Tubulin Folding Cofactors as GTPase-activating Proteins
1999; Elsevier BV; Volume: 274; Issue: 34 Linguagem: Inglês
10.1074/jbc.274.34.24054
ISSN1083-351X
AutoresGuoling Tian, Arunashree Bhamidipati, Nicholas J. Cowan, Sally A. Lewis,
Tópico(s)Ubiquitin and proteasome pathways
ResumoIn vivo, many proteins must interact with molecular chaperones to attain their native conformation. In the case of tubulin, newly synthesized α- and β-subunits are partially folded by cytosolic chaperonin, a double-toroidal ATPase with homologs in all kingdoms of life and in most cellular compartments. α- and β-tubulin folding intermediates are then brought together by tubulin-specific chaperone proteins (named cofactors A–E) in a cofactor-containing supercomplex with GTPase activity. Here we show that tubulin subunit exchange can only occur by passage through this supercomplex, thus defining it as a dimer-making machine. We also show that hydrolysis of GTP by β-tubulin in the supercomplex acts as a switch for the release of native tubulin heterodimer. In this folding reaction and in the related reaction of tubulin-folding cofactors with native tubulin, the cofactors behave as GTPase-activating proteins, stimulating the GTP-binding protein β-tubulin to hydrolyze its GTP. In vivo, many proteins must interact with molecular chaperones to attain their native conformation. In the case of tubulin, newly synthesized α- and β-subunits are partially folded by cytosolic chaperonin, a double-toroidal ATPase with homologs in all kingdoms of life and in most cellular compartments. α- and β-tubulin folding intermediates are then brought together by tubulin-specific chaperone proteins (named cofactors A–E) in a cofactor-containing supercomplex with GTPase activity. Here we show that tubulin subunit exchange can only occur by passage through this supercomplex, thus defining it as a dimer-making machine. We also show that hydrolysis of GTP by β-tubulin in the supercomplex acts as a switch for the release of native tubulin heterodimer. In this folding reaction and in the related reaction of tubulin-folding cofactors with native tubulin, the cofactors behave as GTPase-activating proteins, stimulating the GTP-binding protein β-tubulin to hydrolyze its GTP. GTPase-activating protein deoxy GTP dideoxy GTP subtilisin-truncated tubulin chicken erythrocyte tubulin 4-morpholineethanesulfonic acid guanosine 5′-3-O-(thio)triphosphate N-[2-hydroxy-1,1-bis(hydroxymethyl)ethyl]glycine Tubulin is an α/β heterodimer. Both α- and β-tubulins are GTP-binding proteins; α-tubulin binds GTP nonexchangeably, whereas the GTP-binding site on β-tubulin is exchangeable (1Spiegelman B.M. Penningroth S.M. Kirschner M.W. Cell. 1977; 12: 587-600Abstract Full Text PDF PubMed Scopus (95) Google Scholar). GTP hydrolysis by β-tubulin is a key element in determining the dynamic behavior of microtubules; the hydrolysis of GTP is coupled to tubulin polymerization such that variation in the size of the "cap" of GTP-bound subunits on a given microtubule determines whether it will continue to grow or undergo transition to a rapidly depolymerizing phase (2Kirschner M.W. Mitchison T.J. Cell. 1986; 45: 329-342Abstract Full Text PDF PubMed Scopus (1021) Google Scholar, 3Desai A. Mitchison T.J. Annu. Rev. Cell Dev. Biol. 1997; 13: 83-117Crossref PubMed Scopus (2006) Google Scholar, 4Tran P.T. Joshi P. Salmon E.D. J. Struct. Biol. 1997; 118: 107-118Crossref PubMed Scopus (74) Google Scholar). GTP hydrolysis is also essential for the biogenesis of native tubulin. Recent analysis of the tubulin folding pathway has shown that the α and β polypeptides must interact with