Syk-dependent phosphorylation of microtubules in activated B-lymphocytes

B-cell activation through antigen binding to the B-cell receptor is a multistep process involving receptor aggregation followed by internalization of the antigen-receptor complex (Taylor et al., 1971). Aggregation of the B-cell receptor leads to the activation of protein-tyrosine kinases of the Src and Syk families (reviewed in Kurosaki, 1997; Chan and Shaw, 1995). Members of both kinase families are widely expressed in hematopoietic cells and their catalytic activities are required for the normal functioning of cells within the immune system. Syk kinase has been shown to play a key role in several signaling pathways in which Syk is thought to phosphorylate substrates that are critical for propagating signals downstream of the cell receptors. These pathways include those in B-cells, platelets, mast cells and macrophages, which all fail to respond to signals from the antigen, collagen, IgE, and IgG receptors, respectively (Turner et al., 1995; Cheng et al., 1995; Poole et al., 1997; Costello et al., 1996; Kiefer et al., 1998; Crowley et al., 1997), in the absence of endogenous Syk kinase.

Tubulin is one of the proteins identified as a substrate for Syk in vivo (Peters et al., 1996). Syk plays a key role in the tyrosine-phosphorylation of tubulin, as (1) a Syk-selective inhibitor blocks receptor-stimulated tubulin phosphorylation (Peters et al., 1996) and (2) tubulin is not phosphorylated in B-cell lines lacking endogenous Syk kinase (Fernandez et al., 1999). In vitro, Syk phosphorylates α-tubulin once, on the conserved tyrosine residue (Tyr⁴³²) located 20 residues upstream of the carboxyl terminus (Peters et al., 1996). These earlier studies on Syk phosphorylation of tubulin were performed under non-physiological conditions in which the microtubules were depolymerized. Under these conditions, active Syk is distributed in the soluble fraction, where it associates with and phosphorylates soluble α-tubulin (Fernandez et al., 1999). An important and unresolved question is whether or not cellular microtubules are also phosphorylated by Syk and, if so, what the consequence of that phosphorylation might be on the integrity of the cytoskeleton.

Cellular microtubules can provide an important matrix upon which are arrayed signaling molecules (reviewed in Gundersen and Cook, 1999). Phosphorylation of microtubules might affect cell signaling by enhancing SH2 domain-mediated interactions between microtubules and signaling molecules. A second way by which tubulin phosphorylation may affect cell signaling is by altering the polymerization properties of the tubulin. For example, the phosphorylation of α-tubulin on the extreme carboxyl-terminal tyrosine inhibited microtubule assembly in vitro (Wandosell et al., 1987). Further, in activated T-cells, the tyrosine-phosphorylated α-tubulin is restricted to the soluble fraction and apparently does not incorporate into microtubules (Ley et al., 1994b).

When considered together, these observations invite important new questions. What is the distribution of phosphorylated tubulin under physiological conditions, when the microtubule cytoskeleton is intact? Is the microtubule cytoskeleton phosphorylated by Syk kinase in response to B-cell activation? And what is the consequence of Syk-mediated tubulin phosphorylation in terms of the assembly properties of the tubulin?

Our approach to answering these questions is with the chicken pre-B-cell line DT40 (Takata et al., 1994). These cells have several advantages for the analysis of Syk kinase: (1) A variant of DT40 in which the Syk gene has been disrupted (Syk^−/−) is available, enabling us to compare directly the tyrosine phosphorylation patterns in wild-type cells and cells lacking Syk kinase. (2) DT40 cells contain only one member of the Src family of kinases, and Syk is the major contributor to tyrosine phosphorylation of proteins in these cells. (3) In DT40 cells, Syk kinase can be activated either by the clustering of B-cell surface IgM receptors or by the treatment of the cells with pervanadate. Pervanadate activation of Syk kinase requires the expression of antigen receptors, but is independent of receptor clustering (Wienands et al., 1996). Because pervanadate treatment of DT40 cells strongly shifts the equilibrium to the phosphorylated state, it provides a way to readily identify the substrates of Syk kinase and to determine the effect, if any, of tubulin phosphorylation on microtubule assembly. Further, because pervanadate activation does not involve the clustering of receptors, pervanadate treatment provides the opportunity to use immunofluorescence microscopy to visualize the cytoskeletons of Syk-activated cells without the interference of aggregates of immunoglobulin receptors.

