Solution NMR structure of S100B bound to the high-affinity target peptide TRTK-12

The solution NMR structure is reported for Ca(2+)-loaded S100B bound to a 12-residue peptide, TRTK-12, from the actin capping protein CapZ (alpha1 or alpha2 subunit, residues 265-276: TRTKIDWNKILS). This peptide was discovered by Dimlich and co-workers by screening a bacteriophage random peptide display library, and it matches exactly the consensus S100B binding sequence ((K/R)(L/I)XWXXIL). As with other S100B target proteins, a calcium-dependent conformational change in S100B is required for TRTK-12 binding. The TRTK-12 peptide is an amphipathic helix (residues W7 to S12) in the S100B-TRTK complex, and helix 4 of S100B is extended by three or four residues upon peptide binding. However, helical TRTK-12 in the S100B-peptide complex is uniquely oriented when compared to the three-dimensional structures of other S100-peptide complexes. The three-dimensional structure of the S100B-TRTK peptide complex illustrates that residues in the S100B binding consensus sequence (K4, I5, W7, I10, L11) are all involved in the S100B-peptide interface, which can explain its orientation in the S100B binding pocket and its relatively high binding affinity. A comparison of the S100B-TRTK peptide structure to the structures of apo- and Ca(2+)-bound S100B illustrates that the binding site of TRTK-12 is buried in apo-S100B, but is exposed in Ca(2+)-bound S100B as necessary to bind the TRTK-12 peptide.     

The S100B Protein Inhibits Phosphorylation of GFAP and Vimentin in a Cytoskeletal Fraction from Immature Rat Hippocampus

The S100B protein belongs to a family of small Ca²⁺-binding proteins involved in several functions including cytoskeletal reorganization. The effect of S100B on protein phosphorylation was investigated in a cytoskeletal fraction prepared from immature rat hippocampus. An inhibitory effect of 5 M S100B on total protein phosphorylation, ranging from 25% to 40%, was observed in the presence of Ca²⁺ alone, Ca²⁺ plus calmodulin or Ca²⁺ plus cAMP. Analysis by two dimensional electrophoresis revealed a Ca²⁺/calmodulin-dependent and a Ca²⁺/cAMP-dependent inhibitory effect of S100B, ranging from 62% to 67% of control, on the phosphorylation of the intermediate filament proteins glial fibrillary acidic protein (GFAP) and vimentin. The fact that S100B binds to the N-terminal domain of GFAP and that the two proteins are co-localized in astrocytes suggests a potential in vivo role for S100B in modulating the phosphorylation of intermediate filament proteins in glia.     

Microheterogeneity of tubulin proteins in neuronal and glial cells from the mouse brain in culture

The microheterogeneity of the alpha and beta isoforms of tubulin in brain cells in culture was studied. The cells were prepared from two precise regions of the embryonic mouse brain (ED15), the striatum and the mesencephalon. It was possible to maintain virtually pure cultures of neuronal or glial cells up to 1 and 4 weeks in vitro, respectively. The tubulin heterogeneity of striatal and mesencephalic neurons was found to be very similar after a few days in culture. More precise examination of pure neurons from the striatum revealed that their tubulin content after 7 days in vitro exhibited the same degree of complexity as a control extract from a 4 day-old mouse brain. In fact, we could detect the presence of at least six alpha and nine beta tubulin isoforms. Among these isoforms a specific family of beta proteins (beta' tubulin) and the more acidic alpha proteins were present. Since these isoforms have, up to now, been found only in tubulin extracts prepared from the nervous system, our experiments suggest that they belong to the neuronal subpopulation of this tissue. This point is reinforced by their complete absence from the tubulin proteins extracted from pure glial cells even after several weeks in vitro. These results lead us to propose that brain tubulin microheterogeneity is associated with the presence of neurons and not of glia and may, therefore, play a specific role in maintaining neuronal shape and function.     

Interaction of microtubule-associated proteins with microtubules: yeast lysyl- and valyl-tRNA synthetases and tau 218-235 synthetic peptide as model systems.

