Autonomous and phosphorylation-responsive microtubule-regulating activities of the N-terminus of Op18/stathmin

Microtubules (MTs) are polar polymers composed of α/β tubulin heterodimers that are inherently dynamic, a property referred to as dynamic instability (for a review, see Desai and Mitchison, 1997). Two protein families have been identified that increase the dynamics of microtubules (MTs) in animal cells. The kinesin-related protein XKCM1 and oncoprotein 18/stathmin (Op18) represent prototypes for each of these two families (for reviews, see Cassimeris, 1999; Walczak, 2000).

Op18 is a cytosolic protein that promotes transition from a growing to a shrinking MT; this transition is referred to as a catastrophe. Op18 appears to lack a defined tertiary structure in solution but can be divided into three distinct regions, namely a largely unstructured N-terminus and a tandem repeat of two weakly homologous α-helical regions (see Fig. 1A)(Gigant et al., 2000; Wallon et al., 2000). Op18 binds soluble tubulin, and the overall shape of the complex has been revealed by transmission electron microscopy and a low-resolution X-ray structure(Gigant et al., 2000; Steinmetz et al., 2000). The Op18-tubulin complex can be described as two tubulin heterodimers in a slightly curved head-to-tail alignment (i.e. tandem-tubulin dimers) with each one of the tandem Op18 helical repeats contacting one heterodimer. These tandem tubulin dimers are also stabilized by longitudinal α and βtubulin subunit interactions. In accordance with these cooperative interactions within the ternary complex, Op18 binds two tubulins according to a two-site positive cooperative model, which minimizes complexes with single tubulin heterodimers (Larsson et al.,1999). The 4 Å resolution X-ray structure allowed unambiguous alignment of the relative locations of the subunits and thus rejection of alternative models of the complex(Segerman et al., 2000; Wallon et al., 2000). The N-terminus was not visible in the 4 Å structure, and the resolution was insufficient to resolve the orientation of the extended Op18 helix relative to the tandem tubulin dimers. However, it was recently shown that a peptide corresponding to the N-terminal 10 residues of Op18 can be cross-linked close to helix 10 of α-tubulin, which is involved in both longitudinal and lateral contacts within the MT lattice(Muller et al., 2001). This suggests that the N-terminus of Op18 is located at the α-tubulin subunit end of the tandem tubulin dimers.

The β-tubulin subunit of the heterodimer contains an exchangeable GTP site, termed the E site, and hydrolysis at the E site provides the energy that drives the dynamic properties of MTs. Within the MT polymer, hydrolysis of the E-site-bound GTP of β-tubulin is believed to be triggered by a catalytic Glu residue in a loop on the preceding α-tubulin subunit(Nogales, 2001). We have shown that Op18-tubulin complex formation inhibits nucleotide exchange and stimulates GTP hydrolysis (Larsson et al.,1999). On the basis of these data and the observed longitudinal arrangement of the two tubulin heterodimers, it has been proposed that Op18 stimulates tubulin GTPase activity by promoting interactions between the E site of β-tubulin and the catalytic loop on α-tubulin within the tandem tubulin dimer complex (Steinmetz et al., 2000).

Three major models have been proposed for how Op18 destabilizes MTs, namely that (i) Op18 acts as a pure tubulin-sequestering protein(Curmi et al., 1997; Jourdain et al., 1997), (ii)Op18 is a specific catastrophe promotor(Belmont and Mitchison, 1996)and (iii) Op18 mediates at least two distinct activities, namely catastrophe promotion, which requires the N-terminus of Op18, and a tubulin-sequestering activity observed in vitro, which requires both of the helical repeats but not the N-terminus (Howell et al.,1999b; Larsson et al.,1999). Recent structural data readily explain how the two helical repeats of Op18 form a tubulin-sequestering complex(Gigant et al., 2000). However,the role of the unstructured N-terminus is still elusive but, given that three out of four Ser phosphorylation sites are located in this region, it has been speculated that the N-terminus is of regulatory importance (for a review, see Lawler, 1998). Consistently,Op18 activity is switched off by either multi-site phosphorylation during mitosis (Larsson et al.,1997), which is essential to allow cell division(Marklund et al., 1996), or in response to single or dual phosphorylation by two distinct kinase systems during interphase of the cell cycle (Gradin et al., 1998; Melander Gradin et al., 1997).

