Tài liệu Báo cáo khoa học: The sequentiallity of nucleosomes in the 30 nm chromatin fibre pptx

In the present study, we have chemically crosslinked adjacent nucleosomes along the helix of chicken erythrocyte oligonucleosome fibres, digested the inter-nucleosomal linker DNA and then

chromatin fibre

Dontcho Z Staynov1and Yana G Proykova2

1 Imperial College London, National Heart and Lung Institute, UK

2 School of Earth and Environmental Sciences, University of Portsmouth, UK

The DNA is packed on several levels as chromatin in

the eukaryotic nucleus The first level of packing,

the highly conserved nucleosome, allows

transcrip-tion, after remodelling and⁄ or histone modifications ⁄

replacements The nucleosome core particles have been

reconstituted and crystallized and their structure solved

in detail at 1.9 A˚ resolution [1–3] The second level of

packing is the transcriptionally dormant 30 nm

chro-matin fibre Understanding its structure, as well as the

processes that determine its folding and unfolding, is a

prerequisite for studying the epigenetic mechanism,

which leads to poised-for-transcription or dormant

chromatin [4] The fibre consists of the entire

chroma-tin of the nucleated avian erythrocytes and comprises

approximately 85% of the chromatin in other cell

types [5]

The structure of chicken erythrocyte chromatin is

the most widely studied in the whole nucleus, as well

as in solution Using small angle X-ray and neutron

scattering, it has been shown that all the high

mole-cular weight material that diffuses out of the nuclei after micrococcal nuclease (MNase) digestion is in the

30 nm fibre conformation It consists of a regular helix with a diameter of approximately 33 nm and a variable mass per unit length, which approaches 0.6 nucleo-somesÆnm)1with an 11 nm pitch at 80 mm salt concen-trations This implies that there are seven nucleosomes per helical turn with their flat surfaces almost parallel

to the fibre axis [6–11] The unusually small cross-sectional radius of gyration (9.5 nm at 80 mm salt) suggests a very compact structure with close nucleo-some–nucleosome contacts

There are several basic models for the structure of the fibre that were proposed in the late 1970s and early 1980s, and some variants have been published subse-quently [4,5,12] They all comprise regular helices of more or less seven nucleosomes per turn and thus approximately satisfy the results obtained by small angle X-ray and neutron scattering and low resolution electron microscopy with respect to the packing of

Keywords

30 nm fibre; chromatin structure;

nucleosome

Correspondence

D Z Staynov, Imperial College London,

National Heart and Lung Institute, Guy

Scadding Building, Dovehouse Street,

London SW3 6LY, UK

Tel: +44 207 6223644

E-mail: d.staynov@imperial.ac.uk

(Received 29 March 2008, revised 20 May

2008, accepted 23 May 2008)

doi:10.1111/j.1742-4658.2008.06522.x

The folding of eukaryotic DNA into the 30 nm fibre comprises the first level of transcriptionally dormant chromatin Understanding its structure and the processes of its folding and unfolding is a prerequisite for under-standing the epigenetic regulation in cell differentiation Although the shape of the fibre and its dimensions and mass per unit length have been described, the path of the internucleosomal linker DNA and the sequential-lity of the nucleosomes in the fibre are poorly understood In the present study, we have chemically crosslinked adjacent nucleosomes along the helix of chicken erythrocyte oligonucleosome fibres, digested the inter-nucleosomal linker DNA and then examined the digestion products by sucrose gradient sedimentation We found that the digestion products con-tain considerable amounts of mononucleosomes but less dinucleosomes, which suggests that there are end-discontinuities in the fibres This can be explained by a nonsequential arrangement of the nucleosomes along the fibre helix

Abbreviations

as, acid soluble; DSP, dithiobis-(succinimidyl propionate); EDC, 1-ethyl-3(3-dimethylaminopropyl)-carbodiimide; MNase, micrococcal nuclease.

nucleosomes in the fibre However, they were proposed

before the crystal structure of the nucleosome was

solved and do not take into account the topological

constraints imposed on the relationship with respect to

nucleosome orientation and tilt versus chromatin

repeat length Thus, they differ with respect to the

ori-entation of the nucleosomes and the path of the linker

DNA within the fibre

High definition structures have not been achieved

because the native fibres comprise a mixture of

differ-ent repeat lengths and could not be crystallized To

avoid this problem, several studies recently

reconsti-tuted oligonucleosome arrays on a nucleosome by

positioning DNA sequence-repeats differing by

multi-ples of 10 or 11 bp [13–16] Dorigo et al [14] used

cys-teine substituted recombinant core histones with or

without linker histone The electron micrographs of

their reconstitutes show flat ribbons with

approxi-mately five instead of seven nucleosomes per 11 nm,

which do not fold into helical fibres, although they

refer to them as two-start helices To study nucleosome

sequentiallity in their reconstitutes, Dorigo et al [14]

