Interaction of nerve growth factor with tubulin. Studies on binding and induced polymerization

The interaction of the nerve growth factor with the neurotubule protein has been studied with the aim of elucidating the nature of the large complexes that they form when incubated together and the factors that control this event. The results show that the binding of nerve growth factor to tubulin is followed by the formation of large structures that, in certain experimental conditions, accelerate the rate of tubulin polymerization to form microtubules or catalyze their assembly in conditions where this process does not occur spontaneously. The formation of large nerve growth factor-tubulin complexes starts to occur only at a molar ratio of 1.0–1.5 NaCl or GTP strongly inhibit this process without a detectable effect on NGF binding. Two hypotheses are postulated to explain these finding. Firstly, that tubulin has two sites with different affinity for nerve growth factor and the polymerization occurs only when the second NGF molecule has interacted with the microtubule protein. Alternatively, free tubulin in solution is the limiting factor of the polymerization by hindering a site of tubulin-factor complexes present in solutio at a 1 : 1 molar ratio. In both cases, GTP, Na⁺ or H⁺ will affect the formation of large unsoluble, tubulin-NGF complexes, by changing their conformation or by decreasing electrostatic interactions.     

2'-Deoxy-GTP in the microtubule cytoskeleton of neuronal cells cultured with nerve growth factor.

Tubulin, widely recognized as a GTP/GDP-binding protein, has been isolated in its polymerized state from rat PC12 cells and embryonic chick dorsal root ganglion neurons by Triton X-100 detergent extraction of the cytoskeletal fraction. Perchloric acid extraction and deproteinization of this fraction permitted subsequent analysis of nucleotide identity and content by high performance liquid chromatography. PC12 cells grown in the absence of nerve growth factor (NGF) contained ADP, ATP, GDP, and GTP at levels consistent with the actin and tubulin content of the cytoskeletal fraction. Microtubules from PC12 cells cultured in the presence of NGF contain an additional nucleotide that we have identified as dGTP. Analysis of whole cell nucleotide extracts from PC12 cells grown in the absence or presence of NGF revealed no evidence for the presence of dGTP at 4 and 14 days, respectively. We have determined that embryonic chick dorsal root ganglion neurons also contain this deoxyribonucleotide, and we found virtually no ADP or ATP in the extracted dorsal root ganglion cytoskeletal fraction. On the basis of metabolic labeling studies with [14C] guanine, we have inferred that the presence of dGTP in NGF-treated PC12 cells probably arises either from binding to the nonexchangeable nucleotide site of tubulin undergoing dynamic assembly/disassembly or from binding to the exchangeable site of tubulin subsequently incorporated into highly stabilized microtubules.     

Differential effects of magnesium on tubulin-nucleotide interactions

Magnesium-depleted 2-(N-morpholino)ethanesulfonate (Mes), glutamate, tubulin and microtubule-associated proteins were prepared and used to study the effects of exogenously added MgCl2 on tubulin-nucleotide interactions in 0.1 M Mes with microtubule-associated proteins and in 1.0 M glutamate. Endogenous levels of Mg2+ in the systems studied were approximately stoichiometric with the tubulin concentrations and largely derived from the tubulin. We examined the effects of added Mg2+ on tubulin polymerization, GDP inhibition of polymerization, binding of GDP and GTP to tubulin, and GTP hydrolysis. Exogenously added Mg2+ had markedly different effects on these reactions. The order of their sensitivity for a requirement for added Mg2+ was as follows: GTP binding greater than GTP hydrolysis greater than polymerization greater than GDP binding. Inhibition of polymerization by GDP varied inversely with the Mg2+ concentration and was greatest in the absence of the cation. These results indicate that GDP and GDP-Mg2+ interact with similar affinity at the exchangeable site, while GTP-Mg2+ has a higher affinity for tubulin than does free GTP. Nevertheless, under appropriate conditions, free GTP can interact sufficiently well with tubulin to permit both nucleation and elongation reactions.     

Polymerization and colchicine binding. Two independent properties of tubulin

When stored frozen in 1 M sucrose and 1 mM GTP, tubulin loses polymerizing ability exponentially. Since addition of diethiothreitol does not change the decay half-life, this decrease in activity can not be attributed to disulfide bond formation. When tubulin is stored frozen in dithiothreitol and GTP only, the decay half-life increases by a factor of four, indicating that sucrose destabilizes polymerizing ability. Frozen storage in sucrose has the opposite effect on colchicine binding, which remains at 100% for 40 days. This temporal divergence indicates that colchine binding and polymerization are two independent properties of tubulin.     