The respective contributions of electrostatic interaction and specific sequence recognition in the binding of microtubule-associated proteins (MAPs) to microtubules have been studied, using as models yeast valyl- and lysyl-tRNA synthetases (VRS, KRS) that carry an exposed basic N-terminal domain, and a synthetic peptide reproducing the sequence 218-235 on tau protein, known to be part of the microtubule-binding site of MAPs. VRS and KRS bind to microtubules with a KD in the 10(-6) M range, and tau 218-235 binds with a KD in the 10(-4) M range. Binding of KRS and tau 218-235 is accompanied by stabilization and bundling of microtubules, without the intervention of an extraneous bundling protein. tau 218-235 binds to microtubules with a stoichiometry of 2 mol/mol of assembled tubulin dimer in agreement with the proposed binding sequences alpha[430-441] and beta[422-434]. Binding stoichiometries of 2/alpha beta S tubulin and 1/alpha S beta S tubulin were observed following partial or complete removal of the tubulin C-terminal regions by subtilisin, which localizes the site of subtilisin cleavage upstream residue alpha-441 and downstream residue beta-434. Quantitative measurements show that binding of MAPs, KRS, VRS, and tau 218-235 is weakened but not abolished following subtilisin digestion of the C-terminus of tubulin, indicating that the binding site of MAPs is not restricted to the extreme C-terminus of tubulin.     

Tubulin, hybrid dimers, and tubulin S. Stepwise charge reduction and polymerization

Limited proteolysis of rat brain tubulin (alpha beta) by subtilisin cleaves a 1-2-kDa fragment from the carboxyl-terminal ends of both the alpha and beta subunits with a corresponding loss in negative charge of the proteins. The beta subunit is split much more rapidly (and exclusively at 5 degrees C), yielding a protein with cleaved beta and intact alpha subunit, called alpha beta s, which is of intermediate charge. Further proteolysis cleaves the carboxyl terminus of the alpha subunit leading, irreversibly, to the doubly cleaved product, named tubulin S, with a composition alpha s beta s. Both cleavage products are polymerization-competent and their polymers are resistant to 1 mM Ca2+- and 0.24 M NaCl-induced depolymerization. The two polymers differ in that the alpha beta s polymer is stable to cold, GDP, and podophyllotoxin, whereas tubulin S polymer is disassembled by these agents; moreover, alpha beta s forms ring-shaped polymers, whereas alpha s beta s forms filaments associated into bundles and sheets. Tubulin S co-polymerizes with native tubulin yielding a mixed product of intermediate stability. The presence of low mole fractions of tubulin S leads to a marked reduction in the critical concentration for polymerization of the mixture.     

Chaperone activity of alpha-crystallins modulates intermediate filament assembly.

Intermediate filaments are generally regarded as one of the most insoluble and resilient cytoskeletal structures of eukaryotic cells. In extracts from the ocular lens, we noticed an unusually high level of vimentin in a soluble, non-filamentous form. Immunoprecipitation of this soluble vimentin resulted in the co-precipitation of alpha-crystallins. The alpha-crystallins are homologous to the small heat shock proteins (sHSPs) and have recently been identified as molecular chaperones, capable of preventing the heat-induced aggregation of proteins. We find that the alpha-crystallins dramatically inhibit the in vitro assembly of GFAP and vimentin in an ATP-independent manner. This inhibition is also independent of the phosphorylation state of the alpha-crystallin polypeptides and each one of the four polypeptides, either alpha A1-, alpha A2-, alpha B1- or alpha B2-crystallin, are equally effective in this inhibition. Furthermore, we show that alpha-crystallins can increase the soluble pool of GFAP when added to preformed filaments. Electron microscopy demonstrated that alpha-crystallin particles could bind to intermediate filaments in a regular fashion, the spacing coinciding with the molecular length of GFAP. This is the first report, as far as we are aware, of a chaperone being involved in intermediate filament assembly and implicates chaperones in the remodeling of intermediate filaments during development and cell differentiation.     

Cytoskeleton changes following differentiation of N1E-115 neuroblastoma cell line

Summary. No systematic approach to detect expression of differentiation-related elements was published so far. The undifferentiated N1E-115 neuroblastoma cell line was switched into a neuronal phenotype by DMSO treatment and used for proteomic experiments. We used two-dimensional gel electrophoresis followed by unambiguous mass spectrometrical identification of proteins to generate a map of cytoskeleton proteins (CPs), i.e., to search for differentiation-related structures. Alpha-actin, actin-like protein 6A, gamma-tubulin complex component 2, tubulin alpha 3/alpha 7, CLIP associating protein 2, B4 integrin interactor homolog were detectable in the undifferentiated cell line exclusively and neuron-specific CPs drebrin and presynaptic density protein 95, actin-related protein 2/3, alpha and beta-centractin, PDZ-domain actin binding protein, actinin alpha 1, profilin II, ezrin, coactosin-like protein, transgelin 2, myosin light polypeptide 6, tubulin alpha 2, 6 and 7, beta tubulin (94% similar with tubulin beta-2), tubulin beta 3, tubulin tyrosine ligase-like protein 1, lamin B1 and keratin 20 were observed in the differentiated cell line only. We herein identified differentiation-related expressional patterns thus providing new evidence for the role of CPs in the process of neuronal differentiation.     

