B-cell maturation antigen is modified by a single N-glycan chain that modulates ligand binding and surface retention

Abstract

Glycosylation, an important posttranslational modification process, can modulate the structure and function of proteins, but its effect on the properties of plasma cells is largely unknown. In this study, we identified a panel of glycoproteins by click reaction with alkynyl sugar analogs in plasma cells coupled with mass spectrometry analysis. The B-cell maturation antigen (BCMA), an essential membrane protein for maintaining the survival of plasma cells, was identified as a glycoprotein exhibiting complex-type N-glycans at a single N-glycosylation site, asparagine 42. We then investigated the effect of N-glycosylation on the function of BCMA and found that the dexamethasone-induced apoptosis in malignant plasma cells can be rescued by treatment with BCMA ligands, such as a proliferation-inducing ligand (APRIL) and B-cell–activating factor (BAFF), whereas removal of terminal sialic acid on plasma cells further potentiated the ligand-mediated protection. This effect is associated with the increased surface retention of BCMA, leading to its elevated level on cell surface. In addition, the α1–3,-4 fucosylation, but not the terminal sialylation, assists the binding of BCMA with ligands in an in vitro binding assay. Together, our results highlight the importance of N-glycosylation on BCMA in the regulation of ligand binding and functions of plasma cells.

B-cell maturation antigen (BCMA), classified as tumor necrosis factor receptor superfamily 17 (TNFRSF 17), is primarily expressed on the surface of plasma cells, but is absent on naive B cells and on most memory B cells (1, 2). BCMA can be induced by stimulation with cytokines in the peripheral blood B cells (1, 2) and is important for maintaining the survival of long-lived plasma cells in bone marrow (3). Upon stimulation with a proliferation-inducing ligand (APRIL) or B-cell–activating factor (BAFF) (4, 5), BCMA becomes trimerized, subsequently eliciting a signaling cascade involved in the activation of MAP kinases and the induction of anti-apoptotic proteins, such as Bcl-2 and Bcl-XL (4–7). BCMA is also implicated in the regulation of antigen presentation activity of activated B cells (2) and in the survival, proliferation, and differentiation of adipocytes (8, 9). The expression of BCMA is not restricted to normal tissues. Some cancer cells, such as glioblastoma (10), multiple myeloma (MM) (11), chronic lymphocytic leukemia (12, 13), and Hodgkin lymphoma (14), express BCMA, and blocking its interaction with BCMA ligands is known to reduce cellular survival (10–12, 14, 15). In cells of autoimmune diseases, such as the fibroblast-like synoviocytes of rheumatoid arthritis patients (16) and the plasmablasts from systemic lupus erythematosus (SLE) patients (17), the expression of BCMA is elevated. Particularly, the expression of BCMA on the autoantibody-producing cells is selectively increased in SLE patients (18). Therefore, disruption of the interaction between BCMA and its ligands by either receptor-immunoglobilin fragment crystallizable region (Fc) chimera or antibody was proposed as a therapeutic strategy for the management of malignancies (10, 13, 19–21) and autoimmune diseases (22, 23).

Glycosylation is a common modification process to modulate cell membrane proteins and lipids (24). In addition, the retention of proteins on the cell surface is also regulated by glycosylation, which has potential implications in regulating cellular response to ligand-mediated stimulation (25). In this study, BCMA was identified as a glycoprotein with a single N-glycosylation site in plasma cell lines, and its glycosylation, especially the sialylation, was shown to modulate the function of BCMA.

Results

Use of Sugar Probes in Plasma Cell Lines for Identification of Glycoproteins.

