Interphase and monoastral-mitotic phenotypes of overexpressed MAP4 are modulated by free tubulin concentrations

Microtubules (MTs) are polar dynamic polymers that serve a multitude of cellular functions. For example, MTs segregate chromosomes during mitosis, serve as tracks for cellular transport and regulate cell-type-specific shape and polarity during interphase (reviewed by Desai and Mitchison, 1997). These diverse functions require transitions between stable and dynamic forms of MTs, which are regulated by two main classes of MT regulators. One class comprises MT-associated proteins (MAPs), which serve to stabilize MTs and, in some cases, also to alter specific dynamic properties (reviewed by Cassimeris, 1999). The other class consists of two distinct types of MT-destabilizing protein, namely Op18/stathmin family members and members of the Kin I kinesin subfamily of kinesins such as XKCM1 (termed MCAK in humans) (reviewed by Walczak, 2000). Specific members of these families have been shown to promote transitions from growing to shrinking MTs (i.e. catastrophes) and are termed catastrophe promotors.

It is widely recognized that a balance between MT-stabilizing and MT-destabilizing factors regulates the dynamics of MT polymerization, and that phosphorylation of these factors is important for cell-cycle-specific alterations of MT dynamics (reviewed by Andersen and Wittmann, 2002). Studies in egg extracts and using purified components have shown the importance of the XKCM1 catastrophe promotor and the MT-associated XMAP215 protein (Kinoshita et al., 2001; Tournebize et al., 2000). It has been proposed that much of the observed MT dynamics in intact cells can be attributed to the antagonizing activities of these two proteins (reviewed by Kinoshita et al., 2002).

In addition to the XMAP215 protein (termed TOGp in humans) (Charrasse et al., 1998), the structurally unrelated and ubiquitously expressed MAP4 protein has been implicated as being important for MT stability in vertebrates (reviewed by Olmsted, 1999). It has been reported that MAP4 promotes a large increase in transitions from shrinking to growing MTs (i.e. rescues) during in vitro MT assembly without altering other parameters that define dynamic instability (namely, the rate of growth and shrinkage, and the frequency of catastrophes) (Ookata et al., 1995). Interestingly, it was shown in the same study that phosphorylation of MAP4 results in loss of in vitro rescue-promoting activity without an apparent effect on MT binding. Studies in intact cells have revealed hyperphosphorylation during mitosis, which might suggest functional inactivation. However, given the complexity of MAP4 phosphorylation, the consequences of phosphorylation at specific sites remain largely unknown (Chang et al., 2001; Ookata et al., 1995; Ookata et al., 1997). MAP4 appears to be associated with MTs to the same extent throughout the cell cycle and has been proposed to localize the mitotically active cyclin-dependent kinase/cyclin B complex to the mitotic spindle by serving as a cyclin B binding site (Ookata et al., 1995). Determination of net polymer levels in HeLa cells expressing antisense RNA constructs has indicated that MAP4 has a role in maintenance of normal MT polymer levels during interphase (Nguyen et al., 1999). However, an antibody injection study has suggested that MAP4 ablation does not have an immediate detrimental effect on interphase or mitotic MTs (Wang et al., 1996). Thus, the physiological role of MAP4 in the cell remains unclear.

Previous reports on the outcome of overexpressed MAP4 in various cell types have been somewhat inconsistent. Whereas some have reported increased cellular MT content (Yoshida et al., 1996; Zhang et al., 1998), others have reported that the MT content is unaltered (Barlow et al., 1994; Nguyen et al., 1997). We have recently analyzed the effect of ectopic MAP4 in human leukemic cells with microtubules partially destabilized by either ectopic tubulin-sequestering proteins or proteins that promote catastrophe (Holmfeldt et al., 2002). It was found that coexpression of tubulin-sequestration-specific truncation derivatives of Op18/stathmin family members abolish microtubule stabilization by MAP4, whereas MAP4 successfully counteracts MT destabilization by two distinct and specific catastrophe promoters, namely XKCM1 and a low-sequestering truncation derivative of Op18/stathmin.