After Syk kinase activation, microtubule cytoskeletons were isolated using standard conditions and probed with antibodies to phosphotyrosine and to tubulin. In vitro assembly experiments with purified tubulin were performed in order to determine the effect of Syk-phosphorylation on the extent of microtubule assembly and whether or not preassembled microtubules are a substrate for the Syk kinase. In this paper, we show that α-tubulin is tyrosine-phosphorylated in a Sykdependent manner, and tyrosine-phosphorylated α-tubulin is present within cytoskeletal microtubules. Syk-phosphorylation appears to only slightly affect the extent of microtubule assembly in vitro. Microtubules can incorporate prephosphorylated tubulin, and tubulin in preassembled microtubules also can be phosphorylated by Syk kinase.

Wild-type and Syk^−/− chicken DT40 cells were generously supplied by Dr Tomohiro Kurosaki, Kansai Medical University, Osaka, Japan (Takata et al., 1994). Cells were cultured in RPMI 1640 supplemented with 100 i.u./ml penicillin, 50 U/ml streptomycin, 10% fetal calf serum and 1% chicken serum. To activate cells through clustering of the B-cell receptor, cells were incubated with 50 μg/ml anti-chicken IgM (Bethyl Labs, Montgomery, TX, USA) at 37°C for 3 minutes. To bypass B-cell receptor clustering, cells were activated with pervanadate at the indicated concentrations in tissue culture medium at 37°C for 30 minutes. Working concentrations of pervanadate were prepared from a 5 mM stock solution of sodium orthovanadate in RPMI 1640 supplemented with 5 mM hydrogen peroxide. Activation of the cells by either method was terminated by fixation of the cells in paraformaldehyde or by addition of microtubule stabilizing lysis buffer (see below). To inhibit protein tyrosine phosphatases, all subsequent steps were performed in the presence of 2 mM sodium orthovanadate.

Cells were pelleted and the tissue culture medium removed. The cells were lysed in a large excess volume of warm microtubule stabilizing lysis buffer (0.1 M Pipes, pH 6.8, 2 mM MgCl₂, 1 mM EGTA, 4 M glycerol and 0.1% Triton X-100) at 37°C for 15 minutes. The insoluble cytoskeletons were then collected by centrifugation. To purify tubulin from the isolated cytoskeletons, the cytoskeletons were disrupted in 50 mM Hepes, pH 7.4, 1 mM PMSF, 2 mM sodium orthovanadate, and 0.05% SDS on ice for 15 minutes. During the solubilization, the sample was sheared by 10 passages through a 23-G needle. Particulate material was removed by centifugation at 27,000 g at 4°C for 10 minutes. The supernatant was adjusted to 150 mM NaCl and 1% Triton X-100, then incubated with colchicine-Sepharose (60 nM colchicine) at 4°C for 1 hour. Colchicine-Sepharose was prepared as described previously (Schmitt and Littauer, 1974). The resin was washed four times in the same buffer, and bound proteins were eluted with Laemmli SDS sample buffer (Laemmli, 1970).

