Phosphorylation of MAP4 affects microtubule properties and cell cycle progression

Microtubules (MTs) are dynamic polymers of tubulin that are involved in many diverse functions, including spindle formation, vesicle and organelle transport, and cell motility. The fact that MTs polymerized in vivo are more dynamic than those polymerized in vitro (Desai and Mitchison, 1997EF10; Verdé et al., 1990EF30) has been attributed to the actions of cellular factors that destabilize MTs, such as Op18 (Belmont et al., 1996EF4; Larsson et al., 1999EF16), and the opposing action of factors that stabilize MTs, such as microtubule-associated proteins (MAPs) (Mandelkow and Mandelkow, 1995EF17; Andersen, 2000EF3). MTs undergo a dramatic increase in dynamics during mitosis, relative to interphase (Saxton et al., 1984EF24; Hamaguchi et al., 1985EF12), possibly due to activation of MT-destabilizing factors, deactivation of MT-stabilizing MAPs, or a combination of both. MAP4, a 210 kDa MAP present in all proliferating cells, binds to and stabilizes MTs both in vitro (e.g. Bulinski and Borisy, 1980EF6) and in vivo (e.g. Nguyen et al., 1997EF19). Thus, altered MAP4 activity is a mechanism that might contribute to increased MT turnover at the mitotic stage (Ookata et al., 1995EF22).

Consistent with this hypothesis, MAP4's level of phosphorylation has been shown to increase at the G₂/M transition (Vandré et al., 1991EF29). In vitro, MAP4 interacts directly with cyclin B-cdc2, the major kinase responsible for mitosis-specific phosphorylation of structural proteins (Ookata et al., 1995EF22). In addition, in vivo phosphorylation of MAP4 by cyclin B-cdc2 has been shown to occur at numerous sites throughout the molecule (Ookata et al., 1997EF23). Phosphorylation of MAP4 by cyclin B-cdc2 kinase renders MTs more dynamic in vitro, apparently by decreasing the frequency of rescue (Ookata et al., 1995EF22). This mechanism is consistent with the observed in vivo behavior of MTs during mitosis in mammalian cells.

Mitotic behavior of MAP4 homologs has been studied both in mammalian cells and in Xenopus. However, the Xenopus MAP4 homolog (XMAP4; XMAP230) (Cha et al., 1999; Andersen et al., 1994; Shiina et al., 1992) shows only limited sequence identity with mammalian MAP4 (∼30% sequence identity between Xenopus and human MAP4) (Shiina et al., 1999; Andersen, 2000). Other data suggest that there may also be mechanistic differences between the cellular actions of these homologs. For example, XMAP4 was reported to dissociate from MTs during mitotic prophase (Andersen et al., 1994), in contrast to mammalian MAP4, which remains associated with MTs throughout the cell cycle (Bulinski and Borisy, 1980; Olson et al., 1995). However, it should be noted that other investigators (Cha et al., 1999; Shiina et al., 1999) demonstrated continuous association of XMAP4 with MTs during mitosis. Because human MAP4 is a MT-stabilizing protein that is continuously localized on MTs throughout the cell cycle, we and others have hypothesized that mitosis-specific phosphorylation of MAP4 toggles its MT stabilizing function during portions of the cell cycle.

Accordingly, we focused on phosphorylation events within MAP4's MT-binding domain. We previously showed that cyclin B-cdc2 kinase phosphorylates serine-787 in vivo during mitosis (Ookata et al., 1997EF23), and that in vitro phosphorylation by cyclin B-cdc2 is sufficient to reduce MAP4's capacity to stimulate in vitro polymerization of MTs (Kitazawa et al., 2000EF15). By contrast, we showed that serine-696 is continually phosphorylated throughout the cell cycle in proliferating cells (Ookata et al., 1997EF23), whereas phosphorylation at this site is not detectable in quiescent, serum-starved cells (Srsen et al., 1999EF27). Because we predicted that phosphorylation at sites within the MT-binding domain would manifest effects on MT dynamics, we focused on the function of MAP4 phosphorylation at the serine-696 and -787 sites.

