The Interaction of TOGp with Microtubules and Tubulin*

TOGp is the human homolog of XMAP215, aXenopus microtubule-associated protein that promotes rapid microtubule assembly at plus ends. These proteins are thought to be critical for microtubule assembly and/or mitotic spindle formation. To understand how TOGp interacts with the microtubule lattice, we cloned full-length TOGp and various truncations for expression in a reticulocyte lysate system. Based on microtubule co-pelleting assays, the microtubule binding domain is contained within a basic 600-amino acid region near the N terminus, with critical domains flanking a region homologous to the microtubule binding domain found in the related proteins Stu2p (S. cerevisiae) and Dis1 (S. pombe). Both full-length TOGp and the N-terminal fragment show enhanced binding to microtubule ends. Full-length TOGp also binds altered polymer lattice structures including parallel protofilament sheets, antiparallel protofilament sheets induced with zinc ions, and protofilament rings, suggesting that TOGp binds along the length of individual protofilaments. The C-terminal region of TOGp has a low affinity for microtubule polymer but binds tubulin dimer. We propose a model to explain the microtubule-stabilizing and/or assembly-promoting functions of the XMAP215/TOGp family of microtubule-associated proteins based on the binding properties we have identified.

Microtubule assembly is regulated in cells to generate a relatively stable interphase microtubule array or the much more dynamic microtubules of the mitotic spindle. In either cell cycle stage, the major pathway of microtubule turnover is dynamic instability, where microtubules exist in persistent phases of growth or shortening with the abrupt transitions, termed catastrophe and rescue, between these phases (1). Several classes of microtubule assembly regulators have been identified that can be broadly classified as microtubule stabilizers (e.g. tau, MAP2, 1 or MAP4) or destabilizers (e.g. XKCM1 or oncoprotein 18; reviewed in Refs. 2 and 3). Together, the activities of these accessory proteins generate the dynamic microtubules observed in vivo (reviewed in Refs. 3 and 4).
The stabilizing MAP, XMAP215, was initially isolated based on its preferential promotion of microtubule plus end assembly rates (5). Remarkably, this protein speeds the microtubule plus end growth rate by 7-10-fold, primarily through an increase in the apparent on-rate constant (5,6). In contrast, other stabilizing MAPs, such as tau or MAP2, modestly increase growth rates ϳ2-fold at both microtubule ends, primarily through a decrease in the off-rate constant (7)(8)(9). More recent studies have demonstrated that the plus end stabilizing activity of XMAP215 can also counterbalance the catastrophe-promoting activity of XKCM1 (10). The mechanisms responsible for assembly promotion and catastrophe protection by XMAP215 are not known.
Given the potent and unique effects of XMAP215 on microtubule assembly in vitro and in Xenopus egg extracts (5,6,10,11), it is not surprising that a number of homologs have been identified in other organisms. The yeasts S. cerevisiae and S. pombe express Stu2p (12) and Dis1 (13), respectively, which are similar to the N-terminal half of XMAP215. Additional homologs have been identified in C. elegans (ZYG-9; Ref. 14), Drosophila (Msps; Ref. 15), and human (TOGp; Refs. 16 and 17). A number of observations suggest that these proteins function in mitosis. First, mutations in Stu2p (12), Dis1 (13), ZYG-9 (14), and Msps (15) result in mitotic defects. Second, experiments using mitotic Xenopus egg extracts and HeLa cell extracts have demonstrated a role for XMAP215 and TOGp, respectively, in spindle or mitotic aster formation (10,11,18). Loss of XMAP215 or TOGp in mitotic cell extracts results in a decrease in the number and average length of microtubules. Taken together, the XMAP215/TOGp family of MAPs appears necessary for mitotic spindle assembly, probably through its microtubule-stabilizing activity.
Based on the unique plus end assembly-promoting activity of XMAP215, it is likely that this family of MAPs interacts differently with microtubules compared with the tau, MAP2, and MAP4 group. Consistent with this idea, little sequence homology exists between the XMAP215/TOGp family and other stabilizing MAPs. The exception is one 18-amino acid stretch centered on a KXGS motif in the C terminus of TOGp and XMAP215 (16). This region is ϳ50% identical to the microtubule binding repeats of MAP2, MAP4, and tau, and the KXGS motif appears to be a key regulatory domain in these other MAPs (19).
