Biochimica et Biophysica Acta 888 (1986) 49-61 49 Elsevier

BBA11789

Tubulin-chromatin interactions: evidence for tubulin-binding sites on chromatin a n d i so la ted o l i g o n u c l e o s o m e s

Gi l l es M i t h i e u x a . , B e r n a r d R o u x b a n d B e r n a r d R o u s s e t

o Institut National de la Santb et de la Recherche Mbdicale, U. 197, Facultb de Mbdecine Alexis Carrel, rue G. Paradin, F-69372 Lyons Cedex 08, and l, Laboratoire de Physicochimie Biologique, Unioersitb Claude Bernard, Lyons (France)

(Received April 16th, 1986)

Key words: Tubulin-binding site; Chromatin; Oligonucleosome; Tubulin-chromatin interaction; (Rat)

The interaction of tubulin with chromatin has been studied using (a) a radiolabeled tubulin binding assay and (b) velocity sedimentation analysis on isokinetie sucrose gradients. Soluble chromatin was prepared by mild micrococcai nuclease digestion of rat liver nuclei and tubulin was purified from rat brain by temperature-de- pendent assembly-disassembly and phosphocellulose chromatography. The tubulin-binding assay is based on the ability of chromatin to precipitate quantitatively at physiological ionic strength allowing separation of free tubulin from chromatin-bound tubulin. The binding of tubulin to unfractionated soluble chromatin was rapid, reversible and saturable. Saturation of binding sites was obtained using tubulin concentrations ranging from 0.5 to 400/~g/ml , in the presence of a high concentration (2.5 mg / ml ) of another acidic protein, bovine serum albumin. The Scatchard and Hill plots showed that tubulin bound to a single class of non-interacting sites and yielded values of (0.5-0.6) • 107 M - l for an apparent K, and a maximal binding capacity of 0.8 nmoi tubul in/mg DNA, i.e. about 1 molecule of tubulin/10 nucleosomes. Similar binding parameters were obtained when binding experiments were performed with insoluble chromatin in 0.15 M NaCI. Velocity sedimentation analysis of tubulin-chromatin complexes revealed that tubulin bound to all classes of chromatin oligomers, irrespective of the length of the nucleosomal chain. Tubulin-trinueleosome complexes formed from isolated trinucleosome in the presence of an excess of tubulin were separated from free reactants. It was found that 10-15% of the starting oligonucleosomal species reacted with tubulin, in a stoichiometry of about 0.8 molecule of tubulin/nucleosome. Given the characteristics of the binding and the expected cellular free tubulin concentration, the tubulin-chromatin interaction could possibly take place in vivo, when the nuclear membrane breaks down during the first steps of mitosis.

Introduction

It is now established that a connection exists between microtubules and chromosomes, and be- tween microtubule protein and chromatin compo- nents [1-7]. Microtubule-associated proteins are able to bind to DNA [8-10]. Among them, high molecular weight polypeptides, which show a pref- erential binding for the sequences present in satel-

* To whom correspondence should be addressed.

lite DNA [9-14], have been proposed as the linker between pericentromeric DNA and microtubules. The major microtubule protein, tubulin, has also been found associated with chromatin [15-18] and kinetochore [19], but the question of its physio- logical role is still a matter of controversy. Re- cently, we have reported that the tubulin dimer possesses binding sites for histones, the essential chromatin protein [20]. The question which arises is whether tubulin is able to bind to histones when histones are involved in the DNA folding in the

0167-4889/86/$03.50 © 1986 Elsevier Science Publishers B.V. (Biomedical Division)

50

chromatin fiber. In the present study, using a radiolabeled tubulin-binding assay and velocity sedimentation analysis, we give evidence for the existence of a limiter number of high-affinity tubulin-binding sites on chromatin.

Experimental procedures

Purification and labeling of tubulin Tubulin was purified from rat brain by two

cycles of polymerization/depolymerization in 100 mM Mes/0.5 mM MgCI2/1 mM EGTA/0.1 mM GTP, pH 6.4 (buffer A), according to Shelanski et al. [21]. Polymerizing steps were performed in buffer A supplemented with 4 M glycerol and 1 mM GTP. Tubulin of purity higher than 98% was prepared by phosphocellulose chromatography [22] from two-cycle microtubule protein. Microtubule- associated proteins (MAP2 + Tau proteins) were prepared from two-cycle microtubule protein according to the procedure of Francon et al. [23]. Purified tubulin was labeled by conjugation with the Bolton-Hunter reagent (N-succinimidyl-3-(4- hydroxy[5-125I]iodophenyl)propionate) as previ- ously described [24]. The specific radioactivity of 125I-labeled tubulin was about 300 Ci /mmol . Be- fore use, purified tubulin (labeled or unlabeled) was made 1 mM phosphate/1 mM EDTA, pH 7.4 (buffer B) by gel filtration on Sephadex G-25.

Preparation of soluble chromatin Nuclei were isolated from rat liver as described

by Hewish and Burgoyne [25]. Oligonucleosomes were prepared by mild micrococcal nuclease di- gestion [26]. Briefly, nuclei were suspended at a concentration of 3 • 108 per ml in a buffer contain- ing 15 mM Tris-HCl/0.34 M sucrose/60 mM KC1/15 mM NaC1/15 mM 2-mercapto- ethanol/0.15 mM spermine/0.5 mM spermidine, pH 7.4. The suspension was incubated for 5 min at 37°C, made 1 mM CaC1 z and then 5 units (Sigma units) of micrococcal nuclease (Sigma, St Louis, MO) were added per ml of suspension. The reaction was stopped after 2.5 min by addition of EDTA to a final concentration of 2 mM and quickly chilled in ice for 10 min. After centrifuga- tion at 1500 × g for 5 min, pelleted nuclei were lysed in 0.2 mm EDTA/0.2 mM phenylmethyl- sulfonylfluoride, pH 7.0, by agitation at 4°C for

10 min. Soluble chromatin was separated from nuclear debris by centrifugation at 1500 × g for 5 min.

