Specific interaction between mouse liver non-histone chromosomal proteins and mouse DNA demonstrated by a sequential DNA-protein binding procedure

117

Biochimica et Biophysica Acta, 521 ( 1 9 7 8 ) 1 1 7 - - 1 2 5 © E l s e v i e r / N o r t h - H o l l a n d Biomedical Press

BBA 9 9 2 9 7

SPECIFIC INTERACTION BETWEEN MOUSE LIVER NON-HISTONE CHROMOSOMAL PROTEINS AND MOUSE DNA DEMONSTRATED BY A SEQUENTIAL DNA-PROTEIN BINDING PROCEDURE

BARRY

H. L E S S E R * and D A V I D E. C O M I N G S

Department of Medical Genetics, City of Hope National Medical Center, Duarte, Calif. 91010 (U.S.A.)

Summary The binding of mouse liver chromosomal proteins to DNA has been investigated using the nitrocellulose filter binding technique. Careful purification of the DNA involving nuclease Si digestion and prefiltration through a nitrocellulose filter is used to reduce background binding in the absence of protein to less than 1%. Procedures involving direct binding of protein to labeled DNA, competition of binding of labeled DNA by unlabeled DNA, and dissociation of DNA • protein complexes with time do not indicate significant preference for binding to mouse DNA relative to Escherichia coli DNA. This specificity is demonstrated much more clearly by a novel type of procedure, which we call a sequential binding procedure. In this procedure non-specific binding proteins are sequestered by incubation with an excess of unlabeled E. coli DNA prior to addition of labeled DNA. Under these conditions, labeled mouse DNA is bound to filters to a 3- to 4-fold greater extent than labeled E. coli DNA.

Introduction

The interaction of proteins with DNA is believed to be responsible for the direct regulation of gene expression in both prokaryotes and eukaryotes. In eukaryotes it appears that the non-histone nuclear proteins in particular are the elements involved in the fine control of genetic activity [1,2], while the histones are involved at a coarser level via their effects on chromatin structure [3]. Regulation of gene expression in prokaryotes has been shown to be accomplished by interaction of regulatory proteins with specific sequences of DNA * Present address: Biochemistry Group, D e p a r t m e n t o f Chemistry, University of Calgary, Calgary, Alberta T2N IN4, Canada. A b b r e v i a t i o n : A T , a d e n i n e plus thymine.

118

base pairs in controlling regions adjacent to the structural genes [4]. Because these regulatory proteins have a much smaller but still measurable affinity for nonspecific DNA sequences, and because the specific DNA binding sites are only a small fraction of the total genome, direct demonstration of sequencespecific DNA-protein interaction has required the use of purified components in the binding assays [5]. In eukaryotes the study of these types of interactions is made much more difficult by the greater complexity of cellfilar proteins, some of which (such as histones) bind to DNA tightly but non-specifically, and by the much greater sequence complexity of the genome. In order to determine the optimal conditions for isolation of specific DNA binding proteins, we have used the nitrocellulose filter binding technique [6] to monitor the interaction of mouse liver chromosomal proteins with homologous (mouse) and heterologous (bacterial} DNA. Previous studies [7--11] involving nitrocellulose filter assays of DNA-protein binding have employed three types of experiments: (1) direct binding in which protein is mixed with labeled DNA and the fraction of DNA bound is measured by the amount of radioactivity retained after filtration and washing; (2) binding competition in which protein is mixed simultaneously with labeled DNA and increasing amounts of unlabeled competing DNA; and (3) binding dissociation in which protein is first allowed to bind labeled DNA, after which an excess of unlabeled DNA is added and the dissociation of the DNA ' protein complexes is followed with time. Direct binding provides an indication of the relative affinity of various protein fractions for a given DNA. The competition and dissociation procedures have been used to detect specificity of DNA-protein binding on the assumption that DNAs with specific binding sites will compete better and retain bound proteins longer than those lacking such sites. All three of these procedures have the drawback that proteins which bind to DNA with high affinity but little or no specificity, such as histones or nuclear matrix proteins (Comings, D.E. and Wallack, A., unpublished results), may obscure the binding of specific proteins. We wish to report a fourth type of procedure, a sequential binding procedure, in which proteins are first incubated with unlabeled heterologous DNA to bind proteins which interact tightly but non-specifically. This is followed by addition of labeled homologous or heterologous DNA, and the proteins are allowed to equilibrate between the two DNAs. In the presence of an excess of unlabeled heterologous DNA, proteins that bind non-specifically are expected to remain mostly bound to the unlabeled DNA, while proteins that have a higher affinity for sites on the labeled homologous DNA should bind preferentially to that. We find that while the direct, competition and dissociation procedures show no clear indications of specific binding to homologous DNA, the sequential procedure does. Methods

