THE JOURNAL OF BIO~~ICAL CHEMISTRY Vol. 253, No. 8, Issue of April 25, pp. 2825-2833, 1978

Printed in U.S.A.

Characterization of Microtubule Protein Oligomers by Analytical Ultracentrifugation’

(Received for publication, July 27, 1977)

J. MICHAELMARCUM~ AND GARYG. BORISY~

From the Laboratory of Molecular Biology, University of Wisconsin, Madison, Wisconsin 53706

Samples of microtubule protein prepared by repeated cycles of assembly-disassembly were examined at low tem- peratures by sedimentation velocity ultracentrifugation. Sedimenting boundaries corresponding to the 6 S tubulin subunit as well as to two oligomeric species with s!~,~ values of 18.6 S and 30.6 S were observed. The 30 S to 6 S mass ratio varied with total protein concentration, suggesting that a concentration-dependent equilibrium exists between 6 S and oligomeric species of tubulin. A study of the effects of pH on the mass distribution among the species demonstrated that the 6 S species was favored at low pH values (5.8 to 6.51, the 30 S oligomer was favored at moderate pH values (6.5 to 7.41, and the 18 S oligomer was formed in increasing proportions at higher pH values (7.4 to 8.2). Incubation of purified microtubule protein solutions with increasing con- centrations of NaCl initially favored conversion of the 30 S oligomer to the 18 S species with further increases in salt concentration resulting in the dissolution of both the 30 S and 18 S oligomers. The depolymerizing effects of high salt concentrations were substantially reversible, providing fur- ther evidence for 6 S-oligomer equilibria. The manipulation of the solution variables of pH and ionic strength in a systematic fashion led to the construction of a “phase diagram” for the microtubule protein species which pro- vided a relatively complete description of the mass distribu- tion among the 6 S, 18 S, and 30 S species over a range of physiological and near-physiological solution conditions.

Since the first report of ring-shaped oligomers in prepara- tions of microtubule protein’ by Borisy and Olmsted (1972), rings and their role in the self-assembly of microtubules have been the subject of intense investigation (Weingarten et al., 1974; Rebhun et al., 1974; Kirschner et al., 1974, 1975; Frigon and Timasheff, 1975). Various laboratories have described

* This work was supported by Grant GB-36454 from the National Science Foundation and Grant GM-21963 from the National Insti- tutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

t Recipient of National Institutes of Health Predoctoral Trainee- ship GM-01874. Present address, Department of Cell Biology, Baylor College of Medicine, Houston, Texas 77025.

P To whom correspondence should be addressed. 1 Microtubule protein refers to tubulin and associated proteins

which are co-purified by the temperature-dependent assembly-disas- sembly procedure (Borisy et al., 1975).

rings which differ in their appearance in the electron micro- scope, in their hydrodynamic characteristics, and in their content of tubulin and nontubulin proteins. Several structural models have been proposed to account for the observations.

Weingarten et al. (1974) and Scheele and Borisy (1976) have observed a 36 S oligomeric species in samples of microtubule protein prepared by the glycerol purification method of She- lanski et al. (1973). This species was reported to be a double- walled ring (Weingarten et al., 1974) and to require the presence of a nontubulin protein of molecular weight about 68,000 for its formation (Weingarten et al., 1975). Weingarten et al. (1974) have proposed a structural model for the 36 S ring which accounts for its electron microscopic appearance and sedimentation coefficient in terms of the positions of 42 tubulin protomers, each of molecular weight 55,000, in a single layer of two concentric rings. Frigon and Timasheff (1975) employed tubulin purified by ion exchange chromatography and found an oligomer species of s&,w = 42 f 2 S under conditions of very high Mg*+ concentration. This oligomer, which contains no nontubulin proteins, was also reported to correspond to a double-walled ring.

Our own preparations of microtubule protein, prepared by a reversible assembly procedure not involving glycerol (Borisy et al., 1975) have been observed to contain single-walled rings (Olmsted et al., 1974) and show two oligomer peaks in the ultracentrifuge. The sedimentation coefficient of the faster sedimenting species has been reported as 30 S (Marcum and Borisy, 1974; Scheele and Borisy, 1976). The observation of two oligomeric species has also been made in the preparations of protein prepared by the glycerol procedure of Shelanski et al. (19731, Scheele and Borisy (19761, and Doenges et al. (1976). In addition to the 36 S species already considered to correspond to a double-walled ring-shaped structure, a 20 S species was described and this species was reported to correspond to a single-walled ring (Doenges et al., 1976). Thus, rings have been reported as being single-walled or double-walled, with or without nontubulin proteins, and coming in a variety of sizes, including species of 20 S, 30 S, 36 S, and 42 S. The differences among these species appear to be real and dependent upon the conditions for preparing the microtubule protein.

The present study was undertaken to characterize the oligomeric species observed in our preparations, to determine under what conditions each of the two species predominates, and to determine the mechanism of their formation. A second report (Vallee and Borisy, 1978) will examine the effect of nontubulin protein components on ring formation and on the hydrodynamic properties of the ring. A third report (Scheele

2825

2826 Centrifugation of Microtubule Protein Oligomers

and Borisy, 1978) investigates the structure of the rings by electron microscopy of critical point-dried, metal-shadowed specimens. A final report (Marcum and Borisy, 1978) exam- ines the effect of pressure on ring assembly and addresses the question of the reversibility of ring formation.

EXPERIMENTAL PROCEDURES

Protein Preparation - Purified preparations of microtubule protein were obtained from porcine brain tissue by the reversible, tempera- ture-dependent assembly procedure (Borisy et nl., 1975). Purification was carried out through two cycles of assembly-disassembly and pelleted microtubules were frozen in liquid nitrogen and stored at -80”. For a particular experiment frozen pellets were thawed and resuspended at 5” in Buffer A (0.05 M 1,4-piperazinediethanesulfonic acid buffer containing 0.1 mM MgSG, and 1 mM GTP at pH 6.65 at 24”) unless otherwise noted. In some experiments 0.1 M buffer was used. The resuspended microtubule protein solutions were then generally purified by an additional cycle of assembly-disassembly to remove any inactive protein that might have been produced during storage. Protein concentrations were determined by the procedure of Lowry et al. (1951) using bovine serum albumin as a standard.