Oligonucleosomes were purified as described by Finch et al. [27]. Micrococcal-nuclease-digested chromatin was layered on isokinetic sucrose gradi- ents (5-28.2%, w/v), prepared in buffer B. Centrifugation was performed at 26 000 rpm in a Beckman SW28 rotor for 20 h at 4°C. Fractions of 0.5 ml were collected and the absorbance at 260 nm was recorded centinuously. Fractions corre- sponding to a given oligonucleosome were pooled and dialysed at 4°C against buffer B.

Tubulin-binding assay Soluble chromatin (25-50 /zg DNA) was in-

cubated with 125 I-labeled tubulin (80 000-140 000 cpm, 24-42 ng) with or without unlabeled tubulin in a total volume of 500/~1 buffer B supplemented with 2.5 mg/ml bovine serum albumin. Incuba- tion was performed for 15 min at 25°C. Chro- matin was made insoluble by addition of NaC1 (final concentration 0.15 M) and sedimented at 10000 × g for 5 min. Pellets were counted for radioactivity in a Packard scintillation gamma counter. For competitive binding studies, increas- ing concentrations of unlabeled pure tubulin (0.5-1400 ~g/ml) were added at the beginning of the incubation period. The binding of 125I-labeled tubulin obtained in the presence of a large excess of unlabeled tubulin (1.4 mg/ml) was taken as the nonspecific binding.

Protein analysis Proteins and nucleoproteins were precipitated

by 10% (w/v) trichloroacetic acid; the acid-insolu- ble material was pelleted by centrifugation at 4000 x g for 30 rain, washed twice with acetone and dried under vacuum. Polyacrylamide slab gel elec- trophoresis was performed as described by Laem- mli [28], with slight modifications [29]. Gels stained with Coomassie brilliant blue R250 were scanned using a Vernon recorder densitometer at 582 nm. The amount of protein in a given band was de- termined by measurements of the area of the band on the densitometer tracing and conversion of the area into an absolute amount using a standard curve generated with the given protein.

Protein was assayed according to Lowry et al.

[30] using bovine serum albumin as standard. Pro- tein profiles of gradients were determined accord- ing to Bradford [31].

DNA analysis Solutions containing DNA were made 3 mM

EDTA, 1% (w/v) SDS and 1 M NaCI, and incubated overnight at 37°C in the presence of 100 /~g/ml proteinase K (Merck, Darmstadt, F.R.G.). This treatment was followed by four extractions with chloroform/isoamyl alcohol (24:1, v/v) . Purified DNA was precipitated by addition of 2 vol. ethanol at - 2 0 °C and stored at - 20 ° C for 48 h. The precipitate was recovered by centrifugation at 3500 × g for 30 min, washed once with ethanol and dried under vacuum. The analysis on agarose/polyacrylamide gels was per- formed as previously described [29]. Gels were stained with 5 mg/1 ethidium bromide and photo- graphed under ultraviolet illumination using an orange filter.

DNA amounts were calculated from ab- sorbance measurements knowing that 1 unit of absorbance at 260 nm corresponds to a DNA concentration of 50/~g/ml.

Results

Polymerization of pure tubulin and 125I-labeled tubulin

Tubulin, purified by phosphocellulose chro- matography, was tested for its ability to poly- merize as a criterion of structural integrity. Fig. 1 (left panel) shows that purified tubulin was able to polymerize, provided that assembly-promoting factors (thermostable microtubule-associated pro- teins) were added. The right panel shows that tubulin molecules labeled by conjugation with 125I-Bolton Hunter reagent were able to incorpo- rate in microtubules in the same manner as unlabeled tubulin. Tubulin treated in low ionic strength buffer (particularly in buffer B) retained this property. Samples containing unlabeled tubu- lin and 125I-labeled tubulin were filtered on a Sephadex G-25 colunm for buffer exchange. After 30 min at 20°C in either buffer A (100 mM Mes), 25 mM Mes, 5 mM Mes or buffer B (1 mM phosphate/1 mM EDTA), they were mixed with microtubule proteins maintained at 4°C and po-

51

0.5-

0.4-

0.3-

<1 0.2-

0.1-

O- 4 8 12

TIME (rain)

Fig. 1. Left panel: polymerization kinetics of tubulin purified by phosphocellulose chromatography, followed by turbidime- try measurements at 350 nm. Purified tubulin (2 mg/ml) was incubated at 37°C in buffer A + 1 mM GTP in the absence (a) or in the presence of microtubule-associated proteins at the concentration of 0.2 mg /ml (b) and 0.3 mg /ml (c). (d) Poly- merization of microtubule protein (2.5 mg/ml) in the same conditions. Right panel: SDS-polyacrylamide gel analysis of protein content of microtubules assembled from microtubules proteins and a25I-labeled tubulin (left lane, Coomasie blue- stained gel; right lane, gel autoradiography). Microtubules were polymerized in buffer A supplemented with 1 mM GTP and 4 M glycerol for 30 rain at 37°C and pelleted by centrifu- gation at 150000× g for 1 h at 25°C. The microtubule pellet contained about 80% of total starting protein and 80% of radioactivity. Note that a and /~ subunits of tubulin are labeled to the same extent.

lymerized. Whatever the pretreatment of labeled tubulin, the amount of radioactivity incorporated into microtubules was similar (about 80% of the total) and the specific radioactivity of polymerized tubulin in each condition was not significantly different from the specific radioactivity of the initial mixture. Therefore, as far as the polymeri- zation capacity is concerned, 125I-labeled tubulin and tubulin treated in buffer B behave as native tubulin.

