Temporal regulation of Lsp1 O-GlcNAcylation and phosphorylation during apoptosis of activated B cells

Crosslinking of B-cell receptor (BCR) sets off an apoptosis programme, but the underlying pathways remain obscure. Here we decipher the molecular mechanisms bridging B-cell activation and apoptosis mediated by post-translational modification (PTM). We find that O-GlcNAcase inhibition enhances B-cell activation and apoptosis induced by BCR crosslinking. This proteome-scale analysis of the functional interplay between protein O-GlcNAcylation and phosphorylation in stimulated mouse primary B cells identifies 313 O-GlcNAcylation-dependent phosphosites on 224 phosphoproteins. Among these phosphoproteins, temporal regulation of the O-GlcNAcylation and phosphorylation of lymphocyte-specific protein-1 (Lsp1) is a key switch that triggers apoptosis in activated B cells. O-GlcNAcylation at S209 of Lsp1 is a prerequisite for the recruitment of its kinase, PKC-β1, to induce S243 phosphorylation, leading to ERK activation and downregulation of BCL-2 and BCL-xL. Thus, we demonstrate the critical PTM interplay of Lsp1 that transmits signals for initiating apoptosis after BCR ligation.

B cell activation is critical for mounting antibody responses. Following antigen engagement with B-cell receptors (BCRs), a series of signalling cascades are triggered that involve the activation of spleen tyrosine kinase (Syk) and Lck/Yes-related Novel protein tyrosine kinase (Lyn), which phosphorylate phospholipase C gamma 2 (PLC-g2) and Bruton's tyrosine kinase, respectively. PLC-g2 and phosphoinositide-3kinase (PI3K) further activate downstream pathways, including PLC-g2/calcium/NFAT, AKT, IKK/NF-kB and ERK, resulting in the inducible transcription of genes that regulate the functions of activated B cells 1 . In addition, BCR crosslinking initiates an apoptosis programme for the elimination of auto-reactive B cells 2 . Thus, the coordinated regulation between apoptotic and BCR signalling controls both immune responses and homeostasis. However, the molecular switch that integrates the apoptotic pathway with BCR activation signalling remains unclear. In this regard, we speculated whether the onset of apoptosis following BCR activation could be modulated via crosstalk between posttranslational modifications (PTMs) of critical apoptosis regulators.
O-GlcNAcylation is the addition of O-linked b-N-acetylglucosamine (O-GlcNAc) to serine/threonine residues on cytosolic and nuclear proteins. Two enzymes, O-GlcNAc transferase (OGT) and O-GlcNAcase (OGA), are responsible for the addition and removal of O-GlcNAc, respectively 3 . Altered regulation of O-GlcNAc modification has been found in diseases such as diabetes 4 and Alzheimer's disease 5 . Because O-GlcNAcylation modifies proteins at serine/threonine residues that may overlap with the substrate sites of serine/threonine kinases, O-GlcNAcylation has been shown to play a role in modulating protein function by affecting protein phosphorylation. For example, competitive O-GlcNAcylation and phosphorylation at the S16 site of murine oestrogen receptor b mediates the degradation or stabilization of oestrogen receptor b 6 . In addition, O-GlcNAcylation of p53 at the S149 site, which abolishes phosphorylation at the adjacent T155 site, is critical for the ubiquitination and degradation of p53 and thereby results in p53 stabilization 7 . However, among the O-GlcNAcylated proteins identified thus far, only a few have been revealed to possess key regulatory and/or biological roles 8 .
In the immune system, it has been shown that the O-GlcNAcylation levels of nuclear proteins promptly increase during the early phase of T-lymphocyte activation, whereas O-GlcNAcylation on a portion of cytosolic proteins rapidly decreases 9 . Furthermore, nuclear factor of activated T cells (NFAT) and nuclear factor kappa-B (NF-kB) are modified by O-GlcNAcylation following the activation of either the Jurkat T or BJAB B-cell lines 10 , suggesting that, in addition to phosphorylation, protein O-GlcNAcylation may participate in the regulation of lymphocyte activation. However, the key regulatory proteins, their functional O-GlcNAcylation sites and the underlying significance of their crosstalk with other types of PTMs in B-cell activation remain unknown. To identify the molecular switch integrating BCR activation and the apoptosis pathway, we report the proteome-scale dissection of the functional interplay between protein O-GlcNAcylation and phosphorylation dynamics in activated primary B cells. We show that the inhibition of OGA promotes apoptosis in activated B cells, which is crucially dependent on the O-GlcNAcylation of lymphocyte-specific protein-1 (Lsp1), the recruitment of its kinase, PKC-b1, and the phosphorylation-mediated transduction of apoptotic signals.

OGA inhibition promotes B-cell activation and apoptosis.