Analytical Ultracentrifugation - Purified microtubule protein was analyzed by sedimentation velocity using a Spinco model E analyti- cal ultracentrifuge (Beckman Instrument Co., Palo Alto, Calif.) equipped with schlieren optics and temperature control. Samples were sedimented in 12- or 30-mm double-sector cells, and the schlieren patterns were recorded on strips of Kodak Contrast Process Ortho film (Eastman-Kodak Co., Rochester, N.Y.). Distances on the film were measured with a Gaertner microcomparator (Gaertner Scientific Corp., Chicago, 111.). Sedimentation coefficients were corrected to standard conditions (20” and water) by using a correction factor containing the density and viscosity of the buffer and a value of d = 0.736 for the partial specific volume of tubulin (Lee and Timasheff, 1974). The mass fraction of each sedimenting component was determined from planimetry of tracings derived from the pho- tographs by taking the ratio of the area under a particular peak to the total area and corrected for radial dilution.

pH and Ionic Strength Determinations- For the initial pH experi- ments (see Fig. 5) protein solutions in 0.1 M Buffer A (pH 6.94 at 24” containing 0.1 mM MgSO, and 1 mM GTP) were directly titrated at 5” with either 1 N NaOH or 1 N HCl to the desired pH. The pH values of samples used in the construction of the “phase diagram” (Fig. 8) were adjusted by resuspending the frozen pellets in Buffer A at a variety of pH values and carrying out one purification cycle in the experimental buffer. The pH of a protein solution was measured at 24”. The ionic strength of a sample was calculated from the concen- tration of buffer used, the amount of NaOH added to adjust the buffer pH, the amount of added NaCl, if any, and from the Hender- son-Hasselbalch equation applied to the ionization properties of the buffer.

Electron Microscopy-Samples for electron microscopy were pre- pared by negative staining with 1% aqueous uranyl acetate and subsequently examined in a Philips 300 electron microscope.

RESULTS

Znitial Observations- When solutions of purified microtu- bule protein were examined at low temperatures (5-15”) in the analytical ultracentrifuge under solution conditions which would support rapid and extensive microtubule assembly at elevated temperatures (20-45”), two or three schlieren peaks were observed. Fig. 1 shows a schlieren photograph of a microtubule protein solution examined by ultracentrifugation at 5” in which three schlieren peaks were present. The large, slowest sedimenting boundary in Fig. 1 was universally ob- served under all solution conditions examined and corresponds to the colchicine-binding moiety previously described as the dimeric subunit of tubulin (Weisenberg et al., 1968). The two faster sedimenting boundaries shown in Fig. 1 represent the oligomers of microtubule protein previously described by Olmsted et al. (1974).

Protein Concentration-The sedimentation coefficients of

FIG. 1. Sedimentation pattern of a purified microtubule protein preparation. A microtubule protein solution at 5.0 mg/ml was prepared at pH 6.94 and was centrifuged at 5” at 59,780 rpm in a 12- mm double sector cell. This schlieren photograph was taken 6 min after reaching speed at a phase plate angle of 80”.

these species have been determined at a variety of protein concentrations and the results of one such experiment are shown in Fig. 2. The sedimentation coefficient of the faster sedimenting species displayed very marked concentration-de- pendent sedimentation. Extrapolation of the sedimentation coefficient of this peak to infinite dilution yielded an s!&,~ value of 30.6 2 1.4 S and we designate this as the 30 S species. The Sag,, value reported above was determined from the results of six concentration experiments similar to that shown in Fig. 2, and the presence or absence of the second, more slowly sedimenting species had no effect upon the sedimenta- tion coefficient of the 30 S species. Although both positive and negative values of the slopes of plots of the sedimentation coefficient of the slowest peak uersus protein concentration occurred (g = 0.003 f 0.012 (mg/ml)-‘) no readily definable trend has been observed. The sedimentation coefficient there- fore appears to be essentially independent of protein concen- tration over the range examined (0.43 to 12.0 mglml). Extrap- olation to infinite dilution yielded an s&, value of 5.9 -t 0.6 S and we designate this as the 6 S species. The intermediate sedimenting species displayed an intermediate level of concen- tration dependence of sedimentation rate, and extrapolation to infinite dilution yielded an s!&,, of 18.6 rf- 0.5 S, and we designate this as the 18 S species.2 The s&U, value obtained for this species was determined from the results of four concentra- tion experiments in which varying amounts of the 30 S species were present, and the sedimentation coefficient was found not to depend on the concentration of the 30 S oligomer over the range examined (10 to 40% 30 S). However, the concentration of the 18 S species was more sensitive to the effects of dilution than was the concentration of the 30 S species. As indicated in Fig. 2 by the absence of data at low protein concentrations, the 18 S species was not observed as a discrete schlieren peak below approximately 3.5 mg/ml, whereas the 30 S oligomer was still observed at the lowest protein concentration exam- ined (0.43 mg/ml). The differential stability of these species with respect to protein concentration was found for each of four concentration experiments, even though the initial con- centrations of the species varied over a wide range. Thus, the oligomeric microtubule protein species showed clear differ- ences in stability in addition to differences in sedimentation coefficients and concentration dependence of sedimentation.

Analysis of the concentration data by the standard relation

* We have previously referred to this species as a 20 S species (Marcum and Borisy, 1974; Scheele and Borisy, 1976). We rename it here in order to reflect increased precision in determining the value and to avoid confusion with another species whose sedimentation coefficient is closer to 20 S (Vallee and Borisy, 1978).

Centrifugation of Microtubule Protein Oligomers 2827

I I I I I

25

n 0 2o - x F 2 15

ul

6s . l - .