Linkages between the effects of taxol, colchicine, and GTP on tubulin polymerization

A comparative study has been carried out of the effects of taxol on the polymerizations into microtubules of microtubule-associated protein-free tubulin, prepared by the modified Weisenberg procedure, and of the tubulin-colchicine complex into large aggregates. Taxol enhances, to a much greater extent, the stability of microtubules than that of the tubulin-colchicine polymers so that, with highly purified tubulin, assembly into microtubules takes place at 10 degrees C, even in the absence of exogenous GTP. The polymerization of tubulin-colchicine requires both heat and GTP, and the process is reversed by cooling. These results indicate that in both systems polymerization is linked to interactions with taxol and GTP, the interplay of linkage free energies imparting the observed polymer stabilities. In the case of microtubule formation, the linkage free energy provided by taxol binding is approximately -3.0 kcal/mol of alpha-beta-tubulin dimer, whereas this quantity is reduced to approximately -0.5 kcal/mol in tubulin-colchicine, indicating the expenditure of much more binding free energy in the latter case for overcoming unfavorable factors, such as steric hindrance and geometric strain. The difference in the effect of GTP on the two polymerization processes reflects the respective abilities of the bindings of taxol to the two states of tubulin to overcome the loss of the linkage free energy of GTP binding. Analysis of the linkages leads to the conclusions that taxol need not change qualitatively the mechanism of microtubule assembly and that tubulin with the E-site unoccupied by nucleotide should have the capacity to form microtubules, the reaction being extremely weak.     

Modulation of tubulin-nucleotide interactions by metal ions: comparison of beryllium with magnesium and initial studies with other cations.

With microtubule-associated proteins (MAPs) BeSO4 and MgSO4 stimulated tubulin polymerization as compared to a reaction mixture without exogenously added metal ion, while beryllium fluoride had no effect (E. Hamel et al., 1991, Arch. Biochem. Biophys. 286, 57-69). Effects of both cations were most dramatic at GTP concentrations in the same molar range as the tubulin concentration. We have now compared effects of beryllium and magnesium on tubulin-nucleotide interactions in both unpolymerized tubulin and in polymer. Polymer formed with magnesium had properties similar to those of polymer formed without exogenous cation, except for a 20% lower stoichiometry of exogenous GTP incorporated into the latter. In both polymers the incorporated GTP was hydrolyzed to GDP. Stoichiometry of GTP incorporation into polymers formed with beryllium or magnesium was identical, but much of the GTP in the beryllium polymer was not hydrolyzed. The beryllium polymer was more stable than the magnesium polymer. Beryllium also differed from magnesium in only weakly enhancing the binding of GTP in the exchangeable site of unpolymerized tubulin, while neither cation affected GDP exchange at the site. If both cations were present in a reaction mixture, polymer stability was little changed from that of the beryllium polymer, but most of the GTP incorporated into polymer was hydrolyzed. Six additional metal salts (AlCl3, CdCl2, CoCl2, MnCl2, SnCl2, and ZnCl2) also stimulated MAP-dependent tubulin polymerization, but enhanced polymer stability did not correlate with polymer GTP content. We postulate that enhanced polymer stability is a consequence of cation binding directly to tubulin and/or polymer while deficient GTP hydrolysis in the presence of beryllium, as well as aluminum and tin, is a consequence of tight binding of cation to GTP in the exchangeable site.     

Microtubular protein reaction with nucleotides.

The dissociation constants for GTP and GDP with tubulin were determined to be equal to 1.1 ± 0.4 × 10^?7 M and 1.5 ± .6 × 10^?7 (4°), respectively. A lower limit for the dissociation constant for ATP was established as equal to 6 × 10^?4 M. The equivalent binding of GTP and GDP is not readily consistent with a mechanism in which the role of GTP in microtubule assembly is to bind to the protein to induce a conformation which is able to polymerize. An ATP-induced polymerization of tubulin apparently involves a transphosphorylation reaction in which GTP is formed and mediates the assembly. For this reaction to occur with desalted tubulin trace amounts of GDP are required; in the reaction of 0.1 mM ATP with 22.0 μM tubulin, 0.1 μM GDP induces about 80% as much tubule formation as is seen with 0.1 mM GTP alone.     