Here we demonstrate phosphorylation-regulated autonomous tubulin-directed activities by the N-terminus of Op18. The data suggest that autonomous N-terminal interactions with tubulin have a different functional outcome from formation of a sequestering ternary Op18-tubulin complex by the first and second helical repeats of Op18. Thus, the results provide a framework for better understanding of the reported functional dichotomy of Op18.

Construction of truncated Op18-derivatives encoding amino acids 1-99[Op18(1-99)] and amino acids 46-99 [Op18(46-99)] have previously been described (Segerman et al.,2000). As a part of the general cloning strategy, the two most C-terminal residues (i.e. residue 148 and149) were preserved in all C-terminally truncated Op18 derivatives. For simplicity, these two amino acids are ignored in the naming of derivatives in the present study, and for consistency, the derivative previously denoted Op18-Δ100-147 is renamed to Op18(1-99). Op18(1-57) `non-related helix' derivatives (NR-helix) were constructed by PCR using the 5′-primer 5′-CCG GGA GCT CCT GCA GCC CTG GCC AGA GTG GAA GAG-3′ and the 3′-primer 5′-AGA ACT AGT GGA TCC CTA TTA CAG CAC CTT CAC CTC GTT G-3′ with a cDNA encoding the non-muscle myosin heavy chain [NMMHC-A isoform(Simons et al., 1991)] as a template. The coding region of the resulting PCR fragment covers amino acids 1088 to 1237 of the rod region of NMMHC-A, and the primers introduce a PstI site in the 5′ end and BamHI site in the 3 end. The digested PCR fragment was used to replace the corresponding fragment of Op18 cDNA in pBluescript SK(+). The resulting derivative,pBS-Op18(1-57)-NR-helix, contained the coding region for the first 57 residues of Op18 fused to a 149 residue part of the rod region of NMMHC-A. To construct Op18-triA(1-57)-NR-helix, in which the Ser-16, Ser-25, Ser-38 phosphorylation sites are substituted with Ala, Op18(25-57)-NR-helix and control NR-helix, the NR-helix PstI to BamHI fragment was used to replace the corresponding fragments of pBluescript derivatives of Op18-S16,25,38A,Op18-Δ4-24-F and Op18-Δ4-55-F(Holmfeldt et al., 2001). It should be noted that in each one of the resulting NR-helix derivatives, the first three residues (Met, Ala, Ser) of Op18 were included in the fusion. The coding sequence of the PCR-generated fragment was confirmed by nucleotide sequence analysis using an ABI PRISM dye terminator cycle sequencing kit from Perkin Elmer. Op18-NR-helix derivatives were expressed in E. coliusing pET-3d and purified by a combination of ion-exchange chromatography and gel filtration as previously described for native Op18(Brattsand et al., 1993). The Op18(1-57) derivatives was prepared by inserting a double-stranded oligonucleotide into the PstI site of human Op18, which introduced two consecutive stop codons after residue 57. The Op18(1-57) protein was expressed in E. coli using pET-3d with a six-residue His-tag at the N-terminus. The recombinant protein was absorbed onto a HiTrap chelating HP column (Amersham Pharmacia Biotech) and eluted using a linear gradient of imidazole as recommended by the manufacturer.

Analysis of tubulin GTPase activity was performed in PEM buffer adjusted to pH 6.8 with NaOH (80 mM piperazine-N,N′-bis[2-ethanesulfonic acid], 1 mM EGTA, 4 mM Mg²⁺) containing 5 mM adenyl-5′-yl imidodiphosphate (AMP-PNP; to inhibit non-specific ATPase activity) as described previously (Segerman et al.,2000). In brief, tubulin (TL238, Cytoskeleton, Inc) was incubated with [α-³²P]GTP, the resulting tubulin-α-[³²P]GTP complexes recovered by centrifugation through a desalting column (P-30 Micro Bio-Spin, Bio-Rad) and single-turnover GTP hydrolysis followed at 37°C for up to 40 minutes. Control experiments showed that the Op18 preparations used neither bind nor hydrolyze[α-³²P]GTP or cause dissociation of tubulin bound[α-³²P]GTP. Nucleotide hydrolysis was quantified by ascending chromatography as described previously(Segerman et al., 2000), which allows reproducible analysis of less than 0.2% hydrolysis of[α-³²P]GTP.

