INTRODUCTION

CD19 represents a 95-kDa, B-cell-specific, transmembrane glycoprotein that belongs to the Ig superfamily (1). It is expressed throughout B-cell differentiation except for the plasma cell stage and may reside in a multimolecular complex including CD21, CD81 (TAPA-1), and Leu13 (2, 3, 4). Moreover, CD19 is physically associated with the B-cell antigen receptor (BCR)¹ in activated B-cells (5). CD19 contains an extensive, highly conserved cytoplasmic domain that is encoded by nine exons and consists of 243 amino acids. In addition to several serine and threonine residues, nine conserved tyrosine residues encoded by seven exons are incorporated within the cytoplasmic domain (1). Several lines of evidence suggest that CD19 is involved in B-cell signaling. CD19 is associated with the protein-tyrosine kinase lyn and lck (6, 7, 8) and the protooncogene product Vav (9), and CD19 cross-linking induces protein-tyrosine kinase activity and intracellular Ca²⁺ (Ca²⁺_i) mobilization (8, 10, 11). Moreover, several tyrosine residues in the CD19 cytoplasmic domain can become phosphorylated, after both CD19 and BCR ligation (12, 13). Phosphorylation of two of these tyrosine residues has been found to result in the recruitment of phosphatidylinositol 3-kinase (12). Although CD19 monoclonal antibodies (mAb) are known as potent antagonists of B-cell proliferation (14, 15, 16), Carter and Fearon have reported that immobilized CD19 mAb lowers the threshold for BCR-mediated B-cell proliferation (17). It is still unclear whether these findings reflect the role of CD19 in vivo. A CD19 ligand has not been identified yet.

We and others have previously suggested that physical approximation of the BCR and the CD19/CD21 complex may be advantageous in antigen-dependent B-cell activation (18, 19). To test the signaling consequences of this approximation, we have generated bispecific antibodies (bsAbs) that bind membrane IgM (mIgM) and CD19 simultaneously.

MATERIALS AND METHODS

The mAbs specific for µH chain (CLB-MH15), CD3 (CLB-T3.4/2A and CLB-T3.2), CD5 (CLB-T1/1), CD14 (CLB-mon/1), CD16 (CLB-gran/1), CD19 (CLB-CD19), CD22 (CLB-CD22), CD40 (CLB-CD40), CD45 (CLB-CD45) and FITC-conjugated goat anti-mouse Ig (CLB-GM17F), horseradish peroxidase-conjugated goat anti-mouse Ig (CLB-GM17E), and horse anti-rabbit Ig (CLB-PK17E) antibodies were generated at the Central Laboratory of the Netherlands Red Cross Blood Transfusion Service (CLB). mAb specific for H chain was obtained from Southern Biotechnology Associates, Inc. (Birmingham, AL). Rabbit CD19 antiserum was kindly provided by D. Fearon (Cambridge, UK). The phycoerythrin-conjugated streptavidin was from Becton Dickinson (San Jose, CA). The phosphotyrosine mAb (RC20) and PY20 were purchased from Transduction Laboratories (Lexington, KY) and UBI (Lake Placid, NY). Syk, HCP, and Grb2 antibodies were purchased from Santa Cruz (Santa Cruz, CA). Recombinant human IL-2 was kindly provided by Hoffman-La Roche (Nutley, NJ), recombinant IL-4 was purchased from Genzyme Co. (Cambridge, MA), and recombinant IL-10 was kindly provided by R. de Waal Malefyt, DNAX Research Institute (Palo Alto, CA).

bsAbs were prepared according to a modified protocol as described by Vossebeld et al. (20). Briefly, F(ab)₂ fragments were prepared by digestion with 2% (w/w) pepsin at pH 3.7 (MH15), pH 4.7 (CD19), pH 4.5 (CD22), or pH 4.4 (CD45) for 16 h at 37 °C, followed by protein-A affinity chromatography to remove free Fc fragments and intact antibodies. Fab-SH fragments of CD19, CD22, or CD45 were prepared by reducing F(ab)₂ fragments (2-3 mg/ml in PBS) with 15 mM 2-mercaptoethanol for 30 min at 30 °C. The reduced product was passed through a G-25 Sephadex column (PBS, pH 6.7, and 2 mM EDTA) to remove the 2-mercaptoethanol. After elution, the Fab-SH fraction was incubated with the cross-linker BMH (1 mM for 1 h at 30 °C; Pierce). The reaction mixture was concentrated to the initial volume in C30 Amicon microconcentrators (Amicon, Beverly, MA). The concentrated reaction mixture was passed through a G-25 Sephadex column (PBS, pH 6.7, and 2 mM EDTA) to collect Fab-SH and to remove unbound BMH. Fab-SH fragments of MH15 were prepared as described above, and this fraction was immediately added to the Fab-BMH fraction. The mixture was concentrated to one-third of its volume in C30 Amicon microconcentrators and incubated for 14 h at room temperature. To separate the conjugated heterodimers from the unconjugated parental fragments, the reaction mixture was passed through a fast protein liquid chromatography Superose 12 column equilibrated with PBS. Appropriate fractions were pooled and analyzed on SDS-PAGE. The fraction containing the heteroconjugates was taken for further characterization as bsAb.