Although it has been demonstrated that protein O-GlcNAcylation may play an important role in T-and B-cell activation 11 , the underlying mechanisms remain poorly understood. To further illuminate the mode of action of O-GlcNAc modification in B-cell activation, we purified mouse splenic B cells ( Supplementary  Fig. 1a) and demonstrated that the level of OGA was decreased in primary mouse splenic B cells beginning at 0.25 h after treatment with anti-IgM F(ab 0 )2 (Fig. 1a), which is consistent with the elevated levels of O-GlcNAcylation in splenic B cells after this treatment (Fig. 1b). There is also a dose-dependent effect of anti-IgM on the accumulation of O-GlcNAcylated proteins in splenic B cells (Supplementary Fig. 1b). However, the level of OGT was not changed after anti-IgM treatment ( Supplementary Fig. 1c). We next treated mouse B cells with thiamet-G (TG), a specific inhibitor of OGA that causes increased O-GlcNAc modification 12 . As expected, immunoblotting showed that TG treatment increased O-GlcNAc levels and that O-GlcNAc immunoreactive signals were blocked after co-incubating anti-O-GlcNAc antibody with GlcNAc (Fig. 1c). By monitoring the expression of the activation surface marker CD86, we next examined the effects of TG treatment on B-cell activation caused by BCR crosslinking and found that treatment with TG caused modest increases in CD86 expression (Fig. 1d); this result was consistent with a previous work showing that pretreatment with PUGNAc, a less specific OGA inhibitor 13 , sensitizes B-cell activation 11 . Similarly, two other parameters that are used to indicate B-cell activation, the influx of Ca 2 þ and the phosphorylation of mouse Lyn at tyrosine 397 that represents the active form 14 , were also increased in the presence of TG (Fig. 1e,f). Notably, TG treatment of anti-IgM-stimulated B cells enhanced activation-induced cell death, as determined by the number of viable cells and an analysis of 7-AAD/Annexin V staining by flow cytometric analysis (Fig. 1g,h, respectively). Consistently, knockdown of OGA by shRNA potentiates anti-IgM-induced B-cell activation and apoptosis, as demonstrated by enhanced phosphorylation of human Lyn at tyrosine 396 and Syk at tyrosine 525/526 (Fig. 1i), and Annexin V þ frequency in stimulated Ramos B cells (Fig. 1j). Taken together, we found that the TG-induced accumulation of O-GlcNAc modifications enhanced B-cell activation and apoptosis after BCR crosslinking.
O-GlcNAc-dependent phosphoproteome in activated B cells. We next analysed how O-GlcNAcylation perturbs phosphorylation-mediated B-cell activation/apoptosis and the potential interplay between O-GlcNAcylation and phosphorylation. The immunoblotting of mouse splenic B cells treated with TG was used to evaluate whether increased O-GlcNAcylation globally affects protein phosphorylation during B-cell activation. However, global changes in phosphorylation, as determined by immunoblotting with antibodies against pan-phosphoserine, phosphothreonine and phosphotyrosine, were not evident ( Supplementary Fig. 2), suggesting that the accumulation of O-GlcNAcylation because of TG treatment may target specific phosphorylation events in stimulated B cells. Thus, a quantitative phosphoproteomic analysis was performed to globally identify site-specific phosphorylation changes in response to TG treatment. Mouse splenic B cells were treated with either anti-IgM (condition B) or TG before the anti-IgM treatment (condition C); untreated B cells were used as the control (condition A) ( Supplementary Fig. 3a). The quantitative phosphoproteomic approach used in this study combined our previously developed automatic pH/acid-controlled IMAC procedure for phosphopeptide enrichment 15 and the IDEAL-Q algorithm for label-free quantification 16 (see workflow in Supplementary Fig. 3a).
A total of 3,752 non-redundant phosphopeptides that mapped to 3,079 unique phosphorylation sites originating from 1,266 proteins were identified (Supplementary Data 1). On the basis of the s.d. of our label-free quantitation strategy 17 , phosphopeptides showing more than twofold changes in abundance were considered to be altered phosphorylation sites with either upregulation or downregulation based on changes in either protein expression or phosphorylation stoichiometry 18 . Compared with the control (condition A), B-cell activation on anti-IgM stimulation (condition B) induced differential levels of 1,225 (32.6%) phosphopeptides from 632 phosphoproteins (B/A ratio). Furthermore, increased O-GlcNAc modification because of TG treatment altered 326 (8.7%) phosphopeptides, corresponding to 313 O-GlcNAcylation-dependent phosphosites from 224 phosphoproteins, in activated B cells (condition C/B), presumably representing subsets of the identified phosphoproteome sensitized to the effects of increased O-GlcNAcylation in activated B cells (Table 1).