Cells were lysed in the warm microtubule stabilizing lysis buffer, then settled onto polylysine-coated (1 mg/ml poly-L-lysine in distilled water) coverslips at room temperature for 10 minutes. The cytoskeletons on the coverslips were fixed in methanol at −20°C for 5 minutes, then rehydrated in phosphate-buffered saline (PBS). The samples were incubated with various combinations of primary antibodies: (1) monoclonal anti-tubulin antibodies (Asai et al., 1982), (2) anti-phosphotyrosine antibody (Hutchcroft et al., 1992) and (3) anti-γ-tubulin antibody (Sigma, St. Louis, MO, USA). After incubation with the primary antibodies, the coverslips were washed with PBS, and the samples were incubated with the appropriate fluorochrome-conjugated secondary antibodies (Kirkegaard and Perry, Gaithersburg, MD, USA). After washing with PBS, the coverslips were mounted in 50% glycerol in 0.2 M sodium borate, pH 9, and supplemented with 0.7% N-propylgallate. The samples were viewed by laser scanning confocal fluorescence microscopy on a BioRad MRC1024 mounted on a Nikon Optiphot upright fluorescence microscope. Images were captured by LaserSharp (BioRad) and formatted in Adobe Photoshop.

Cells or cytoskeletons were solubilized in SDS-sample buffer and electrophoresed in 7% SDS-polyacrylamide gels (Laemmli, 1970). Proteins were transferred to nitrocellulose and probed with antitubulin and anti-phosphotyrosine antibodies. After primary antibody binding, the blots were incubated with peroxidase-conjugated secondary antibodies (Amersham, Piscataway, NJ, USA), and developed using the enhanced chemiluminescence system (Pierce, Rockford, IL, USA) followed by exposure to X-ray film.

Phosphocellulose-purified tubulin was isolated from twice-cycled porcine brain microtubules (Borisy et al., 1975; Sullivan and Wilson, 1984). Recombinant GST-p35, which is a fusion protein containing the Syk catalytic domain, was isolated from transfected Sf9 cells by affinity-purification over glutathione-Sepharose (Peters et al., 1996). The MAP-free tubulin was phosphorylated with the recombinant Syk kinase in a buffer containing 0.1 M Pipes, pH 6.8, 1 mM MgCl₂, 1 mM EGTA, 1 mM NaVO₃ and 2 mM MgATP²⁻ at 37°C for 50 minutes. Under these conditions, the stoichiometry of phosphorylation was approximately 1 mole phosphate per mole of α-tubulin. The tubulin was supplemented with 1 mM GTP and assembled onto Tetrahymena ciliary axonemes (Mitchison and Kirschner, 1984). After 10 minutes of assembly at 37°C, the samples were fixed in glutaraldehyde and processed for immunofluorescence microscopy utilizing anti-tubulin and anti-phosphotyrosine antibodies. Fluorescent samples were viewed by confocal fluorescence microscopy and the lengths of the assembled microtubules were measured using NIH Image (http://rsb.info.nih.gov/nih-image/).

B-cells can be activated by crosslinking the B-cell receptor with anti-IgM antibodies. In a previous study, we recovered tyrosine-phosphorylated soluble α-tubulin from extracts of receptor-activated B-cells (Peters et al., 1996). Because these earlier experiments were performed under conditions in which the cellular microtubules were disassembled prior to analysis, we first asked whether tyrosine-phosphorylated α-tubulin was present in the cytoskeletal fraction of anti-IgM-activated DT40 cells. In the experiments described in this paper, the cytoskeletal fraction was defined as the portion of the cell remaining after extraction with 0.1% Triton X-100, as described in Materials and Methods.

Wild-type and Syk^−/− DT40 cells were activated by anti-IgM treatment and then extracted with 0.1% Triton X-100 in the microtubule-stabilizing lysis buffer. The insoluble fractions were analyzed in western blots (Fig. 1). In both cell types, several tyrosine-phosphorylated proteins were detected after activation. The number and intensity of phosphotyrosine proteins were greater in the wild-type cells than in the Syk^−/−?cells, which suggested that in DT40 cells Syk kinase is a major contributor to protein tyrosine phosphorylation in response to receptor clustering. A significant increase in antiphosphotyrosine reactivity was detected in wild-type cells after anti-IgM activation (compare lanes 2 and 4 in the anti-P-tyr blot in Fig. 1). Among the phosphorylated cytoskeletal proteins were four proteins (approx. 55 kDa, 66 kDa, 90 kDa and 120 kDa). These proteins correspond to the proteins previously shown to be associated with tubulin from activated B-cells (Fernandez et al., 1999). The 55 kDa phosphoprotein was not detected in the unstimulated wild-type cells or in either of the samples from Syk^−/− cells. The blot was stripped and reprobed with an anti-α-tubulin antibody, revealing that, as expected, tubulin was present in all four samples. These results suggested that at least a portion of the 55 kDa phosphoprotein was tubulin.