In the present study, we have identified the microtubule binding domain within TOGp and explored how this MAP interacts with the microtubule lattice and tubulin dimers. It is important to note that TOGp and XMAP215 are ϳ80% identical at the amino acid level (10,20). Although partially purified TOGp did not promote assembly to the same extent as XMAP215 (17), our binding studies are relevant to understanding how XMAP215 and other family members interact with the microtubule lattice. Furthermore, these studies allow us to formulate models of how XMAP215 regulates microtubule assembly and protects plus ends from depolymerization by XKCM1. Construction of TOGp Expression Plasmids-The original cloning of the coding sequence for the N-terminal half of TOGp has been previously described (pTOGN; Ref. 17). A plasmid for the expression of the full-length protein, pBSTOG, was generated by cloning the entire coding sequence into the pBluescript vector (Stratagene) at the 5Ј-SacII and 3Ј-KpnI sites of the multicloning region. A plasmid for the expression of the C-terminal half of TOGp, pBSTOG(⌬14 -1229), was generated by removing the coding sequence for amino acids 14 -1229 from pBSTOG by digestion with BclI and religating the remaining plasmid.
Plasmids for the expression of various N-terminal fragments of TOGp were generated using the parent pTOGN construct. pTOGN(⌬383-907) was constructed by removing the coding sequence for amino acids 383-907 by digestion with PstI and religation of the remaining plasmid. pTOGN(⌬144 -331) was generated by removing the coding sequence for amino acids 144 -331 by digestion with XcmI and religation. pGEMT ϩ TOGN(⌬383-596) was generated by removing the coding sequence for amino acids 383-907 by digestion with PstI and SalI and replacing it with the coding sequence for amino acids 597-907 using polymerase chain reaction mutagenesis to create an in-frame PstI restriction site (shown in boldface type) with the primers 5Ј-GACCTG-CAGTCAGCTGTTCTTCCCCCTACCTGT-3Ј sense and 5Ј-TCTAGA-CAGGATATTCAGCGTTTGCTGTAC-3Ј antisense. pTOGN(⌬383-799) was also generated by removing the coding sequence for amino acids 383-907 by digestion with PstI and SalI and replacing it with the coding sequence for amino acids 800 -907 using polymerase chain reaction mutagenesis to create an in-frame PstI restriction site (shown in boldface type) with the primers 5Ј-GACCTGCAGGCCCTCCTATC-CCAGATAGATGCAGAA-3Ј (sense) and 5Ј-TCTAGACAGGATAT-TCAGCGTTTGCTGTAC-3Ј (antisense). The coding sequence for amino acids 144 -596 was removed from the pGEMT ϩ TOGN(⌬383-596) construct by digestion with XcmI and PstI and replaced with the coding sequence for amino acids 144 -237 using polymerase chain reaction mutagenesis to create an in-frame PstI site (shown in boldface type) with the primers 5Ј-GGATCCTGGAAAGCAAGGTTAAGTGGGTAT-3Ј sense and 5Ј-GACCTGCAGTTCTTGTTGGGAACGAAGAAATCG-3Ј antisense to generate pTOGN(⌬238 -596).
