Thermodynamics of the Op18/Stathmin-Tubulin Interaction*

Op18/stathmin (stathmin) is an intrinsically disordered protein involved in the regulation of the microtubule filament system. One function of stathmin is to sequester tubulin dimers into assembly incompetent complexes, and recent studies revealed two tubulin binding sites per stathmin molecule. Using high sensitivity isothermal titration calorimetry, we document that at 10 °C and under the conditions of 80 mm PIPES, pH 6.8, 1 mm EGTA, 1 mm MgCl2, 1 mm GTP these two binding sites are of equal affinity with an equilibrium binding constant of K0 = 6.0 × 106 m-1. The obtained large negative molar heat capacity change of ΔCp0 = -860 cal mol-1 K-1 (referring to tubulin) for the tubulinstathmin binding equilibrium suggests that the hydrophobic effect is the major driving force of the binding reaction. Replacing GTP by GDP on β-tubulin had no significant effect on the thermodynamic parameters of the tubulin-stathmin binding equilibrium. The proposed pH-sensitive dual function of stathmin was further evaluated by circular dichroism spectroscopy and nuclear magnetic resonance. At low temperatures, stathmin was found to be extensively helical but devoid of any stable tertiary structure. However, in complex with two tubulin subunits stathmin adopts a stable conformation. Both the stability and conformation of the individual proteins and complexes were not significantly affected by small changes in pH. A 4-fold decrease in affinity of stathmin for tubulin was revealed at pH 7.5 compared with pH 6.8. This decrease could be attributed to a weaker binding of the C terminus of stathmin. These findings do not support the view that stathmin works as a pH-sensitive protein.

Microtubule (MT) 1 filaments are dynamic polymers made of ␣/␤-tubulin heterodimers that are essential for a wide variety of central cellular functions in all eucaryotes, including cell transport, cell motility, and mitosis. One of the key properties of MTs is that of "dynamic instability" (1). Dynamic instability comprises the continuous switching between catastrophes (depolymerization or shrinkage phase) and rescues (polymeriza-tion or growing phase) of individual MTs and is central to MT function. In recent years, it has become clear that the balance between MT-stabilizing and -destabilizing factors is responsible for the observed switching between growth and shrinkage of MTs in vivo (2). Therefore, obtaining molecular insights into the mechanisms of action of MT regulatory factors is essential to understand how they contribute to the dynamic state of MTs during the cell cycle.
The members of the phylogenetically well conserved Op18/ stathmin family are phosphorylation-controlled MT-destabilizing proteins (reviewed in Refs. [3][4][5][6][7][8]. They are important for proper cell cycle progression in many types of proliferating eucaryotic cells (3,4,8), were found to be crucial for the development of the nervous system in Drosophila (9), promote neurite outgrowth through regulation of MT dynamics in growth cones (10), and are implicated in a wide variety of cancers (11,12). Human Op18/stathmin (referred to as stathmin (3)) is an evolutionary well conserved 17-kDa cytoplasmic phosphoprotein. The monomeric protein consists of a N-terminal capping domain (13,14) and a C-terminal helical interaction domain (13,(15)(16)(17)(18)(19). Originally, stathmin was described as a protein that binds to tubulin dimers and increases the catastrophe frequency of MT in vitro (20). Subsequent studies have identified two putative mechanisms for stathmin: (i) Tubulin dimer sequestration that slows MT growth rate (21,22) and (ii) direct stimulation of MT plus-end catastrophes (23)(24)(25)(26)(27)(28)(29). The sequestering mechanism is largely supported by recent structural and biophysical studies (13, 14, 16 -18, 30 -32). As illustrated in Fig. 1, the C-terminal helical domain of the stathmin-like domain (SLD) of the neural isoform RB3 can bind two head-to-tail aligned ␣/␤-tubulin heterodimers. The tubulin subunits in the ternary complex are tilted with respect to each other by ϳ25°. The N-terminal part of stathmin was found to cap one ␣-tubulin monomer to prevent further longitudinal tubulin assembly. Hence, binding of stathmin to tubulin makes these sequestered tubulin subunits efficiently assembly incompetent.
However, other in vivo and in vitro studies converged to the hypothesis that stathmin works as a pH-sensitive bifunctional protein: At pH 6.8 stathmin is supposed to act as a sequestering protein, whereas at pH 7.5 it primarily stimulates MT plus-end catastrophes without significantly interacting with tubulin subunits (24,28). This view is supported by deletion mapping, which indicated that the N terminus of stathmin is required for plus-end catastrophe promotion, whereas the function of the C-terminal helical domain is to bind tubulin dimers (24,29). In contrast to the well characterized sequestering mechanism, the molecular details of how stathmin interacts with MT ends and triggers catastrophes are not clear.