In order to confirm that phosphorylated tubulin was recovered in the cytoskeletons of the activated DT40 cells, the Triton-insoluble fraction was dissolved in 0.05% SDS and then the solubilized tubulin was absorbed to colchicine-Sepharose. As expected, α-tubulin was present in these samples, as shown by probing the western blot with anti-tubulin. Reprobing the stripped blot showed that the α-tubulin was tyrosinephosphorylated (Fig. 2). The α-tubulin in IgM-activated cells exhibited an increased reactivity with the anti-phosphotyrosine antibody as compared to the α-tubulin from unstimulated cells. Thus, α-tubulin present in the Triton-insoluble fraction was tyrosine-phosphorylated after surface IgM-mediated activation of B-cells. An examination of the cytoskeletal fractions prepared from the anti-IgM-treated cells by immunofluorescence with anti-tubulin antibodies indicated that the tubulin was in the form of microtubules (data not shown). However, due to the low steady state level of phosphorylated tubulin, we were unable to visualize anti-phosphotyrosine immunofluorescence staining of the microtubule cytoskeletons in anti-IgM-activated cells.

The treatment of B-cells with pervanadate leads to the activation of Syk in a manner that is dependent on the expression of antigen receptors, but not on receptor clustering (Wienands et al., 1996). We took advantage of this property to evaluate the phosphorylated proteins in DT40 cells. Total cell lysates and cytoskeletons were prepared from untreated and pervanadate-treated wild-type and Syk^−/− cells. The proteins were separated by SDS-PAGE, blotted to nitrocellulose, and probed with anti-phosphotyrosine antibodies (Fig. 3). This method detected a number of phosphoproteins in the total cell lysates from both cell types (lanes 2 and 6 in Fig. 3). Most of these phosphoproteins appeared not to be part of the cytoskeleton (compare lanes 2 and 4, lanes 6 and 8). When the cytoskeletal fractions were analyzed, a significant difference between the Syk^−/− cells and wild-type cells was observed (lanes 4 and 8 in Fig. 3). The predominant tyrosinephosphorylated species in the cytoskeletons prepared from pervanadate-treated wild-type cells was a protein of approx. 55 kDa; this phosphorylated protein was absent in Syk^−/− cytoskeletons. Cytoskeletons from untreated and pervanadatetreated wild-type cells were isolated and processed for western blotting (Fig. 4). Probing the blot with a mixture of antibodies to α- and β-tubulins identified these proteins in all the cytoskeleton samples, as expected. The blot was then stripped and reprobed with the anti-phosphotyrosine antibody. Only a single band, corresponding to α-tubulin, was reactive with the phosphotyrosine antibody. The cytoskeletal α-tubulin was tyrosine-phosphorylated only in cells treated with pervanadate. Thus, as was the case in the anti-IgM-mediated activation of B-cells, pervanadate treatment resulted in the Syk-dependent tyrosine-phosphorylation of cytoskeletal α-tubulin. However, unlike the case with anti-IgM-activated cells, the pervanadate treatment resulted in a steady state from which a high level of phosphorylated tubulin was recovered.

The biochemical studies summarized above demonstrated that a substantial portion of the phosphorylated α-tubulin was present in the Triton-insoluble cytoskeletal fraction of the pervanadate-treated DT40 cells. The apparent stability of the Syk-dependent phosphorylated tubulin under these conditions prompted us to attempt to visualize the distribution of the phosphorylated tubulin. A further advantage was that pervanadate treatment of cells results in Syk-dependent tyrosine-phosphorylation of target proteins without the complication of receptor-antibody aggregates.