Isolation of chromosomal proteins. White Swiss mice were used in all experiments and were killed by cervical dislocation. All procedures were carried out at 0--4°C. All buffers dontained either 1/~g/ml soybean trypsin inhibitor (Sigma Chemical Co., St. Louis, Mo.) or 2" 10 -4 M p h e n y l m e t h y l s u l f o n y l

fluoride (Sigma) to inhibit proteolysis.

119 Nuclei were purified from mouse liver as described previously [12,13] or by a modification of the procedure of Blobel and Potter [14]. Following two washes in 0.075 M NaCl/0.025 M EDTA, pH 7, with centrifugation at 1500 × g for 10 min in the Sorvall HB-4 rotor, loosely bound non-histones were extracted by two washes in 0.35 M NaC1/0.01 M Tris/pH 8. The supernatants were centrifuged at 200 000 X g for 60 min in the Spinco SW65Ti rotor. Histones and tightly bound non-histones were extracted from nuclei using 2 M NaC1 and 5 M urea and fractionated using hydroxyapatite, essentially as described by MacGillivray et al. [15]. Before use in binding experiments, all protein fractions were dialyzed against 0.1 mM EDTA/0.01 M Tris, pH 7.5, containing 0.1 M or 0.35 M NaC1, and centrifuged at 16 000 × g for 10 min. Protein concentrations were determined by a trichloroacetic acid turbidity assay [16]. Nucleic acid contamination was estimated by measurement of A280 nm/A260 nm [17]. Polyacrylamide gel electrophoresis. Electrophoresis was performed using the sodium dodecyl sulfate (SDS) slab gel technique described previously [12,13]. Labeling of DNA. Escherichia coli strain thymine-JG108 was grown in Media 9 [18] supplemented with 0.4% glucose, 0.4% Casamino acids (Difco Laboratories, Detroit, Mich.), and 0.5 #g/ml thymidine. To label DNA, exponentially growing cells were innoculated into the above medium without thymidine but containing 5 ~zCi/ml [3H]thymidine (40--60 Ci/mmol, New England Nuclear, Boston, Mass.), and incubation continued for 12--14 h. To label mouse DNA, Krebs ascites cells were incubated for 12--14 h with gentle shaking in McCoy's medium (Grand Island Biological Co., Grand Island, N.Y.) supplemented with 10% fetal calf serum, 1% non-essential amino acids, penicillin and streptomycin, and containing 5 pCi/ml [3H]thymidine. Isolation o f DNA. Sources of unlabeled DNA were pooled mouse livers, spleens and kidneys, frozen plateau phase E. coil purchased from Miles Laboratories (Elkhart, Ind.), and cultures of Neisseria catarrhalis purchased from Carolina Biological Supply Co. (Burlington, N.C.) and grown in tryptic soy broth (Difco Laboratories, Detroit, Mich.). DNA was isolated essentially by the procedure of Marmur [19]. The labeled DNAs were further purified by centrifugation to equilibrium in cesium chloride. Density of cesium chloride was 1.696 g]ml, and centrifugation was for 60--72 h at 130 000 Xg in a Spinco SW41Ti rotor. The purified DNA preparations were dialyzed against 0.1 M NaC1/0.03 mM ZnSO4/0.03 M NaCH3COO, pH 4.5, sheared by three passages through a 25 gauge needle, and digested at 37°C for 30 min with 10 units per ~g DNA of nuclease $1 (Miles Laboratories, Elkhart, Ind.). After three extractions with chloroform]isoamyl alcohol (24 : 1, v/v) the DNA was dialyzed against 0.03 M sodium phosphate, pH 6.8, and loaded onto a column of hydroxyapatite [20] (4 ml bed volume per mg DNA). The column was run at 60°C. The column was washed with 0.03 M sodium phosphate, pH 6.8, and 0.12 M sodium phosphate, 6.8, and then DNA was eluted with 0.48 M sodium phosphate, pH 6.8. The DNAs were dialyzed against 0.1 M NaCl/0.1 mM EDTA/0.01 M Tris, pH 7.5, and pre-filtered through a 0.45 ~um Mfllipore nitrocellulose filter (Millipore Corp., Bedford, Mass.) using a syringe and a Swinnex 25 filter holder (Millipore Corp., Bedford, Mass.) before use in filter binding assays.

