Cell, Vol. 12, 587-800,

Turnover Hamster

November

1977,

Copyright

0 1977 by MIT

of Tubulin and the N Site GTP in Chinese Ovary Cells

Bruce M. Spiegelman, Stephen M. Penningroth* and Marc W. Kirschner Department of Biochemical Sciences Moffett Laboratories Princeton University Princeton, New Jersey 08540

Summary Radioactively labeled tubulin from Chinese hamster ovary (CHO) cells can be isolated by copolymerization with nonradioactive porcine brain mlcrotubule protein. 75% of the soluble tubulin in CHO extracts co-polymerizes with the porcine protein through several cycles, without preferential loss of either CHO or porcine subunits. After phosphocellulose chromatography of the co-polymerized microtubules, the CHO tubulin is radiochemically homogeneous, as judged by SDSpolyacrylamide gel electrophoresis. CHO tubulin purified in this way has 1 mole of nucleotide per mole of protein noncovalently bound at the nonexchangeable or N site. Thin-layer chromatography indicates that the N site nucleotide is entirely ribo-GTP. Label and chase experiments show that the N site GTP exchanges intracellularly with a half-time of 33 hr in growing cells which have a generation time of 17 hr, while the tubulin polypeptides are degraded with a half-time of 48 hr. Intracellular hydrolysis of the y-phosphate of the N site nucleotlde can be detected but occurs very slowly, with a half-time of 24 hr. These results suggest that the N site nucleotide may function in vlvo as a stable structural co-factor of the tubulin molecule and render Improbable the possibility that it has a regulatory role in microtubule assembly. Introduction Microtubules, polymerized forms of the globular protein tubulin, are cylinders of diameter 25 nm and indeterminate length which function as supporting structures and perhaps as motile elements in cell division, cell morphology and other processes (see reviews by Olmsted and Borisy, 1973; Soifer, 1975; Snyder and McIntosh, 1978; Goldman, Pollard and Rosenbaum, 1976). Studies of the assembly of purified microtubules in vitro have suggested methods of regulating polymerization which could involve accessory proteins, calcium concentration or the state of the nucleotides on tubulin. Purified tubulin from mammalian brain is * Present Princeton,

address: Department New Jersey 08540.

of Biology,

Princeton

University,

known to bind 2 mole of nucleotide per mole of protein (Weisenberg, Borisy and Taylor, 1968). 1 mole of GTP or GDP exchanges with nucleotide in the medium and is therefore said to be bound at the exchangeable or E site. The other mole of guanosine nucleotide does not exchange with nucleotide in the medium and is therefore said to be bound at a nonexchangeable or N site. Microtubule assembly in vitro requires the addition of a nucleoside triphosphate, ATP or GTP (Weisenberg, 1972). It is now known that GTP added exogenously to purified microtubule protein is bound at the E site and is hydrolyzed to GDP during microtubule assembly (Kobayashi, 1975; Penningroth and Kirschner, 1977). We have recently shown that nonhydrolyzable nucleotide analogs such as GMPPCP can also bind to the E site of tubulin and induce assembly in vitro, suggesting that the E site nucleotide functions by binding to tubulin and inducing a conformational alteration in the protein which favors assembly (Penningroth, Cleveland and Kirschner, 1978). It has proven considerably more difficult to study the role of the N site nucleotide in vitro. Even the identity of this nucleotide has been the subject of conflicting reports. Several investigators have reported that the N site of brain tubulin purified by ion-exchange chromatography may contain GDP or GTP (Berry and Shelanski, 1972; Jacobs, Smith and Taylor, 1974), while recent reports using protein purified by temperature-induced cycles of polymerization have claimed that only GTP is bound at the N site, and that this GTP is not hydrolyzed during microtubule assembly in vitro (Kobayashi, 1975; Penningroth and Kirschner, 1977). Because of its apparent lack of reactivity in vitro, it has not been possible to determine what role the N site nucleotide has in the function of the tubulin molecule. Since this nucleotide seems to be a stable part of the tubulin molecule, however, we have suggested that this site could serve as a control point for the assembly of tubulin in vivo (Penningroth et al. 1976). The lack of reactivity of the N site nucleotide in vitro has induced us instead to seek methods for studying the metabolism of this nucleotide on tubulin within cells. To date, no information has been reported concerning the intracellular utilization of the N site nucleotide. While methods of purifying microtubule proteins by temperature-dependent cycles of polymerization have been successfully used to purify microtubule proteins from some cultured cells (Wiche and Cole, 1976), the very large numbers of cells required make this technique less effective for experiments which require physiological manipulations of the cells. We report here on the use of a method of co-polymer-

Cell 500

king radioactively labeled CHO tubulin which permits the separation of radiolabeled microtubule proteins of cultured cells from other proteins in a radioactive cell extract. This method is well suited to studies of tubulin and accessory proteins which require physiological manipulations of cells. We describe here some of the properties of this purification system and use it to examine the turnover of CHO tubulin and the role of the N site nucleotide.

Results Isolation of CHO Tubulin by Co-Polymerization with Porcine Brain Microtubule Protein Radioactive tubulin from CHO cells was purified by cycles of polymerization and depolymerization using nonradioactive porcine brain microtubules as carrier. An extract of CHO cells labeled with 35S-methionine was prepared and mixed with unlabeled, purified porcine brain microtubule protein in a ratio of approximately 50 mg of microtubule protein per mg of total labeled extract protein. We subjected this mixture to repeated cycles of thermally induced assembly, disassembly and sedimentation to purify 35S-labeled CHO tubulin by copolymerization with the unlabeled carrier microtubule protein. The effectiveness of the co-polymerization procedure for purifying CHO tubulin is demonstrated by analysis of the products by SDS-polyacrylamide gel electrophoresis as shown in Figure 1. Sample a in Figure 1 shows a fluorogram of 35Smethionine-labeled total cell extract. Slots b, c and d present electrophoretic analyses of the 35Slabeled proteins following, respectively, one, two and three rounds of assembly and disassembly with unlabeled carrier microtubule protein. We observed a progressive enrichment for two 35Smethionine-labeled, closely spaced polypeptides of apparent molecular weight 54,000 daltons in successive microtubule pellets. These two polypeptides, which represented a constant 90% of the total radioactivity after the second cycle, co-migrate with (Y- and /3-tubulin from the unlabeled porcine brain carrier. The major CHO extract polypeptide, which has an apparent molecular weight of 47,000 daltons and co-migrates with rabbit muscle actin, is not purified by this procedure and decreases in successive microtubule pellets. If the CHO extract in the absence of carrier microtubule protein is warmed with 1 mM GTP and centrifuged exactly as in a and b, no enrichment for the tubulin polypeptides is observed (Figure 1, slots g and h). To purify 35S-labeled tubulin further, preparations which had been subjected to three cycles of co-assembly with carrier porcine microtubules were fractionated by phosphocellulose chromatography. The fraction unadsorbed to phosphocellu-

-T ‘- A

TA-

Figure 1. Co-Polymerization Proteins with Porcine Brain

of 58S-Labeled Microtubules

CHO

Microtubule

Labeled CHO microtubule proteins were isolated by co-polymerization with porcine brain microtubules and phosphocellulose chromatography as described in Experimental Procedures. Aliquots of the mixture were taken at various points throughout the procedure and quickly frozen. They were later thawed, TCAprecipitated and centrifuged, and the precipitated protein was analyzed by SDS-lo% polyacrylamide gel electrophoresis and fluorography. T and A indicate the positions of porcine tubulin and rabbit actin standards. We have followed the nomenclature of Borisy et al. (1975) in describing stages of assembly-disassembly purification. H refers to a sedimentation at 25°C and C refers to a sedimentation at 4°C. The numerical subscript refers to the cycle number. (a) crude extract; (b) H,P; (c) H*P; (d) HIP; (e) HIP material not absorbed to phosphocellulose column; (1) 1.0 M NaCl wash from phosphocellulose column; (g) crude extract with no added porcine carrier microtubule protein; (h) H,P of sample with no added porcine carrier microtubule protein.