MT destabilization by catastrophe promotion and tubulin sequestering have opposite effects on the concentration of free tubulin heterodimers. Hence, a possible explanation for our previous data is that MAP4 is active at the increased free tubulin concentrations that result from catastrophe promotion, whereas its activity is reduced under tubulin-sequestering conditions. Here, we have evaluated this model by coexpression of MAP4 with truncated and/or chimeric Op18/stathmin family member derivatives with either catastrophe-promoting or tubulin-sequestering properties and distinct cellular localizations. The results support a model in which MAP4 activity is decreased in direct response to the lowered free tubulin concentrations that are still within a range that permits spindle formation.

pMEP4 shuttle vector derivatives expressing full-length human MAP4 (isoform III) (Chapin et al., 1995), Op18(1-99)-FLAG-tetraA and Op18(25-149)-FLAG-triA have been described previously (Holmfeldt et al., 2002; Larsson et al., 1999). MAP4-4R cDNA was kindly provided by Jeannette Chloe Bulinski. The two Op18 truncation derivatives, both containing an 8 amino acid FLAG epitope tag at their C-terminus, are denoted by numbers within brackets, which indicate the amino acid residues present. The RB3(1-145)-FLAG derivative lacks its native hydrophobic N-terminus and the amino acid sequence is numbered according to the Op18 sequence. This derivative was prepared by a PCR strategy using a full-length cDNA clone of human RB3 (accession number AL534520; purchased from Research Genetics, Huntsville, AL) as template with primers 5′-GC CGG TTC ACC ATG GCC TAC GAT GAC ATG GAA GTC ATC GAG CTG-3′ and 5′-CGC GGA TCC CTA CTT GTC GTC ATC GTC CTT GTA GTC CCT GGA GGC CTC TTC CTT CAG C-3′. The resulting PCR fragment was digested with NcoI and BamHI, and used to replace the corresponding fragment of a human Op18 cDNA inserted into the EcoRI site of pBluescript. The CD2 chimeras were prepared by a PCR strategy using a full-length cDNA clone of rat CD2 [accession number X05111; kindly provided by Allan Williams (He et al., 1988)] as template with primers 5′-GCG GCG CCA TGG CGA GAT GTA AAT TCC TAG GG-3′ and 5′-CGC GCG CCA TGG CGC ACG CGT TCC GTT TTT TCC TCT TGC-3′. The resulting PCR fragment contained NcoI sites at both ends and a coding sequence corresponding to amino acids 1-212 of rat CD2. The NcoI-digested fragment was inserted into the corresponding site of Op18(56-149)-FLAG or RB3(1-145)-FLAG to create CD2-Co and CD2-RB3, respectively. The Op18(56-149)-FLAG derivative does not bind tubulin with significant affinity and served as a negative control (Larsson et al., 1999). The coding sequence of the PCR-generated fragment was confirmed by nucleotide sequence analysis using the ABI PRISM dye terminator cycle sequencing kit from Perkin Elmer. For expression in cell lines, coding regions for each MT regulator were subcloned as HindIII to BamHI fragments into the Epstein-Barr virus (EBV)-based shuttle vector pMEP4 (Invitrogen). Conditional expression/coexpression was induced from the hMTIIa promoter, which can be suppressed by cultivation in a specifically formulated medium and subsequently induced by Cd²⁺ (Marklund et al., 1996). Single and cotransfection of human K562 erythroleukemia cells using pMEP4 derivatives, and selection of hygromycin-resistant cell lines, were performed as described (Gradin et al., 1998) using 18 μg of DNA in total and the indicated ratio of each specific derivative. Ectopic expression was induced in transfected K562 cells by changing from culture medium containing 15 μM EDTA to a culture medium containing 0.5 μM Cd²⁺. The DG75 Burkitt's lymphoma and Jurkat acute T-cell leukemia cell lines were transfected and selected by the same protocol, and high-level expression was induced by the addition of 15 μM Cd²⁺.