A potential role for the COOH-terminal domain in the lateral packing of type III intermediate filaments

To identify sites of self-association in type III intermediate filament (IF) proteins, we have taken an "anti-idiotypic antibody" approach. A mAb (anti-Ct), recognizing a similar feature near the end of the rod domain of vimentin, desmin, and peripherin (epsilon site or epsilon epitope), was characterized. Anti-idiotypic antibodies, generated by immunizing rabbits with purified anti-Ct, recognize a site (presumably "complementary" to the epsilon epitope) common among vimentin, desmin, and peripherin (beta site or beta epitope). The beta epitope is represented in a synthetic peptide (PII) modeled after the 30 COOH-terminal residues of peripherin, as seen by comparative immunoblotting assays. Consistent with the idea of an association between the epsilon and the beta site, PII binds in vitro to intact IF proteins and fragments containing the epsilon epitope, but not to IF proteins that do not react with anti-Ct. Microinjection experiments conducted in vivo and filament reconstitution assays carried out in vitro further demonstrate that "uncoupling" of this site-specific association (by competition with PII or anti-Ct) interferes with normal IF architecture, resulting in the formation of filaments and filament bundles with diameters much greater than that of the normal IFs. These thick fibers are very similar to the ones observed previously when a derivative of desmin missing 27 COOH-terminal residues was assembled in vitro (Kaufmann, E., K. Weber, and N. Geisler. 1985. J. Mol. Biol. 185:733-742). As a molecular explanation, we propose here that the epsilon and the beta sites of type III IF proteins are "complementary" and associate during filament assembly. As a result of this association, we further postulate the formation of a surface-exposed "loop" or "hairpin" structure that may sterically prevent inappropriate filament-filament aggregation and regulate filament thickness.     

The evolution of the structure of tubulin and its potential consequences for the role and function of microtubules in cells and embryos

This paper discusses the results of homology modeling and resulting calculation of key structural and physical properties for close to 300 tubulin sequences, including alpha, beta, gamma, delta and epsilon -tubulins. The basis for our calculations was the structure of the tubulin dimer published several years ago by Nogales et al. (1998), later refined to 3.5 resolution by Lowe et al. (2001). While, it appears that the alpha, beta and gamma-tubulins segregate into distinct structural families, we have found several differences in the physical properties within each group. Each of the alpha, beta and gamma- tubulin groups exhibit major differences in their net electric charge, dipole moments and dipole vector orientations. These properties could influence functional characteristics such as microtubule stability and assembly kinetics, due to their effects on the strength of protein-protein interactions. In addition to the general structural trends between tubulin isoforms, we have observed that the carboxy-termini of alpha and beta-tubulin exists in at least two stable configurations, either projecting away from the tubulin (or microtubule) surface, or collapsed onto the surface. In the latter case, the carboxy-termini form a lattice distinctly different from that of the well-known A and B lattices formed by the tubulin subunits. However, this C-terminal lattice is indistinguishable from the lattice formed when the microtubule-associated protein tau binds to the microtubule surface. Finally, we have discussed how tubulin sequence diversity arose in evolution giving rise to its particular phylogeny and how it may be used in cell- and tissue-specific expression including embryonal development.     

Structural changes in the C-terminus of Ca2+-bound rat S100B (beta beta) upon binding to a peptide derived from the C-terminal regulatory domain of p53.