lose contained nearly homogeneous (>98% pure) 35S-methionine-labeled material which co-migrated with (Y- and P-tubulin (Figure le). A number of nontubulin CHO polypeptides, as well as 5% of the CHO material co-migrating with tubulin, remained bound to the phosphocellulose column and could be eluted with a single salt cut of 1 .O M NaCl

Physiological

Studies

of Tubulin

569

(Figure If). The nontubulin CHO polypeptides are the subject of a separate communication (B. M. Spiegelman and M. W. Kirschner, manuscript in preparation). The identity of the major CHO polypeptides purified by co-polymerization can be further demonstrated to be tubulin by the peptide mapping procedure of Cleveland et al. (1977). As a standard, porcine brain tubulin was methylated with Xformaldehyde (Rice and Means, 1971) and analyzed by the same peptide mapping procedure as the 54,000 dalton molecular weight CHO polypeptides (Figure 2a). Comparison of Figures 2a and 2b shows that the 54,000 dalton molecular weight CHO polypeptides yielded a banding pattern which is virtually identical with the pattern obtained from the porcine brain tubulin. This provides strong evidence that the major polypeptides purified from CHO extracts by cycles of co-polymerization with porcine brain microtubules are CHO (Y- and ptubulin. For comparison, the 35S-labeled CHO material which co-migrated with rabbit muscle actin in the H,P step was peptide mapped by the same procedure and compared to the peptide products obtained from 14C-labeled rabbit muscle actin. As can be seen in Figures 2c and 2d, the banding pattern obtained from the 35S-labeled 47 CHO protein is similar to that obtained from the 14C-labeled actin, and dissimilar to tubulin. Quantitation of Co-Polymerization of CHO Tubulin We have quantitated the co-polymerization reaction to determine the efficiency of interaction of CHO tubulin with the porcine microtubule protein. In addition, we have sought to investigate whether a class of inactive subunits, as has often been suggested (Borisy et al., 1975), could be detected in the CHO extracts. In the experiment presented in Figure 3, the ratio of 35S radioactivity (entirely from the CHO cells) to protein (>97% from the porcine microtubule protein) was followed through five cycles of co-polymerization purification. The ratio of radioactivity to protein, referred to as the “activity ratio,” drops rapidly by 95% from an initial value of 2.4 x lo3 cpm/pg to 0.12 x lo3 cpm/kg at the H,P step. No further decrease in the activity ratio is observed during three additional cycles of co-polymerization, however, indicating that active purified CHO tubulin subunits are capable of copolymerizing and depolymerizing with porcine microtubule protein without preferential loss of either CHO or porcine tubulin. In contrast to the activity ratio, the level of radioactivity in the co-polymerization mixture decreases during each cycle and reaches a value of 0.1% of the initial extract radioactivity at the H,P step (not shown). The stable plateau which the activity ratio reaches at 4.5% of

the initial activity ratio in the extract indicates that 4.5% of the 3 radioactivity in the extracts is present in active CHO microtubule proteins. Since approximately 90% of the purified CHO material is tubulin, it can be estimated that 4.1% of the 35Smethionine-labeled material in the extract is an active form of tubulin. Previous attempts to quantitate the fraction of cell extract proteins which are active microtubule components have been frustrated by the instability of the tubulin molecule in

b

cl

Figure

2. Peptide

Maps

c

of the CHO Tubulin

and Actin

CHO YS microtubule proteins were purified by one cycle of copolymerization, subjected to polyacrylamide gel electrophoresis, stained and destained. The material which co-migrated with rabbit muscle actin and porcine brain tubulin was cut out of the gel, eluted electrophoretically into a dialysis bag and precipitated with 20% (final concentration) TCA after addition of 100 rg of cold carrier rabbit muscle actin or porcine brain tubulin. The use of other cold carrier proteins did not alter the results. This material was ether extracted and dissolved in 250 r.d of digestion buffer which contained 0.125 M Tris-HCI (pH 6.6). 1 mM EDTA, 0.5% SDS and 10% glycerol. The peptide fragments were generated and analyzed by the procedure of Cleveland et al. (1977). 0.4 pg of chymotrypsin (Worthington) were added to 50 pl samples, and this was incubated at 37°C for 30 min. Incubation was terminated by adding 5 ~1 of 10% SDS, 5 ~1 of mercaptoethanol and boiling for 5 min. 1 ~1 of 0.1% bromophenol blue was added, and samples were electrophoresed on 15% polyacr-ylamide gels and fluorographed as described. Nonradioactive proteins were labeled with ‘*C-formaldehyde (Rice and Means, 1971). In each case, the slot to the right was proteolyzed and the slot to the left was not. (a) “C porcine brain tubulin; (b) KS-CHO tubulin; (c) “C rabbit muscle actin; (d) “‘S-CHO actin.

Cdl 590

vitro, and by the monomer-polymer equilibrium nature of microtubule assembly. The presence of the cold carrier porcine microtubule proteins, which remain in constant stoichiometry with CHO components during cycles of polymerization, serves as an internal control for both these factors. This assumes, of course, that both types of subunits behave in the extract as they do at later purification steps. To determine whether the 4.1% of the extract radioactivity which was in active tubulin represented all or part of the cellular pool of tubulin, we measured the total amount of CHO tubulin in cell extracts by gel electrophoresis assay. Comparison of lines 1 and 2 in Table 1 demonstrates that 14% of the total extract radioactivity co-electrophoresed with tubulin. That most or all of this radioactivity in the tubulin band is tubulin, rather than a coelectrophoresing contaminant, is indicated by its 3.0

r

Edict Figure

1

I I I I I I HIP CfS H2P C2S H3P H4P CYCLES OF PURIFICATION

3. Efficiency

of Cycles

I H5P

of Co-Polymerization

Labeled CHO microtubule proteins were purified by cycles of polymerization-depolymerization and sedimentation with porcine brain microtubule proteins as described in Experimental Procedures. At various stages of the purification, aliquots were taken from the mixture and quickly frozen. These were later thawed, TCA-precipitated and centrifuged, and the precipitate was analyzed for protein and radioactivity. The initial co-polymerization mixture contained 1.27 x 10’ cpm and 54 mg of protein. Activity ratio (cpm/pg) is plotted versus stage of purification.

sedimentation behavior during the first cycle of co-assembly (Table I, lines 3 and 4). The 4.1% of the extract radioactivity which is present in active tubulin therefore represents a subclass of tubulin comprising approximately 30% of the total tubulin in the extract, while 70% of the CHO tubulin represents an inactive class. We attempted to find out whether this “inactive” CHO tubulin represented an artifact of preparation, or a class of nonpolymerizable monomers or oligomers, by characterizing the effect of temperature and colchicine on its sedimentation behavior. The results in Table 1 show that following resuspension of the H,P material (line 4) and depolymerization at 4°C for 30 min, 66% of the labeled tubulin in the extract (84% of the labeled tubulin in the H,P) sedimented in the cold (line 5). By contrast, only 10% of the unlabeled porcine tubulin was not depolymerized by cooling. The addition of 5 x 1O-5 M colchicine to the initial co-polymerization extract, followed by incubation at 37°C and sedimentation at 25”c, resulted in the sedimentation of 65% of the CHO tubulin, while only 10% of the porcine tubulin sedimented (Table I, line 6). Thus treatment either with low temperature or with colchicine inhibited the sedimentation of only about 15% of the CHO tubulin, but inhibited 86% of the porcine tubulin. In a number of separate experiments, there was very close agreement between the amount of protein which was insensitive to cold and that which was insensitive to colchicine, suggesting that this represents a single class of protein. In another experiment (not shown), the sedimentation coefficient of the colchicine-insensitive material present in both CHO extracts and in purified porcine brain microtubule protein was estimated by velocity gradient centrifugation through 5-65% sucrose gradient. On these gradients, unsheared microtubules showed a sedimentation coefficient of approximately 500 S. The colchicine-insensitive CHO and porcine tubulin, which pelleted through the gradient, had a sedimentation coefficient larger than 10,000 S. When the activity ratio of tubulin in the extract is corrected for the presence of these aggregates in Table 1, line 5 or 6, the initial value of 1.06 x IO’ cpm/3.30 mg or 3.2 x IO3 cpmlwg (line 2) is decreased to 1.2 x IO3 cpm/pg. This would decrease the expected activity ratio of tubulin stably retained through cycles of co-polymerization from 14 to 5.3%. The close agreement between this corrected value and the fraction of the activity ratio, which is actually retained through multiple cycles of co-polymerization (4.1%), suggests that the large inactive tubulin aggregates may account for most or all of CHO tubulin which cannot be purified by this procedure.

