Modulation of the stathmin-like microtubule destabilizing activity of RB3, a neuron-specific member of the SCG10 family, by its N-terminal domain.

RB3 is a neuron-specific homologue of the SCG10/stathmin family proteins, possessing a unique N-terminal membrane-associated domain and the stathmin-like domain at the C terminus, which promotes microtubule (MT) catastrophe and/or tubulin sequestering. We examined herein the contribution of the N-terminal subdomain of RB3 to the regulation of MT dynamics. To begin with, we determined the effects of full-length (RB3-f) and short truncated (RB3-s) forms of RB3 on the polymerization of MT in vitro. RB3-s had a deletion of amino acids 1-75 from the N terminus, leaving the so-called stathmin-like domain, consisting of residues 76-217. Although both RB3-f and RB3-s exhibited MT-depolymerizing activity, RB3-f was less effective. The binding affinity for tubulin was also lower in RB3-f. Direct observation of the dynamics of individual MTs using dark field microscopy revealed that RB3-s slowed MT elongation velocity, increased catastrophes, and reduced rescues. This effect is almost identical to that by stathmin/oncoprotein 18. On the other hand, the MT elongation rate increased at lower concentrations of RB3-f. In addition, RB3-f, indicated higher rescue frequency than control as well as the catastrophe in a dose-dependent manner. The functionality of RB3-f indicated that full-length RB3 has not only stathmin-like MT destabilizing activity but also MT-associated protein-like MT stabilizing activity. Possibly, the balance of these activities is altered in a concentration-dependent manner in vitro. This interesting regulatory role of the unique N-terminal domain of RB3 in MT dynamics would contribute to the physiological regulation of neuronal morphogenesis.

regulated, depending on various cellular activities, and a variety of regulatory proteins have been identified so far. Stathmin/ oncoprotein 18 family proteins are one of the protein families that make MTs labile (1)(2)(3). It has been demonstrated that one stathmin molecule forms a ternary complex with two tubulin heterodimers, termed the "T2S complex," via hydrophobic interaction (4,5). The formation of the T2S complex can alter conditions leading to the depolymerization of MTs (4, 6 -10). Stathmin has also been found to act on GTP-tubulin at MT ends to stimulate the hydrolysis of GTP and to promote the dissociation of GTP-tubulin from the growing ends of MTs (8,11). Based on these results, two models were proposed to account for the stathmin-dependent destabilization of MTs: 1) sequestration of tubulin dimers to prevent their assembly or 2) stimulation of MT catastrophe at MT tips by removing a "GTP cap" that protects the MT from catastrophe. SCG10 family proteins, including SCG10, SCLIP, and RB3, show a high degree of sequence homology to stathmin in their C terminus, termed the "stathmin-like domain." The various stathmin-like domains including the predicted ␣-helix display 65-75% amino acid identity with stathmin (7,12). They have properties in common with of stathmin, although they differ slightly in activity (7,13). In addition to their stathmin-like domain, family members possess unique N-terminal domains, which contain two Cys residues that serve as palmitoylation sites (14). These sites mediate the association to intracellular membranes and the transport to growth cones (15,16). In contrast to the ubiquitous cytosolic stathmin, these proteins, excluding SCLIP (17), are highly expressed in the nervous system (18) and localized to the Golgi apparatus and the growth cone (14 -16, 19). Stathmin is also abundant in the developing nervous system, but its regional expression in the brain is quite distinct from that of SCG10 (20). The level of SCG10 protein is very low in native PC12 cells but is strongly increased upon the nerve growth factor (NGF)-dependent induction of differentiation (21,22). Additionally, PC12 cells that constitutively expressed SCG10 showed a dramatic increase in the tendency to form elongated neurites upon NGF-induced neuronal differentiation, although no neurite outgrowth was observed in the absence of NGF (23). On the other hand, Di Paolo et al. (21) showed in PC12 cells that a selective blockade of stathmin expression by phosphorothioate antisense oligonucleotides prevented the differentiation-promoting actions of NGF. Stathmin-depleted PC12 cells produce SCG10 protein normally upon NGF treatment but do not differentiate, suggesting that SCG10 cannot compensate for the function of stathmin (23). These results have led to the proposal that the functions of stathmin family proteins are considered to be diverged during neural evolution (1).