Assembly of Microtubules and Altered Lattice Structures for Copelleting Assays-The following polymers were assembled from purified tubulin. 1) Taxol-stabilized microtubules were prepared by assembling 20 -50 M tubulin in PEM buffer supplemented with 1 mM GTP and 20 -50 M taxol at 37°C for 10 min. Microtubules were then placed on ice and rewarmed at 37°C for 5 min before use in pelleting assays. 2) Sheared microtubules were prepared by passing taxol microtubules through a 26-gauge needle 10 -15 times immediately before the addition to the reticulocyte lysate samples (22). 3) GMPCPP microtubules were prepared by assembling 10 M GMPCPP-tubulin in PEM buffer supplemented with 0.2 mM GMPCPP at 37°C for 30 min (21). GMPCPP microtubules were then used directly in pelleting assays. 4) Subtilisincleaved microtubules were prepared as described previously (23) with slight modifications. Microtubules were assembled from 20 M tubulin in PEM buffer supplemented with 1 mM GTP and 20 M taxol at 37°C for 30 min. Microtubules were then cleaved by the addition of 30 g/ml subtilisin and incubated at 37°C for 3 h. The enzyme was inactivated with phenylmethlylsulfonyl fluoride (2 mM), and microtubules were pelleted through a 4 M glycerol cushion in PEM buffer (supplemented with 2 mM phenylmethlylsulfonyl fluoride, 1 mM GTP, and 20 M taxol), for 10 min at 20 p.s.i. in an Airfuge (Beckman). The pellet was resuspended in PEM buffer and then used in pelleting assays. 5) Subtilisincleaved tubulin was prepared as described previously (24 -26) with slight modifications. Tubulin (20 M) was first cleaved with 40 g/ml subtilisin in PEM buffer at 30°C for 30 min. The enzyme was inactivated with 2 mM phenylmethlylsulfonyl fluoride and incubated at room temperature for 30 min. Polymer was assembled by the addition of 1 mM GTP and 20 M taxol and incubation at 37°C for 30 min. 6) Zinc sheets were assembled essentially as described previously (27). 10 M tubulin was assembled in 100 mM Pipes buffer supplemented with 1 mM ZnCl 2 , 0.5 mM MgCl 2 , 0.5 mM EGTA, 1 mM GTP, and 20 M taxol at 30°C for 20 min. 7) GDP protofilament rings were assembled as described by others (28). Subtilisin cleavage of nucleotide-free tubulin was performed as described above. Rings were assembled from 15 M subtilisincleaved tubulin in PEM buffer supplemented with 1 mM GDP and 10 mM MgCl 2 at 30°C for 30 min. Rings were then pelleted in an Airfuge for 30 min at 20 p.s.i. and resuspended in PEM buffer supplemented with 1 mM GDP and 10 mM MgCl 2 before use in co-pelleting assays. To pellet protofilament rings, no glycerol cushion was used. Despite this, little to no TOGp pelleted in the absence of rings (data not shown).
Microtubule Co-pelleting Assays-[ 35 S]methionine-labeled fulllength TOGp and TOG peptides were expressed using the TNT Coupled Reticulocyte Lysate System (Promega) according to manufacturer's directions. Reticulocyte lysate samples containing expressed full-length TOGp or TOG peptides (TOG-lysates) were diluted 3-fold in PEM buffer and clarified for 10 min at 20 p.s.i. (ϳ100,000 ϫ g) in an Airfuge. 25 l (maximum of 80 ng of expressed protein, based on the manufacturer's estimate of maximum expression level) of the clarified TOG-lysate was added to the indicated concentration of microtubule polymer. Based on XMAP215 binding to microtubules (5), these samples should maintain a substoichiometric ratio of TOG peptide to tubulin. Samples containing TOG-lysate, microtubules (concentrations given here), 1 mM GTP, 20 M taxol, and 0.1 mg/ml BSA in a total volume of 50 l of PEM buffer were incubated at 37°C for 15 min and then pelleted through an equal volume of a 4 M glycerol cushion (in PEM buffer supplemented with 1 mM GTP, 20 M taxol, and 0.1 mg/ml BSA) for 10 min at 20 p.s.i. in an Airfuge. The supernatant was combined with the cushion in order to recover all soluble TOG peptide and prepared for SDS-PAGE by adding 25 l of 5ϫ sample buffer without glycerol. The pellets were washed in PEM buffer and resuspended in 125 l of 1ϫ sample buffer. Samples were examined by SDS-PAGE and autoradiography as described below.
Electron Microscopy-To confirm the assembly of microtubule polymer and of altered lattice structures, polymer samples were assembled as described above and then applied to formar/carbon-coated grids and stained with 2% uranyl acetate. Grids were examined on a Phillips EM 400-T microscope operated at 80 kV.
Tubulin and BSA Affinity Chromatography-The interaction of various domains of TOGp with tubulin dimer was examined using tubulin affinity chromatography. BSA columns were used as a nonspecific binding control, since BSA and tubulin are both acidic proteins (29). Tubulin and BSA affinity columns were prepared using CNBr-activated Sepharose (Amersham Pharmacia Biotech) following instructions provided by the manufacturer. After coupling and blocking the remaining reactive groups, the resin was washed and resuspended in PEM buffer and stored at 4°C for up to 2 months.
