Electron-microscopy of intact nuclear dna from human cells

DNA is very fragile so that it is difficult to spread large molecules of DNA for electron microscopy without breaking them (Burgi & Hershey, 1961; Levinthal & Davison, 1961). Since a single nick in a constrained or circular molecule of DNA releases all supercoiling (Bauer & Vinograd, 1974), it is not surprising that spreads of the extremely long molecules found in the nuclei of higher cells contain little, if any, evidence of supercoiling (Du Praw, 1970; Paulson & Laemmli, 1977; McCready, Cox & McLaughlin, 1977; Pinon & Salts, 1977). Recently we have developed techniques for handling nuclear DNA without breaking it (Cook & Brazell, 1975; Cook, Brazell & Jost, 1976). Living cells are lysed in a detergent and 2 M salt to release histone-free nuclear DNA packaged within a flexible cage of RNA and protein. Such nucleoids from cells of insects, amphibians, birds and mammals sediment in the presence of intercalating agents in the manner characteristic of circular molecules of DNA (Cook & Brazell, 1975, 1976): mammalian nucleoids also exhibit the distinctive ethidium-binding capacity of circles (Cook & Brazell, 1978). We concluded that the DNA of higher cells was supercoiled and made quasi-circular by organization of linear molecules into loops. Others, who are probably working with similar structures, have confirmed that the DNA of Drosophila, yeast and Chinese hamster cells sediments in the characteristic manner (Benyajati & Worcel, 1976; Pinon & Salts, 1977; Hartwig, 1978). We now show that the DNA of HeLa nucleoids spread for electron microscopy using the Kleinschmidt (1962) procedure is extensively supercoiled.

HeLa nucleoids were isolated from cells growing in suspension using ‘step’ gradients containing 1·95 M NaCl (Cook et al. 1976). Procedures for handling and counting nucleoids and for monitoring by fluorometry the superhelical status of their DNA have been described (Cook et al. 1976; Colman & Cook, 1977; Cook & Brazell, 1978). Nucleoid DNA was broken by γ-radiation from a caesium 137 source (Cook & Brazell, 1976).

Untwisting (nicking-closing) enzyme essentially free of nicking activity was purified from rat liver by phosphocellulose and CM-sephadex chromatography (Champoux & McConaughy, 1976). Fractions eluted from CM-sephadex were concentrated by pressure filtration and used directly or after storage at —70 ° C. Untwisting and nicking activity were assayed either using supercoiled PM2 DNA and 0 7 % agarose gels (Barnes, 1977) or nucleoids and a spectrofluorometric method based on a procedure for detecting single-strand breaks in DNA (Cook & Brazell, 1978). The method is outlined below and will be published in detail elsewhere. The fluorescence of ethidium is enhanced when it binds to DNA so that the amount of bound dye is readily measured spectrofluorometrically. In 8 μg/ml ethidium, more dye binds to γ-irradiated (9·6 J kg⁻) nucleoids containing broken and relaxed DNA than to their unirradiated counterparts containing superhelical DNA. In the presence of dye, treatment with untwisting or nicking activities equalizes the amount of dye bound by unirradiated and irradiated nucleoids. If ethidium is added after the reaction, nicking again equalizes the amounts bound but untwisting increases the difference. Untwisting assays were carried out in 0·2 M NaCl, 1 mM EDTA, 10 mM Tris (pH 8·0) and nicking assays in 0·2 M NaCl, 1 mM MgCl₂ and 10 mM Tris (pH 8·0). Since the quasi-circles in nucleoids are on average 22 times larger than PM2 DNA, and since only one nick releases all supercoiling in the quasi-circle, the fluorometric assay using nucleoid DNA is extremely sensitive. Concentrated fractions eluted from CM-sephadex contained no nicking activity detectable by either method at the concentrations used to treat the nucleoids.

Nucleoids containing untwisted DNA were prepared as follows. Nucleoids isolated in 1·95M Nacl were diluted to 0·2 × 10^β nucleoids/ml (i.e. 2·4μgDNA/ml) and 0·2 M NaCl using 1 MM EDTA, 10 mM Tris (pH 8·0) and the appropriate salt concentration. 20 vol. of nucleoids were incubated for 15 min at 37 °C with 1 vol. of untwisting enzyme. Sufficient enzyme was used to untwist completely all the nucleoid DNA. The reaction was stopped by adding 3 M ammonium acetate (adjusted to pH 5·5 with glacial acetic acid) to give a final concentration of 1 M.