The stable ternary complex with ␣␤ tubulin heterodimers is formed via the stathmin-like domain (4,6,24), but N-and C-terminal flanks of the domain are required for cooperative binding to the tubulin heterodimers (8). This suggests that the terminal flanks in addition to the binding motifs in contact with tubulin heterodimers play important roles in the regulation of the tubulin-binding and/or MT dynamics. Since the unique N-terminal domain is one of the most important characteristics of SCG10 family proteins, it is possible that it too affects the interaction with tubulin and/or MT dynamics. Although studies of the SCG10 family have provided important information on their stathmin-like actions, the effects of fulllength neural SCG10 proteins on the regulation of MTs are still unresolved. In this paper, we investigated the role of the Nterminal domain of RB3, which is the longest N-terminal domain among member of the SCG10 family, to examine whether the unique N-terminal domain of SCG10 proteins is involved in the regulation of MT dynamics. We compared the effects of full-length RB3 and an N-terminally truncated RB3 in vitro using MT assays and found that the full-length RB3 plays a unique role in regulating MT dynamics.

EXPERIMENTAL PROCEDURES
DNA Construction, Protein Expression, and Purification-The isolation and manipulation of DNA were performed using standard techniques. Three RB3 gene transcripts have been identified, but since the gene expression of RB3 splicing variants is similar in rat brain and in PC12 cells stimulated with NGF (12), we used the longest form. Bacterial expression plasmids for the rat full-length RB3 (RB3-f) and an N-terminally truncated mutant (RB3-s, comprising residues 76 -217) were constructed by subcloning the cDNAs into the SalI and EcoRI sites of the HAT fusion vector pHAT10 (Clontech) (see Fig. 1). The RB3 derivatives were expressed with a HAT tag at the N terminus and purified from Escherichia coli as follows. The collected cells were resuspended in a lysis buffer (8 M urea, 50 mM sodium phosphate (pH 7.0), and 0.3 M NaCl) and incubated for 15 min at room temperature. Then the samples were centrifuged at 30,000 ϫ g for 30 min at 10°C. After adding TALON metal affinity resins (Clontech) were added to the supernatants, and samples were incubated for 1 h at room temperature. The HAT-RB3-bound resins were washed once with a wash buffer (50 mM sodium phosphate, pH 7.0, and 0.3 M NaCl) containing different concentrations of urea (8, 4, 2, 1, 0.5, and 0 M, in that order) and once with 50 mM sodium phosphate, pH 7.0, containing 5 mM imidazole. Finally, protein was eluted from the resins in 50 mM sodium phosphate, pH 7.0, containing 100 mM imidazole. The buffer was changed to PM buffer (0.1 M PIPES, pH 6.9, 1 mM EGTA, and 0.5 mM MgSO 4 ) by centrifugal filtration. Tubulin was isolated from porcine brain by two cycles of polymerization and depolymerizatin and was purified by DEAE-Sepharose column chromatography using fast protein liquid chromatography (Amersham Biosciences) as described previously (25). Protein concentrations were determined by SDS-PAGE with Coomassie Brilliant Blue staining using bovine serum albumin as a standard. Purified tubulin and RB3 recombinant proteins were stored at Ϫ80°C prior to use.
Cosedimentation Assay of MTs-The MT destabilizing activity and binding with RB3 derivatives were examined in a co-sedimentation experiment as follows. Tubulin (10 M) was assembled in PM buffer containing 1 mM GTP in a final volume of 20 l in the presence of various concentrations of RB3-f or RB3-s (0 -10 M) for 15 min at 37°C. The polymerized MTs were sedimented by ultracentrifugation at 70,000 rpm for 60 min at 36°C. The resulting supernatants and pellets were examined with 4 -20% gradient SDS-PAGE. After Coomassie Brilliant Blue staining, the band signals were analyzed using the program NIH Image. The data were fit by linear or nonlinear regression using the standard binding equation for a macromolecule to obtain the apparent dissociation equilibrium constant. Curve fitting was performed using a Kaleida Graph (Synergy Software).
