Proc. Natd. Acad. Sci. USAVol. 91, pp. 5277-5280, June 1994Biochemistry

Linker DNA accessibility in chromatin fibers of differentconformations: A reevaluationJORDANKA ZLATANOVA*t, SANFORD H. LEUBA*, GUOLIANG YANGt, CARLOS BUSTAMANTEf,AND KENSAL VAN HOLDE**Department of Biochemistry and Biophysics, Oregon State University, Corvallis, OR 97331-7305; and 4nstitute of Molecular Biology and Department ofChemistry, University of Oregon, Eugene, OR 97403

Contributed by Kensal van Holde, February 22, 1994

ABSTRACT New studies on chromatin fiber morphology,using the tehque of s ig force microscopy (SFM), havecaused us to xa e recent analysis of Ofchromatin. Chicken erythrocyte chromatin fibers, gntalde-hyde-fixed at 0, 10, and 80 mM NaCI, were Imaged with thehelp of Si1M The du omatin fibers ply_ssess a bose three-

30-nm str even in the absence of added salt.This structure Iht d upon addition of 10 mMNaCI, and highly acted, Irregularly segmented fiberswere observed at 80 mlM NaCl. This sheds new light upon ourprevionsly reported analysis Of the kinetics of d on bysoluble and membrae-obilized micrococcal nucease[Leuba, S. H., Zblatna, J. & van Holde, K. (1994) J. Mol.d.1235, 871-880]. While the low-ionic-strength fibers werereadily d igestA, the hhy compacted tructure formed at 80mM NaCI was refto to nuc s attack, Implying that thelinkers were tilly sible in the low-ionic-strength confor-mation but not in the condensed fibers. We now find thatdeavgSe of the linker DNA by a small molecule, metidm-propyl-EDTA-Fe(li), roeeds for a types of onfortions atsimilar rates. Thus, steric dan e is reoble for the lackof essibly to m i c neas in the condensed fiber.Taken In total the data that reexamination of existingmodels of o n o sormation Is warranted.

Micrococcal nuclease (MNase) has been widely used for thestudy of chromatin structure. It preferentially attacks thelinker regions between nucleosomes to give the well-known"ladder" patterns corresponding to DNA fragments whichare of multiples of nucleosomal length (1, 2). Usually, how-ever, the enzymatic digestion has been performed underconditions which led to rapid degradation of the fiber. Undersuch circumstances, subtle effects of fiber structure on thedigestion might be obliterated. In an attempt to look forpossible differences in the accessibility of the linkerDNA insoluble chromatin fibers of different conformation, we haverecently performed a systematic study using extremely milddigestion conditions (3). Long chromatin fibers from chickenerythrocytes were brought to different states of compactionby increasing the concentration of monovalent ions. Diges-tions were performed with extremely low enzyme concen-trations, in the absence of added Ca2+ ions. Under suchconditions, the "extended" fiber conformations which existat low salt were readily digested with both soluble andmembrane-immobilized MNase, whereas the compact fiberpresent at high salt proved to be almost refractory to nucle-olytic attack. The results were interpreted in the light of thethen-existing models ofchromatin fiber structure in solutionsof different ionic strengths: open zig-zag, closed zig-zag, and30-nm fiber.

Recently, we have successfully imaged chicken erythro-cyte chromatin fibers with tapping-mode scanning forcemicroscopy (SEM) (unpublished work). Using both unfixedand glutaraldehyde-fixed fibers and a variety of ingconditions, we have unequivocally shown that the fibers existin the absence of salt as irregular loosely organized three-dimensional arrays of nucleosomes; the average diameter ofthese fibers is about 34 nm. Addition of salt leads to pro-gressive condensation of the fiber to structures of enormouscompaction and segmented appearance (Fig. 1). This infor-mation indicates that previous ideas about chromatin struc-ture, particularly at lower ionic strength, need to be recon-sidered. Furthermore, our digestion data must be reevaluatedin terms of these results.

