attraction, phasing and neighbour effects of histone octamers on curved dna

J. Mol. Biol. (1990) 216, 363-374

Attraction, Phasing and Neighbour Effects of Histone Octamers on Curved D N A

Giovanna Costanzo ~, Ernesto Di M a uro 1'2, Ga©tano Salina 3 and R o d o l f o Negri ~

ICentro Acidi Nucleici (CNR), Rome, italy

2Dipartimento di Genetica e Biologia Molecolare Universitd di Roma "La Sapienza", Rome, Italy

3Istituto Nazionale di Fisica Nucleare Sezione di Roma "'Tot Vergata", Rome, Italy

(Received 12 January 1990; accepted 27 July 1990)

Nucleosome core particles were reconstituted on various DNA fragments containing a Crithidia fasciculata kinetoplast curved tract.

The results show that, on curved DNA, nucleosome core particles form six- to sevenfold preferentially, relative to bulk sequences. The preferential deposition occurs at multiple periodic positions, whose distribution reveals a unique rotational setting of DNA with respect to the histone octamer surface and whose average periodicity is 10-26_0-04.

Evidence is provided for a context effect in histone octamer deposition: octamers bound to a segment of curved DNA influence the positions of neighbour octamers. Taken together, the preferential formation of nucleosome core particles and the influence on the localization of neighbouring particles suggest for intrinsically bent sequences the biologically relevant role of organizers of nucleosomal arrays.

1. I n t r o d u c t i o n

The nucleosome subunit of chromatin consists of DNA folded around a histone core as a 1-8-turn left- handed solenoid. Corollaries of the wrapping of DNA around the protein core are that the formation of a nucleosome should be favoured by (1) pre- existing negative supercoiling or (2) preformed DNA bends in the same direction. Negative supercoiling does indeed allow nucleosome core particle (ncpt) formation in topologically stringent conditions (Negri et al., 1989). The results presented here prove the validity of the second corollary and provide information relevant to a more detailed understanding of the DNA-histone core interaction.

The local conformation of DNA determines the selectivity of the interaction of histone octamers (Drew & Travers, 1985; Satchwell et al., 1986; Travers & Klug, 1987; Travers, 1989). Sequence-dependent bendability (or anisotropic flexibility) has been indicated as the major cause of a rotationally related positioning of octamers, as originally shown by DNase I footprinting of nucleo-

Abbreviations used: ncp, nucleosome core particle; bp, base-pair(s); ExoIII, exonuciease III.

0022-2836/90/220363-12 $03.00/0

somes on unique-sequence DNA (Drew & Travers, 1985) and by the specific positions of (A). and (T). runs relative to the octamer (Satehwell et al., 1986). These studies have shown that the 3' end of homo (A), or (T), is preferentially positioned at the inside of the ncp (Satchwell et al., 1986) and have estab- lished the principle of the rotational relationship between different possible locations of the histone octamer. The conclusions are based on the analysis of a large number of cloned DNA fragments from chicken erythrocyte nucleosomes and made use of an a posteriori logic; this approach has safely estab- lished the fact that preferential formation of nucleosomes had occurred in vivo on the analysed DNA segments and has defined the sequence-related conformational properties of these DNA segments. A similar study that uses the same logic on DNA from the same organism has been reported (Pennings et al., 1989); interestingly, one of the two segments analysed had properties of intrinsic curvature. I t has been proposed that the strain of bending DNA about the histone core might be relieved if the DNA contained a sequence-directed curvature (Trifonov 1980, 1985; Trifonov & Sussman, 1980).

The present study reaches similar conclusions on the relevance of local anisotropy in determining the

363 © 1990 Academic Press Limited

364 G. Costanzo et al.

positioning of ncps by an approach tha t uses a specular logic. In the search for the conformational signals tha t could locate ncps on DNA, we have not analysed sequence and phase requirements of reconst i tuted ncps on DNA segments selected for having been par t of in vivo nucleosomes. We have instead analysed the preferential i ty of formation and the phasing of ncps on a DNA endowed of a priori defined periodical anisotropy.

Preferential localization of ncps on large DNAs containing curved inserts has been reported (Hsieh & Griffith, 1988). I t was shown tha t ncps reconstituted preferentially on the curved segments and tha t under non-restrictive conditions reconsti tution occurred both in the curved and adjacent regions. These observations, made by electron microscopy, clearly indicate preferential i ty on curved sequences but suffer from unavoidable approximations (i.e. low resolution, ncp localization made relative to the nearest end, which may cause incorrect assign- ments; ncps shorten DNA).

The DNA used in this s tudy is the Crithidia fasciculata kinetoplast DNA, a left-handed intrinsically and highly curved molecule. We show tha t in vitro this curved DNA segment preferentially a t t racts and phases ncps.