Here we have combined isothermal titration calorimetry (ITC), transmission electron microscopy (TEM), circular dichroism (CD) spectroscopy, and nuclear magnetic resonance (NMR) spectroscopy to investigate the biophysical and structural characteristics of the interaction of stathmin with tubulin. We report that stathmin binds two tubulin subunits with equal affinities under all conditions investigated. The thermodynamic data suggest that the hydrophobic effect represents the major driving force governing the tubulin-stathmin binding reaction. The affinity of stathmin for tubulin as well as the stabilities and conformations of the individual proteins are not significantly affected by small changes in pH. Nevertheless, the small affinity change that was observed between pH 6.8 and 7.5 can be attributed to the weaker binding of the C terminus of stathmin. The results presented are discussed in the context of current functional and structural data on the tubulin-stathmin system.

EXPERIMENTAL PROCEDURES
Protein Preparations-The cloning of the human full-length stathmin cDNA into the bacterial expression vector pET-16b (Novagen) is outlined in Ref. 13. Recombinant stathmin was expressed in LB medium using the Escherichia coli host strain JM109(DE3) (Promega), affinity-purified as His 6 -tagged fusion protein by immobilized metal affinity chromatography on Ni 2ϩ -Sepharose (Amersham Biosciences), and separated from the N-terminal His 6 tag by proteolytic cleavage with an immobilized snake venom prothrombin activator (for details, see Ref. 13). After dialysis in 20 mM Tris-HCl, pH 8.0, a RESOURCE Q (Amersham Biosciences) anion exchange chromatography step using a NaCl gradient from 0 to 500 mM was performed.
All protein samples were centrifuged at 4°C and at full speed for 10 min in a tabletop Eppendorf centrifuge prior to the experiments. Concentrations of protein samples were determined by the Advanced Protein Assay (Cytoskeleton Inc.).
High Sensitivity ITC-ITC experiments were performed using a VP-ITC calorimeter (Microcal Inc., Northampton, MA). For each experiment, the sample cell (volume 1.4 ml) was filled with a ϳ10 M tubulin solution in G buffer pH 6.8 or 7.5 supplemented with either 1 mM GTP or 2 mM GDP. A 300-l syringe was filled with a solution of ϳ100 M stathmin (present in the same buffer as tubulin). The reference cell contained buffer only. Typically, 5-l stathmin aliquots from the stirred syringe (305 rpm) were injected 40 times into the sample cell. At each injection, tubulin was bound to stathmin, leading to a characteristic heat signal. Integration of the individual calorimeter traces yielded the heat of binding, h i , of each injection step. The heats of dilution of stathmin were subtracted from the titration data. During the course of the titration the total tubulin concentration is slightly diluted due to the addition of stathmin. This effect is small but nevertheless was taken into account in the evaluation. The binding isotherms were fitted via a non-linear least squares minimization method to determine the binding stoichiometry, n, the equilibrium binding constant, K 0 , and the change in enthalpy, ⌬H 0 (see "Results"). For all titrations, the experimental conditions warranted that tubulin remained in the soluble heterodimeric state.
Glycerol Spraying/Low-angle Rotary Metal Shadowing and TEM-Protein samples (0.5 mg/ml) in G buffer, pH 6.8, supplemented with 1 mM GTP were incubated for 1 h at room temperature after which glycerol was added to a final concentration of 30%. The samples were immediately sprayed onto freshly cleaved mica and rotary shadowed in a BA 511M freeze-etch apparatus (Balzers) with platinum/carbon at an elevation angle of 3-5° (36). Electron micrographs were taken in a Philips Morgagni TEM operated at 80 kV equipped with a Megaview III charge-coupled device camera. Approximately 400 particles were counted from single micrographs and visually classified into globular and elongated specimens, 8 -10 nm and 8 -10 ϫ 16 -20 nm, respectively.