Unirradiated and irradiated nucleoids bind equal amounts of ethidium at a critical concentration at which supercoiling is minimized (Cook & Brazell, 1978). The critical concentrations which equalized binding in the spreading buffers were determined by repeating the fluorometric experiments described by Cook & Brazell (t978). They were 0·5 and 4·0 μg/ml ethidium for hypophase and hyperphase respectively. Concentrations of 0·5–1 μg/ml and 24 μg/ml were used in both hyper- and hypophases during spreading to prepare nucleoids containing unwound DNA and positively super-twisted DNA respectively.

Nucleoids were suspended in 100 μl1 M ammonium acetate (pH 5·5) containing 10 μg/100 μl cytochrome c (Sigma, type VI) and spread onto 0·3 M ammonium acetate (pH 5·5) using the standard method of Kleinschmidt (2962). The spread nucleoids were picked up onto a carbon-coated parlodion support film on copper grids, stained with uranyl acetate (Davis & Davidson, 1968), rotary shadowed with gold/palladium (6o%/40%) and examined in an AEI 8or electron microscope.

Fig. 1 is an electron micrograph of a typical HeLa nucleoid which has been spread as described in Materials and methods. In such spreads the RNA and protein which initially encaged the DNA has collapsed to form a disk about 14 μm in diameter. Although the cage contains fibrous elements of the cytoskeleton (i.e. actin and intermediate filaments; Levin, Brazell and Cook, unpublished observations) most of the filaments in the disk are probably DNA fibres which have spilled out of the cage and then condensed on to it. Surrounding the collapsed cage and obviously emanating from it is a tangled fibrous network. The thicker branched fibres are probably groups of DNA duplexes which have not bound cytochrome c and so condense during ethanol dehydration (Lang, 1969). The skirt of fibres surrounding the collapsed cage has an average radius of about 32 μm, but occasionally fibres extend over 40 μm from the centre. (The shape of the skirt varies widely from nucleoid to nucleoid and presumably depends upon the forces acting during spreading.)

Individual thin fibres can be resolved from the tangle only at the very edge of the skirt (Figs. 2, 3). Much of the DNA appears highly twisted and coiled in the manner characteristic of superhelical DNA (Bauer & Vinograd, 1974; Delius & Worcel, 1974; Kavenoff & Ryder, 1976; Kavenoff & Bowen, 1976). Many crossovers occur between 0·08– 0·12 μm apart. There are few, if any, free ends. Where paired fibres in superhelices are so tightly intertwined that they appear to form only one fibre they can be recognized as ‘twigs’ projecting from the main ‘branch’. Any relaxed fibres present can be distinguished readily since they follow much less convoluted paths, often appearing as if they had been stretched across the field from a node, broken and unravelled during spreading.

All the nuclear RNA is contained within unspread nucleoids concentrated in the nucleolus and in a perinuclear rim (Cook et al. 1976). Few transcription complexes are observed in the skirt of spread DNA. Perhaps the RNA is torn from the DNA during spreading; alternatively it might remain within the collapsed cage.

A variety of control experiments indicate that the superhelical state of nucleoid DNA influences the shapes adopted by the DNA fibres in our spreads and that they cannot be artifacts produced by spreading high concentrations of long but broken and relaxed DNA fibres. (See Lang (1973) for some examples of coiling of linear DNA induced by dehydration with ethanol.) A dose of 9·6 J kg⁻¹ of γ-radiation introduces at least one single-strand break into most of the quasi-circles in HeLa nucleoids, relaxing the DNA (Cook & Brazell, 1975; 1978). Electron microscopy confirms this result (Fig. 4). The collapsed cage is unaffected but the fibres are relaxed. Again there are few ends: the very low dose of radiation used is insufficient to introduce any double-strand breaks.

Supercoiling was also removed from nucleoid DNA using a nicking-closing enzyme purified from rat liver. Nucleoids treated with the enzyme contain intact and quasi-circular DNA which is essentially free of single or double-strand breaks (see Methods). Their DNA appears relaxed (Fig. 5).