Quantitative Analysis of the Binding between Tubulin and RB3 Derivatives-The binding affinity between RB3 derivatives and tubulin was determined by Scatchard analysis. RB3 derivatives (1 M) and purified tubulin (0.5-10 M) were mixed in a final volume of 35 l in a reaction buffer (0.1 M PIPES, pH 6.9, 0.5 mM MgSO 4 , 1 mM GTP, 5% Tween, and 1 mg/ml bovine serum albumin) and were incubated for 15 min on ice. RB3-tubulin complexes were captured with 5 l of TALON metal affinity resin for 30 min at 4°C. In the present study, we used a low concentration of tubulin heterodimers and an incubation temperature of 4°C, because these conditions favor the formation of tubulin heterodimers rather than polymers (26). To confirm the amount of tubulin bound to RB3, the bound proteins were separated by 4 -20% gradient SDS-PAGE, subjected to silver staining, and analyzed using NIH Image. The data were normalized by subtraction of the nonspecific binding signals.
Optical Microscopic Observation of MT Dynamics-To determine the exact effects of RB3s on MT dynamics, individual MTs were monitored using a dark field microscope with several modifications (27)(28)(29). First, 24 M of tubulin was added to allow the polymerization of MTs in PM buffer containing 1 mM GTP at 37°C for 5 min. An equal volume of prewarmed RB3 recombinant protein solution (0 -9 M) was added to the MT solution, the mixture was transferred to a glass slide, and the slide was immediately sealed. The final tubulin concentration was 12 M. The optical microscopy was performed at 24°C. During observations, bovine serum albumin (1.8 mg/ml) was added to prevent the nonspecific attachment of proteins to the glass surface. The dynamic behavior of the MTs was observed using a dark field microscope (ECLIPSE E600; Nikon) with a plan fluor ϫ 100 objective lens and an oil immersion dark field condenser equipped with a 100-watt high pressure mercury lamp as the illumination source. Images were observed with a CCD camera (IR-1000; DAGE-MTI) and were recorded onto a one-half inch SVHS videotape with a Panasonic model AG 3750 videocassette recorder for 40 min. Changes in MT length were measured from the video images incorporated into a Power Mac G4 with the aid of an image capture board (LG3; Scion) in real time using NIH Image. The change in length at each end of a given MT was followed by determining the distance between each end and an arbitrary fixed point on the MT as a function of time. The two ends of MTs indicated the different polymerization rates; the faster growing end is referred to as the plus end and the slower growing end as the minus end (30). Whereas the minus ends of MTs are relatively stable (31), the plus ends undergo variable phases of assembly and disassembly, also referred to as dynamic instability (32). The dynamic instability of minus ends is probably not physiologically relevant, because minus ends in cells are either capped by other proteins (e.g. at centrosomes) or depolymerized when free in the cytoplasm (33). Therefore, we examined only the MT plus ends. Data were shown as means Ϯ S.D., and an unpaired t test was used to make comparisons between the control value and each means with the Macintosh software Statview (SAS Institute Inc.).

FIG. 1. Schematic representation of the RB3-f and RB3-s used in this study.
At the top, RB3-f, the longest isoform of RB3, has a stretch of hydrophobic residues at the N terminus (white box) that is responsible for membrane targeting attained by palmitoylation at two cysteines. RB3-s has a deletion of the unique N-terminal portion up to residue 75. Both recombinant proteins contained a HAT epitope tag at the N terminus (white circle). At the bottom, native stathmin is depicted with a nonhelical N-terminal region (40N region), which appears to have a low degree of secondary structure, and an extended ␣-helical region containing two low homology repeats (C-terminal domain) (4, 6). The longitudinally aligned two ␣␤-tubulin heterodimers are bound to the C-terminal domain, and the cap formed by the 40N region ensures the specificity and integrity of the complex by preventing further longitudinal stathmin-tubulin complex aggregation (23).

MT Destabilizing Activity and Tubulin Binding Affinity Were
Weakened in Full-length RB3 as Compared with RB3 Stathmin-like Domain-The activity of the recombinantly produced RB3 derivatives in MT assembly and the direct interaction of RB3 with MTs were tested using an MT sedimentation assay. Both recombinant proteins effectively inhibited tubulin assembly in a dose-dependent manner ( Fig. 2A). This indicated that both proteins have MT destabilization activity. As shown in Fig. 2B, the linear regression analysis of the sedimented MTs and the concentration of RB3 used indicated that RB3-s reduces the concentration of MTs with a slope of Ϫ1.95, suggesting that the depolymerization was mainly due to the formation of the T2S complex and its sequestering activity. On the other hand, the slope of RB3-f, Ϫ1.37, is smaller than that of RB3-s, and complete inhibition of 10 M tubulin polymerization required a higher concentration of RB3-f (about 9 M) as compared with RB3-s (about 5 M) (Fig. 2, A and B). These results suggest that the MT depolymerization activity is suppressed in RB3-f.