Physiological 591

Table

Studies

1. Efficiency

of Tubulin

of Co-Polymerization

of CHO Tubulin Counts Protein

Extract

through

in s5S x 10’”

One Cycle

of Assembly

% Recovery

7.6

Total (w)”

Protein % Recovery

3.67

Tubulin

in Extract

1.06

100

3.30

100

Tubulin

in H,S

0.23

22

1.46

33

Tubulin

in H,P

0.62

76

2.21

67

Tubulin

in C,P

0.70

66

0.32

10

Tubulin

in H,P

0.69

65

0.32

10

(+5 x 1O-5 M Colchicine before Warming) a Per 1.6 x IO7 cells after labeling for 36 hr with IO &i/ml of Y+methionine. b Reflects primarily (297%) porcine microtubule proteins. Labeled CHO microtubule proteins were purified by one cycle of co-polymerization with porcine brain microtubule proteins. At various stages of the purification, aliquots were taken and frozen. They were later electrophoresed on SDS-10% polyacrylamide gels, stained, destained and fluorographed as described in Experimental Procedures. The tubulin band was removed with a razor blade, and the radioactivity was determined directly by liquid scintillation counting of the gel slice. Efficiency of recovery was determined to be 37% with gel-purified YQ-actin. Protein was determined by the method of Lowry et al. (1951), and nonradioactive tubulin was quantitated by microdensitometry of Coomassie-stained and destained gels in a Zeiss spectrophotometer equipped with a linear transport.

Isolation and Stoichiometry of the N Site GTP on CHO Tubulin The co-polymerization purification of 32POrlabeled CHO tubulin provides an opportunity to label the N site nucleotide selectively. This was accomplished by growing cells in the presence of 32P04 and purifying the CHO tubulin through three cycles of co-polymerization with porcine brain tubulin. Since each cycle of co-polymerization was effected with 1 mM of unlabeled GTP, which is known to replace >90% of the nucleotide at the E site (Kobayashi, 1975), any radioactively labeled nucleotide present at the E site in the cell extract should be reduced to 0.1% of the total labeled nucleotide following three cycles of co-polymerization. By contrast, the radioactively labeled nucleotide at the N site in the cell extract, which does not exchange demonstrably with nucleotide in the medium, should retain 100% of its label through the co-polymerization purification procedure. Following the third round of co-polymerization, CHO tubulin was separated from microtuble accessory proteins by chromatography on phosphocellulose. The tubulin-associated nucleotides were released by ethanol precipitation of the protein and analyzed by thin layer chromatography, using a chromatographic system capable of resolving ATP, GTP, GDP and GMP. Autoradiography of thin layer chromatograms of tubulin-associated nucleotides from 32POrlabeled, purified CHO tubulin revealed a single spot co-migrating with a GTP standard (Figure 4a). No other components were detected. Analysis of the CHO N site nucleotide in another solvent system, which separates ribo- and deoxyribo-GTP (Gonzales and Geel, 1975), showed ribo-GTP as the only visible component (Figure 4b). These results indicated that the N

site of active CHO tubulin binds ribo-GTP specifically. The stoichiometry of the N site nucleotide was estimated by growing cells for 60 hr in medium containing both 80 #Xml of 32P04 and 8 &i/ml of 35S-methionine. In this medium, the cell number doubled twice in a period of 36 hr, but after that time, there was no further increase in cell number, although the cells remained viable as judged by their ability to exclude trypan blue stain. The 4 fold increase in cell number during the labeling period, plus the endogenous turnover of intracellular components which also occurs (see below), should result in near radiochemical equilibration of the cells with the medium. The stoichiometry of the N site nucleotide in relation to the CHO tubulin was estimated by determining the 32Pcounts in GTP and the 35S counts in tubulin purified by phosphocellulose chromatography following three cycles of copolymerization. The results in Table 2 show that a value of 0.7-1.1 mole of GTP per mole of CHO tubulin was obtained. Comparison of Intracellular Tubulin Turnover and N Site GTP Exchange The ability to isolate active CHO tubulin containing stoichiometric amounts of selectively labeled N site GTP provided an opportunity for measuring tubulin turnover and N site GTP exchange in the cells. By using a 35S/32P double label, tubulin turnover and GTP exchange were determined in the same experiment, thus providing a direct comparison between these two processes. Two labelchase experiments were performed. In the first experiment, CHO cells were labeled with 10 &I/ ml of 32P and 1 &i/ml of 35S. Cells grew logarith-

Cdl 592

Table 2. Stoichiometr-y

of GTP Associated

cpm per Aliquot” ‘*P in GTP 5sS in Tubulin

Q-

j 1

-Q

Figure 4. s*P Nucleotide Associated with CHO Tubulin Cells were labeled for 24 hr with 60 &i/ml SPP-orthophosphate, and CHO tubulin was isolated by three cycles of co-polymerization with porcine carrier microtubule protein and phosphocellulose chromatography. Protein was ethanol-precipitated (50% ETOH) and sedimented, and thin-layer chromatography was performed on the supernatant on polyethyleneimine-cellulose thin-layer plates (Bakerflex). Cold standard nucleotides were added to the radioactive sample and chromatographed simultaneously. (a) developed in 1.2 M LiCl as described by Penningroth, Cleveland and Kirschner (1976); (b) developed in 1.5 M LiCI, saturated with boric acid (pH 7.0 with NH,OH) as described by Gonzales and Gee1 (1975).

mically during the pulse and the chase periods with a doubling time of 15-17 hr. The results in Figure 5a show that during the chase, tubulin decayed with a half-life of 48 hr, while the half-life of GTP decay was 15 hr. The 32P radioactivity found covalently associated with the CHO phosphocellulose tubulin fraction decayed rapidly, with 70% of the counts lost in 3 hr of cold chase. This value further decreased to 18% of the initial value by 12 hr of cold chase, but did not decrease beyond this level in another 12 hr of nonradioactive chase. This rapid decay of covalent phosphate radioactivity indicates that the cold chase was effective in rapidly decreasing the specific activity of the intracellular phosphate pool, and suggests that the phosphate covalently associated with the

31.2 477.6

mole per Aliquot* (X 10-5)

with CHO Tubulin Stoichiometry (mole GTP/mole Tubulin)

1.6 2.5’ 1 .Bd 1.6’

0.7’ 1 .Od 1.1’

a Radioactivity is expressed as counts per minute as GTP or tubulin in a 100 ~1 aliquot. 35S-tubulin purity was estimated by microdensitometry of gel electrophoresis autoradiographs of parallel, single labeled samples. Nucleotide purity was estimated by thin-layer chromatography on polyethyleneimine-cellulose sheets using 1.2 M LiCl as a solvent. b The specific activity of the phosphate in the medium was determined by the scintillation counting and total phosphate determination method of Ames (1966). The specific activity of the CHO tubulin was estimated from the specific activity of the methionine in the medium, determined with a Beckman amino acid analyzer and scintillation counting, and the methionine content of tubulins given below: c Bryan and Wilson (1971). 17 mole of methionine per mole of chick brain tubulin. d Luduena and Woodward (1973), 26 mole of methionine per mole of sea urchin tubulin. r Eipper (1974). 29 mole of tubulin per mole of rat brain tubulin. Cells were grown for 60 hr in 60 @/ml 32P-orthophosphate and 6 &i/ml YB-methionine. CHO tubulin was purified by three cycles of co-polymerization and phosphocellulose chromatography. 32P-nucleotide and 36S-labeled protein were separated by TCA precipitation as described in Experimental Procedures.