MATERIALS AND METHODSChicken Erythocyte Cr . Chromatin and mononu-

cleosomes were prepared essentially as described by Yageret al. (4). To obtain sufficiently long fibers and to reduce thelevel of residual MNase activity in the final preparation, theamount of enzyme used for solubilization of nuclear chro-matin was reduced by a factor of about 500 (3). The finalchomatin preparation was dialyzed versus either 10 mMTris/HC1 (pH 7.5) or 5 mM triethanolamine/HC1 (pH 7.0) asindicated in the figure legends. The minimal length of thefibers was 50-60 nucleosomes (3). When glutaraldehydefixation was employed, the method used was that of Thomaet al. (5) with some modifications. Chromatin (0.1 mg/ml) inthe above buffers, containing 0, 10, or 80mM NaCl, was fixedwith 0.1% (vol/vol) glutaraldehyde overnight at 49C and thendialyzed extensively versus 10 mM Tris/HCI (pH 7.5) andstored on ice. Chromatin depleted ofhistones Hi and H5 wasobtained by stripping the linker histones with 0.35M NaCl inthe presence of CM-Sephadex C25 (Pharmacia) (6).

Digeston with Solable and mobied MNase. Immobi-lization of MNase (Worthington) on Immobilon membranes(Millipore) has been described (3). The digestion conditionsand the quantitative analysis ofthe agarose gels were exactlyas described (3). Partially or totally linearized (EcoRI) plas-mid DNAs used for the control digestions were pBR322 or itsderivative pML2aG (7) (a gift from F. Rougeon, InstitutePasteur, Paris).

Cleavage ofCh withMe my EDTA-Fe(H)(MPE). Chromatin preparations were cleaved with MPE asdescribed (8). Equal volumes of 0.2% MPE and freshlyprepared 2 mM ferrous ammonium sulfate were mixed, and1 ;1 of the mixture was added to 135 1 of chromatin (A260 =2). At time zero, sodium ascorbate (pH 7.0) was added to 1mM. Aliquots were removed at the indicated times and thereaction was stopped by the addition ofbathophenanthroline

Abbreviations: MNase, micrococcal nuclease; MPE, methidi-umpropyl-EDTA-Fe(II); SFM, scanning force microscopy.tTo whom reprint requests should be addressed.

5277

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

20, 2

020

5278 Biochemistry: Zlatanova et al.

to 5 mM. Electrophoretic analysis of the aliquots was asdescribed (3).SFM. SFM images were obtained on a Nanoscope III

(Digital Instruments, Santa Barbara, CA) operated in thetapping mode (unpublished work). Glutaraldehyde-fixed fi-bers were deposited on freshly cleaved mica which had beenpretreated with 1 nM spermidine for 5 min, washed withNanopure water (Barnstead, Dubuque, IA), blotted, and thendried with N2 gas. Spermidine treatment increased the ad-hesion of fibers to the mica surface. The chromatin sampleswere incubated on the surface of the mica for 1 min and thenwashed and dried as above. Purified mononucleosomes wereadded to some ofthe samples to serve as internal controls formeasurements of fiber diameters.

RESULTS AND DISCUSSIONSFM Images of Long Chromatin Fibers. The structure of

isolated chromatin fibers as a function of monovalent saltconcentration in the medium was studied by using the three-dimensional imaging capabilities of SFM, which made itpossible to obtain images under conditions much less harshthan those used in conventional electron microscopic tech-niques. The fibers are likely to retain at least the stronglybound and structurally essential water under the conditionsused (room temperature and 30-60% relative humidity). Fig.1 shows representative images of chromatin fibers that hadbeen fixed with glutaraldehyde before deposition on the solidsupport. The fixation step was necessary to avoid changingthe conformation of the fibers during the abrupt reduction ofsalt concentration which occurs during the necessary wash-ing with water, following deposition. Glutaraldehyde fixationis known to preserve chromatin structure as it exists duringthe fixation step, even when the salt present at the time of

() 10

fixation is removed afterwards (3, 5, 9, 10), or when addi-tional salts are added (10). That glutaraldehyde fixation doesnot cause any gross artifacts is supported by the observationthat images of fixed and unfixed fibers at zero salt werepractically indistinguishable (unpublished work).