120

The DNAs had spectral ratios of 1.9--2.1 for A260 nm/A2s0 nm and of 2.1--2.2 for A260 ,m/A23o nm. Hyperchromicity was 30--35%. The molecular weight of the sheared DNA was determined by sedimentation velocity using absorption optics of the Model E analytical ultracentrifuge (Beckman Instruments, Fullerton, Calif.) [21]. DNAs used in the binding experiments had a molecular weight of 4 • 106--5 • 106. Filter binding assays. Filter binding assays were performed essentially as described by Riggs et al. [22]. Protein and DNA were incubated in a total volume of 0.2 ml containing 20 pg bovine serum albumin (Sigma Chemical Co., St. Louis, Mo.) in 0.1 mM EDTA/3 mM MgC12/1 mM dithiothreitol/10 mM Tris, pH 7.5, and an appropriate concentration of NaC1. Incubation was carried out at 0°C. After an appropriate time, samples were diluted to 1 ml with 0.1 mM EDTA/3 mM MgC12/1 mM dithiothreitol/10 mM Tris, pH 7.5, plus NaCI, applied to a 25 mm nitrocellulose filter (pore size 0.45 pm) that had been pre-soaked in 0.1 mM EDTA/3 mM MgC12/1 mM dithiothreitol/10 mM Tris, pH 7.5, plus NaC1, and filtered at a rate of 1 ml/20--30 s. The filters were washed three times with 1 ml of 0.1 mM EDTA/3 mM MgC1/1 mM dithiothreitol/10 mM Tris, pH 7.5, plus NaC1, and counted in an LS-150 scintillation counter (Beckman Instruments, Fullerton, Calif.) using as scintiUant 10 ml of dioxane containing 100 g napthalene and 5 g diphenyloxazole/1. The amount of DNA bound to the filters in the absence of protein was 0.5--1% of input, and this background binding was subtracted. The specific activity of labeled DNA was 10--12 000 cpm/pg for E. coli DNA and 20-23 000 cpm//~g for mouse DNA. The amount of labeled DNA used per assay was routinely 0.1 gg. The number of cpm corresponding to 100% binding was determined by adding the appropriate amount of DNA to a scintillation vial containing a nitrocellulose filter that had been pre-soaked and washed as described above. Results

