Incorporation of tubulin subunits into dimers requires gtp hydrolysis

Protein assembly often involves interactions of unfolded proteins with preexisting proteins that function as molecular chaperones. Different groups of molecular chaperones have been involved in assisting protein folding, including the heat-shock protein 70 group, and the chaperonins (Langer et al., 1992; Ellis and van der Vies, 1991). By means of non-denaturing electrophoresis and anion-exchange chromatography it has been shown that newly synthesized β-tubulin interacts with cellular factors forming at least two different complexes (the high molecular mass form C₉₀₀ and the low molecular mass form C₃₀₀) before it can incorporates into dimers (Yaffe et al., 1988, 1992; Zabala and Cowan, 1992). A similar behaviour has been observed for α-tubulin subunits, although the kinetics of appearance and proportion of the different molecular forms are different from those of β-tubulin (Zabala and Cowan, 1992).

A toroid multisubunit particle of about 800-900 kDa is present in the cytosol of different animal cells. This particle contains TCP-1, a protein highly related in sequence to TF55, which has ATPase activity and is involved in protein folding (Trent et al., 1991) with four to six other unidentified proteins and two Hsp70 heat-shock proteins (Lewis et al., 1992). These oligomeric particles have been implicated in folding of completely denatured actin and α- and β-tubulin (Yaffe et al., 1992; Gao et al., 1992) constituting, therefore, a chaperone complex (C₉₀₀). The chaper-one complex binds actin or tubulin polypeptides,giving rise to a binary complex. These complexes formed between purified particles and denatured actin polypeptides are able to release functional actin monomers in the presence of Mg-ATP (Gao et al., 1992). The complexes formed between the C₉₀₀ and tubulins are also able to release functional tubulin in the presence of Mg-ATP (Yaffe et al., 1992; Fry-dman et al., 1992; Gao et al., 1993).

In vitro translation in reticulocyte extracts of tubulin polypeptides gives rise to the formation of complexes of about 180-300 kDa termed C₃₀₀. The molecular composi-tion of these complexes that tubulin forms with reticulo-cyte proteins needs to be further investigated, although they have been functionally implicated in tubulin dimer-ization (Yaffe et al., 1988; Zabala and Cowan, 1992). The interaction of β-tubulin with the 900 kDa complex appears to be a fast initial step in the tubulin folding and assem-bly process: prolonged incubation at 37°C of in vitro trans-lation reactions reduced the amount of C₉₀₀ complexes compared to the amount of C₃₀₀ complexes (Yaffe et al., 1992). Kinetic analysis of tubulin dimer formation has also shown that, in a slower fashion, the proportion of C₃₀₀ complexes, monomers and dimers changed: the amount of dimers formed was related to the disappearance of C₃₀₀ complexes and monomers (Zabala and Cowan, 1992). We have investigated the process by which newly synthesized β-tubulin monomers and C₃₀₀ complexes give rise to the formation of tubulin dimers. We present evidence that het-erodimers can form only after GTP is hydrolyzed and that this process is dependent on the presence of Mg²⁺. More-over, the incorporation of tubulin subunits from C₃₀₀and/or monomers into dimers is independent of ATP hydrolysis. These data are discussed in the context of a two-step mechanism implicated in the tubulin folding and assembly process.

Reticulocyte extracts were obtained from Promega, [³⁵S]methio-nine (>1,000 Ci/mmol) from Amersham and GTPγS and ATPγS from Boehringer Mannheim.

Full-length cDNA encoding wild-type β₃-tubulin (Wang et al., 1986) cloned into pGEM-2 vector was used as template for in vitro transcription (Melton et al., 1984). Transcribed mRNA was translated in a rabbit reticulocyte cell-free system (Pelham and Jackson, 1976) in the presence of [³⁵S]methionine (Zabala and Cowan, 1992). In vitro synthesized products were centrifuged at 100,000 g for 30 min at 4°C in a Beckman TLX centrifuge and extensively dialyzed against 50 mM MES, pH 6.7.