Most of the tubulin and RB3 proteins were recovered in the supernatants when the polymerization of MTs was completely inhibited by a sufficient amount of RB3 protein. Neither RB3-f nor RB3-s was sedimented in the absence of tubulin ( Fig. 2A), indicating that the "T2S" complex with RB3 or RB3 itself did not form large aggregates in the present conditions. However, both RB3-f and RB3-s were found in pellets at low concentrations when MTs were present ( Fig. 2A). This suggests that the sedimented RB3 is bound to MTs. As shown in Fig. 2C, the molar ratios of both RB3 proteins to MT increased as the MT concentration decreased (i.e. as the amount of RB3 used increased) and was higher for RB3-f at all concentrations. Although all unbound RB3 is not practically "free" due to the formation of T2S complex, we further plotted the molar ratio against unbound RB3 concentrations, which was recovered in supernatant, as in Fig. 2D. This plot is not a real Langmuir isothermal plot, but we found apparent saturation of binding for each RB3 construct. From these curves, we obtained apparent equilibrium dissociation constants to MTs for RB3-f (0.43 M) and for RB3-s (0.81 M). These values suggest that the interaction between RB3 and MTs is enhanced by the existence of the unique N-terminal domain.
It has been shown that the tubulin heterodimer-stathmin affinity in vitro is dependent on pH (10,11,34), and there is relatively tight interaction between them at pH Ͻ7.0 (11). Hence, the tubulin binding activity of the stathmin-like domain may also play an important role in the MT regulatory mechanism of RB3 in the present buffer at pH 6.9. Thus, a Scatchard analysis was performed to determine the affinity between tubulin heterodimers and RB3 derivatives. As shown in Fig.  3, the dissociation constants (K d values) for RB3-f and RB3-s were 0.57 Ϯ 0.10 and 0.33 Ϯ 0.10 M, respectively. In previous studies (8,35), the Scatchard conversion of the binding between tubulin dimers and stathmin was shown to be a complex process with a nonlinear data point distribution typical of two-site positive cooperativity in binding. However, the data point distributed along a linear regression in the present study. This would not be due to the difference in the mode of binding between stathmin and our RB3 constructs. The difference is rather due to a weaker sensitivity for the detection of bound proteins in our experiments than in previous ones using a radioisotope as the protein tracer (8). Nevertheless, we emphasize that the K d values were found to be suppressed in RB3-f as compared with RB3-s in this study.
The present results suggested that the affinity of RB3-f for tubulin heterodimers is somehow suppressed, but that for MTs is enhanced as compared with RB3-s. The effects of RB3-f could be attributed to the unique N-terminal domain.
Full-length RB3 Stabilizes MTs by Enhancing the Rescue from MT Depolymerization but Simultaneously Promotes MT Catastrophe-We used dark field microscopy to examine whether the N-terminal domain of RB3 alters the characteris-  Table I. Four parameters of MT dynamics were assessed: parameter 1, rate of elongation (Fig. 5A); parameter 2, rate of shortening (Fig.  5B); parameter 3, elongation length required for a catastrophe (the transition from elongation to shortening) (Fig. 6A); and parameter 4, shortening length required for a rescue (the transition from shortening to elongation) (Fig. 6B). Elongation or shortening length required for a transition event (catastrophe or rescue) was calculated by dividing the summed length of elongation or shortening phases by the total number of transition events, respectively (28). These parameters were calculated to express how the level of ease (or difficulty) that ended each phase in the given condition, and the values might be rather comparable with the dynamicity that reflects how dynamic each microtubule is. Parameter 5 (mean length of elongation phase) (Fig. 6C) and parameter 6 (mean length of shortening phase) (Fig. 6D) were also calculated to know the variance of the data. Mean length is the simple average of the lengths of each phase lying between the starting transition event and ending transition event. Since each phase does not always end with the event and since the number of the transition event is smaller than the number of corresponding phases, it is impossible to obtain the S.D. value in obtaining the phase (elongation or shortening) length required for an event (catastrophe or rescue). In addition, the number of phases shown in Fig. 6, C and D, is usually smaller than that shown in Fig. 6, A and B, respectively. Especially, the number of the phase used in the calculation of the mean length of the shortening phase in the presence of 1.5 M RB3-s (Fig. 6D) is obviously smaller than that of the shortening length required for a rescue (Fig. 6B). Parameter 7 (catastrophe or rescue frequency) was obtained by dividing the total number of catastrophe or rescue events by the total time spent in elongation or shortening phases, respectively (25), and was shown in Table I. In the present study, a clear difference in the dynamics profiles of MTs was detected between RB3-f and RB3-s, although the MT elongation rate seemed to be low throughout the experiment.