The KA8 cell line, a transfected derivative of K562 cells, expresses an ectopic integrin α8 subunit (provided by Louis Reichardt)(Muller et al., 1995). By plating this transfected K562 derivative on a plastic surface coated with the bacterial Yersinia pseudotuberculosis invasin protein, which activates β1-intergins (Arencibia et al., 1997), the otherwise spherical KA8 cells spread out on the surface, which facilitated morphological analysis of mitotic cells. For expression of Op18 deletion mutants in human cell lines, coding regions were cloned as HindIII to BamHI fragments into the corresponding sites of the EBV-based shuttle vector pMEP4 (Invitrogen)(Groger et al., 1989). Transfection of the pMEP4 derivatives into KA8 cells was performed as previously described for the original K562 cell line(Marklund et al., 1994). Conditional expression was achieved by employing the hMTIIa promoter, which can be suppressed by low concentrations of EDTA (20 μM) and induced by Cd(Marklund et al., 1994). For transfection, 12 μg DNA was used, and expression was induced with 0.5 μM Cd.

To analyze Op18-NR-helix phosphoisomers, we employed a native PAGE system that separates Op18 according to the charge differences introduced by each of the four identified phosphorylations(Marklund et al., 1993). To determine ectopic expression levels, cell extracts were separated by SDS-PAGE together with graded amounts of recombinant Op18 and Op18-NR-helix proteins as described previously (Marklund et al.,1994). The ratio of ectopic proteins versus endogenous Op18 was determined by probing western blots with affinity-purified rabbit antibodies and comparing them with a standard of recombinant proteins with known protein content (Brattsand et al.,1993).

The cellular content of MT polymers was determined essentially as described previously (Holmfeldt et al.,2001). In brief, cells were resuspended in an MT-stabilizing buffer containing 0.05% saponin, to extract soluble tubulin, and subsequently fixed in 4% paraformaldehyde/0.05% glutaraldehyde. The remaining polymerized tubulin was stained with anti-α-tubulin (clone B-5-1-2, Sigma) and fluorescein-conjugated rabbit anti-mouse immunoglobulin. Fluorescence was quantified by flow cytometry using a FACS-caliber together with the Cell Quest software (Becton-Dickinson). To allow calculation of the percentage of polymerized tubulin in relation to the total amount of cellular tubulin, the polymerization status in vector-control-transfected cells was determined by quantitative western blotting (mean of three independent determinations:58±12%). Relative fluorescence intensities of extracted cells were normalized in each experiment assuming that the level in vector-control cells corresponds to 58% polymerized tubulin. This procedure faithfully reproduced the results obtained by quantification of soluble and particulate tubulin by western blot analysis (Marklund et al.,1996) but with increased reproducibility. Within the time limits of the experiments, ectopic Op18 does not alter the cellular levels of total tubulin in K562 or KA8 cells. Analysis of DNA content and quantification of mitotic cells, using the MPM-2 antibody, was performed by flow cytometric analysis as described previously (Marklund et al., 1996). Immunofluorescence analysis was performed as described elsewhere (Holmfeldt et al.,2001).

The assembly of individual MTs seeded from axoneme fragments was visualized using video-enhanced differential interference contrast (DIC) microscopy as described previously (Howell et al.,1999b; Vasquez et al.,1994). Assembly conditions were such that MTs assembled only from axonemes, and the total amount of tubulin incorporated into MT polymer was insignificant compared with the total tubulin concentration. A PEM buffer adjusted to pH 7.5 was used in the present study. To prevent any possible adherence of the NR-helix to the slide or coverslip, glass surfaces were blocked with 5 mg/ml casein for 5 minutes(Vasquez et al., 1994). We estimate that the detection limit for MTs assembled from axonemes was approximately 0.3 μm.

Extensive C-terminal truncations into the first helical repeat of Op18 result in unstable polypeptides and consequently very low expression levels in mammalian cells (data not shown). Therefore, to determine potential MT regulatory properties of the N-terminus in human cell lines it was necessary to construct a fusion derivative. For this purpose we replaced the major part of the extended helix of Op18 with a 149-residue α-helical portion from the rod region of non-muscle myosin heavy chain, which lacks significant homology to the helical region of Op18 (9.8% identity if aligned at the point of fusion and without gaps) and is therefore termed a non-related helix(NR-helix). The Op18(1-57)-NR-helix derivative was prepared with either the wild-type sequence or with the three N-terminal Ser phosphorylation sites substituted with Ala (Ser-16, Ser-25, Ser-38, termed Op18-triA(1-57)-NR-helix)to obtain a derivative that could not be phosphorylated. To confirm that the NR-helix does not modulate interactions between the N-terminus of Op18 and tubulin, we also prepared a protein derivative of Op18 that only contained the first 57 residues Op18(1-57). This derivative, although unstable in mammalian cells, is produced in sufficient quantities to allow purification when expressed in E. coli. The Op18 derivatives used in this study are depicted in Fig. 1A together with an outline of the general structure of the tubulin-binding Op18 protein.