Properties of soluble chromatin The characteristics of compaction of chromatin

maintained in buffer B have been studied by velocity sedimentation analysis and circular dichroism. Chromatin oligomers were separated on isokinetic sucrose gradients and the sedimenta- tion coefficient of each species was calculated using the equation of Fritsch [32]. The relationship between the s value and the number of nucleo- somes for each oligomer is reported in the double

52

30 ¸

J

CO

, , , r ,

A

I I [ I I I 2 3 4 5 6 7

n

2 6 0 2 8 0 3 0 0

)~ (nm)

2

- 0

% x

- -1 ( D

Fig. 2. Sedimentation and circular dichroism properties of soluble chromatin at low ionic strength. A, double-logarithmic plot of the sedimentation coefficient of chromatin oligomers versus the number of nucleosomes of the chain. The relation- ship is described by a straight line the equation of which is: s = 10.6 × n °St (10.6 and 0.51 representing the cross-over point of the straight line with the ordonate axis and the slope of the straight line, respectively). B, circular dichroism spectrum of soluble chromatin in the near ultraviolet. Spectra were re- corded between 250 and 320 nm at room temperature with a Jobin & Yvon mark IV dicrograph, using chromatin solutions at a concentration of about 50 /~g DNA per ml, in 1 cm path-length cuvettes. Results are presented in terms of molar ellipticity [ O] expressed in deg. cm 2. dmol - 1.

logarithmic plot of Fig. 2A. The experimental data are described by a straight line the slope of which was 0.51, a value within the range (0.48-0.55) of those reported to be characteristic of a compact folding of the nucleosomal chain [33-37]. Unfrac- t ionated chromatin a n d / o r isolated oligonucleo- somes (purified on sucrose gradients) were studied by circular dichroism. A representative spectrum in the near-ultraviolet region is reported in Fig. 2B with extrema at 282 nm and 295 nm. Total chro- matin and isolated oligomers exhibited the same spectra. The molar ellipticity at 282 nm, a parame- ter characterizing the propor t ion of free D N A in nucleosomes [38], was 1820 + 65 d e g . cm 2. dmol-1 . This value is similar to those previously reported [39-42] for chromatin oligomers exhibit- ing a native D N A compaction.

Binding of 1:5i.labele d tubulin to chromatin We have taken advantage of the fact that solu-

ble chromat in self-assembles, under precise ionic conditions, into large-size sedimentable com- plexes, to develop a tubulin-binding assay in which

chromat in-bound tubulin could be separated f rom free tubulin by a 10000 × g centrifugation. At a low ratio of tubulin to 125I-labeled tubulin (30 ng), the chromatin pellet (about 30 /~g D N A ) con- tained 20% of the labeled tubulin. When increas- ing amounts of unlabeled tubulin were added dur- ing the incubation, the amount of pelleted 125I- tubulin was progressively decreased to about 2% of the total; this residual binding was taken as the nonspecific binding. The specificity of the interac- tion is documented by the fact that chromatin bound nanograms of labeled tubulin in the pres- ence of milligrams of an acidic protein, serum albumin, which was included in the binding buffer. The interaction of tubulin with chromatin is a rapid process: the max imum binding was obtained within a few (2-5) minutes of incubat ion at 25°C and remained stable for at least 45 min. The determinat ion of the precise time-course of the reaction has not been possible, since measurement of binding requires a 5 min centrifugation under condit ions where the reaction goes on. Compe- titive binding experiments between 125I-labeled and unlabeled tubulin have been performed under conditions where stable and reproducible binding was obtained (15 rain at 25°C). Results are re- ported in Fig. 3. Tubulin bound to chromat in at different concentrat ions of tubulin was plotted versus total tubulin concentrat ion in Fig. 3A. Binding of tubulin to chromatin was maximum at 400 /~g/ml tubulin. Binding parameters were ob- tained from the titration values by Scatchard and Hill analysis (Fig. 3B and C). Scatchard plots are linear, indicating the presence of one type of tubulin binding site on chromatin with an ap- parent overall affinity constant K a = (0.54 _+ 0.05) • 107 M -] . The maximum binding capacity was 0.82 + 0.15 nmol of tubulin d i m e r / m g D N A ; as- suming that each nucleosome contains a D N A fragment of 200 basepairs, this value corresponds to 1 molecule of tubulin bound per 10 nucleo- somes. The Hill plot of the binding data gave similar values for K a and a Hill coefficient close to 1.0. The reversibility of the tubu l in -chromat in interaction was studied in a two-step incubation procedure. Soluble chromatin was first incubated with 123I-labeled tubulin for 15 min, then un- labeled tubulin was added and the mixture was further incubated for 30 rain. The amount of

53

2

, m 0 lO0

¢'-" 0.2

,.n 0.1- 2 e~ z

o

TOTAL TUBULIN (!Jg/ml)

2 3 4 BOUND TUBULIN (pg/ml)

/ / - - I 9O0

24

14

_~ 01

C

Fig. 3. Equilibrium binding studies of 125I-labeled tubulin to chromatin. A, specific binding of tubulin to chromatin as a function of tubulin concentration. Chromatin (27.5 /tg DNA) was incubated with 125I-labeled tubulin (100000 cpm) and various concentrations of unlabeled tubulin. Bound tubulin was calculated from bound 125I-labeled tubulin and the specific radioactivity of tubulin in the assay. Nonspecific binding was determined as the residual 125I-labeled tubulin binding in the presence of 1.4 mg/ml tubulin in the assay. B, Scatchard plot of binding data. C, Hill plot of binding data (Bma x, maximal binding capacity; B, bound tubulin concentration; F, free tubulin concentration). Each point represents the mean value of duplicate incubations from a representative experiment.

15-

o~ 10-

l -

e a 5-

\

O. " ' ~ ' ~ ' ~ / - - - - - - e i I / / i 100 200 650

TUBULIN (Hg/ml)

Fig. 4. Reversibility of the binding of tubulin to chromatin. Chromatin (34 /~g DNA/ml) was first incubated with 125I-labeled tubulin (40000 cpm) in the presence of a low concentration of unlabeled tubulin (2/~g/ml). After 15 min at 25 o C, increasing concentrations of tubulin in buffer B (10-650 /tg/ml) or buffer B alone were added. At the end of the second incubation of 30 min at 25°C, chromatin was precipitated, sedimented and pellets were counted for radioactivity. Non- specific binding was taken as the residual radioactivity of a pellet of chromatin incubated in one step in the presence of 1 mg/ml tubulin. Each point represents the mean value of duplicate incubations.