First, the fractions containing the bsAb were tested for their ability to inhibit the binding of the parental antibodies to Daudi B-cells. Therefore, cells were incubated with CD19, CD22, or CD45 bsAb (5 µg/ml) for 30 min at 4 °C. After washing the cells were incubated for 30 min at 4 °C with goat anti-mouse Ig-FITC or the respective parental biotinylated mAb (5 µg/ml). After another washing step, the latter samples were incubated with phycoerythrin-conjugated streptavidin. Finally, cell-associated fluorescence was measured in a FACScan (Becton Dickinson, Mountain View, CA). Second, it was tested whether both binding sites were present in the bsAb. For this purpose a previously described mIgM Daudi variant was used (21). Cells were incubated with the CD19, CD22, or CD45 bsAb, respectively. Because of the absence of mIgM, binding of the bsAb could only be mediated via the anti-CD19, anti-CD22, or anti-CD45 component of the bsAb. After washing, the cells were incubated with human serum as a source of soluble human IgM. After another washing step, the IgM bound to the anti-IgM part of the bsAbs was detected with FITC-conjugated MH15 mAb.

The Burkitt lymphoma cell line Daudi and the previously described mIgM Daudi variant (21) were routinely cultured in Iscove's modified Dulbecco's medium supplemented with 10% fetal calf serum and antibiotics. Tonsillar B-cells were isolated from tonsils of healthy donors.

For purification of B-cells peripheral blood mononuclear cells were incubated with saturating amounts of CD3 (CLB-T3.4/2A), CD14, and CD16 mAbs. After washing, the cells were incubated with goat anti-mouse Ig-coated magnetic particles (Dynabeads-M450, Dynal A.S., Oslo, Norway), and cell-coated beads were removed by magnetic separation. Immunophenotypical analysis indicated that the purity of the final cell populations was >98%.

Daudi cells were loaded with indo-1-AM (1 µmol/liter; Molecular Probes, Eugene, OR) for 60 min at 37 °C in HEPES medium as described (22). Cells were stimulated with µF(ab)₂ or bsAb, and ratios of 405/485 nm fluorescence per individual cell were calculated with the software Lysys (Becton Dickinson). Results were expressed as percentages of cells responding with a ratio above background level. Background levels were determined in resting cells and set at a level that included >95% of the cells.

B-cells were activated, and immunoprecipitates were analyzed as recently described (23). Briefly, after incubation of the B-cells with any of the antibodies, the reaction was stopped by the addition of ice-cold Nonidet P-40 immunoprecipitation buffer. Following the immunoprecipitation and SDS-PAGE, proteins were transferred to Hybond C nitrocellulose membranes (Amersham Corp.), and after blocking with 1% bovine serum albumin, the reactivity with specific antibodies was detected by chemiluminescence (Amersham Corp.) to visualize the bound proteins. Autoradiography was performed with Kodak X-Omat S films.

Purified tonsillar B-cells (5×10⁴ cells/well) were cultured in 96-well round bottom plates (Greiner, Nurtingen, FRG) in Iscove's modified Dulbecco's medium with 10% fetal calf serum in the presence of various concentrations of µF(ab)₂ or bsAbs combined with CD40 mAb (ascites 1:1000 dilution) and recombinant human IL-2 (50 units/ml), IL-4 (10 ng/ml), or IL-10 (100 units/ml), respectively. Proliferation was determined by a 16-h pulse with [³H]thymidine after 72 h of culture.