Fifty l of TOG-lysate was incubated with 1 mM GTP and 25 l of tubulin or BSA resin and mixed on a Nutator for 15 min at room temperature. The resin was pelleted for 1 min at 6000 rpm in a microcentrifuge, washed three times with 500 l of PEM buffer, and resuspended in 50 l of 1ϫ gel sample buffer for SDS-PAGE and autoradiography (see below).
SDS-PAGE and Autoradiography-Proteins were separated by SDS-PAGE on 7.5% or 3-10% gradient polyacrylamide gels using a discontinuous buffer system. Coomassie Blue or silver staining (30) was used for protein detection. Gels were dried and exposed to X-Omat AR (Eastman Kodak Co.) x-ray film with an intensifying screen.

RESULTS
Binding of Full-length TOGp to Microtubules-Using a reticulocyte lysate system, we expressed a 35 S-labeled 215-kDa protein ( Fig. 1) that reacted with TOGp antibodies on Western blots (data not shown). Several lower molecular weight 35 Slabeled proteins are also present and probably represent partial translation products, since they are not observed in the absence of added plasmid. The expressed TOGp co-pelleted with microtubules, as shown in the autoradiogram (Fig. 1).
To further define microtubule binding, we measured an apparent K d , as defined previously by Butner and Kirschner (22). Here, K d is defined as the concentration of polymerized tubulin required to pellet half of the total added MAP (22). We determined the apparent binding affinity of full-length TOGp by adding diluted and clarified [ 35 S]TOG-lysate to varying concentrations of taxol-stabilized microtubules. A typical experiment is shown in Fig. 1, A and B, and demonstrates that the K d for TOGp is ϳ2 M. Based on the results of five identical assays performed on different days, we conclude that the apparent K d of TOGp for microtubules is 2-5 M. This analysis also shows that the interaction of TOGp with taxol-stabilized microtubules is specific; TOGp does not pellet in the absence of microtubules (Fig. 1B) and is the only detectable protein in the lysate mix that co-pellets with microtubules (Fig. 1A). Moreover, this association is inhibited by salt (data not shown), a property that is characteristic of the ionic interaction of other MAPs with microtubules (31).
To ensure that species-specific or tissue differences in ␣and ␤-tubulin did not alter the TOGp interactions, microtubule co-pelleting assays with full-length TOGp were also performed using purified HeLa cell tubulin. The K d values in these assays were also in the 2-5 M range (data not shown), suggesting that binding properties were not altered by any potential posttranslational modification of tubulin in brain tissue (32).
TOGp Preferentially Binds to Microtubule Ends-Since TOGp and several of its homologues have been shown to either localize near microtubule ends in vivo (12)(13)(14)(15)(16), interact with MT ends in vitro (33), or specifically affect microtubule plus end growth in vitro (5, 6), we modified the binding assay to determine if TOGp bound preferentially to microtubule ends. First we compared the amount of [ 35 S]TOGp that co-pelleted with taxol-stabilized microtubules with the amount that co-pelleted with an equal concentration of polymer that was sheared to increase the number of ends. Using differential interference contrast (DIC) microscopy, we estimated that the sheared samples contained ϳ5 times more ends compared with unsheared microtubules (data not shown). At 5 M microtubule polymer, TOGp was equally distributed between the supernatant and pellet in assays using unsheared microtubules (Fig. 2). In contrast, a greater fraction of TOGp was found co-pelleting with sheared microtubules (Fig. 2). We further examined a possible preference of TOGp for microtubule ends by nucleating microtubule assembly with the nonhydrolyzable GTP analog, GMPCPP. Because tubulin-GMPCPP readily nucleates into microtubules (34), the number of nuclei is greater, and the resulting number of microtubule ends is also greater, compared with samples containing GTP-tubulin. We find that tubulin-GMPCPP assembles into shorter microtubules with an ϳ10 time greater concentration of ends compared with tubulin-GTP assembled with taxol (data not shown). When GMPCPP microtubules were used in co-pelleting assays, TOGp was almost completely found in the pellet fraction (Fig. 2).
The Microtubule Binding Domain of TOGp Is Found within the N Terminus-The preferential binding of TOGp to microtubule ends means that the apparent K d will be highly dependent on the number of ends in a sample. For molecular dissection of the microtubule binding domains within TOGp, we used apparent K d as a convenient means for comparison, but the values should not be compared with the reported binding affinities of other MAPs, since our values are dependent upon microtubule end concentration.