We first used our designed alkynyl sugar probes as substrates for fucosylation and sialylation in cell cultures followed by the triazole-forming click reaction using an azido-biotin analog (26) to identify the labeled glycoproteins of plasma cells (Fig. 1). We first investigated if the MM cell lines, including H929, RPMI8226, and U266, the malignant counterpart of plasma cells, were incorporated with the sugar analogs. Indeed, we were able to detect the fluorescent signal in cells fed with alkyne sugars, but not the control, after performing the click reaction with azido-biotin, followed by staining with alexa-488–conjugated streptavidin (Fig. S1A). Furthermore, immunoblot analysis showed that, after azido-biotin labeling and streptavidin-HRP blotting, the proteins extracted from the control had only some nonspecific background luminescence, but those extracted from cells cultured with the probes showed strong luminescence, indicating that the alkyne sugar analogs were efficiently incorporated into proteins (Fig. S1B). The cell surface glycoproteins labeled by the alkyne sugar probes were also confirmed by flow cytometry analysis (Fig. S1C). Because the sugar analogs were incorporated into proteins by plasma cell lines, we identified the glycoproteins that labeled by the sugar analogs with mass spectrometry analysis. The list of proteins labeled by the alkynyl fucose and alkynyl N-acetylmannosamine (ManNAc) are shown in Table S1. However, the functions of most of the identified proteins in plasma cells were unclear. Among these proteins, we were particularly interested in TNFRSF 17 (International Protein Index: IPI00293877), also known as BCMA, because it is essential for normal and malignant plasma cell survival (8, 10–15).

Fig. 1.

Flowchart of identification of glycoproteins. Cell extracts from sugar probe-fed cells were subjected to click reaction with an azido-biotin analog. Products of the click reaction were purified by streptavidin, followed by trypsin digestion. Sugar probe-labeled glycopeptides were released after PNGase F treatment and then analyzed by mass spectrometry.

BCMA Is a Glycoprotein with a Single N-Glycosylation Site.

Given that a previous study reported that BCMA was not subjected to N-glycosylation in the MM cell lines (27) and that, according to the protein database, UniProt (www.uniprot.org), BCMA is a nonglycosylated protein, we sought to further investigate our finding. Purified BCMA from H929 cells was subjected to mass spectrometry analysis to confirm the presence of glycans. A trypsin-digested fragment, YCNASVTNSVK (amino acid 40–50), was identified as the sole glycopeptide of BCMA. Consistent with the result of incorporation of sugar probes, both fucose and sialic acid were present in the identified complex-type N-glycans that contain abundant biantennary structure (Fig. 2A). In addition, high-mannose type N-glycans were also found in BCMA (Fig. 2A). We then investigated if BCMA contains N-glycans by peptide-N-glycosidase F (PNGase F) treatment to remove the N-glycans from proteins and found that, before PNGase F treatment, two or multiple bands were immunoreactive to the anti-BCMA antibody in cell lysates from H929, U266, and RPMI8226 cells (Fig. 2B). After PNGase F treatment, only one band with the predicted molecular weight (20.165 kDa) was detected in these three cell lines (Fig. 2B). To further confirm this result, we transfected 293T cells with a cDNA encoding the FLAG-tagged BCMA. Consistently, the protein bands immunoreactive to the anti-FLAG antibody in the cell lysates from the transfected cells were sensitive to PNGase F treatment (Fig. 2C). These findings suggested that BCMA contains N-glycans.

Fig. 2.

BCMA is a glycoprotein. (A) Glycans identified on BCMA from H929 cells by glycopeptide profiling. Percentages represent the glycan type identified from the analysis. (triangle, Fuc; diamond, NeuAc; square, HexNAc; circle, Hex) (B) PNGase F treatment caused mobility shift of BCMA detected by immunoblotting. Cell extracts from H929, RPMI8226, and U266 cells were incubated with PNGase F, followed by immunoblotting with anti-BCMA antibody. (C) N42 of BCMA is the N-glycosylation site. Cell extracts from 293T cells transfected with WT or N42A BCMA-expressing vector were treated without or with PNGase F and then subjected to immunoblotting with the indicated antibodies.

Because the trypsin-digested glycopeptide of BMCA contains the N-glycosylation consensus sequence Asn-Ala-Ser (NAS) (Fig. 2C), we suspected that the asparagine (N) residue at amino acid 42 is likely the N-glycan site. The cDNA encoding the mutated BCMA with the 42nd amino acid residue changed to Ala (N42A BCMA) was generated for expression. Immunoblot results with anti-FLAG antibody show that only one band could be detected in the extract from 293T cells transfected with the N42A BCMA-expressing vector, that its molecular weight was identical to the lower band of wild type (WT) BCMA (Fig. 2C), and that this single protein band was resistant to PNGase F treatment (Fig. 2C). These data demonstrated that BCMA is an N-glycan modified protein and N42 is the sole N-glycosylation site.