Differentiation factors, including nerve growth factor, fibroblast growth factor, and interleukin-6, induce an accumulation of an active Ras.GTP complex in rat pheochromocytoma PC12 cells.

Ras has been thought to be involved in neuronal differentiation of rat pheochromocytoma PC12 cells. PC12 cells are immature adrenal chromaffin-like cells which undergo differentiation to sympathetic neuron-like cells in response to nerve growth factor (NGF). Fibroblast growth factor (FGF) and interleukin (IL)-6 can also induce differentiation of PC12 cells. In this paper, we report that NGF, FGF, and IL-6 induce an accumulation of an active Ras.GTP complex. In the serum-starved culture of PC12 cells, 6% of the Ras protein was complexed with GTP. Upon stimulation with NGF, the percentage of Ras.GTP increased to 24% after 2 min, and the high level of Ras.GTP was maintained for at least 16 h. On the other hand, the activation of Ras by FGF and IL-6 showed distinct kinetics; about 3-fold increase of Ras.GTP was detected at 10 min, and afterward, the level returned to the basal level within 60 min. These observations provide direct evidence that activation of Ras is involved in signal transduction from these differentiation factors. In addition, it was found that growth factors, including epidermal growth factor, insulin, and insulin-like growth factor-I, and a tumor promoter, 12-O-tetradecanoylphorbol-13-acetate (TPA), can also activate Ras under the same conditions. A tyrosine kinase-specific inhibitor, genistein, inhibited the increase of Ras.GTP induced by NGF and other factors. On the other hand, down-regulation of protein kinase C (PKC) by prolonged treatment with TPA, which sufficiently blocked TPA-induced Ras activation, did not abolish the formation of Ras.GTP by NGF. These results suggest that tyrosine kinases rather than PKC play a major role in the NGF-induced activation of Ras in PC12 cells.     

Taxol effect on tubulin polymerization and associated guanosine 5'-triphosphate hydrolysis

Taxol has been used as a tool to investigate the relationship between microtubule assembly and guanosine 5'-triphosphate (GTP) hydrolysis. The data support the model previously proposed [Carlier, M.-F., & Pantaloni, D. (1981) Biochemistry 20, 1918] that GTP hydrolysis is not tightly coupled to the polymerization process but takes place as a monomolecular process following polymerization. The results further indicate that the energy liberated by GTP hydrolysis is not responsible for the subsequent blockage of GDP on polymerized tubulin. When tubulin is polymerized in the presence of 10-100 microM taxol, the rapid formation of a large number of very short microtubules (l less than 1 micron) is accompanied by the development of turbidity to a lesser extent than what is observed when the same weight amount of longer microtubules (l = 5 microns) is formed. A slower subsequent turbidity increase corresponds to the length redistribution of these short microtubules into 3-5-fold longer ones without any change in the weight amount of polymer. The evolution of the rate of length redistribution with the concentration of taxol suggests a model within which taxol would bind to dimeric tubulin and to tubulin present at the ends of microtubules with a somewhat 10-fold lower affinity than to polymerized tubulin embedded in the bulk of microtubules. In agreement with this model, binding of taxol to the tubulin-colchicine complex in the dimeric form could be measured from the increase in the GTPase activity of the tubulin-colchicine complex accompanying taxol binding.     

Binding of guanine nucleotides and Mg2+ to tubulin with a nucleotide-depleted exchangeable site

Binding of GTP and GDP to tubulin in the presence or absence of Mg2+ was measured following depletion of the exchangeable site--(E-site) nucleotide. The E-site nucleotide was displaced with a large molar excess of the nonhydrolyzable GTP analogue, GMPPCP, followed by the removal of the analogue. Using a micropartition assay, the equilibrium constant measured in 0.1 M 1.4-piperazinediethanesulfonic acid (Pipes), pH 6.9, 1 mM ethylene glycol bis(beta-aminoethyl ether) N,N,N',N'-tetraacetic acid, 1 mM dithiothreitol, and 1 mM MgSO4 at 4 degrees C was 9.1 x 10(6) M-1 for GTP and 4.4 x 10(6) M-1 for GDP. Removal of Mg2+ reduced the binding affinity of GTP by 160-fold while the affinity of GDP remained essentially unchanged. Similar values were obtained if 0.1 M Tris, pH 7.0, was used instead of Pipes. Binding of Mg2+ to tubulin containing GTP, GDP, or no nucleotide at the E-site was also examined by the micropartition method. Tubulin-GTP contained one high affinity Mg2+ site (K alpha = 1.2 x 10(6) M-1) in addition to the one occupied by Mg2+ as tubulin is isolated, while only weak Mg2+ binding to tubulin-GDP and to tubulin with a vacant E-site (K alpha = 10(3) M-1) was observed. It is suggested that Mg2+ binds to the beta and gamma phosphates of GTP, and only to the beta phosphate of GDP, as shown for the H. ras p21 protein.     