a series of chaperone proteins before reaching the native state (reviewed in Ref. 5Lewis S.A. Tian G. Cowan N.J. Trends Cell Biol. 1997; 7: 479-485Abstract Full Text PDF PubMed Scopus (138) Google Scholar). The first step in this pathway is the binding of newly synthesized (or denatured) tubulin molecules to cytosolic chaperonin (6Gao Y. Thomas J.O. Chow R.L. Lee G.H. Cowan N.J. Cell. 1992; 69: 1043-1050Abstract Full Text PDF PubMed Scopus (427) Google Scholar) (also referred to as TRiC (7Frydman J. Nimmesgern E. Erdjument-Bromage H. Wall J.S. Tempst P. Hartl F.U. EMBO J. 1992; 11: 4767-4778Crossref PubMed Scopus (345) Google Scholar) and CCT (8Kubota H. Hynes G. Carne A. Ashworth A. Willison K. Curr. Biol. 1994; 4: 89-99Abstract Full Text Full Text PDF PubMed Scopus (274) Google Scholar)). Following one or more rounds of ATP hydrolysis by cytosolic chaperonin, the resulting quasi-native tubulin intermediates interact with tubulin-specific chaperones named cofactors A–E (9Gao Y. Melki R. Walden P.D. Lewis S.A. Ampe C. Rommelaere H. Vandekerckhove J. Cowan N.J. J. Cell Biol. 1994; 125: 989-996Crossref PubMed Scopus (75) Google Scholar, 10Tian G. Huang Y. Rommelaere H. Vandekerckhove J. Ampe C. Cowan N.J. Cell. 1996; 86: 287-296Abstract Full Text Full Text PDF PubMed Scopus (243) Google Scholar, 11Tian G. Lewis S.A. Feierbach B. Stearns T. Rommelaere H. Ampe C. Cowan N.J. J. Cell Biol. 1997; 138: 821-832Crossref PubMed Scopus (173) Google Scholar). Native tubulin is released from a supercomplex that contains both α- and β-tubulin and cofactors C, D, and E and that hydrolyzes GTP as part of this reaction (5Lewis S.A. Tian G. Cowan N.J. Trends Cell Biol. 1997; 7: 479-485Abstract Full Text PDF PubMed Scopus (138) Google Scholar, 11Tian G. Lewis S.A. Feierbach B. Stearns T. Rommelaere H. Ampe C. Cowan N.J. J. Cell Biol. 1997; 138: 821-832Crossref PubMed Scopus (173) Google Scholar, 12Fontalba A. Paciucci R. Avila J. Zabala J.C. J. Cell Sci. 1993; 106: 627-632Crossref PubMed Google Scholar). We sought to determine the nature of the GTP hydrolysis reaction performed by the tubulin-cofactor supercomplex and its role in producing native tubulin. Here we show that both in the folding reaction and in the related reaction of cofactors with native tubulin, the cofactors act as GTPase-activating proteins (GAPs),1 stimulating many-fold the negligible intrinsic GTPase activity of the tubulin to which they are bound. To generate purified tubulin 35S-labeled in the α- or β-subunit, full-length tubulin cDNAs encoding mouse α2 (13Villasante A. Wang D. Dobner P. Dolph P. Lewis S.A. Cowan N.J. Mol. Cell. Biol. 1986; 6: 2409-2419Crossref PubMed Scopus (234) Google Scholar) or mouse β3 (14Wang D. Villasante A. Lewis S.A. Cowan N.J. J. Cell Biol. 1986; 103: 1903-1910Crossref PubMed Scopus (225) Google Scholar) were 35S-labeled by coupled transcription/translation in rabbit reticulocyte lysate (Promega Inc., Madison, WI) and purified by anion exchange chromatography on DEAE-Sephacel (11Tian G. Lewis S.A. Feierbach B. Stearns T. Rommelaere H. Ampe C. Cowan N.J. J. Cell Biol. 1997; 138: 821-832Crossref PubMed Scopus (173) Google Scholar). Tubulin folding cofactors A, B, C, and D were either purified from bovine testis tissue or from Escherichia coli (as cloned recombinant proteins) as described (9Gao Y. Melki R. Walden P.D. Lewis S.A. Ampe C. Rommelaere H. Vandekerckhove J. Cowan N.J. J. Cell Biol. 1994; 125: 989-996Crossref PubMed Scopus (75) Google Scholar, 10Tian G. Huang Y. Rommelaere H. Vandekerckhove J. Ampe C. Cowan N.J. Cell. 1996; 86: 287-296Abstract Full Text Full Text PDF PubMed Scopus (243) Google