crosslinked the samples and digested the linker DNA

with a restriction enzyme The resultant

oligonucleo-somes migrate in a ‘native’ gel as a ladder with dimers

up to half the size of the original material and support

a two-start helix In a subsequent study by Schalch

et al [16], a reconstituted tetranucleosome with a

20 bp linker DNA was crystallized It was speculated

that this construct allows a two-stranded helix with

close nucleosome contacts in which the flat surfaces of

the nucleosomes are perpendicular instead of parallel

to the fibre axis

Very different results were obtained by Robinson

et al.[15] using native chicken erythrocyte core histones

and H5 linker histone They observed two helical

arrays: one for repeat lengths below 210 bp with a

diameter of 33 nm and another for repeat lengths above

210 bp with a diameter of 45 nm and with the flat

surfaces of the nucleosomes close to parallel to the fibre

axis The overall shapes of their reconstitutes are very

similar to the fibres observed in ‘native’ chromatin

Most striking are the two very different structures

presented by the two groups for the 177 bp as well as

the 207 and 208 bp repeat lengths, which differ by the

presence⁄ absence of the linker histone Apparently,

the reconstitutes of the two groups cannot represent

the same structure and additional evidence is needed

Both groups have discussed their results with respect

to discriminating between single-start (sequentially

arranged nucleosomes) and two-start nonsequential

helices Other possible nonsequential helices were

ignored Neither group considered the very

informa-tive results obtained by DNase I digestion of nainforma-tive chromatin, which produces ‘dinucleosome repeat’ pat-terns Such patterns in which the even multiples are strong can be produced only if the adjacent nucleo-somes are digested at alternating sites and, thus, the odd multiple fragments are attenuated and the even multiple fragments dominate the pattern These results unambiguously show that there is a common structure

of the fibre in which the consecutive nucleosomes in samples of several different repeat lengths have alter-nating orientations, as extensively discussed elsewhere [5,12,17,18]

The question of the sequentiallity of the nucleo-somes in the fibre is essential Because a variety of higher order structures might be capable of reconstitu-tion with a repeat-sequence DNA, a key quesreconstitu-tion is how do the reconstitutes of the two groups compare with the fibres obtained from natural chromatin?

In the present study, we examined the sequentiallity of the nucleosomes in 30 nm fibres, which diffuse out of chicken erythrocyte nuclei after a mild MNase diges-tion without further manipuladiges-tions We used the ratio-nale of Dorigo et al [14], which involved crosslinking and nuclease digestion To demarcate the adjacent nucleosomes along the fibre, we used nonspecific protein–protein crosslinkers with two different spans: (a) dithiobis-(succinimidyl propionate) (DSP) (also known as Lomant’s reagent), a cleavable bifunctional ester with 1.2 nm span and a noncleavable, contact-site crosslinker and (b) 1-ethyl-3(3-dimethylaminopropyl)-carbodiimide (EDC) Subsequently, the samples were redigested with MNase and fractionated by sedimenta-tion on sucrose gradients Instead of obtaining half the size of the original material, we observed only a slight decrease of its size and a considerable number of mononucleosomes These results do not support the two-start helix arrangement, but a higher order nonse-quential arrangement of the nucleosomes in the fibre with end-defects as described below

Experimental rationale

It has been shown that, at moderate ionic strength dif-ferent, crosslinkers can covalently crosslink linker- and core-histones beyond a single nucleosome and thus are able to demarcate adjacent nucleosomes along the helix of the 30 nm fibre [5] In the present study, we used internucleosome histone crosslinking and subse-quent nuclease digestion to distinguish between differ-ent arrangemdiffer-ents of the nucleosomes in the fibre The rationale is illustrated in Fig 1 Three topologically different arrangements of the nucleosomes along the fibre have been suggested [5]

A sequential arrangement

The order of nucleosomes along the fibre helix follows

their order along the DNA If a 9-mer fragment of

continuous helix of nucleosomes (Fig 1Aa) is

exten-sively crosslinked, the adjacent nucleosomes will be

crosslinked and the continuity of the linker DNA will

not be required to keep them together Figure 1Ab

shows the sedimentation profile of an oligonucleosome

sample comprising 7- to 9-mers and half the quantity

of 6- to 10-mers After 100% crosslinking and nuclease

digestion to mononucleosome size DNA, the

sedimen-tation profile of the sample remains the same The

80% crosslinked sample will show a decrease of the

average size of the original oligonucleosome sample and smaller size oligomers will appear (Fig 1Ac) If the nucleosomes are interdigitated, as shown by Rob-inson et al [15], there will be more internucleosomal crosslinks, which will stabilize the fibre structure, and the sedimentation profiles of digestion products will be intermediate between those shown in Fig 1Ab,Ac

A multi-start helix Nucleosomes are arranged in a multistrand sequence and are not consecutive Figure 1Ba illustrates a rib-bon, which can fold into a two-start helix fibre, with linkers either parallel or perpendicular to its axis

Fig 1 Schematic presentation of different nucleosome arrangements in the 30 nm fibre and the expected sucrose gradient sedimentation profiles after 100%, 90% or 80% chemical crosslinking of nucleosomes and digestion of the DNA to mononucleosome size of samples con-sisting of a mixture of hepta-, octa- and nonanucleosomes and half the amount of hexa- and decanucleosomes (6- to 10-mer sample) (A) (a) Nonamer of a sequential single helix (b, c) Sedimentation profiles after crosslinking and subsequent digestion of all DNA linkers of the 6- to 10-mer sample: (b) 100% crosslinked and (c) 80% crosslinked (B) (a) Nonamer in a two start nonsequential (ribbon) arrangement Numbers indicate the number of consecutive nucleosomes along the DNA chain (b–d) Sedimentation profiles of the oligonucleosome sample before digestion (b) and after digestion of 100% crosslinked (c) and 90% crosslinked sample (d) (C) (a) Hexamer in a single helix nonsequential arrangement (b–d) Sedimentation profiles of the 6- to 10-mer sample: (b) original, (c) after digestion of 100% crosslinked and (d) 90% cross-linked sample The thick red line demarcates adjacent nucleosomes, expected to be crosscross-linked Numbers under the horizontal axes of the sedimentation profiles denote mono- di-, etc nucleosomes.