CD Spectroscopy-Protein samples were in 8 mM PIPES-KOH, pH 6.8 or 7.5, 1 mM MgCl 2 , 1 mM EGTA, and 1 mM GTP. Far-UV CD spectra and thermal unfolding profiles were recorded on a Jasco J-810 spectropolarimeter (Jasco Inc.) equipped with a temperature-controlled quartz cell of 0.1-cm path length. The spectra shown are the averages of five accumulations and were evaluated with the Jasco and Sigma Plot (Jandel Scientific) software. A ramping rate of 1°C⅐min Ϫ1 was used to record the thermal unfolding profiles. The apparent midpoints of the transitions, values of T m , were taken as the maximum of the derivative, NMR Spectroscopy-NMR experiments were carried out at 23°C on a Varian UnityPlus spectrometer operating at a 1 H frequency of 600 MHz. Samples of 2 H/ 13 C/ 15 N-labeled stathmin and unlabeled tubulin were in G buffer at pH 6.8 or 7.5. Both 15 N-1 H HSQC and 15 N-1 H TROSY experiments were recorded. However, the TROSY spectra were not superior to the HSQC spectra, probably due to the high proton density in non-deuterated tubulin. The data were analyzed with the VNMR software.
FIG. 1. Current structural view of the ternary tubulin-stathmin complex. The structure of the complex has been investigated by electron microscopy (13), by a 4-Å resolution x-ray structure of the tubulin-RB3-SLD complex (note that RB3-SLD exhibits 72% identity with stathmin (17)), and by chemical cross-linking (14). The tubulinstathmin complex consists of two head-to-tail aligned ␣/␤-tubulin heterodimers (represented as light and dark gray ribbons) that are tilted with respect to each other by ϳ25°. The 4-Å x-ray structure shows the C-terminal domain of stathmin folded into an extended 91-residue helix (represented as a blue ribbon) that interacts in a regular fashion along the two longitudinally aligned tubulin dimers. Because of the lack of resolution, the interaction face of the helix with tubulin could not be defined. The N-terminal domain (schematically represented as yellow slip knot), which was not resolved in the x-ray electron density map, was found to bind close to helix 10 (red helix at the tip of the ␣1-tubulin monomer) and to the loop connecting helix 10 with beta strand 9 of the ␣1-tubulin monomer (14). These represent critical secondary structural elements that are involved in establishing both longitudinal as well as lateral protofilament contacts of tubulin subunits within the MT wall (48). Four contact points (denoted T ␣1 -S N , T ␣1␤1 -S h1 , T ␤1 -T ␣2 , and T ␣2␤2 -S h2 ; see also "Discussion") between the two tubulin subunits and stathmin are indicated by dotted blue (for the first tubulin subunit) and red (for the second tubulin subunit) circles.

Analysis of the Tubulin-Stathmin Binding Equilibrium by
ITC-To assess the thermodynamic binding parameters of the tubulin-stathmin binding equilibrium, ITC studies were performed. Fig. 2A shows a typical titration profile in which 12.8 M GTP-tubulin (referred to as the ␣/␤-tubulin heterodimer with GTP occupying the exchangeable E-site on ␤-tubulin (2)) is titrated with stathmin. Each titration peak corresponds to the injection of 5-l aliquots of a 84.0 M stathmin solution. At 10°C, the reaction is endothermic, and the reaction heats decrease with increasing injection numbers as less and less free tubulin is available for stathmin binding. The total stathmin concentration in the calorimeter cell varies during the course of the titration and increases from 0.3 M after the first injection to 10.5 M after 40 injections. After ϳ30 injections, all tubulin is bound to stathmin, and no further heat is produced. Fig. 2B shows the integrated heats of reaction as a function of the stathmin:tubulin molar ratio.
The analysis of the binding isotherm is based on a model that assumes n independent and equal binding sites on stathmin for tubulin. If C S 0 , C T,bound , and C T,free denote the concentration of total stathmin in the calorimeter cell, and the concentration of bound and free tubulin, respectively, then the binding isotherm can be described as (37), where K 0 is the intrinsic equilibrium binding constant for a single binding site. Equation 1 can be solved for C T,bound via, where C T 0 is the total concentration of tubulin in the calorimeter cell. If we denote with h i the measured heat of the ith injection step and by r (k) the fraction of bound tubulin after a total of k injections, then Equation 3 results.
The numerator gives the cumulative heat of the first k injections, the denominator is the total heat produced when the supply of the tubulin is exhausted after a total of m injections. The concentrations of bound and free tubulin are then given by Equations 4 and 5.
C T,bound and C T,free can thus be determined without any assumptions about the binding mechanism. Because C T,bound is directly accessible via the heats of titration, it is possible to calculate unambiguously the stoichiometry, n, the apparent molar heat of reaction, ⌬H 0 , and the equilibrium binding constant, K 0 , via a non-linear least squares fit to Equation 2 (38,39). The solid line in Fig. 2B calculated with n ϭ 2 represents the best fit to the experimental data and yielded ⌬H GTP-tubulin 0 ϭ 15.0 Ϯ 0.14 kcal mol Ϫ1 GTP-tubulin, and K 0 ϭ 6.0 ϫ 10 6 Ϯ 0.61 ϫ 10 6 M Ϫ1 .