To determine if the first 57 residues of Op18 are sufficient for functional interaction with tubulin, modulation of tubulin GTP hydrolysis was compared with that of wild-type Op18 (Op18-wt) and the C-terminally truncated Op18(1-99) derivative (Fig. 1B). The result reveals the expected threefold stimulation of tubulin GTPase hydrolysis in the presence of Op18-wt. This stimulation has previously been shown to be independent from the N-terminal 46 residues of Op18 (Larsson et al., 1999),and the mechanism may involve GTPase-productive interactions between the E site of β-tubulin and the catalytic loop of α-tubulin within the tandem tubulin dimer complex (Steinmetz et al., 2000). Moreover, we also show that the Op18(1-99) derivative,which lacks the second helical repeat, has the opposite activity, namely inhibition of the basal GTP hydrolysis of soluble tubulin. Inhibition of basal tubulin GTP hydrolysis by C-terminally truncated Op18 has previously been shown to be a consequence of loss of cooperative binding of the two tubulin heterodimers. Thus, the truncated Op18 derivative is inefficient in forming tandem tubulin dimer complexes and binds preferentially to single tubulin heterodimers, which in turn inhibit the basal GTP hydrolysis of tubulin in solution (Segerman et al.,2000). Given that the isolated first helical repeat by itself is inactive [see Op18(46-99), Fig. 1B], it appears that the N-terminus is essential for the observed inhibition by Op18(1-99).

Taken together, the data in Fig. 1 indicate that the N-terminus of Op18 retains functional interactions with tubulin. The low affinity of the Op18(1-57)-NR-helix, as indicated by the dose response, suggests that the first helical repeat of Op18 serves to increase the affinity of the isolated N-terminus towards tubulin heterodimers. It follows that the Op18(1-99) derivative can be functionally dissected into an N-terminus that blocks basal GTP hydrolysis and the first helical repeat that increases the otherwise very low binding affinity of the N-terminus to single tubulin heterodimers. The presence of both the first and second helical repeat, as in Op18-wt, further increases the tubulin-binding affinity by two-site positive cooperative binding along two tubulin heterodimers, which in turn result in a GTPase-productive complex.

Op18 acts as a specific catastrophe factor during MT assembly in vitro at pH 7.5, a property that requires an intact N-terminus(Howell et al., 1999a). To evaluate whether the N-terminus alone is sufficient for catastrophe promotion,individual MTs were analyzed in the presence or absence of Op18-NR-helix derivatives. Low concentrations of these derivatives (2 and 10 μM) were without significant activity (data not shown), which was not unexpected given that the N-terminal fragment appeared to have very low tubulin-binding affinity as evaluated by modulation of tubulin GTP hydrolysis (see Fig. 1B). However, at 30 and 60μM, the Op18(1-56)-NR-helix caused a twofold and sixfold increase in the rate of catastrophe (Fig. 2,lower panel). At these concentrations Op18(1-57)-NR-helix also caused a slight but significant reduction in growth rate(Fig. 2, upper panel). This effect on growth rates may represent some steric interference with tubulin polymerization at the high concentrations used, which would be expected from low-affinity inhibition of basal GTP hydrolysis of soluble tubulin shown in Fig. 1B (similar levels of inhibition were obtained at pH 6.8 and pH 7.5, data not shown). Importantly,doubling the concentration of Op18(1-57)-NR-helix did not further decrease the growth rate but did increased the catastrophe rate almost threefold. This suggests that a major part of the observed catastrophe promotion at the highest concentration is specific and not due to growth rate inhibition via a simple sequestering-type mechanism. As a control for specificity, a derivative lacking the first 24 residues of the N-terminus (Op18(25-57)-NR-helix) was analyzed at 60 μM. This derivative appeared to be essentially inactive since it has no effect on the growth rate and only a minor effect on the catastrophe rate. Hence, the data suggest that high concentrations of the N-terminus alone are sufficient for catastrophe promotion in vitro.