Typical examples of micrographs and the dynamics profiles of single MTs are shown in Fig. 4. Although the number of MTs slightly decreased within a 10-min observation, most of the MTs observed initially remained at 15 min in the absence of RB3 constructs (Fig. 4A, upper panel). However, only a small number of MTs were found 15 min after the addition of 1.5 M RB3-s (Fig. 4A, middle panel). In contrast, most MTs remained 15 min after the addition of 1.5 M RB3-f (Fig. 4A, lower panel), although most vanished in the presence of 4.5 M RB3-f (data not shown). These observations were consistent with the re-sults of the co-sedimentation assay shown in Fig. 2, again indicating that both constructs exhibit MT destabilizing activity, but the activity of RB3-f is weaker than that of RB3-s. Since these results provided the upper limit for the concentration of each RB3 protein to observe MT dynamics by dark field microscopy, we observed MT dynamics over the range 0.75-1.  (Fig. 6, A and B). In addition, the elongation rate was considerably decreased in the case of 1.5 M RB3-s (0.32 Ϯ 0.14 m/min) in contrast to the control (0.45 Ϯ 0.11 m/min) (Fig.  5A). These results indicated that catastrophe is occurs more easily and rescue occurs with more difficulty in the RB3-s than in its absence, suggesting that the effects of RB3-s on the MTs resemble those of stathmin (11). However, these obvious differences cannot be found by the comparison of a simple average of the phase lengths, although slight tendency is somehow detected (Fig. 6, C and D). Because shortening phases that did not end with rescue is omitted in the calculation for the mean length of shortening phase for instance, the clear tendency is not detected. In the case of elongation phase, each microtubule elongated only at the beginning of the observation, and such a phase is ignored by the definition of the calculation for the mean length of elongation phase. Since the concentration dependence of mean values (Fig. 6, C and D) resembles that of the phase length required for either event (Fig. 6, A and B), the former parameters somehow reflect the overall characteristic property of microtubule dynamics at given concentrations. However, we consider that they are inadequate to consider the present extreme cases (e.g. where the majority of microtubules  In contrast to RB3-s, RB3-f at lower concentrations significantly increased the MT elongation rate: 0.53 Ϯ 0.13 m/min at 0.75 M and 0.54 Ϯ 0.16 m/min at 1.5 M (Fig. 5A). This effect of RB3-f at lower concentrations on MT dynamics is quite different from that found in RB3-s, and is an unknown function as stathmin family protein. At higher concentrations of RB3-f, the elongation length required for a catastrophe also decreased to 2.1 m at 3 M and 1.8 m at 4.5 M RB3-f (Fig. 6A). These values were comparable with those found in the presence of 1.5 M RB3-s. Shortening length was also decreased in a dosedependent manner of RB3-f (Fig. 6B). In these cases, significant changes in the mean length of depolymerization were also obtained (Fig. 6, C and D). This is due to that the quite frequent repeats between elongation and shortening are enhanced at the increased concentrations of RB3-f as shown in Fig. 4D. Additionally, the rate of shortening was significantly suppressed in the presence of RB3-f at 3 M. A similar rate was indicated at 4.5 M RB3-f, although it was not significantly different from the control (Fig. 5B). RB3-f seems to act partly as an MTstabilizing factor, like MAP4, which shows catastrophe-and rescue-promoting activity during in vitro assembly of microtubules (28). However, catastrophe promotion activity of stathmin finally overcame at higher concentrations of RB3-f, which completely depolymerized MTs, as shown in Fig. 2.