CHO phosphocellulose tubulin fraction turns over rapidly. Unfortunately, the specific activity of the samples was too low to determine with which proteins the 32P was associated. In a second experiment, CHO cells were labeled with 80 pCi/ml of 32P and 8 &i/ml of 35S. Under these conditions, cells doubled twice during the labeling period but showed no increase in cell number during the chase period. Cells remained viable during the chase, however, as shown by their ability to exclude trypan blue stain. When the cell population was stationary during the chase period, tubulin decayed with a half-life of 34 hr and GTP with a half-life of 9 hr (Figure 5b).

Hydrolysis of GTP at the N Site Possible hydrolysis of the N site GTP on tubulin was measured in the same label-chase experiments as those measuring tubulin turnover and GTP exchange by comparing the specific activity of theyphosphate to the LY- and P-phosphates. At various times during the chase period, CHO tubulin was purified by three cycles of co-polymerization followed by phosphocellulose chromatography, and the 32P-labeled N site nucleotide was released by denaturing tubulin at 100°C for 5 min. An aliquot of the nucleotide released was hydrolyzed to GDP by incubation for 2 hr at 37°C with fresh porcine

Physiological 593

Studies

of Tubulin

brain microtubule protein (see Experimental Procedures), which contains a GTPase activity (Penningroth, 1977). The quantitative hydrolysis of labeled CHO tubulin N site to GDP was confirmed by thin-layer chromatography on PEI-cellulose followed by autoradiography (not shown). To measure the rate of GTP turnover, the amount of labeled y-phosphate released by the GTPase was determined following removal of GDP by adsorption to charcoal. Controls indicated that >98% of the total 32P counts were adsorbed to charcoal if the hydrolysis step was omitted. The results are expressed as the fraction of the total 32P nucleotide counts isolated for each time point which are present as the y-phosphate. As expected on the basis of complete radiochemical equilibration of intracellular pools during the labeling period, one third of the 32P associated with N site GTP is present in the y-phosphate at the start of the chase period (Figure 6). Approximately half of the yphosphate was found to turn over during a 24 hr chase in both growing and nongrowing cells. The use of the labeling procedure permitted us to examine only those nucleotides which remained bound to tubulin throughout the chase period. This estimate of y-phosphate turnover, therefore, is not affected by N site GTP exchange.

of the physiology and regulation of microtubules. It permits the isolation of tubulin and other microtubule-associated proteins from a small homogeneous population of cultured cells which can be readily manipulated for physiological studies. As shown in Figure 1, radioactive tubulin can be isolated from a small number of CHO cells in a nearly pure form by repeated polymerization, depolymerization and sedimentation, using nonradioactive porcine brain microtubules as carrier. Several other minor components are also isolated, and these can be separated from tubulin by chromatography on phosphocellulose. The peptide

Discussion Purification of CHO Tubulin by Co-Polymerization Physiological studies of microtubule proteins and their regulation require systems which are capable of being manipulated physiologically, and from which one can conveniently isolate microtubule components. While microtubule assembly processes occurring within animal cells have been described in several systems (Hsie and Puck, 1971; Johnson, Friedman and Pastan, 1971; Fuller, Brinkley and Boughter, 1975), they have not been extended to the biochemical level for lack of methods for isolating and examining the active microtubule components from manageable quantities of cells. Actively polymerizing microtubule proteins have been isolated from mammalian brain (Weisenberg, 1972; Shelanski, Gaskin and Cantor, 1973; Borisy et al., 1974), a source particularly rich in microtubules, and a great number of studies have described the properties of these components in vitro (see Soifer, 1975; Goldman, Pollard and Rosenbaum, 1976). Adult brain is static, however, in that it is not possible to correlate any particular biological process occurring within such a heterogeneous tissue with the microtubule components of that tissue. The co-polymerization of radioactively labeled protein from tissue culture cells with brain microtubule protein is well suited for studies

'OO

Figure

5. Turnover

I

I

I

6 TIME

12 OF COLD CHASE

18 (hrs)

24

with

CHO Tubulin

of 3*S and 12P Associated

I

4 x lo5 CHO cells were labeled with %S-methionine and 52Porthophosphate for 36 hr. The cells were then rinsed and grown for various times in nonradioactive medium before harvesting. Duplicate aliquots of phosphocellulose-purified. co-polymerized CHO tubulin were TCA-precipitated and analyzed for radioactivity as described in Experimental Procedures. (A) growing cells: cells 52P-orthophosphate and 1 Ci/ml YSlabeled with 10 &i/ml methionine. (O--Z) TCA-precipitable YG; (O-O) nonTCAprecipitable a2P; (A-A) TCA-precipitable 3*P after 5 min of incubation at 37°C with IO mg/ml RNAase A and 1 mM phenylmethylsulfonyl fluoride. (8) nongrowing cells: cells labeled with 60 $Xml 32P-otthophosphate and 6 &i/ml of YS-methionine. (m+) TCA-precipitable YS; (O--O) nonTCA-precipitable ,*P. TCA-precipitable 32P was not determined.

Cell 594

maps presented in Figure 2 unambiguously identify the major CHO protein which is purified by the copolymerization procedure as tubulin. Since the initiation of these studies, co-polymerization with nonradioactive microtubule proteins has been used to identify putative radioactively labeled tubulin in Aspergillus (Sheir-Neiss et al., 1976) and mammalian cell membranes (Bhattacharyya and Wolff, 1976). It is clear that the co-polymerization procedure selects the subclass of CHO tubulin molecules (4.1% out of 14%) which can repeatedly co-polymerize with porcine brain microtubule proteins. For physiological studies, such a selection process is both useful and potentially hazardous. It will eliminate protein which has been denatured during in vitro manipulations, and which could obscure correlations between biological processes and the physical or chemical state of tubulin. It is possible, however, that such a procedure makes a biased selection of tubulin within the cell, thus also obscuring correlations between biological processes and chemical properties of tubulin. In the present system, it seems improbable that the nonpolymerizable CHO tubulin represents a biologically significant class of inactive monomers or oligomers. The nonpolymerizable tubulin, approximately 65% of the total CHO tubulin in the extract, shows no sensitivity to cold or colchicine during the first cycle of polymerization (Table 1) and sediments as huge aggregates. Thus it is probable that this material represents tubulin which has been denatured in vitro. The strong tendency of tubulin to aggregate in an amorphous, noncolchicine-sensitive manner has been noted (Shelanski and Taylor, 1968). When this inactive aggregated material is excluded from the calculation and the purified porcine tubulin carrier is used as a standard to normalize for the loss of active CHO extract tubulin during the purification, 77% of the remaining CHO tubulin appears competent to co-polymerize and depolymerize repeatedly (Figure 3, Table 1). [We have recently found that in co-polymerization extracts prepared without the addition of 0.75 M NaCl (see Experimental Procedures), 70% of the CHO tubulin is present in a nonaggregated form and >75% of this nonaggregated tubulin is competent to co-polymerize repeatedly. The N site nucleotide associated with these CHO tubulin preparations is entirely rGTP.1 Thus although strictly speaking we have two classes of tubulin molecules, we have found no evidence for a large class of tubulin which is nonaggregated, and yet is incompetent to polymerize and depolymerize repeatedly in vitro. In fact, such a class may not exist. We cannot exclude the possibility, however, that tubulin may be subtly regulated in vivo in a

fu,o~ 0

4

8

12

16

TIME OF COLD CHASE

20

24

(hrs)

Figure 6. Turnover of the y-Phosphate of N Site GTP The percentage of the total 32P counts isolated from each time point in the figure which were not adsorbable to charcoal after hydrolysis with porcine brain microtubule protein (see ExperimentalProcedures)is shown. Cells were labeledwith IO &i/ml a2P and 1 &i/m Y3 (O-O), or 80 $2i/ml asp and 6 &i/ml 3% (e-0).