Sialylation at the Terminal N-Glycans of BCMA.

Given that the cell surface of murine plasma cells displayed increased sialylation compared with the stimulated murine mature B cells (28) and that BCMA could be labeled by ManNAcyne, we next examined the sialosides linkage of BCMA on MM cells. Sambucus nigra lectin (SNA), which preferentially binds to the α2–6-linked sialosides, and Maackia amurensis lectin (MAL), which selectively binds to the α2–3-linked sialosides, can bind to H929 and RPMI8226 cells as determined by flow cytometry analysis, but the binding was significantly reduced on cells pretreated with sialidase, which hydrolyzes the α2–3- and α2–6-linked sialosides (Fig. 3 A and B). We next examined if these sialoside linkages occur on BCMA. A solution of purified FLAG-tagged BCMA isolated from RPMI8226 or H929 transfectants was added into anti-FLAG–coated plates, followed by incubation with SNA or MAL. The results in Fig. 3 C and D indicated that both lectins recognized intact BCMA better than the sialidase-treated BCMA. Reciprocally, we preadded lectins onto streptavidin-coated plates before incubation with BCMA, and the result showed that intact BCMA bound better to SNA or MAL than the sialidase-treated BCMA did (Fig. 3 E and F). These data suggest that the N-glycans of BCMA are composed of both α2–3- and α2–6-linked sialosides at the glycan terminal.

Fig. 3.

Glycans on BCMA are terminally modified by sialic acid. (A and B) FACS shows the binding with SNA (A) or MAL (B) on H929 and RPMI8226 cells pretreated with or without sialidase. (C and D) ELISA shows the binding of lectins with BCMA. Full-length BCMA purified from transfectants was pretreated with or without sialidase and added to anti-FLAG antibody coated plates, followed by the biotinylated SNA (C) or MAL (D). (E and F) ELISA shows the binding of BCMA to lectins. Biotinylated SNA (E) or MAL (F) was bound to streptavidin-coated plates before mixing with sialidase-treated or untreated full-length BCMA. Results in A and B are representative of three or four independent experiments, and the number in the histogram indicates the mean of fluorescence. Results in C−F represent mean ± SEM of three or four independent experiments. *P < 0.05 and **P < 0.01.

Sialylation Suppresses the Prosurvival Activity of BCMA Ligands.

Because glycosylation can be involved in the regulation of protein function (24), and the major function of BCMA is to promote cell survival (3), we sought to study whether sialylation participates in BCMA-mediated cell survival in the protection of apoptosis induced by dexamethasone (DEX). DEX, a glucocorticoid analog, is a therapeutic drug used to treat patients suffering from MM. MM cells undergo apoptosis following DEX treatment, which can be protected by the treatment with the ligands of BCMA (29). In the basal state, RPMI8226 cells treated with or without sialidase have a similar number of apoptotic cells (Fig. 4A). Although DEX caused more apoptosis in sialidase-treated cells than in sialidase-untreated cells, the difference is not statistically significant (Fig. 4B). Indeed, the BCMA ligand APRIL can partially protect sialidase-untreated and -treated cells from DEX-induced apoptosis as previously reported (29) (Fig. 4A), but the extent of rescue is not significant between these two groups (Fig. 4C). It is noted that another BCMA ligand, BAFF, can protect sialidase-treated cells from DEX-induced apoptosis more effectively (Fig. 4 A and B) and that the extent of rescue increased about 15% compared with sialidase-untreated cells (Fig. 4C), suggesting that sialylation on BCMA may reduce the survival effect of the BAFF-BCMA pathway. We next examined which sialylation—α2–3 or α2–6 sialylation—was important for this regulation. The results in Fig. S2 show that removal of α2–3 sialylation can further enhance the protective effect of BAFF on DEX-induced apoptosis, although the extent of rescue was only marginal (Fig. S2C). These results suggest that both α2–3 and α2–6 sialylation on BCMA may inhibit BAFF-mediated cell survival.