S100B(beta beta) is a dimeric Ca2+-binding protein that interacts with p53, inhibits its phosphorylation by protein kinase C (PKC) and promotes disassembly of the p53 tetramer. Likewise, a 22 residue peptide derived from the C-terminal regulatory domain of p53 has been shown to interact with S100B(beta beta) in a Ca2+-dependent manner and inhibits its phosphorylation by PKC. Hence, structural studies of Ca2+-loaded S100B(beta beta) bound to the p53 peptide were initiated to characterize this interaction. Analysis of nuclear Overhauser effect (NOE) correlations, amide proton exchange rates, 3J(NH-H alpha) coupling constants, and chemical shift index data show that, like apo- and Ca2+-bound S100B(beta beta), S100B remains a dimer in the p53 peptide complex, and each subunit has four helices (helix 1, Glu2-Arg20; helix 2, Lys29-Asn38; helix 3, Gln50-Asp61; helix 4, Phe70-Phe87), four loops (loop 1, Glu21-His25; loop 2, Glu39-Glu49; loop 3, Glu62-Gly66; loop 4, Phe88-Glu91), and two beta-strands (beta-strand 1, Lys26-Lys28; beta-strand 2, Glu67-Asp69), which forms a short antiparallel beta-sheet. However, in the presence of the p53 peptide helix 4 is longer by five residues than in apo- or Ca2+-bound S100B(beta beta). Furthermore, the amide proton exchange rates in helix 3 (K55, V56, E58, T59, L60, D61) are significantly slower than those of Ca2+-bound S100B(beta beta). Together, these observations plus intermolecular NOE correlations between the p53 peptide and S100B(beta beta) support the notion that the p53 peptide binds in a region of S100B(beta beta), which includes residues in helix 2, helix 3, loop 2, and the C-terminal loop, and that binding of the p53 peptide interacts with and induces the extension of helix 4.     

Integrin alpha 6 beta 4 forms a complex with the cytoskeletal protein HD1 and induces its redistribution in transfected COS-7 cells.

The integrin alpha 6 beta 4 is a major component of hemidesmosomes, in which it is linked to intermediate filaments. Its presence in these structures is dependent on the beta 4 cytoplasmic domain but it is not known whether beta 4 interacts directly with keratin filaments or by interaction with other proteins. In this study, we have investigated the interaction of GST-cyto beta 4A fusion proteins with cellular proteins and demonstrate that a fragment of beta 4A, consisting of the two pairs of fibronectin type III repeats, separated by the connecting segment, forms a specific complex containing a 500-kDa protein that comigrates with HD1, a hemidesmosomal plaque protein. A similar protein was also bound by a glutathione S-transferase fusion protein containing the cytoplasmic domain of a variant beta 4 subunit (beta 4B), in which a stretch of 53 amino acids is inserted in the connecting segment. Subsequent immunoblot analysis revealed that the 500-kDa protein is in fact HD1. In COS-7 cells, which do not express alpha 6 beta 4 or the hemidesmosomal components BP230 and BP180, HD1 is associated with the cytoskeleton, but after transfecting the cells with cDNAs for human alpha 6 and beta 4, it was, instead, colocalized with alpha 6 beta 4 at the basal side of the cells. The organization of the vimentin, keratin, actin, and tubulin cytoskeletal networks was not affected by the expression of alpha 6 beta 4 in COS-7 cells. The localization of HD1 at the basal side of the cells depends on the same region of beta 4 that forms a complex containing HD1 in vitro, since the expression of alpha 6 with a mutant beta 4 subunit that lacks the four fibronectin type III repeats and the connecting segment did not alter the distribution of HD1. The results indicate that for association of alpha 6 beta 4 with HD1, the cytoplasmic domain of beta 4 is essential. We suggest that this association may be crucial for hemidesmosome assembly.     

The role of microtubules in platelet secretory release

The role of microtubules in platelet aggregation and secretion has been analyzed using platelets permeabilized with digitonin and monoclonal antibodies to alpha (DM1A) and beta (DM1B) subunits of tubulin. Permeabilized platelets were able to undergo aggregation and secretory release. However, threshold doses of agonists capable of eliciting a second wave of aggregation and the platelet release reaction were higher than in control platelets exposed to dimethyl sulfoxide, the solvent for digitonin. Both antibodies to alpha and beta tubulin caused a further increase in the threshold concentration of agonists and inhibited the secretory release of permeabilized platelets, but were ineffective using intact platelets. Neither monoclonal antibody inhibited polymerization or depolymerization of platelet tubulin in vitro. Antibodies to platelet actin and myosin also exhibited an inhibitory activity on platelet aggregation albeit less severe than that observed with the antibodies to alpha and beta tubulin. There was evidence of an interaction between DM1A and DM1B and the antibodies to actin and myosin. The interaction of platelet tubulin and myosin was investigated by two different methods. (1) Coprecipitation of the proteins at low ionic strength at which tubulin by itself did not precipitate and (2) affinity chromatography on columns of immobilized myosin. Tubulin freed of its associated proteins (MAPs) by phosphocellulose chromatography bound to myosin in a molar ratio which approached 2. Platelet actin competed with tubulin for 1 binding site on the myosin molecule. MAPs also reduced the binding stoichiometry of tubulin/myosin. Treatment of microtubule protein with p-chloromercuribenzoate or colchicine did not influence its binding to myosin. DM1A and DM1B inhibited the interaction of tubulin and myosin. This effect could also be demonstrated by reaction of electrophoretic transblots of extracted platelet tubulin with the respective proteins. We interpret these results as evidence for an interference of the two monoclonal antibodies to the tubulin subunits (DM1A and DM1B) with the translocation of microtubule protein from its submembranous site to a more central one during the activation process.     