Composition of protein fractions. The electrophoretic patterns of the various fractions used in the binding studies are shown in Fig. 1. The non-histone fractions were free of detectable histone. The loosely bound and tightly bound non-histone fractions were electrophoretically very similar. R N A content was 0.08 rag/rag protein in the loosely bound and 0.04 rag/rag protein in the tightly bound fraction. Direct, competition and dissociation procedures. Direct, competition and dissociation binding procedures were performed as described by Riggs et al. [22,23]. With all three procedures, none of the chromosomal protein fractions examined showed a significant preference for binding to mouse D N A , as opposed to E. coli D N A (unpublished results). Sequential binding. To minimize the effects of proteins that bind to D N A tightly but non-specifically, proteins have first been incubated to equilibrium with 10 ~g unlabeled D N A followed by addition of 0.1 /2g labeled D N A (i.e., a 100-fold excess of unlabeled D N A is present), and the system allowed to re-equilibrate. Maximal binding of labeled D N A is observed after an incubation period of 1 h. The results with the tightly bound non-histones are shown in

121

Fig. I. Po lyacry lamid e gel electrophoresis of proteins used in the DNA binding experiments. A, loosely b o u n d non-histones; B, tightly b o u n d non-histones; C, total histones were eleetrophoresed using 12.5% 'SDS-polyacrylamide gels [ 12,13]. Each sample c o n t a i n e d 15 ~g protein.

Fig. 2. Binding of labeled mouse DNA is 3- to 4-fold greater than binding of labeled E. coli DNA after incubation with unlabeled E. coli (Fig. 2A) or Neisseria catarrhalis (Fig. 2B) DNA. After incubation with unlabeled mouse DNA (Fig. 2C), this difference is greatly reduced, especially at protein concentrations of 10 /~g or less, conditions where the unlabeled DNA (10 gg) is in excess. These results have been repeated with four different preparations. When the concentration of NaC1 is increased from 0.2 M to 0.35 M, binding is reduced by a factor of 10 (Fig. 2A, C). The results o f sequential binding studies with the loosely bound non-histone proteins are shown in Fig. 3. A small preference for labeled mouse DNA relative to labeled E. coli DNA is observed after incubation with unlabeled E. coli DNA (Fig. 3A), and this difference is eliminated after incubation with unlabeled mouse DNA (Fig. 3B). However, these differences are much smaller than those observed with the tightly bound non-histone fraction. Sequential binding studies have also been performed using histones and

122

log

A

80

/•~ /

//

/ 60

/ /

40

/

/

/

/

/

/

•/

/

!

/

/

/

/

/

60

/

B 40

.fq

/

I

20

"'

•

-i I ,•"

/

/ /

/

// =ms.

/

i

•

j

2o

/&

I

if/"

/' • /

/

p,W"

E

o

s/ /

210

10

20

lo

~g

lO

PROTEIN

20

lo

2o

o

~o

2o

)zg PROTEIN

Fig. 2. B i n d i n g o f D N A b y t h e t i g h t l y b o u n d n o n - h i s t o n a s in t h e s e q u e n t i a l b i n d i n g p r o c e d u r e . T h e indi c a t e d q u a n t i t i e s o f p r o t e i n s w e r e i n c u b a t e d f o r 1 h in e i t h e r 0.2 M NaC1 ( e , m) o r 0 . 3 5 M NaCI (o, o) cont a l n i n g ( A ) 10 /Jg u n i a b e l e d E. call D N A ; (B) 1 0 # g u n l a b e l e d N. catarrhalis D N A ; (C) 1 0 #g u n l a b e l e d m o u s e D N A . T h e n 0.1 # g o f 3 H - l a b e l e d E. call D N A ( e . . . . . o) o r 3 H - l a b e l e d m o u s e D N A (¢ •) w a s a d d e d , a n d b i n d i n g o f l a b e l e d D N A d e t e r m i n e d 1 h l a t e r using t h e n i t r o c e l l u l o s e filter b i n d i n g t e c h nique. Fig. 3. B i n d i n g of D N A b y t h e l o o s e l y b o u n d n o n - h i s t o n e s in t h e s e q u e n t i a l b i n d i n g p r o c e d u r e . T h e indic a t e d q u a n t i t i e s o f p r o t e i n s w e r e i n c u b a t e d f o r 1 h in 0 . 2 M NaC1 c o n t a i n i n g ( A ) 1 0 # g u n l a b e l e d E. call D N A ; (B) 10 /~g u n i a b e l e d m o u s e D N A . T h e n 0.1 /~g o f 3 H - l a b e l e d E. coli D N A (o . . . . . a ) o r 3 H - l a b e l e d m o u s e D N A (¢ -') w a s a d d e d , a n d b i n d i n g o f l a b e l e d D N A d e t e r m i n e d 1 h l a t e r b y t h e n i t r o c e l lulose filter b i n d i n g t e c h n i q u e .