Except as noted, all chemicals were purchased from Sigma (St Louis, MO) or from Fisher Scientific (Tustin, CA). All tissue culture materials were from Gibco Life Sciences (Gaithersburg, MD). Immunochemicals were obtained from Organon Teknika (Durham, NC). All restriction enzymes were purchased from Promega Biotech (Madison, WI).

MAP4 cDNA was mutated at amino acids serine-696 and serine-787 and inserted into pEGFP-C1 vector, modified to be dexamethasoneinducible. Full-length MAP4 cDNA encoding the five-repeat isoform of human MAP4 (isoform IV; Chapin et al., 1995) was mutagenized using the Muta-Gene in vitro Mutagenesis Kit (Biorad, CA) and the following primers for the S696A/S787A (AA-MAP4) construct: AACAAGGAGCTCCCACCAGCCCCAGAG and TGGCTTGGATGGTGCGGCCCGCTT. Corresponding primers for S696K/S787K and S696E/S787E were substituted, respectively, with AAA/TTT and GAG/CTC at the positions underlined in the AA-MAP4 primers. All mutant and wild-type MAP4 cDNAs were then cut with SmaI and partially digested with BglII, which deleted the N-terminal 669 amino acids corresponding to the projection domain. Each fragment was then cloned into a modified pEGFP-C1 (Clontech, CA), generated by replacing its promoter with MMTV-LTR, a dexamethasone (dex)-inducible promoter from the pMAMneo vector (Clontech, CA). Transfection of wild-type and mutant MAP4 into mouse fibroblast L^tk- cells (ATCC, MD) and selection of stably transfected cell lines were performed according to protocols described previously (Nguyen et al., 1997). Cell lines were induced to express WT or mutant forms of MAP4 by adding 1 μM dex for 36 hours prior to analysis.

Following induction with dex, confluent cell monolayers containing equal numbers of cells in 60 mm dishes were washed once with Earle's buffered salt solution (EBSS) at 37°C and extracted with and without saponin (200 μg/ml) in PEM buffer (100 mM PIPES, pH 6.9, 5 mM MgCl₂, 1 mM EGTA) for 2 minutes at 37°C. Cells were scraped off the plates in SDS-gel sample buffer and boiled for 5 minutes. Identical volumes of sample buffer were added to pellets containing total cell protein (extracted in PEM alone) and MT polymer (extracted in PEM with saponin). Western blots of 20 μg extract protein (Chapin and Bulinski, 1991EF8), stained with anti-β-tubulin antibody (3F3, courtesy of James Lessard, University of Cincinnati, OH), were scanned with an HP ScanJet 6200C scanner and intensities of each immunostained band determined with MetaMorph software (Universal Imaging Corp., West Chester, PA). The percentage of tubulin in polymer was defined as the ratio (tubulin in MT polymer sample):(tubulin in total protein extract).

Cells in 100 mm tissue culture plates, induced with dex, were released from the substratum with Viokase solution, and counted in a haemocytometer before and after an additional 36 hours of growth. Cell cycle stages of each cell line were determined following 36 hours of dex-induction on coverslips. The coverslips were fixed in ice-cold methanol for 15 minutes, mounted in the presence of 1 μg/ml DAPI, and cell cycle stages were determined by scoring GFP-MAP4-labeled spindles and DAPI-labeled condensed chromosomes. Fluorescence-activated cell sorting (FACS) was performed as previously described (Nguyen et al., 1999EF20).

Following dex induction, cells on coverslips were incubated in medium containing the MT-depolymerizing drug nocodazole (10 μM) for 0-30 minutes at 37°C, washed once with Hanks' Balanced Salt Solution (HBSS) containing 10 μM nocodazole, and immediately fixed in ice-cold methanol for 15 minutes. GFP-expressing cells were scored for the presence of a MT array at each time of nocodazole treatment. Since all expressing cells showed an extensive MT array initially (T=0) and no MTs at T=30, the proportion of expressing lines with a MT array at the 10-minute time point (t=10), when a disparity was noted, was quantified.