Immunoblotting and subsequent detection using the electrochemiluminescence (ECL) detection system (Amersham Pharmacia Biotech) were performed using anti-MAP4 (M75820; Transduction Laboratories), anti-α-tubulin (B-5-1-2, Sigma) and anti-rat CD2 (MRC OX-34) (He et al., 1988). To allow simultaneous and equivalent detection of all Op18/stathmin family members, rabbit antibodies were raised against a peptide corresponding to a completely conserved region of Op18 (SLEEIQKKLEAAE) corresponding to residues 46 to 58. The resulting antibodies were affinity purified by adsorption to native Op18 coupled to Sepharose. Analysis of cellular MT content by flow cytometry (>95% of all cells were included in the acquisition gate and >200,000 cells were collected) was performed using a FACS Calibur instrument (Becton and Dickinson) as described previously (Holmfeldt et al., 2001), but with the following modifications: first, soluble tubulin was pre-extracted in a saponin-containing MT-stabilizing buffer modified by increased pH (pH 7.4), omission of glycerol, and reduced paclitaxel concentration (from 4 μM to 20 nM). These modifications minimize nonspecific MT polymerization during the fixation step. In experiments including cells transfected with plasma-membrane-associated CD2-RB3, saponin in the MT-stabilizing buffer was replaced with 0.2% Triton X-100 to ensure that tubulin-associated CD2-RB3 complexes were solubilized. This protocol ensures that MT-specific fluorescence is not obscured by retention of tubulin heterodimers by CD2-RB3. The final modification involved addition of an equal volume of 4% paraformaldehyde dissolved in PEM buffer (80 mM piperazine-N,N'-bis[2-ethanesulfonic acid], 1 mM EGTA, 4 mM Mg²⁺, pH 7.4) added directly to cells resuspended in MT-stabilizing buffer. After gentle mixing and 15 minutes incubation at 37°C, the cells were washed and stained for α-tubulin and DNA as described (Holmfeldt et al., 2001). In some experiments, these cells were also stained in parallel with anti-phospho-histone H3 (Ser28; Upstate Biotechnology). Quantification of MAP4 expression by flow cytometry was performed on cells chilled on ice to depolymerize MTs, followed by paraformaldehyde fixation (4%) and staining with anti-MAP4 (5 μg/ml). Fluorescein-conjugated rabbit anti-mouse immunoglobulin was used as secondary antibody. For immunolocalization of centrosomes, cells were fixed with methanol at -20°C and stained with rabbit anti-pericentrin (PRB-432; http://www.babco.com). For characterization of spindles by immunofluorescence analysis, cells were permeabilized with saponin (0.2%) in MT-stabilizing buffer and subsequently fixed in 4% paraformaldehyde/0.5% glutaraldehyde, followed by quenching with NaBH₄. MTs and DNA were co-stained using Alexa Fluor488-conjugated anti-α-tubulin and propidium iodide, and analyzed by epifluorescence. To evaluate tubulin sequestering in cells expressing CD2 chimeras, cells were fixed directly in 4% paraformaldehyde at 37°C. CD2 chimeras and tubulin were co-stained with biotinylated anti-CD2/R-phycoerythrin-conjugated streptavidin and Alexa Fluor488-conjugated anti-α-tubulin, respectively, and analyzed using a Leica SP2 confocal imager system (Marklund et al., 1996). To estimate cytosolic levels of nonpolymerized tubulin in mitotic cells, fluorescence intensities of Alexa Fluor488 within ten randomly chosen circular areas (radius approx. 0.25 μm) in the cytosol, which were free from MTs, were evaluated in a confocal section. To ascertain that only tubulin-specific fluorescence was quantified, the analysis was performed on cells stained with Alexa Fluor488-conjugated anti-α-tubulin alone. Using the same general approach, plasma-membrane-associated tubulin was analyzed from the same confocal section within ten randomly chosen rectangular areas of the plasma membrane. The cytosolic and plasma membrane fluorescence intensities were averaged for each of a total of 50 cells analyzed. Special care was taken not to bleach the fluorescence signal prior to confocal scanning.

We have recently reported that ectopic MAP4 stabilizes the interphase MT array in the human K562 leukemia cell line (Holmfeldt et al., 2002). Provided that overexpressed MAP4 acts similarly during mitosis, this activity would seem likely to interfere with spindle formation. By employing the replicating pMEP vector for stringent regulation of ectopic MAP4 expression from the hMTIIa promotor, the mitotic overexpression phenotype was evaluated. Induction of pMEPMAP4-transfected K562 cells with Cd²⁺ resulted in a rapid six-to tenfold increase in MAP4 levels from a basal level close to the endogenous level (Fig. 1A). Since the total levels of α-tubulin were unaltered (Fig. 1A), it appears that ectopic MAP4 does not significantly alter tubulin content in K562 cells within a 24 hour period. However, ectopic MAP4 does cause a mitotic block, as demonstrated, first, by a time-dependent accumulation of more slowly migrating MAP4 phosphoisomers characteristic of mitotic cells (Fig. 1A, right) (Ookata et al., 1997) and, second, by the fact that most cells have G2/M content of DNA after 24 hours (Fig. 1B). Immunofluorescence staining of MAP4-overexpressing cells revealed the appearance of mitotic cells with a striking and uniform morphology characterized by monoastral spindles containing both kinetochore and non-kinetochore MTs (a confocal section is shown in Fig. 1B). Confocal sectioning of cells revealed that the chromosomes have spherical distribution around a pair of closely and centrally located centrosomes, as detected by pericentrin staining (Fig. 1C). For comparison, we also treated K562 cells with the drug monastrol, which has been reported to generate a monoastral spindle by inhibition of the mitotic kinesin Eg5 (Kapoor et al., 2000; Mayer et al., 1999). Confocal images reveal monoastral spindles with features that appear distinct from the monoastral spindles generated in MAP4-overexpressing cells. For example, in monastrol-treated cells, all MTs appear attached to kinetochores and most spindle MTs are not anchored to a centrally located centrosome (Fig. 1B).