In detail, the results show tha t in vitro: (1) the intrinsic periodic anisotropy of this DNA favours the reconstitution of ncps relative to mixed- sequence DNA; (2) the ncps form precisely in phase with the phased anisotropy; (3) ncps bound to a segment of curved DNA influence the position of neighbour octamers.

2. Materials and Methods

(a) Materials

Restriction endonucleases were from New England Biolabs and Boehringer-Mannheim; bacteriophage T4 polynucleotide kinase was from USB; Klenow enzyme, exonuclease III and microcoecal nuclease were from Boehringer-Mannheim. P~adiochemicals were from NEN.

(b) DNAs

The DNA segments used were obtained by various restrictions from the pSP65 (3005 bp) or pPK201/CAT (3230 bp; kindly provided by P.T. Englund) plasmids. pPK201/CAT contains the StuI-AccI 211 bp bent segment from the kinetoplast DNA of the trypanosoma- tidae protozoan C. fasciculata cloned in the BamHI site of the polylinker of pSP65 by fill-in and additions of BamHI linkers (Kitchin el al., 1986).

The sequence of the insert and of the surrounding linker segments are reported in Fig. 3(f). Map, size and relevant restriction sites of the fragments used are reported in Fig. 1.

(c) Reconstitution of ncp

Reconstitution of ncp was performed according to the salt dilution protocol described by Drew & Travers (1985), starting from 0"9 M-NaC1 to 100 mM-NaC1. The nueleosome core particles were: (1) from Saccharomyces cerevisiae

I H ~ ~ _ ~,,,~r 705

le-com ~ H Z N.-uI ... . ~ o ~ c ~ ''nd~ I I l i - - i - i ~_. - - ~ - -

5 ~bp ~RZ ~ 45l

4 , ~ - - 5 2 4

5 ~ x ~ox 25 I

6 ,.~*x ~ ' ~ 480

Figure 1. A map of the DNA segments used. Thin line, pSP65 vector sequences; thick line, linker; broken line, curved insert. The restriction sites that define the external boundaries of the larger fragment used are NheI and PvuII. NheI is at position 2763, PvuII is at position 238 of the vector pSP65. Position I is the 1st position of the linker and is the RNA initiation site of the plasmid. In fragment l, the size of the left and right arms of the plasmid are 242 bp and 178 bp long, respectively, the linker+insert is 283 bp. The total length (703 bp) of the fragment is indicated on the right. Fragments from 2 to 6: symbols as above. In fragment 2, the underlining indicates the A-tracts (see more detailed representations in Figs 3 and 4). The scale (100 bp) refers to fragments l and 3 to 6. Fragment 2 is enlarged 10 times. The sizes indicated refer to fragments with filled-in extremities.

chromatin (strain RSI90, deficient in DNA topoisomerase I, was kindly provided by R. Sternglanz, State University of New York at Stony Brooke), digested with micrococcal nuclease and purified according to a standard procedure through 5% to 20% (w/v) sucrose gradients made 500mM-NaC1; (2) from chicken erythrocyte, prepared according to Forte et al. (1989). Reconstitution was moni- tored by gel electrophoresis in 0"75% (w/v) agarose in 0-5×TBE buffer (TBE buffer is 0"045 M-Tris-borate, 0-001 M-EDTA (pH8)). The amount of reconstituted DNA was evaluated by scanning densitometry of underexposed autoradiograms.

5'-Terminal labelling and nick-translation were performed according to standard procedures. Fragments labelled at one extremity were obtained as follows: plasmid DNA was restricted with a 5'-protruding enzyme, labelled and restricted with the appropriate secondary enzyme. The fragments were separated by standard polyacrylamide gel electrophoresis and the band of interest was recovered.

DNA curvature refers to the non-linear trajectory of the helix axis due to certain sequence arrangements, DNA bending refers to the change of the helix axis imposed by protein bending.

3. Results

(a) Reconstitution of nucleosome core particles occurs preferentially on curved D N A tracts

The ncp reconstitutes, obtained according to the s tandard salt dilution protocol, were analysed, on agarose gel electrophoresis in a band re tardat ion assay (Fig. 2).

On a relatively large DNA fragment (524 bp, Fig. 2 (a)), six bands are formed. On a f ragment of

Histone Octamers on Curved D N A 365

1 2 3 4 5 6 7

,i i ' l "1 t u o ' I I , -

2 l

- • ]7::~

(b)

' , : ; ~ * . ' - ~ : : - . . - .