results are shown in Fig. 4. These results are for binding in 0.35 M NaC1, but identical results are obtained in 0.2 M NaC1 (unpublished results). With 0.5 #g histone or less, binding o f labeled D N A is totally suppressed by incubation with 100 A

80 E3

z 0

/q

6O

/ / / // / s

/

Q

40

//

a/

/ ZO

a/

/

/Iss" / • / //

•"

•//J

i/e/" 0

/

/J/ J

,"

/

~

lO

k"

5

~o

)zg PROTEIN Fig. 4. B i n d i n g o f D N A b y h i s t o n e s in t h e s e q u e n t i a l b i n d i n g p r o c e d t t r e . T h e i n d i c a t e d q u a n t i t i e s o f prot e i n s w e r e i n c u b a t e d f o r 1 h in 0 . 3 5 M NaCI e o n t a i n i n g ( A ) 1 0 # g u n l a b e l e d E. co|i D N A ; (B) 1 0 / ~ g unl a b e l e d m o u s e D N A . T h e n 0.1 //g o f 3 H - l a b e l e d E. coil D N A (o . . . . . o) o r 3 H - l a b e l e d m o u s e D N A (= --) w a s a d d e d , a n d b i n d i n g o f l a b e l e d D N A d e t e r m i n e d 1 h l a t e r using t h e n i t r o c e l l u l o s e filter binding technique.

123

10 pg unlabeled DNA. With larger amounts of histone, preference is observed for labeled mouse DNA relative to labeled E. coli DNA, but the differences (1.5 to 2-fold) are less than the 3- to 4-fold the differences obtained with the tightly bound non-histones (Fig. 2). Unlike the tightly bound non-histones, the preference for labeled mouse DNA is not eliminated after incubation with unlabeled mouse DNA. Discussion

The nitrocellulose filter binding assay is a rapid and convenient method for monitoring DNA binding proteins [22,23]. In situations where sequencespecific DNA binding proteins are purified free of non-specific binding proteins, as is the case with the lac repressor [22,23], specificity can be demonstrated simply by direct interaction between the protein and DNA. However, in complex mixtures of proteins such as the chromosomal proteins, specific binding is obscured by proteins that bind to DNA tightly but non-specifically. We have developed a procedure, the sequential binding procedure, that enables the filter binding assay to be used to detect specific binding proteins in the presence of non-specific binding proteins. We have designed this procedure to isolate proteins that have DNA binding properties analogous to those of the lac repressor, that is, a measurable affinity for heterologous DNA along with an affinity for a homologous binding site that is several orders of magnitude greater [24]. Exposing the proteins to an excess of unlabeled heterologous DNA allows all DNA binding proteins to bind and proteins that bind tightly but non-specifically will remain bound to this DNA. Upon addition of tracer quantities of labeled homologous DNA, only proteins that have the desired properties, that is, a much higher affinity for specific sites on homologous DNA relative to affinity for heterologous DNA, can overcome the concentration excess of heterologous DNA and equilibrate onto the homologous DNA. Unlike the tandem column procedure [7,25], the sequential binding procedure does not depend on absolute specificity of the specific DNA binding proteins. The tandem column procedure utilizes a column of bacterial DNA immobilized on cellulose to remove non-specific binding proteins, presumably leaving the specific binding proteins to bind to a subsequent column of homologous DNA. However, greater than 90% of a protein with an affinity equal to that of the lac repressor for non-operator DNA [24] would be expected to bind to an E. coli DNA-cellulose column where the DNA concentration is about 2.5 mg/ml [25]. Other procedures which have been developed to demonstrate specific DNA binding chromosomal proteins have employed the cumbersome and time consuming techniques of gradient dialysis [26--29] and density gradient centrifugation [27--29]. The sequential binding procedure is much more suitable for routine monitoring of specific DNA binding proteins. We have found that treatment of labeled DNAs used in binding studies with nuclease $1 and pre-filtration through nitrocellulose filters is essential to reduce background binding of labeled DNA to less than 1% and to greatly improve reproducibility of protein binding (unpublished results). The only protein fraction that shows a dramatic preference for mouse DNA relative to E. coli DNA is the tightly bound non-histones (Fig. 2). This prefer-