It is shown in Fig. 3A that all the Op18-NR-helix derivatives tested can be rapidly induced from the hMTIIa promotor in transfected KA8 cells to produce 35- to 50-fold higher expression levels than endogenous Op18. This is substantially higher than the level of native Op18 expressed from the same expression system, which indicates the high stability of the Op18-NR-helix protein(Fig. 3A). Native PAGE of Op18 can be used to determine the stoichiometry of Op18 phosphorylation owing to the contribution of two negative charges from each phosphate group, which has a strong impact on the migration of a small protein. It is shown in Fig. 3B that the majority of Op18(1-57)-NR-helix is phosphorylated on up to three distinct sites in cells blocked in mitosis by the MT-destabilizing agent nocodazole. As expected, the phosphorylation-site-deficient Op18-triA(1-57)-NR-helix derivative migrates at the position of the non-phosphorylated Op18(1-57)-NR-helix. Hence, we conclude that replacement of the first and second helical repeats of Op18 with the NR-helix allows high level expression and specific phosphorylation of all three physiologically relevant sites at the N-terminus of Op18.

Inducible and robust overexpression of Op18-NR-helix derivatives in a human leukemia cell line allows analysis of a phenotype on the level of the MT system. As shown in Fig. 4,shortly after induced expression Op18-wt causes essentially complete destabilization of MTs, whereas Op18(1-99) is somewhat less efficient. It should be noted that native Op18 and Op18 with the second helical repeat removed, that is Op18(1-99), are expressed at similar levels(Larsson et al., 1999), which are twofold to threefold lower than the high expression levels of NR-helix-fused derivatives (Fig. 3A). It is evident from Fig. 4 that expression of either Op18(1-57)-NR-helix or non-phosphorylatable Op18-triA(1-57)-NR-helix at these high levels causes a less dramatic but still significant destabilization of MTs. The destabilizing activity of Op18(1-57)-NR-helix is sensitive to a 24-residue deletion from the N-terminal end as evidenced by the Op18(25-57)-NR-helix derivative, which appears as inactive as the NR-helix alone. The observation that Op18(1-57)-NR-helix is less active than Op18(1-99) in intact cells is in line with in vitro data on inhibition of basal GTP hydrolysis of soluble tubulin(Fig. 1B).

It is evident from the flow cytometric analysis shown in Fig. 4B that, although the Op18(1-57)-NR-helix derivative is much less efficient than wild-type Op18, 6 hours of expression of either derivative causes a homogeneous decrease in MT content in the cell population. A 6 hour period is too short a time to detect a mitotic block caused by interference with mitotic spindle MTs, and it follows that the results at this early time point reflect the MT content in interphase cells.

Earlier studies have shown that Op18 is phosphorylated at multiple site with high stoichiometry at mitosis, which results in its inactivation(Larsson et al., 1997; Larsson et al., 1995). Hence,owing to mitotic phosphorylation, ectopic expression of wild-type Op18 destabilizes the interphase array of MTs without interfering with spindle formation during mitosis. However, mutants with the N-terminal Ser phosphorylation sites substituted with Ala are non-phosphorylatable and consequently constitutively active, and such mutants block formation of the mitotic spindle (Marklund et al.,1996). The data in Fig. 4 shows that the overexpressed N-terminus of Op18 is sufficient to significantly destabilize the interphase array of MTs. To interpret these data it was important to know if the N-terminus of Op18, which has been taken out of its normal context of the first and second helical repeat, is (i) still regulated by phosphorylation and (ii) is sufficient to block formation of the mitotic spindle. For this purpose the cell cycle profiles of transfected cells were evaluated after 24 hours of induced expression(Fig. 5). As expected from the low degree of leakage from the hMTIIa promotor, no significant alteration of the DNA profile is observed prior to induced expression. However, after 24 hours of induced expression, it is evident that the phosphorylation-site-deficient Op18-triA(1-57)-NR-helix derivative causes a major accumulation of cells in the G2/M phase, whereas the Op18(1-57)-NR-helix causes only a minor shift in the DNA profile. Since cells overexpressing Op18(1-57)-NR-helix readily accumulate in mitosis in the presence of the MT-directed drug nocodazole (see insert on Fig. 5), it is clear that the data are not biased by a potential general cell cycle block or interference with a metaphase checkpoint by this derivative. Finally, as expected from the inability of the Op18(25-57)-NR-helix and NR-helix to interfere with interphase MT, these two derivatives did not cause any detectable alteration of the DNA profile.