1:7 1

f!;. i '

-(8)

2 3 4 5.: :':~:

i / i l (c:i

~o 50 •

D

I I I I I 0~05 0.1 0.25 0.5 1.0

Nucleosomal core porficles (/~g)

2"0

Figure 2. Curved DNA or curve-containing DNA fragments reconstitute preferentially over vector sequences. (a) ncps reconstituted (as described in Materials and Methods) on 10 ng of 5'-terminally labelled bent-containing fragment 4 (NheI-HindIII, 524 bp) with 0, @05, 0-1, 0"25, 0"5, l and 2 #g (lanes l to 7) orS. cerevisiae ncps. U, unreconstituted DNA. (b) Bent DNA (5 ng: SmaI-HindIII, 257 bp) uniformly labelled by nick-translation reconstituted with 0, 0-05, 0-1, 0-25, 0"5 and l pg ncp/assay. (c) As in (b), on piasmid segment NheI-SmaI (267 bp). U, linear unreconstituted DNA. The 6 different complexes are indicated (see the text). (d) Quantitative evaluation of the DNA-dependent differential reconstitution. Values obtained by scanning autoradiograpby of the data reported in (a) to (e). Ordinate, percentage unreconstituted DNA; abscissa, amount of ncp in the reconstitution assay. (I--1) The 524 bp curve-containing segment; (O) the 257 bp curved segment; (A) the 267 bp vector segment.

this size, only a maximum of tbree yeast ncps can be accommodated ((146 + short linker) x 3). The formation of a complex population has been observed and explained by Linxweiler & Horz (1985) on the basis of the differential gel migration of DNA-protein complexes in which the nucleosomes occupy different positions (i.e. centrally or terminally located). A simple combinatory scheme shows tha t six bands can be generated by three different types of occupancy (single, double or triple) with two variable positions (internal or terminal). On shorter segments (Fig. 2(b) and (c)), fewer combinations are possible, terminal effects (i.e. nucleosomal protru- sions) are more relevant and lower resolution of bands is obtained.

The difference of the relative efficiency of reconst i tution among various DNAs is safely evaluated by measuring the disappearance of free DNA. Reconsti tution on plasmid DNA (Fig. 2(c)) requires more ncp than on curved (Fig, 2(b)) or curve- containing (Fig. 2(a)) segments. As judged from the plot of the da ta (Fig. 2(d)), the affinity of curved DNA for histone octamers is six- to sevenfold higher.

Shrader & Crothers (1989) have produced synthetic oligomeric sequences tha t form ncps with high levels of efficiency and have compared the free energy of reconstitution on these sequences relative

to various other nucleosome-positioning sequences of natural origin. Let us make, for comparative purposes, the likely assumption tha t the free energy of reconstitution on the plasmid sequences tested by Shrader & Crothers (1989) is similar to tha t of our plasmid sequence. I f so, our da ta show tha t the C.fasciculata curved DNA favourably compares with the synthetic te t ramer and pentamer sequences and is several-fold more efficient than other natural positioning sequences (namely, 5 S sequences of various origins: see Fig. 3 of Shrader & Crothers (1989) and references therein for comparison).

(i) The reconstituted particles are ncps

The reconstitutes were analysed for resistance to micrococcal nuclease digestion. I t was observed (data not shown) tha t the size of the material protected from endonucleolytic at tack corresponds to the expected value (146 + 3), therefore providing evidence tha t bona fide properly reconstituted neps are formed (Linxweiler & Horz, )985). The size of protected fragments was constant, irrespective of the origin of the nucleosomal core particles used for reeonstitution (yeast or chicken erythrocytes). Fragments corresponding to dimers and trimers were also observed.

366 G: Costanzo et al.

2 3 4 5

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

:, ~a2'

6 7 8 9 I0 M

- - I - - 2

- - 3 - - 4

- - 5 - - 6 - - ( 7 )

- - 8

- - 9 - - I 0

--11 - - I Z

- -15

- -14

- - ! 5

/

~ ~.. --20

--21

--22

23

i _ _ - - 25

l 2 3 4 5 6

O

= , ~ 4 e Q

ep

7 M

Q

- - I - - 2

- - 3 - - 4

- - 5 - - 6 - - 7

- - 8

- - 9 - - I 0

- - I I

--12

--13

--15

- 1 7

I - - 1 8

--19 ,ti

- - 20

- - 2 ! q

I :*~ - - 22

- - - - 23

,l=, ~ I I I ' O - 24

( IIIIpO'Q - - 25

M

622

527

404

509

242 238

217

201

190

180

t60

147

123

(o)

F i g . 3.

( b )

(b) In mixed DNA domains ncps reconstitute preferentially on curved sequences and adopt a

defined periodical distribution

The localization of ncps was analysed on a DNA fragment containing a curved segment (524bp, fragment 4).

(i) Kinetics of Exol I l digestion. Stable borders are due to ncps

The localization of ncps on reconstitutes obtained at a high ncp to DNA ratio was analysed by ExoI I I digestion. Figure 3(a), lanes 9 and ]0, shows tha t a

composite pattern made of 25 resistant borders is (brined. The pattern remains constant for digestions lasting 15 and 40 minutes (lanes 9 and 10), pointing to the stabili ty of the system and showing tha t the ExoI I I borders are actually due to impediments to the passage of the enzyme. These impediments are absent in the control lanes (] to 5).