Since a higher concentration of RB3-f was necessary to achieve similar effects of RB3-s on the catastrophe promotion, the N-terminal domain of RB3 is likely to be a negative regulator of the MT destabilizing activity of the "stathmin-like domain." At higher concentrations, the elongation velocity reverted to the level of the control at these concentrations, contrary to the prediction that the elongation velocity should be unchanged or rather increased at higher concentrations of RB3-f (Fig. 5A). Since the affinity for tubulin hetrodimers is reduced in RB3-f (Fig. 3), one can assume that the tubulin sequestering activity is also weaker in RB3-f. Weakened sequestering activity is also assumed from the abolishment of the decrease in MT elongation velocity, because the elongation velocity has been reported to depend on the tubulin sequestering activity, which reduces the concentration of tubulin available for polymerization (8,11). Due to the decrease in sequestering activity, a higher concentration of RB3-f is required to induce an inhibitory effect on the elongation velocity similar to that of RB3-s. Only at higher concentrations of RB3-f, the enhanced polymerization would be counterbalanced by the tubulin sequestering activity, and consequently, the elongation velocity reverted to the control level. Since the shortening length required for a rescue was relatively low throughout the concentration range, the MT stabilizing activity was not lost by increasing the RB3-f concentration. In fact, the shortening rate was significantly suppressed at a higher concentration of RB3-f (Fig. 5B). These MT stabilizing effects could be attained by the direct interaction of RB3-f with MTs. However, MTs were promptly depolymerized at higher concentrations of RB3-f. Perhaps the stronger catastrophepromoting activity overcame the stabilizing effect at these concentrations, suggesting that the bifunctional regulation of MT dynamics by RB3-f could be altered in a concentrationdependent manner in vitro.

DISCUSSION
In the present study, we investigated the role of the Nterminal domain of RB3 in the regulation of MT dynamics in vitro and found an extra function. As described previously (7,13,36), RB3-s, the majority of which comprises the stathminlike domain, behaved as a MT-destabilizing protein just like stathmin. Indeed, the polymerization of MT was completely inhibited in the presence of approximately half the amount of RB3-s as tubulin. 1.5 M RB3-s added to 12 M tubulin significantly decreased the MT-elongation rate and markedly decreased the elongation length required for a catastrophe. In contrast, the MT destabilizing activity was partly inhibited in RB3-f. It seems that this inhibition is mainly due to the attenuated binding affinity of RB3 for tubulin heterodimers shown in the Scatchard conversion of binding data. This suggests that the unique N-terminal domain of RB3 affects negatively the binding of tubulin heterodimers. The intramolecular interaction between the N-terminal domain and stathmin-like domain may inhibit the formation of the T2S complex.
The binding affinity of RB3 for MTs is enhanced by the existence of the N-terminal domain. Although MTs tended to depolymerize in the presence of either RB3-s or RB3-f, the two constructs were found to bind the remaining MTs as in Fig. 2A. This indicates that both RB3-f and RB3-s are able to bind to the MT lattice. Moreover, the molar ratio of RB3-f to MTs was FIG. 7. Presumed mechanism of action of stathmin and fulllength RB3 on tubulin heterodimers and MTs. On the bases of the results presented here and the data available from the literature (see "Results" and "Discussion"), we derived a possible mechanism of how the actions of RB3-s and RB3-f differ on tubulin and MTs. Typically, MTs are assembled from the tubulin heterodimers (1). In the presence of stathmin, assembly of MTs is inhibited by the sequestering activity of stathmin, and the 40N region acts as a cap of tubulin subunits at the site normally engaged in the formation of longitudinal contacts within protofilaments (2). Stathmin also acts on GTP-tubulin at MT ends to stimulate GTP hydrolysis and to promote GTP-tubulin dissociation off the growing MT ends (3). Previously, it has been shown that RB3-s has common properties of stathmin, although they differ slightly in these activities (7,13). In the full-length form of RB3, the comprehensive MT destabilizing activity, resulting from both the sequestering and catastrophe promoting activity, is suppressed as compared with RB3-s (4). Both RB3-s and RB3-f can bind to MTs, but the binding affinity is higher in RB3-f. Four possible models for RB3 binding on MT are shown (5). RB3-s may bind on MT lattice as it forms a T2S complex (5, a); the N-terminal domain and stathmin-like domain may bind along a single protofilament (5, b); the N-terminal domain may bind to the tubulins in the way of bridging between adjacent protofilaments (5, c); and the N-terminal domain can bind to MTs independently of the stathmin-like domain (5, d). However, we have no information to judge which model is the most likely one. Anyway, the N-terminal domain of RB3 is required to reduce the stathmin-like MT destabilizing activity and to attain the MAP-like MT stabilizing activity. higher than that of bound RB3-s at all concentrations used. Several lines of evidence suggest that the depolymerization of MTs by stathmin is due to the effect on MT ends (11), although no direct evidence has been provided for the binding region of stathmin on MT ends. Furthermore, a structural analysis revealed that the C-terminal ␣-helical domain of stathmin (or stathmin-like domain) is required to bind with two tubulin heterodimers (4, 6) (see Fig. 1). However, the present results suggest that the stathmin-like domain could be bound to polymerized MTs. It has been shown that the 40 N region of stathmin promotes catastrophe at MT ends (11) and acts as a "cap" of tubulin subunits at the site normally engaged in the formation of longitudinal contacts within protofilaments to ensure the specificity and integrity of the complex (24). Stathmin deleted of this cap region stabilized a kinked protofilament-like tubulin tetramer, and this complex was sedimented by ultracentrifugation (24). The N-terminal domain of RB3 may have modified the function of the cap region in the stathmin-like domain, resulting in the enhancement of the binding affinity for MTs.