Immunofluorescence microscopy of whole cells with the anti-phosphotyrosine antibody revealed that pervanadatetreated cells were stained in the cytoplasm, with wild-type cells staining more brightly than Syk^−/− cells (Fig. 5A-D). When cytoskeletons were prepared from these cells and then stained with anti-phosphotyrosine, a fibrous pattern was detected in the wild-type cells but almost no staining was seen with the Syk^−/− cells (Fig. 5E-H). The reduction in overall staining of the cytoskeletons compared to the total cells indicated that many of the phosphoproteins were not part of the cytoskeleton and was consistent with the biochemical analysis summarized above and in Fig. 3.

To determine whether the cytoskeletal network was composed of microtubules, cytoskeletons were double-stained with tubulin and phosphotyrosine antibodies (Fig. 6A,B). Both antibodies stained the same cytoskeletal network; in addition, the anti-phosphotyrosine antibody but not the anti-tubulin stained the periphery of the nucleus. To further confirm that the cytoskeletal network was composed of microtubules, wild-type cells were first treated with pervanadate, then with 30 μM nocodazole to disassemble the microtubules. After 30 minutes of nocodazole treatment, the cells were lysed and the cytoskeletons were double-stained with tubulin and phosphotyrosine antibodies (Fig. 6C,D). The nocodazole treatment eliminated the fibrous staining pattern. Thus, Syk kinase phosphorylates α-tubulin that is part of the microtubule cytoskeleton.

Different concentrations of pervanadate were examined for their ability to stimulate the tyrosine-phosphorylation of microtubules (Fig. 7). Treatment of wild-type cells, but not Syk^−/− cells, with 10 μM pervanadate produced bright antiphosphotyrosine staining of a spot located near the nucleus in each cell (Fig. 7C). Occasionally, untreated cells also exhibited similar but fainter staining (data not shown). Syk^−/− cells treated with 100 μM pervanadate showed only faintly staining spots (Fig. 7B), whereas wild-type cells treated with 100 μM pervanadate possessed brightly staining microtubules and the spots (Fig. 7D). A good cytological marker for centrosomes is γ-tubulin (Stearns et al., 1991). Wild-type cells treated with 10 μM pervanadate and double-stained with anti-phosphotyrosine and anti-γ-tubulin antibodies demonstrated that the two antigens were colocalized (Fig. 7E,F). These results suggest that Syk phosphorylates a component of the centrosome even under conditions in which the cytoskeletal array of microtubules is only weakly phosphorylated.

Syk kinase specifically phosphorylates the conserved tyrosine residue located near the C terminus of α-tubulin (Peters et al., 1996). In order to explore the possible effects of Syk-mediated tubulin phosphorylation on microtubule assembly, we performed in vitro assembly experiments. The strategy was first to phosphorylate soluble MAP-free tubulin with recombinant Syk kinase, then assemble the phosphorylated tubulin onto Tetrahymena ciliary axonemes. The microtubules were then processed for immunofluorescence microscopy, and the lengths of the (+) end microtubules were measured from the fluorescence images. The phosphorylation reaction was optimized in order to obtain the maximum extent of phosphorylation. Under the conditions of these experiments, the extent of phosphorylation was approximately 1 mole of phosphate incorporated per mole of α-tubulin.

Recombinant GST-p35, which is the Syk kinase domain, was affinity-purified on and eluted from glutathione-Sepharose. The protocol was to incubate tubulin with or without the p35, always in the presence of ATP. In two separate experiments, there was only a slight (12-13%) reduction in assembly, and in a third experiment there was no significant change in the average length of microtubules formed from Sykphosphorylated tubulin compared to mock-phosphorylated tubulin (Table 1). The incorporation of the phosphorylated tubulin into the microtubules was confirmed by double-staining with anti-phosphotyrosine and anti-tubulin antibodies (Fig. 8) and by western blotting (data not shown). Thus, even under conditions in which the tubulin was heavily phosphorylated by Syk kinase, the phosphorylated tubulin was assembled into microtubules only slightly less well than untreated tubulin. We infer that tubulin phosphorylated by Syk kinase should be able to be incorporated into cellular microtubules.