The penetrance of the monoastral mitotic overexpression phenotype of MAP4 in K562 cells is very high and it is evident from Table 1 that the same fraction of mitotic cells is generated as with treatment using the MT-stabilizing drug paclitaxel (Table 1), which indicates that ectopic MAP4 imposes a complete M-block. However, paclitaxel does not cause monoastral spindles in K562 cells. Finally, it is evident from the data shown in Fig. 1D that the mitotic overexpression phenotype of MAP4 is not unique for K562 cells, since high frequencies of monoastral spindles are also observed in Burkitt's B-cell lymphoma (DG75) and acute T-cell leukemia (Jurkat) cell lines.

To compare MAP4-mediated elevation of MT content during interphase and mitosis, we devised a modification of a previous method to quantify MT-specific fluorescence by flow cytometry. This modified protocol was used to analyze the transfected cell populations presented in Table 1. As shown in Fig. 2A, separation of G1-, G2- and M-phase populations is obtained by the criteria of DNA content and disrupted nuclear envelope, as detected by propidium iodine staining of DNA and decreased granularity, respectively (Epstein et al., 1988). The acquisition gate was set to include all viable cells (>95%) and the fraction of cells identified by the M-phase gate is consistent with microscopic determination of the mitotic index of each cell population (Table 1). To confirm the accuracy of G2- and M-phase discrimination, a histone H3 phosphoepitope antibody was used to stain mitotic cells. The data in Fig. 2B show that essentially no mitotic cells are detected within the G2 gate and that a large proportion of cells within the M-phase gate contain phosphorylated histone H3. It should be noted that H3 dephosphorylation already begins during anaphase (Hendzel et al., 1997). Therefore, the fraction of H3 phosphoepitope-negative mitotic cells in the control population is readily explained by the presence of post-anaphase cells in this unsynchronized population. However, essentially all mitotic cells overexpressing MAP4 were H3 phosphoepitope-positive, which indicates that ectopic MAP4 blocks cells in a metaphase-like stage similar to paclitaxel treatment (Fig. 2B).

The distribution of MT content was determined within the G2- and M-phase gates validated above. The histograms derived from control and MAP4-expressing cells reveal relatively homogeneous distributions of MT content and, consistent with a previous study on normal cycling cells (Zhai et al., 1996), that the average MT content becomes somewhat reduced during mitosis (Fig. 2C; nonspecific fluorescence was <1% and is not shown). A comparison between an unsynchronous mitotic control population and cells blocked at mitosis by overexpressed MAP4 reveals that the latter results in a more symmetrical and narrow distribution of MT content. It is also noteworthy that this distribution is even narrower than the distribution of MT content among cells blocked in mitosis by the MT-stabilizing drug paclitaxel, again indicating the homogeneity of the cell population blocked by ectopic MAP4. Given that the majority of cells within each cell-cycle-stage-specific gate show a relatively tight distribution (Fig. 2C; data for G1 cells are not shown), the mean MT-specific fluorescence intensity provides a good estimate of the relative MT content at different stages of the cell cycle. A plot of these data reveals that ectopic MAP4, similar to paclitaxel, causes an increased average level of MT polymers in all phases of the cell cycle (Fig. 2D), which in turn shows that MAP4 stabilizes MTs all through the cell cycle. It should be noted that the observed cell-cycle fluctuations of MT content shown in Fig. 2D faithfully reproduce data of others (Zhai and Borisy, 1994; Zhai et al., 1996), with G2 cells having double the MT content of G1 cells, which is consistent with a doubling of cell mass, and with the MT content dropping somewhat during mitosis. Given that immunoblotting reveals no increase in the total cellular tubulin level in K562 cells within the time period analyzed (Fig. 1A) (Holmfeldt et al., 2002), it appears that ectopic MAP4 increases MT polymers in both interphase and mitotic K562 cells at the expense of the available pool of free tubulin. This stabilization seems likely to interfere with spindle formation on many different levels, including the process of centrosome separation.