Scholar, 11Tian G. Lewis S.A. Feierbach B. Stearns T. Rommelaere H. Ampe C. Cowan N.J. J. Cell Biol. 1997; 138: 821-832Crossref PubMed Scopus (173) Google Scholar). Tubulin folding cofactor E (which generates insoluble inclusion bodies upon expression in E. coli) was purified following expression as a biologically active protein in insect Sf21 cells using the BacPAK baculovirus expression system (CLONTECH Inc., Palo Alto, CA). For experiments using ribose-modified GTPs, native tubulin was exchanged with the corresponding nucleotide by anion exchange chromatography of twice-cycled bovine brain microtubules on DEAE-Sephacel (10Tian G. Huang Y. Rommelaere H. Vandekerckhove J. Ampe C. Cowan N.J. Cell. 1996; 86: 287-296Abstract Full Text Full Text PDF PubMed Scopus (243) Google Scholar) using buffer containing 50 μm dGTP or ddGTP. Microtubules were purified from calf brain by the method of Shelanski et al. (15Shelanski M.L. Gaskin F. Cantor C.R. Proc. Natl. Acad. Sci. U. S. A. 1973; 70: 765-768Crossref PubMed Scopus (2003) Google Scholar). Subtilisin-truncated tubulin (S-tubulin) and tubulin free from associated proteins were prepared as described (16Murphy D.B. Vallee R.B. Borisy G.G. Biochemistry. 1977; 16: 2598-2605Crossref PubMed Scopus (120) Google Scholar, 17Sackett D.L. Bhattacharyya B. Wolff J. J. Biol. Chem. 1985; 260: 43-45Abstract Full Text PDF PubMed Google Scholar). Chicken erythrocyte tubulin (E-tubulin) was purified from heparinized chicken blood (Pelfreeze Inc., Rogers, AK) by the procedure of Murphy and Wallis (18Murphy D.B. Wallis K.T. J. Biol. Chem. 1983; 258: 8357-8364Abstract Full Text PDF PubMed Google Scholar). In experiments to determine the potential for spontaneous tubulin subunit exchange, DEAE-Sephacel-purified bovine brain tubulin and S-tubulin (2 μm each) were incubated at 30 °C for 1 h in folding buffer (20 mm MES, pH 6.8, 50 mm KCl, 1 mm MgCl2, 1 mm EGTA, 1 mm GTP). To assess the role of cofactors in the exchange reaction, tubulin 35S-labeled in its α-subunit (1 μm) was incubated under the same conditions with various concentrations of E-tubulin or with 0.9 or 1.4 μm S-tubulin in the presence or absence of cofactors C, D, and E (each present at 1 μm) together with either 1 mm GTP or 1 mm GTPγS. Reaction products were analyzed by nondenaturing polyacrylamide gel electrophoresis as described (6Gao Y. Thomas J.O. Chow R.L. Lee G.H. Cowan N.J. Cell. 1992; 69: 1043-1050Abstract Full Text PDF PubMed Scopus (427) Google Scholar, 19Gao Y. Vainberg I.E. Chow R.L. Cowan N.J. Mol. Cell Biol. 1993; 13: 2478-2485Crossref PubMed Google Scholar). Unlabeled ddGDP was prepared by incubation of 5 mm ddGTP, 0.05 mm ADP, 1 unit of nucleoside diphosphate kinase (bovine liver), and 1 unit of hexokinase (bakers' yeast) (both from Sigma) in 20 mm Tris-HCl buffer, pH 7.6, containing 20 mm glucose and 1 mmMgCl2 at 30 °C for 16 h. Reaction products were purified by extraction with phenol and chloroform. ADP was converted to ATP by the addition of 1 mm ddGTP and 1 unit of nucleoside diphosphate kinase and further incubation at 30 °C for 1 h. ddGDP was purified from other nucleotides by anion exchange chromatography on a column (5/5 15Q; Amersham Pharmacia Biotech) developed with a linear gradient (0.02–1.0 m) of ammonium bicarbonate and checked by its mobility on TLC. [γ-32P]dGTP and [γ-32P]ddGTP were prepared in reactions containing 0.2 mCi of [γ-32P]ATP (specific activity, 3,000 Ci/mmol) and either 50 μm dGDP or ddGDP in 5 μl of 20 mm Tricine buffer, pH 7.2, containing 1 mm MgCl2 and 1 unit of bovine liver nucleoside diphosphate kinase. After incubation at 30 °C for 30 min, reaction