- . - . 5-

I 1 I I I I 0 2.0 4.0 6.0 6.0 10.0

CONCENTRATION (mg/ml)

FIG. 2. Concentration dependence of the sedimentation coeffr- cients of the microtubule protein oligomers. A solution of microtu- bule protein was prepared at pH 6.94 and high protein concentration. Dilutions from the concentrated stock were made just prior to analytical ultracentrifugation at 50,740 rpm at 5” in 12- to 30-mm double sector cells. Sedimentation coefficients were measured and corrected as described under “Experimental Procedures.” The lines were derived from least squares analyses of the sedimentation coefficients of the 30 S (W), 18 S (A), and 6 S (0) species at the indicated protein concentrations.

s = so (1 - gc), where s is the sedimentation coefficient at a given protein concentration (c); so is the sedimentation coeffr- cient at infinite dilution, and g is a factor which depends upon the size and asymmetry of the sedimenting species (Schach- man, 19591, resulted in values ofg = 0.054 r 0.006 ml/mg for the 30 S oligomer and g = 0.034 f 0.008 ml/mg for the 18 S species. The value of g reported here for the 18 S tubulin oligomer is similar to the value of g = 0.02 to 0.03 ml/mg reported by Weingarten et al. (1974) for the 36 S tubulin ring from glycerol-purified microtubule protein and somewhat larger than the value ofg = 0.018 ml/mg for the 42 S tubulin oligomer described by Frigon and Timasheff (1975). The value ofg for the 30 S species is, however, significantly larger than the values for the 18 S, 42 S, and 36 S oligomers and suggests that the 30 S particle is appreciably more asymmetric than the other oligomers (Schachman, 1959; Creeth and Knight, 1965).

The greater asymmetry of the 30 S species was also sug- gested by the shape of the 30 S boundary. Fig. 3 shows sedimentation patterns at early and later times during cen- trifugation of two solutions of microtubule protein. The first solution, shown in Fig. 3, a and b, contained only the 6 S and 30 S species, and at this rotor velocity the 30 S species showed a relatively symmetrical, hypersharp schlieren peak which retained its hypersharp character throughout the course of centrifugation. In contrast, the 18 S component, shown in Fig. 3, c and d, from a solution containing only 18 S and 6 S species displayed a more diffuse boundary which became markedly flattened as centrifugation continued. These observations are

consistent with the notion that the 30 S species is a more highly asymmetric structure with a larger excluded volume than the 18 S oligomer. It should also be noted that since the 30 S and 18 S species each constituted approximately half the protein present in the solutions and, since gel electrophoretic analyses have shown that these preparations contained 70 to 80% tubulin (Borisy et al., 1975), it is clear that both the 30 S and 18 S species must be tubulin-containing oligomers.

The effect of variations in the total protein concentration on the mass fractions of the 6 S and 30 S species was investigated and the results of one such experiment are shown in Fig. 4. In this experiment, protein at pH 6.7 was prepared at high protein concentration, and dilutions from this stock were made prior to analytical ultracentrifugation at 5”. Only the 6 S and 30 S peaks were present under these solution condi- tions; however, similar results have been obtained from mix- tures of the oligomeric species when the sum of the 18 S and 30 S mass fractions were plotted versus total protein concen- tration. The mass fraction measurements were corrected for radial dilution; however, no attempt was made to correct for the Johnston-Ogston effect. Since the sedimentation coefficient of the 6 S species was essentially invariant with protein concentration (Fig. 21, it is not expected that the Johnston- Ogston effect will be large for this system (Schachman, 1959).

As the protein concentration was decreased by dilution, the mass fraction of the 30 S oligomer decreased with a concomi- tant increase in the relative concentration of the 6 S species (Fig. 41. The mass fraction of the 30 S species was relatively stable above 4 mglml, while below this concentration it

FIG. 3. Schlieren photographs showing the effect of time of cen- trifugatidn on the boundary shapes of the 30 S and 18 S tubulin oligomers. A solution of microtubule protein at 5 mglml at pH 6.8 was sedimented at 42,040 rpm at 5” and photographed at a phase angle of 75” at 15 (a) and 59 min (b) after attaining speed. A second solution at 5 mg/ml prepared at pH 7.43 was incubated with 50 mM NaCl at 0” and centrifuged 1 h later at 42,040 rpm at 5”. Schlieren photographs were taken at a phase plate angle of 75” at 17 (c) and 57 min (d) after reaching speed.

2828 Centrifugation of Microtubule Protein Oligomers

1.00 1 I I I I I I \ \

i

: I

I I I I I I

0 2.0 4.0 6.0 6.0 10.0 12.0

TOTAL PROTEIN CONCENTRATION (mg/ml)

FIG. 4. The variation in the mass fraction of the 30 S and 6 S tubulin species with total protein concentration. A microtubule protein solution was prepared at pH 6.65 at a high protein concentra- tion. Dilutions from the concentrated stock solution were made prior to analytical ultracentrifugation at 42,040 rpm at 5”. Mass fraction measurements of the 30 S (W) and 6 S (0) species for the different concentrations were made from photographs taken at equivalent run times and were corrected as described under “Experimental Procedures.” - - -, extrapolation of the data to zero total protein concentration.

showed a sharp decrease to less than 0.18 at 0.43 mg/ml. The change in the mass fraction of oligomer upon dilution is suggestive of an equilibrium between the 6 S and 30 S species with increasing concentration favoring the oligomer as would be expected from La Chatelier’s rule. Additional evidence for an equilibrium association among the 6 S species and the oligomers comes from the shapes of the schlieren peaks. As shown in Fig. 3, c and d, the base-line between the 6 S and 18 S peaks was elevated and remained so throughout the experiment, as would be expected for an associating system (Gilbert and Gilbert, 1973). The elevation in base-line for conditions where only the 6 S and 30 S species were present (Fig. 3, a and b) was less evident, although still detectable, suggesting that the equilibrium constant for formation of the 30 S oligomer was considerably greater than for formation of the 18 S species. The pressure-dependent nature of the distri- bution of the 6 S and 30 S species also supports the idea that these species are in rapidly reversible association with each other, and this phenomenon is discussed more fully in an accompanying report (Marcum and Borisy, 1978).