At low concentrations the intercalating ligand, ethidium, binds to circular DNA, removing negative superhelical turns. At a critical concentration, which depends upon the initial degree of supercoiling, all superhelical turns are removed. At higher concentrations binding induces positive supercoiling (Bauer & Vinograd, 1974). As the ç quasi-circles of DNA in nucleoids are probably supercoiled to different degrees (Cook & Brazell, 1975, 1977) it is impossible to select one concentration of ethidium sufficient to remove supercoiling from all. We therefore spread nucleoids in the range of concentrations (i.e. 0·5–1 μg/ml) that minimizes supercoiling in nucleoid DNA in the buffers used during spreading. Fig. 6 illustrates a spread made in º’75 μg/ml; some fibres are relaxed, whilst others are slightly supercoiled, suggesting that different regions of nucleoid DNA have different superhelical densities.

A concentration of ethidium of 24 μg/ml induces many tightly interwound and ‘twig’-like forms which are characteristic of highly supertwisted DNA, confirming that most of the DNA is torsionally constrained (Fig. 7).

The fields presented here are typical of the majority that we have seen. Although regions richer or poorer in superhelical fibres can be found in spreads of unirradiated nucleoids, almost invariably some supercoiling can be detected within a field the size of that presented in Fig. 2. Conversely, and again almost invariably, none can be found in spreads of irradiated nucleoids. The control experiments using DNA untwisted with ethidium or nicking-closing enzyme, or supertwisted with excess ethidium confirm that the nucleoid DNA is supercoiled. As biophysical studies indicate that all nucleoid DNA is torsionally constrained (Cook & Brazell, 1978), the little relaxed DNA seen in the electron micrographs of unirradiated nucleoids probably arose during spreading.

We conclude that histone-free DNA from nuclei can be isolated without breaking it. No special measures have been taken in handling nucleoids to prevent breakage so that the fragile DNA must be exceptionally well-packaged within nucleoids: it must be protected from shear by a flexible cage of RNA and protein (Cook et al. 1976). The DNA from a range of different cell-lines of various species is similarly encaged but as the cage is derived mainly from cytoplasmic fibres, the DNA of diploid cells does not seem to be so well protected, probably because the diploid ceils do not have such well-developed cytoskeletons (Cook et al. 1976; Levin, Brazell and Cook, unpublished observations).

Biophysical studies have shed light on the degree of supercoiling in nucleoid DNA (i.e. in 2 M NaCl one supercoil every 90–180 base-pairs or 0·03–0·06 μm) and the average size of the quasi-circles (i.e. 2·2× 10⁵ base pairs or 75 μm) (Cook & Brazell, 1977, 1978). The superhelical density influences the number of crossovers per unit length in interwound superhelices and the results from the electron micrographs are consistent with the biophysical findings. (But see Vinograd, Lebowitz & Watson (1968) and Wang (1969) for a discussion of the inaccuracies inherent in measuring superhelical density by counting crossovers.) The complexity of the spread fibres precludes estimation of the average size of the quasi-circles, but if they are anchored in or to the cage, then some must be at least 64 μm or twice the radius of the skirt. On the other hand, if they are formed by specific DNA-DNA interactions, for example at the nodes, their minimum size will be correspondingly smaller. (See Cook & Brazell (1978) for a discussion of the nature of the constraining mechanism. We do not know how the quasi-circles measured biophysically correlate with the loops seen by Paulson & Laemmli (1977).)

As spreads of nucleoids from interphase cells are so complex we have also spread individual chromosomes. Conventional methods for the isolation of chromosomes invariably relax the DNA (Warren, 1978). However, when nucleoids are prepared from mitotic cells and stained with ethidium, chromosome aggregatescan be identified in the light microscope: their DNA is supercoiled (Warren & Cook, 1978). Spreads of such mitotic nucleoids contain superhelical fibres and will be the subject of a separate paper.

We thank Professor H. Harris, F.R.S., and Dr B. S. Cox for their continued support and encouragement and I. A. Brazell and M. Simpkins for their help. This research was supported by grants from the Cancer Research Campaign and the Science Research Council.