Digestion of the same mixture of 6- to 10-mers

(Fig 1Bb) of a 100% crosslinked sample will produce

half the size of the original sample (Fig 1Bc) If the

crosslinking is not complete, smaller size oligomers will

appear (Fig 1Bd) but the maximum of the main peak

will be approximately half the size of the original

sample

Single-stranded helices with nonsequentially arranged

nucleosomes

Due to nucleosomes’ nonsequentiallity, the fibres have

end-defects (not envisaged in either of the structures

shown in Fig 1A,B); namely, one or two missing

end-nucleosomes, as well as one or two end-nucleosomes

separated from the rest by longer distances Thus,

these end-nucleosomes are probably non-interacting

with the continuous helix and may not be crosslinked

to the rest One such arrangement, the (–3,5)

arrange-ment, is shown in Fig 1Ca [18] Thus, nuclease

diges-tion will shorten the size of the original sample on

average by three nucleosomes and will produce a

frac-tion of 3⁄ n mononucleosomes, where n is the average

number of nucleosomes per fragment The expected

sedimentation profiles of the same 6- to 10-mer

sam-ple, before and after nuclease digestion, are shown in

Fig 1Cb–d Because the closely interacting

nucleo-somes are always an even number (Fig 2), digestion

will produce a mononucleosome fraction and a mix-ture of even multiples If some of the end-nucleosomes are crosslinked to the rest via linker–linker or linker– core histone crosslinks, the mononucleosome fraction will be less than 3⁄ n and crosslinking will produce some odd number oligonucleosomes in the digest Thus, the sedimentation profile might not be as clear-cut as shown in Fig 1Cc,d, but there would be a mononucleosome fraction and enriched even-multiples

of oligonucleosomes in the main fraction Incomplete (90%) crosslinking will also produce some odd number oligomers

As shown in Fig 1, the differences among the expected sedimentation profiles of the oligonucleosome samples after crosslinking and MNase digestion are expected to be considerable and some incomplete crosslinking, or cross-chain crosslinking, will not change their characters

Figure 2 shows that, in the nonsequential (–3,5) arrangement, the number of nucleosomes making close contacts is always even [18] Thus, there are no close contacts in the di- and trinucleosomes, whereas only two nucleosomes are close to each other in tetra- and pentanucleosomes In hexanucleosomes, there are four nucleosomes in close proximity

Results

Chicken erythrocyte nuclei were digested with MNase and the material that diffused out of the nuclei was fractionated by sedimentation through sucrose gradi-ents All chromatin samples from dinucleosomes to high molecular weight material contained indistin-guishable ratios of linker to core histones (see supple-mentary Fig S1)

Chromatin crosslinked with DSP DSP has been used previously for histone crosslinking

to establish the proximity of different histones inside the nucleosome or of nucleosomes in the fibre [5,19] It

is a cleavable crosslinker with two succinimidyl groups, which react with lysines independently The maximum span of crosslinking is 1.2 nm

A sample of oligonucleosomes, consisting mainly of tetra-, penta-, hexamers, minor tri- and heptamer com-ponents with an average number of nucleosomes in the main peak of 4.9, was extensively crosslinked It was digested with MNase for different lengths of time and agarose gel electrophoresis of the DNA exhibited the characteristic nucleosome repeat (not shown) After removal of free crosslinker by dialysis, it was fraction-ated on sucrose gradients Figure 3A,B shows that the

hexa

1

2 2

2

3

3

3

3

4

5

5

6

Fig 2 Schematic presentation of di- to hexanucleosomes in the

(–3,5) arrangement The thick red lines illustrate closely spaced

nucleosomes Numbers indicate consecutive nucleosomes along

the DNA chain.

crosslinking resulted in a partial loss of resolution and

a drop in the sedimentation velocities Digestion with

MNase produced 10% mononucleosomes and as little

as 3.5% dinucleosomes (Fig 3C) The average number

of nucleosomes per chain in the main peak decreased

to 4.1 One third of the optical density sedimented slower than the mononucleosome fraction It com-prised an acid soluble (as) oligo-nucleotide fraction (14%) at the top of the gradient, and a well-defined band, S, of mononucleosome-size naked DNA (17%) Further MNase digestion (Fig 3D,E) led to an increase of band S by up to 50%, but did not change the overall result; the proportion of mononucleosomes increased to 15% and dinucleosomes to 7% The main peak was centered at tetranucleosomes (approximately 40%) with the even numbers (2-, 4- and 6-mers) slightly more pronounced than the odd numbers (3- and 5-mers) At longer times of digestion, more than 50% of the sample was converted into the frac-tion S Breaking the disulfate bond in the middle of the crosslinker produced only mononucleosomes and the band S DNA gel electrophoresis of the fractions from Fig 3D,E showed that approximately 90% of the DNA in the main peak as well as the mononucleo-somes and band S are all in the 140–160 bp size interval (not shown) The high percentage of mono-nucleosomes obtained after MNase digestion with only