Fig. 4.

Sialylation affects BCMA ligand-mediated protection from apoptosis induced by DEX. RPMI8226 cells pretreated with or without sialidase were treated with DEX in the absence or presence of APRIL or BAFF. Three days later, cells were subjected to annexin V staining by FACS analysis. (A) One representative result of three independent experiments is shown; the number in the dot plot indicates the percentage of annexin V–positive cells. (B) The mean value ± SEM of three independent experiments from A. (C) Percentage of rescue of apoptosis as determined by (% of apoptosis caused by DEX–% of apoptosis induced by DEX in the presence of BCMA ligand)/(% of apoptosis induced by DEX). (D and E) Histograms of FACS analysis show that removal of sialylation increases the binding of ligands for BCMA with cell surface. H929 or RPMI8226 cells were pretreated with sialidase and then incubated with Fc-APRIL (D) or Fc-BAFF (E), followed by detection with FACS analysis. The number in the histogram indicates the mean of fluorescence. Bar graphs below D and E show statistical analysis of ligand binding after treatment of cells with sialidase in three independent experiments. Results are mean ± SEM. *P < 0.05 and **P < 0.01.

Removal of Sialylation Increases the Binding of Ligands to Plasma Cell Surface.

We next examined if sialylation modulates the binding between BCMA and its ligands. Compared with control cells, the binding of APRIL with RPM8226 and H929 cells was slightly increased after removal of sialic acid (Fig. 4D). It is noted that the binding of BAFF to RPMI8226 or H929 cell surface was enhanced by sialidase treatment and the enhancement was more significant than APRIL (Fig. 4E).

Sialylation Reduces the Retention of BCMA on the Cell Surface.

The enhancement of ligand binding to the cell surface after sialidase treatment may result from the increased levels of receptor on the cell surface. To test this possibility, we measured the receptor level of BCMA. Interestingly, compared with untreated control, treatment with sialidase resulted in an elevated amount of BCMA on the plasma cell surface (Fig. 5A). Likewise, H929 or RPMI8226 cells treated with α2–3 sialidase displayed an increased level of surface BCMA (Fig. S3A). This finding implied that both α2–3 and α2–6 sialylation may hamper the presence of BCMA on the cell surface. This elevated BCMA on the cell surface caused by treatment with sialidase was sustained for at least 24 h (Fig. S3 B and C). In addition, by treating the cells with cycloheximide (CHX) to inhibit the synthesis of new proteins, we were able to calculate the rate of BCMA accumulated on the cell surface after removing the terminal sialic acid. We found that treatment with CHX did not affect the accumulation of BCMA on RPMI8226 and H929 cells, but after sialidase treatment, the accumulation of BCMA was enhanced up to 160–180% (Fig. 5B). Consistently, treatment with α2–3 sialidase also promoted the retention of BCMA on the surface of RPMI8226 and H929 cells (Fig. S3D). Furthermore, we examined if removal of N-glycans by PNGase F can abrogate the effect of sialidase on promoting the surface retention of BCMA. Although PNGase F only partially removed surface N-glycans (Fig. S4), we were still able to observe that treatment with PNGase F abolished the enhanced retention of BCMA resulted from sialidase treatment (Fig. 5C). In addition, in the presence of CHX, treatment with sialidase cannot promote the surface retention of N42A BCMA (Fig. 5D and Fig. S5), indicating that sialylation on the N-glycans of BCMA at N42 reduces the retention of BCMA on the cell surface.

Fig. 5.