We have shown previously that tubulin sequestering, but not catastrophe promotion, is functionally dominant over MAP4-mediated MT stabilization during the interphase of the cell cycle (Holmfeldt et al., 2002). To compare counteraction of MAP4 activity during interphase and mitosis, MAP4 was coexpressed with either the tubulin-sequestering Op18(25-149)-triA or catastrophe-promoting Op18(1-99)-tetraA derivatives outlined in Fig. 3A. Both of these Op18 truncation derivatives are non-phosphorylatable (i.e. they posses Ser to Ala substitutions at phosphorylation sites) and are consequently not inactivated by phosphorylation during mitosis, which is the fate of endogenous Op18 (Larsson et al., 1997; Marklund et al., 1996). For conditional coexpression, a pMEP-vector-based cotransfection approach was used that would allow stringently regulated expression of two gene products from the hMTIIa promoter (Gradin et al., 1998). As compared with our previous report (Holmfeldt et al., 2002), the ratio of pMEP DNAs was altered such that (1) the levels of coexpressed Op18 derivatives were increased at the expense of MAP4, and (2) the MT destabilization by the two derivatives when expressed alone is equivalent (Fig. 3B). These coexpression conditions caused a modest increase in MT content by ectopic MAP4 and an approximately 50% decrease in MT content by either of the two truncated Op18 derivatives (Fig. 3B). The data depict MT content determined specifically in G2 cells after 20 hours of induced expression, and it is clear that our previous findings on a mixed cell-cycle-stage population induced for 8 hours are faithfully reproduced under these modified conditions. Thus, coexpressed MAP4 did not counteract MT destabilization caused by the coexpressed tubulin-sequestering Op18(25-149)-triA derivative, whereas the action of the catastrophe-promoting Op18(1-99)-tetraA was counteracted (Fig. 3B).

Analysis of DNA profiles, shown in Fig. 4, confirms that the two Op18 truncation derivatives interfere with spindle formation to different extents, and only Op18(1-99)-tetraA causes a prominent mitotic block. To uncover both suppressive and enhancing effects on monoastral-mitotic phenotypes, the cotransfection conditions were optimized so that the MAP4 levels would not cause a complete mitotic block. Interestingly, the cell-cycle profiles of cotransfected cells show that the mitotic block mediated by MAP4 is somewhat suppressed by coexpressed Op18(25-149)-triA, whereas coexpressed Op18(1-99)-tetraA has the opposite effect (Fig. 4, compare - and + MAP4). This apparent suppression by coexpressed Op18(25-149)-triA was not due to a general cell-cycle arrest or potential interference with a metaphase checkpoint, as evidenced by accumulation of G2/M cells in the presence of paclitaxel (see insert in Fig. 4, top panels). Thus, MT destabilization by two distinct mechanisms has opposite effects on the potency shown by the MT-stabilizing MAP4 protein in blocking cell division.

Since the monoastral spindle is a hallmark of the mitotic phenotype of ectopic MAP4, we evaluated cotransfected mitotic cells with respect to monoastral or bipolar spindles. For simplicity, data on bipolar spindles are presented in Fig. 4 without consideration either of MT density or whether the spindles appeared aberrant or not. Consistent with our previous report (Holmfeldt et al., 2001) and the DNA profiles, the results show that expression of Op18(25-149)-triA alone only causes a modest increase in the mitotic index. Moreover, as previously reported (Holmfeldt et al., 2001), Op18(1-99)-tetraA causes a much more pronounced mitotic block, and a major fraction of all spindles were bipolar. These spindles were clearly abnormal, with MTs appearing as two small star-like asters (data not shown) (Holmfeldt et al., 2001). Analysis of cells coexpressing MAP4 with either of the Op18 truncation derivatives revealed a large increase in the frequency of mitotic cells. The appearance of the mitotic figures was heterogeneous, some of which were difficult to categorize as normal or aberrant. Nevertheless, a major proportion of the abundant bipolar spindles found in MAP4/Op18(1-99)-tetraA-coexpressing cells appeared similar to the abnormal mitotic figures of cells expressing the catastrophe-promoting Op18 derivative alone (data not shown) (Holmfeldt et al., 2001).

Monoastral spindles were scored as being either intermediate-to-large or small (see Fig. 4, lower panel). In the latter category, MTs appeared as a single, centrally located bright dot without or with very short, individually distinguishable MTs. Importantly, and in agreement with a partial relief of a mitotic block, the data in Fig. 4 demonstrate that coexpression of the tubulin-sequesterer Op18(25-149)triA with MAP4 causes a 50% reduction in the total frequency of monoastral spindles. The data also show that the frequency of intermediate-to-large monoastral spindles in MAP4-expressing cells is reduced by a factor of 4 to 5 by coexpressed Op18(25-149)-triA, which reveals a general decrease in the MT content of monoastral spindles.

As shown in Fig. 4, the fraction of small monoastral spindles in MAP4-overexpressing cells is increased by coexpression of Op18(1-99)-tetraA. However, despite this apparent MAP4-antagonizing destabilizing activity, the total frequency of monoastral spindles was unaltered and the potency of the mitotic block was further increased by the catastrophe-promoting Op18(1-99)-tetraA derivative. It follows that the observed suppression of MAP4-mediated monoastral-mitotic phenotype by the tubulin-sequesterer Op18(25-149)-triA cannot be explained by a general mechanism involving antagonistic activities. This suggests that suppression is caused by downregulation of MAP4 activity, which would be consistent with the observed inability of MAP4 to antagonize MT destabilization by Op18(25-149)-triA in interphase cells (Fig. 3B). To address whether the mechanism involves direct interactions with Op18(25-149)-triA, we lysed MAP4/Op18(25-149)-triA-Flag-coexpressing cells and used antibodies directed against the Flag-epitope tag to precipitate the Op18 derivative. These experiments revealed the expected co-precipitation of tubulin but not MAP4 (data not shown). Thus, it appears that MAP4 activity is not downregulated by direct interactions with Op18(25-149)-triA.