Parameters of the binding of tubulin to chromatin at low (soluble chromatin) or physiological (in- soluble chromatin) ionic strength were compara- ble, although small differences were found: the

125I-labeled tubulin in the chromatin pellet was decreased in the samples containing unlabeled tubulin. The phenomenon was dependent on the concentration of tubulin (Fig. 4). Displacement of chromatin-bound labeled tubulin was obtained using tubulin concentrations similar to those used for the saturation of binding sites in competitive binding experiments.

In order to determine whether tubulin binding sites were similarly exposed and exhibited the same properties in soluble and insoluble chro- matin, binding experiments were performed with insoluble chromatin, i.e. chromatin in 0.15 M NaCI. Results are reported in Table I. Although insoluble, chromatin at physiological ionic strength gave homogeneous suspension. Measurement of the binding of 125I-tubulin to insoluble chromatin was peformed as described for soluble chromatin.

TABLE I

PARAMETERS OF TUBULIN BINDING TO CHRO- MATIN AT LOW AND PHYSIOLOGICAL IONIC STRENGTHS

Experiments were performed as described in Fig. 3. Bmax, maximal binding capacity; h, Hill coefficient; Ka, apparent affinity constant. Results are expressed as the means + S.E. of three (buffer B+150 mM NaC1) or five (buffer B) different experiments.

Binding parameters Incubation conditions

buffer B buffer B + 150 mM NaC1

Hillplot h 1.01+0.02 1.12+0.02 K a >(10 -7 (M -x) 0.56+0.06 1.29+0.38

Scatchard plot Bma x (nmol tubulin

bound)/mg DNA 0.82 + 0.15 0.55 + 0.08 K a >( 10 -7 (M -1) 0.54-t-0.05 1.03_+0.13

54

apparent affinity constant and Hill coefficient were slightly increased at 0.15 M NaC1, whereas the maximal binding capacity was decreased.

The presence of a limited number of tubulin- binding sites on chromatin was cnfirmed by a non-radioisotopic approach: chromatin-tubulin complexes were analyzed by polyacrylamide gel electrophoresis and the amount of tubulin on the Coomassie blue-stained gel was determined by quantitative densitometry using a standard curve generated with pure tubulin on parallel gels. Re- sults are reported in Fig. 5. The saturability of tubulin binding appears clearly; however, an ac- tual plateau was not reached; this is likely due to a fraction of nonspecific binding, the importance of which increased with the concentration of tubulin. This nonspecific binding could not be estimated using this approach; consequently, the actual tubulin-binding capacity of chromatin could not be accurately determined. However, at a tubulin concentration of 0.5 mg/ml, for which binding sites are expected to be saturated (see above), the amount of tubulin bound to chromatin was 9.8/~g per 100 ~g DNA, which corresponded to about

0.9 nmol tubulin dimer bound per mg DNA. This value is in agreement with the value obtained by the 1251_labeled tubulin-binding assay.

Velocity sedimentation analysis of tubulin-chromatin complexes

In order to know whether tubulin binds to all classes of oligonucleosomes or to selective oligomers, tubulin-chromatin complexes prepared at low ionic strength in the presence of an excess of tubulin were analyzed on isokinetic sucrose gradients. Tubulin and chromatin in solution in buffer B were mixed by adding slowly chromatin to tubulin. The interaction was allowed to proceed for 15 min at 25°C under mild agitation. The incubation mixtures were layered on sucrose gradients and centrifuged as described in the methods section. In the absence of tubulin, chro- matin oligomers containing from one to seven nucleosomes were correctly separated from each other (Fig. 6). After incubation in the presence of tubulin, the separation between the different oligomers wa markedly altered; there was no clear separation of discrete oligomeric species. In ad-

~" 10- ::3-

Z _ J

rn

I--- 5

a Z

o I11

TUBULIN

0 I i I I 0 0.2 0.4 0.6 0.8

T U B U L I N ( m g / m l )

1.0

Fig. 5. Determination of tubulin bound to chromatin at different tubulin concentra- tions, by quantitative densitometry per- formed on Coomassie blue-stained protein gels. Soluble chromatin (0.25 mg D N A / m l ) was incubated for 15 rain at 20°C, alone or with increasing concentrations of pure tubu- lin (15-1000 ~tg/ml). Chromatin was pre- cipitated and pelleted at 10000×g . Pellets were suspended in Laemmli suspension buffer and analyzed on SDS-polyacrylamide gels. Inset: protein content analysis of tubulin- chromatin complexes on 8% polyacrylamide gel. Lane 1: molecular mass markers: 94, 67, 56, 54 and 42 kDa, including a and fl sub- units of tubulin (15 /tg); lanes 2 to 7: chro- matin incubated with 0, 15, 60, 200, 500 and 1000 # g / m l tubulin (the equivalent of 100 pg D N A was analysed in each lane). The amount of tubulin in each lane was de- termined by comparison with a standard gel run in parallel, using tubulin amounts rang- ing from 0.5 to 30 ~g. Histones were stacked at the migration front when analyzed in this low-percentage polyacrylamide gel.