Sialylation influences surface level of BCMA. (A) Histograms show that removal of sialic acid results in the increase of surface BCMA. H929 or RPMI8226 cells treated with or without sialidase were subjected to FACS analysis with APC-conjugated anti-BCMA antibody. (B) Removal of sialic acid results in the accumulation of preexisting BCMA on the cell surface. H929 or RPMI8226 cells were treated with CHX in the absence or presence of sialidase, followed by FACS analysis with APC-conjugated anti-BCMA antibody. (Right) Relative level of BCMA on H929 and RPMI8226 cell surface following treatment with sialidase (C) Removal of N-glycans by PNGase F abolishes the effect of sialidase on surface BCMA. CHX-treated H929 or RPMI8226 cells were treated with or without PNGase F in the presence of sialidase. The level of preexisting BCMA on the cell surface was detected by APC-conjugated anti-BCMA antibody by FACS analysis. (D) The level of N42A BCMA expressed on RPMI8226 cells did not change after sialidase treatment. RPMI8226 cells transfected with mock control and N42A-BCMA expression vector were treated with sialidase in the presence of CHX, followed by FACS analysis with APC-conjugated anti-BCMA antibody. (A–D) Each histogram is based on data from three independent experiments; the number in the histogram indicates the mean of fluorescence. (Right) Statistical analysis of three independent experiments. Results are mean ± SEM. *P < 0.05 and **P < 0.01.

N-Glycan of BCMA Is Required for Ligand Binding.

Although the increased binding of ligand may result from the elevated level of receptors on the cell surface, it remains possible that the binding between BCMA and the ligand per se was enhanced in the presence or absence of certain types of glycans. Therefore, we examined if sialylation and other types of N-glycans modulate ligand–receptor interaction by sequential enzymatic digestion with glycosidases and an in vitro binding assay (Fig. 6A). We found that removal of sialic acid on the purified BCMA had no effect on the binding with APRIL and BAFF (Fig. 6 B and C, respectively). Given that the N42 residue of BCMA is close to the domain required for ligand binding (30), we sought to examine whether the N-glycans participate in ligand binding. Indeed, hydrolysis of the whole N-glycan by PNGase F reduced the binding of BCMA with exogenously added APRIL or BAFF compared with mock-treated BCMA (Fig. 6 D and E). Compared with wild-type BCMA, the N42A BCMA mutant also consistently exhibited decreased binding with APRIL and BAFF (Fig. 6 F and G, respectively). Following the consequence of sequential enzymatic digestion with various glycosidases, we found that digestion with sialidase and β-galactosidase to remove the galactose residue decreased the binding of APRIL and BAFF with BCMA (Fig. 6 B and C, respectively). Likewise, sequentially treating BCMA with N-acetylglucosaminidase and mannosidase further reduced its binding with ligands (Fig. 6 B and C). Fucosylation, normally attaching to the N-acetylglucosamine residue, may also affect ligand binding. Indeed, BCMA treated with l-fucosidase showed reduced ability to bind with APRIL and BAFF (Fig. 6 H and I, respectively). Together, these data indicate that the N-glycans, but not terminal sialylation, aid in the binding of BCMA with ligands in vitro.

Fig. 6.

N-glycans modulate the binding of BCMA with ligands. (A) Illustration of glycosidase treatment. (B and C) Sequential digestion of glycosidases reveals the glycan motifs required for binding with BCMA. Extracts from H929 and RPMI8226 transfectants expressing FLAG-tagged BCMA were prebound in 96-well plates coated with anti-FLAG antibody and then treated with glycosidases (as indicated in A) followed by adding Fc-APRIL (B) or Fc-BAFF (C). (D and E) PNGase F treatment impairs ligand binding to BCMA. Extracts from FLAG-tagged BCMA-expressed RPMI8226 transfectants were prebound by anti-FLAG antibody-coated plates and then treated with PNGase F, followed by addition of Fc-APRIL (D) or Fc-BAFF (E) to detect the binding of ligands by HRP-conjugated anti-human IgG-Fc antibody. (F and G) Binding of ligands with N42A BCMA was less than with wild-type BCMA. Extracts from 293T transfectants expressing either WT or N42A FLAG-tagged BCMA were prebound by anti-FLAG antibody in 96-well plates, followed by addition of Fc-APRIL (F) or Fc-BAFF (G). (H and I) Fucosylation is involved in ligand binding of BCMA. Extracts from H929 or RPMI8226 transfectants expressing exogenous FLAG-tagged BCMA were prebound in anti-FLAG antibody-coated 96-well plates and then treated with L-fucosidase followed by addition of Fc-APRIL (H) or Fc-BAFF (I). Statistical analysis of three or four independent experiments is shown by mean ± SEM. *P < 0.05, **P < 0.01, and ***P < 0.001.