Fig. 1 shows that the chromatin fiber possesses a loosethree-dimensional structure even in the absence of addedsalt. This structure slightly condenses when the concentra-tion ofNaCl is raised to 10 mM, with almost no change in theapparent diameter (34 ± 5 nm at zero salt versus 32 ± 5 nmat 10mM NaCl). A major change, however, occurs when thesalt is further raised to 80 mM NaCl. The fiber condenses toa highly compact, irregularly segmented structure, with ap-parent diameter of 46 ± 6 nm. Experiments using graduallyincreasing salt concentrations to follow the condensationprocess more closely show fibers ofintermediate compaction(unpublished data). The segmentation is recognizable alreadyat 20 mM NaCl.

Digestions with Soluble and Membrane-I biMN. In earlier studies (3) fibers at 0, 10, and 80mM NaClwere subjected to extremely mild digestion with solubleMNase. To take into account the dependence of the enzy-matic activity on the concentration of salt, pure DNA wasdigested in parallel, under exactly the same conditions.Quantitations of the digestions by scanning the gels showedthat the differences in the digestion kietics under the dif-ferent conditions were only partially due to differences in theenzymatic activity with salt and depended primarily onchromatin structure (for more detailed discussion, see ref. 3).To completely avoid the complication due to the depen-

dence of the enzymatic activity on the concentration of salt,these experiments were repeated with fibers that were glu-tarldehyde-fixed at the appropriate salt concentrtion andthen dialyzed to remove the salt. The fixed fibers were then

SO mM NaCl

FIG. 1. SFM images of long chromatin fibers fixed with glutaaldehyde in 5 mM triethanolamine/HC1 (pH 7.0) containg 0, 10, or 80 mMNaCl, as denoted at the top. After fixation the samples were extensively dialyzed versus 5 mM triethanolamine/HCI (pH 7.0) and applied tothe mica surface as described in Materials and Methods. Some of the samples contained purified mononucleosomal particles added as internalcontrol for measurements of fiber widths. (Bar = 100 nm.)

Proc. Nad. Acad. Sci. USA 91 (1994)

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

20, 2

020

Proc. Nadl. Acad. Sci. USA 91 (1994) 5279

digested in the absence of salt, so that the digestion kineticsnow depended only on the structure of the fiber at the saltconcentration at which fixation took place. Also, the use offixed fibers for both the MNase digestions and the SFMimaging makes the results from both approaches comparable.Digestions offixed fibers gave essentially the same results asdigestions of unfixed ones (Fig. 2).The SFM images ofthe fibers (Fig. 1) do not usually reveal

the location of the linker DNA; this is especially true of thehigh-salt structure, in which even individual nucleosomes arenot discernible, probably because of their extremely closeproximity. The linker DNA is clearly inaccessible to thesoluble enzyme [3.4 nm in diameter (11)] at 80 mM salt butcan be digested at low salt. This could be explained if at lowsalt the linkers were externally situated or, alternatively, ifthey were centrally located but accessible through an axial"hole" in a fiber with peripherally located nucleosomes (Fig.1; see also ref. 12). To discriminate between these twopossibilities, we also used MNase immobilized on Immobilonmembranes (3). This form of the enzyme should not cut fromthe interior ofthe fiber but should be able to cleave externallysituated linkers. The immobilized enzyme gave essentiallythe same results as did the soluble one: equally rapid diges-tion of fibers at 0 and 10 mM salt and no visible digestion at80mM (3). This implies that the linker DNA in the low-ionic-strength 30-nm fibers is organized so that it is easily acces-sible to digestion from the fiber exterior, whereas in highionic strength it is not accessible at all. Indeed, the SFMpictures clearly show that the chromatin fibers formed at lowionic strength, while possessing three-dimensional structure,are open and loosely organized. In contrast, the structuresformed at 80 mM NaCl are extremely compacted. As notedabove, individual nucleosomes can be visualized in the fibersobtained either in the absence ofadded salt orin 10mM NaCl;they cannot be recognized in 80 mM salt (Fig. 1).WFE Cleavage. The above experiments suggest that the