To evaluate how the non-phosphorylatable Op18-triA(1-57)-NR-helix derivative causes accumulation of cells with G2/M content of DNA, mitotic cells and mitotic spindle morphologies were manually inspected and graded into three classes (normal and abnormal type I and II) described and depicted in Fig. 6. The percentage of normal versus abnormal spindles was quantified and normalized to the frequency of mitotic cells (Table 1). The data show that only the Op18-triA(1-57)-NR-helix causes a significant increase in the frequency of mitotic cells. It should be noted that K562 cells have poor spindle assembly checkpoint control, so that interference with the mitotic spindle only results in a transient block in metaphase, which is followed by entry into a tetraploid pseudo-G1 state with G2/M content of DNA(Marklund et al., 1996). In the present study we have employed the chicken integrin α8 subunit expressing KA8 subclone of K562 (Muller et al., 1995), since these cells spread on plastic coated with a integrin-β1-activating ligand and thereby facilitate morphological examination of mitotic cells. However, owing to the very transient metaphase checkpoint of the KA8 subclone, only about 10% of all cells are blocked at metaphase after 24 hours of induced expression(Table 1). Nevertheless, it is still clear that overexpression of Op18-triA(1-57)-NR-helix interferes dramatically with spindle assembly since the majority of metaphase cells (96%)show severe spindle abnormalities. Taken together, the results in Fig. 6 and Table 1 show that non-phosphorylatable Op18-triA(1-57)-NR-helix causes accumulation of cells with G2/M content of DNA owing to potent interference with the mitotic spindle. As expected by mitosis-specific phosphorylation and inactivation of Op18-wt (Larsson et al.,1997), the Op18(1-57)-NR-helix derivative is extensively phosphorylated in M-blocked cells (Fig. 3B) and shows minimal interference with the mitotic spindle(Table 1). Since destabilization of interphase MTs is not dependent on mutated N-terminal phosphorylation sites, it follows that the N-terminus of Op18 can be regulated by mitotic phosphorylation independently from the major tubulin-binding regions represented by the first and second helical repeats. From this finding it can be deduced that Op18 phosphorylation can regulate Op18 activity by a mechanism that is independent from formation of the tandem tubulin dimer complex. Finally, the morphological appearance of the majority of spindles in Op18-triA(1-57)-NR-helix-expressing cells (i.e. type II represented by asters containing dense short MTs) is compatible with autonomous high-level catastrophe-promoting activity by the overexpressed non-phosphorylatable Op18-N-terminus.

The present study demonstrates that an N-terminal region of Op18, which encompasses the unstructured N-terminal 46 residues and 11 residues of the first helical repeat, is sufficient and directly responsible for inhibition of the low-rate basal tubulin GTP hydrolysis in solution. The basal tubulin GTP hydrolysis, which occurs under non-polymerizing conditions at 37°C, is linearly dependent on the free tubulin concentrations and approaches zero at very low concentrations (i.e.= first-order with respect to tubulin GTP concentration) (Carlier et al.,1997). This indicates a requirement for interactions between tubulin heterodimers that involves the catalytic loop located onα-tubulin and the E site on β-tubulin. The simplest explanations for the observed inhibitory activity by the Op18 N-terminal portion is either binding close to and consequent masking of the catalytic loop located onα-tubulin or, alternatively, directing modulation of the E site onβ-tubulin. Since cross-linking experiments suggest that the N-terminus of Op18 is located at the α-tubulin of the tandem tubulin dimers(Muller et al., 2001), the former hypothesis appears the most likely. Thus, the N-terminus of Op18 may inhibit basal tubulin GTP hydrolysis by masking the catalytic loop and thereby indirectly protecting the E site from GTPase-productive interactions. It follows that native Op18 has the potential to both stimulate tubulin GTP hydrolysis by formation of GTPase-productive tandem tubulin dimer complexes and inhibit basal tubulin GTP-hydrolysis by a mechanism that may involve masking of the catalytic loop located on α-tubulin.