RESULTS AND DISCUSSION

Until now modulation of BCR signaling by CD19 has been studied using heteroconjugates of cross-linked bivalent antibodies. Carter et al. (10) demonstrated that these heteroconjugates induce both a synergistic activation of phospholipase C and an increase in [Ca²⁺]_i. We recently showed that co-ligation of CD19 with the cross-linked BCR enhanced the phosphorylation of SHC and the formation of multimeric SHC complexes (24).² In the case of T-cell-dependent antigens, the BCR is likely be ligated by an antigen (or antigenic fragment) with nonrepetitive epitopes that is opsonized by complement. Therefore, co-ligation of a single BCR with a single CD19 molecule would more closely reflect physiological B-cell triggering. To study the signaling consequences of this approximation of single receptor complexes, we choose to construct bispecific antibodies that ligate BCR and CD19 simultaneously.

Binding efficiency and bispecificity of the distinct bsAb were analyzed. First, preincubation of Daudi cells with µF(ab)₂ or any of the bsAbs was found to effectively inhibit the subsequent binding of the original bivalent mAbs (Table I). Second, to confirm the bispecificity of the individual bsAb, an mIgM Daudi variant was stained with any of the three bsAbs. In this case, binding of the bsAb could only be mediated via the anti-CD19, CD22, or CD45 component of the bsAb. Subsequently, the cells were incubated with normal human serum to permit binding of serum IgM to the anti-IgM-specific part of the cell-bound bsAb. The bsAb-bound soluble IgM was detected with FITC-conjugated MH15, and indeed, all bsAb were found to be able to bind (soluble) IgM (Table I).

Table I. Characterization of the bsAbs Daudi cells were incubated with control IgG1 mAb, µF(ab)2, or any of the bsAb (5 µg/ml) and subsequently with goat anti-mouse (GAM)-FITC or biotinylated CD19, CD22, or CD45 mAb (5 µg/ml) followed by streptavidine-phycoerythrin (PE). IgM Daudi cells were consecutively incubated with one of the bsAb, PBS, or normal human serum (NHS) and finally with MH15-FITC. The values represent mean fluorescence intensity. The results are representative for two separate experiments. NT, not tested. GAM-FITC CD19/PE CD22/PE CD45/PE MH15/PE PBS/MH15-FITC NHS/MH15-FITC C 39 411 535 782 992 16 21 µF(ab)2 975 405 NT NT 56 17 22 CD19 bsAb 873 45 547 763 153 16 159 CD22 bsAb 892 421 49 NT 135 15 142 CD45 bsAb 846 409 NT 67 123 16 297

To analyze the effect of bsAbs on mIg-induced tyrosine phosphorylation, Daudi B-cells were activated for 1 min with suboptimal amounts of µF(ab)₂ (1 µg/ml; Fig. 1, lane C) or any of the bsAbs (lanes D-F). Subsequently, the tyrosine-phosphorylated proteins were recovered from the lysates with phosphotyrosine mAb, and analyzed in anti-phosphotyrosine Western blots. Several tyrosine-phosphorylated proteins could be detected after activation with µF(ab)₂ (Fig. 1, lane C). Although no tyrosine phosphorylation was detected following activation with µFab or CD19Fab (data not shown), stimulation with the CD19 bsAb resulted in a more pronounced tyrosine phosphorylation of most of these proteins and preferentially induced tyrosine phosphorylation of a 90-95 kDa protein. In contrast, in the B-cells activated with CD22 bsAb this protein was hardly detectable, but instead a prominent 135-140 kDa tyrosine-phosphorylated protein was precipitated. Stimulation with the CD45 bsAb almost completely prevented the induction of protein tyrosine phosphorylation.

Based on their apparent molecular weight, their specific appearance, and the observation that both phosphoproteins resided in the membrane fraction of activated B-cells (data not shown), the 90-95-kDa and the 135-140-kDa phosphoproteins could represent CD19 and CD22, respectively. To investigate this possibility the CD19 protein was isolated by immunoprecipitation with CD19 Ab from B-cells that were activated with the distinct bsAbs and analyzed in anti-phosphotyrosine blots. Indeed, the 90-95-kDa CD19 protein was strongly and predominantly tyrosine-phosphorylated after activation with the CD19 bsAb (Fig. 2A). An indirect approach was taken for the identification of the major 135-140-kDa protein in CD22 bsAb-stimulated cells. It has been recently shown that in mIg-activated B-cells, tyrosine-phosphorylated CD22 specifically recruits the tyrosine phosphatase HCP (25, 26, 27). Therefore, the amount of HCP-associated CD22 may directly reflect the tyrosine phosphorylation level of CD22. The HCP protein was isolated from bsAb-activated B-cells by immunoprecipitation with HCP Ab and subjected to anti-phosphotyrosine blotting. The tyrosine-phosphorylated 135-140-kDa protein representing CD22 was almost exclusively detected after stimulation with CD22 bsAb (Fig. 2B). Together, these findings show that intimate co-ligation of mIg and accessory molecules results in the phosphorylation of specific substrates.