Next, we focus on targets associated with BCR signalling. In Fig. 2, we illustrate the differential phosphorylation of 22 previously characterized proteins involved in the BCR signalling pathway (depicted in circles with a solid line), including Lyn, Syk, SHP1 and BLNK, as well as proteins in the PLC-g2/calcium/ NFAT, PI3K, IKK/NF-kB and ERK pathways. Through using a web-based bioinformatics software, Ingenuity Pathways Analysis, and literature mining, we further identified 17 phosphoproteins from our proteomics results that are not reported to be involved in the BCR signalling pathway but are closely related to the downstream molecules of BCR signalling, including Dock-2, insulin receptor substrate (IRS-2) and HDAC-1 (circles with a dashed line). The absence of a few known molecules in BCR signalling, including Bruton's tyrosine kinase and PI3K, may be a result of the prompt transfer of phosphate groups or the limited detection sensitivity of current proteomics technology. On the basis of the quantitation results, we filtered potential targets of O-GlcNAcylation-phosphorylation interplay in BCR signalling. Ideally, such molecules exhibit elevated phosphorylation in response to anti-IgM stimulation but display altered phosphorylation levels in response to TG pretreatment. Using Nop56 as an example, TG pretreatment reduced Nop56 phosphorylation at S513 and S543 ( Supplementary Fig. 3b).
Site-specific alteration was observed for Map4k1: enhanced O-GlcNAcylation suppressed two phosphorylation sites, S452 and S455, whereas phosphorylation at S373/S375 was not affected, and S419 showed slightly increased phosphorylation levels compared with that under basal-level O-GlcNAcylation ( Supplementary Fig. 3c). In summary, among the 44 identified phosphoproteins in BCR signalling, as many as 15 (34.1%) phosphoproteins showed altered phosphorylation levels, suggesting that O-GlcNAcylation exerts global effects on the dynamic phosphorylation-mediated BCR signalosome. Site-specific alterations also indicated that the crosstalk between O-GlcNAcylation and phosphorylation may alternatively occur at the same sites of a given protein but may not be a general event.   Figure 2 | Perturbed protein phosphorylation in the BCR signalosome in response to TG treatment. The quantitative change in the levels of phosphopeptides in the BCR signalosome after anti-IgM treatment is indicated on the left half of the circle for each protein; the change from pretreatment with TG is shown on the right half. The colours red, green and grey represent overexpression (ratio 42), downregulation (ratio o0.5) or no change in the phosphorylation site of each protein, respectively. Sample A represents B cells without treatment. Sample B represents activated B cells on treatment with anti-IgM (10 mg ml À 1 ) for 10 min; sample C was also pretreated with TG (0.1 mM) for 8 h. Circles with solid lines represent known molecules in BCR signalling, while those depicted in dashed lines have not been reported in this pathway but are added through using Ingenuity Pathways Analysis and literature mining (CD45 is PTPRC, BCAP is Pik3ap1, SHIP is Inpp5D, PKC is PRKC and Bam32 is DAPP1). Temporal O-GlcNAcylation and phosphorylation on Lsp1.
Among the 15 proteins in the BCR signalling pathway that showed potential interplay between O-GlcNAcylation and phosphorylation, Lsp1 drew our attention because this protein is known as a pro-apoptotic regulator in anti-IgM-induced apoptosis 19 even though the underlying mechanisms are unclear. Thus, we further investigated whether the site-specific O-GlcNAcylation of Lsp1 interferes with its phosphorylation and plays a role in anti-IgM-induced apoptosis. Here we have identified 17 unique phosphopeptides of Lsp1, including 12 phosphorylation sites (S77, T166, S168, T175, T179, S180, S184, T186, T187, T193, S195 and S243; Fig. 3a). All of them were known phosphorylation sites except for T193. Sites that are sensitive to TG treatment were illustrated ( Supplementary  Fig. 4a). We then validated that Lsp1 was present in anti-O-GlcNAc immunoprecipitates from the lysates of BCL-1 cells, a murine B-cell lymphoma line ( Supplementary Fig. 4b).