a small amount of dinucleosomes, as well as the grad-ual decrease of the number of nucleosomes in the main peak, indicates that crosslinking is almost complete in the middle of the fibre, but some of the end-nucleo-somes are not crosslinked to the rest and therefore must originate from end-defects in the fibre In differ-ent experimdiffer-ents, the mononucleosome fraction was always in the range 0.9–1.6 per chain (and often higher than 1.0)

A sample of eight to 12 nucleosomes was cross-linked, digested with MNase, and further digested with trypsin for different lengths of time The sedimentation profiles are shown in Fig 4 It is seen that MNase (Fig 4B) produces a similar profile as in Fig 3B, with prominent fractions as, S and mononucleosomes, but that di- to penta-nucleosomes are of negligible amounts due to the larger size of the starting material

F

E

D

C

B

A

as

Distance from top of the gradient

Fig 3 UV absorbance profiles of sucrose gradients of an

oligonu-cleosome sample mainly comprising tetra- penta- and hexamers

and minor tri- and heptamer components (A) Control (no

crosslink-ing) (B–E) Extensively crosslinked with DSP and digested with

20 unitsÆmL)1 MNase for 0, 8, 16 and 32 min respectively (F)

Showing the sample used in (E) but reduced to break the

crosslin-ker Numerals 1, 2, etc., denote mono-, di-, etc., nucleosomes.

Fig 4 UV absorbance profiles of sucrose gradients of a sample of eight to 12 nucleosomes (A) Control (no crosslinking) (B–F) Exten-sively crosslinked with DSP, digested with 20 unitsÆmL)1 MNase for 20 min and subsequently digested with trypsin for 0, 1.5, 5, 15 and 45 min Numerals 1, 2, etc., denote mono-, di-, etc., nucleo-somes.

Short trypsin digestion times caused an increase of the

mononucleosome fraction without appearance of di- to

penta-nucleosomes (Fig 4C,D,E) and all the

oligonu-cleosome fractions appeared only after a long digestion

(Fig 4F)

Histone gel electrophoresis of the samples

cross-linked with DSP showed that crosslinking of core

particles produced a histone octamer, whereas

cross-linking of oligonucleosomes produced higher multiples

corresponding to their size, as reported previously [5]

It is not clear why treatment with DSP slows down the

sedimentation of all oligonucleosomes including the

mononucleosome fraction The final product of MNase

digestion of all the fractions, including S, is

mononu-cleosome size DNA and the ten nucleotide periodicity

is preserved in the DNase I digests (not shown)

Tryp-sin digestion leads to digest products of the same limit

as that observed in the untreated nucleosomes (see

supplementary Fig S2) This suggests that, in all

prob-ababilty, DSP changes the buoyant density of the

samples without changing the structure of the

nucleo-somes Bearing in mind that the two-succinimidyl

groups react independently, some lysines that normally

interact with DNA are bound by one of these groups

Therefore, several of their positive charges are

neutral-ized and the histone-DNA interactions are weakened,

with or without the establishment of covalent bonds

with other lysines Apart from the as oligonucleotides

at the top of the gradients, the remainder of the DNA,

including the naked DNA in fraction S, comes from

digested nucleosomes Thus, after crosslinking, the

nucleosomes must have been intact However, these

nucleosomes become less stable and some of them do

not survive the subsequent dialysis

Chromatin crosslinked with EDC

The water-soluble carbodiimide EDC has been used to

crosslink H1-histone to itself and to core histones

Although it is noncleavable and does not allow easy

identification of the crosslinked products, it offers

some important advantages over the cleavable

crosslin-kers First, it is a contact-site (zero length) crosslinker

and thus it excludes long-range bridges between

non-interacting amino acids Second, it binds to an acidic

aminogroup first, and only subsequently makes a

pep-tide bond with an adjacent lysine [20] Thus, it does

not interfere with the majority of the lysines that

inter-act with DNA and the chromatin structure is less

likely to be damaged

In a repetition of the experiments shown in Figs 3

and 4, a sample comprising 6–18 nucleosomes per

chain was crosslinked with EDC and sedimented in a

sucrose gradient (Fig 5A) Extensive digestion with MNase, which broke more than 90% of the DNA linkers according to DNA electrophoresis, produced 13.5% mononucleosomes (approximately two per chain), 0.5% dinucleosomes and a negligible amount

of tri- and tetranucleosomes (Fig 5B) The average size of the oligonucleosome fraction decreased from approximately 14 to 12 nucleosomes per chain This result is principally as that obtained by DSP cross-linking, although crosslinking does not change the sedimentation velocity of the samples and does not produce the free DNA fraction S It is interesting that MNase digestion produces less than three mononucleo-somes per chain, even after crosslinking with this zero length crosslinker Most probably, some of the end-nucleosomes are crosslinked to the rest via linker– linker or linker–core histone links Digestion of this material with trypsin initially caused a gradual decrease of the number of nucleosomes per chain in the main peak, with a corresponding increase in the

