Images were captured with a Leica DC 500 digital camera on a Leica DMR microscope (Leica Microsystem Digital Imaging, Cambridge, UK).

    Immunogold EM

    Cells were prepared for EM by standard methods [36]. Briefly, cell pellets were postfixed in osmium tetroxide (1% w/v in 0.1 m sodium phosphate buffer), contrasted with uranyl acetate (2% w/v in distilled water), dehydrated through increasing concentrations of ethanol (70–100%), and embedded in LR Gold resin (Agar Scientific, Reading, UK). Ultrathin sections (50–80 nm) were prepared by use of a Reichert Ultracut S microtome (Reichert, Vienna, Austria), and mounted on 200-mesh nickel grids. For immunogold detection of CD98 and galectin 3, sections were incubated with either goat anti-(human CD98) (1 : 100) or rabbit anti-(human galectin 3) (1 : 100) for 2 h and for 1 h with protein A−15 nm gold complex. For control sections, the primary antibody was omitted and replaced with a matching dilution of the respective nonimmune serum. Sections were then lightly counterstained with uranyl acetate and lead citrate. All antibodies were diluted in 0.1 m phosphate buffer containing 1% w/v egg albumin. The sections were viewed with a JEOL 1010 transmission electron microscope (JEOL, Peabody, MA, USA), and representative micrographs were prepared. The area of each cellular compartment of interest was determined by point counting morphometry, and the number of gold particles over each compartment was counted. The density of immunogold (particles per µm2) was then calculated. For double labeling of CD98 and galectin 3, sections were sequentially stained as above using rabbit anti-(goat IgG)−10 nm gold complex to detect anti-CD98 and goat anti-(rabbit IgG)−15 nm gold complex to detect anti-galectin 3.

    Statistical analysis

    Results are presented as means ± SE. The significance of the differences between means was assessed using the two-tailed Student's t-test or one-way anova. P values < 0.05 were considered to be significant.

    Acknowledgements

    We are grateful to Dr Paul Klenerman (Peter Medawar Building for Pathogen Research, Oxford University) for assistance with the flow cytometry, and we thank Lynne Scott for expert technical help with the immuno-EM. This work was funded by the Wellcome Trust.