variety of ways, but that such regulation is obscured under the in vitro conditions of assembly. Previously, Borisy et al. (1975) have estimated that 50% of the soluble tubulin in porcine brain extracts is competent to polymerize by sedimentation assay. These investigators used a preliminary centrifugation of cell extracts in the cold before analysis, so that tubulin aggregates, even if initially present, were not observed. If the inactive aggregated tubulin in the CHO extracts does represent a specific class of tubulin molecules within the cell which does not interact with the general tubulin pool, then all the results presented here would have to be interpreted as pertaining to only a subclass of the total tubulin. The exact molecular relationship between CHO and porcine subunits during the procedures used here is unknown. The results, however, are most consistent with the formation of true co-polymers between the CHO and porcine subunits. This is suggested by the observation that a stable ratio of radioactivity to protein is reached and maintained after two cycles of assembly and disassembly. If the CHO and porcine subunits are segregating into separate polymers during assembly, this would require that the assembly reaction proceed to the same extent in the two distinct tubulin pools, which differ in concentration by as much as 1000 fold. This behavior would then differ from the strong concentration dependence noted for the in vitro assembly of brain microtubules (Borisy et al., 1975). In addition, the concentration of CHO tubu-

Physiological 595

Studies

of Tubulin

lin in our extracts (~10 kg/ml) is far below the reported critical concentration needed for assembly (0.2 mg/ml; Borisy et al., 1975). The dependence on carrier microtubule protein which the purification of CHO tubulin shows (Figure l), therefore, can be most simply explained by the interaction of CHO subunits with porcine subunits which are present at levels well above the critical concentration required for assembly.

site. The extreme difficulty in removing all of the E site GDP by gel filtration can be predicted from its tight binding constant (Silhavy et al., 1975; Dixon, 1976) and has been previously described for tubulin (Penningroth and Kirschner, 1977). It is also possible that the brain system differs from the fibroblast system, or that N site GTP is partially hydrolyzed to GDP during the ion-exchange procedure.

The N Site Nucleotide rGTP

Intracellular Turnover of CHO Tubulin and Exchange of N Site GTP

Bound to CHO Tubulin Is

There have been conflicting reports concerning the identity of the nucleotide bound to the N site of tubulin. Weisenberg, Borisy and Taylor (1966), Berry and Shelanski (1972), and Jacobs et al. (1974) have found both GDP and GTP bound to ion-exchange purified tubulin from brain. Others, such as Kobayashi (1975) and Penningroth et al. (1976), using protein purified by temperature-induced cycles of assembly-disassembly, have found only GTP on the N site. The present co-polymerization scheme presents a useful system for investigating the N site of tubulin because the introduction of a radioactive label into the N site nucleotide permits it to be readily distinguished from the cold nucleotide which is exchanged onto the E site during the microtubule purification. Radioactive nucleotide bound to the E site is removed by exchange with cold GTP during the three cycles of co-polymerization where 1 mM cold GTP is added at each disassembly step. It is apparent from the chromatograms shown in Figure 4 that all of the N site nucleotide bound to the purified CHO tubulin is ribo-GTP. The absence of any GDP further supports the notion that the nucleotide isolated originated from the N site, and not the E site, because GTP at the E site of tubulin is hydrolyzed to GDP during microtubule assembly in vitro (Kobayashi, 1975). Although it is possible that GDP present at the N site of tubulin in CHO cells is phosphorylated in vitro by extract enzymes and radioactive nucleotides, this is ruled out because the application of a cold chase to labeled cells, which quickly reduces the specific activity of the cellular phosphate pool, does not greatly affect the specific activity of the y-phosphate of N site GTP (Figures 5 and 6). We therefore conclude that tubulin in CHO cells probably contains only ribo-GTP at the N site. The stoichiometry determinations presented in Table 2 indicate that approximately 1 mole of rGTP is bound per mole of CHO tubulin. Since the purification of tubulin by ion-exchange chromatography may not completely remove GDP from the E site, it is possible that all of the GDP which other investigators have observed tightly bound to ion-exchange purified tubulin is actually bound at the E

It has not been possible to remove the N site nucleotide without denaturing tubulin (Weisenberg et al., 1968), nor does it apparently exchange through repeated cycles of assembly and disassembly with nonhydrolyzable GTP analogs as shown by Penningroth et al. (1976). What is not known is whether it exchanges readily in vivo either due to specific conditions or aided by some enzymatic activity. The more rapid loss of radioactivity from the N site GTP, as compared with that of tubulin during application of a cold chase to the labeled CHO cells, suggests that this nucleotide can exchange within cells. Figure 5 shows that in growing cells, tubulin decays with a half-life of 48 hr, while in 32P-arrested cells, this time is reduced to 34 hr. The radioactivity in the N-site GTP on CHO tubulin decays considerably faster than radioactivity in tubulin, having a half-life of 15 and 9 hr in growing and nongrowing cells, respectively. Fine and Taylor (1976) have reported a half-life of 32 hr for tubulin in nongrowing confluent 3T3 cells, and this value was increased to >72 hr in growing cells. Experiments investigating the specific activity of the free methionine pool (B. M. Spiegelman, unpublished data) and the 32P covalently associated with tubulin (Figure 5a) indicate that the specific activity of both pools is reduced by at least 80% after 3 hr of cold chase. Differences in the rate of dilution of the precursor pools and reutilization of 32P04- or 35S-methionine, however, could affect the exact turnover rates to some extent. Although the preferential turnover of the N site GTP above that of tubulin can be explained by simple exchange of radioactive GTP, it is possible that this preferential decay of the radioactivity from the nucleotide is facilitated by specific enzymes which remove the N site GTP from tubulin. No such activity, however, has been reported. Only a small part of the loss of 32P counts during the cold chase is caused by y-phosphate hydrolysis of the N site GTP (15% of the total 32P counts in 24 hr of cold chase; Figure 6) as determined by measuring the y-phosphate independently. During the same 24 hr period, the observed loss of 66% of the 32P

Cdl 596

radioactivity in growing cells, and 84% in the nongrowing cells during the cold chase (Figure 5), must be accounted for either by hydrolysis of the (Y--, p- and y-phosphates or more probably, by exchange of the complete molecule of GTP. The actual rate of exchange of GTP can be calculated if it is assumed that 32P in GTP is lost when tubulin decays by an irreversible process, such as denaturation, when GTP exchanges off native tubulin and when the y-phosphate of GTP on tubulin is hydrolyzed in situ (see Appendix). Half-times for the intrinsic exchange rate of GTP are calculated to be 15 hr for nongrowing cells and 33 hr for growing cells. Thus although detectable levels of N site GTP exchange occur within CHO cells, the rate is remarkably slow when compared to a generation time of 17 hr for growing cells. The previous inability of investigators to observe exchange of this nucleotide in vitro seems to reflect accurately its exchange behavior in vivo. The reason for the faster exchange rate of the N site GTP in 32Parrested cells than in growing cells is not known, but it could be related to the physiology of such an abnormal state. Hydrolysis of N Site GTP Although measurements of N site GTP show exchange at a very slow rate (having a half-life of 33 hr in growing cells), this represents the total 32P radioactivity lost from the N site and does not distinguish the rates at which radioactivity is lost from the three separate phosphate groups in GTP. The presence of nucleoside diphosphate kinase (NDP-kinase) activities in purified brain microtubule preparations which can phosphorylate GDP to GTP bound to the E site of tubulin suggests that similar NDP-kinase activities and phosphatases could act at the N site of tubulin and regulate tubulin function. We have tested this notion by measuring the preferential turnover rate of the yphosphate of N site GTP. By isolating radioactive GTP bound to tubulin from cells which had been exposed to a cold chase for various lengths of time, we were able to examine only that GTP which had not exchanged off tubulin during the cold chase. Thus the y-phosphate turnover of N site GTP could be estimated without bias due to exchange. It is evident in Figure 8 that the y-phosphate represents a smaller percentage of total N site GTP counts with progressively longer periods of cold chase, indicating that some preferential turnover of the y-phosphate does occur in CHO cells. The rate, however, is quite slow, having a half-time of 24 hr in both growing and 32P-arrested cells. The fact that y-phosphate turnover can be detected in nucleotide molecules which have remained bound to tubulin throughout the chase