Fig 5 UV absorbance profiles of sucrose gradients of an oligonu-cleosome sample of six to 18 nuoligonu-cleosomes per chain crosslinked with EDC (A) Crosslinked but undigested The profile of the un-crosslinked sample is identical to (A) (not shown) (B–D) Digested with 20 unitsÆmL)1MNase at 37 C for 40 min and with trypsin for (B) 0 min; (C) 0.5 lgÆmL)1trypsin for 6 min; and (D) 2 lgÆmL)1 tryp-sin for 30 min.

percentage of mononucleosomes by up to 25%;

approximately 2.7 nucleosomes per chain (Fig 5C)

The oligonucleosome fraction has a maximum at ten

nucleosomes and two shoulders around six and eight

nucleosome sizes There are other shoulders beyond

ten nucleosomes but their sizes cannot be estimated

accurately (Fig 5C) As with the samples that were

crosslinked with DSP, after extensive digestion with

trypsin, almost all of the material was converted into

mononucleosomes (Fig 5D) This experiment was

repeated several times with preparations consisting of

a different number of nucleosomes per chain and

crosslinked for different lengths of time (from 30 min

up to 5 h) in a 30–80 mm Na+ion concentration The

proportion of mononucleosomes in the sucrose

gradi-ents after MNase digestion was always more than two

per chain Most probably, the histones were

cross-linked mainly via their N- and C-terminal tails and

extensive digestion with trypsin separated them from

each other (Fig 5D)

Because the EDC crosslink is not cleavable, how

far the histones are digested by trypsin in the

oligo-nucleosome fraction cannot be determined but trypsin

digestion of crosslinked mononucleosomes shows very

similar digest limits to the noncrosslinked

mono-nucleosomes (see supplementary Fig S3) Crosslinking

with EDC does not change the sedimentation

veloci-ties, nor does it cause sliding of the nucleosomes

along the DNA The repeat length is preserved and

the background is low (see supplementary Fig S4A)

It only slows down the digestion rate of the MNase

to approximately one quarter of the original rate (see

supplementary Fig S4B) Prior trypsin digestion of

this material increases the rate of digestion by MNase

several fold without causing nucleosome sliding (see

supplementary Fig S4C) There is no indication that

DNA is crosslinked to histones and thereby protected

from nuclease digestion Some crosslinked samples

were dialyzed against distilled water and dried

Poly-acrylamide gels of this material were stained first for

DNA and subsequently for proteins It was seen that

the DNA and histones moved independently

Further-more, when this material was first dissolved in 0.25 m

HCl and the supernatant was precipitated with 20%

trichloroacetic acid, no DNA was observed in the gels

(results not shown) Thus, the histones are not

cross-linked to DNA and the reduction in digestion rate

and its recovery after partial trypsin treatment

suggests that the histone–histone crosslinking has

introduced some steric hindrance to nuclease around

the DNA linkers (i.e the linkers are buried inside the

fibre)

Discussion

When compared with the expected results from the three topologically different arrangements in the Experimental rationale, our results are incompatible with the single-helix sequential and the two-start helix arrangements (Fig 1A,B) because neither would yield end-of-fibre discontinuities The results are consistent, however, with the (–3,5) nonsequential arrangement shown in Fig 1C [18] or some other unenvisaged non-sequential nucleosome arrangement The sedimentation profiles of the chromatin fragments digested with MNase after crosslinking with two very different cross-linkers show remarkable similarities and must reflect the actual proximities of the nucleosomes in the fibre There is a considerable amount of mononucleosomes and much less di-⁄ trinucleosomes in the products The mononucleosomes evidently come from the ends

of the fibre because of the corresponding decrease of the average number of nucleosomes per chain in the main peak Digestion of the EDC crosslinked samples with MNase produced less than the three mononucleo-somes per chain expected for the (–3,5) arrangement, but partial trypsin digestion, which cuts the linker his-tone tails first, increased their number to approxi-mately three per chain, with a negligible increase of di- and trinucleosomes In other similar experiments with EDC, there were between two and three mono-nucleosomes per chain Thus, some of the end-nucleo-somes are crosslinked to the rest, even by the zero-length crosslinker, perhaps via linker histones H1 and H5 Indeed, there is evidence that the tails of these histones follow the path of the linker DNA [5] They are crosslinked first to each other first and subse-quently to the core histones [19] The even number nucleosome fragments (six, eight and ten) is more prominent in the main peak, as expected from Fig 2 The rest of the nucleosomes must interact either directly or via histone tails and are also crosslinked by the zero-length crosslinker Early in trypsin digestion, when the tails of all H1 and H5 and some H3 histones were digested, only end-nucleosomes become separated (Fig 5C), whereas, after breaking the core histone tails, all oligonucleosome sizes appeared and, when the trypsin limit digest was approached, the whole sample was converted to mononucleosomes (Fig 5D) Two alternative explanations of these results were consid-ered but both appear very unlikely