products were isolated by extraction with phenol and chloroform. Remaining [γ-32P]ATP was converted to ADP by addition of unlabeled ATP to 50 mm together with 10 mm glucose and 1 unit of hexokinase and a further incubation at 30 °C for 10 min. Following a further extraction with phenol and chloroform, the final reaction products were purified by anion exchange chromatography as described above. [α-32P]GDP was generated by incubating [α-32P]GTP (specific activity, 8 Ci/mmol) and GDP (each at 0.1 mm) in a 50-μl reaction in the presence of 1 unit of nucleoside diphosphate kinase for 10 min at 30 °C, followed by extraction with phenol and chloroform and purification by anion exchange chromatography as described above. To measure the rates of hydrolysis of cofactors C, D, and E at different tubulin concentrations, D (0.32 μm), E (0.67 μm), and C (1.1 μm) were incubated at 30 °C in folding buffer with [γ-32P]GTP (22.5 μm) and various concentrations of DEAE-purified tubulin. Aliquots (2 μl) were withdrawn from the 12-μl reaction into 1 m perchloric acid at 1, 2, 3, 4, and 5 min, and the amount of phosphate released was quantitated (20Carlier M.F. Pantaloni D. Biochemistry. 1981; 20: 1918-1924Crossref PubMed Scopus (200) Google Scholar). In all cases less than 10% of the input GTP was hydrolyzed. The hydrolysis of [γ-32P]GTP, [γ-32P]dGTP, and [γ-32P]ddGTP by the tubulin-chaperone supercomplex was measured in the same way using 0.5 μm bovine brain tubulin, 0.1 μm each cofactors C, D, and E, and 20 μm nucleotide. In some experiments (see "Results"), the tubulin was preincubated with podophyllotoxin (20 μm) for 10 min at 30 °C. 32P-Labeled cofactor D-β-tubulin complexes resolved on nondenaturing gels (10Tian G. Huang Y. Rommelaere H. Vandekerckhove J. Ampe C. Cowan N.J. Cell. 1996; 86: 287-296Abstract Full Text Full Text PDF PubMed Scopus (243) Google Scholar) were located by autoradiography of the wet gels and excised. Nucleotide was extracted by maceration in 50% aqueous methanol. Gel fragments were removed by centrifugation, and the supernatants were dried under vacuum. The residue was dissolved in H2O and spotted onto phosphoethyleneimine TLC plates. The plates were developed with 1.2m LiCl (1Spiegelman B.M. Penningroth S.M. Kirschner M.W. Cell. 1977; 12: 587-600Abstract Full Text PDF PubMed Scopus (95) Google Scholar). Native α- and β-tubulins reportedly exist in free equilibrium with the heterodimer, with an apparent dissociation constant of about 1 μm (21Detrich H.W.D. Williams R.C. Biochemistry. 1978; 17: 3900-3907Crossref PubMed Scopus (150) Google Scholar, 22Mejillano M.R. Himes R.H. Biochemistry. 1989; 28: 6518-6524Crossref PubMed Scopus (29) Google Scholar, 23Panda D. Roy S. Bhattacharyya B. Biochemistry. 1992; 31: 9709-9716Crossref PubMed Scopus (33) Google Scholar, 24Sackett D.L. Zimmerman D.A. Wolff J. Biochemistry. 1989; 28: 2662-2667Crossref PubMed Scopus (24) Google Scholar). However, the fact that tubulin-specific chaperones bring together α- and β-tubulin (Fig.1 A) is puzzling if the subunits can freely exchange. To examine the dissociation of tubulin, we used S-tubulin (25Serrano L. Avila J. Maccioni R.B. Biochemistry. 