linker DNA in the loose fibers formed at low ionic strength0 10 80 mM NaCI

aoda' time

1 2 3 4 5 1 2 3 4 5 1 2 3 4 5

FiG. 2. Agarose gel electrophoretic analysis of the digestion ofglutaraldehyde-fixed long chromatin fibers in 10 mM Tris/HCl (pH7.5) containing 0, 10, or 80 mM NaCi with soluble MNase. Thetriangles above the lanes denote increasing digestion times: 0, 30, 60,120, and 240 min of digestion in lanes 1, 2, 3, 4, and 5, respectively.The amount of MNase used for each sample was only 0.01 unit perpg ofDNA. Average DNA size in the initial undigested material wasbetween 15 and 20 kbp; the smallest fiagments on the gel are about200 bp long.

is accessible from the exterior of the fiber and that the failureof the compacted structure at 80 mM NaCl to be digested isdue to drastically reduced accessibility to the nuclease. Todefinitively determine that the lack of digestion of the com-pact fiber was due to steric hindrance and not caused by someotherfactors, digestions were repeated withMPE as cleavageagent (13). MPE has been successfully used in studies ofchromatin organization, as it preferentially cleaves DNA inthe linker regions, producing patterns similar to those ob-tained upon MNase digestion (8). This small molecule (mo-lecular mass, =1000) can penetrate into structures stericallyinaccessible to large probes. Indeed, MPE did digest allconformations ofthe fiber with indistinguishable rates. Slow-ing down the digestion by performing the experiments atlower temperature (Fig. 3) or lower MPE concentrations stilldid not discriminate among the different fiber conformations.Our MPE results are consistent with those of Houssier et al(14), who performed diffusion-enhanced fluorescence ener-gy-transfer experiments with terbium chelates as donor mol-ecules and linker-intercalated ethidium as acceptors. Thoseauthors found equal accessibility of the linker DNA toterbium chelate (0.8 nm in diameter) in both extended (5 mMNaCl) and condensed (80 mM NaCl) fibers.Both MNaae and MPE Digest the Chromatin Fiber Nonran-

domly. At this point we would like to draw attention to a verypeculiar feature of the digestion patterns which was evidentindependent of whether soluble MNase, immobilizedMNase, or MPE was used. In all cases where the fiber wasdigested under sufficiently mild conditions, a gap existedbetween the nucleosomal ladder of one to six nucleosomesand the high molecular mass material. The gap remained evenas the large fragments were gradually degraded to somewhatlower molecular mass fragments. Such a gap has, to ourknowledge, never been reported before and we attribute itsappearance to the extremely mild digestion conditions used.Removal of the linker histones led to blurring ofthe gap (Fig.4), which suggests some role of histone H1/H5 in its gener-ation. This extreme nonrandomness of linker DNA cleavagein the loosely organized fiber may be determined by linkerhistone molecules missing or by other protein molecules

110 S ) nllk Na.]

II_I

FiG. 3. Agarose gel electrophoretic analysis ofthe MPE cleavageoflong chromatin fibers in 10mM Tris/HCl (pH 7.5) containing 0, 10,or 80mM NaCl. Cleavage was performed on ice for 0, 8, 32, 64, and128 min (lanes 1, 2, 3, 4, and 5 in each series, respectively). For adescription of the DNA sizes, see legend to Fig. 2.

Biochemistry: Zlatanova et A

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

20, 2

020

5280 Biochemistry: Ziatanova et al.

U10 mM

1( I III lt

123 4 2 3 4

FiG. 4. Agarose gel electrophoretic analysis of the digestion of

H1/HS-depleted long chromatin fibers with immobilized MNase.

Digestion was in 10 mM Tris/HCl (pH 7.5) or in the same buffer

cnainingl 10mM NaCl, as indicated. The triangles above the lanes

denote increasing digestion times: 0,30,60, and 120 mi of digestion

in lanes 1, 2, 3, and 4, respectively. For a description ofDNA sizes,

see legend to Fig. 2.

[e.g., hg-mobility group 1 (HMG1)J present at the most

accessible sites.