o 2 .0

l i | i J i i 2 3 4 5 6 7 8 r9 7', BOTTOM oP ! i 3'° l ,"=i o.~5

: , ' " 0 . 5 0 Ln

. . , ~ ,-, . ~ , , , ,

.: ".. , ~_ 1.0- : ; , 0 . 2 5

: : r " : r

o w o 0 10 20 3 0 4 0 5b 6 0 70

T U B E N U M B E R

Fig. 6. Isokinetic sucrose gradient sedimentation analysis of micrococcal-nuclease-digested chromatin preincubated with ( - - ) or without ( . . . . . . ) tubulin. Soluble chromatin (50 units of absorbance at 260 nm) was incubated for 10 min at 25°C without or with tubulin (11 mg) in a final volume of 4 ml and layered on sucrose gradients. Absorbance at 260 nm was recorded continuously during collection of fractions. A control gradient contained tubulin alone ( . . . . . . ); each tube sample was assayed for protein using the Coomassie method (A595). The different fractions were pooled as indicated at the top of the figure for subsequent analysis•

F R A C T I O N N U M B E R

2 3 4 5 6*7 8 9

55

Vl l - ~ V l ' - " - - V

~ I V

~ I I I

- 4- -- 4- -- 4- -- 4- - 4- -- 4- -- 4-

Fig. 7. Agarose-polyacrylamide gel analysis of oligonucleo- somal DNA fragments in chromatin fractions separated on sucrose gradient after incubation in the presence or in the absence of tubulin. For each gradient fraction (fractions 2 to 9), the lane corresponding to the tubulin-chromatin gradient is indicated under the gel by ( + ) and the lane corresponding to the chromatin gradient by ( - ) . Fraction 1 did not contain DNA. Roman numerals from I to IX refer to the migration of mononucleosomal to nonanucleosomal DNA fragments, re- spectively. Conditions of chromatin and tubulin concentration were those of Fig. 2. 5/~g DNA were analysed in each lane.

dition, a shoulder in the A260 profile appeared before the mononucleosome peak, revealing the presence of molecular species with a sedimenta- tion coefficient lower than that of the monomer. Whatever the conditions (presence or absence of tubulin), there was no material as pellets.

Fractions corresponding to the different chro- matin oligomers were pooled as indicated in Fig. 6 and analyzed in terms of DNA content (Fig. 7). The HaelII-digested q~X174 DNA fragments were used to check the mean DNA repetitive length, which was approx. 195-200 base pairs. The com- position of a given chromatin fraction in terms of oligonucleosomes was determined by analysis of the DNA length distribution. In the absence of tubulin, fractions corresponding to an oligonuc- leosome of class n (n nucleosomes) exhibited the correct purity, as evidenced by the limited con- tamination by the n-1 and n + 1 oligomers. The oligonucleosomal composition of the correspond- ing fractions of the gradient containing chro- matin-tubulin complexes was similar. However,

the contamination of a given oligonucleosomal fraction by adjacent species appeared to be larger. This is particularly clear for oligonucleosomes of low order (n = 1-4). It can be seen in Fig. 7 that: (a) oligonucleosomal DNA fragments of lower order were present in all the fractions (for exam- ple, dinucleosomal DNA was present in the tri- nucleosomal fraction) (b) higher-order oligonuc- leosomal DNA fragments were present in the frac- tions corresponding to low-order oligomers, up to the trimer (for example, dinucleosomal DNA in the mononucleosomal fraction), and (c) mono- and dinucleosomal DNA appeared in a fraction sedimenting slower than the mononucleosome. Therefore, if the majority of oligonucleosomes from soluble chromatin preincubated with tubulin sedimented at the expected place in the gradient, a population of oligonucleosomes sedimented faster and another population sedimented slower.

Proteins were analyzed in each gradient frac- tion (Fig. 8). In the control chromatin gradient, all five histone classes were present in each fraction,

56

FRACTION NUMBER

1 2 3 4 5 6 7 8 9

~,~ T U B .

" - H1

H3 ~,~,H2B "-"H2A -~'H4

Fig. 8. Analysis by electrophoresis on 16% polyacrylamide gels of chromatin fractions separated on sucrose gradients. For each fraction, the fight lane refers to chromatin incubated with tubulin and the left lane to the control chromatin. The amount of protein analyzed was 10/~g when samples contained tubulin or histones and 20/~g when samples contained tubulin and histones.

except that the mononucleosomal and dinucleo- somal fractions contained a "small proportion of histone H1. In the tubulin-chromatin gradient, tubulin was found in all the fractions: in a free form in fractions 1, 2 and 3, and bound to oligonucleosomes in fractions 2-9. Fraction 2 did not contain any histone in the control gradient. But, in the tubulin-chromatin gradient, histones were present in fraction 2. This fraction also con- tained D N A (Fig. 7) and tubulin. Since soluble tubulin-histone complexes cannot exist in an ex- cess of tubulin [20], histones were associated to D N A and therefore material of fraction 2 likely consisted of tubulin-mononucleosome (about 90%) and tubulin-dinucleosome ( = 10%) complexes in addition to unbound tubulin.

The relative amount of tubulin in each gradient fraction was estimated by densitometric analysis of polyacrylamide gels. Except in the mono- nucleosome fraction (fraction 3), which contained a significant amount of free tubulin, tubulin repre- sented a constant proportion of the total protein (tubulin +his tones) in all other fractions (4-9). It was not possible to determine the stoichiometry of the tubulin-chromatin interaction, since both free oligonucleosomes and tubulin-oligonucleosome complexes coexisted in all fractions. The fraction- ation of tubulin-chromatin complexes on sucrose gradients clearly shows that tubulin bound to all classes of oligonucleosomes irrespective of the length of the nucleosomal chain. Similar experi- ments have not been performed at higher ionic strengths, since at NaC1 concentrations above 20 mM, interoligonucleosomal interactions hinder a

clear separation of chromatin oligomers on the basis of their length of associated DNA [43].