period suggests that such hydrolysis does not require nucleotide exchange. These data, which indicate that the N site GTP exchanges only slightly more slowly (half-time of 33 hr in growing cells) than it is hydrolyzed (halftime of 24 hr in growing cells), argue against yphosphate turnover in situ as a general mechanism for regulating tubulin assembly within cells. It is possible, however, that a particular subclass of tubulin molecules-for example, those in a polymerized state-could experience a greatly reduced exchange rate. In such a case, it would still be theoretically possible to regulate some aspect of microtubule function by y-phosphate turnover. Could hydrolysis or exchange of GTP at the N site of tubulin be involved in microtubule assembly? Despite its lack of a role in vitro, such processes could, in principle, be involved in vivo. The surprisingly slow rate of exchange and hydrolysis at this site, as presented, indicates that if such processes are correlated with microtubule assembly, each tubulin molecule participates in assembly considerably less often than once per cell generation. The extent of assembly and disassembly reactions of tubulin in cells is not known. lnoue and Sato (1967) have proposed a model in which microtubules are in a constant assembly-disassembly equilibrium, but little experimental evidence is available in cultured cells. lmmunofluorescent evidence of a disassembly-assembly process has been presented for the interphase-mitosis conversion in the cell cycle, as the cytoskeletal microtubules disappear and are replaced by spindle tubules (Fuller et al., 1975). CHO cell morphology, which has been shown to be dependent upon microtubule function (Hsie and Puck, 1971), varies throughout the cell cycle (Porter et al., 1974), and this probably reflects the dynamic nature of the microtubule assembly-disassembly process. Similarly, the directionality of movements of cultured cells on a substratum, which seems dependent upon micro tubules, normally changes several times in a cell cycle (Gail and Boone, 1970, 1971). Given that a large fraction (lo-40%) of CHO tubulin exists in the polymer form. intracellularly at the cell densities reached in these experiments (Rubin and Weis, 1975), the demonstration that half the y-phosphate of the N site GTP preferentially turns over in 24 hr and exchanges in 33 hr is probably consistent with a regulatory role for this site only if microtubule assembly-disassembly occurs very infrequently or in a small fraction of tubulin subunits. We consider it most probable that microtubule assembly is a dynamic process, but that y-phosphate hydrolysis or exchange of the total N site GTP has no role in regulating assembly. On the basis of the very slow intracellular ex-

Physiological

Studies

of Tubulin

597

change and hydrolysis rates, we propose that the N site GTP has a solely structural, as opposed to a regulatory, role in the function of the tubulin molecule. Nucleotide co-factors, whose hydrolysis is not an obligatory part of protein function, have been described in actin (Cooke and Murdoch, 1973), ribosomal proteins involved in some protein synthesis reactions (Haselkorn and RothmanDenes, 1973) and the E site of tubulin (Penningroth et al., 1976). For reasons which are not understood, however, hydrolysis and exchange of these nucleotides do normally occur during protein synthesis reactions and during actin and tubulin assembly. Thus as far as is presently known, the tubulin N site GTP may be unique among nucleoside triphosphate protein ligands in its apparent lack of reactivity. Since, however, these studies were performed on a fibroblast cultured cell line, which would not be expected to display specialized microtubule functions such as are believed to occur in nerve, pigment and lymphocyte cells (Soifer, 1976), a final conclusion must await studies of N site nucleotide in other cell types. The co-polymerization system described here should be useful in this regard. Experlmental Cell Culture

Procedures and Radlolabellng

Chinese hamster ovary cells K, clone (CHO-K,; Puck, Cieciura and Robinson, 1958) were a gift from Dr. Arthur Pardee (Sidney Farber Cancer Center, Boston, Massachusetts). They were grown at 37°C in monolayer culture in nutrient mixture F-10 buffered with 25 mM HEPES, containing 10% calf serum, 5% fetal calf serum, 40,000 units/l penicillin and 40,000 pg/I streptomycin [all from Grand island Biological Co. (Gibco). Grand Island, New York]. Cell stocks were maintained in 75 cm2 flasks (Falcon Plastics, Division of Bioquest, Oxnard, California) and were harvested by trypsinization. Cells were radioactively labeled by plating 2-8 x IO5 cells on 100 mm diameter plastic petri dishes (Falcon Plastics) in 10 ml of F-10 medium. After 12 hr, this medium was replaced with phosphate-free or methionine-free F10 medium (Gibco) supplemented with serum and antibiotics as above. The desired level of 32P-orthophosphate (carrier-free, NEX-054; New England Nuclear, Boston, Massachusetts) was added to the phosphate-free medium without the addition of carrier phosphate. 35S-methionine (100-400 Ci/mmole. NEGOOQH; New England Nuclear) was added to methionine-free medium with 1.2 mg/l of cold methionine. Double label experiments were performed in phosphate-free medium.

Preparatkn

of Porcine

Braln

Mkrotubule

Proteln

Microtubule protein was purified from porcine brain by two cycles of polymerization and depolymerization according to the method of Shelanski et al. (1973). as modified by Weingarten et al. (1974). The protein was stored at -20°C in 8 M glycerol and purified through a third cycle of polymerization immediately before use. When a pellet of microtubules was required (see below), the protein in 8 M glycerol was diluted 1:1 with polymerization buffer [PB; 0.1 M Mes, 0.5 mM MgCI,, 2 mM EGTA, 0.1 mM EDTA, 1 mM &mercaptoethanol (pH 6.5 with NaOH), and aggregates in the stored, two cycle-purified protein were removed by centrifugation at 4°C before the microtubules were assembled and sedimented.

Co-Polymerlzatlon of Labeled Mkrotubule Protelns

CHO Extracts

wlth

Porclna

After the cells (typically six 100 mm petri dishes) were labeled for the desired period of time, the radioactive medium was removed and the cells were rinsed twice with ice-cold phosphate-buffered saline (PBS). After harvesting the cells by scraping the dishes with a rubber serum stopper in a small volume of PBS, the cells were collected by centrifugation and resuspended in 1.5 ml of ice-cold PB. The cell suspension was homogenized on ice in a 3.0 ml glass homogenizer using a teflon pestle (Kontes Glass Co., Vineland, New Jersey) driven by an Omnimixer (Ivan Sorvall Co., Norwalk, Connecticut). Twenty passes of the pestle in 1.5 min yielded a preparation which was approximately 95% nuclei and 5% intact cells. Intact cells, nuclei and large debris were removed by centrifugation at 12,000 x g for 10 min at 4°C. Since RNA has been shown to inhibit microtubule assembly (Bryan, Nagle and Donges, 1975), the supernatant was incubated with 10,000 units of RNAse A or T, for 5 min at 37°C before mixing with a pellet of about 15 mg of porcine microtubules (see above). After solubilization of the microtubule pellet with the cell extract in 1.5 ml PB, the mixture was made 0.75 M in NaCl to dissociate oligomeric forms of tubulin to 6 S tubulin subunits (Weingarten et al., 1974). This mixture was then sedimented at 120,000 x g for 90 min at 4”C, and the supernatant was desalted on a 1 .O x 20 cm Biogel P-6 column equilibrated with PB. The desalted solution was made 1 mM in GTP and warmed for 20 min at 37”C, and microtubules were isolated by centrifugation at 100,000 x g for 40 min at 25°C. The supernatant was discarded, and the microtubule pellet was carefully rinsed with warm PB + mfvl GTP to ensure complete removal of the supernatant. The pellet was resuspended in ice-cold PB + 1 mM GTP, chilled on ice for 30 min to depolymerize microtubules and centrifuged at 100,000 x g for 40 min at 4°C to remove cold-insensitive aggregates. The pellet was discarded, the supernatant was transferred to a new tube and the temperature-dependent cycling procedure was repeated twice more for a total of three rounds of polymerization/ depolymerization. After the third round of polymerization/depolymerization, tubulin was separated from microtubule accessory proteins by chromatography on a 0.5 x 2.0 cm phosphocellulose column, as described by Weingarten et al. (1975). Experiments in which 4 M glycerol were added to PB throughout the co-polymerization purification did not markedly improve yields or purity of tubulin, so it was not used in the experiments described here.

Proteln

Determlnatknr

and Sclntlllatlon

Countlng

Protein was determined by the method of Lowry et al. (1951) using bovine serum albumin (BSA) as a standard. Radioactivity was determined by adding IO-200 ~1 of the sample to 4.5 ml of a toluene-ethanol mixture (12:7) containing 0.4% 2,5-diphenyloxazole (PPO), 0.01% 1,4-bis[2(5-phenyloxazolyl)]benzene (Winkler and Wilson, 1968), and counting in a Packard 2 channel scintillation spectrometer.