pH and Ionic Strength -The response of the various species to changes in conditions of pH was examined and the results of one such study are shown in Fig. 5. Here, the mass fractions of the microtubule protein species are plotted as a function of the pH of the protein solutions measured at 5”. Low pH values favored the 6 S species, intermediate pH values favored the 30 S oligomer, while the 18 S species was formed in increasing proportions at higher pH values. At low pH (5.8), two schli- eren peaks were observed, the slower of which sedimented at 6 S and corresponded to the tubulin subunit, while the second peak had a sedimentation coefficient greater than 30 S. The rapidly sedimenting material also appeared to be very hetero- geneous and the shape of the boundary changed dramatically with minor pH variations. Electron microscopic examination of this material revealed bead-like structures of variable diameter as well as amorphous protein aggregates, which

perhaps were aberrant assembly products of tubulin. Above pH 5.8, the 30 S oligomer formed with a concomitant

reduction in the amount of both the 6 S and the rapidly sedimenting material. The 30 S species was observed over the remainder of the pH range examined, and its concentration displayed a rather broad maximum centered around pH 6.9. From pH 6.6 to 8.5, the 18 S oligomer appeared and became the dominant constituent in solutions above pH 7.5. Since the concentration of the 6 S subunit was relatively constant between pH 7 and 8, it is possible that the 18 S species formed by direct breakdown of the 30 S oligomer over this range. Thus, these results support the idea that equilibria exist among the microtubule protein species and suggest that proton binding is important as a control over the mass distribution of protein among the species.

The ionic strength of the microtubule protein solutions was also an important solution variable for determining mass distribution. Fig. 6 shows the results of an experiment in which a sample of microtubule protein at pH 6.7 measured at 5” was incubated with increasing amounts of NaCl in the cold prior to sedimentation analysis. The 30 S oligomer was rela- tively sensitive to small changes in NaCl concentration as evidenced by the consistent decrease in 30 S mass fraction with increasing salt concentration. At pH 6.7, the 30 S oligomer was completely eliminated by incubation in buffer containing 190 mM Na+. The initial appearance of the 18 S oligomer with increasing Na+ concentration was directly cor- related with the diminution of the 30 S oligomer. The 18 S concentration was maximal at a total Na+ concentration of approximately 210 mM. The results of this experiment indicate that ionic interactions are important for the integrity of the oligomers. Additionally, the correlation of increasing 18 S concentration with decreasing 30 S concentration, while the concentration of the 6 S species remained constant, again I I

pH at 5’C

FIG. 5. Sedimentation velocity analysis of the dependence of oligomer mass fractions on pH. Aliquots of a microtubule protein solution at 7.5 mglml prepared at pH 6.94 at 5” were titrated to the desired pH with either 1.0 N HCl or 1.0 N NaOH and were centri- fuged at 50,740 rpm at 5” in 1%mm double sector cells. Mass fractions ofthe high S to),30 S (W), 18 S (A), and 6 S (0) species were derived from the schlieren photographs and corrected for radial dilution.

Centrifugation of Microtubule Protein Oligomers 2829

0.2

150 200 Not CONCENTRATION (mM)

250

FIG. 6. Ultracentrifugal analysis of the dependence of oligomer mass fractions on Na+ concentration. A solution of microtubule protein at 5.6 mglml was prepared in Buffer A at pH 6.7 and aliquots of this solution were incubated with varying concentrations of NaCl for 2 h in the cold before centrifugation at 50,740 rpm at 5” in 12-mm double sector cells. The mass fractions of the 30 S (W, 18 S (A), and 6 S (0) components were determined and plotted uersus the total Nat concentration in each sample. - - -, extrapolation of the data based on the results of similar experiments at slightly different pH values (see Fig. 8).

supports the notion of a direct interconversion between the 30 S and 18 S species.

The depolymerizing effect of exposure to high salt concen- trations (-0.25 M NaCl) on the oligomers was essentially reversible. Fig. 7 shows the results of an experiment in which microtubule protein at pH 6.7 and 24” was incubated in the cold with sufficient added NaCl to eliminate the tubulin oligomers. The salt-treated sample was subsequently dialyzed in the cold against two changes of buffer at pH 6.7 to remove the added salt prior to sedimentation velocity analysis. The top schlieren pattern is of the untreated sample which con- tained 59% 30 S oligomer. Incubation of this material in the NaCl buffer at a total ionic strength of 0.3 led to the complete elimination of the 30 S species (Fig. 7, second pattern). Dialysis of the salt-treated material against buffer without added NaCl (p = 0.095) resulted in the reappearance of the 30 S oligomer at 73% of the untreated concentration. Similar results have been obtained from similar experiments per- formed on solutions which contained mixtures of 30 S and 18 S oligomers. In one such experiment, a preparation at pH 7.22 and 24” which contained initially 30% 30 S, 20% 18 S, and 50% 6 S species was converted entirely to 6 S material by addition of salt Q..L = 0.30). After subsequent dialysis to remove the salt, the resulting solution had the following composition: 17% 30 S, 29% 18 S, and 54% 6 S tubulin. Hence, these results demonstrate that both the 18 S and 30 S species are reversibly

dissociated by salt and they provide additional evidence for the equilibrium nature of the interconversion of the various microtubule protein species.

“Phase Diagram” -The results of the pH and ionic strength experiments showed that these two solution variables were critical in determining the equilibrium concentrations of the several tubulin components. However, a complete description of the solution conditions which delineate the 30 S to 18 S transition was not available from these studies. In order to document more fully the nature of the interconversion, it was necessary to vary pH and ionic strength in a systematic fashion and to construct a phase diagram for these solution variables.