Suspension cultures of each stably transfected cell line, induced with 1 μM dex for 20-28 hours, were used to prepare Taxol-stabilized MTs, as described previously (Chapin et al., 1991). After washing MT pellets four times in buffer containing 0.1 M PIPES, pH 6.9, 1 mM dithiothreitol, 1 mM EGTA and 1 mM MgCl₂ (PDEM), the MTs were resuspended in fresh PDEM plus the applicable concentration of NaCl (0-0.6 M), incubated for 10 minutes at 37°C, and centrifuged (40,000 g, 15 minutes). Supernatant fractions containing unbound MAP were assayed by western blotting with guinea pig antibodies prepared against the MT-binding domain of human MAP4 (MTB antibody; Nguyen et al., 1997) or to the bacterially expressed prolinerich domain of murine MAP4 (MuPro antibody; used at 1:1000 dilution). MAP4 released at each salt concentration was quantified and normalized to that eluted with 0.6 M NaCl, which was set at 100%.

In previous work, we identified sites within MAP4 that were phosphorylated in vitro and in vivo by cyclin B-cdc2 kinase (Ookata et al., 1997EF23). Two of the major spots detectable in 2D phosphopeptide maps were of special interest, since they corresponded to sites within the MT-binding domain of the MAP, and their sequences were conserved amongst species (Fig. 1). To determine in vivo sequelae of phosphorylation at these sites, we performed in vitro mutagenesis of the two relevant serines (ser-696 and ser-787). Fig. 2 shows that cyclin B-cdc2 phosphorylation at these sites was abolished in constructs containing mutations of either or both serines to alanine residues, thus confirming that these constructs are appropriate for the proposed studies.

Human MAP4 with serines at positions 696 and 787 (WT) or with residues 696 and 787 mutated to alanines (AA, imparting a neutral charge), lysines (KK, introducing two constitutive positive charges) or glutamates (EE, adding two constitutive negative charges, mimicking phosphorylation) were cloned into pEGFP-C1. Since MAP4 is believed to bind to MTs through ionic interactions (Vallee, 1982EF28; Ookata et al., 1995EF22), we hypothesized that substitution of amino acids of different charges within the MAP4 MT-binding domain might change MAP4's binding to and stabilization of MTs. To simplify interpretation of phenotypes caused by altered phosphorylation, in each mutant and wild-type construct we also deleted the N-terminal 669 amino acids (i.e. most of the projection domain of MAP4), which is known to contain additional phosphorylation sites (Ookata et al., 1997EF23).

We were unable to isolate L^tk- stable transfectants that constitutively expressed any of the GFP-MAP4 forms. Transfectants appeared to exhibit a growth disadvantage, and were consequently overgrown quickly by non-expressing cells (data not shown). For this reason, we cloned a dexamethasone (dex)-inducible promoter sequence from pMAMneo into pEGFP-C1 along with the MAP4 wild-type and mutant constructs. Stably transfecting this construct into L^tk- cells yielded L^tk- cells whose expression of MAP4 forms was inducible. Interestingly, cells expressing any of the dex-inducible mutant MAP4 forms decreased expression by at least twofold after three days of induction in dex. By contrast, there was no detectable decrease in GFP-MAP4 expression observed after dex-induction of WT-MAP4 for an identical period (data not shown). This was the first suggestive evidence that expression of the MAP4 mutants was more deleterious to cell growth than expression of WT-MAP4. Because of these observations, we induced cells expressing each construct with dex for 24-36 hours in all experiments.