1984; 23: 4675-4681Crossref PubMed Scopus (218) Google Scholar, 26Bhattacharyya B. Sackett D.L. Wolff J. J. Biol. Chem. 1985; 260: 10208-10216Abstract Full Text PDF PubMed Google Scholar), which migrates at a position distinct from that of untreated tubulin heterodimers upon native gel electrophoresis (Fig. 1 B). If subunit exchange among heterodimers is indeed spontaneous, then co-incubation of untreated tubulin and S-tubulin might be expected to generate a population of hybrid heterodimer molecules consisting of full-length α- and truncated β-tubulin or truncated α- and full-length β-tubulin having a mobility intermediate between that of native brain tubulin and S-tubulin. We detected no such hybrid heterodimers (Fig. 1 B, lanes 1 and 2). However, when the same reaction was repeated using brain tubulin 35S-labeled in its α-subunit in the presence of the three tubulin-folding cofactors (C, D, and E) that are common to the α- and β-tubulin folding pathways (Fig.1 A), a shift of label to an electrophoretic position between tubulin and S-tubulin was observed (Fig. 1 C, lanes 3 and 4). This shift was inhibited by the inclusion of the slowly hydrolyzable analog GTPγS (Fig. 1 C, lanes 5 and 6). (In these reactions, most of the counts are found in the slowly migrating tubulin-cofactor supercomplex). These results imply that exchange of subunits between tubulin and S-tubulin requires the action of cofactors and the hydrolysis of GTP. Because it is possible that treatment of tubulin with subtilisin significantly alters its exchange properties, we repeated this experiment using E-tubulin. Native brain tubulin and E-tubulin (27Murphy D.B. Wallis K.T. Machlin P.S. Ratrie III, H. Cleveland D.W. J. Biol. Chem. 1987; 262: 14305-14312Abstract Full Text PDF PubMed Google Scholar) migrate with different mobilities upon native gel electrophoresis (Fig.1 D) because of sequence differences in the β-subunits; however, the sequence of the α-subunit of E-tubulin (28Rudiger M. Weber K. Eur. J. Biochem. 1993; 218: 107-116Crossref PubMed Scopus (39) Google Scholar) is 99% identical to that of the mouse α-tubulin, which we labeled to test for subunit exchange. If free exchange of subunits among heterodimers can indeed occur, then incubation of brain tubulin labeled in its α-subunit with unlabeled E-tubulin should result in a shift of the label to the electrophoretic position characteristic of the E-tubulin dimer. Once again, no such shift was observed in this experiment (Fig.1 E, lane 1); the same result was obtained over a range (0.6–9.6 μm) of tubulin concentrations (data not shown). However, when the same reaction was repeated in the presence of cofactors C, D, and E, a shift of label to the electrophoretic position of E-tubulin was observed (Fig. 1 E, lane 2), showing that in this case α-subunit exchange had occurred between the two types of tubulin. The generation of hybrid heterodimers in this experiment was also dependent upon the hydrolysis of GTP, because the reaction was inhibited by GTPγS (Fig. 1 E, lane 3). Thus, the subunits of the α/β-tubulin heterodimer can only exchange by transit through the supercomplex, which is a part of the de novo tubulin folding pathway. We have previously shown that when tubulin and either cofactors C and D or cofactors C, D and E are mixed, GTP hydrolysis ensues (11Tian G. Lewis S.A. Feierbach B. Stearns T. Rommelaere H. Ampe C. Cowan N.J. J. Cell Biol. 1997; 138: 821-832Crossref PubMed Scopus (173) Google Scholar). To understand the nature of this reaction, we measured the rate of GTP hydrolysis as a function of the concentration of added tubulin (Fig.2 A). We found that the cofactors behave with Michaelis-Menton kinetics, with tubulin-GTP as substrate and Pi as product. The cofactors have a Km for tubulin of 0.1 μm and a turnover number of about 1.6 min−1 (6 pmol Pi/min/3.8 pmol cofactor D monomer, where D is the limiting cofactor). Because the Km is about 200-fold lower than the critical concentration for tubulin polymerization and because the rate of GTP hydrolysis is saturable with increasing tubulin, it is clear that