Discussion

Here, we used sugar analogs coupled with mass spectrometry analysis (26) to identify the critical glycoproteins involved in plasma cell function and revealed that BCMA is an N-glycosylated protein. However, a previous study by Gras et al. (27) suggested that BCMA is not a glycoprotein in human MM cells. This discrepancy could be due to the use of different antibodies and protein extraction protocols. Another possibility is that the amount of glycan-modified BCMA is not sufficient to be noted in the previous report. The notion that BCMA is a glyocoprotein is further confirmed by our analysis of the glycan profiles and identification of the N-glycosylstion site of BCMA at the N42 residue. Moreover, disruption of the N-glycosylation site by site-directed mutagensis abolished the presence of the glycosylated form of BCMA.

De-sialylation by knockdown of sialylatransferase or treatment with sialidase enhanced the downstream signaling of epidermal growth factor receptor (EGFR) by promoting dimerization and internalization of EGFR, leading to the augmentation of cell proliferation or migration (31, 32). Apoptosis induced by the activation of the Fas death receptor is also repressed by sialylation through the reduction of receptor internalization that is required for transduction of apoptotic signals (33). Additionally, sialylation prevented the binding of galectin, a type of animal lectin, with CD45 or integrins modified with N-acetyl-d-lactosamine (LacNAc), thereby inducing apoptosis (34, 35). We also found that sialylation suppresses ligand-mediated cell survival by modulating surface retention of BCMA and that sialylation diminished the apoptosis of MM cells triggered by administration of the death ligand of the TNF-related apoptosis-inducing ligand (TRAIL) receptor (Fig. S2D). All these results support the notion that sialylation negatively regulates the function of protein substrates.

The glycans on the plasma cell surface could be involved in binding with lectins. Depletion of galectin-3, for example, reduced the surface expression of EGFR (36), and the retention of the renal epithelial calcium channel transient receptor potential cation channel subfamily V member 5 was increased by binding with galectin-1, leading to the decreased dynamin-dependent internalization (37). In our case, because a previous study showed that removal of sialic acids on the plasma cell surface enhanced the binding of galectins, such as galectin-1 and galectin-8, with the cell surface (28), we have thus hypothesized that the increased retention of BCMA on the cell surface following sialidase treatment may be due to the effect of galectins. However, we found that addition of LacNAc, the pan-inhibitor of galectins (38), did not influence the surface retention of BCMA on plasma cells regardless of the sialidase treatment (Fig. S6). Previous study also showed that the number of N-glycans and their degree of branching on membrane proteins affect the binding of galectin, which counteracts the constitutive endocytosis of membrane proteins (39). Because BCMA has only one N-glycan chain with a low percentage of tri- and tetra-antennary glycans, BCMA may not be targeted by galectins. Another possibility is that the sialic acid-binding Ig-like lectin (siglec), another type of animal lectin with the ability to recognize sialic acids (40), may be involved. Recent studies indicated that siglecs may act as endocytic receptors, such that ovalbumin or cytotoxin conjugated with antibody/sugar ligands specific for siglecs can be delivered into B cells, macrophages, or mouse embryonic fibroblasts (41–43). Especially in the case of CD22, which is expressed on B cells (44), the siglec level on the cell surface can be regulated by clathrin-mediated endocytosis and recycles between the cell surface and endosomal compartments (42). Hence, we assume that removal of sialylation on BCMA may disrupt the binding with siglecs, thereby preventing the endocytosis of BCMA along with siglecs and resulting in the increased retention of BCMA on the cell surface. However, further experiments are required to prove our hypothesis.