Site-directed mutagenesis was performed to generate a mutant Lsp1 in which amino acid 209 is replaced by Ala (S209A). Lysates from Ramos B cells transfected with vectors expressing either Flag-EGFP-tagged wild-type (WT) or S209A Lsp1 were subjected to a pull-down assay using succinylated wheat germ agglutinin (sWGA), which is a lectin that has high affinity for O-GlcNAc (ref. 20). The results shown in Fig. 3c indicate that compared with WT Lsp1, S209A Lsp1 can only be slightly pulled down by sWGA, suggesting that S209 is the major O-GlcNAc site on Lsp1.  13 12 KgSQPTLP IST I DER    We further confirmed that the lower O-GlcNAcylation level found on S209A Lsp1 was not a result of an impaired interaction with OGT because nearly equal amounts of OGT were immunoprecipitated by WT and S209A Lsp1 (Fig. 3d). During anti-IgM stimulation in splenic B cells, kinetic measurements indicated that Lsp1 O-GlcNAcylation, as detected by immunoprecipitation with anti-Lsp1 antibody crosslinked to protein A agarose, occurred rapidly and peaked at 1 h after stimulation (Fig. 3e, upper panel). Likewise, sWGA pull-down assay also showed similar kinetics of Lsp1 O-GlcNAcylation in stimulated B cells (Supplementary Fig. 4e). Among the 12 identified phosphorylation sites of Lsp1, site S243 is upregulated most markedly on TG treatment (Supplementary Fig. 4a). Therefore, Lsp1 phosphorylation was examined using a commercially available antibody against Lsp1 phosphorylated at S243. Following anti-IgM stimulation, S243 phosphorylation was continuously augmented at 16-24 h (Fig. 3e, lower panel); this occurred later than the rapid rise in O-GlcNAcylation. More importantly, the Lsp1 mutation S209A significantly reduced S243 phosphorylation in anti-IgM-stimulated Ramos B cells (Fig. 3f), showing that Lsp1 O-GlcNAcylation at the S209 site is required for the phosphorylation of S243. Consistently, after BCR crosslinking, the phosphorylation of Lsp1 at S243 was gradually enhanced in TG-treated splenic B cells at various time points after anti-IgM stimulation (Fig. 3g). Furthermore, TG treatment in stimulated B cells resulted in the O-GlcNAcylation of Lsp1 at earlier time points, which is correlated with the earlier induction of phosphorylation of Lsp1 at S243 (Supplementary Fig. 4f).
Apoptosis requires Lsp1 O-GlcNAcylation and phosphorylation. Lsp1 appears to be crucial for the induction of apoptosis in B cells after IgM stimulation because knockdown of Lsp1 reduced the frequency of anti-IgM-induced apoptosis in Ramos B cells (Fig. 4a). To further examine the biological outcome of the interplay between the O-GlcNAcylation and phosphorylation of Lsp1 in B-cell activation, we ectopically expressed Flag-EGFPtagged WT or various mutated Lsp1 variants using a lentiviral vector in Ramos B cells and Flag-tagged WT or various variants of Flag-tagged Lsp1 via a retroviral vector in splenic B cells (Fig. 4c,e). As reported previously 19 , ectopic expression of WT Lsp1 significantly enhanced anti-IgM-induced B-cell apoptosis in Ramos B cells and mouse primary splenic B cells, as determined by Annexin V staining (Fig. 4b-e). However, Lsp1 did not affect B-cell activation ( Supplementary Fig. 5a). Remarkably, ectopic expression of S209A Lsp1 induced much less apoptosis in Ramos B cells compared with WT Lsp1 (Fig. 4b,c). Mutation with an aspartic acid at the 243 site of Lsp1 (S243D), which acts as a phospho-mimicry mutation, resulted in enhanced anti-IgMinduced apoptosis compared with WT Lsp1 (Fig. 4b,c). TG treatment caused increased levels of apoptosis after anti-IgM stimulation in Ramos B cells carrying vector alone or exogenous WT Lsp1 (Fig. 4b,c). In contrast, TG did not enhance apoptosis in S209A Lsp1-expressing Ramos B cells (Fig. 4b,c). This result is consistent with another line of finding that TG did not increase the phosphorylation at S243 on S209A Lsp1 ( Supplementary  Fig. 5b). Likewise, S209A Lsp1-expressing mouse splenic B cells showed reduced anti-IgM-induced apoptosis (Fig. 4d,e). Mutating serine to alanine at the 243 site of Lsp1 (S243A) reduced anti-IgM-induced apoptosis, while ectopic expression of S243D resulted in enhanced apoptosis compared with S243A that is similar to the effect caused by the double mutant S209A and S243D (Fig. 4d,e). These combined results suggest that O-GlcNAcylation at the Lsp1 S209 site and O-GlcNAcylationdependent phosphorylation at the S243 site contribute to anti-IgM-induced apoptosis.