To investigate whether the potent induction of early B-cell activation events by CD19 bsAb could be translated in late events, the effect of these bsAbs on B-cell proliferation was analyzed. Purified tonsillar B-cells were stimulated with various concentrations of µF(ab)₂ or any of the bsAb, in combination with CD40 mAb and IL-2 (Fig. 3). Whereas a strong proliferative response was induced with µF(ab)₂, the CD19 bsAb were unable to induce B-cell proliferation. Similar results were obtained when IL-2 was replaced by either IL-4 or IL-10 (data not shown). The CD22 bsAb was found to be moderately agonistic at low concentrations, but this was converted to an inhibitory effect at the highest concentration. In accordance with its inability to activate protein-tyrosine kinases, no proliferative response was induced with CD45 bsAb. Similar results were obtained when purified peripheral blood B-cells were used (data not shown).

Thus, the initial biochemical analyses and functional assays appeared to be contradictory. CD19 bsAb are excellent inducers of protein-tyrosine kinase activity, yet they fail to work as agonists for mitogenesis. To further analyze this apparent discrepancy we investigated the effect of CD19 bsAb on two signaling pathways that are known to be crucial for mitogenic activation: 1) Ca²⁺_i mobilization and 2) the assembly of Grb2 complexes that may couple the BCR to p21^ras.

The mobilization of Ca²⁺_i is one of the cellular activation events that is critically dependent on protein-tyrosine kinase activity (28, 29, 30) and may lead to the transcriptional activation of early response genes (31). To compare the efficiency of µF(ab)₂ and CD19 bsAb to evoke Ca²⁺_i mobilization, a titration experiment was performed. After stimulation of Daudi cells with µF(ab)₂ (2.5 µg/ml), a rapid increase in Ca²⁺_i, followed by a plateau phase was observed. The percentage of responding cells, both in the initial peak (Fig. 4) and the plateau phase (data not shown), were found to be modestly enhanced following activation with the CD19 bsAb. In contrast, both peak and plateau values were considerably lower in cells activated with CD22 bsAb (data not shown). Moreover, at lower concentrations, the percentage of responding cells was found to be clearly higher after stimulation with the CD19 bsAb. This suggests that CD19 co-ligation may amplify the mIg-induced signal(s) that is (are) responsible for the Ca²⁺_i mobilization.

Recently genetic evidence has been provided that activation and tyrosine phosphorylation of protein-tyrosine kinase syk is an indispensible step in the process of BCR-induced Ca²⁺_i mobilization (32). In addition, we have found that the absence of mIg-induced Ca²⁺_i mobilization and proliferation was correlated with lower levels of syk expression in particular B-CLL (23). We therefore analyzed whether the aforementioned differences in Ca²⁺_i mobilization are correlated with the tyrosine phosphorylation status of syk. The measurements of mIg-induced Ca²⁺_i mobilization indicated that the amplifying effect of CD19 co-ligation was particularly evident at lower levels of mIg occupancy. Interestingly, analysis of syk tyrosine phosphorylation after activation with decreasing concentrations of µF(ab)₂ or CD19 bsAb revealed a similar pattern (Fig. 5). Already at a very low concentration (0.3 µg/ml), stimulation with the CD19 bsAb but not with the µF(ab)₂ resulted in tyrosine phosphorylation of syk and an associated 120-kDa protein (pp120). Only at optimal concentrations (5 µg/ml) a similar syk phosphorylation was induced by both stimuli, which appears to be in accordance with the comparable levels of Ca²⁺_i mobilization. The CD19 bsAb-induced tyrosine phosphorylation of pp120 was found to be relatively reduced in the latter case. Similar effects were observed in tonsillar B-cells (data not shown). The identity of pp120 remains to be determined but was excluded to be phospholipase C based on a distinct molecular mass and nonreactivity with specific antibodies (data not shown). However, based on the recently described association between the syk-related protein-tyrosine kinase ZAP70 and p120 c-cbl in T cells (33), pp120 may represent c-cbl assembling in syk complexes upon BCR cross-linking. Together, these findings indicate that intimate co-ligation of CD19 and mIg specifically lowers the threshold for mIg-induced tyrosine phosphorylation of syk and the subsequent Ca²⁺_i mobilization.