Our finding that the N-glycans on BCMA affect its binding to ligands is reflected by the fact that the N-glycosylation site is proximal to the ligand-binding domain of BCMA. The reduction of ligand binding is not due to the alternation of protein folding when the glycosylation site is disrupted because we have compared the secondary structure of N42A BCMA with WT BCMA by circular dichroism (CD) and have found that their overall structures were indistinguishable (Fig. S7A). We also compared the CD spectrum of PNGase F-treated BCMA with intact BCMA and found that the PNGase F treatment did not generally affect the secondary structure of BCMA (Fig. S7B). The possibility that the reduced binding to ligand is due to protein stability was excluded because BCMA with inner-core trisaccharides capable of maintaining protein stability (45) still showed reduced ligand binding. We also showed that fucosylation, but not sialylation, directly affects ligand binding and that different motifs of glycans have distinct effects. Nevertheless, we suspect that the overall N-glycan structure of BCMA may facilitate ligand binding.

In summary, here we used sugar analog probes to identify BCMA as a glycoprotein with a single N-glycan site that is involved in the regulation of surface retention and ligand binding. Thus, manipulation of sialylation on normal or malignant plasma cells may affect the life span of plasma cells or the treatment outcome of diseases. On the other hand, the N-glycan and α1–3,-4 fucosylation enhance the binding of ligand to BCMA, shedding light on the exploitation of strategies to improve the efficacy of receptor-based therapeutics.

Materials and Methods

Cell Culture and Reagents.

Human multiple myeloma cells H929, U266, and RPMI8226 were maintained in RPMI-1640 medium (Invitrogen) supplied with 10% (vol/vol) FBS (Biological Industries). Human 293T cells were cultured in DMEM (Invitrogen) supplied with 10% (vol/vol) FBS. 293F cells were cultured in freestyle medium (Invitrogen). β-Mercaptoethanol (50 μM; Invitrogen) was added to the medium for H929 cells. For sugar analog incorporation, cells were fed with alkyne fucose (200 μM), alkyne ManNAc (25 μM), or unmodified sugar. All cells were cultured at 37 °C with 5% (vol/vol) CO₂, except 293F cells that were cultured at 37 °C with 8% (vol/vol) CO₂.

Flow Cytometry.

For sugar analog incorporation, click reaction was performed by incubation with azido-biotin, followed by incubation with alexa488-conjugated streptavidin (1:100 dilution; BD Biosciences) as previously described (46). For ligand binding, cells were incubated with Fc-APRIL or Fc-BAFF (1 μg/mL) for 30 min on ice, followed by incubation with biotin-conjugated anti-human IgG-Fc antibody (1:200 dilution; eBioscience) for 15 min on ice and streptavidin-allophycocyanin (APC) (1:200 dilution; BD Biosciences) for 15 min on ice sequentially. For lectin binding, cells were incubated with biotin-conjugated SNA (1 μg/mL; Vector Laboratories), MAL (1 μg/mL; Vector Laboratories), or Phaseolus vulgaris leucoagglutinin (L-PHA) (1 μg/mL; Vector Laboratories) for 30 min, followed by incubation with streptavidin-APC for 15 min. For detection of surface BCMA, cells were incubated with APC-conjugated anti-BCMA antibody (1:20 dilution; R&D Systems) for 30 min. In some cases, CHX (10 μM; Sigma-Aldrich) was added to culture. The fluorescent intensity of stained samples was detected by FACS Canto (BD Biosciences), and results were analyzed by FlowJo (TreeStar).

Glycosidase Treatment.

Cells were washed with fasting medium [RPMI1640 with 0.5% (vol/vol) BSA] and then incubated with PNGase F (50,000 U/mL; New England Biolabs), α2–3 sialidase (1,000 U/mL; New England Biolabs), or sialidase (1,000 U/mL; New England Biolabs) for 4 h at 37 °C. Heparinase I (10 U/mL; Sigma-Aldrich) was added in APRIL ligand-binding assay. For the in vitro ligand-binding assay or lectin binding, sialidase (50 mU/mL; QA-Bio), galactosidase (30 mU/mL; QA-Bio), glucosaminidase (400 mU/mL; QA-Bio), mannosidase (100 mU/mL; QA-Bio), or L-fucosidase (5 mU/mL; QA-Bio) was incubated with cell extracts or purified BCMA in supplied buffer at 37 °C for overnight.

Other Methods.

Please see SI Materials and Methods.

Statistics.

Two-tailed Student t test was used for all experiments. P < 0.05 was considered significant.