Interaction of tubulin with isolated oligonucleosomes The formation of tubulin-chromatin complexes

was further studied using isolated nucleosomal species with two main purposes: (a) the separation of the complex (generated from a given oligonuc- leosome and tubulin) from the oligonucleosome itself and free tubulin, (b) the determination of the stochiometry of the interacting species. Fig. 9 shows that isolated oligonucleosomes of low order (mono-, di- and trinucleosome) exhibited a correct purity. Conditions of interaction were the same as for total soluble chromatin. The D N A concentra- tion of each oligonucleosomal fraction was about 0.1 mg /ml , which corresponded approximately to its concentration in the experiments with unfrac- tionated chromatin. Tubulin-isolated oligonucleo- some mixtures were analyzed on isokinetic sucrose gradients. For the three isolated oligomers, in- cubation with tubulin led to the formation of new sedimenting species with sedimentation coefficient lower than that of the oligomers alone (Fig. 9A, B, C) as revealed by the DNA pofile (A260). About 10-15% of the oligonucleosomal D N A was found in the complex fraction, in the three cases. The formation of faster sedimenting species was only a minor event.

Protein profiles showed that there was a rather good separation between free tubulin, tubulin-tri- nucleosome, complexes and the free trinucleosome (Fig. 9F). Fractions were pooled as indicated in Fig. 9F, and total protein was quantitatively

57

o ¢o 04

<(

I 1 l I J I I

1.0 /~t A

top /_ ', eorroM

i'i 0 . 5

0 a_ u..~-- 7 . . . . . . ~ t I I I ) I I I

1.o- C

I

0.5 /~ 0-

I I I / I I 0 10 20 30 40 50 60 70

TUBE

i i i i

I I I I ) I

1 i

D

BOTTOM

E /r ' - , \

/ f " ~ - -

I ) I I I I

1 2 3 4 5 i ~ i i t i ) ) r - -

"%

- ,, \

- / 'I

i I I I I o ,o 2o 3'o io

N U M B E R

I

F "t

I 70

-0.6

-0.4

0.2

-0

0.6

-0.4 i,n u~

• 0.2 ~

-0

-0.6

0 . 4

0 . 2

0

Fig. 9. Sucrose gradient analysis of tubulin-oligonucleosome complexes generated from free oligonucleosomes. Purified oligonucleo- somes (5 units of absorbance at 260 nm) were incubated in the presence of tubulin (1.1 mg) in a final volume of 4 ml. Absorbance at 260 nm was recorded continuously during the collection of fractions (panels A, B and C). Protein was assayed in each tube using the Coomassie method (A595) (panels D, E and F). Panels A and D refer to experiments with mononucleosome, panels B and E to experiments with dinucleosome and panels C and F to experiments with trinucleosome. - - , oligonucleosome preincubated with tubulin; . . . . . . , oligonucleosome alone. Protein profile of a gradient containing only tubulin (1.1 mg) is given in panel F ( . . . . . . ). Gradient fractions were pooled as indicated at the top of panel F in order to determine their protein composition.

analyzed on polyacrylamide gels (Fig. 10). In the absence of tubulin, histone could only be detected in fraction 4. In the presence of tubulin, both histone and tubulin were present in fraction 3, whereas fraction 4 still contained only histones. The amount of tubulin in fraction 3 was estimated from the densitometric analysis of the gel and a standard curve generated with tubulin. The ab-

solute amount of D N A was determined from ab- sorbance measurement at 260 nm. It was calcu- lated that the complex contained 650 /~g tubulin dimer (approx. 6 nmol) per mg D N A i.e. 0.8 molecule of tubulin dimer bound per nucleosome. This represented about 12 tubulin molecules bound per 100 nucleosomes, when considering the total starting trinucleosome fraction.

58

F R A C T I O N N U M B E R

1 2 3 4 5 1 2 3 4 5

' - Tubu l i n

Histone

Trinucleosome Trinucleosome alone Tub+ulin

Fig. 10. Determination of the protein composition of tubulin trinucleosome complexes by electrophoresis on 16% poly- acrylamide gels. Experimental conditions (tubulin and tri- nucleosome concentration) were those of Fig. 9. Fractions 1 to 5 correspond to those defined in Fig. 9.30% of total protein of each fraction was analyzed.

Discussion

In a preceding work [20], we found that tubulin exhibits a strong affinity for histones. The strength of the interaction was shown to be dependent on the respective content of lysine and arginine re- sidues of each histone class: the higher the arginine/lysine ratio, the higher the strength of the binding of histone to tubulin. Similarly, the tightness of the binding of histone to D N A in chromatin structure is known to be related to the arginine and lysine content of histone. H1, which is lysine-rich, exhibits the weaker binding to tubu- lin and is also the most weakly bound histone to D N A in chromatin. H3 and H4, which are arginine-rich and show the strongest binding to tubulin, are known to be the most tightly bound histones to nucleosomal DNA. Moreover, H2A and H2B, with an intermediate arginine/lysine ratio, present a binding to tubulin of intermediate

strength and an intermediate activity (compared to H1 and H3, H4) in the DNA folding occurring in chromatin structure.

In the present study, we have examined the ability of tubulin to bind to chromatin which contains about 50% histone by weight. Most of interaction experiments have been carried out in a low-salt phosphate buffer. In this buffer, inter- oligonucleosomal interactions, which could obscur the results, did not take place [43] and each chro- matin oligomer studied retained a native compac- tion of DNA. We have shown that tubulin fully retained its ability to polymerize after treatment in this buffer.

Using a radioiodinated tubulin-binding assay, we have found that chromatin (a) binds tubulin with a high affinity (apparent K a = (0.5-0.6). 107 M - t for soluble chromatin and about 1 • 1 0 7 M - 1 for insoluble chromatin) and (b) possesses a limited number of tubulin-binding sites: one site for tubulin dimer per 10 nucleosomes. The analysis of tubulin-chromatin complexes by velocity sedimen- tation indicates that tubulin binds to all classes of chromatin oligomers irrespective of the length of the nucleosomal chain but only to a small propor- tion (about 10%) of each oligonucleosomal species.