Synthetic chimeras of mouse growth factor-associated glandular kallikreins. II. Growth factor binding properties.

Six chimeric constructs of the sequentially similar growth factor-associated kallikreins-epidermal growth factor binding protein (EGF-BP) and the gamma-subunit of nerve growth factor (gamma-NGF)--have been expressed, and their ability to generate complexes with epidermal growth factor (EGF) and beta-NGF, analogous to the high molecular weight forms (7S NGF and HMW-EGF) found in the mouse submaxillary gland, evaluated. The chimeras are distinguished by the interchange of three regions composing the amino, middle, and carboxyl terminal regions that encompass four surface loops possibly involved in specific growth factor interactions. Native beta-NGF (along with native alpha-NGF) formed complexes indistinguishable from naturally occurring 7S NGF, characterized by an alpha 2 beta gamma 2 structure (where beta-NGF is itself a dimer), with recombinant (r) gamma-NGF and with a chimera in which the amino terminal region from EGF-BP was substituted. Two other chimeras containing either the middle or carboxyl terminal regions of gamma-NGF showed weaker ability to form 7S complexes. Thus, all chimeras containing two segments from gamma-NGF retained at least some ability to form the 7S complex. rEGF-BP reacted weakly with EGF, but the chimera composed of the amino and middle segments of EGF-BP and the carboxyl terminal segment of gamma-NGF formed a nativelike HMW-EGF complex. None of the other chimeras appeared to bind EGF. These results identify amino acid positions within each kallikrein that participate in strong growth factor interactions and demonstrate that, outside of active site contacts, different regions of the kallikreins are involved in the binding of EGF and beta-NGF, respectively.     

Phosphofructokinase from Dictyostelium discoideum is a potent inhibitor of tubulin polymerization

We identified the nonallosteric phosphofructokinase from the slime mold Dictyostelium discoideum as a potent protein factor that inhibits the rate of polymerization of tubulin at a molar ratio of 1 molecule to about 300 tubulin dimers for half-maximal action (IC50 = 32 nM). This effect was (i) assessed by turbidity measurements, pelleting of microtubules, and electron microscopy, (ii) observed when tubulin assembly was induced by taxol as well as by GTP in the presence of microtubule-associated proteins or glutamate, and (iii) specific as it was not produced by the phosphofructokinase from rabbit muscle. Also in contrast to the latter, neither tubulin nor microtubules modified the catalytic activity of the slime mold isozyme. Immunoelectron microscopy provided further evidence that D. discoideumphosphofructokinase physically interacts with tubulin, leading to the formation of aggregates. The process seems to be reversible since microtubules eventually formed in the presence of the inhibitor with concomitant reduction of tubulin aggregates. Limited proteolysis by subtilisin showed that the hypervariable C-termini of tubulin is not involved in the interaction with the enzyme. The possible physiological relevance of this novel function of D. discoideum phosphofructokinase different from its glycolytic action is discussed.     

Dense precipitate of brain tubulin with skeletal muscle myosin

Purified tubulin from porcine brain formed a dense precipitate at 37 degrees C with muscle myosin filaments from rabbit skeletal muscle; this effect was greater than that with partially purified tubulin. ATP or GTP, which prevented the myosin filaments from precipitating, inhibited the formation of the dense precipitate, but did not dissociate the dense precipitate once formed. The dense precipitate was found by thin-section electron microscopy to be composed to side-by-side aggregates of myosin filaments whose projections might be decorated by tubulin. The decoration was also seen by negative-stain electron microscopy. The binding of tubulin to myosin filaments decreased the Mg2+- and Ca2+-GTPase activity of the myosin by about half, but did not affect either Mg2+- or Ca2+-ATPase activity. The binding ratio of tubulin to myosin in the presence of 5 mM MgCl2 was 2.2 mol/mol using purified tubulin and 1.8 mol/mol using partially purified tubulin. Five mM ATP and GTP in the presence of 5 mM MgCl2 decreased the tubulin binding by 1.6-2.0 and 1.1-1.3 mol/mol, respectively, when added before an encounter of tubulin with myosin filaments, but did not cause any decrease when added after such an encounter.     