In vitro translation reactions were carried out for 1 hour at 30°C and centrifuged at 100,000 g for 30 minutes at 4°C. C₃₀₀ com-plexes were partially purified by a modification of the method pre-viously described (Zabala and Cowan, 1992). Aliquots of in vitro translation reactions were loaded onto a (Pharmacia) Superose 6 column equilibrated with buffer (100 mM MES, pH 6.7) and run at a flow rate of 0.25 ml min⁻¹. Fractions of 200 μl were collected and C ₃₀₀ complexes eluted in ml 13.4.

Purified brain tubulin was prepared as described (Weingarten et al., 1974; Shelanski et al., 1973), by consecutive passages through phosphocellulose, cation-exchange FPLC (Zabala and Cowan, 1992), and G-25 equilibrated in 150 mM MES, pH 6.7. Reticu-locyte extracts were centrifuged at 100,000 g for 30 minutes at 4°C and depleted of low molecular mass components by passage through G-25 equilibrated in 150 mM MES, pH 6.7.

Aliquots from reactions carried out as described in the legend of the figures were diluted one-fith with 25% sucrose (final concen-tration of 5%) in 50 mM MES, pH 6.7 and immediately loaded onto a 7% non-denaturing polyacrylamide gel, containing ME buffer (100 mM MES, pH 6.7, 1 mM EGTA) with 0.1 mM GTP and 1 mM MgCl₂, 0.1 mM GTPγS and 1 mM MgCl₂, or 0.1 mM GTP. Following electrophoresis, gels were fluorographed, dried and exposed to film. The radioactive bands were excised and quan-tified by scintillation counting.

Protein samples were electrophoresed through 4.5% and 7% non-denaturing polyacrylamide gels and electrophoretically transferred to nitrocellulose membranes in transfer buffer without methanol. The membranes were stained with Ponceau S (Sigma), and after several washes blocked with 5% dried milk in Tris-buffered saline with 0.05% Tween-20 (Sigma).

Monoclonal anti-β-tubulin antibodies were purchased from Amersham and the rat mAb 91A anti-mouse TCP-1 was gener-ously provided by K. Willison. Primary antibodies were incubated with the blot for 1 hour. These primary antibodies were detected with horseradish peroxidase-conjugated donkey anti-mouse or anti-rat secondary antibody, followed by chemiluminiscent pro-cessing using Amersham ECL reagents. Exposure to Amersham ECL films was carried out for one second to several minutes.

The C₃₀₀ multimolecular complexes that α- and β-tubulin form with unknown factors present in reticulocyte extracts might represent intermediates in the biosynthesis of tubulin monomers; the extent of monomer incorporation into dimers depends on the amount of exogenous brain tubulin added (Zabala and Cowan, 1992). To investigate the requirements for tubulin dimer formation, the fate of the newly synthesized β-tubulin was followed after addition of exogenous purified brain tubulin under different conditions (Fig. 1). The proportion of C₃₀₀ complexes, monomers (M) and dimers (D) was analyzed by non-denaturing gel elec-trophoresis under conditions where the amount of C₉₀₀ complexes was reduced compared with the amount of C₃₀₀ complexes (i.e. prolonged incubation at 37°C of in vitro translation reactions). The proportion of newly synthesized β-tubulin in these different molecular forms (C₃₀₀, D and M) was calculated by measuring the associated radioactiv-ity after the bands were excised and quantified by scintil-lation counting, since the densitometric analysis did not give a linear response. Increased incorporation of tubulin subunits into dimers is observed in the presence of GTP and Mg²⁺ (compare Fig. 1A and B), suggesting that GTP or Mg²⁺ or both are implicated in the formation of the het-erodimer.

To analyze the role of Mg²⁺, the experiments were repeated in the presence of 0.3 mg ml⁻¹ of brain tubulin, GTP, and increasing concentrations of Mg²⁺. The incorpo-ration of newly synthesized β-tubulin within the het-erodimer was found to be a function of the concentration of Mg²⁺ (Fig. 2). Magnesium is probably required for the binding of the guanine nucleotide to tubulin (Correia et al., 1987).