(ii) The patterns of Exol I I borders change as a function of the increasing concentration of donor ncps

Variations of the distribution of ncps were obtained upon varying the concentration of ncps in


o I /

t6 ', '2,, ,L ' /Us_A.~

I 0.2 ~ i

° ~ U v I 17 14 6~A5 4. 2

(c)

1 2 3 4 5 6 ~ I ~ • V V V V

I I to 0 tst 61

i

4

5

2 / 2 ~ / %""

/ I

• I

0"1 0"2 0"5 I '0 0"15 0 '25

ncp (p4)

(d)

2l 23 25 8 91011121:~14151617181920 22 24 • V V V V V V V T V v ~ I ' V V v V v V

lllllllllllllmli~ili~F:-- ---- lllli,~,,.llll

1200 I 300', I 1400 I 5001 I 0 I

8F I

(e)

Fig. 3.

91 I01

II1" 121

131 141

151 161

171 181

191 201

~, ~_, 231

241

I I

I I

I I

I I

I I

i I

I I

I

368 G: Costanzo et al.

243 ~, SfuI GAATACACGG AATTCGAGCT CGCCCGGGGA TCCGGCCTAA

10 11 12 13 2 4 0 252 263 275

283 AATTCCAACC GAAAATCGCG AGGTTACTTT TTTGGAGCCC

303 313 323 GAAAACCACC CAAAATCAAG GAAAAATGGC CAAAAAATGC

18 19 20 21 325 337 346 357

363 CAAAAAATAG CGAAAATACC CCGAAAATTG GCAAAAATTA

22 23 24 25 367 376 386 395

4 0 3 ACAAAAAATA GCGAATTTCC CTGAATTTTA GGCGAAAAAA

4 4 3 CCCCCGAAAA TGGCCAAAAA CGCACTGAAA ATCAAAATCT

4 8 3 ~.Accl GAACGTCGCC GGATCCTCTA GAGTCGACCT GCAGCCCA

( f )

Figure 3. Localization of ncps on the curve-containing DNA fragment. (a) Kinetics of ExoII I digestion. The NheI-HindIII 524 bp fragment (4), 5'-labelled at the HindIII extremity (see the map in (e)) was digested with 50 units ExoIII/ml (30°C) for 0, 2, 5, 15 and 40 rain (lanes 1 to 5). Lanes 6 to 10, the same on the product of reconstitution (0.5 pg of yeast neps). The numbers on the right of lane 10 indicate the stops of ExoII I caused by ncps. Stop sites (from 1 to 25) are on the lower strand (see the map in (e)). Stop 7 appears occasionally and is reported in parentheses. M, markers; HpaII digest of pBR322. The size of the markers is reported at the right of (b). Polyacrylamide gel electrophoresis in standard DNA sequencing conditions. (b) Localization of ncps borders as a function of the increasing amount of ncps in the reconstitution. DNA as in (a). ncp, 0, 0"1, 0"15, 0-2, 0'25, 0"5, 1 pg (lanes 1 to 7). Markers and numbering as in (a). ExoIII , 50 units/ml, 40 min. In the photographic reproduction, in certain positions the bands due to spontaneous ExoII I stops are not clearly distinguishable from the bands due to the presence of an ncp border, i.e. at positions 4, 17, 19 and 25. The band assignment was made on the basis of underexposed autoradiograms (not shown). (c) Scanning densitometry of data reported in (b). Lines from top to bottom correspond to lanes 1 to 7. The top of the lane is on the right side of the line. The numbering system of the ncps borders is as in (a) and (b). The peaks not numbered are those caused by spontaneous (i.e. not induced by neps) ExoII I stops. (d) The selective positioning of ncps as a function of the ncp concentration in the reconstitution assay. The amount of ncps formed on the probe at each concentration of donor ncps was evaluated singly by the areas of the peaks shown in (c). Selected samples are reported: 24 is the ncp that forms the first; 2 forms the last. At high concentrations of donor neps, octamers form close to the NheI extremity (plasmid sequences), thus impeding access to ExoIII . The fact that the internal particles form first allows their analysis. The fact that NheI-terminal ncps form last interferes with the analysis, at high concentrations of ncps, of the distribution of internal oetamers. The broken lines in the profiles of ncps 24 and 14 refer to this problem. (e) The localization of the reconstituted ncps. Data are from (a) and (b). The arrow indicates the direction of ExoI I I digestion. The triangles are the upstream ncp border, the thin bars indicate the region of 146 bp putatively encompassed by an ncp (except for ncps from 1 to 4, not reported for graphical reasons). The thin bars under triangles 1 and 2 point to the uncertainty of the mapping procedure in the region. The dot is the 5'-labelled position. (f) The localization of reconstituted ncps on the sequence of the insert segment. The numbering refers to the 524 bp fragment (see Fig. 1 and Materials and Methods). The StuI and AccI cleavage sites used to clone the curved fragment (Kitchin et al., 1986) in pPK201/CAT are indicated. The 1st nucleotide indicated is number 1 in the pSP65 vector numbering system (transcription start) and number 243 in the 524 bp DNA fragment used. The triangles indicate the upstream border of the reconstituted ncps, on the lower strand.

the reconst i tut ion assay, as shown by the var ia t ion of the pa t te rn of the E x o I I I borders (Fig. 3(b), lanes 1 to 7). The map and the sequence position of the E x o I I I - r e s i s t a n t borders are shown in Figure 3(e) and (f). Two major facts are evident from the analysis of the differential distr ibution of the E x o I I I borders. (1) The sites from 8 to 25 are regularly spaced. (2) They form at a lower concentrat ion of ncps, before the format ion of the sites 1 to 4. (For sites 5 to 7, see below.)