Guanosine 5'-O-(3-thiotriphosphate), a potent nucleotide inhibitor of microtubule assembly

Guanosine 5'-O-(3-thiotriphosphate) (GTP gamma S) and the two diastereoisomers of guanosine 5'-O-(2-thiotriphosphate) (GTP beta S) were prepared enzymatically, and their interactions with tubulin and microtubule-associated proteins (MAPs) in 0.1 M 2-(N-morpholino)ethanesulfonate, 0.5 mM MgCl2 were examined. GTP gamma S did not support microtubule assembly but instead inhibited the reaction. This analog was 1.5-2 times more potent than GDP in inhibiting both tubulin polymerization and GTP hydrolysis under conditions in which these reactions were dependent on MAPs. In contrast to the analog's inhibitory effects on polymerization and hydrolysis, however, radiolabeled GTP gamma S was only feebly bound by purified tubulin at 0 degrees C relative to the binding of GDP and GTP. There was a marked increase in the amount of GTP gamma S bound when the reaction temperature was raised to 37 degrees C or when MAPs were included in the reaction mixture. Only when both MAPs were present and the higher reaction temperature was used did the binding of GTP gamma S exceed that of GDP. Since substitution of sulfur for oxygen in a molecule should decrease its hydrophilic properties, these findings suggest that the exchangeable nucleotide binding site of tubulin becomes more hydrophobic at higher temperatures and in the presence of MAPs. The two isomers of GTP beta S were able to support MAP-dependent polymerization, although a 50-100-fold higher concentration of the analogs as compared to GTP was required. Neither isomer of GTP beta S had a significant inhibitory effect on GTP hydrolysis dependent on tubulin + MAPs.     

Regulation of microtubule protein levels during cellular morphogenesis in nerve growth factor-treated PC12 cells

Nerve growth factor induces neurite process formation in pheochromacytoma (PC12) cells and causes the parallel increase in levels of the microtubule-associated proteins, tau and MAP1, as well as increases in tubulin levels. Mechanisms to insure balanced accumulation of microtubule proteins and make their levels highly responsive to nerve growth factor were investigated. The effects on tau, MAP1, and tubulin are due to changes in protein synthesis rates, which for tau and tubulin we could show are due in part to changes in the mRNA levels. Whereas tubulin shows feedback regulation to modulate synthesis up or down, tau protein synthesis is not affected in a straightforward way by microtubule polymerization and depolymerization. The degradation of tau, MAP1, and both tubulin polypeptides, however, are stimulated by microtubule depolymerization caused by colchicine, or nerve growth factor removal. Combined feedback on synthesis and stability make tubulin levels highly responsive to assembly states. In addition, the linkage of tau and MAP1 turnover with the state of microtubule polymerization amplifies any change in their rate of synthesis, since tau and MAP1 promote microtubule polymerization. This linkage lends itself to rapid changes in the state of the system in response to nerve growth factor.     

Thrombin, epidermal growth factor, and phorbol myristate acetate stimulate tubulin polymerization in quiescent cells: a potential link to mitogenesis.

Previous studies suggest that alterations in the microtubule (MT)-tubulin equilibrium during G0/G1 affect mitogenesis. To determine the effect of growth factors on the MT-tubulin equilibrium, we developed a radioactive monoclonal antibody binding assay (Ball et al.: J. Cell. Biol. 103:1033-1041, 1986). With this assay, 3H-Ab 1-1.1 binding to cytoskeletons in confluent populations of cultured cells is proportional to the number of tubulin subunits polymerized into MTs. We now show that purified alpha-thrombin increases 3H-Ab 1-1.1 binding to cytoskeletons of serum-arrested mouse embryo (ME) fibroblasts from 1.5- to 3-fold. This stimulation is dose-dependent and correlates with concentrations of thrombin required for initiation of DNA synthesis. Other mitogenic factors, epidermal growth factor (EGF) and phorbol 12-myristate 13-acetate (PMA), also stimulate MT polymerization. Addition of colchicine (0.3 microM) eight hours after growth factor addition, blocks stimulation of 3H-thymidine incorporation by thrombin, EGF, or PMA, suggesting that tubulin polymerization or subsequent events triggered by MT polymerization are required for cells to enter a proliferative cycle. Consistent with models for autoregulation of tubulin synthesis, thrombin, EGF, and PMA all increase tubulin synthesis 9 to 15 hr after growth factor addition, raising the possibility that the decrease in free tubulin and subsequent stimulation of tubulin synthesis is linked to progression of cells into a proliferative cycle. Colchicine addition to these cells also stimulates DNA synthesis, but colchicine-stimulated cells enter S phase 6 to 8 hr later than those stimulated by growth factors. This delayed stimulation may be related to the time required for degradation of tubulin-colchicine complexes below a critical level. These data suggest that regulation of cell proliferation may be linked to increased MT polymerization and the resulting decrease in free tubulin pools.     