To establish the role of GTP in the formation of the het-erodimer, newly synthesized β-tubulin was incubated with 0.3 mg ml⁻¹ of brain tubulin, in the presence of increasing concentrations of GTP (Fig. 3A). In this case, no differ-ences in the relative amounts of monomers, dimers, and C₃₀₀ multimolecular complexes were detected: in the pres-ence of 0.3 mg ml⁻¹ tubulin, interconversion between these forms does not change (see Fig. 1). The high amount of GTP present in reticulocyte lysates and the difficulty of dia-lyzing them efficiently might explain the apparent ineffi-cacy of GTP in this reaction. To address this point we repeated the experiment in the presence of GTPγS, a non-hydrolyzable GTP analog, which can substitute for GTP during microtubule assembly (Kirsh and Yarbrough, 1981). When GTPγS was used instead of GTP, incorporation of newly synthesized tubulin into dimers was prevented in a concentration-dependent manner (Fig. 3B). To confirm that the effect was due to GTPγS itself (and not to a possible contamination with ATPγS), the experiments were repeated in the presence of both ATP and GTPγS. Similar results to those shown in Fig. 3B were obtained. Thus, the incorpo-ration of newly synthesized β-tubulin within the het-erodimer is dependent on GTP hydrolysis.

The overall process of tubulin dimer formation requires among others, ATP and magnesium ions. The proportion of C₉₀₀ complexes increases when ATPγS or EDTA were added to in vitro translation reactions, indicating the impor-tance of ATP hydrolysis and magnesium ions in the reac-tion involving the C₉₀₀ complexes. To investigate the sig-nificance of GTP and ATP hydrolysis in the reaction step concerning the C₃₀₀ complexes, we carried out the follow-ing experiment (Fig. 4): aliquots from partially purified C₃₀₀ complexes were incubated at 37°C for 30 minutes in the presence of Mg²⁺ for control reactions (track 1), or with reticulocyte extracts (Materials and Methods) and purified native brain tubulin and either added with GTP (track 2), ATP and GTPγS (track 3), and GTP plus ATPγS (track 4). The results confirm that the release of tubulin monomers from C₃₀₀ complexes and incorporation into dimers requires the presence of unknown factors in the reticulocyte extracts (Release Factors) and GTP hydrolysis. This process is inde-pendent of ATP hydrolysis.

Native electrophoresis through 7% gels of in vitro-synthe-sized β-tubulin isotypes resulted, in addition to substantial quantities of radioactive material that remained at or close to the origin, in a number of bands, which have been assigned as monomers (M), dimers (D), and C₃₀₀ multi-molecular complexes (Zabala and Cowan, 1992; Fig. 5A). These different molecular forms apparent upon native gel electrophoresis of in vitro β-tubulin translation reactions were examined and compared to the endogenous forms present in reticulocyte lysates by using specific antibodies. Immunodetection using monoclonal anti-β-tubulin antibod-ies showed that reticulocyte extracts contained all these dif-ferent molecular forms of β-tubulin. The pattern obtained using anti-β-tubulin antibodies was indistinguishable from the pattern obtained after fluorography of the native gel con-taining ³⁵S-labelled in vitro translation products, indicating the close relationship existing between the in vitro synthe-sized products and the in vivo endogenous forms of β-tubu-lin (Fig. 5A, compare tracks 1 and 3).

When in vitro translation reactions were subjected to 4.5% polyacrylamide gel electrophoresis under the same conditions, most of the material that remains at or close to the origin under analysis on 7% non-denaturing gels, gen-erated two slowly migrating bands. These bands contained both β-tubulin and TCP-1 peptides as shown by immun-odetection using specific monoclonal antibodies suggesting that both represent multimolecular complexes identified as C₉₀₀ binary complexes (Fig. 5B). Anti-TCP1 monoclonal antibodies react with the C₉₀₀ but not with the C₃₀₀ com-plexes (Fig. 5A and B) suggesting that the C₃₀₀ complexes are devoid of TCP-1 peptides.