Sequence comparisons of TOGp with other MAPs have revealed one region in the C terminus that is similar to the microtubule binding repeat of tau, MAP2, and MAP4 (16). This motif is present in TOGp only once (amino acids 1773-1790) and is centered around a KIGS domain (16). In order to determine if this region is also important for TOGp binding to microtubules, constructs containing the N-terminal half (amino acids 1-925) and a C-terminal region containing the KIGS motif (amino acids 1230 -1972) were assayed for microtubule binding. The results are shown in Fig. 3A. TOGN(1-925) has an apparent K d Յ 5 M, similar to that of the full-length protein. This peptide also has a higher affinity for microtubule ends (data not shown), suggesting that the N terminus is sufficient for preferential binding to microtubule ends. In contrast, less than half of the C-terminal peptide, TOG(⌬14 -1229), pellets with 25 M taxol-stabilized microtubules, suggesting that this region does not directly contribute to microtubule binding.
Additional deletion constructs of the amino-terminal half of TOGp were constructed and assayed for microtubule binding. The K d values for these peptides are summarized in Fig. 3B. These results suggest that a 650-amino acid region with a pI ϭ 8.92 is necessary for microtubule binding. This region contains a stretch of sequence homologous to the microtubule binding domains of the yeast homologues of TOGp, Dis1 (35), and Stu2p (12) (amino acids 493-596 in TOGp), but deletion of this domain alone had a minimal effect on TOGp binding to microtubules (see the results for construct TOGN(⌬383-596).
TOGp Binds to Sheet-like Polymers and Rings of Protofilaments-We next determined whether the acidic C-terminal region of ␣ and ␤ tubulin was necessary for TOGp binding. This region of tubulin is required for binding of tau, MAP2, and MAP4 (23,25,26,36). Subtilisin digestion of taxol-stabilized microtubules removes the extreme C terminus of ␣and ␤-tubulin while the microtubule structure remains intact (23) (Fig. 4A). TOGp co-pellets with subtilisin-cleaved microtubules  2. TOGp has a higher affinity for microtubule ends. An autoradiogram is shown of a microtubule co-pelleting assay comparing the binding of full-length TOGp to 5 M microtubule samples containing varying numbers of ends. Samples containing shorter microtubules (i.e. more ends) were generated by shearing taxol microtubules through a needle or assembling tubulin in the presence of the nonhydrolyzable GTP analog, GMPCPP. Relative numbers of microtubule ends in each sample are indicated (1ϫ, 5ϫ, 10ϫ). T-MTs, unsheared taxol-stabilized microtubules. (Fig. 4C), suggesting that the TOGp binding site on tubulin differs from MAP2, MAP4, and tau.
Several reports have proposed models for the interaction of stabilizing MAPs (MAP2, MAP4, and tau) with the microtubule lattice, which involves the cross-linking of tubulin subunits of adjacent protofilaments (37,38). This type of interaction would probably require the normal parallel arrangement of protofilaments and the cylindrical morphology of microtubules to allow binding. Therefore, we determined whether TOGp binding was dependent on an intact microtubule structure. When tubulin is cleaved with subtilisin prior to assembly into polymer, the result is the formation of sheets and ribbons of protofilaments rather than cylindrical microtubules (Fig. 4A) (24). As shown in Fig. 4C, TOGp co-pelleted with sheets of protofilaments assembled from subtilisin-cleaved tubulin.
Inducing tubulin polymerization in the presence of zinc ions (39) can alter polymer morphology and arrangement of protofilaments. Sheets of protofilaments in an antiparallel arrangement with opposite surfaces exposed (inside out versus outside in) are produced (40,41) (Fig. 4A). TOGp was also able to bind to these sheet-like polymers (Fig. 4C).
The binding of TOGp to protofilament sheets suggested that it may bind along single protofilaments (see "Discussion"). To further test this hypothesis, double rings of coiled protofilaments were assembled (28,42) (Fig. 4A). As shown in Fig. 4C, TOGp was also able to co-pellet with the protofilament rings.