To further understand why the O-GlcNAcylation of Lsp1 at S209 potentiates anti-IgM-induced apoptosis, we monitored the expression of CD95, which controls extrinsic apoptotic B-cell signalling 21 , in S209A and WT Lsp1-expressing anti-IgMstimulated splenic B cells. However, CD95 levels were not affected when the O-GlcNAc site was mutated ( Supplementary  Fig. 5c). Given that Lsp1 was shown to translocate to the cytoplasmic side of the B-cell plasma membrane after anti-IgM treatment 22 and that Lsp1 binds with F-actin 23 , the polymerization of which contributes to BCR crosslinkinginduced apoptosis 24 , we suspected that the F-actin-mediated translocation of Lsp1 and apoptosis may be coordinately controlled by Lsp1 O-GlcNAcylation. Indeed, higher level of O-GlcNAcylated Lsp1 was detected in the F-actin-enriched fraction that was pulled down by phalloidin 25 (Fig. 4f). Moreover, much less F-actin was co-immunoprecipitated with Flag-EGFP-tagged S209A Lsp1 in lysates from Ramos transfectants as compared with Flag-EGFP-tagged WT Lsp1 (Fig. 4g). We next examined the expression of BCL-2 and BCL-xL, two important anti-apoptotic molecules that protect against anti-IgM-induced apoptosis 26,27 , and found that their levels were reduced in cells expressing WT Lsp1, but not in S209A, compared with control vector-transduced cells (Fig. 4h). Lsp1-mediated apoptosis is associated with the activation of ERK kinase 28 , and we found that the overexpression of WT Lsp1 caused significant increases in ERK phosphorylation in anti-IgM-stimulated B cells; in contrast, S209A-expressing cells only exhibited modest increases in ERK phosphorylation (Fig. 4i). Together, these data suggest that following anti-IgM stimulation, O-GlcNAcylation at S209 of Lsp1 causes increased ERK phosphorylation and the reduced expression of BCL-2 and BCL-xL, leading to the apoptosis of activated B cells.
Lsp1 O-GlcNAcylation recruits PKC-b1 to induce apoptosis. Given that S209 O-GlcNAcylation mediates S243 phosphorylation and promotes anti-IgM-induced apoptosis, we sought to further delineate the underlying molecular mechanisms that link the effects of S209 O-GlcNAcylation and S243 phosphorylation. In B cells, Lsp1 is a substrate of PKC 29 and recruits PKC-b1 to enforce apoptosis 28 ; Lsp1 is also phosphorylated by MAPKPKA kinase (MK2) at the S243 site in neutrophils 30 . To identify which kinase contributes to the phosphorylation of Lsp1 at S243, primary mouse splenic B cells were pretreated with MK25, an MK2 kinase inhibitor, or Gö6983, a PKC inhibitor. We found that treatment with Gö6983 dose-dependently reduced Lsp1 phosphorylation at S243 (Fig. 5a, left panel), but treatment with MK25 did not reduce phosphorylation (Fig. 5a, right panel). Treatment with Gö6983 also attenuated the pro-apoptotic effects of WT Lsp1 on anti-IgM-induced apoptosis (Fig. 5b). These data suggest that PKC-b1 mediates the phosphorylation of Lsp1 at S243, which may result in enhanced anti-IgM-induced apoptosis. In agreement with the observation that PKC-b1 is O-GlcNAcylated in rat hepatocytes, PKC-b1 was also found to be rapidly O-GlcNAcylated, peaking at 1 h after BCR crosslinking in B cells, as detected by sWGA pull-down (Fig. 5c). Because PKC-b1 regulates the S243 phosphorylation of Lsp1 and O-GlcNAcylation at S209 affects S243 phosphorylation, we assessed whether O-GlcNAcylation at S209 modulates the recruitment of PKC-b1 to phosphorylate Lsp1 at S243. We examined the amounts of PKC-b1 binding to Lsp1 in the absence or presence of TG. Of note, the elevated O-GlcNAc modification of Lsp1 by TG treatment resulted in increased interactions with PKC-b1 because more PKC-b1 was detected in anti-Lsp1 immunoprecipitates when using the lysates of TG-treated cells (Fig. 5d). Accordingly, we found that less PKC-b1 interacts with S209A Lsp1 compared with WT Lsp1 (Fig. 5e). These data suggest that O-GlcNAc modification at S209 of Lsp1 promotes the recruitment of PKC-b1.

Discussion
O-GlcNAcylation has been shown to be involved in several cellular events and to regulate many important cellular functions.