We localized both WT and mutant MAP4 forms in vivo by imaging the fluorescence of the GFP attached to each. All transfected MAP4 species localized to MTs, and the pattern of those MTs was unremarkable when compared with MTs in untransfected cells, visualized via tubulin immunofluorescence (Fig. 3A). The morphology of the transfectants was also indistinguishable from that of untransfected cells. In mitotic cells, both wild-type and mutant MAP4 forms were localized to spindle MTs throughout cell division; spindles appeared to be normal in morphology. Note that in the images in Fig. 3A, since the GFP-EE-MAP4 fluorescence is weak to start with and weaker after the fixation, which is necessary for performing immunofluorescence, regions with dense or bundled MTs are accentuated. By contrast, from live cell imaging, in which single MTs are better visualized, it was clear that the MT array in EE-MAP4 cells is indistinguishable from that of naive or WT-MAP4 cells (data not shown).

We isolated two clones each of cells expressing WT and mutant versions of MAP4. Expanded clones showed significant cell-to-cell heterogeneity, even though FACS sorting or selection of clones was performed as closely as possible to the time of the experiment. In each clone we measured the average level of transfected MAP4 detected by western blot, and the percentage of expressing cells detected by GFP fluorescence (this varied from 27-96%). Fig. 3B shows blots of the cell lines; when corrected for the proportion of expressing cells, expression levels varied from 0.28- to 2.2-fold the MAP4:tubulin ratio (mol:mol) found in HeLa, the human cell line used for comparison (Nguyen et al., 1997EF19). As shown in Table 1, we used these data to categorize each line as low (<0.4-fold expression relative to HeLa), medium (0.5-1.5-fold), or high (>1.6-fold) expressers.

We determined that the tubulin content of each transfectant line was indistinguishable from the other transfectants or from naive L^tk- cells (determined as a proportion of total soluble protein, shown in Fig. 3B). These data are consistent with results of a previous study of four cell lines induced to express varying levels of wild-type MAP4 (full- or partial-length); only the line whose MAP4 expression level was eightfold higher than the level endogenous to HeLa cells showed any increase in tubulin level (Nguyen et al., 1997). In cells expressing MAP4 mutant forms, the proportion of tubulin that was polymeric was similar in all of the transfected cell lines (Table 1), again consistent with previous results on MAP4 transfected cells. Thus, expression of modest levels of mutant MAP4 forms in L^tk- cells did not significantly alter cell morphology, content of tubulin, or appearance of the MT array, nor did it affect the distribution of the MAP4 itself, during the cell cycle.

Next, we measured the growth rate of each transfectant line to determine if expression of mutant MAP4 species changed cell growth properties. Cell doubling times of 28-30 hours were measured for all lines, regardless of the mutant or its expression level (Table 2). Because growth rate data could be skewed by the growth of non-expressing cells within the population, we analyzed the cell cycle further, using FACS. As shown in Table 2, cells expressing a medium-level of EE-MAP4 exhibited a higher proportion of cells in G₁ phase and a lower proportion in G₂/M phase, suggesting that EE-MAP4 mutant cells either transit G₁ phase more slowly or pass through G₂/M phase more quickly, or both, as compared with cells expressing WT-MAP4 or other mutant MAP4 species.

FACS analysis is only amenable to the quantification of progression through three broadly defined categories of cell cycle events, that is, G₁, G₂/M, and S phases. To delineate cell cycle progression of individual cells more precisely, we used the localization pattern of their GFP fluorescence to assay each MAP4-expressing transfectant morphologically. We scored the proportion of log phase MAP4-expressing cells that contained mitotic spindles, separated centrosomes with an intact interphase MT array and condensed chromatin, and midbodies connecting daughter cells. These three morphologically-defined categories corresponded to cells in M phase, in late G₂ or the G₂/M transition, and in early G₁ phase, respectively. As shown in Fig. 4, KK-MAP4 cells possessed more than twice the proportion of cells in the late G₂ stage compared with any other cell line, suggesting that KK-MAP4 cells transit the G₂/M transition half as quickly as cells expressing the other MAP4 mutants. By contrast, neither the duration of transit through M-phase itself, nor through early G₁ differed significantly among the cell lines. Note that the morphological scoring of the low-expressing EE-MAP4 clone showed a higher proportion of cells in all three stages. Because these are very faintly expressing cells, we believe that this measurement is an artifact caused by the difficulty of scoring dimly fluorescent cells. In these cells, EE-GFP-MAP4 fluorescence may have been more readily detectable when it was concentrated into discrete structures such as separated centrosomes, spindles and midbodies.