Sites 8 to 25 are borders of ncps located on the

curved segments, i.e. sites 1 to 4 on the plasmid sequence. (Sites 5 and 6 are borders of ncps t ha t encompass mos t ly vector sequences, site 7 is observed occasionally.)

Figure 3(c) contains the scanning dens i tomet ry of the t i t ra t ion shown in (b). The numbered peaks are the E x o I I I borders. Figure 3(d) repor ts selected instances of ncps format ion as a function of the ncps concentrat ion in the reconst i tut ion assay. D a t a were obta ined by quan t i t a t ive evaluat ion of the areas of the peaks repor ted in Figure 3(c}. S tar t ing


I 2 5 4 5 6 7 8 M I 2 5 4

. . - , ~ : / ~

' ~ ~-:~; - 1 6

. .-, ..:~:;.~! . '.,~ ~;..~..}

- I - 2 -3 - 4 - 5

- 6

- 7

- 8

- 9

- I 0

- I I

-12

-13

Fig, 4.

5 6 7 8 M

O

(a ) b~

from the more preferential, the order of formation is: 24, 17, 6, 12, 14, 5, 20, 21, 15, 16, 13, 22, 10, 8, 9, l l , 4, 3, l, 2.

Preferentiality does not require complete overlap with curved sequences, nor does it increase as a function of the increasing amount of the curved sequences encompassed in the 146bp unit. For preferential formation to occur, it appears that only a short tract of curved DNA is necessary.

From the data described so far, we conclude that the preferentiality of formation of the neps that encompass intrinsically curved sequences is evident and that the relationship between preferentiality and curvature does not follow simple rules. The conformational requirements for optimal formation of ncps are relatively complex and are discussed in section (d), below.

ncps cluster on the right extremity of the fragment in the proximity of the Hindl I I end (Fig. 3(e)) and therefore should prevent access to ExoIII from that extremity. We have experimentally verified that this is indeed the case {data not shown).

(c) The localization of ncps in the curved D N A tract is not affected by flanking sequences

(i) The localization on the curved fragment analysed in the absence of other sequences

The ExoIII pattern observed for neps reconstituted on the curved segment flanked only by short linker tracts (EcoRI-XbaI, 251 bp, fragment 5) is similar to the pattern observed on the same sequences when analysed embedded in a longer DNA domain. Figure 4 shows the regular pattern


EcoRI

!

4 I

I

L 12 I I I 9 8 7 6 5 4 V T T T T ,

, . , . . ~ ... ,,,m n, ,,,, ,,,, ,m, ,m~, . . . . . • • • " ' l i '4;' & s 6789,0,, ,z ,3 ,4 ,5 ,6h4

146

(c)

5 2

,i Xbol

11,$3 142

142 147

Figure 4. Localization of ncps on the curved DNA fragment, ncps were reconstituted, as described, on the curved EcoRI-XbaI fragment (251 bp) with 0-5/~g of yeast ncps. (a) Lower strand. Reconstitution was performed on DNA 5'-labelled at the XbaI extremity (the label is indicated by the filled dot in the map in (c)). ExoIII digestion, 50 units/ml for 0, 5, 15 and 40 min. Lanes l to 4, no ncps; lanes 5 to 8, ncps. The numbem on the right refer to ncps borders. M, markers; HinfI digest of pBR322 + Bsp1286+ HaeIII + Sinai digests of the EcoI~I-XbaI fragment. (b) Upper strand. As in (a) for the same fl'agment 5'-labelled at the EcoRI extremity. Digestion times: 2"5, 5, 15 and 40 rain. M = A +G sequence of the same fragment. (c) Localization of the ncps borders (filled triangles) on both strands. Data from (a) and (b). The lines that connect the borders on the lower and the upper strand identify the presumptive position of each ncp. Thin vertical bars: A-tracts. Thick lines: polylinker.

formed by the borders of the ncps identified on both strands: (a) lower strand; (b) upper strand. In the map (Fig. 4(c)) we connect the borders localized on both strands, relying on the observation (not shown) tha t the two borders are separated by 146( + 3) bp.