The C Terminus of TOGp Contains a Tubulin Dimer-binding Domain-Previous studies demonstrated that XMAP215 increased the plus end elongation rate by an increase in the association rate constant (5,6). Gard and Kirschner (5) proposed that an interaction of XMAP215 with tubulin dimers or oligomers that oriented and enhanced assembly at the plus end could explain the dramatic effect on tubulin on-rate. Because of the high sequence homology between XMAP215 and TOGp, we looked for a tubulin-binding domain in TOGp to test their hypothesis.
To identify tubulin binding domains, we used tubulin affinity chromatography (29,43). Tubulin was coupled to Sepharose beads under conditions that favored dimer rather than polymer formation. The specificity of the tubulin dimer interaction was tested by comparing binding to BSA-coupled Sepharose. Fulllength TOGp showed a higher affinity for tubulin than BSA (Fig. 5). In order to localize the domain responsible for tubulin binding, the N-and C-terminal halves of TOGp were analyzed separately (constructs TOGN(1-925) and TOG(⌬14 -1229), respectively). The results are also shown in Fig. 5 and demonstrate that the C terminus binds tubulin dimer but not microtubule polymer (compare with Fig. 3A).
The interaction between TOGp and tubulin was confirmed by co-immunoprecipitation and chemical cross-linking. [ 35 S]TOG expressed in reticulocyte lysates was combined with 10 M tubulin at 4°C (to prevent polymer assembly). The C terminus of TOGp co-immunoprecipitated with tubulin in the presence or absence of the cross-linker EDC but not in the absence of tubulin antibody (data not shown). These results confirm that the C terminus contains a tubulin dimer-binding domain.

TOGp Contains Distinct Microtubule Polymer and Tubulin
Dimer-binding Domains-Our binding studies with various deletion constructs showed that TOGp contains a microtubule binding site within the N terminus. This binding site is relatively large (ϳ600 amino acids), has a basic pI (8.92), and contains a central region similar to the microtubule binding domain of the yeast homologues, Stu2p (12) and Dis1 (35). Deletion of this region alone (amino acids 493-596) had only a minimal effect on microtubule binding, but additional deletions of flanking sequences significantly decreased binding to microtubules. Sequences of ϳ200 amino acids in regions N-and C-terminal to amino acids 493-596 appear to be equally necessary for microtubule binding (Fig. 3B). Sequence alignment of these flanking regions (shown as gray boxes in Fig. 3B) showed little homology between them (Ͻ20% identity; not shown).
TOGp also binds tubulin dimers or oligomers in a region distinct from the microtubule binding domain (Fig. 5). Several previous studies have also identified tubulin binding by MAPs (44,45), but these studies did not identify separate tubulin and microtubule binding regions. Gard and Kirschner (5) previously hypothesized that XMAP215 contains a tubulin binding domain, and they suggested that this domain could speed microtubule elongation by delivering tubulin dimers or oligomers to microtubule plus ends. Our data support this general model, since TOGp contains separate microtubule and tubulin binding domains. It is not known whether TOGp (or XMAP215) binds first to tubulin dimers or microtubule ends; these possibilities are outlined in Fig. 6, A and B. A general mechanism where XMAP215 or TOGp delivers tubulin dimers/oligomers to microtubule ends is a theoretically possible way to increase microtubule growth rate, since recent modeling studies suggested that tubulin addition to a microtubule end is limited by the orientation of the dimer as it approaches the microtubule tip and not limited by the rate of tubulin diffusion to the microtubule end (46).
Another characteristic of the interaction of XMAP215 and TOGp with microtubules is their preference for microtubule ends (Fig. 2) (33). Both XMAP215 and TOGp also bind along the length of microtubules, since immunofluorescent observations showed continuous labeling along microtubules (5,10,16). Taken together, these results suggest that both XMAP215 and TOGp binding are enhanced at microtubule ends but are not limited to ends as observed with CLIP-170 (50,51) or adenomatous polyposis coli (APC) (52). It is unlikely that TOGp binds differently to GTP or GDP tubulin subunits, since all microtubules were in a GTP-like conformation in our experiments; the addition of taxol or incorporation of GMPCPP results in the assembly of a GTP-like lattice (53,54). Under our experimental conditions, the enhanced binding to microtubule ends is observed after the TOGp addition to preassembled polymers. Therefore, TOGp binding to microtubule ends is not dependent on concurrent microtubule assembly, also in contrast to that observed for CLIP-170 (51). It is possible that XMAP215 and TOGp bind preferentially to the protofilament sheets observed extending from microtubule ends (55)(56)(57), perhaps along the edges of exposed protofilaments.