The experiments summarized in Fig. 8 and Table 1 indicate that prephosphorylated tubulin was able to assemble into microtubules in vitro. A second important question is whether or not Syk can phosphorylate preformed microtubules, or if the phosphorylated microtubules arise only from assembly of phosphorylated soluble tubulin. Taxol was utilized to induce the assembly of MAP-free, unphosphorylated tubulin. The taxol-stabilized microtubules were then incubated with recombinant Syk kinase with or without ATP, the microtubules collected by centrifugation, and the proteins evaluated by western blotting utilizing anti-tubulin and antiphosphotyrosine antibodies (Fig. 9). A substantial amount of phosphorylation was detected on the α-tubulin present in the microtubules treated with Syk kinase and ATP (Fig. 9, lane 2). In the immunofluorescence images of phosphorylated tubulin assembled onto axoneme seeds (e.g. Fig. 8), we noted bright staining of the axoneme. When the axonemes alone were incubated with Syk kinase and ATP, the axonemal tubulin was not phosphorylated, but autophosphorylated Syk kinase was bound to the axonemes (data not shown). These data suggest that microtubules but not axonemes can be tyrosinephosphorylated by Syk kinase.

Together, the experiments reported here demonstrate that Syk-phosphorylated α-tubulin can be incorporated into microtubules by two separate pathways in vitro. Prephosphorylated tubulin can be assembled into microtubules; alternatively, preassembled microtubules can be subsequently tyrosine-phosphorylated by Syk kinase.

In this paper we report three findings. (1) Syk kinase phosphorylates α-tubulin that is part of the cytoskeleton in DT40 cells activated either with anti-IgM or with pervanadate. (2) α-tubulin phosphorylated with Syk kinase can be assembled into microtubules in vitro. (3) Preassembled microtubules can also be phosphorylated by Syk kinase in vitro.

It had been reported previously that both Syk and its T-cell homolog, ZAP-70, are able to bind to and phosphorylate tubulin in B- and T-cells, respectively (Peters et al., 1996; Ley et al., 1994b; Isakov et al., 1996). In those earlier studies, the tubulin that was phosphorylated was isolated from the soluble fraction, leading to the speculation that soluble tubulin alone served as a substrate. However, here we show that Syk kinase activity leads to phosphorylated tubulin that appears to be part of the cytoskeleton, suggesting that the previously described Syk-tubulin interaction (Fernandez et al., 1999) extends to microtubules.

In the present study, we show that Syk kinase activity in anti-IgM-stimulated cells led to phosphorylated tubulin that partitioned in the insoluble fraction under conditions that stabilize the microtubule cytoskeleton. This phosphorylation was restricted to cells expressing the Syk kinase and occurred on the α-tubulin subunit, consistent with the substrate specificity of Syk kinase (Peters et al., 1996). Tubulin phosphorylation occurred in both anti-IgM and pervanadatestimulated cells, although the tyrosine-phosphorylation of tubulin elicited by receptor engagement was much less than that produced by pervanadate. While the data demonstrate that phosphorylated tubulin was incorporated into microtubules in pervanadate-treated cells, the exact location of the phosphorylated tubulin after receptor engagement was less clear. The low steady state level of tubulin phosphorylation in response to anti-IgM treatment prevented the direct visualization of phosphorylated tubulin by immunofluorescence microscopy. An alternative strategy to address this issue would be to disassemble the microtubules after their phosphorylation and then to biochemically evaluate the solubilized proteins. Nocodazole treatment would not be sufficient, because a phosphorylated insoluble aggregate remained in cells after nocodazole treatment (see Fig. 6); however, our earlier study provides evidence that the insoluble phosphorylated tubulin is in microtubules (Peters et al., 1996). In that study, B cells were first stimulated with anti-IgM, then lysed by Dounce homogenization on ice in a detergent-free buffer. These conditions disassembled the cytoskeleton and most of the phosphorylated tubulin was found in the soluble fraction. Thus, it is likely that receptor engagement resulted in a low level of phosphorylation of cytoskeletal tubulin.