The different molecular forms present when α-or β-tubu-lins are translated in vitro (complexes among tubulin and reticulocyte proteins (C₉₀₀ and C₃₀₀), dimers and monomers) represent progressive stages of the folding path-way of tubulin proteins. We have shown here that the final step of this folding reaction, the formation of the het-erodimer, is dependent on tubulin concentration in the pres-ence of magnesium ions and GTP hydrolysis and does not require ATP.

On the other hand, in the absence of reticulocytes, Mg²⁺or GTP the equilibrium does not take place as predicted (see Figs 1, 2 or 3). Monomers partially purified through Superose-12 (according to the method of Zabala and Cowan, 1992) are not incorporated into preformed dimers in the absence of reticulocyte extracts (data not shown). Therefore, we consider the monomers observed by gel per-meation and non-denaturing electrophoresis as incom-pletely folded monomers. Similar results were obtained by Gao et al. (1993) who showed that the monomeric product is not competent for exchange or coassembly with added carrier microtubules and suggested that correctly folded monomeric β-tubulin is unstable unless it can exchange with available tubulin heterodimers.

In the tubulin folding pathway, the accumulation of C₉₀₀ complexes when ATPγS was added to in vitro translation reactions suggests the need for ATP hydrolysis at least in the first part of the process (Yaffe et al., 1992). Here we present evidence that incorporation of monomeric β-tubu-lin into dimers in reactions containing partially purified C₃₀₀ complexes requires GTP hydrolysis, but is independent of ATP (Fig. 4). The use of prolonged incubations at 37°C of the translation reactions reduced the amount of C₉₀₀ com-pared to C₃₀₀ complexes (Yaffe et al., 1992 and results not shown); under these conditions the residual C₉₀₀ complexes are not participating in the formation of the heterodimer as is also suggested by the fact that the addition of ATPγS to the translation reactions did not preclude the formation of the dimer (A.F. and J.C.Z. unpublished results). Further-more, TCP-1 could not be detected in C₃₀₀ complexes as assayed by immunoblotting with an anti-TCP-1 monoclonal antibody; since TCP-1 is an ATP-dependent protein this result might substain our finding. Although further charac-terization of the release factors and the composition of the C₃₀₀ complexes will facilitate a better understanding of the process of tubulin folding and dimer formation, the presently available data are compatible with a two-step mechanism. A first step involves the interaction of newly synthesized polypeptides with the C₉₀₀ complex, its disso-ciation is dependent on ATP (Yaffe et al., 1992). A second step would include the transition from the large complex to the C₃₀₀ complex and monomers and, finally, the assembly of the heterodimer. This second step requires GTP binding and hydrolysis. GTP also might be required for tubulin to adopt the active conformation. Although tubulin is a GTP-binding protein (Jacobs et al., 1974; Weisenberg et al., 1976), the involvement of another GTP-binding protein in the folding process is conceivable. From preliminary results we have indications that GTP hydrolysis is also necessary for the incorporation of α-tubulin into the tubulin dimer. Since α-tubulin is a GTP-binding protein and probably does not hydrolyze GTP (Carlier, 1982), it would suggest that a protein different from tubulin is responsible for the hydrol-ysis of GTP. This putative factor might correspond to one of the two protein cofactors required for the proper folding of tubulin, as recently reported (Gao et al., 1993). In addition, multiprotein complexes have been recently iso-lated from nocodazole-arrested CHO cells, which may be involved in mitotic spindle nucleation. They have been shown to contain α, β, and γ-tubulin, HSP70, and a 50 kDa protein identified as elongation factor 1α, a GTP-binding protein (Marchesi and Ngo, 1993).

In summary, comparison of the proposed folding mech-anisms for actin, and α- and β-tubulin suggests a common or similar first ATP-dependent step, and a second step in which GTP hydrolysis is required for the proper confor-mation of tubulin subunits and the assembly of the het-erodimer (Fig. 6).

We thank Juan M. Garcia-Lobo, Fernando de la Cruz and Nicholas J. Cowan for helpful discussions, advice and critical comments on the manuscript, and Rafael Campo for technical assistance. The TCP-1 specific monoclonal antibody was gener-ously provided by Keith Willison. This work was supported by DGICYT Proyecto no. PB91-0825 (to J.C.Z.).