this hydrolysis is not the result of the cofactors acting as microtubule-associated proteins and promoting microtubule polymerization with its concomitant GTP hydrolysis. Rather, GTP hydrolysis is the result of the interaction of cofactors with tubulin dimers themselves. To underscore this point, we found that these two types of GTP hydrolysis are additive (Fig. 2 B). At 9 μm tubulin but not at 3 μm tubulin, some polymerization-dependent GTP hydrolysis is observed. At the former concentration there is no net polymerization of microtubules, but there is hydrolysis because of tubulin-tubulin interactions (29Carlier M.F. Didry D. Pantaloni D. Biophys. J. 1997; 73: 418-427Abstract Full Text PDF PubMed Scopus (37) Google Scholar). The addition of cofactors C, D, and E to tubulin at these two concentrations results in approximately the same increase in the rate of hydrolysis in each case, showing that the two types of hydrolysis are proceeding independently of each other. We conclude that the cofactors are acting on the G-protein tubulin as GAPs. In the absence of cofactor E, the combination of cofactors C and D have GAP activity; C and D stimulate GTP hydrolysis by tubulin, although they have a higher Km for tubulin (0.19 μm) and a 6-fold lower turnover number (0.3/min) (Fig. 2 A). To gather more direct evidence that the GTP hydrolysis step in the tubulin heterodimerization reaction is performed by the β-tubulin subunit and not by one of the cofactor proteins contained in the supercomplex, we looked at the reaction of cofactors with tubulin in the presence of ribose-modified forms of GTP; both dGTP and ddGTP support microtubule growth in vitro and are hydrolyzed as efficiently as GTP by the β-tubulin subunit upon polymerization (30Hamel E. Lustbader J. Lin C.M. Biochemistry. 1984; 23: 5314-5325Crossref PubMed Scopus (38) Google Scholar). This property is diagnostic of the tubulin GTPase; polymerases and transferases, for example, utilize deoxy- and dideoxyribonucleotides very poorly. We found that dGTP and ddGTP also support the formation of tubulin-cofactor D complexes and can be used in the reaction that generates the heterodimer from these complexes by the addition of cofactors C and E and tubulin dimer (as a source of the α-subunit) (Fig. 2 C). This result is independent of the concentration of nucleotide in the range 10–300 μm (data not shown). Furthermore, both dGTP and ddGTP are hydrolyzed as rapidly or slightly more rapidly than is GTP by the supercomplex (Fig. 2 D). This unusual behavior therefore supports the idea that it is the β-tubulin in the supercomplex that performs the hydrolysis step leading to heterodimer release. We also examined the effect of microtubule poisons on the rate of hydrolysis by the supercomplex. Podophyllotoxin, which inhibits tubulin-dependent GTP hydrolysis in microtubule polymerization reactions (31Hamel E. Med. Res. Rev. 1996; 16: 207-231Crossref PubMed Scopus (356) Google Scholar) and suppresses hydrolysis-dependent dynamic instability (32Schilstra M.J. Martin S.R. Bayley P.M. J. Biol. Chem. 1989; 264: 8827-8834Abstract Full Text PDF PubMed Google Scholar), also slows GTP hydrolysis by the supercomplex (Fig. 2 D) to a remarkably similar extent (33Lin C.M. Hamel E. J. Biol. Chem. 1981; 256: 9242-9245Abstract Full Text PDF PubMed Google Scholar). These data provide further evidence that it is the tubulin in the supercomplex that hydrolyzes GTP when stimulated by cofactors. The data presented in Fig. 1 show that GTP hydrolysis is necessary for the reaction in which cofactors act on native tubulin to scramble the heterodimers. We have previously shown