Mechanistically distinct types of MT destabilization have opposite effects on the concentration of free tubulin heterodimers. Thus, while catastrophe promotion increases the free tubulin concentration, tubulin sequestering has the reverse effect. It follows that suppression of MAP4 activity by the tubulin-sequestering Op18(25-149)-triA derivative might be explained if MAP4 is less active under conditions of reduced free tubulin concentrations. To assess directly the effect of decreased free tubulin concentrations on MAP4 activity, we took advantage of the RB3 neural member of the Op18/stathmin family, which binds tubulin with affinity that is orders of magnitude higher than that of Op18, and consequently forms much more stable tubulin-sequestering complexes (Charbaut et al., 2001). To localize sequestering complexes to a defined cellular location as far as possible from the mitotic spindle, RB3 was located specifically to the plasma membrane by creating a chimera between RB3 and the extracellular and transmembrane region of the T-cell-specific cell-surface protein CD2 (Fig. 5A). As a control, a CD2 chimera was prepared in which the RB3 part was replaced by a 95-residue C-terminal part of Op18 lacking significant tubulin affinity (CD2-Co). A confocal section through a mitotic cell shows that CD2-Co and CD2-RB3 were expressed to similar degrees at the plasma membrane, as detected by an anti-CD2 antibody (Fig. 5B). It is also evident that the CD2-RB3, but not CD2-Co, chimera causes membrane localization of a significant fraction of cytosolic tubulin (Fig. 5B).

To estimate alterations in the pool of cytosolic tubulin heterodimers in CD2-RB3-expressing cells directly, the average fluorescence intensity of cytosolic and membrane-associated tubulin was determined by confocal microscopy (see Materials and Methods). The data indicated that, on average, CD2-RB3 expression causes a 30% decrease in the diffuse staining of non-polymerized tubulin (Fig. 5C). The significance of this estimate is indicated by a clear negative correlation between the tubulin-specific fluorescence intensities at the plasma membrane and the cytosol among CD2-RB3-expressing cells, but not among CD2-Co-expressing cells (Fig. 5D). Given that an unknown fraction of nonpolymeric tubulin may be in a sequestered state, the estimated 30% decrease in soluble tubulin should be considered to be a minimum estimate of decreased free tubulin concentration.

Immunoblot analysis, using antibodies against a peptide corresponding to a completely conserved region among members of the Op18/stathmin family (anti-SLEEIQ), showed that the CD2-RB3 chimera is expressed at higher levels than endogenous Op18 (Fig. 6). Since the CD2 cell-surface receptor is extensively glycosylated, CD2 chimeras migrate as a broad heterogeneous band, as detected by either anti-CD2 or by anti-SLEEIQ. Arbitrary quantification of CD2-RB3 relative to endogenous Op18 was achieved from serial dilutions of cell lysates (data not shown). Integration of the heterogeneous band signal indicated that CD2-RB3 is expressed at a fourfold higher level than endogenous Op18. On the basis of previous estimates of cytosolic concentration of Op18 in K562 cells (i.e. 10 μM) (Larsson et al., 1999), it can be predicted that the CD2-RB3 chimera is expressed at approximately 40 μM, which should be compared with an estimated total tubulin heterodimer concentration of 23 μM in K562 cells (Larsson et al., 1999). These approximations suggest that CD2-RB3 is expressed at an approximately twofold molar excess over total tubulin. Under these conditions of decreased free tubulin concentrations, determination of MT content in interphase (G2) cells revealed a 65% reduction in polymerized tubulin and, most significantly, that this drop is not counteracted by coexpressed MAP4 (Fig. 6).

The RB3 protein lacks the cyclin-dependent kinase phosphorylation sites of native Op18 and is not inactivated by phosphorylation during mitosis. Analysis of cell-cycle profiles revealed that expression of CD2-RB3 causes a 2.5-fold increase in the mitotic index as compared with Vector-Co cells (Fig. 7, percentage of mitotic cells is shown below the cell-cycle profiles). This result is not surprising given the high tubulin-sequestering potential of this derivative, as revealed by its potent destabilization of interphase MTs. The DNA ratio used for coexpression was adjusted such that MAP4 expression caused a major accumulation of G2/M cells. Significantly, coexpression of the CD2-RB3 chimera potently and specifically suppressed this accumulation of G2/M cells. This was not due to a general cell-cycle arrest, as shown by accumulation of G2/M cells in the presence of paclitaxel (see insert in Fig. 7, top panels).