We also found that the MT elongation rate was significantly increased in the presence of a 1.5 M or lower concentration of RB3-f. In addition, MTs easily underwent rescue without a significant change in the mean length of elongation phase, which reflects the catastrophe promotion activity, in the presence of 1.5 M RB3-f. It is impossible to predict this effect from the stathmin-like activity of RB3, although the level of activity was decreased in RB3-f. These functional roles of RB3-f seem to resemble those of MT-stabilizing factors, like MT-associated proteins (MAPs) (28,33,37), which exert their effects primarily by binding to the MT lattice and cross-linking adjacent tubulin subunits (27,38). In the present study, RB3 bound to MTs with high affinity and displayed MT stabilizing activity like classical heat-stable MAPs. In addition to the well characterized stathmin-like domain of RB3, which must be the binding site for tubulin heterodimers, one may assume that the unique Nterminal domain of RB3 also interacted with MTs independently of the stathmin-like domain. If so, it is easy to explain the enhanced binding affinity of RB3-f for MTs, although we have no evidence of direct interaction between the N-terminal domain and MTs because it was too hard to obtain recombinant protein with only the N-terminal domain from E. coli. Thus, RB3-f might, at least in part, regulate MTs by a MAP-like function.
MT in the presence of 3-4.5 M RB3-f underwent catastrophe more easily, as easy as in the presence of 1.5 M RB3-s. Furthermore, the elongation rate was decreased as compared with that at 0.75-1.5 M RB3-f, although it was similar to that of the control. This suggests that RB3-f promotes catastrophe. However, the length of shortening phases was also significantly decreased, suggesting that the rescue occurred more easily. Moreover, the shortening rate was suppressed at higher concentration of RB3-f. The MT stabilizing and destabilizing activities of RB3-f may be promoted independently, because the predicted mechanisms for each are quite different, as mentioned above. However, the two events happen simultaneously, and the priority seems to be altered in a concentration-dependent manner. Our presumable model for the action of RB3-s and RB3-f on tubulin and MTs is shown in Fig. 7. The N-terminal domain of RB3 may act as an enhancer of MAP-like activity and suppressor of stathmin-like activity.
A recent study (39) showed that growth cones of hippocampal neurons transfected with palmitoylation-deficient mutated SCG10 are characterized by longer filopodia, whereas overexpression of wild type SCG10 results in growth cone collapse. However, PC12 cells constitutively expressing wild type SCG10 exhibited a dramatic increase in the elongation of neurites upon NGF-induced neuronal differentiation (23). The major difference between these studies is the protein expression system used. The different systems could explain the different protein induction levels in cells. Hence, the SCG10 family may play an important part in the regulation of growth cone filopodia and neurite extension in a bifunctional role as well as specific intracellular localization by the N-terminal domain. Since phosphorylation is an important regulatory mechanism for the activity of stathmin family proteins in vivo (6,9,40), further study is required to elucidate the effect of phosphorylation of the full-length SCG10 family proteins on this bifunctional regulation of MTs. This interesting regulatory role of the unique N-terminal domain of the SCG10 family in MT dynamics would contribute to the physiological regulation of neuronal morphogenesis.