The ncps are numbered for comparat ive purposes. The position of ncp l in this short fragment (fragment 5) corresponds to tha t of ncp 13 of the long fragment reported in Fig. 3 (fragment 4): 1 - 5 = 13- 4, 2 - 5 = 14-4, etc.

The same comparison has been made for the ncps localized on the curved segment when encompassed in another DNA domain (DNA fragment 3) and has produced the same results (data not shown).

In conclusion, the localization of ncps in a curved segment is not influenced by the surroundings plasmid sequences.

(ii) The localization of ncps on the curved fragment remains constant in a different sequence context

Fragment 3 contains the curved sequence m a different context (joined to the right arm of the plasmid). Analysis similar to those already reported has shown tha t the localization of ncps in the curved sequence remains constant (see the previous section). In addition, it was observed tha t the position of 11 ncps in this large fragment corresponds to tha t of ncps in the other large f ragment analysed (fragment 4) (not shown).

We have compared the pat tern of the positions of the reconsti tutes obtained with yeas t and chicken ery throcytes ncps. The resulting pat terns were very similar (not shown).

(d) The localization of ncps in plasmid sequences is affected by the presence of ncps in the in cis curved

DN A tracts

Plasmid sequences reconsti tute ncps with lower efficiency than the curved DNA tracts, both when present ill composite domains (Fig. 3 and sections (c)(i) and (ii), above) and when alone (Fig. 2). We have localized the positions of the ncps tha t form (with low efficiency) ill the plasmid sequences in the absence of curved inserts. Figure 5 shows tha t in the relatively large segment composed by the plasmid sequences tha t surround the polylinker (480 bp, 6), only five E x o I I I borders can be detected (Fig. 5(a), lane 8, the 5 arrowed positions), in spite of the high concentrat ion of ncps used. The map shows the localization of the ncps positions, ncps form only on the left arm of the plasmid (upper mapping).

We have reported (Fig. 3) the position of the ncps that form on the same plasmid region (NheI- EcoRI), when par t of a differently composed DNA domain (reported in Fig. 5 in the lower mapping to allow direct comparison). The distribution of the ncps on the same sequences (left segment) when embedded in the two different domains is different.

Why do the ncps in the plasmid component of the composite domain (p lasmid+curved DNA) select positions different from those tha t are selected on the same plasmid sequences when present in a domain made only of plasmid tracts? To formulate an answer, one should recall tha t the ncps form with higher efficiency (Figs 2 and 3) on curved sequences. Therefore, ncps tha t form on the plasmid t rac t of the composite domain do not form on unoccupied

w

o

0 b

l o

o -

-~

~

t rt

t t

0

S"


Nhe T 2 3 TT

2t 31

41

5 T

I

51

I

! I

II IIIIII

PvuTT I

NheZ I I T

21

5 4 5 6 8 ?T T T T

31 4~

I

51 61

8t

910 II 1215 . . , e t c . T T V Y T ............................................................................................... ,

H ind lrr I

I I

............. |

I I

91 I I01 I

III I 121 I

131" I ",

".e~c.

(b)

Figure 5. Localization of ncps on vector sequences. (a) NheI-PvuII plasmid DNA (480 bp, fragment number 6, • labelled at the NheI extremity). Lanes 1 and 5: untreated DNA. ExoIII, 50 units/ml for 5, l0 and 40 rain on free DNA (lanes 2-4) and on the product of reconstitution obtained with 0-5/~g of yeast ncp. Markers: HpaII digest of pBR322. The 5 arrowed positions on the right are the only ncps borders that can be detected. (b) Upper: map assignment of the ncps on the NheI-PvuII fragment (data from (a)). Lower: the ncps map on the same plasmid sequence, when joined to the curved segment (data from Fig. 4), reported to allow direct comparison (see the text).

DNA. On the contrary, they occupy a region of a molecule on which other ncps are already present elsewhere, and which has already been altered, at least in terms of residually available positions. More complex kinds of alterations (i.e. long distance effects, co-operativity) should not be a priori excluded.

This experiment provides evidence that curved sequences may function as organizers of nucleosome arrays.

4. D i s c u s s i o n

We have shown that on intrinsically curved DNA ncps (1) form preferentially and (2) are rotationally phased. Relative to bulk DNA sequences, preferentiality is six- to sevenfold. The preferential formation occurs at multiple positions. The distribution of these positions reveals a unique rotational setting of DNA with respect to the protein surface.

The mechanisms that could cause or influence nucleosome positioning have been discussed (Kornberg, 1981; Palen & Cech, 1984; Strauss & Varshavsky, 1984; AImer et al., 1986; Benezra et al., 1986; Richard-Foy & Hager, 1987; Thoma & Zatchej, 1988). In outline, a set of defined positions could be determined by one or more different causes. (I) Boundary effects and size of the domain. In a limited space, only a defined number of nucleosomes can fit; this spatial limitation would allow only a defined number of alternative positions. The borders of the domain could be set by DNA special structures, by pre-existing nucleosomes or other proteins. (2) Sequentiality of formation: the position of the second nucleosome could be defined by the

position of the first and so on; co-operativity of the process (Forte et at., 1989) could be a relevant component of this mechanism. (3) Higher affinity for one (or for a set of) defined sequence-related conformational properties. Specific combinations of sequences have been identified in the unit of DNA engaged in vivo in nucleosome particles: defined sequences are clearly favoured in certain positions in the 146 bp unit, indicating the role of the local conformation in the preferential positioning of nucleosomes (Travers, 1989 and references therein).