The phase diagram (Fig. 8) was constructed from the results of sedimentation velocity analyses on a large number (-60) of solutions of microtubule protein at 5 mg/ml (see “Experimen- tal Procedures” for details). The diagram consists of two families of curves, and each curve represents a region of constant oligomer concentration (0 to 50% for 30 S and 0 to 40% for 18 S). The 6 S species exists throughout the diagram and its mass fraction at any given point may be derived by subtracting the per cent of the oligomeric forms present from 100 %. These curves cover a range of pH values which is coincident with the effective buffering range of 1,4-pipera- zinediethanesulfonic acid (Good et al., 1966), and also cover the range of physiological and near-physiological pH and ionic strength values. As shown in Fig. 8, between pH 6.1 and 6.5 and below ionic strengths of approximately 0.20 only the 30 S oligomer was observed together with the 6 S species. The concentration of the 30 S material was favored at low ionic strengths and was maximal at values of p < 0.10 where it constituted some 40 to 50% of the total protein in solution. The increase in 30 S concentration with decreasing ionic strength which was observed throughout the pH range investigated is consistent with the proposition that ionic bonding plays an important role in 30 S oligomer formation and stability.

FIG. 7. Ionic strength reversibility of 30 S oligomer formation. A solution of microtubule protein at 7.0 mg/ml was prepared at pH 6.66 (top pattern). An aliquot of this solution was incubated with NaCl at a total ionic strength of 0.3 for 1 h at 0” (second pattern). A portion of the salt-treated material was dialyzed against two changes of buffer at pH 6.7 at a buffer volume 100 times that of the sample (bottom pattern). The three samples were centrifuged in an AnF rotor at 42,040 rpm at 5” and were photographed 26 min after reaching speed at a phase plate angle of 75”.

2830 Centrifugation of Microtubule Protein Oligomers

30

I 60

I I I 6.5 7.0 7.5

pli at 24 *

FIG. 8. Phase diagram of the microtubule protein oligomer sys- tem. Microtubule protein solutions at 5.0 mg/ml were prepared at a variety of pH values. Aliquots of these solutions were incubated with varying concentrations of NaCl for 1 h at 0” prior to ultracen- trifugation at 42,040 rpm at 5”. Mass fraction measurements of the tubulin components were made from schlieren photographs taken at a phase plate angle of 75”. The pH of aliquots of the protein solutions were measured at 24” and the ionic strength calculated as outlined under “Experimental Procedures.” ---, regions of constant 30 S oligomer concentration from 0 to 53% of the total protein concentra- tion; p, regions of constant 18 S oligomer concentration ranging from 0 to 40% of the total. The 6 S species was observed throughout the diagram and its approximate concentration at any point can be obtained by subtracting the percentages of the oligomer concentra- tions from 100%.

As the pH was increased above 6.5, the 18 S oligomer formed and, for ionic strengths between approximately 0.10 and 0.20, it was in equilibrium with the 30 S and 6 S species. Through- out this region of overlap of the two families of curves, a three- species system was observed in the analytical ultracentrifuge with a maximum total oligomer concentration (30 S + 18 S) of approximately 65% of the total protein concentration. In the overlap region a reciprocal relationship was found between 30 S concentration, 18 S concentration, and ionic strength; in- creasing ionic strength favored the formation of 18 S oligomer at the expense of the 30 S species. Increasing the ionic strength above approximately 0.20 resulted in the disappearance of the 30 S species and the presence of only the 18 S and 6 S species in solution. For pH values greater than 6.5, the concentration of the 18 S oligomer increased from zero at ionic strengths greater than approximately 0.28 to more than 40% of the protein at pH 7.2 and an ionic strength of approximately 0.18.

The phase diagram results support the notion that the 18 S oligomer can form by direct breakdown of the 30 S species; however, the reverse reaction of 30 S formation from solutions containing a high concentration of the 18 S species was not shown. In order to demonstrate that the 18 S species was not an irreversible breakdown product of the 30 S species, we attempted to convert the 18 S oligomer to the 30 S form. A microtubule protein solution in Buffer A at pH 7.3 at 24” and p 1 0.18 was analyzed by sedimentation and found to contain 17% 30 S, 21% 18 S, and 62% 6 S species. After dialysis against two changes of buffer at pH 6.15 in the cold (conditions under which no 18 S would be initially present), the sample was analyzed again and showed greater than 60% of the protein in the 30 S form with no 18 S detectable and the remainder as the 6 S species. The conversion of the 18 S material into the

30 S species by this treatment indicated that the 30 S to 18 S transition is indeed reversible.

DISCUSSION

Effects of pH, Ionic Strength, and Concentration on Oligo- mer Equilibria-Our results confirm that the two species previously observed in the ultracentrifuge in addition to the 6 S tubulin dimer (Olmsted et al., 1974) represent two oligo- mers with discrete properties (Figs. 1 and 2). The two must contain tubulin in view of their high relative concentration in our protein preparations (Figs. 3, 4, and 8) and the amount of tubulin in our preparations (Borisy et al., 1975). The 30 S species is decidedly the more stable oligomer. It is the more prominent oligomer under conditions of pH and ionic strength that correspond to the optima for microtubule assembly (Olmsted and Borisy, 19751, suggesting that it may be the physiologically more important species. Our results suggest that the 30 S oligomer is reversibly converted to the 18 S species by increasing ionic strength or pH (Figs. 5, 6, and 8) and that both oligomers can be reversibly converted to sub- units (Fig. 7), indicating that the forces involved in maintain- ing these structures are noncovalent and may involve ionic interactions.

Self-association Behavior of Oligomers -The variation in the relative concentrations of 6 S and 30 S or 6 S and 18 S species with total concentration is consistent with the reversi- ble association of subunits to form rings. However, the reac- tions cannot be of the simple, rapidly reversible type nM e P where n subunits, M, associate to form a polymer species, P. For such a reaction the product concentration would approach 100% at sufficiently high concentrations of total protein. In the present case the 30 S species clearly does not exhibit this behavior, but appears to reach a maximum relative concentra- tion of about 65% (Fig. 4), indicating that the fraction of the total protein capable of participating in the formation of rings is limited in some way under the conditions employed. How- ever, this limitation is not due simply to the presence of appreciable amounts of inactive tubulin since other studies (Johnson and Borisy, 1975) have shown that greater than 95% of the protein in our preparations is competent to be incorpo- rated into microtubules.