The classic way of representing biochemical binding data is by using the Scatchard plot, which plots, where K D ϭ 1/K 0 denotes the dissociation constant, ϭ C T,bound /C S 0 , and C T,free can be determined according to Equations 3-5. Fig. 2C shows the corresponding plot. A straight line is obtained with n ϭ 1.99 and K 0 ϭ 8.2 ϫ 10 6 M Ϫ1 , which agrees with the non-linear least squares fit shown in Fig. 2B. All titration experiments performed were also analyzed by Scatchard plots (not shown). It should be noted, however, that the evaluation of the measured h i via a non-linear least squares fit to Equation 2 is more accurate. Next, the GTP-tubulin-stathmin binding reaction was monitored as a function of temperature. Fig. 3 (A and B) illustrates the variation of the reaction enthalpy and the binding constant with temperature. The reaction enthalpy was found to be endothermic at low temperatures and decreased linearly with temperature according to ⌬H GTP-tubulin 0 (kcal mol Ϫ1 GTP-tubulin) ϭ Ϫ0.86 T(°C) ϩ 24.0 (Fig. 3A). Extrapolation to higher temperatures predicts ⌬H GTP-tubulin 0 ϭ 0 kcal mol Ϫ1 at 28°C and exothermic ⌬H GTP-tubulin 0 values above this temperature. Hence, below 28°C the interaction between GTP-tubulin and stathmin is completely entropy-driven, above this temperature both enthalpy and entropy contribute favorably to the binding reaction. The slope of the straight line in Fig. 3A corresponds to the molar heat capacity change of the binding reaction ⌬C p 0 ϭ Ϫ860 cal mol Ϫ1 K Ϫ1 (referring to tubulin). This is a very large negative change in heat capacity indicative of a hydrophobic reaction. Considering the temperature dependence of the binding reaction, K 0 reaches a maximum value of K 0 ϭ 1. .
The free energy can then be calculated according to ⌬G 0 ϭ ϪRT ln K, and the entropy term is T⌬S 0 ϭ ⌬H 0 Ϫ ⌬G 0 . The corresponding evaluations are summarized in Table I. It has been reported that stathmin only modestly enhances the slow basal GTPase activity of tubulin (26, 28, 29, 40 -42). To ensure that GTP hydrolysis does not influence the binding isotherm within the time scale of the experiment, ITC measurements with GDP-tubulin (referred to as the ␣/␤-tubulin heterodimer with GDP occupying the exchangeable E-site on ␤-tubulin (2)) were carried out. As summarized in Table I, the thermodynamic parameters remained almost constant at pH 6.8. The reaction enthalpy varied linearly with ⌬H GTP-tubulin 0 (kcal mol Ϫ1 GDP-tubulin) ϭ Ϫ0.85 T(°C) ϩ 21.2 and extrapolation to higher temperatures predicts ⌬H GTP-tubulin 0 ϭ 0 kcal mol Ϫ1 at 26°C (Fig. 3). Hence, within the accuracy of the measurements, the binding constants are identical for both GTP-and GDP-tubulin. Consistent with previous reports (30,31), changing the pH from 6.8 to 7.5 had little influence on the thermodynamic parameters (Table I). At 10°C and pH 7.5 the equilibrium binding constant was K 0 ϭ 1.4 ϫ 10 6 M Ϫ1 , four times smaller than the one measured at pH 6.8.
In summary, the ITC results revealed that under all conditions investigated stathmin binds two tubulin subunits with equal intrinsic affinities. They further suggest that at physiological temperatures, the hydrophobic effect is the major driving force governing the tubulin-stathmin binding reaction.
In the presence of stathmin at a 0.5:1 molar ratio, ϳ20% of the particles were still 8 -10 nm in diameter, however, ϳ80% of them were distinctly elongated with dimensions of 8 -10 ϫ 16 -20 nm (Fig. 4B). The elongated particles have been previously identified as ternary tubulin-stathmin complexes (13). In the presence of stathmin at a 1.25:1 molar ratio, ϳ70% of the particles were 8 -10 nm in diameter and only ϳ30% revealed the elongated shape of 8 -10 ϫ 16 -20 nm (Fig. 4C). Note that TABLE I Thermodynamic binding parameters derived from the titration of tubulin with stathmin ITC measurements were carried out in 80 mM PIPES, 1 mM EGTA, 1 mM MgCl 2 supplemented with either 1 mM GTP (for tubulin-GTP) or 2 mM GDP (for tubulin-GDP) at the indicated temperature (T) and pH value.  because of its elongated and thin structure (1 nm in diameter (13)) the excess of unbound stathmin does not affect the evaluation. Together, these findings suggest that the 8-to 10-nm particles seen in the presence of equimolar amounts of stathmin correspond predominantly to binary tubulin-stathmin complexes.