However, the role of O-GlcNAcylation in immune cells at the molecular level has not been extensively studied to date. B cells rapidly increase glucose uptake and glycolysis during the early phase of BCR engagement 31 . We found that global O-GlcNAcylation is dynamically altered after BCR crosslinking, which may reflect the rapid uptake and utilization of glucose that can be used as the source for UDP-GlcNAc synthesis via the hexosamine biosynthetic pathway and for O-GlcNAc   ARTICLE modification via OGT. Our results therefore imply that glucose acquisition and metabolism may contribute to the regulation of antigen-induced B-cell activation and apoptosis. Treatment with PUGNAc has been shown to sensitize lymphocyte activation 11 . PUGNAc is an inhibitor of a variety of N-acetylhexosaminidases, and its specificity for OGA is 35,000fold less than that of TG 32 . In our study, we found that O-GlcNAcylation accumulation resulting from TG treatment not only potentiates B-cell activation but also promotes anti-IgMinduced apoptosis. Although our mass spectrometry results did not reveal the potential O-GlcNAcylation and phosphorylation interplay of NF-kB and NFAT, the O-GlcNAcylation of which was previously reported to be important for controlling B-cell activation 10 , our data did suggest other TG-responsive molecules that are known to be critical to B-cell activation and apoptosis, such as PKC, PLC-g2 and Lsp1. We also identified the mediator, PKC-b1, as the kinase of Lsp1 that links its site-specific O-GlcNAc and phosphorylation. The O-GlcNAcylation of Lsp1 at S209 is required to bind to PKC-b1, thereby inducing the phosphorylation of Lsp1 at S243. It is still obscure why there is a time lag between the peaks of O-GlcNAcylation and S243 phosphorylation of Lsp1 as our data showed that O-GlcNAcylation of Lsp1 peaks at 1 h and the phosphorylation of Lsp1 at S243 occurs at later time points. We suspect this could be due to changes in the cellular localization of Lsp1. Supporting this notion, we found that O-GlcNAcylation of Lsp1 promotes the interaction with F-actin and its recruitment to the F-actin compartment. A previous study reported that Lsp1 S243 is involved in neutrophil polarization 30 ; however, the role of Lsp1 phosphorylation at the S243 site in cell survival has not yet been reported. We here show that phosphorylation of Lsp1 at S243 promotes anti-IgM-induced apoptosis in B cells. Taken together, it is conceivable that, on antigen ligation of BCRs, O-GlcNAcylated Lsp1 and PKC-b1 translocate to the plasma membrane of B cells, resulting in PKC-b1-mediated increased levels of Lsp1 phosphorylation at S243, which causes the activation of a signalling complex containing ERK and apoptosis (Fig. 5f).
In previous studies, protein O-GlcNAcylation was proposed to antagonize the effects of protein phosphorylation because both modifications occur at Ser/Thr sites. Accumulating evidence also suggests other types of interplay between O-GlcNAcylation and phosphorylation, such as reciprocal modification 33 . In the present study, quantitative phosphoproteomic results revealed 8.7% alteration in the levels of total phosphorylation of identified phosphopeptides on accumulated O-GlcNAcylation. We also found that PKC is O-GlcNAcylated after BCR crosslinking. However, whether OGT, PKC-b1 and Lsp1 form a complex after B-cell activation remains unknown. It is notable that PKC-b family proteins are indispensable for BCR-induced NF-kB activation and survival signalling 34 . Therefore, it is plausible that PKC-b1 may be a double-edged sword in promoting or preventing apoptosis during B-cell activation, depending on the status of its interactions with Lsp1. The expression of Lsp1 has been shown to be of diagnostic value for some lymphoid tumours, such as Hodgkin's disease 35 . Moreover, Lsp1 is aberrantly expressed in hairy cell leukaemia 36 . Our results revealed that PTMs of Lsp1 by O-GlcNAcylation and phosphorylation may affect its function and the survival of activated primary B cells. Whether such PTM interplay exists in B-cell malignancies and can be applied to control treatment outcomes warrants further study. In summary, we demonstrate the temporal regulation among Lsp1 O-GlcNAcylation, its kinase recruitment and O-GlcNAcylation-dependent phosphorylation, which coordinately transduces apoptotic signals in activated B cells.

Methods
Animals and cell culture. Six-to eight-week-old C57BL/6 mice were purchased from the National Laboratory Animal Center in Taiwan. The animal experimental protocol was approved by the Institutional Animal Care and Utilization Committee at Academia Sinica. Ramos B cells from Bioresource Collection and Research Center, Taiwan (BCRC) (BC-60252) and BCL-1 cells (ATCC, CRL-1669) were cultured in RPMI 1640 medium (Life Technologies) containing 10% fetal bovine serum (FBS), 100 U ml À 1 penicillin and 100 mg ml À 1 streptomycin (Life Technologies) at 37°C with 5% CO 2 . 293T (American Type Culture Collection, CRL-3216) and 3T3 cells (BCRC, BC-60071) were maintained in DMEM (Dulbecco's modified Eagle's medium, Life Technologies) containing 10% FBS, 100 U ml À 1 penicillin and 100 mg ml À 1 streptomycin at 37°C with 5% CO 2 . Cell lines used in this study are free from mycoplasma contamination as examined by using EZ-PCR mycoplasma test kit (Biological Industries, 20-700-20). Cell viability was monitored by trypan blue staining. Mouse splenic B cells were purified using anti-B220 microbeads (Miltenyi Biotec) from 6-to 8-week-old C57BL/6 mice and cultured in RPMI 1640 medium supplemented with 10% charcoal/dextran-treated FBS, 100 U ml À 1 penicillin, 100 mg ml À 1 streptomycin and 50 mM 2-ME at 37°C with 5% CO 2 . For mouse splenic B-cell stimulation, purified B cells at 2 Â 10 6 cells per ml were treated with goat-anti-mouse IgM F(ab 0 )2 (Jackson ImmunoResearch Laboratories). For blocking OGA activity, TG (0.1 mM, Cayman) was used to pretreat mouse splenic B cells or Ramos B cells 8 h before anti-IgM stimulation and was re-added into cell culture 1 day later. In some experiments, an MK2 inhibitor, MK25, or a PKC inhibitor, Gö6983, both purchased from Cayman, were added 1 h before anti-IgM stimulation. In some experiments, Ramos B cells were stimulated with a goat anti-human IgM F(ab 0 )2 fragment (Jackson ImmunoResearch Laboratories).