In activated T-cells, tyrosine-phosphorylated α-tubulin was found to be excluded from the microtubule fraction (Ley et al., 1994b), whereas in the present study, anti-IgM activation of DT40 cells produced a detectable level of tyrosinephosphorylated tubulin in the Triton-insoluble fraction. It is possible that the two results may be due to methodological differences in the two studies. For example, the colchicine-Sepharose affinity method we used may be more efficient at accumulating the phosphorylated tubulin than the immunoprecipitation used in the T-cell experiment. Further, the methods to solubilize the cytoskeletons may be an important difference. In the T-cell study, 5 mM calcium was used, whereas in the DT40 study, we used SDS and shearing of the sample through a 23-G syringe needle. In our hands, 5 mM calcium did not appear to be a reliable method to disassemble all of the microtubules; this was also noted by Ley et al. (1994b). Regardless of the reason for the apparent discrepancy between the two studies, it is clear that the level of tyrosinephosphorylation of tubulin in anti-IgM-activated cells is very low, estimated to increase from 1.2% of the total tubulin in unactivated cells to 4.7% after activation (Fernandez et al., 1999). It is possible that, in the present study, we were unable to distinguish between phosphorylated tubulin actually assembled into microtubules versus tubulin otherwise associated with or trapped on the cytoskeleton. Our data, however, clearly indicate that when pervanadate was used to stimulate Syk kinase, phosphorylated tubulin was present in microtubules.

What might be the effect of the phosphorylation of microtubules? In theory, the added negative charge conferred by the phosphate might substantially inhibit the subsequent incorporation of the phosphorylated tubulin into microtubules, as was shown in experiments in which the C-terminal tyrosine was phosphorylated (Wandosell et al., 1987). We addressed this question by treating the cells with pervanadate, which, by shifting the equilibrium to the phosphorylated state, is a convenient way to exaggerate any effect of tubulin phosphorylation. We reasoned that, if there were a significant effect of tubulin phosphorylation on the cytoskeleton, we would detect it in pervanadate-treated cells. In our studies, hoewever, there was no noticeable difference in the microtubule polymer contents of wild-type versus Syk^−/− cells, as judged by immunofluorescence microscopy. Our results are in general agreement with an earlier study, which found that phosphorylation of tubulin by pp60^c-src did not cause significant changes in the polymerization of microtubules (Simon et al., 1998).

In an effort to quantify any subtle differences in total polymer amount, we utilized fluorescence-activated cell sorting of cytoskeletons stained with anti-tubulin antibodies to quantify the total microtubule polymer levels in pervanadatetreated wild-type and Syk^−/− cells (data not shown). No significant difference in polymer levels was observed between the two cell types. Unfortunately, pervanadate treatment caused a time-dependent destabilization of the microtubules that was independent of Syk expression or tubulin phosphorylation, complicating the FACS analysis of microtubule stability. Further, in the in vitro studies, even under conditions in which the tubulin was highly phosphorylated by Syk kinase, we found that the tyrosine-phosphorylated tubulin incorporated into microtubules only slightly less efficiently than the unmodified monomer. These results indicate that α-tubulin phosphorylation on Tyr⁴³² does not have a profound effect on microtubule assembly. Because anti-IgM activation apparently resulted in a lower level of microtubule phosphorylation than pervanadate treatment, it is unlikely that phosphorylation of tubulin significantly affects the integrity of the cytoskeleton during normal B-cell activation.