CD and NMR Measurements on the Tubulin-Stathmin System-It has been proposed that a shift in pH from 6.8 to 7.5 switches the function of stathmin from tubulin sequestration to specific stimulation of MT catastrophes (8,24,28). To probe for conformational differences of the tubulin-stathmin system at these two pH values, far-UV CD and NMR measurements were carried out. As shown in Fig. 5A, CD spectra recorded at 5°C from stathmin and pH 6.8 or 7.5 revealed two minima centered at 207 and 222 nm characteristic of proteins with ϳ60% helical content for both pH values. The differences in the mean molar ellipticities [⌰] between the two spectra were less than 4% throughout the wavelength range. Similar, fully reversible non-cooperative thermal unfolding transitions were obtained for both pH values (Fig. 5B). The profiles indicate that stathmin unfolds rapidly with increasing temperature. Spectra recorded at different temperatures revealed a constant shift in the wavelength of the first minimum from 207 at 5°C to 200 nm at 70°C, consistent with a shift in the equilibrium from helix to random coil (not shown). Throughout the temperature range from 5 to 70°C, the conformational difference of stathmin between pH 6.8 and 7.5 is marginal. Similar experiments were carried out with tubulin in the absence and presence of stathmin. CD spectra recorded from GTP-tubulin at 5°C and pH 6.8 or 7.5 revealed minima centered at 208 and 222 nm characteristic for the presence of ϳ40% helical structure (Fig.  5C). The differences in the mean molar ellipticities [⌰] between the two spectra were less than 5% throughout the wavelength range. Similar cooperative thermal unfolding profiles with single midpoint of transitions T m centered at 59.0 and 58.4°C for pH 6.8 and pH 7.5, respectively, were obtained (Fig. 5D). The profiles were not reversible upon cooling. As for stathmin, the conformational difference of GTP-tubulin between pH 6.8 and 7.5 is marginal throughout the temperature range from 5 to 50°C. In the presence of a 0.5:1 molar ratio of stathmin to GTP-tubulin, a slight increase of 10% in the helical signal at 15°C compared with free GTP-tubulin was apparent (Fig. 5D). This finding suggests that, at low temperatures, extensive helix formation in stathmin is not induced upon binding to tubulin. The change in helical conformation can be estimated by the , respectively. This corresponds to a 3-7% increase in the helical structure of stathmin upon binding to tubulin within this temperature range if it is assumed that the tubulin subunits do not contribute to ⌬[⌰] h . The thermal unfolding profiles obtained in the presence of a 0.5:1 molar ratio of stathmin to GTP-tubulin at pH 6.8 and 7.5 revealed single transitions centered at T m ϭ 61.2 and 59.6°C, respectively (Fig. 5D). The progressions and corresponding T m values of the profiles were very similar to their counterparts recorded from pure GTP-tubulin, suggesting that the bound stathmin melts in parallel with the tubulin subunits and does not significantly increase their thermal stabilities.
NMR of uniformly 2 H/ 13 C/ 15 N-labeled stathmin and unlabeled GTP-tubulin was used to assess possible pH-induced conformational differences of stathmin when bound to tubulin at the single residue level. At pH 6.8, the two-dimensional 15 N, 1 H correlation spectrum (HSQC or TROSY) of unbound stathmin revealed ϳ50 sharp and ϳ90 broader backbone N-H resonances ( Fig. 6A; see also Ref. 13). The sharp resonances are characteristic of unstructured and flexible residues. The broader resonances with limited chemical shift dispersion indicate that these residues are in rapid exchange between random coil and helical conformations. In the presence of a 1:0.4 molar ratio of GTP-tubulin to stathmin, the 13 C/ 15 N labels allowed the selective monitoring of stathmin residues in the predominantly formed ternary tubulin-stathmin complex. As expected for a large 200-kDa complex, at pH 6.8 most resonances disappeared in the HSQC spectrum indicating that most stathmin residues adopt a stable conformation upon binding to tubulin (Fig. 6B). Eight stathmin backbone N-H resonances remained visible in the HSQC spectrum characteristic of flexible residues. Additional signals are not seen in TROSY spectra, probably due to the high proton density in non-deuterated tubulin. Heteronuclear three-dimensional NMR experiments identified these eight resonances as the C-terminal stathmin sequence Asp 141 -Asp 149 (note that Pro 142 has no amide proton and therefore no HSQC peak). At pH 7.5 (Fig. 6C), five intense and five to ten weaker N-H resonances in addition to Asp 141 -Asp 149 were observed in the HSQC spectrum. Three out of the five intense resonances were identified as residues Glu 138 -Lys 140 . It is reasonable to speculate that the remaining two intense and the five to ten weaker resonances also stem from the C terminus. No N-terminal stathmin residues were observed in the HSQC spectrum indicating that these residues adopted a stable conformation in the tubulin-stathmin complex at pH 7.5. The observed 4-fold decrease in affinity upon changing the pH from 6.8 to 7.5 (Table I) can thus be structurally explained on a per-residue basis and attributed to a weaker interaction of the C terminus of stathmin to tubulin.