Retroviral and lentiviral transduction. To ectopically express mouse Lsp1 (mLsp1) or mouse OGT (OGT) in mammalian cells, a cDNA encoding full-length mLsp1 or OGT was cloned by RT-PCR from mouse splenic B cells. mLsp1 was then cloned into the pEGFP-N1 expression vector (BD Biosciences Clonetech), and OGT was cloned into the pFlag-CMV2 expression vector (Sigma). The detailed cloning procedures for the site-directed mutagenesis of various mLsp1 mutants, including S209A, S243A, S243D and S209AS243D, are available on request. For retroviral or lentiviral transduction, various forms of mLsp1, including WT and mutants, were cloned into the pGC-YFP retroviral vector 37 or pFUGW lentiviral vector 38 . The detailed cloning procedures are available on request. The procedures for the production of retroviral or lentiviral vector were as described 38 . Mouse splenic B cells or B-cell lines were transduced with viral vectors at a multiplicity of infection of 1-5 in the presence of 5 mg ml À 1 polybrene (Sigma). YFP þ or GFP þ cells were examined or sorted using a FACSCanto or FACSAria (Becton Dickinson) at 48 or 72 h after transduction.
Flow cytometry and calcium flux. The procedure for the flow cytometric analysis was performed as described 37 . The antibodies used in this study were as follows: anti-phycoerythrin-conjugated anti-mouse CD86 (clone GL1; BD PharMingen) and anti-phycoerythrin-Cy7 (PE-Cy7)-conjugated CD95 (clone Jo2, BD PharMingen). The cells were analysed using a FACSCanto (Becton Dickinson) with FACS Express 3.0 software. For detecting apoptotic cells, Annexin V and 7-AAD staining (BD PharMingen) were performed according to the manufacturer's protocols. The levels of calcium flux after anti-IgM stimulation were examined by labelling primary splenic B cells with Fluo-4AM dye (Invitrogen, 1 mM) at a density of 2 Â 10 6 cells per ml in RPMI for 30 min at room temperature. The labelled cells were then washed and resuspended in Hank's balanced salt solution and stimulated by anti-IgM (25 or 10 mg ml À 1 ) immediately before FACS analysis using a FACSCalibur (Becton Dickinson).
To perform purification of full-length mLsp1 for mass spectrometric analysis, expression vector encoding Flag-EGFP-mLsp1 was transfected to 293T cells for 48 h, and the recombinant Flag-EGFP-Lsp1 proteins were then collected and purified by anti-Flag antibody using previously described protocols 40 . To pull down O-GlcNAcylated Lsp1 and PKC-b1, 50 ml sWGA agarose (Vector Laboratories, AL-1023S) were added to cell lysate and incubated at 4°C for 12 h, and washed five times with IP lysis buffer. Finally, O-GlcNAcylated proteins were eluted by 5 Â sample buffer containing 250 mM Tris-Cl (pH 6.8), 4% bromophenol blue, 2% SDS, 50% glycerol and 10% b-mercaptoethanol at 95°C for 5 min for following immunoblotting analysis. For detection of O-GlcNAcylated levels of endogenous Lsp1 in primary mouse splenic B cells after anti-IgM stimulation, 15 mg anti-Lsp1 antibody (Santa CruZ, sc-23804) was crosslinked to 30 ml protein A agarose by disuccinimidyl suberate and immobilized to a spin column according to the manufacturer's instructions (Thermo Fisher Scientific, 26147). After crosslinking, 250 mg stimulated cell lysates were added to the antibody/agarose complex in a spin column, and incubated at 4°C for 12 h. Purified Lsp1 protein was obtained by washing the column five times with IP lysis buffer and then eluted with 50 ml elution buffer (Thermo Fisher Scientific, 26147).