The possibility that simple electrostatic interac- tions could account for the binding of tubulin to chromatin can reasonably be excluded, since the characteristics of the binding were not very differ- ent when experiments were carried out in the presence of 0.15 M NaC1. In addition, binding studies with 125I-labeled tubulin have been per- formed in a medium containing a high concentra- tion of another acidic protein, bovine serum al- bumin. It must be noted that the maximal binding capacity was slightly lower at physiological ionic strength than at low ionic strength. One explana- tion would be that some binding sites are masked following self-condensation of chromatin frag- ments. Consequently, the small increase in Hill coefficient at 0.15 M NaCI, which might indicate some cooperativity increasing in the system, could be explained by the hypothesis according to which the fixation of a tubufin molecule on its binding site would reveal a site previously masked in the salt-induced condensed chromatin structure. It is interesting to notice that insoluble chromatin in 0.15 M NaCI (conditions close to the physiological

situation) binds tubulin with a somewhat higher affinity than soluble chromatin at low ionic strength. Variations of self-association states of pure tubulin, similar to those reported for micro- tubule proteins in salt conditions close to the physiological situation [44], could also account for the small observed differences between character- istics of the binding in the presence and in the absence of salt. Studies in high ionic strength medium ( > 200 mM NaC1) have not been done because of the instability of the his tone-DNA interaction in such conditions.

The stoichiometry of the interaction (1 mole- cule of tubulin bound per 10 nucleosomes) indi- cates that, if histone-tubulin interactions are pos- sibly involved in the binding of tubulin to chro- matin, such interactions do not represent the key mechanism. Indeed, each nucleosome possesses the same histone content, but only a small propor- tion of nucleosomes interacts with tubulin. As mentioned above, the existence of a tubulin-bind- ing site does not seem to be dependent on the length of the nucleosomal chain. The existence of a chromatin-associated non-histone protein, the presence of which in the nucleosome could gener- ate a tubulin-binding site, must be considered. However, this hypothesis seems very unlikely, since it was not possible to detect any additional pro- tein band in the trinucleosome-tubulin complex fraction as compared to the free trinucleosome fraction; one would expect such a protein to be present at least stochiometrically with tubulin in the complex. One possibility could be that the presence of a minor primary-structure variant his- tone in the nucleosome generates a tubulin-bind- ing site; the existence of primary-structure sub- types has been shown to occur for almost all histone classes [45]. Another possibility could be that the occurrence of a tubulin-binding site on a given nucleosome depends on its internal organi- zation. The presence of tubulin-binding sites could be related to discrete structural alterations of chromatin secondary to chemical modifications such as histone acetylation, histone phosphoryla- tion or DNA methylation [45,46].

Experiments carried out using isolated oligonu- cleosomes, especially trinucleosomes, suggest that binding sites for tubulin are located in a given region or in several discrete portions of chromatin

59

but not randomly distributed on chromatin. This hypothesis is supported by the fact that a small proportion of trinucleosome (10-15%) binds 2-3 tubulin molecules. Indeed, a random distribution would have led to the formation of a higher pro- portion of complexes containing less tubulin per trinucleosome.

By interacting with chromatin, tubulin induces modifications in the sedimentation behaviour of oligonucleosomes. Two groups of tubulin-oligonu- cleosome complexes have been found: (1) a first group of complexes which sediment faster than the corresponding oligonucleosome alone, an ob- servation in keeping with a mass increase, (2) a second group of complexes which sediment slower than the oligonucleosome alone. In the latter com- plexes, the effect of mass increase should be cancelled by additional modifications. Indeed, the dependence of the sedimentation coefficient on the compaction and the permeability to solvant of the nucleosomal chain is now well understood. There is a relationship between the decrease in the sedimentation coefficient of oligonucleosomes and the unfolding of the chromatin structure [34,35,39]. The tubulin-induced decrease in the sedimentation coefficient of oligonucleosomes complexed to tubulin may be interpreted in terms of such struct- ural changes. The hypothesis according to which tubulin could induce the removal of histones from chromatin oligomers to yield free or partially free D N A and tubulin-histone complexes appears very unlikely. Indeed, neither soluble nor insoluble tubulin-histone complexes are formed under the experimental conditions used in this study. The decrease in the sedimentation coefficients of tubulin-bound oligonucleosomes seems therefore assignable to the unfolding of the nucleosomal chain resulting in an extended conformation.

The fact that a part of tubulin-oligonucleosome complexes sediments faster than the correspond- ing oligonucleosome may be interpreted in two different ways: (1) the binding of tubulin to a given oligomer results in a mass increase and consequently in an increase in the sedimentation coefficient, provided that the complex retains a compact structure; (2) tubulin is able to crosslink two (or possibly more) oligomers containing n and m nucleosomes which gives an apparent oligomer of (n + m) nucleosomes. The first hy-

60

pothesis seems to be the most probable, since the augmentation of sedimentation coefficient of a given oligomer due to tubulin binding is always of limited extent and compatible with a limited mass increase. According to the second hypothesis, one would predict the presence of low-order oligonuc- leosomal D N A in fractions corresponding to oligomers of much higher sedimentation coeffi- cient. It is apparent on the D N A gels that this is not a significant feature in unfractionated chro- matin.

In conclusion, we have found that chromatin exhibits tubulin-binding activity. The apparent dissociation constant of tubulin-chromatin com- plexes is within the range of expected intracellular free tubulin concentrations. This suggests that tubulin-chromatin interactions could possibly take place in vivo, under conditions where cytosolic and intranuclear material are located in the same intracellular compartment, i.e. in the first steps of mitosis when the nuclear membrane breaks down. The kinetochores, which presumably contain chro- matin in the outer layer [4], were shown to bind tubulin in vitro [6], a binding which enhanced microtubule nucleation [6]. Even if additional sig- nals are required in vivo during mitosis since tubulin binds to kinetochores and not to chro- matin present in the rest of metaphase chro- mosome, the in vitro binding of tubulin to chro- matin studied in this paper could have biological significance.