124 ence is eliminated when preincubation is with unlabeled mouse DNA (Fig. 2C) rather than with unlabeled E. coli DNA (Fig. 2A). The loosely bound nonhistones do not show such preference (Fig. 3). One possible explanation of this difference might be the greater RNA content of the loosely bound non-histone fraction (RNA/protein = 0.08) compared to the tightly bound non-histone fraction (RNA/protein = 0.04). RNA has been shown to interfere with the binding of non-histone proteins to DNA [29] although the amount of RNA used (RNA/protein = 1.0) was much higher than the amount present in either the loosely bound or tightly bound non-historic fraction. Histones exhibit a significant preference for mouse DNA, but considerably less than the tightly bound non-histones (Fig. 4A). If this preference were truly specific, the difference should be greatly reduced following preincubatlon with unlabeled mouse DNA, but this is not the case (Fig. 4B). It is possible that the preference of histones for mouse DNA reflects base composition differences. Histones tend to bind more strongly to adenine plus thymine (AT)-rich DNA [30] ; mouse DNA contains 60% AT while E. coli DNA contains only 50% AT [31]. This possibility could be tested directly by examining the binding of a bacterial DNA that has the same base compositions as mouse DNA, such as DNA from N. catarrhalis which contains 59% AT [31]. However, attempts to label DNA from N. catarrhalis have not been successful since the organism does not incorporate [3H]thymidine in vivo. In the case of the tightly bound nonhistones, the preference for mouse DNA cannot be accounted for by differences in base composition since preferential binding to mouse DNA is observed after preincubation with unlabeled N. catarrhalis DNA (Fig. 2B) as well as after preincubation with E. coli DNA (Fig. 2A). The following evidence indicates that histones are not affecting the results of the sequential binding procedure: (1) The binding of labeled DNA by UP to 0.5 pg histone is completely suppressed by preincubation with 10/~g of unlabeled DNA (Fig. 4A and B). This amount of histone would represent at least 5% contamination of the quantities of non-histones used in sequential binding experiments, and polyacrylamide gel electrophoresis (Fig. 1) indicates undetectable historic contamination of the non-histone fraction. (2) Sequential binding of non-histones is greatly reduced when the NaC1 concentration is increased from 0.2 M to 0.35 M (Fig. 2A and C), whereas sequential binding of histories is unaffected by this change in NaC1 concentration. We conclude that the most suitable type of procedure for examination of specificity of interaction of proteins with DNA is the sequential binding procedure, which apparently reduces the effects of non-specific interactions and allows specific interactions to be expressed more clearly. In conjunction with the nitrocellulose filter binding assay, this procedure is a rapid and convenient method of screening large numbers of protein fractions for specific DNA binding proteins.

Acknowledgements This work was supported by National Institutes of Health Grants GM23199 and GM15886 and by a Fellowship to B.L. from the Medical Research Council of Canada. We gratefully acknowledge the invaluable technical assistance of Ms. Ann G. Miguel.