SDS-Polyacrylamlde

Gel Electrophoreslr

and Fluorography

Sodium dodecylsulfate (SDS)-polyacrylamide gel electrophoresis was performed as described by Laemmli (1970) on 10% polyacrylamide slab gels. Proteins were stained with Coomassie blue and destained by diffusion in 5% methanol, 7.5% acetic acid. Gels were impregnated with PPO, according to the fluorographic modification of autoradiography (Bonner and Laskey, 1974), and the gels were subsequently dried under vacuum. Kodak RP Royal X-omat film was exposed to the dried, PPO-impregnated gels at -70°C.

Ouantitatlon

of Radloactlvlty

In Polyacrylamide

Gels

Fluorographs can be unreliable for quantitating radioactivity in bands in a gel (Laskey and Mills, 1975); thus bands which were localized by aligning the dried gel with the fluorograph were removed from the dried gel with a razor blade and placed directly into a scintillation vial with 4.5 ml of the toluene-ethanol scintilla-

Cell

tion cocktail and counted. The radioactivity detected by this method was shown to be linear with respect to disintegrations in a band between 50 and 5000 dpm. The overall efficiency of detection of electrophoretically pure polypeptides (calibrated with “C-BSA or V-actin) was about 35% by this method.

Measurement 01 Turnover of N Slte Nucleotlde

Rate of CHO Tubulin

concentration of tubulin,

of tubulin-GTP then schematically:

k,L

T-G '

and Exchange

CHO cells were grown in phosphate-free F-10 medium containing both 3?3-methionine and 52P-orthophosphate for 36 hr. This medium was removed, the cells were rinsed 4 times with warm sterile PBS, and fresh complete nonradioactive medium was added to the cells for various periods of time. CHO microtubule proteins were isolated by the co-polymerization method, and the tubulin was purified by phosphocellulose chromatography (see above). Phosphocellulose-purified tubulin was divided into two equal aliquots for analysis. Bound nucleotide was released by precipitating tubulin with 20% TCA. In some experiments, nucleotide was released by ethanol or heat precipitation. After removal of the precipitate by centrifugation. BSA at 0.5 mg/ml was added to the supernatant. followed by a second centrifugation step to ensure complete removal of tubulin. The microtubule pellet obtained from the first TCA precipitation was washed twice with ice-cold 10% TCA to remove any trapped low molecular weight material. The washed pellets was dissolved in 0.2 M NaOH and neutralized with acetic acid. This material and the supernatants remaining after two precipitations were analyzed for both YS and 3zP by scintillation counting. In all cases, the duplicate aliquots agreed within 10% of each other for both isotopes.

Analysis of Intracellular y-Phosphate Hydrolysis Uslng Porcine Brain Mlcrotubule Proteln GTPase

If T.G is the initial T is the concentration

complex

'

and

T + GTP

k-2

T-GDP + Pi degraded tubulin

+

GTP

During cold chase, we assume that k-,, k-, and k-S are negligible for GTP, and that the specific activities of the (x-, pand y-phosphates of GTP are.identical. The experimentally available rates are: -the molar loss of YS-labeled tubulin by degradation: z=

k,[T.G]

This is equal to the molar

loss of 32P-labeled

GTP by this process:

of N Slte GTP

32P-labeled nucleotide was released from phosphocellulose-purified tubulin of the cold-chased CHO cells by heating at 100°C for 5 min. Aggregated protein was removed by centrifugation at 12,000 x g for 20 min at 4°C. 500 ~1 of this supernatant were made 0.1 mM in GTP. and 100 ~1 of three cycle-purified porcine brain microtubule proteins were added to a final concentration of approximately 2.0 mg/ml. The mixture was incubated at 37°C for 2 hr. After this incubation, the sample solutions were made IO mM in NaH,P04 and 6% in perchloric acid, and the resulting precipitate was removed by centrifugation. The supernatant was removed, and the GDP and inorganic phosphate were separated by the addition of 3 mg of acid-washed chacoal and incubation on ice for 5 min. The charcoal was sedimented by centrifugation at 10,000 x g for 10 min. This treatment removed >96% of the OD at 260 nm in control samples containing only GDP. Aliquots of the supernatant of the centrifugation step directly before the addition of the charcoal, and the supernatant after the addition and sedimentation of the charcoal, were analyzed by liquid scintillation counting, giving measures of the total nucleotide radioactivity and the y-phosphate radioactivity, respectively, in each sample.

-the

molar

d(y-%) ~-

= k3 [y-PO,]

dt

-the g,

loss of +*POr

total

molar

by the

and GTP

due to hydrolysis:

= ks [T. G]

loss of 32P04 in GTP.

processes

exchange

of tubulin given

where

+ b)P

y-PO,

hydrolysis

below:

where k, is the rate constant contains three phosphates, dP dl = (k, + ;

degradation,

for GTP exchange. Since every GTP [P] = 3 [T.G]. and (4) can be rewritten:

= bP

1 k, = k, + k2 + - k3.

3 Since all of these are first-order processes, can be expressed in terms of its half-time:

each

rate

constant

Appendix Calculation of the lntrlnslc Exchange Rate of N Slte GTP from % and 3zP Decay Rates (Figure 5) and an Independent Measurement 01 Turnover of the y-PO, on the N Slte GTP F&w3 6) We assume first, that the primary mode of 35S decay is an irreversible process such as protein degradation, and second, that there are three modes of ‘*P nucleotide decay: -The release of 32P nucleotide in the first case above. This nucleotide quickly equilibrates with the cold GTP pool. This process occurs at the same rate as the loss of ‘?S in the first case above. -The exchange of szP N site GTP with nonradioactive GTP from the cold pool. -The specific turnover of the y-phosphate of N site GTP bound to tubulin.

(6) From the half-time of Yi decay (T,), the half-time of the yphosphate hydrolysis rate (TV) and the half-time for the overall N site 32P decay rate (~~3. one can calculate a half-time for the intrinsic exchange rate of GTP (T?). For growing cells where 7, is 46 hr, 73 is 24 hr and 7T is 15 hr, T* is calculated to be 33 hr. For nongrowing cells where T, is 34 hr, TV is 24 hr and 7T is 9 hr, 72 is calculated to be 15 hr.

Acknowledgments We wish to acknowledge General Medical Sciences supporting this work. Received

June

29, 1977;

the USPHS, the and the American

revised

National Cancer

July 29, 1977

Institute Society

of for

Physiological

Studies

of Tubulin

599

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

nucleotide

References Ames, B. N. (1966). Assay of inorganic phate and phosphatases. In Methods York: Academic Press), pp. 115-116.

8. and Wolff, J. (1976). Nature 264, 576-577.

Polymerization

Bonner, W. M. and Laskey, R. A. (1974). for tritium-labeled proteins and nucleic gels. Eur. J. Biochem. 46, 63-88. Borisy, G. G., Olmsted, Microtubule assembly

Laemmli, assembly

Wilson, Proc.

Bryan, J., Nagle. tubulin assembly required protein.

L. (1971). Nat. Acad.

Olmsted, Biochem.

Dixon, H. B. macromolecule

B. W. and Dbnges, K. Ii. (1975). Inhibition of by RNA and other polyanions: evidence for a Proc. Nat. Acad. Sci. USA 72, 3570-3574.

Eipper, B. A. (19741. Chem. 249, 1407-1416.

Properties

of

rat

brain

tubulin.

a

J. Biol.

Fine, R. E. and Taylor, L. (1976). Decreased actin and tubulin synthesis in 3T3 cells after transformation by SV40 virus. Exp. Cell Res. 702. 162-166. Fuller, G. M., Brinkley, B. R. and Boughter, J. M. (1975). nofluorescence of mitotic spindles by using monospecific bodies against bovine tubulin. Science 187, 946-950. Gail, M. H. and Boone, C. W. (1970). fibroblasts in tissue culture. Biophys. Gail, M. fibroblast

Ii. and motility.

The locomotion J. 10, 960-993.