References

1 Teltzer, B.R., Moses, M.J. and Rosenbaum, J.L. (1975) Prec. Natl. Acad. Sci. USA 72, 4023-4027

2 Gould, R.R. and Borisy, G.G. (1978) Exp. Cell. Res. 113, 369-374

3 Pepper, D.A. and Brinkley, B.R. (1979) J. Cell. Biol. 82, 585-591

4 Ris, H. and Witt, P.L. (1981) Chromosoma (Berl.) 82, 153-170

5 De Brabander, M. (1982) Cell Biol. Int. Rep. 6, 901-915 6 Mitchison, T.J. and Kirschner, M.W. (1985) J. Cell. Biol.

101,755-765 7 Mitchison, T.J. and Kirschner, M.W. (1985) J. Cell. Biol.

101,766-777 8 Corces, V.G., Salas, J., Salas, M.L. and Avila, J. (1978) Eur.

J. Biochem. 86, 473-479 9 Villasante, A., Corces, V.J., Manso-Martinez, R. and Avila,

J. (1981) Nucleic Acids Res. 9, 895-908

10 Corces, V.G., Manso, R., De la Torre, J., Avila, J., Nasr, A. and Wiche, G. (1980) Eur. J. Biochem. 105, 7-16

11 Wiche, G., Corces, V.G. and Avila, J. (1978) Nature (Lond.) 273, 403-405

12 Avila, J., Montejo de Garcini, E., Wandosell, F.. Villasante, A., Sogo, J.M. and Villanueva, N. (1983) EMBO J. 2. 1229-1234

13 Marx, K.A., Denial, T. and Keller, T. (1984) Biochim. Biophys. Acta 783, 283-292

14 Wanadosell, F., Rodriguez-Campos, A., Piqueras, J.F. and Avila, J. (1984) Chromosoma (Berl.) 90, 68-71

15 Douvas, A.S., Harrington, C.A. and Bonner, J. (1975) Prec. Natl. Acad. Sci. USA 72, 3902-3906

16 Menko, A.S. and Tan, K.B. (1980) Biochim. Biophys. Acta 629, 359-370

17 Farr, R.M., Luckow, V. and Sundharadas, G. (1979) Exp. Cell. Res. 121,428-432

18 Armbruster, B.L., Wunderli, H., Turner, B.M., Raska, I. and Kellenberger, E. (1983) J. Histechem. Cytochem. 31, 1385-1393

19 Pepper, D.A. and Brinkley, B.R. (1977) Chromosoma (Berl.) 60, 223-235

20 Mithieux, G., Alquier, C., Roux, B. and Rousset, B. (1984) J. Biol. Chem. 259, 15523-15531

21 Shelanski, M.L., Gaskin, F. and Cantor, C.R. (1973) Proc. Natl. Acad. Sci. USA 70, 765-768

22 Weingarten, M.D., Lockwood, A.H., Hwo, S.Y. and Kirschner, M.W. (1975) Prec. Natl. Acad. Sci. USA 72, 1858-1862

23 Francon, J., Fellous, A., Lennon, A.M. and Nunez, J. (1978) Eur. J. Biochem. 85, 43-53

24 Bernier-Valentin, F. and Rousset, B. (1982) J. Biol. Chem. 257, 7092-7099

25 Hewish, D.R. and Burgoyne, L.A. (1973) Biochem. Bio- phys. Res. Commun. 52, 504-510

26 Noll, M., Thomas, J.O. and Kornberg, R.D. (1975) Science 187, 1203-1206

27 Finch, J.T., Noll, M. and Komberg, R.D. (1975) Proc. Natl. Acad. Sci. USA 72, 3320-3322

28 Laemmli, U.K. (1970) Nature (Lond.) 227, 680-685 29 Mithieux, G. and Roux, B. (1984) Biechim. Biophys. Acta

781,286-293 30 Lowry, O.H., Rosebrough, N.J., Farr, A.L. and Randall,

R.J. (1951) J. Biol. Chem. 193, 265-275 31 Bradford, M.M. (1976) Anal. Biochem. 72, 248-254 32 Fritsch, A. (1973) Anal. Biochem. 55, 57-71 33 Ramsay-Shaw, B. and Schmitz, K.S. (1976) Biochem. Bio-

phys. Res. Commun. 73, 224-232 34 Noll, M. and Kornberg, R.D. (1977) J. Mol. Biol. 109,

393-404 35 Osipova, T.N., Pospelov, V.A., Svetlikova, S.B. and

Vorob'ev, V.I. (1980) Eur. J. Biochem. 113, 183-188 36 Pearson, E.C., Butler, P.J.G. and Thomas, J.O. (1983)

EMBO J. 2, 1367-1372 37 Gottesfeld, J.M. and Butler, P.J.G. (1977) Nucleic Acids

Res. 4, 3155-3173 38 Watanabe, K. and Iso, K. (1981) J. Mol. Biol. 151,143-163 39 Mithieux, G., Roux, B. and Marion, C. (1984) Biophys.

Chem. 20, 111-119

40 Marion, C., Bezot, P., Hesse-Bezot, C., Roux, B. and Bernengo, J.C. (1981) Eur. J. Biochem. 120, 169-176

41 De Murcia, G., Das, G.C., Erard, M. and Daune, M. (1978) Nucleic Acids Res. 5, 523-537

42 Rill, R.L. (1979) in Molecular Genetics III (Taylor, J.M., ed.), pp. 247-313, Academic Press, New York

43 Str~itling, W.H. (1979) Biochemistry 18, 596-603

61

44 Marcum, J.M. and Borisy, G.G. (1978) J. Biol. Chem. 253, 2825-2833

45 Isenberg, I. (1979) Annu. Rev. Biochem. 48, 159-191 46 Cartwright, I.L., Abmayr, S.M., Fleischman, G.,

Lowenhaupt, K., Elgin, S.C.R., Keene, M.A. and Howard, G.C. (1982) CRC Crit. Rev. Biochem. 13, 1-86