Other predictions for a rapidly reversible association scheme of the type nil4 Z? P are that the sedimentation coefficient of the oligomer peak in such a system would decrease at suffi- ciently low total protein concentration, and that the oligomer peak would disappear completely when the total concentration was below a “critical” concentration level (Gilbert, 1955, 1959). These observations have been borne out for the 42 S rings observed by Frigon and Timasheff (1975) formed from pure tubulin at high concentrations of Mgz+. But we observed neither effect for the 30 S rings at the lowest concentrations at which our preparations were examined in the ultracentrifuge (0.43 mg/ml, Fig. 4). However, the 18 S oligomers were absent at low protein concentrations, and a tendency toward a de- creased sedimentation coefficient may be in evidence at the lowest concentrations at which the 18 S species was detectable (Fig. 2). Thus, the 18 S oligomer may behave as a Gilbert system and it is possible that the 30 S oligomer would also show the predicted behavior if it were examined at sufficiently low concentrations of protein.

On the other hand, the scheme, nM + P, may be inadequate to describe the 18 S-30 S interconversion. While the concentra- tion of these oligomers, when examined independently, de-

Centrifugation of Microtubule Protein Oligomers 2831

pended on total protein concentration in a manner that could be consistent with a self-associating structure, the behavior of the two together is not as readily explained. If we assume that the 18 S species represents a less extensive degree of associa- tion, i.e. contains fewer subunits of tubulin, then it is surpris- ing that the 30 S structure would appear at a lower concentra- tion than the 18 S structure. This observation would appear to

indicate that the 30 S structure is not a simple aggregate of 2 or more 18 S units. Thus, while it appears from our results that more than one tubulin-containing oligomer is formed in our preparations and that the structures formed are intercon- vertible, the behavior of the system represents a departure from that previously described for associating systems.

An element of the association scheme not described by the scheme, nM z? P, is the role of nontubulin proteins in the formation of oligomers. Murphy and Borisy (1975) have re- ported that, in addition to tubulin, nontubulin factors promote ring formation. These factors would be expected to influence the association properties of the oligomers and might account for the deviation from the behavior predicted by Gilbert (1955, 1959) for a simple self-associating system. Vallee and Borisy (1978) have fractionated microtubule protein by gel filtration chromatography into oligomeric and 6 S subunit fractions and

have demonstrated that nontubulin proteins of high molecular weight are associated preferentially with the 30 S fraction.

These authors have estimated that the stoichiometry of HMW” to tubulin in the 30 S fraction was 1 HMW molecule/five or six tubulin dimers. Based on the above stoichiometry and the composition of purified microtubule protein they have calcu- lated that the HMW concentration could be responsible for limiting the amount of 30 S formation to mass fractions in the range of the observed data (Fig. 4). Thus, the deviation from the theoretical mass distribution predictions of Gilbert (1955, 1959) may be interpreted in terms of a limiting concentration of a required ligand (HMW). This possibility is considered further in an accompanying paper (Vallee and Borisy, 1978).

Structure of Oligomers -Preliminary evidence has indi- cated the microtubule protein oligomers to be single-walled, ring-shaped structures (Olmsted et al., 1974). Vallee and Borisy (1978) confirm this result for the 30 S ring and indicate the outer diameter of the ring to be 39 nm. Although the oligomeric nature of the rings is readily apparent from elec- tron micrographs (Olmsted et al., 1974), the details of struc- ture are not obvious. In addition, it is possible that the 18 S oligomer is not a ring (Vallee and Borisy, 1978). We therefore pursue the question of the substructure with regard to the 30 S species only.

We have made calculations for the sedimentation coeffi- cients of several oligomeric models using an expression for the frictional coefficient

f = nf{l + [f/(6 qn)] i i [R,‘]}-’ I=1 .s=*

where n is the number of subunits, f’ is the subunit frictional coefficient, n is the viscosity of water, and R,, is the center-to- center distance between any two subunits (Kirkwood, 1954; see also Weingarten et al., 1974). The value off’ used was 8.19 X lo-* g/s as calculated from the Svedberg equation f’ = M(1 ~ Up)/Ns, where N is Avogadro’s number, p is the density of water at 20”, and b, M, and s are the partial specific volume, molecular mass, and sedimentation coefficient of the dimeric

3 The abbreviation used is: HMW, high molecular weight class of microtubule-associated proteins (Murphy and Borisy, 1975).

tubulin subunit, respectively. The values used for d, M, and x were 0.736 ml/g (Lee and Timasheff, 1974), 1.08 x lo5 g/mol (Lee et al., 1973), and 5.8 S (Weisenberg et al., 1968; Frigon and Timasheff, 1975), respectively. For each model considered, the value of n was chosen and the double sum of reciprocal distances was calculated from the model geometry. The result- ing value off was then used to calculate the sedimentation coefficient of the oligomer by use of the Svedberg equation, using for M the total mass of the oligomer. The calculated model sedimentation coefficients were compared to our mea- sured value of 30.6 S to assess the suitability of the model as a representation of the structure of the oligomer.

The simplest model for the 30 S species, suggested by the electron microscopic observations of Olmsted et al. (1974) and Vallee and Borisy (1978), is a single flat ring of 6 S tubulin dimers arranged end-to-end as in the protofilament of a microtubule. Since the length of a dimer is of the order of 8 nm (Amos and Klug, 1974) and the outer diameter of the ring is of the order of 39 nm (median radius = 17.25 nm; Vallee and Borisy, 1978), reasonable choices for the value of n are from 12 to 15. Accordingly, we have used the procedure outlined above to calculate the sedimentation coefficient for an oligomer consisting solely of 12 to 15 tubulin dimers situated with their centers evenly distributed around a ring of radius equal to 17.25 nm. The resulting values for the frictional and sedimen- tation coefficients are shown in Table I.

It is seen from the values in Table I that the single flat ring model with 12 to 15 subunits does not give sedimentation coefficients satisfactorily close to the experimental value of 30.6 S. The calculated sedimentation coefficients are 16 to 33% lower than the desired value. We note without further com- ment at this time, however, that the values calculated for the single ring are sufficiently close to the experimental value for the 18 S oligomer for this model to be considered for this species.