A possible explanation for the increased abundance of late G₂ stage cells in KK-MAP4 cell lines is that the KK-MAP4 mutant is a more potent stabilizer of MTs than other MAP4 forms. KK-MAP4 would be expected to stabilize MTs more effectively than other forms if it bound to MTs more tightly. If this were the case, the interphase MT array containing these stable MTs might break down more slowly, and this could result in a delayed progression through late G₂ and into M phase, as we observed for KK-MAP4-expressing cells. Accordingly, we tested the MT-binding properties of the MAP4 mutant forms, exploiting the capacity of MAP4, which binds ionically to MTs, to be eluted from Taxol-stabilized MTs with moderate salt concentrations (Vallee, 1982EF28). By quantifying the efficacy of various concentrations of salt in eluting MAP4 from crude MTs, we assessed how tightly each MAP4 mutant form bound to MTs. Unlike a conventional affinity measurement, this assay, which relied upon MTs assembled in cell extracts, did not require the availability of pure proteins.

Results of the salt-elution assay (Fig. 5C) showed that KK-MAP4 bound to MTs more tightly than any other MAP4 form, mutant or WT (P<0.05 at salt concentrations <0.3 M). AA-MAP4 bound less tightly than KK-MAP4, but significantly more tightly than WT-MAP4 or EE-MAP4 (e.g. P<0.05 at salt concentrations <0.25 M). MT binding of the latter two forms was indistinguishable from one another, or from the endogenous (murine) MAP4 in L^tk- cells (Fig. 5B).

Inducing expression of transfected MAP4, either mutant or wild-type forms, did not significantly alter either the total amount of tubulin in the cells (Fig. 3) or the proportion of tubulin that existed in the polymeric state (Table 1). However, from MT-binding data described above, in which KK-MAP4 and AA-MAP4 exhibited tighter in vitro binding to MTs than either WT- or EE-MAP4, one might predict that these mutants would show corresponding differences in MT stabilization in vivo, that is, in the stability or longevity of their MTs. To test for differences in MT stabilization, we measured the capacity of MAP4 mutant forms to protect MTs from in vivo depolymerization by the MT-depolymerizing drug, nocodazole.

We assayed the nocodazole sensitivity of MTs to determine whether any of the cell lines expressing MAP4 mutants possessed MTs that differed in their sensitivity to nocodazole. This assay entailed exposing cells, all of which initially contained the same quantity of MT polymer, to nocodazole in cell culture medium for a limited time (10 μM, 10 minutes), and then scoring individual cells for the presence or absence of a nocodazole-resistant array of MTs. Table 3 shows that, of the four MAP4 forms, cells expressing AA-MAP4 were least sensitive to nocodazole, followed by those expressing KK-MAP4. MTs in either WT or EE-MAP4 cells were equivalent to each other in their nocodazole-sensitivity, and they were significantly more nocodazole-sensitive than MTs in cells expressing AA- or KK-MAP4 (P<0.05 for each line). These results are consistent with MT binding data for the two mutants and suggest that MT stability can be varied by altering the charge of residues within the MAP's MT-binding domain.

Previous studies demonstrated in vivo phosphorylation of two residues (ser-696 and ser-787) within the MT-binding domain of MAP4 by cyclin B-cdc2 kinase (Ookata et al., 1997EF23). Here, we examined effects of mutating these sites on cell cycle progression and MT stabilization. We found that stably transfecting mouse cells with MAP4 in which ser-696 and ser-787 had been replaced with lysines increased the proportion of cells in late G₂ or at the G₂/M transition, suggesting that expression of KK-MAP4 retards progression through this portion of the cell cycle.