We find that the N-terminal half of TOGp is sufficient for preferential binding to microtubule ends, suggesting that the microtubule binding region is sufficient for enhanced end binding. The S. cerevisiae homolog, Stu2p, is approximately the size of our N-terminal construct, yet studies with Stu2p failed to detect enhanced binding to microtubule ends (12). While this is possibly due to assay conditions, the different results may also point to subtle differences in functions between these proteins.
A Model for the Structural Interaction of TOGp with Microtubules-TOGp appears to bind along individual protofilaments, since the expressed protein co-pelleted with protofilament sheets (parallel orientation), Zn 2ϩ -induced antiparallel sheets, and protofilament rings. These results suggest that the cylindrical morphology of a microtubule, the polarized alignment of adjacent protofilaments, and the groove between protofilaments were not required for TOGp binding to polymer. XMAP215 is predicted to be an elongate molecule of approximately 100 nm in length (5). If TOGp is also an elongate molecule, it would extend along 12 or 13 dimers of a protofilament. If this size is correct, the microtubule binding domain would be ϳ30 nm long, and thus TOGp could span 3 or 4 tubulin dimers (8 nm/dimer) along a protofilament. Further support for a model where XMAP215 or TOGp binds along protofilaments comes from studies of microtubule rigidity; microtubules assembled with XMAP215 showed no difference in stiffness compared with microtubules assembled from purified tubulin. 2 In contrast, tau is thought to cross-link adjacent protofilaments (22,(47)(48)(49), and tau substantially increases microtubule rigidity (49).
The binding of TOGp along protofilaments suggests several 2 P. Tran, L. Cassimeris, and E. D Salmon, unpublished observations.
FIG. 5. TOGp contains a tubulin dimer-binding region in the C terminus. The presence of a tubulin dimer-binding domain within TOGp was examined by tubulin affinity chromatography. Autoradiograms comparing the binding of various TOGp constructs to tubulin-Sepharose resin are shown. BSA-Sepharose was used as a control for nonspecific ionic interactions, since the pI of tubulin and BSA are both acidic. The position of each 35 S-labeled peptide is marked with an asterisk.
The C-terminal construct binds specifically to tubulin-Sepharose.
FIG. 6. Model for TOGp interactions with microtubules and tubulin dimers and mechanisms to stimulate assembly. In each model, TOGp is diagrammed as an elongate molecule (red) with a tubulin dimer binding region (yellow). A, TOGp is shown bound first to the microtubule lattice, which may orient incoming tubulin dimers. B, TOGp may first bind tubulin dimers, and this complex may subsequently bind to the microtubule lattice and deliver tubulin to the microtubule end. The microtubule binding domain is shown in light blue. In models A and B, delivery of tubulin dimers to microtubule ends would stimulate assembly, and polarized binding of TOGp to the microtubule lattice would expose the tubulin dimer binding region only at one microtubule end. C, TOGp binds along protofilaments; this binding orientation may stabilize elongating protofilaments, generating a stable seed that favors the addition of dimers to the adjacent protofilaments. Propagation of elongating protofilaments laterally could stimulate assembly. A structurally similar model was proposed previously for microtubule nucleation by ␥-tubulin (59). For all diagrams, TOGp is not drawn to scale, and for clarity, only one TOGp is shown per microtubule end. models for how this protein and XMAP215 can stimulate assembly and protect growing ends from depolymerization by XKCM1. First, TOGp binding along a protofilament may stabilize nascent protofilaments extending from microtubule ends. This could form a stable "seed" for addition of subunits laterally to form adjacent protofilament sheets (Fig. 6C). Second, TOGp could bind in a polarized orientation that extends the C-terminal, dimer binding region toward the microtubule plus end. In this orientation, TOGp could facilitate tubulin addition specifically at plus ends (Fig. 6, A and B). This type of model could apply to XMAP215 and provide a mechanism to specifically enhance assembly at microtubule plus ends. Either of these models could also account for the increase in the apparent association rate constant measured during XMAP215-stimulated plus end assembly (6). Third, binding of TOGp along a protofilament could block binding sites for a kinesin such as XKCM1 (58), protecting the ends from destabilization and maintaining microtubules in an elongation phase.