Increasing evidence suggests a major role for microtubules to act as scaffolds upon which components of signal transduction pathways are organized (Gundersen and Cook, 1999). In fact, Syk and several of its substrates in B-cells share the ability to bind tubulin (Fernandez et al., 1999). The phosphorylation of tubulin on tyrosine might serve to provide additional modes of interaction with molecules that possess SH2 (Src-homology 2) or PTB (phosphotyrosine-binding) domains. For example, Fyn associates with phosphotubulin in an interaction mediated by the Fyn SH2 domain (Marie-Cardine et al., 1995). Since the sequence surrounding the Sykcatalyzed site of phosphorylation on α-tubulin (pYEEV) (Peters et al., 1996) matches perfectly the optimal peptidebinding specificity of Src-family SH2 domains (pYEEI/V) (Songyang and Cantley, 1995), it is likely that Syk (or in T cells, ZAP-70) is responsible for the covalent modification of the Fyn-binding site on tubulin. Interestingly, Fyn has been localized in T cells to the centrosome and the microtubule arrays that emanate from it, the same sites preferentially phosphorylated in Syk-expressing cells (Ley et al., 1994a).

Thus, cytoskeletal elements such as microtubules may allow for the compartmentalization of kinases enhancing specificity by restricting accessibility of kinases to substrates (Faux and Scott, 1996). Indeed, B-cell activation due to receptor engagement leads to the translocation of tyrosine kinases to the cytoskeleton as well as changes in the organization of cytoskeletal networks. For example, Lyn and Syk are rapidly recruited to the membrane cytoskeleton after B-cell activation (Jugloff and Jongstra-Bilen, 1997). In macrophages, Syk translocates to the region where FcγR-mediated phagocytosis initiates concomitant with actin polymerization (Strzelecka et al., 1997). Macrophages derived from Syk knockout mice fail to undergo FcγR-mediated phagocytosis. This same step is also blocked by inhibitors of phosphatidylinositol 3-kinase (Crowley et al., 1997), which also has been reported to bind α- and β-tubulin in microtubules and γ-tubulin in centrosomes (Kapeller et al., 1993).

Finally, it is interesting that the present study indicates that centrosomes may be sites for tyrosine phosphorylation by Syk. Consistent with this observation is the recent report that Syk is located at the centrosomes in B-cells (Navara et al., 1999). At present, we do not know the identity of the phosphoprotein(s) that accumulates at the centrosome; it may be α-tubulin, or γ-tubulin, Syk, or another protein. Tyrosine phosphorylation of centrosomes may be involved in linking the microtubule cytoskeleton to events in B-cell activation. For example, in activated B-cells, the anti-IgM aggregated receptors form ‘caps’ that are normally positioned over the centrosomes (Schweitzer and Brown, 1984). This spatial interaction can be disrupted by microtubule active drugs such as colchicine or taxol (Rogers et al., 1981; Paatero and Brown, 1982). Further, in T-cells a dominant negative ZAP-70 mutation can block the reorientation of the microtubule cytoskeleton toward the site of aggregation of the T-cell receptor (Lowin-Kropf et al., 1998). Therefore, the tyrosine-phosphorylation of centrosomes may be involved in transmitting receptor induced polarity changes required for lymphocyte function to the microtubule cytoskeleton. In preliminary studies, we have not seen an absolute requirement of Syk for capping, centrosomedependent positioning of the cap, or cap internalization in DT40 cells since Syk^−/− cells are able to form caps in close proximity to the centrosome and to internalize the aggregated receptors. The kinetics of these events may be altered in Syk^−/− cells, however, and studies are underway to investigate these differences.

This research was supported in part by National Cancer Institute Grant CA37372 (R.L.G.), American Cancer Society grant RPG 96-017-CSM (D.J.A.) and National Science Foundation grants MCB-9728207 and BIR-9512962 (D.J.A.). S.F. was supported in part by a fellowship from the Indiana Elks Charities, Inc. awarded through the Purdue Cancer Center.