Cotransfected cells were evaluated with respect to monoastral and bipolar spindles according to the criteria described in Fig. 4. It is shown in Fig. 7, lower panel, that essentially all mitotic CD2-RB3-expressing cells contained bipolar spindles. Under these conditions, expression of MAP4 alone or coexpressed with CD2-Co resulted in an essentially homogeneous population of mitotic cells with monoastral spindles of intermediate-to-large size. Consistent with the cell-cycle profile, this phenotype of MAP4 was efficiently suppressed by the tubulin sequestration-specific CD2-RB3 derivative.

It seems reasonable to assume that suppression of monoastral spindles and consequent formation of functional spindles is a stringent criterion for downregulation of MAP4 activity by the coexpressed CD2-RB3 chimera. This chimera has a defined plasma membrane location, which precludes direct contact with the mitotic machinery. As expected, analysis of MAP4-specific fluorescence revealed that MAP4 was present both in the cytosol and in association with MTs, and this pattern of localization appeared to be unaltered by CD2-RB3 expression (data not shown). Thus, since MAP4 and CD2-RB3 did not colocalize, the data indicate that the decline in free tubulin concentration was a direct cause of downregulating MAP4 activity.

Previous studies have revealed extensive phosphorylation of MAP4 at multiple sites during mitosis (Ookata et al., 1995; Ookata et al., 1997). On the basis of mobility shift on low-percentage SDS-PAGE, MAP4 is extensively phosphorylated at mitosis even at the overexpression levels that cause mitotic arrest of K562 cells (Fig. 1). The functional implications of mitotic phosphorylation of MAP4 remain unclear and it appears that MAP4 is equally associated with MTs during both interphase and mitosis (Bulinski and Borisy, 1980; Ookata et al., 1995), but it is still possible that the relative affinity of MAP4-MT interactions is decreased by mitotic phosphorylation (Chang et al., 2001). Overexpressed MAP4 has the potential to regulate the MT system in all cell-cycle phases, as shown by increased MT content during both interphase and mitosis (Fig. 2D). This provides a general mechanism for the observed block in centrosome separation and the associated arrest in a metaphase-like stage. Since excessive stabilization of the MT system by MAP4 can be predicted to interfere at many levels during the complex process of spindle formation, the block in centrosome separation is most probably only one of several deleterious effects. Nevertheless, regardless of the mechanism involved, the monoastral spindle provides a clear-cut homogeneous overexpression phenotype, and was therefore used in the present study as a marker for excess MAP4 activity during mitosis. It is striking that numerous reports of MAP4-overexpression phenotypes have not noted generation of monoastral spindles previously [Chang et al. (Chang et al., 2001), and references therein]. At present, the extent to which these differences can be explained by different ectopic expression systems is unclear, as is the question of whether the suspension-growing human leukemia/lymphoma cell lines analyzed in the present study (Fig. 1D) are in general more sensitive than other cell lines to excess MAP4 activity.

The net balance of opposing MT-regulatory factors seems likely to determine the length of MTs, and thus the total MT content (Andersen and Wittmann, 2002; Kinoshita et al., 2001; Tournebize et al., 2000). We have previously shown that MAP4 counteracts MT destabilization by XKCM1 and a C-terminally truncated Op18 derivative (two catastrophe promotors that both lack significant tubulin-sequestering activity), but not MT destabilization caused by tubulin-sequestering competent regulators (Holmfeldt et al., 2002). This suggested that MAP4 differentiates between mechanistically distinct types of MT destabilization. However, from these analyses of the MT content of interphase cells, it could not be excluded that MAP4 coexpressed with tubulin-sequestering regulators modulated MT dynamics without altering cellular MT content to a significant extent. However, in mitotic cells, interference with dynamic parameters of MTs by coexpressed stabilizing/destabilizing proteins will cause a mitotic block. Thus, if the combined phenotype of coexpressed ectopic MT regulators is manifested as a suppression of MAP4-mediated spindle defects to allow subsequent cell division, it seems reasonable to assume that essential dynamic parameters of spindle MTs in coexpressing cells are close to the normal range.