that GTP is likewise necessary in the tubulin folding reaction in which cofactors bring together newly folded α- and β-tubulin subunits (11Tian G. Lewis S.A. Feierbach B. Stearns T. Rommelaere H. Ampe C. Cowan N.J. J. Cell Biol. 1997; 138: 821-832Crossref PubMed Scopus (173) Google Scholar). These two reactions share the GTP hydrolysis-dependent release of heterodimer from the supercomplex (Fig. 1). We hypothesized that GTP hydrolysis may directly lead to release of dimer if cofactors have a much lower affinity for GDP-tubulin than GTP-tubulin. To test this idea, we incubated tubulin with cofactors in the presence of either [α-2P]GTP or [α-2P]GDP. Tubulin-cofactor D complexes form very inefficiently when only GDP is present (Fig.3 A). To confirm this result, we mixed tubulin and cofactor D with equal quantities of [α-32P]GTP and [α-32P]GDP and analyzed by TLC the nucleotide content of the complexes thus formed. Only GTP was found in such complexes (Fig. 3 B). These experiments show that cofactor D has a much lower affinity for GDP-β-tubulin than for GTP-β-tubulin. We infer that hydrolysis of GTP by β-tubulin in the supercomplex leads directly to its release from that complex (in the form of tubulin heterodimer). The simple and direct data presented here showing that α- and β-tubulin subunits do not exchange in the absence of tubulin-folding cofactors imply either that the subunits are very tightly associated in the heterodimer or that they denature as they dissociate. The latter explanation cannot be true, however, because this would imply that tubulin is highly unstable, which it is not, even at concentrations below the reported dissociation constant of the heterodimer, long thought to be in the micromolar range (21Detrich H.W.D. Williams R.C. Biochemistry. 1978; 17: 3900-3907Crossref PubMed Scopus (150) Google Scholar, 22Mejillano M.R. Himes R.H. Biochemistry. 1989; 28: 6518-6524Crossref PubMed Scopus (29) Google Scholar, 23Panda D. Roy S. Bhattacharyya B. Biochemistry. 1992; 31: 9709-9716Crossref PubMed Scopus (33) Google Scholar, 24Sackett D.L. Zimmerman D.A. Wolff J. Biochemistry. 1989; 28: 2662-2667Crossref PubMed Scopus (24) Google Scholar). The conclusion is inescapable that the subunits of the tubulin heterodimer are tightly bound to each other and are not in free equilibrium. There are compelling recent data that support this surprising finding; when tubulin is bound to a column via antibody specific to the α-subunit, the β-subunit cannot be dissociated unless it is exposed to Tris buffer at high pH (34Giraudel A. Lafanechere L. Ronjat M. Wehland J. Garel J.-R. Wilson L. Job D. Biochemistry. 1998; 37: 8724-8734Crossref PubMed Scopus (10) Google Scholar). How then can we explain the corpus of results that have been interpreted to reflect a micromolar dissociation constant for the heterodimer? If tubulin undergoes some concentration-dependent conformational change (but does not dissociate), this would account for many of these experimental results. Our data imply that the process whereby tubulin heterodimers are assembled includes a quality control step. We know that only dimer, not individual subunits, can be released from the supercomplex, because dimers can only exchange subunits via the supercomplex (Fig. 1). Furthermore, dimer is released only when it has demonstrated its ability to hydrolyze GTP (Fig. 1, C and E). In support of this interpretation, tubulin molecules with mutations that are predicted to abolish GTP hydrolysis are not assembled into microtubules when expressed in vivo and remain complexed with cofactors when made in vitro (35Zabala J.C. Fontalba A. Avila J. J. Cell Sci. 1996; 109: 1471-1478Crossref PubMed Google Scholar). Additionally, in a genetic screen for viable yeast harboring β-tubulin mutations that altered rates of GTP hydrolysis, only mutations that increased the rate of hydrolysis were found (36Davis A. Sage C.R. Dougherty C.A. Farrell K.W. Science. 