In contrast to single-cell assays, neither measurements of population doubling times nor FACS analysis detected any differences in cell cycle progression among cells expressing the MAP4 mutants. This is not surprising since these methods are not sufficiently sensitive to detect differences of only 15-30 minutes, within a 28-30 hour cell cycle time. All transfectants showed heterogeneous expression levels of MAP4 mutant forms, no matter whether the cells had been cloned, drug-selected, or pre-sorted by FACS. Diversity in expression would tend to obscure differences in average growth rate of the population. Similarly, bulk analyses by FACS, which demarcates only three cell cycle stages, would not be expected to discriminate as finely among defined cell cycle stages as would morphological analysis of individual cells.

Expression of MAP4 mutants with differently charged residues also allowed us to test the hypothesis that the phosphorylation state of MAP4's MT-binding domain modulates its strength of MT binding and ultimately, its ability to stimulate MT polymerization. Specifically, we predicted that mutation of ser-696 and ser-787 to positively charged lysines (KK-MAP4) would increase the affinity of positively charged MAP4 for negatively charged tubulin within the MT. Conversely, we proposed that adding negatively charged glutamic acid residues to mimic the phosphorylated state of wild-type MAP4 (EE-MAP4) might decrease the affinity of MAP4 for MTs. Our results showed that, among the MAP4 mutants, KK-MAP4 indeed displayed the tightest binding to MTs, followed by neutrally charged AA-MAP4. EE-MAP4 bound the least tightly, which was also as expected. However, EE-MAP4's binding profile was indistinguishable from that of WT-MAP4. The similarity in behavior of EE- and WT-MAP4 was consistent with previous findings that ser-696 in wild-type MAP4 is continuously phosphorylated throughout interphase (Ookata et al., 1997EF23). This result also implies that the phosphorylation state of ser-696 impacts more on binding strength of MAP4 than does the phosphorylation state at ser-787.

Both KK-MAP4 and AA-MAP4 bound more tightly to MTs than did WT-MAP4 in the salt elution assay, and both KK- and AA-MAP4 showed greater in vivo MT stabilizing activity than WT-MAP4 in the nocodazole depolymerization assay. However, KK-MAP4 bound to MTs more tightly than AA-MAP4, whereas AA-MAP4 showed greater in vivo MT stabilization than KK-MAP4. We are uncertain why the binding and stabilization activities of KK-MAP4 and AA-MAP4 were not better correlated. One possibility is that the extra positively charged residues in KK-MAP4 resulted in higher MT affinity but also brought about conformational changes that led to less optimal MT stabilization.

MAP4 homologs exist in many species, including humans, cow, chicken, mouse and, more recently discovered, Xenopus. The Xenopus MAP4 homolog, XMAP230 (XMAP4), shows only 30% sequence identity with mammalian MAP4, possibly explaining behavioral differences between it and mammalian MAP4. For example, injecting XMAP4 function-blocking antibodies into Xenopus blastomeres disrupted assembly of mitotic spindles (Cha et al., 1999), whereas removal of human MAP4 from MTs by function-blocking antibodies, its depletion by antisense, or its overexpression in mammalian cells did not detectably affect spindle formation (Wang et al., 1996; Nguyen et al., 1997; Yoshida et al., 1996; Olsen et al., 1995; Nguyen et al., 1999). Finally, XMAP4 and mammalian MAP4 also differ in the effects each manifests on the parameters of MT dynamics: XMAP4 was found to increase MT growth rate, decrease the rate of MT shrinkage, and suppress catastrophes in vitro (Andersen et al., 1994), whereas human MAP4 was shown to increase the frequency of rescue events in vitro under steady-state conditions (Ookata et al., 1995).