A fragment containing only the first 57 residues of Op18 has only low binding activity towards tubulin, as evidenced by the high concentrations required for inhibition of basal tubulin GTPase activity(Fig. 1B). The low tubulin-binding affinity of the N-terminal portion precluded direct analysis of binding; however, derivatives containing both the N-terminus and the entire first helical repeat [i.e., Op18(1-99)] bind tubulin with an appreciable affinity (Larsson et al.,1999). Given the 10- to 20-fold difference in dose response between Op18(1-57) and Op18(1-99), and the fact that the first helical repeat in isolation is devoid of activity (see inhibition of basal GTP hydrolysis, Fig. 1B), it seems likely that the first helical repeat simply stabilizes binding of the N-terminus.

The present study demonstrates that the N-terminal region of Op18 expresses autonomous microtubule-destabilizing activity in intact cells. It has been shown that Op18 is inactivated by phosphorylation at four Ser residues during mitosis, which allows formation of the mitotic spindle in Op18-overexpressing cells (Larsson et al., 1997). The present study shows that the N-terminus expressed in the absence of tandem helical repeats can also be phosphorylation-inactivated during mitosis. Thus,expression of the Op18-triA(1-57)-NR-helix resulted in a mitotic block,whereas the corresponding phosphorylatable wild-type sequence allowed productive spindle formation during mitosis. Although overexpression of the N-terminus alone is sufficient for destabilization of interphase MTs and the mitotic spindle, the first helical repeat increases the potency of the effect of the N-terminus in intact cells (see Fig. 4). Thus, it seems likely that the first helical repeat stabilizes the interaction between the N-terminus and α-tubulin in intact cells in an analogous manner to that discussed above for the in vitro situation.

Previous work has shown that Op18 derivatives truncated at either the N-terminus or the C-terminus exhibit distinct defects in intact cells. Both types of truncated Op18 proteins destabilize interphase MTs to a similar extent but differ completely in their action during mitosis(Holmfeldt et al., 2001). For example, whereas non-phosphorylatable (i.e. Ser to Ala substitutions)Op18-tetra-A(1-99) blocked formation of functional mitotic spindles as expected, overexpression of a non-phosphorylatable Op18-triA(25-149)derivative allowed formation of normal spindles and subsequent cell division. The N-terminal truncated Op18-triA(25-149) derivative, containing the complete helical region of Op18, is more efficient in destabilizing interphase MTs than the Op18-triA(1-57)-NR-helix derivative used here. Since the latter derivative, but not the tandem tubulin dimer complex forming the Op18-triA(25-149) derivative, blocks cells in mitosis(Table 1), this result emphasizes the importance of the N-terminus for Op18-mediated destabilization of the mitotic spindle (Holmfeldt et al.,2001).

If the two mechanisms by which Op18 regulates the MT system operate independently from each other, it follows that they may be independently regulated by phosphorylation. We have recently analyzed the phenotype of a`pseudo-phosphorylation' derivative of Op18 with four Ser to Glu substitutions at phosphorylation sites (denoted Op18-tetraE)(Holmfeldt et al., 2001). It is noteworthy that three of these substitutions are located within the unstructured N-terminus. Consistent with an altered function of the N-terminus, it was found that Op18-tetraE does not promote catastrophes in vitro, and detailed analysis of tubulin-directed activities indicated the Glu substitutions at phosphorylation sites have a similar functional outcome to N-terminal truncations. Accordingly, Op18-tetraE efficiently destabilizes the interphase array of MTs but, as predicted by its lack of catastrophe activity,does not interfere with formation of the mitotic spindle. Hence, it seems likely that phosphorylation of N-terminal sites primarily results in attenuation of N-terminal-mediated functional activities, such as catastrophe promotion.

As outlined above, both studies in vitro and in intact cells have indicated a functional dichotomy of Op18. Examples include (i) sequestering versus catastrophe-promoting activity during MT assembly in vitro(Howell et al., 1999a), (ii)various types of modulation of tubulin GTP turnover in vitro(Larsson et al., 1999), (iii)differential MT-directed activities during interphase and mitosis(Holmfeldt et al., 2001) and(iv) efficient suppression by Op18 derivatives with intact tubulin sequestering activity but not by MT-destabilizing catastrophe-proficient Op18 derivatives that are deficient in tubulin sequestering activity of microtubule stabilization by the MAP4 protein(Holmfeldt et al., 2002). Functional dichotomy of Op18 was originally suggested by analysis of MT-directed activities of truncated Op18 derivatives, which demonstrated that the N-terminus is required for catastrophe promotion but not tubulin-sequestering activity, which only requires the first and second helical repeats of Op18 (Howell et al.,1999b). This report was subsequently extended by studies on modulation of tubulin GTP hydrolysis by truncated Op18 derivatives, which demonstrated that the first and second helical repeats are both sufficient and necessary for efficient formation of a ternary tubulin complex and to stimulate a low-rate GTP hydrolysis within this complex(Larsson et al., 1999). Since the C-terminally truncated Op18(1-99) derivative promotes catastrophes, but not formation of GTPase-productive tandem tubulin dimer complexes, GTPase stimulation within the ternary tubulin complex is clearly not involved in catastrophe promotion.