In nucleosomes, DNA should be able to accommodate the deformation imposed by the interaction with a protein surface that causes narrowing of both major and minor grooves on the inside of the curve, and widening on the outside. Therefore, sequence-dependent positioning of nucleosomes is explained in terms of the differential flexibility of different sequences and of departures from smooth bending. Support for this explanation is provided by the facts that DNA is characterized by anisotropic flexibility, that the DNA curvatures caused by this anisotropic property depend on the distribution of differently flexible sequences and that in vivo DNA sequences adopt defined locations with respect to the histone octamer. We have presented evidence that curvature is a property of DNA relevant for the understanding of the preferential locations of histone octamers as predicted by Calladine & Drew (1986).

In our in vitro system, DNA curvature appears to be a major phasing determinant, strong enough to define localization and preference over bulk DNA. The rotational constraints (i.e. the relationship of the grooves to the direction of the curvature) of the


C.fasciculata curved DNA are so strong and so regularly repeated that the analysis of asymmetri- cally located translational signals (Hogan et al., 1987; Satchwell & Travers, 1989) is not possible.

Comparison of in vivo and in vitro nucleosome positioning has been reported (Linxweiler & Horz, 1985) and it was shown that additional constraints or mechanisms act in vivo to determine nucleosome position. The fact that in certain instances DNA bendability appears to be a major energetic determinant of nucleosome formation and localization in vivo (Travers, 1989) and in vitro (this study) suggests that the additional constraint may serve fine tuning or regulatory purposes of a DNA-protein interaction system whose major determinant is DNA conformation itself.

(a) The preferential formation of ncps on a D N A characterized by the regular repetition of sequence

motifs offers a contribution to the interpretation of the linking number paradox

Measurements of closed circular DNA extracted from simian virus (SV40) chromatin (Germond et al., 1975; Simpson et al., 1985) indicated a linking number change near to one per nucleosome, whereas X-ray crystal analysis has shown that in a nucleosome the double helix of DNA is wound into about 1.8 turns of superhelix (Finch et al., 1977). One would therefore expect that the linking number change is 1"8 instead of the observed l'0. The proposed explanation of the paradox requires that the screw of the DNA on the nucleosome and when free in solution differ (Finch et al., 1977; Klug & Lutter, 1981). Upon nucleosome formation, the helical repeat h (the number of base-pairs per unit winding number h = N / ¢ , where N is the total number of base-pairs involved and (I) is the winding number) changes and from 10.6 bp per turn (Wang, 1979; Rhodes & Klug, 1980; Peck & Wang, 1981) becomes 10"17 (Drew & Travers, 1985), This value has been calculated with an external probe (DNaseI) and is consistent with the observed change in linking number.

A recent theoretical analysis of the problem reaffirms, for the solution of the apparent topologi- cal paradox, that the helical repeat changes, upon nucleosome formation, from the value in solution (10"6) to a value closer to 10"0 bp per turn (White & Bauer, 1986, 1989; Klug & Travers, 1989). The possibility that the twist of DNA does not actually change in the DNA wrapped around a histone core has been raised (Morse & Simpson, 1988; Zivanovic et at., 1988).

We have presented evidence that ncp formation is highly favoured by the phased repetition of defined sequence motifs. This is predicted by the observation that certain motifs, i.e. (A), and (T), runs, assume specific positions relative to the octamer surface (Satehwell et al., 1986). The DNA analysed here allows the preferential formation of a large and regular ensemble of rotationally phased octamers. This offers, for the first time, the possibility of

correlating such preference with a periodic struc- tural property. Independently from the precise nature of this property, what is actually of interest here is the period of the phase (i.e. 10"5 or different from 10"57) and its correlation with the experimentally observed period of the deposited octamers. A best-fit analysis (not detailed) of the positions of the ncps with the positions of the (A), and (T), runs shows a periodicity of 10.26+0-04. Therefore, sequences with a physical property that repeats itself with a period of 10-26 preferentially attract octamers. This observation provides independent support to the interpretation that the solution of the linking number paradox is in the reduction of the helical repeat of DNA in the nueleosome.

In summary, we have shown that the constrained topology of DNA on the surface of the octamer complex is stabilized by intrinsic curvatures whose direction and amplitude are similar to that of the DNA on the nucleosome. Such intrinsic curvatures require the correct alignment of flexible DNA sequences and result in the decrease of the binding energy necessary for ncp formation. Therefore, in curved sequences ncps form preferentially relative to regions where intrinsic curvature is absent.