Shiina et al. used Xenopus cultured cells to test the effects of mutating ten XMAP4 consensus sites for cyclin B-cdc2 or MAP kinase, including sites homologous in position and sequence to human ser-696 and ser-787 (Shiina et al., 1999). Mutating a minimum of six cyclin B-cdc2 consensus sites within the projection domain (i.e. not including the sites homologous to human ser-696 and ser-787) affected chromosome separation during mitotic anaphase A. In particular, Shiina et al. observed that chromosome segregation towards spindle poles was pre-empted by spindle elongation in about 20% of mitotic cells; they also observed a phenomenon they termed `back-and-forth peristaltic movement', whereby multiple cleavage furrows prematurely formed prior to spindle elongation and appeared to push the spindle from one side to another (Shiina et al., 1999). These experiments show that the projection domain of Xenopus MAP4 and/or its appropriate phosphorylation is required for some aspect of anaphase chromosome separation.

It is appropriate to compare our results with those Shiina et al. obtained when they mutated the two MT-binding region sites in XMAP that are homologous to human ser-696 and ser-787, as well as six of the projection domain phosphorylation sites of XMAP discussed above (Shiina et al., 1999EF26). Mutants of all eight sites phenotypically resembled mutants of the six projection domain sites alone, suggesting that the phenotype(s) of MT-binding site mutations were obscured by the striking phenotypic defects that arose from mutations in the projection domain sites. Therefore, we tested the role(s) of phosphorylation of MT-binding domain sites directly, focusing specifically on ser-696 and ser-787. As expected, mutating these two residues in each of three possible ways yielded more subtle, yet mechanistically more understandable effects than those Shiina et al. observed for their XMAP projection domain mutants (Shiina et al., 1999EF26). We can only speculate on whether expression of XMAP4 MT-binding domain mutants would yield the same phenotypes we observed for human MAP4 MT-binding domain mutants, since no one has yet dissected the role(s) of the MT-binding domain sites of XMAP4. Since the MT-binding domain functions not only in binding to, but also in stabilizing MTs, cells may use subtle modulation of MT-binding activity in concert with other changes in MAPs or MTs to fully effect the MT transitions needed for error-free mitosis.

Others have studied phosphorylation of MAP4 by other kinases, such as protein kinase C (Mori et al., 1991EF18), MAP kinase (Shiina et al., 1992EF25; Hoshi et al., 1992EF13), and MARK (Ebneth et al, 1999EF11; Illenberger et al., 1996EF14). These data, as well as the large number of spots detected in 2D phosphopeptide mapping of MAP4 (Shiina et al., 1992EF25; Ookata et al., 1997EF23; Andersen et al., 1994EF2; Mori et al., 1991EF18; Illenberger et al., 1996EF14) show that MAP4 undergoes multiple phosphorylation events involving several kinases, in addition to phosphorylation of ser-696 and ser-787 sites. Indeed, Ookata et al. showed that, in human MAP4, at least six additional sites, some within the MT-binding domain and some not, are phosphorylated in vivo during M-phase (Ookata et al., 1997EF23). These phosphorylation events may be carried out by cyclin B-cdc2, by other kinases studied in vitro, or by kinases not yet identified. Determining the effects of each phosphorylation event on mitosis or MT function is a complicated and daunting task.

Because expression of mutants showed that phosphorylation at each site affected MT stability and MAP binding, it may be informative to perform time-lapse studies to observe how each affects MT dynamics. MAP4 was previously shown in vitro to affect MT dynamics by increasing the frequency of rescue, that is, the transition between MT depolymerization and pausing or polymerization (Ookata et al., 1995EF22). By contrast, XMAP4 was shown to decrease the frequency of catastrophe, the transition between MT polymerization or pausing and depolymerization (Andersen et al., 1994EF2). Studies of in vivo MT dynamics may provide a clearer picture of how MAP4 controls the dynamics of individual MTs and whether mutant MAP4 forms with different MT-binding capabilities exert different effects on MT dynamics.

The authors are grateful to Michael Sheetz and Clare Waterman-Storer, for invigorating discussions and to Aleksandar Micevski, Aparna Prasad and Pallavi Prasad for technical assistance. This research was supported by a National Institutes of Health grant to J.C.B. (HL 62617) and W.C. was supported by a NIH Traineeship (T32 GM 08224 and T32 CA 09503) during part of this work.