Since additional mass would increase the sedimentation coefficient, it is instructive to consider a model for the 30 S species with more than 12 to 15 subunits. We have considered a double-layered ring model in which an additional layer of subunits is added in the axial direction in order to preserve

TABLE I

Calculated frictional and sedimentation coefficients for rings of tub&in dimers

Subunit (dimer) frictional coefficient f = 8.19 x 10mL g SK’ was calculated from M = 1.08 x lo5 g mall’ (see Lee et al., 1973), d = 0.736 ml g-’ (see Lee and Timasheff, 19741, and sf,, = 5.8 S (Weisenberg et al., 1968; Frigon and Timasheff, 1975).

n (dimers/layer) f sB0.w

(g s-‘) x I07 s

Single rings” 12 2.799 20.4 13 2.799 22.1 14 2.795 23.8 15 2.788 25.6

Axially doubled ringsb 12 2.920 39.1 13 2.925 42.2 14 2.926 45.5 15 2.925 48.7

’ Dimers in the single ring are evenly distributed with their centers on a circle of radius 17.25 nm.

’ Dimers in the axially doubled rings are arranged such that the distance between the center of a dimer and the center of the nearest dimer in the other ring is 5.0 nm, the width of a microtubule protofilament (Erickson, 1974). Dimer centers are positioned on lines perpendicular to the ring planes.

2832 Centrifugation of Microtubule Protein Oligomers

the apparent single-walled character of the 30 S species. The results of the calculations for this model are also shown in Table I where, again, cases having 12 to 15 subunits/layer are considered.

In Table I it is observed that the double-layered ring model with 12 to 15 subunits/layer is also inadequate to describe the 30 S species, the sedimentation coefficients in these cases

being 28 to 60% higher than desired. The values calculated for the axially doubled ring model are in the range observed experimentally by Weingarten et al. (1974) and by Scheele and Borisy (1976) for oligomers of microtubule protein isolated by the method of Shelanski et al. (1973) (36 S), or by Frigon and Timasheff (1975) for oligomers of purified tubulin (42 S). However, this similarity of the calculated and experimental values cannot be used to argue that the axially doubled ring is a suitable model for these oligomers since electron microscopic evidence indicates that they are radially doubled rings (Wein- garten et al., 1974; Frigon and Timasheff, 1975).

Since both the single- and double-layered models consisting solely of tubulin subunits fail to describe adequately the sedimentation behavior of the 30 S oligomer, it becomes appropriate to consider models which are less simple in concept. This study has shown that a simple polymerization scheme of the type nM e P is not adequate to describe the formation of the microtubule protein oligomers and that consideration of nontubulin proteins in the polymerization scheme may be necessary. Vallee and Borisy (1978) have shown that the HMW proteins are, in fact, integral compo- nents of the 30 S structure. We therefore have considered the possible contributions of these proteins in the hydrodynamic models as well. We have made calculations of the sedimenta- tion coefficients of the single-layered ring model with addi- tional mass provided by the major nontubulin protein associ- ated with microtubules, which is reported by Murphy and Borisy (1975) to have a molecular weight of 2.86 x 105. First, we assumed, in the absence of any information regarding the geometry of the association of the nontubulin proteins with the ring, that they made no contribution to the frictional coefficient of the oligomer. Calculations showed that addition of 2 molecules of the high molecular weight protein to a ring consisting of 13 tubulin dimers led to a sedimentation coeffi- cient of 31.1 S.

ered ring with added mass and drag cannot be calculated without assumption of a particular geometry, there are exper- imental precedents for the argument that addition of large projecting molecules can increase the frictional coefficient of a structure and reduce the sedimentation coefficient even though the mass is also increased. The study of Lowey and Cohen (1962) on the properties of proteolytically derived frag- ments of myosin provides an indication of the effects of a highly asymmetric projection on the sedimentation properties of a more globular unit. The sedimentation coefficient of native myosin is 6.4 S (Holtzer and Lowey, 1959) and limited

proteolytic digestion results in the formation of two fragments, the highly asymmetric light meromyosin of molecular weight 1.26 x 105, which sediments at 2.9 S, and the more globular heavy meromyosin fragment of molecular weight 3.20 x 105, sedimenting at 7.2 S plus about 2.4 x 10” g/m01 of material which is digested. Thus, myosin, which amounts to the combination of a relatively globular particle and an extended projection, has a mass which is 47% greater and a sedimenta- tion coefficient which is 11% less than that of the more globular moiety alone. This idea is further substantiated by the experimental results of Benbasat and Bloomfield (1975) who studied the joining of bacteriophage T4D heads and tails. These authors found that the sedimentation coefficient of the phage heads was 1025 S and that the addition of the asymmet- ric tail which had a sedimentation coefficient of approximately 130 S (King, 1968) resulted in a tail-fiber-less phage particle with a sedimentation coefficient of 968 S. The addition of the asymmetric tail which constituted an increase of approxi- mately 10% to the mass of the particle increased its frictional coefIicient so that the sedimentation coefficient was approxi- mately 6% less than the sedimentation coefficient of the head alone. Thus, from a consideration of the electron microscopic observations of the high molecular weight molecules project- ing from the surface of microtubules and of the sedimentation behavior of proteolytically derived myosin fragments and of T4D phage particles, we conclude that the attachment of the highly asymmetric HMW molecules to oligomeric rings of tubulin could lead to a sedimentation coefficient which is lower than that which the oligomer would otherwise be expected to display.

Thus, it appears that a single-layered ring with added mass can give an apparently satisfactory fit to the experimentally observed sedimentation coefficient of the 30 S oligomer. How- ever, there is an alternative model which should be consid- ered. It has been shown that the high molecular weight non- tubulin proteins project from the surface of microtubules (Murphy and Borisy, 1975; Vallee and Borisy, 1977), raising the possibility that they may project from the surface of the ring oligomer as well. If this were the case, the presence of these asymmetric, high molecular weight proteins would add frictional drag as well as mass to the oligomer. In the case of the single-layered ring model, it is probably safe to say that enough drag could be provided by the high molecular weight proteins to bring the sedimentation coefficient far enough below 30 S to make this model unrealistic. We have accord- ingly reconsidered the double-layered ring model and propose that such a model with 12 to 15 tubulin dimers/ring, having several high molecular weight proteins as integral parts of its structure and projecting from it (Vallee and Borisy, 1978) could also adequately represent the sedimentation properties of the 30 S microtubule protein oligomer.