Evidence that the tightly bound magnesium in tubulin is associated with the N-site GTP

In an attempt to determine whether the tightly bound Mg2+ found in purified tubulin in associated with the N-site GTP or the E-site GDP or GTP, we removed the E-site nucleotide by several means: (i) alkaline phosphatase treatment; (ii) displacement using excess GMPPCP; and (iii) polymerizing tubulin in the presence of alkaline phosphatase and non-hydrolyzable analogues. The Mg2+ content remained equal to about 1 mol/mol tubulin under conditions where denaturation did not occur. Moreover, the Mg/GTP ratio always remained equal to 1. These results indicate that the Mg2+ is associated with the N-site GTP.     

Low resolution structure of microtubules in solution. Synchrotron X-ray scattering and electron microscopy of taxol-induced microtubules assembled from purified tubulin in comparison with glycerol and MAP-induced microtubules.

The structure of microtubules has been characterized to 3 nm resolution employing time-resolved X-ray scattering. This has revealed detailed structural features of microtubules not observed before in solution. The polymerization of highly purified tubulin, induced by the antitumour drug taxol, has been employed as a microtubule model system. This assembly reaction requires Mg2+, is optimal at a 1:1 taxol to tubulin heterodimer molar ratio, proceeds with GTP or GDP and is intrinsically reversible. The X-ray scattering profiles are consistent with identical non-globular alpha and beta-tubulin monomers ordered within the known helical surface lattice of microtubules. Purified tubulin-taxol microtubules have a smaller mean diameter (approx. 22 nm) than those induced by microtubule associated proteins or glycerol (approx. 24 nm), but nearly identical wall substructure to the resolution of the measurements. This is because the majority of the former consist of only 12 protofilaments instead of the typical 13 protofilaments, as confirmed by electron microscopy of thin-sectioned, negatively stained and ice-embedded taxol microtubules. It may be concluded that taxol induces a slight reduction of the lateral contact curvature between tubulin monomers. The main fringe pattern observed in cryo-electron micrographs is consistent with a simple 12 protofilament 3-start skewed lattice model. Cylindrical closure of this lattice can be achieved by tilting the lattice 0.8 degrees with respect to the microtubule axis. The closure implies a discontinuity in the type of lateral contacts between the tubulin monomers (regardless of whether these are of the -alpha-beta- or the -alpha-alpha-/-beta-beta- type), which indicates that lateral contacts and the subunit specificity of taxol binding are, to a large degree, equivalent.     

Nucleotide binding and phosphorylation in microtubule assembly in vitro.

Two non-hydrolyzable analogs of GTP, guanylyl-β,γ-methylene diphosphonate and guanylyl imidodiphosphate, have been found to induce rapid and efficient microtubule assembly in vitro by binding at the exchangeable site (E-site) on tubulin. Characterization of microtubule polymerization by several criteria, including polymerization kinetics, nucleotide binding to depolymerized and polymerized microtubules, and microtubule stability, reveals strong similarities between microtubule assembly induced by GTP and non-hydrolyzable GTP analogs. Nucleoside triphosphates which bind weakly or not at all to tubulin, such as ATP, UTP and CTP, are shown to induce microtubule assembly by means of a nucleoside diphosphate kinase (NDP-kinase, EC 2.7.4.6.) activity which is not intrinsic to tubulin. The NDP-kinase mediates microtubule polymerization by phosphorylating tubulin-bound GDP in situ at the E-site. Although hydrolysis of exchangeably bound GTP occurs, it is found to be uncoupled from the polymerization reaction. The non-exchangeable nucleotide binding site on tubulin (N-site) is not directly involved in microtubule assembly in vitro. The N-site is shown to contain almost exclusively GTP which is not hydrolyzed during microtubule assembly. A scheme is presented in which GTP acts as an allosteric effector at the E-site during microtubule assembly in vitro.