In summary, the CD and NMR results demonstrate that stathmin, although extensively helical at lower temperatures, is devoid of any stable tertiary structure in aqueous solutions. However, in complex with two tubulin subunits all except its last eight to fifteen C-terminal residues adopt a stable conformation. This conformational behavior classifies stathmin into the growing family of the so-called "intrinsically disordered proteins" (43). The results further suggest that the conformations, thermal stabilities, and dynamics of the tubulin-stathmin system are only modestly sensitive to small changes in pH.

Mechanism of Tubulin-Stathmin Complex Formation-
The binding equilibrium between tubulin and stathmin has been investigated by surface plasmon resonance (17,19,22,32), by pull-down experiments (26,28,41), and most recently by a non-equilibrium-perturbing sequestration assay (31). All these studies indicated that stathmin interacts with two ␣/␤-tubulin heterodimers in the low micromolar range. In analytical ultracentrifugation (21,30) and gel filtration (18,22,32) studies only ternary (denoted T 2 S) but not binary (denoted TS) complexes were isolated. Consistent with this observation, pulldown assays using agarose bead-coupled antibodies against glutathione S-transferase-and FLAG-tagged stathmin variants suggested that the binding of tubulin to stathmin is highly cooperative and that binary TS complexes are labile intermediates (26,28,41). However, only dissociation constants but no other thermodynamic data have been reported in these studies.
The advantage of the ITC equilibrium method is its high sensitivity and precision. As illustrated by Fig. 2A, the binding of tubulin to stathmin is fast and occurs within the response time of the calorimeter. Analysis of the binding isotherms either by a non-linear least squares fitting method or by Scatchard plots revealed a simple two-site binding mechanism. Surprisingly, the two sites appeared to be independent and identical with respect to binding affinity, which implicates that, at a molar excess of stathmin over tubulin, formation of binary TS complexes should occur. This prediction is supported by the TEM analysis shown in Fig. 4, which revealed that, at a 1.25:1 molar ratio of stathmin relative to tubulin, significant amounts of TS complexes are formed indeed in addition to T 2 S. The current structural view of the asymmetric T 2 S complex ( Fig. 1) suggests that, besides tubulin-stathmin interactions, tubulin-tubulin interactions between the ␤1and ␣2-tubulin monomers may also contribute to the overall stability of the ternary complex. Hence, the two tubulin binding sites on stathmin are probably not fully independent. Nevertheless, the present ITC analysis clearly revealed that both sites are characterized by the same intrinsic equilibrium binding constant K 0 . The structure of T 2 S shows that site 1 (denoted s1) comprises contact points between the N-terminal capping domain of stathmin and regions of ␣1-tubulin (denoted T ␣1 -S N ) and contact points between the first half of the C-terminal helical domain of stathmin and parts of the ␣1␤1-tubulin heterodimer (denoted T ␣1␤1 -S h1 ). Site 2 (denoted s2) comprises contact points between parts of the ␤1and ␣2-tubulin monomers (denoted T ␤1 -T ␣2 ) and contact points between the second half of the helical domain of stathmin and regions of the ␣2␤2-tubulin heterodimer (denoted T ␣2␤2 -S h2 ). The finding that both tubulin binding sites on stathmin are characterized by the same binding constant implies that the binding energetic of s1 ϭ T ␣1 -S N ϩ T ␣1␤1 -S h1 is equal to s2 ϭ T ␤1 -T ␣2 ϩ T ␣2␤2 -S h2 (Fig. 1). If it is assumed that the contribution of T ␣1␤1 -S h1 ϭ T ␣2␤2 -S h2 , it follows that the asymmetry caused by the N-terminal domain of stathmin (contact point T ␣1 -S N ) is compensated by the ␤1and ␣2-tubulin monomer interaction T ␤1 -T ␣2 within the T 2 S complex.