The biochemical and hydrodynamic considerations enumer- ated in the preceding discussion have resulted in the develop- ment of two different models for the 30 S microtubule protein ring oligomer. We have designated these two models “single- layered ring with added mass” and “double-layered ring with added mass and drag.” The first of these models is a single ring of 12 to 15 tubulin dimers and one or two of the high molecular weight microtubule-associated proteins, the latter being bound in such a way as to have little effect on the frictional coefficient of the structure and serving to increase the oligomer mass. The second model consists of two rings each with 12 to 15 dimers. The flat rings are stacked one on top of the other and have several of the HMW protein molecules associated with them in such a way as to project from the oligomer surface and to add frictional drag to the particle as well as to contribute mass. The two accompanying papers (Vallee and Borisy, 1978; Scheele and Borisy, 1978) address the problem of distinguishing between and refining these models.

Amos, L. A., and Klug, A. (1974) J. Cell Sci. 14, 523-549 Benbasat, J. A.. and Bloomfield. V. A. (1975) J. Mol. Bid. 95. 335

Although the sedimentation coefficient of this double-lay- 357

REFERENCES

Centrifugation of Microtubule Protein Oligomers 2833

Borisy, G. G., Marcum, J. M., Olmsted, J. B., Murphy, D. B., and Johnson. K. A. (1975)Ann. N. Y. Acad. Sci. 253. 107-132

Borisy, G.‘G., and Olmsted, J. B. (1972) Science 177, 1196-1197 Creeth, J. M., and Knight, C. G. (1965) Biochim. Biophys. Acta 102,

549-558 Doenges, K. H., Biedert, S., and Paweletz, N. (1976) Biochemistry

15. 2995-2999 Erickson, H. P. (1974) J. Cell Biol. 60, 153-167 Friaon, R. P., and Timasheff. S. N. (1975) Biochemistry 14. 4559-

4566’ ”

Gilbert, G. A. (1955) Discuss. Faraday Sot. 20, 68-72 Gilbert. G. A. (1959)Proc. Trans. R. Sot. Land. Ser. A. Metab. Phvs.

Sci. 250, 377-388 Gilbert, L. M., and Gilbert. G. A. (1973) Methods Enzvmol. 27. 273-

296 Good, N. E., Winget, G. D., Winter, W., Connolly, T. N., Izawa, S.,

and Singh, R. M. M. (1966) Biochemistry 5, 467-477 Johnson, K. A., and Borisy, G. G. (1975) in Molecules and Cell

Movement (Inoue. S.. and Steuhens. R. E.. eds) nn. 119-141. Raven Press, New York’ I I_

Holtzer, A., and Lowey, S. (1959) J. Am. Chem. Sot. 81, 1370-1377 King, J. (1968) J. Mol. Biol. 32, 231-262 Kirkwood, J. G. (1974) J. Polymer Sci. 12, 1-14 Kirschner. M. W.. Honia. L.-S.. and Williams. R. C. (1975) J. Mol.

Biol. 99; 263-276 -’ ’ Kirschner. M. W., Williams. R. C.. Weingarten. M.. and Gerhart.

J. C. (1974)Proc. Natl. A&d. Sci. U. S.-A. 71,’ 1159-1163 Lee, J. C., and Timasheff, S. N. (1974) Biochemistry 13, 257-265 Lee, J. C., Frigon, R. P., and Timasheff, S. N. (1973) J. Biol. Chem.

248, 7253-7262

Lowey, S., and Cohen, C. (1962) J. Mol. Biol. 4, 293-308 Lowry, 0. H., Rosebrough, N. J., Farr, A. L., and Randall, R. J.

(1951) J. Biol. Chem. 193, 265-275 Marcum, J. M., and Borisy, G. G. (1974) Fed. Proc. 33, 1231 Marcum, J. M., and Borisy, G. G. (1978) J. Biol. Chem. 253, 2852-

2857 Murphy, D. B., and Borisy, G. G. (1975) Proc. Natl. Acad. Sci. U. S.

A. 72, 2696-2700 Olmsted, J. B., and Borisy, G. G. (1975) Biochemistry 14, 2996-3005 Olmsted, J. B., Marcum, J. M., Johnson, K. A., Allen, C., and

Borisv, G. G. (1974) J. Sunramol. Strut. 2. 429-450 Rebhun,.L. I., Mellon, M., Jemiolo, D., Nath; J., and Ivy, N. (1974)

J. Supramol. Strut. 2, 466-485 Schachman, H. K. (1959) Ultrucentrifugation in Biochemistry, Aca-

demic Press, New York Scheele, R. B. and Borisy, G. G. (1976) Biochem. Biophys. Res.

Commun. 70, l-7 Scheele, R. B., and Borisy, G. G. (1978) J. Biol. Chem. 253, 2846-

2851 Shelanski, M. L., Gaskin, G., and Cantor, C. R. (1973) Proc. N&Z.

Acad. Sci. U. S. A. 70, 765-768 Vallee, R. B., and Borisy, G. G. (1977) J. Biol. Chem. 252, 377-382 VaIIee, R. B., and Borisv, G. G. (1978) J. Biol. Chem. 253.2834-2845 Weingarten, M. D., St&r, M. M., Littman, D. R., and Kirschner,

M. W. (1974) Biochemistry 13, 5529-5537 Weingarten, M. D., Lockwood, A. H., Hwo, S. Y., and Kirschner,

M. W. (1975) Proc. Natl. Acad. Sci. U. S. A. 72, 1858-1862 Weisenberg, R. C., Borisy, G. G., and Taylor, E. (1968) Biochemistry

7, 4466-4479