In the first explanation, oligonucleosomes smaller than 6-mers have different hydrodynamic behaviour compared to longer oligomers and such short oligo-mers might not fold into a fibre [21] Such oligooligo-mers

would not be crosslinked and, after MNase digestion,

they will produce only mononucleosomes, whereas, in

the longer oligomers, adjacent nucleosomes will be

crosslinked to each other If this were the case, the

sucrose gradient profiles would be very different from

those observed Because the mononucleosomes would

be produced from the shorter oligomers, the average

number of nucleosomes in the main peak would

increase, and not decrease as actually observed

More-over, the gradients shown in Fig 3C–E show that

con-siderable amounts of nucleosomes are crosslinked,

even in a mixture of tetra- to hexanucleosomes, and

suggest that there are closely interacting nucleosomes

in such short oligonucleosomes

For the second explanation, it was reported that

oligomers tend to lose some H1 and H5 histones and

the loss is approximately inversely proportional to the

size of the fragments [22] It can be speculated that, in

a sequentially folded fibre, the lost H1 and H5 histones

come from the ends of the fibre and the

end-nucleo-somes therefore no longer interact with the remainder

This explanation is highly unlikely for the two reasons

First, early during trypsin digestion when it is mainly

the linker histones that are cut, more end-nucleosomes

are converted into mononucleosomes These

nucleo-somes must therefore have been crosslinked to the rest

via the linker histones (i.e linker histones were present

at the ends of the fibre) Second, the loss of H1 and

H5 depends on the procedure of chromatin extraction

We used the same protocol of mild digestion with

MNase to footprint H1⁄ H5 histones on the

chromato-some and found that only the mononucleochromato-some

frac-tion loses some of the linker histones The ratio of H5

to H4 histones (which have similar abundances and

can be assessed quantitatively) in the dinucleosome

sample is indistinguishable from that of high-molecular

weight chromatin Furthermore, in the DNase I

foot-print of the dinucleosome, the band resulting from a

cut on the dyad axis (Band S0 at 70 bp) becomes

undetectable as a result of protection by linker histone

[23]

How then do the present results compare with those

from the reconstituted fibres of Dorigo et al [14] and

Robinson et al [15]? Both groups show real regular

structures, as indicated by electon micrographs They

both considered two cases: a continuous helix and a

two-strand fibre and they neglected any possible

non-sequential arrangement of the nucleosomes (Fig 1C)

Nor did they consider the topological constraints

imposed by the length of the linker DNA on the tilts

of the nucleosomes, as discussed elsewhere [24]

Dorigo et al [14] studied the sequentiallity of

the nucleosomes in the fibre using crosslinking and

digestion rationale but with cysteine-substituted recom-binant histone mutants They obtained unambiguous proof that their reconstitutes are two-start ribbons The most probable explanation for the discrepancy between the results obtained by Dorigo et al [14] and those of the present study is that we examined differ-ent structures and the reconstitutes used do not repre-sent the native 30 nm fibre of higher eukaryotes In the selected micrographs of Dorigo et al [14] (see Figs 4 and S2 therein), the samples appear less con-densed and look very different from previously pub-lished micrographs of 30 nm fibres [25], as well as their Fig 1 The reconstituted decamers, which are large enough to be in the fibre conformation [21], are flat ribbons with approximately five instead of seven nucleosomes per 11 nm

The reconstituted oligonucleosomes of Dorigo et al [14] can be crosslinked in the absence of H1 histone, suggesting that H1 is not required for the close nucleo-some–nucleosome contacts However, H1 histone is essential for the fibre stability of chicken erythrocyte chromatin [5,26] The loss of linker histone causes exposure of the linker DNA to DNase I [27] and leads

to shortening of the chromatin repeat length in the mouse [28] It can be speculated that the particular repeat lengths used by Dorigo et al [14] bring the nucleosomes into contact Other causes for the differ-ences between reconstituted and native fibres might be the type of the crosslinking used, which, in the study

by Dorigo et al [14], comprised selective crosslinking using cysteine modified core histones that may facili-tate direct nucleosome interactions The crystallized tetranucleosome [16] from the same laboratory is out-side the repeat length interval of higher eukaryotic chromatin and might have relevance to viral, telomeric

or yeast chromatin in which the presence of linker his-tone is questionable Furthermore, the presence of all oligomers, with dimers up to the half the size of the original samples after digestion of the linkers, was sug-gested by Dorigo et al [14] to indicate incomplete crosslinking However, incomplete but random cross-linking should produce a Poisson distribution of all sizes with a single maximum In their gels, there are two maxima As shown in Fig 3A,B of Dorigo et al [14], both the even multiples (tetra- and hexamers) are more abundant than the pentamers This suggests that either the samples comprised a mixture of different structures, the crosslinking is not random, or the nucleosomes are nonsequentially arranged (not dis-cussed in their study)