Colchicine-sensitive and colchicine-insensitive intermediate filament systems distinguished by a new intermediate filament-associated protein, IFAP-70/280 kD.

A monoclonal antibody was produced, using as antigen a BHK-21 cytoskeletal preparation enriched in intermediate filaments (IF) and their associated proteins. This antibody reacted exclusively with a reproducible set of 70-280 kD polypeptides present in minor quantities in this preparation, as detected by immunoblot analysis. Based upon several criteria, this immunologically related group of polypeptides was designated as IFAP-70/280 kD (IF-Associated Protein): (1) it co-isolated with IF in vitro, (2) it co-localized (by both immunofluorescence and immunoelectron microscopy) with IF in situ in all stages of cell spreading, and (3) it segregated in vitro with the 54/55 kD (desmin/vimentin) structural IF subunit proteins of BHK cells through two cycles of in vitro disassembly/assembly. Immunogold labeling further localized IFAP-70/280 kD to regions of parallel or loosely bundled IF in situ, suggesting a role in regulating the supramolecular organization of IF. When this monoclonal antibody was used for double-label immunofluorescence observations of colchicine-treated BHK cells, it demonstrated the presence of colchicine-sensitive and colchicine-insensitive IF. Anti-IFAP-70/280 kD localized entirely to the drug-induced juxtanuclear IF cap, while a polyclonal antibody directed against the desmin/vimentin structural IF subunits and the previously characterized monoclonal anti-IFAP-300 kD [Yang et al., 1985; J. Cell Biol. 100:620] localized to both the juxtanuclear IF cap and a colchicine-insensitive IF network peripheral to the cap in the same cells. The colchicine-insensitive IF pattern often exhibited similarities to that observed for the actin-based stress fiber system, suggesting that stress fiber association may be an additional factor in IF organization.     

A discrete repeated sequence defines a tubulin binding domain on microtubule-associated protein tau

The protein domain responsible for the interaction of tau with tubulin has been identified. Biophysical studies indicated that the synthetic peptide Val187-Gly204 (VRSKIG-STENLKHQPGGG) from the repetitive sequence on tau binds to two sites on the tubulin heterodimer and to one site on each of the microtubule-associated protein-interacting C-terminal tubulin peptides alpha(430-441) and beta(422-434). The binding data showed a relatively stronger interaction of Val187-Gly204 with beta(422-434) as compared to that with alpha(430-441). The interaction of this tau peptide with either alpha or beta tubulin peptides appears to be associated with conformational changes in both the tau and the tubulin peptides. The beta tubulin peptide also appears to induce a structural change of tau fragment Val218-Gly235. Interestingly, tau peptides Val187-Gly204 and Val218-Gly235 induced tubulin self-assembly in a cold-reversible fashion, and incorporated into the assembled polymers. The specificity of the interaction of the tau peptide was supported by the competition of tau protein for the interaction with the tubulin polymer. In addition, the tau peptide appears to contain the principal antigenic determinant(s) recognized by anti-idiotypic antibodies that react with the tubulin binding domains on microtubule-associated proteins. The present findings together with the demonstration of the presence of multiple sites for the binding of the alpha(430-441) and beta(422-434) tubulin fragments to tau, and the existence of repetitive sequences on tau, strongly support the hypothesis that the region of tau defined by the repetitive sequences is involved in its interaction with tubulin.     