1994; 264: 839-842Crossref PubMed Scopus (55) Google Scholar). Many toxins bind to tubulin and interfere with its polymerization, some by changing the GTPase activity of tubulin (reviewed in Ref. 31Hamel E. Med. Res. Rev. 1996; 16: 207-231Crossref PubMed Scopus (356) Google Scholar). These toxins can poison microtubules substoichiometrically; if poisoned subunits are added to the growing end of a microtubule, the growth and stability of the whole tubule is affected (32Schilstra M.J. Martin S.R. Bayley P.M. J. Biol. Chem. 1989; 264: 8827-8834Abstract Full Text PDF PubMed Google Scholar, 37Vandecandelaere A. Martin S.R. Schilstra M.J. Bayley P.M. Biochemistry. 1994; 33: 2792-2801Crossref PubMed Scopus (34) Google Scholar). By analogy, the quality control of tubulin by cofactors in vivo may be important for protecting the microtubules of the cell from poisoning by misfolded, non-GTP-hydrolyzing subunits. In addition to acting in the de novo folding pathway (10Tian G. Huang Y. Rommelaere H. Vandekerckhove J. Ampe C. Cowan N.J. Cell. 1996; 86: 287-296Abstract Full Text Full Text PDF PubMed Scopus (243) Google Scholar, 11Tian G. Lewis S.A. Feierbach B. Stearns T. Rommelaere H. Ampe C. Cowan N.J. J. Cell Biol. 1997; 138: 821-832Crossref PubMed Scopus (173) Google Scholar), cofactors can interact with native tubulin dimer to convert GTP-tubulin to GDP-tubulin (Fig. 2); this process could function in the quality control of the pool of tubulin in the cell and/or in regulating tubulin polymerization. Many GTP-binding proteins, including the heterotrimeric G proteins and the Ras-related small GTP-binding proteins, hydrolyze GTP at a very low intrinsic rate that can be increased enormously by the binding of specific GAPs. These proteins appear to contribute an "arginine finger" to the catalytic site of the GTP-binding protein and in doing so activate it (38Sceffzek K. Ahmadian M.R. Wittinghofer A. Trends Biochem. 1998; 23: 257-262Abstract Full Text Full Text PDF PubMed Scopus (357) Google Scholar). Although the GTP-binding site of tubulin and the related bacterial protein FtsZ have a different fold from that of all other GTP-binding proteins (39Nogales E. Downing K.H. Amos L.A. Lowe J. Nat. Struct. Biol. 1998; 5: 451-458Crossref PubMed Scopus (443) Google Scholar, 40Nogales E. Wolf S.G. Downing K.H. Nature. 1998; 391: 199-203Crossref PubMed Scopus (1825) Google Scholar), the experiments described here show that, like other GTP-binding proteins, a specific GAP (in this case tubulin-folding cofactors) stimulates the negligible intrinsic GTP hydrolysis rate of tubulin by orders of magnitude. In this sense, α-tubulin also behaves like a GAP when it interacts with the GTP-binding domain of an adjacent β-subunit during microtubule polymerization (39Nogales E. Downing K.H. Amos L.A. Lowe J. Nat. Struct. Biol. 1998; 5: 451-458Crossref PubMed Scopus (443) Google Scholar, 41Erickson H.P. Trends Cell Biol. 1998; 8: 133-137Abstract Full Text Full Text PDF PubMed Scopus (106) Google Scholar). One of the tubulin-folding cofactors and α-tubulin may therefore share some structural homology that allows each to contribute to the catalytic site of β-tubulin polypeptides in tubulin-cofactor supercomplexes and in microtubules, respectively. We thank N. Kallenbach and E. Hamel for helpful discussions.
Referência(s)