Boone, C. W. (1971). Effect Exp. Cell Res. 65, 221-227.

Immuanti-

of mouse

of colcemid

on

Goldman, R., Pollard, T. and Rosenbaum, J., eds. (1976). Microtubules and Related Proteins, Cold Spring Harbor Conferences on Cell Proliferation, 3 C (Cold Spring Harbor, New York: Cold Spring Harbor Laboratory). Gonzales. L. W. and Geel, S. E. (1975). Thin-layer chromatography of brain adenosine nucleotide and nucleotides and determination of ATP specific activity. Anal. Biochem. 63, 400413. Haselkorn, R. and Rothman-Denes, L. B. (1973). Protein synthesis. Ann. Rev. Biochem. 42,397-436. Hsie, A. W. and Puck, T. T. (1971). Morphological transformation of Chinese hamster cells by dibutyryl adenosine 3’.5’-monophosphate and testosterone. Proc. Nat. Acad. Sci. USA 68, 358-361. Ino&, S. and Sato, H. (1967). Cell motility of molecules. J. Gen. Physiol. 50, 259-292. Jacobs,

M. H.,

Smith,

H.

and

Taylor,

by labile E. W.

(1974).

association Tubulin:

J. B. and Borisy. 42, 507-540.

G. G. (1973).

Microtubules.

Ann.

Rev.

A. (1973). The effect of inhibitors on dibutyryl cyclic AMP mediated of Chinese hamster ovary cells in Commun. 50, 566-573.

Penningroth, S. M. (1977). Roles of ligands in microtubule assembly in vitro. 1. The mechanism of nucleotide action. 2. The site of colchicine inhibition. Ph.D. Thesis, Princeton University, Princeton, New Jersey. Penningroth. S. M. and binding and phosphorylation Mol. Biol.. in press.

of actin with 72, 3927-3932.

F. (1976). Removal of a bound ligand from by gel filtration. Biochem. J. 759, 161-162.

N. J., Farr. A. L. and Randall, R. J. with the Folin phenol reagent. J.

Patterson, D. and Waldren. C. of RNA and protein synthesis morphological transformations vitro. Biochem. Biophys. Res.

Are cytoplasmic microtubules Sci. USA 68, 1762-1766.

R. and Murdoch, L. (1973). Interaction of adenosine triphosphate. Biochemistry

of tubulin-bound guaassembly. J. Biochem.

Luduena, R. F. and Woodward, D. 0. (1973). Isolation and partial characterization of a- and /3-tubulin from outer doublets of sea urchin sperm and microtubules of chick embryo brain. Proc. Nat. Acad. Sci. USA 70, 3594-3596.

C. (1974).

Cleveland, D. W., Fischer, S. G.. Kirschner, M. W. and Laemmli, U. K. (1977). Peptide mapping by limited proteolysis in sodium dodecyl sulfate and analysis by gel electrophoresis. J. Biol. Chem. 252, 1102-1106. Cooke, analogs

J. Mol. Biol.89,455-468.

U. K. (1970). Cleavage of structural proteins during the of the head of bacteriophage T4. Nature 227, 660-665.

Lowry, 0. H., Rosebrough, (1951). Protein measurement Biol. Chem. 193,265-275.

A film detection method acids in polyacrylamide

Borisy, G. G., Marcum. J. M., Olmsted, J. B.. Murphy, D. B. and Johnson, K. A. (1975). Purification of tubulin and associated high molecular weight proteins from porcine brain and characterization of microtubule assembly in vitro. Ann. NY Acad. Sci. 253, 107-132. Bryan, J. and heteropolymers?

activity.

R. and Pastan, I. (1971). Restoration of characteristics of normal fibroblasts in with adenosine-3’,5’-cyclic monophosProc. Nat. Acad. Sci. USA 68, 425-429.

Laskey, R. A. and Mills, A. D. (1975). Quantitative film detection of 3H and “C in polyacrylamide gels by fluorography. Eur. J. biochem. 56, 335-341.

of mem-

J. B., Marcum, J. M. and Allen, in vitro. Fed. Proc. 33, 167-174.

and enzymic

Kobayashi. T. (1975). Dephosphorylation nosine triphosphate during microtubule 77,1193-1197.

phosphate, total phosin Enzymology, 8. (New

Berry, R. W. and Shelanski, M. L. (1972). Interaction of tubulin with vinblastine and guanosine triphosphate. J. Mol. Biol. 71, 71-60. BhattacharYya. brane tubulin.

binding

Johnson, G., Friedman, several morphological sarcoma cells treated phate and its derivatives.

Kirschner, M. W. (1977). in microtubule assembly

Nucleotide in vitro. J.

Penningroth, S. M., Cleveland, D. W. and Kirschner, M. W. (1976). In vitro studies of the regulation of microtubules. In Microtubules and Related Proteins, Cold Spring Harbor Conferences on Cell Proliferation, 3 C, R. Goldman, T. Pollard and J. Rosenbaum, eds. (Cold Spring Harbor, New York: Cold Spring Harbor Laboratory), pp. 1233-1257. Porter, K. R., Puck, electron microscope on Chinese hamster

T. T., Hsie, A. W. and Kelley, D. (1974). An study of the effects of dibutyryl cyclic AMP ovary cells. Cell 2, 145-162.

Puck, T. T., Cieciura, S. J: and Robinson, A. (1956). Genetics of somatic mammalian cells. Ill. Long-term cultivation of euploid cells from human and animal subjects. J. Exp. Med. 108,945-956. Rice, R. H. and Means, proteins in vitro. J. Biol.

G. E. (1971). Radioactive Chem. 246, 631-832.

labeling

of

Rubin. R. W. and Weiss, G. D. (1975). Direct biochemical measurement of microtubule assembly and disassembly in Chinese hamster ovary cells. The effect of intercellular contact, cold, Da and N’, 02-dibutytyl cyclic adenosine monophosphate. J. Cell. Biol. 64, 42-53. Sheir-Neiss. G., Nardi, R. V., Gealt, M. A. and Morris, N. R. (1976). Tubulin-like protein from Aspergihs nidulans. Biochem. Biophys. Res. Commun. 69, 285-290. Shelanski, M. L. and Taylor, E. W. (1966). protein subunit of central pair and outer-doublet sea urchin flagella. J. Cell Biol. 38, 304-315. Shelanski, M. L., Gaskin, assembly in the absence Sci. USA 70, 765-766.

Properties of the microtubules of

F. and Cantor, C. R. (1973). Microtubule of added nucleotides. Proc. Nat. Acad.

Silhavy, T. J., Szmelcman, S., Boos, On the significance of the retention Nat. Acad. Sci. USA 72, 2120-2124.

W. and Schwartz, M. (1975). of ligand by protein. Proc.

Snyder, J. A. and McIntosh, J. R. (1976). Biochemistry and physiology of microtubules. Ann. Rev. Biochem. 45, 699-727.

Cell 600

Soifer, D., ed. (1975). The Ann. NY Acad. Sci. 253.

biology

of cytoplasmic

miccrotubules.

Weingarten, M. D., Suter, M. M., Littman, D. R. and Kirschner, M. W. (1974). Properties of the depolymerization products of microtubules from mammalian brain. Biochemistry 73. 55295526. Weingarten, M., Lockwood, A. H., Hwo. S. Y. and Kirschner, M. W. (1975). A protein factor essential for microtubule assembly. Proc. Nat. Acad. Sci. USA 72, 1855-1862. Weisenberg, R. C. (1972). Microtubule formation in vitro in solutions containing low calcium concentrations. Science 777, 11041105. Weisenberg, Ft. C., Borisy, G. G. and Taylor, E. W. (1968). The colchicine binding protein of mammalian brain and its relation to microtubules. Biochemistry 7, 4466-4479. Wiche, G. and Cole, R. D. (1976). Reversible in vitro polymerization of tubulin from a cultured cell line (rat glial cell clone C 6). Proc. Nat. Acad. Sci. USA 73, 1227-1231. Winkler, H. H. and Wilson, T. H. (1966). coupling in the transport of P-galactosides J. Biol. Chem. 241, 2200-2211.

The role of energy by Escherichia co/i.