The finding that stathmin possesses two similar binding sites for tubulin under all conditions investigated (Table I) contrasts with earlier studies suggesting that the second tubulin subunit is bound distinctly tighter than the first (26,28,30,31,41). Notably, such a highly cooperative binding mechanism would explain why only the ternary T 2 S and no binary TS complexes were found so far. At present, we can only speculate on the origin of this discrepancy. Earlier binding assays used different physical and chemical solution conditions and required tagging, immobilization, and separation of the different species. These manipulations might have entailed additional interactions, conformational changes, and perturbations in the equilibrium leading to a stabilization of the T 2 S complex.
Consistent with the present analysis, the tubulin-stathmin binding equilibrium can be described as, with K 1 ϭ K 2 ϭ K 0 as outlined above. The overall binding constant is K S3 T 2 S ϭ K 1 ϫ K 2 ϭ K 0 2 , and the corresponding dissociation constant is K T 2 S3 S ϭ 1/K 0 2 . The most recent data on the tubulin-stathmin binding equilibrium are those reported in Ref. 31. Using a non-equilibrium-perturbing sequestration assay these authors found K T 2 S3 S ϭ 0.1 M 2 at 37°C and pH 6.8 leading to K 0 ϭ K S3 T 2 S Ϫ1/2 ϭ 3.2 ϫ 10 6 M Ϫ1 . Extrapolation of the present data (Table I) to 37°C yields K 0 ϭ 1 ϫ 10 7 M Ϫ1 , in reasonable agreement with the earlier result.
As illustrated in Fig. 3, the interaction of stathmin with tubulin is highly temperature-dependent. Around 28°C the reaction enthalpy is zero and decreases from positive values at low temperatures to negative values at high temperatures. The temperature dependence of the binding constant is coupled to ⌬H 0 through van't Hoff's law. Consequently, K 0 reaches a maximum where ⌬H 0 ϭ 0 kcal mol Ϫ1 . The large negative heat capacity change of ⌬C p 0 ϭ Ϫ860 cal mol Ϫ1 K Ϫ1 (referring to tubulin) is typical of a hydrophobic reaction. When two hydrophobic surfaces come into close contact, they release their hydration water resulting in a reduction of the heat capacity of the complex. Empirical studies on proteins have shown that the transfer of an apolar surface area of 1 nm 2 in size from a polar to a non-polar environment makes a contribution to ⌬C p,app 0 of Ϫ45 cal mol Ϫ1 (44). Accordingly, the ⌬C p 0 value for the tubulinstathmin binding reaction could be interpreted as a hydrophobic surface area of about 19 nm 2 , which is brought from a polar to a non-polar environment. Because both stathmin and tubulin must contribute equally to the process, the estimated hydrophobic contact area of the two molecules is ϳ9.5 nm 2 . The total buried surface area between tubulin-RB3-SLD (note that the N-terminal domain of RB3-SLD was not resolved in the 4-Å x-ray structure of the complex) and ␤1-tubulin-␣2-tubulin was estimated to be ϳ32 nm 2 , i.e. ϳ16 nm 2 per binding site (17). Hence, a rough guess is that ϳ50% of the total buried surface area per binding site might be hydrophobic, which is a reasonable value. Another factor that could contribute to ⌬C p 0 is a change in the hydration state of stathmin upon folding. The CD measurements (Fig. 5) indicated that between 10 and 35°C, stathmin, although not assuming a stable tertiary fold by itself, experiences moderate structure formation upon binding to tubulin. Accordingly, this effect is not expected to largely influence ⌬C p 0 at physiological temperatures. However, the present data do not rule out whether additional factors also contribute to the large negative heat capacity change of the tubulinstathmin binding equilibrium.