In the present study, counteraction of ectopic MAP4 activity was evaluated in mitotic cells by determining (1) suppression of the monoastral phenotype, (2) size of monoastral spindles, and (3) formation of functional dipolar spindles. Our result demonstrated that both catastrophe-promoting and tubulinsequestering Op18 derivatives antagonize MT stabilization by MAP4 in mitotic cells, since coexpression in both cases altered the ratio of monoastral spindles with intermediate-to-large and small MT content (Figs 4, 7). Given that only the tubulinsequestration-competent derivatives suppressed MAP4-mediated formation of monoastral spindles and facilitated cell division in coexpressing cells, it seems clear that an antagonizing activity is not sufficient to suppress the deleterious MT-stabilizing effect of excess MAP4 activity during mitosis. This does not seem surprising, considering that spindle formation and subsequent cell division involves a complex series of events requiring finely tuned regulation of MT dynamics. Thus, the improbability that the spindle-disrupting effect of ectopic MAP4 can be partially reversed by an antagonizing MT-destabilizing activity provided an indication that the cytosolic tubulin-sequestering Op18(25-149)-tetraA derivative might suppress the mitotic phenotype by downregulation of MAP4 activity.

To link downregulation of MAP4 activity directly to a decrease in free tubulin concentrations requires a high-affinity tubulin-sequestering molecule that is spatially separated from both the MT system and the cytosolic/MT-associated MAP4 protein. In interphase cells, the MT system is in extensive contact with the entire cytoplasmic space, whereas mitotic MTs are mainly located in the vicinity of chromosomes and have minimal contact with the plasma membrane. Thus, the transmembrane CD2-RB3 chimera provides an ideal tool for studying how decreased free tubulin concentrations modulate ectopic MAP4 activity in a mitotic cell. Expression of CD2-RB3 alone causes only a modest (2.5-fold) increased frequency of mitotic cells (Fig. 7). This weak mitotic phenotype may appear surprising, but is in line with the previously observed resistance of spindle formation to tubulinsequestering conditions (Holmfeldt et al., 2001). The relative robustness of spindle formation under tubulin-sequestering conditions greatly simplified interpretation of data from coexpression experiments that, taken together, provide compelling evidence that the activity of ectopic MAP4 is downregulated by the coexpressed CD3-RB3 derivative. Since CD2-RB3 and MAP4 are spatially separated in mitotic cells, it seems reasonable to conclude that downregulation of MAP4 activity is directly coupled to decreased free tubulin concentrations. Moreover, since a functional spindle can still be formed at these lowered concentrations of free tubulin, albeit at a decreased rate, our results indicate that the activity of MAP4 is modulated within the physiological range of free tubulin concentrations.

The free concentration of tubulin heterodimers modulates the rate of tubulin gene expression in many, but not all, cellular systems (Cleveland et al., 1981). Paclitaxel or MAP4-mediated over-polymerization of MTs or tubulin sequestration does not result in a detectable increase in tubulin synthesis within a 24-hour time period in the leukemic K562 cell line (Holmfeldt et al., 2002). This property of K562 greatly simplifies interpretation of data in the present study. However, the specific mechanism behind downregulated MAP4 activity by lowered free tubulin concentration is still unclear. Analysis of MAP4-specific fluorescence reveals no obvious redistribution of cytosolic and MT-associated MAP4 in cells expressing tubulin-sequestering regulators (data not shown). Moreover, downregulated MAP4 activity could in principle be explained by either (1) direct physical association with MAP4, (2) phosphorylations by a putative tubulin-sequestration-regulated protein kinase, or (3) reduction of the MT-stabilizing properties of MAP4 at lowered free tubulin concentrations. Given that CD2-RB3 and MAP4 are spatially separated during mitosis, it is highly improbable that direct association suppresses MAP4 activity. However, we are presently unable to distinguish between the two remaining alternatives. For example, lowered tubulin concentrations may activate members of the MAP/microtubule affinity regulating kinase (MARK) family, which have been shown to phosphorylate and thereby downregulate the activity of several MAPs including MAP4 (Drewes et al., 1997; Ebneth et al., 1999). Alternatively, it is also possible that low free tubulin concentration might directly limit MAP4 activity. Thus, increased knowledge on the mechanism by which MAP4 stabilizes MTs, combined with functional dissection of MAP4 phosphorylation sites, is required to resolve these issues.

Downregulation of MAP4 activity at lowered free tubulin concentrations in K562 cells has general implications for regulation of MT dynamics in intact cells. For example, excessive polymerization will lead to deceased MAP4 activity, whereas depolymerization by catastrophe promotion will increase MAP4 activity. Such a negative autoregulatory loop might contribute to the homeostasis of the MT system. Given that the free tubulin concentration could vary in different cell types (e.g. by differences in tubulin partitioning between monomer-polymer pools), the present results provide a rationale for explaining the lack of consensus among previous reports describing the consequences of ectopic MAP4-overexpression phenotypes.

We thank Lynne Cassimeris and Victoria Shingler for helpful discussions and critical reading, and Jeannette Chloe Bulinski and Allan Williams for providing reagents. This work was supported by the Swedish Research Council.