125 References I Paul, J., Gilmour, R.S., Affara, N., Birnie, G., Harrison, P., Hell, A., Humphries, S., Windass, J. and Young, G. (1973) Cold Spring Harbor Syrup. Quant. Biol. 38, 885---890 2 Stein, G., Park, W., Thrall, C., Mans, R. and Stein, J. (1975) Nature 257, 764--767 3 Kornberg, R.D. (1974) Science 184, 868--871 4 Goldberger, R.F., Deeley, R.G. and Mullinix, K.P. (1976) Adv. Gen. 18, 1--67 5 Lin, S.-Y. and Riggs, A.D. (1972) J. Mol. Biol. 72, 671--690 6 Jones, O.W. and Berg, P. (1966) J. Mol. Biol. 22, 199--209 7 Sevall, J.S., Cockburn, A., Savage, M. and Bonnet, J. (1975) Biochemistry 14, 782--789 S Johnson, J.D., St. John, T, and Bonner, J. (1975) Biochim. Biophys. Acta 378, 424--438 9 Umansky, S.R., Kovalev, Y.I. and Tokarskaya, V.I. (1975) Biochim. Biophys. Acta 383, 242--254 10 Thomas, T.L. and Patel, G.L. (1976) Biochemistry 15, 1481--1489 11 Bliithmann, H. (1976) Eur. J. Biochem. 70, 233--240 12 Comings, D.E. and Harris, D.C. (1975) Exp. Cell Res. 96, 161--1"/9 13 Comings, D.E. and Harris, D.C. (1976) J. Cell Biol. 70, 440--452 14 Blobel, G. and Potter, V.R. (1966) Science 154, 1662--1665 15 MacGillivray, A.J., Cameron, A., Krauze, R.J., Rickwood, D, and Paul, J. (1972) Biochim, Biophys. Acta 277, 384--402 16 Comings, D.E. and Tack, L.O. (1972) Exp. Cell. Res. 75, 73--78 17 Warburg, O. and Christian, W. (1941) Biochem. Z. 310, 384--421 18 Miller, J.H. (1972) Experiments in Molecular Genetics, p. 431. Cold Spring Harbor Laboratory, Cold Spring Harbor, New York 19 Marmur, J. (1963) Methods in Enzymology 6, 726--738 20 Miyazawa, Y. and Thomas, Jr., C.A. (1965) J. Mol. Biol. 11, 233--237 21 Doty, P., McGill, B.B. and Rice, S.A. (1958) Proc. Natl. Acad. Sci. U.S. 44, 432--438 22 Riggs, A.D., Suzuki, H. and Bourgeois, S. (1970) J. MoL Biol. 48, 67--83 23 Riggs, A.D., Bourgeois, S. and Cohn, M. (1970) J. Mol. Biol. 53, 401--417 24 Lin, S.-Y. and Riggs, A.D. (1975) Cell 4, 107--111 25 Van den Broek, H.W.J., Nooden, L.J,, Sevall, J.S. and Bonnet, J. (1973) Biochemistry 12, 229--236 26 Kleinsmith, L.J., Heidema, J. and Carroll, A. (1970) Nature 226, 1025---1026 27 Teng, C.S., Teng, C.T. and Allfrey, V.G. (1971) J. Biol. Chem. 246, 3597--3609 28 Chiu, J.-F., Wang, S., FuJitani, H. and Hnillca, L.S. (1975) Biochemistry 14, 4552--4558 29 Rickwood, D. and MacGilHvray, A.J. (1977) Exp. Cell Res. 104, 287--292 30 Hnilica, L.S. (1972) The Structure and Biological Function of Histones, p. 93, Chemical Rubbel Publishing Co., Cleveland, Ohio 31 Szybalski, W. (1968) Methods in Enzymology 12 (Part B), p. 346