Recently, several groups have reported the mIg-induced assembly and membrane translocation of multimeric Shc-Grb2 complexes, which may couple the BCR to p21^ras (24, 34, 35). To investigate the possible influence of the CD19 bsAb on this process, we have analyzed the assembly of Grb2 complexes. Daudi cells were stimulated for distinct periods with suboptimal amounts of µF(ab)₂ or CD19 bsAb. Subsequently, the Grb2 complexes were isolated and analyzed in anti-phosphotyrosine blots. After stimulation with µF(ab)₂, the previously described 145-kDa phosphoprotein (pp145) and the 54-kDa form of Shc (p54) were detected within the Grb2 complex (Fig. 6, left panel) but were less profound than after cross-linking with biotinylated MH15 mAb (Fig. 6, right panel) (24). Both pp145 phosphoprotein and p54 Shc also resided in the Grb2 complex after activation with CD19 bsAb, although the phosphorylation of pp145 was predominant. It has been previously shown that pp145 directly interacts with the phosphotyrosine-binding domain of Shc (36), and recent evidence suggests that pp145 is phosphorylated by protein-tyrosine kinase lyn (37). A possible explanation for the predominance of pp145 in Grb2 complexes after stimulation with CD19 bsAb may thus reside in the CD19-associated lyn activity (6, 7, 8). Strikingly, an additional and very prominent tyrosine-phosphorylated protein of 90-95 kDa (pp90-95) was specifically detected under these conditions. This protein was hardly or not at all detectable after stimulation with µF(ab)₂ or with cross-linked biotinylated MH-15 mAb (Fig. 6, right panel) and has not been previously described in Grb2 complexes (24, 34, 35).

A plausible candidate for pp90-95 might be tyrosine-phosphorylated CD19, because CD19 is predominantly phosphorylated after stimulation with CD19 bsAb (see Figs. 1 and 2A). Moreover, several putative binding motifs for the Grb2 SH2 domain (amino acid code YXN) are present in the cytoplasmic tail of CD19. To analyze this possibility, CD19 and Grb2 proteins were immunoprecipitated from Daudi cells after stimulation with CD19 bsAb and analyzed in anti-phosphotyrosine blots (Fig. 7). The Grb2-associated pp90-95 co-migrated with tyrosine-phosphorylated CD19 and the finding that both phosphoproteins still co-migrated after N-glycanase treatment (data not shown) provide further support for their identity. Given the fact that pp90-95 was never detected in Shc immunoprecipitations (data not shown), it may be possible that CD19 can act as a putative competitor of tyrosine-phosphorylated Shc for binding to Grb2. Interference with the generation of regular Grb2 complexes might explain the lack of mitogenic potential of these bispecific antibodies.

Our data show that co-ligation of particular accessory molecules alters signaling via the BCR both in a qualitative and quantitative fashion. Thus, compared with the mitogenic µF(ab)₂ fragments, CD19 bsAb were unable to induce proliferation. Detailed analysis of early signaling events revealed an intact and even enhanced Ca²⁺_i mobilization but a qualitatively altered composition of Grb2 complexes. Expression of human CD19 in mice resulted in a severely impaired development of immature B-cells in the bone marrow and only limited numbers of peripheral B-cells (38). Whether this is caused by competition for a murine CD19 ligand or by interference with particular signals transmitted via the pre-BCR complex is currently unknown. With respect to the latter option interference with normal activation of the p21^ras pathway seems possible in these animals. In contrast to the CD19 transgenic mice, a relatively normal development of conventional B-cells was observed in CD19 knock-out mice (39, 40). However, the B-1 B-cell subset was found to be largely reduced, probably by interference with the self-renewal capacity (41). Notably, the T-cell-dependent B-cell responses were severely impaired in these mice. Combining data on transgenic and knock-out mice suggest that CD19 is involved in setting a signaling threshold for mIg-dependent B-cell activation. Our current data suggest that threshold setting may differ for individual intracellular signaling pathways. Likewise, the consequences of BCR CD19 co-ligation in vivo may depend on the extracellular context in which the B-cell is activated, e.g. depending on the presence of T cells and follicular dendritic cells.