The current structural view of T 2 S (Fig. 1) together with deletion mapping (13,18,19,26,29,41) and mutational studies (40,42) indicate that the helical domain of stathmin accounts predominantly for the binding free energy. Interestingly, the interaction sites of the two ␣/␤-tubulin heterodimers with the RB3-SLD helix (T ␣1␤1 -S h1 and T ␣2␤2 -S h2 ; Fig. 1) appear conserved in the 4-Å structure of the complex (17). This finding indicates that two homologous sequence sites on the helical domain of stathmin may mediate the interactions with the two ␣/␤-tubulin heterodimers. Indeed, a 35-residue duplication, Glu 48 -Val 82 and Glu 99 -Val 133 , with 40 and 80% identity and similarity, respectively, is found in the sequence spanning residues Lys 41 -Lys 140 of stathmin ( Fig. 7A (17, 45)). Notably, the continuous helix formed by Lys 41 -Lys 140 is amphipathic with most of the bulky hydrophobic residues of Glu 48 -Val 82 and Glu 99 -Val 133 clustering on one side of the helix (Fig. 7B). Due to a lack of resolution, the interacting residues of RB3-SLD in the ternary complex could not be identified. However, considering the large negative ⌬C p 0 value for the tubulin-stathmin equilibrium (see discussion above), it appears reasonable to speculate that binding of the helical domain of stathmin to the tubulin subunits is established by apolar residues present in its hydrophobic seam and on the surface of the two ␣/␤-tubulin heterodimers. The sixteen stathmin residues that separate the two duplicated tubulin binding sites may allow for correct positioning of these hydrophobic sites with respect to the interacting tubulin target residues. The impor-tance of amphipathic helix-mediated interactions between an intrinsically disordered protein and its binding target is a frequently encountered mechanism in other disordered protein systems (43).
Does Stathmin Work as a pH-sensitive Bifunctional Protein?-In vitro studies have proposed that a small change in pH from 6.8 to 7.5 switches the function of stathmin from tubulin sequestration to direct stimulation of MT catastrophes (8,24,28). This change in function implies that at the lower pH value stathmin binds preferentially to tubulin subunits, whereas at the higher pH value specific structural determinants present at MT ends are predominantly targeted. In principle, the structural features of stathmin suggest that the molecule, besides sequestering tubulin subunits, might be able to specifically recognize MT ends (13,17,46). To further clarify this important issue we have systematically addressed the question whether the affinities, stabilities, and conformations of the tubulin-stathmin system assessed at pH 6.8 and 7.5 supports the proposed pH-sensitive dual function of stathmin.
The physicochemical consequences on the stathmin and tubulin polypeptide chains expected in lowering the pH from 7.5 to 6.8 is a change in protonation state of acidic residues. The only residue side chain titrating in a polypeptide near physiological pH is the imidazole group of histidine (pK a value of ϳ6.5-7.0 (47)). The principle effects anticipated by a shift in the protonation state of His are (i) a change in protein conformation and/or stability and (ii) a change in the binding energetics of His side chains engaged at a specific binding site. Both, the long and the short range mechanisms can influence FIG. 7. The C-terminal stathmin helix contains a sequence duplication and is amphipathic. A, primary amino acid sequence of Lys 41 -Lys 140 of stathmin with the 35-residue sequence duplication (40% identity and 80% similarity (17,45)) aligned and highlighted in gray. Identical (perpendicular lines) and similar (plus signs) residues are indicated between the two sequence stretches. Bulky hydrophobic residues (i.e. Ala excluded) are in boldface. B, helical wheel representation of the sequence shown in A. The starting residue, Lys 41 , is underlined. Bulky hydrophobic residues are marked by boxes. The dashed line divides the idealized and continuous helix in two equal parts highlighting the amphipathic character of the sequence stretch. Bulky hydrophobic residues participating in the hydrophobic seam of the helix (left part) and located within the two homologous sequence sites are highlighted in gray.
the affinity of a protein-complex system. Several lines of evidence clearly indicate that the protonation state of His residues only modestly alters the biophysical characteristics of the tubulin-stathmin system. Analytical ultracentrifugation studies (21,30), binding experiments (31), and CD measurements (Fig.  5) conducted at pH 6.8 and 7.5 revealed only small differences in conformation and stability of stathmin, tubulin, and tubulinstathmin complexes at these two pH values. NMR shows that at both pH values most of the stathmin residues, including the N-terminal domain that is proposed to be necessary for catastrophe promotion (24,29), are stably bound to tubulin with a few C-terminal residues remaining disordered and flexible (Fig. 6). Consistent with the structural data, ITC revealed a mere Ϫ0.8 kcal mol Ϫ1 difference in the free binding energies between pH 6.8 and pH 7.5 (Table I). Therefore, together with the observation that stathmin is highly expressed in vivo (note that catastrophe promotion is a substoichiometric process), these findings suggest that within the living cell non-phosphorylated stathmin primarily sequesters tubulin subunits and by doing so destabilizes the MT network regardless of small changes in pH. In this context, it is important to note that there have been no reports so far that directly assess MT-end binding activity or enhanced MT turnover and accompanying steadystate GTPase activity, features that would be essential for stathmin representing an authentic MT catastrophe-promoting factor (30,31).