The two structures reported by Robinson et al [15] are regular helices and conform to the parameters of the natural fibres with corresponding repeat length

intervals Robinson et al [15] did not examine the

sequentiallity of the nucleosomes, nor did they identify

the path of the linker DNA or the location of the

linker histones They have assumed that the

nucleo-somes follow the fibre helix and that the linker

histones define the different paths of the linker DNA

for different repeat lengths It is unclear, however,

if this would be the case because they used the same

linker histone for all repeat lengths [24] Accordingly,

their results do not contradict the results provided in

the present study and, thus, further experiments are

required to check for compatibility Nevertheless, the

reconstitutes of Robinson et al [15] are a good start

for further high definition structural studies

Conclusions

The present study demonstrates that, after crosslinking

of oligonucleosomes from native nuclei with two

different crosslinkers followed by nuclease digestion,

there is a gradual decrease of the size of the main

frac-tion, and mainly mononucleosomes are liberated

These mononucleosomes evidently come from

end-discontinuities in the fibre, which can be explained

only by nonsequential arrangements of the

nucleo-somes along the fibre helix The shoulders in the

digests that represent even-numbered nucleosome

frag-ments (Fig 5), as well as the stronger tetra- and

hexa-nucleosome bands shown in Fig 3A,B in the study by

Dorigo et al [14], suggest a nonsequential arrangement

of the nucleosomes; whether this is the (–3,5)

arrange-ment, or some other as yet unenvisaged structure with

end-defects, remains to be seen

Experimental procedures

Preparation of chromatin samples

To avoid irreversible damage of the fibre and to minimize

the redistribution of linker histones, all crosslinking and

sucrose gradient fractionation experiments were carried out

at Na+ion concentrations in the range 25–60 mm Chicken

erythrocyte nuclei, freshly prepared or frozen at)70 C in

40% glycerol, 10 mm Tris–HCl (pH 7.6), 6 mm MgCl2,

25 mm KCl, 35 mm NaCl, 0.2 mm phenylmethanesulfonyl

fluoride, were washed and suspended at 6 mgÆmL)1 DNA

in digestion buffer [0.25 m sucrose, 1 mm CaCl2, 5 mm

Tris–HCl (pH 7.6), 60 mm NaCl] They were digested with

33 unitsÆmL)1MNase at 37C for 10 min, terminated with

5 mm EDTA at final concentration and pelleted by

centri-fugation at 2300 g for 1 min in a microcentrifuge

The supernatant contained approximately 10–15% of the

DNA and consisted mainly of acid-soluble material and

mononucleosomes with a lower amount of H1 and H5 hi-stones The pellet was resuspended in 1 mm EDTA, 5 mm Tris–HCl (pH 7.6) and 25 mm NaCl, dialysed against the same buffer overnight and pelleted again The supernatant usually contained 60–70% of the total DNA and all the histones It was fractionated according to size in 6–40% isokinetic sucrose gradients in the same buffer on an SW27 Beckman rotor (Beckman Coulter, Fullerton, CA, USA) at

5C, 29 500 g for different times The material consisted

of a mixture of mono- to 30–40 nucleosomes size frag-ments, with the most abundant comprising 10–20 nucleo-somes Dialysis of the fractionated samples against higher salt concentration buffers (15–40 mm NaH2PO4, pH 8.0) was performed when the sucrose was dialysed out In this way, no apparent aggregation was observed We have pre-viously shown that, when using this protocol for the isola-tion of chromatin, only mononucleosomes lost some of the linker histones [23]

All samples that were used contained equal ratios of linker to core histones Histone gel electrophoresis of samples used in Figs 3–5 are shown in the supplementary Fig S1 Although the mean sizes of the three samples are approximately 5, 10 and 14 nucleosomes per chain, respec-tively, the ratio of intensities of the bands of H5 to H4 remains the same

In some experiments, the nuclei were digested with

60 unitsÆmL)1MNase and oligonucleosomes were extracted directly with 60 mm NaCl, 5 mm Tris–HCl (pH 7.6) and

5 mm EDTA The supernatant contained 40–50% of the total DNA up to 15–20 nucleosomes long This material did not show any difference in histone content compared to the samples shown in Fig S1

Crosslinking with DSP Crosslinking with DSP (Pierce, Rockford, IL, USA) was carried out in 15–40 mm NaH2PO4 (pH 8), 1 mm EDTA (31–59 mm Na+ ions) with 3 mgÆmL)1 DSP at room temperature for different times and terminated with 15 mm glycine and the crosslinker dialysed out We started with

5 h of crosslinking, but subsequently found out that 10 min was sufficient

Crosslinking with EDC Crosslinking with EDC (Pierce) was carried out in 22–40 mm NaH2PO4 (pH 6.8), 1 mm EDTA (31–62 mm

Na+ions) with 5 mgÆmL)1 EDC at room temperature for

30 min to 12 h and terminated with 1% 2-mercaptoethanol

No difference was found for the results obtained using different times for crosslinking

Redigestion of crosslinked oligonucleosome samples with MNase was carried out in 30 mm NaH2PO4 (pH 7.6), 1.2 mm CaCl2, either at 37C or at 6 C and terminated

with 5 mm EDTA Digestion with trypsin (type II; Sigma,

St Louis, MO, USA) was carried out at room temperature

and terminated with soybean trypsin inhibitor (Sigma)

Isokinetic sucrose gradients (6–40%) of crosslinked and

(or) redigested material were run in 30 mm NaH2PO4

(pH 6.8), 5 mm EDTA (46 mm Na+ions) in a SW41 rotor

(Beckman) at 5C (200 000 g for 12–18 h)