F-actin enrichment. In all, 3 Â 10 7 Ramos B cells were lysed in 400 ml IP lysis buffer and subjected to immunoprecipitation with 50 mg biotin-phalloidin (Sigma, P8716) at room temperature for 12 h, followed by incubation with 50 ml streptavidin agarose resin (Thermo Fisher Scientific, 20347) at room temperature for 2 h and then spin-down at 1,000g at 4°C for 1 min. The supernatant was collected and used as the F-actin non-enriched fraction. Biotin-phalloidin-/streptavidin agarosebound proteins were resuspended in 400 ml IP lysis buffer and then used as the F-actin-enriched fraction. To detect the O-GlcNAcylation levels of Lsp1 within each fraction, 50 ml sWGA agarose (Vector Laboratories, AL-1023S) was added as described above.
Phosphopeptide enrichment using IMAC. For packing IMAC column, a 10 cm microcolumn (500 mm id PEEK column, Upchurch Scientific/Rheodyne) that has been packed with Ni-NTA resin (Qiagen) was enclosed in a stainless-steel columnend fitted with a 0.5 mm frit disk at one end. Phosphopeptide purification was performed using an autosampler and an HP1100 solvent delivery system (Hewlett-Packard). The flow rate was set at 13 ml min À 1 and the loading/condition buffer was 6% (v/v) AA with pH adjusted to 3.0. Ni 2 þ ions on the resin were removed by an amount of 100 ml of 50 mM EDTA in 1 M NaCl followed by equilibration with loading buffer for 15 min. To equip the IMAC column with Fe 3 þ , 100 ml of 0.2 M FeCl 3 was loaded into the column for another 15 min before sample loading. Peptide samples were resolved in loading buffer and loaded into the Fe 3 þequipped IMAC column for 20 min. A total of 100 ml of 25% (v/v) acetonitrile was used to wash the unbound peptides away for 15 min. The bound peptides were eluted with 100 ml of 200 mM NH 4 H 2 PO 4 and dried by vacuum centrifugation for further use.
Liquid chromatography-MS/MS analysis. Liquid chromatography-MS/MS analysis was performed on an LTQ-Orbitrap XL mass spectrometer (Thermo Fisher Scientific, Bremen, Germany) for phosphoproteome, which equipped with a nanospray interface (Proxeon, Odense, Denmark). Peptides were separated on a nanoAcquity system (Waters, Milford, MA), which was connected to mass spectrometry. Peptide mixtures were loaded onto a 75 mm ID, 25 cm length C18 BEH column (Waters, Milford, MA) packed with 1.7 mm particles with a pore of 130 Å. A segmented gradient in 90 min from 1 to 35% solvent B (acetonitrile with 0.1% formic acid) at a flow rate of 300 nl min À 1 and a column temperature of 35°C were used. Solvent A was 0.1% formic acid in water. Survey full-scan MS spectra were acquired in the orbitrap (m/z 350-1,600) with the resolution set to 60,000 at m/z 400 and automatic gain control (AGC) target at 10 6 . The mass spectrometer was operated in the data-dependent mode. The 10 most intense ions were sequentially isolated for CID MS/MS fragmentation and detected in the linear ion trap (AGC target at 7,000) with previously selected ions dynamically excluded for 90 s. Ions with singly and unrecognized charge state were excluded. To improve the fragmentation spectra of the phosphopeptides, 'multistage activation' at 97.97, 48.99 and 32.66 Thomson (Th) relative to the precursor ion was enabled in all MS/ MS events. All the measurements in the Orbitrap were performed with the lock mass option for internal calibration.
Phosphopeptide identification and label-free quantification. Raw MS/MS data were converted into peak lists (MGF file) using Raw2MSM (ref. 41) with default parameters. The resulting peak lists were searched against the IPI_MOUSE_3.87 database (version 3.87, 68161 entries) via an in-house Mascot search engine (Matrix Science Ltd., UK; version 2.2.1). The search parameters were set as follows: peptide mass tolerance, 10 p.p.m. MS/MS ion mass tolerance, 0.6 Da; enzyme set as trypsin and allowance of up to two missed cleavages; variable modifications included oxidation on methionine and phosphorylation on serine, threonine and tyrosine residues; peptide charge, 2 þ and 3 þ . Phosphopeptide identification results were accepted only if the Mascot scores pass statistically confidence (Po0.05) and the identified sequence ranked as the top match. The search results in MASCOT were exported in eXtensive Markup Language data (.XML) format. The raw data were converted into files of mzXML format. Peptide identification results from each liquid chromatography-MS/MS run and the corresponding XML files were used for quantitative analysis by IDEAL-Q as described previously 16,17 .
Statistics. Statistical significance was analysed by two-tailed unpaired t-tests in which the variances of the groups are similar. A value of Po0.05 was considered significant. All the data shown in this study are the mean ± s.e.m. from at least three biological replicates.