The results from the SFM and the digestion experiments

collectively suggest that even at very low ionic strength, the

structure of the chromatin fiber is a loose irregular three-

dimensional array of nucleosomes, with occasional helical

turns and an apparent diameter of about 30 am. The linker

DNA in these fibers is readily accessible to enzymatic probes

fr-om, the exterior of the fiber. Thus, it is not internalized in

the fiber but is rather situated between adjacent nucleo-

somes. Salt-induced condensation occurs via further folding

of the 30-nm fiber to create extremely compact, irregular,

segmented structures mn which the linker DNA is no longer

accessiible to nuclease attack by even relatively small en-

zymes but is still accessible to chemical cleavage by small

probes.In general, all fiber conformations are much more irregular

that hitherto generally believed. It has to be pointed out,

however, that early work from Hamkalo's laboratory (15, 16)

suggested a variety of nucleosomal packing conformationsand their variable distribution along the length of the chro-matin fiber for both the interphase nucleus and the metaphasechromosomes. Similar views about the intrinsic irregularitiesin the organization of the fiber have recently been expressedby Woodcock and collaborators (17, 18). In view of theseresults we feel that attempts to build regular models ofchromatin structure should be approached with considerableskepticism and that previous physical and biochemical datashould be carefully reexamined.

J.Z. is on sabbatical leave from the Institute ofGenetics, BulmrianAcademy of Sciences, 1113 Sofia, Bulgaria. J.Z. and S.H.L. havecontributed to this work equally. We thank Dr. D. Lohr for helpfuladvice on the MPE digestions and Dr. P. Dervan for the generous giftofMPE. The technical assistance of V. Stanik is acknowledged. TIisresearch was supported by National Institutes of Health GrantGM22916 to K.v.H., National Science Foundation GrantsMCB9118482 and BIR9318945 to C.B., and National Institutes ofHealth Grant GM32543 to C.B.

1. van Holde, K. E. (1988) Chromatin (Springer, New York).2. Tsanev, R., Russev, G., Pashev, I. & Zlatanova, J. (1992)

Replication and Transcription of Chromatin (CRC, Boca Ra-ton, FL).

3. Leuba, S. H., Zlatanova, J. & van Holde, K. (1994) J. Mol.Biol. 235, 871-880.

4. Yager, T. D., McMunay, C. T. & van Holde, K. E. (1989)Biochemistry 28, 2271-2281.

5. Thoma, F., Koller, T. & Klug, A. (1979) J. Cell Biol. 83,403-427.

6. Libertini, L. J. & Small, E. W. (1980) Nucleic Acids Res. 8,3517-3534.

7. Nishioka, Y. & Leder, P. (1979) Cell 18, 875-82.8. Cartwright, I. L. & Elgin, S. C. R. (1989) Methods Enzymol.

170, 359-369.9. De Murcia, G. & Koller, T. (1981) Biol. Cell. 40, 165-174.

10. Russanova, V. R., Dimitrov, S. I., Makarov, V. L. & Pashev,I. G. (1987) Eur. J. Biochem. 167, 321-326.

11. Taniuchi, H., Anfinsen, C. B. & Sodja, A. (1967)J. Biol. Chem.242, 4752-4758.

12. Woodcock, C. L., McEwen, B. F. & Frank, J. (1991) J. CellSci. 99, 107-114.

13. Hertzberg, R. P. & Dervan, P. B. (1984) Biochemistry 23,3934-3945.

14. Houssier, C., Lerho, M., Sarlet, G. & Colson, P. (1992)Proceedings ofthe International Symposium on Colloidal andMolecular Electro-Optics (IOP Publ., Bristol, U.K.), pp. 223-232.

15. Rattner, J. B. & Hamkalo, B. A. (1978) Chromosoma 68,363-372.

16. Rattner, J. B. & Haiko, B. A. (1979) J. Cell Biol. 81,453-457.

17. Woodcock, C. L., Grigoryev, S. A., Horowitz, R. A. & Whit-aker, N. (1993) Proc. Natl. Acad. Sci. USA ", 9021-95.

18. Giannasca, P. J., Horowitz, R. A. & Woodcock, C. L. (1993)J. Cell Sci. 105, 551-561.

Proc. Nad. Acad. Sci. USA 91 (1994)

Dow

nloa

ded

by g

uest

on

Sep

tem

ber

20, 2

020