The present data together with recent structural and functional information suggest a simple model of multiple functional Op18-tubulin interactions, which all contribute to the final tubulin-binding affinity of native Op18. This model seems relevant for the understanding of the functional dichotomy of Op18 since it implies two distinct types of Op18 interactions with tubulin heterodimers, both of which involve multiple stabilizing contact points. One type of interaction can be described as cooperative binding along the longitudinally arranged tandem tubulin dimers via the first and second helical repeats of Op18 (Gigant et al.,2000), which is essential for sequestering of tubulin in GTPase-productive tandem tubulin dimers(Larsson et al., 1999). A second type can be described as an independent interaction between the N-terminus of Op18 and most probably a longitudinal surface on theα-tubulin. This can be viewed as a variant of the `capping model', which was proposed on the basis of the digital image analysis of electron micrographs of Op18-tubulin complexes(Steinmetz et al., 2000) and subsequent cross-linking experiments(Muller et al., 2001). With respect to binding affinity, these two types of interaction are synergistic,with the N-terminus of Op18 mediating very low-affinity binding in the absence of the first helical repeat. If the most C-terminal helical repeat of Op18 is removed, this results in a switch from two-site positive cooperative binding of two heterodimers to non-cooperative binding of a single heterodimer(Larsson et al., 1999; Segerman et al., 2000). Truncation of either the N- or C-terminus results in loss of contact points and therefore a decrease in tubulin affinity, as suggested by both binding analysis (Larsson et al.,1999) and structural studies(Gigant et al., 2000; Muller et al., 2001). However,the functional outcome of the interaction(s) with the N-terminus of Op18 is clearly very different from the functional consequences of formation of a tandem tubulin dimer complex by the first and second helical repeat. Thus, as outlined above, interaction of the Op18 N-terminus with a longitudinal surface on the α-tubulin end of the ternary complex appears essential for catastrophe promotion but not tubulin-sequestering activity. Moreover, it appears that this interaction with α-tubulin is also manifested by inhibition of the basal GTP hydrolysis of soluble tubulin. Since the first and second helical repeats allow two-site positive cooperativity in binding, and thereby facilitate interactions between the Op18 N-terminus andα-tubulin, it probably allows Op18 N-terminal-dependent catastrophe promotions at lower Op18 concentrations. Such synergistic interplay between the N-terminus and the first and second helical repeats of Op18 for catastrophe promotion is supported by the present evidence for specific catastrophe promotion by high concentrations of the Op18(1-57)-NR-helix(Fig. 2).

Since it has been suggested that the N-terminus of Op18 interacts with theα-tubulin end of the α/β heterodimer(Muller et al., 2001), it seems unlikely that this region of Op18 promotes plus-end-specific catastrophes by direct interaction with the β-subunits exposed at the MT plus-end tip. It follows that the mechanism for MT destabilization by the N-terminus of Op18 may involve interactions with soluble α/βheterodimers or possibly even MT polymers. Since catastrophe promotion can clearly be dissociated from tubulin sequestering by distinct activities of truncated Op18 derivatives (Holmfeldt et al., 2002; Holmfeldt et al.,2001; Howell et al.,1999b), it seems to exclude the possibility that the catastrophe-promoting activity can be explained by tubulin sequestering. Thus,the mechanism behind Op18-mediated catastrophe promotion remains elusive. However, the present model of how distinct types of tubulin interaction contribute to the functional dichotomy of Op18 should provide a framework for future functional studies.

We thank Victoria Shingler for discussions and critical reading of the manuscript. B.S., P.H. and M.G. were supported by Swedish Natural Science Research Council and L.C. and J.M. were supported by an NIH grant.