Human S100B protein interacts with the Escherichia coli division protein FtsZ in a calcium-sensitive manner

S100B is a small, dimeric EF-hand calcium-binding protein abundant in vertebrates. Upon calcium binding, S100B undergoes a conformational change allowing it to interact with a variety of target proteins, including the cytoskeletal proteins tubulin and glial fibrillary acidic protein. In both cases, S100B promotes the in vitro disassembly of these proteins in a calcium-sensitive manner. Despite this, there is little in vivo evidence for the interaction of proteins such as tubulin with S100B. To probe these interactions, we studied the expression of human S100B in Escherichia coli and its interaction with the prokaryotic ancestor of tubulin, FtsZ, the major protein involved in bacterial division. Expression of S100B protein in E. coli results in little change in FtsZ protein levels, causes a filamenting bacterial phenotype characteristic of FtsZ inhibition, and leads to missed rounds of cell division. Further, S100B localizes to positions similar to those of FtsZ in bacterial filaments: the small foci at the poles, the mid-cell positions, and between the nucleoids at regular intervals. Calcium-dependent physical interaction between S100B and FtsZ was demonstrated in vitro by affinity chromatography, and this interaction was severely inhibited by the competitor peptide TRTK-12. Together these results indicate that S100B interacts with the tubulin homologue FtsZ in vivo, modulating its activity in bacterial cell division. This approach will present an important step for the study of S100 protein interactions in vivo.     

Intermediate filament expression in non-neoplastic pituitary cells.

Fifty-one non-neoplastic human pituitary glands, including examples with Crooke's hyalinization or amyloidosis, were examined by an immunoperoxidase method using antibodies to keratin, vimentin, neurofilaments (NFs), glial fibrillary acidic protein (GFAP), desmin, actin, S-100 protein and a variety of pituitary hormones. It was confirmed that most of the epithelial cells in the pituitary gland express keratin immunoreactivity. These cells included endocrine cells in the anterior lobe, endocrine cells and squamous metaplastic cells in the pars tuberalis, columnar and ciliated epithelia forming follicular structures and salivary-type epithelium in the pars intermedia, and anterior lobe cells infiltrating the posterior lobe. This study also demonstrated that keratin and NFs may be co-expressed in endocrine cells in the pituitary anterior lobe, that keratin, vimentin and GFAP may be co-expressed in the epithelial cells forming cyst-like follicle in the pars intermedia, and that vimentin and GFAP may be co-expressed in folliculo-stellate cells and pituicytes. In addition, the GFAP and S-100 protein-negative high columnar epithelium in the pars intermedia tended to be positive for adrenocorticotropic hormone and melanocyte stimulating hormone, while the low columnar epithelium with the co-expression of GFAP and S-100 protein was negative for pituitary hormones.     

Preparation of a monoclonal antibody specific for the class IV isotype of beta-tubulin. Purification and assembly of alpha beta II, alpha beta III, and alpha beta IV tubulin dimers from bovine brain.

Tubulin, the 100-kDa subunit protein of microtubules, is a heterodimer of two 50-kDa subunits, alpha and beta. Both alpha and beta subunits exist as numerous isotypic forms. There are four isotypes of beta-tubulin in bovine brain tubulin preparations; their designations and relative abundances in these preparations are as follows: beta I, 3%; beta II, 58%; beta III, 25%; and beta IV, 13%. We have previously reported the preparation of monoclonal antibodies specific for beta II and beta III (Banerjee, A., Roach, M. C., Wall, K. A., Lopata, M. A., Cleveland, D. W., and Luduena, R. F. (1988) J. Biol. Chem. 263, 3029-3034; Banerjee, A., Roach, M. C., Trcka, P., and Luduena, R. F. (1990) J. Biol. Chem. 265, 1794-1799). We here report the preparation of a monoclonal antibody specific for beta IV. By using this antibody together with those specific for beta II and beta III, we have prepared isotypically pure tubulin dimers with the composition alpha beta II, alpha beta III, and alpha beta IV. We have found that, in the presence of microtubule-associated proteins, all three dimers assemble into microtubules considerably faster and to a greater extent than does unfractionated tubulin. More assembly was noted with alpha beta II and alpha beta III than with alpha beta IV. When assembly is measured in the presence of taxol (10 microM), little difference is seen among the isotypically purified dimers or between them and unfractionated tubulin. These results indicate that the assembly properties of a tubulin preparation are influenced by its isotypic composition and raise the possibility that the structural differences among tubulin isotypes may have functional significance.