Agarose gel electrophoresis of DNA was carried out in

2% gels in 30 mm NaH2PO4(pH 6.8), 0.5 mgÆL)1ethidium

bromide

Additional results on linker histone abundance and

MNase and trypsin digestions are presented in the

supple-mentary Figs S1–S4

Acknowledgements

We are grateful to Drs Daniela Rhodes and Venki

Ramakrishnan for useful discussions Funding by

Wellcome Trust grant no 037008 to D Z S is

grate-fully acknowledged

References

1 Davey CA, Sargent DF, Luger K, Maeder AW &

Rich-mond TJ (2002) Solvent mediated interactions in the

structure of the nucleosome core particle at 1.9 A˚

reso-lution J Mol Biol 319, 1097–1113

2 Harp JM, Hanson Bl, Timm DE & Bunick GJ (2000)

Asymmetries in the nucleosome core particle at 2.5 A˚

resolution Acta Crystallogr D Biol Crystallogr 56

(Pt 12), 1513–1534

3 Luger K, Mader AW, Richmond RK, Sargent DF &

Richmond TJ (1997) Crystal structure of the

nucleo-some core particle at 2.8 A˚ resolution Nature 389, 251–

260

4 Wolffe AP (1998) Chromatin: Structure and Function

Academic Press, San Diego, CA

5 van Holde KE (1988) Chromatin Springer-Verlag, New

York, NY

6 Finch JT & Klug A (1976) Solenoidal model for

super-structure in chromatin Proc Natl Acad Sci USA 73,

1897–1901

7 Suau P, Bradbury EM & Baldwin JP (1979)

Higher-order structures of chromatin in solution Eur J

Bio-chem 97, 593–602

8 Bordas J, Perez-Grau L, Koch MH, Vega MC & Nave

C (1986) The superstructure of chromatin and its

con-densation mechanism II Theoretical analysis of the

X-ray scattering patterns and model calculations Eur

Biophys J 13, 175–185

9 Bordas J, Perez-Grau L, Koch MH, Vega MC & Nave

C (1986) The superstructure of chromatin and its

condensation mechanism I Synchrotron radiation

X-ray scattering results Eur Biophys J 13, 157–173

10 Gerchman SE & Ramakrishnan V (1987) Chromatin higher-order structure studied by neutron scattering and scanning transmission electron microscopy Proc Natl Acad Sci USA 84, 7802–7806

11 Greulich KO, Wachtel E, Ausio J, Seger D & Eisenberg

H (1987) Transition of chromatin from the ‘10 nm’ lower order structure, to the ‘30 nm’ higher order struc-ture as followed by small angle X-ray scattering J Mol Biol 193, 709–721

12 Ramakrishnan V (1997) Histone H1 and chromatin higher-order structure Crit Rev Eukaryot Gene Expr 7, 215–230

13 Tse C & Hansen JC (1997) Hybrid trypsinized nucleos-omal arrays: identification of multiple functional roles

of the H2A⁄ H2B and H3 ⁄ H4 N-termini in chromatin fiber compaction Biochemistry 36, 11381–11388

14 Dorigo B, Schalch T, Kulangara A, Duda S, Schroeder

RR & Richmond TJ (2004) Nucleosome arrays reveal the two-start organization of the chromatin fiber Science 306, 1571–1573

15 Robinson PJ, Fairall L, Huynh VA & Rhodes D (2006)

EM measurements define the dimensions of the ‘30-nm’ chromatin fiber: evidence for a compact, interdigitated structure Proc Natl Acad Sci USA 103, 6506–6511

16 Schalch T, Duda S, Sargent DF & Richmond TJ (2005) X-ray structure of a tetranucleosome and its implica-tions for the chromatin fibre Nature 436, 138–141

17 Staynov DZ, Dunn S, Baldwin JP & Crane-Robinson C (1983) Nuclease digestion patterns as a criterion for nucleosome orientation in the higher order structure of chromatin FEBS Lett 157, 311–315

18 Staynov DZ (1983) Possible nucleosome arrangements

in the higher-order structure of chromatin Int J Biol Macromol 5, 3–9

19 Thomas JO & Khabaza AJ (1980) Cross-linking of his-tone H1 in chromatin Eur J Biochem 112, 501–511

20 Kunkel GR, Mehrabian M & Martinson HG (1981) Contact-site cross-linking agents Mol Cell Biochem 34, 3–13

21 Butler PJ & Thomas JO (1980) Changes in chromatin folding in solution J Mol Biol 140, 505–529

22 Ruiz-Carrillo A, Puigdomenech P, Eder G & Lurz R (1980) Stability and reversibility of higher ordered struc-ture of interphase chromatin: continuity of deoxyribo-nucleic acid is not required for maintenance of folded structure Biochemistry 19, 2544–2554

23 Staynov DZ & Crane-Robinson C (1988) Footprinting

of linker histones H5 and H1 on the nucleosome EMBO J 7, 3685–3691

24 Staynov DZ & Proykova YG (2007) Topological con-straints on the possible structures of the 30 nm chroma-tin fibre Chromosoma 117, 67–76

25 Rattner JB & Hamkalo BA (1979) Nucleosome packing

in interphase chromatin J Cell Biol 81, 453–457