Reconstitution experiments on segments that contain both curved and non-curved sequences have shown that site-selection actually occurs and that the presence of ncps in the curved segments affects the deposition of other ncps in the surrounding regions (Fig. 5).

This work was supported by Fondazione Istituto Pasteur-Fondazione Cenci Bolognetti and by Piano Finalizzato Bioteenologie e Biostrumentazione (CNR, Italy).

We thank M. Savino and M. Buttinelli for a generous supply of chicken erythrocyte ncp, and P. T. Englund for the pPK201/CAT plasmid.

References

AImer, R. H., Hinnen, A. & Horz, W. (1986). EMBO J. 5, 2689-2696.

Benezra, R., Cantor, C. R. & Axel, R. (1986). Cell, 44, 697-704.

Calladine, C. R. & Drew, H. R. (1986). J. Mol. Biol. 192, 907-918.

Drew, H. R. & Calladine, C. R. (1987). J. Mol. Biol. 195, 143-173.

Drew, H. P~. & Travers, A. A. (1985). J. Mol. Biol. 186, 773-790.

Finch, J. T., Lutter, L. C., Rhodes, D., Brown, R. S., Rushton, B., Leavitt, M. & Klug, A. (1977). Nature (London). 269, 29-36.

Forte, P., Leoni, L., Sampaolese, B. & Savino, M. (1989). Nud. Acids Res. 17, 8683-8694.

Germond, J. E., Hirt, B., Oudet, P., Gross Bellard, M. & Chambon, P. (1975). Proc. Nat. Acad. Sci., U.S.A. 72, 1843-1847.

Hogan, M. E., Rooney, T. F. & Austin, R. H. (1987). Nature (London), :}28, 554-557.

Hsieh, C. H. & Griffith, J. D. (1988). Cell, 52, 535-544. Kitchin, P. A., Klein, V. A., Ryan, K. A., Gann, K. L.,

Ranch, C. A., Kang, D. S., Wells, R. D. & Englund, P. T. (1986), J. Biol. Chem. 261, 11302-11309.


Klug, A. & Lutter, L. C. (1981). Nucl. Acids Res. 9, 4267-4283.

Klug, A. & Travers, A. A. (1989). Cell, 56, 10-11. Kornberg, R. (1981). Nature (London), 292, 579-580. Linxweiler, W. & Horz, W. (1985). Cell, 42, 281-290. Morse, R. H. & Simpson, R. T. (1988). Cell, 54, 285-287. Negri, R., Costanzo, G., Venditti, S. & Di Mauro, E.

(1989). J. Mol. Biol. 207, 615-619. Palen, T. E. & Cech, T. R. (1984). Cell, 36, 933-942. Peck, L. J. & Wang, J. C. (1981). Nature (London), 292,

375-378. Pennings, S., Muyldermang, G., Meersseman, G. &

Wyns, L. (1989). J. Mol. Biol. 207, 183-192. Richard-Foy, H. & Hager, G. L. (1987). EMBOJ. 6,

2321-2328. Rhodes, D. & Klug, A. (1980). Nature (London), 286,

573-578. Satchwell, S. C. & Travers, A. A. (1989). EMBOJ. 8,

229-238. Satchwell, S. C., Drew, H. R. & Travers, A. A. (1986).

J. Mol. Biol. 191,659-675.

Shrader, J. C. & Crothers, D. M. (1989). Proc. Nat. Acad. Sci., U.S.A. 86, 7418-7422.

Simpson, R. T., Thoma, F. & Brubaker, J. M. (1985). Cell, 42, 799-808.

Strauss, F. & Varshavsky, A. (1984). Cell, 37, 889-901. Thoma, F. & Zatchej, M. (1988). Cell, 55, 945-953. Travers, A. A. (1989). Annu. Rev. Biochem. 58, 427-452. Travers, A. A. & Klug, A. (1987). Phil. Trans. Roy. Soc.

set. B, 317, 537-561. Trifonov, E. N. (1980). Nucl. Acids Res. 8, 4041-4053. Trifonov, E. N. (1985). CRC Crit. Rev. Bioehem. 19,

89-106. Trifonov, E. N. & Sussman, J. L. (1980) Proc. Nat. Acad.

Sci., U.S.A. 77, 3816-3820. Wang, J. C. (1979). Proc. Nat. Acad. Sci., U.S.A. 76,

200-203. White, J. H. & Bauer, W. M. (1986). Proc. Nat. Acad. Sci.,

U.S.A. 85, 772-776. White, J. H. & Bauer, W. R. (1989). Cell, 56, 9-10. Zivanovic, Y., Goulet, I., Revet, B., Le Bret, M. &

Prunell, A. (1988). J. Mol. Biol. 200, 267-290.

Edited by A. Klug

attraction, phasing and neighbour effects of histone octamers on curved dna

Documents