Electron Microscopy of Interphase Chromosomes In Situ: Binding of Permanganate to Chicken Erythrocytes

The structure of the interphase chromosomes in eukaryotic nuclei can be conveniently considered at 4 levels. First, there is the sequence of nucleotide bases (Britten & Davidson, 1971; Strauss, 1974) in the single long DNA molecules now thought (Swift, 1973) to comprise each individual chromosome. Second, the way in which histones and other proteins are associated with the DNA resulting in folding, to form structural units which have been shown to be long and thread-like (Gall, 1966; DuPraw, 1970; Ris & Kubai, 1970), although there has been controversy about their diameter. And further whether all the DNA is assembled into a thread with similar properties along its length. In this latter respect it is clear that the structure will vary during replication and transcription. Thirdly, there is the way in which the thread is arranged within individual chromosomes, and finally the arrangement of the chromosome complement within the confines of the nuclear envelope.

Studies on intact cells, by optical or other methods, are clearly the only way of revealing the organization at level three. Our electron-microscope observations (refs, in Davies, Murray & Walmsley, 1974), particularly on chicken erythrocyte nuclei, in which all the interphase chromosomes are condensed into chromatin bodies, indicate that the bulk of each body consists of a thread-like unit with similar geometry along its length, and further that the thread itself is folded to form one or more layers at the surface in contact with the envelope. In the interior the arrangement appears random except that there are spherical cavities containing material which may be ribonucleoprotein. This material, together with similar roughly spherical bodies associated with large granules lying in the nuclear sap (Walmsley & Davies, 1975) may have resulted from the transcriptional activity, more active at earlier stages of development. During the deformation of the chromatin bodies and the associated flow of the nuclear envelope which occurs naturally in many cell types, the threads maintain their integrity, which is to be expected if they are the structural units. In the process they form monolayers, rarely bilayers. Their constant width irrespective of cell type and species, animal or plant (Davies & Haynes, 1975), suggests a basically similar arrangement of DNA and histone within the unit, and this is quite likely a structural reflexion of the remarkable similarities in amino acid sequences of certain histones in different species (DeLange & Smith, 1971).

Other features pertinent to the organization at level two are found from electron microscopy of the layers. First, the diameter of the structural unit in situ is about 28·0 nm. Second, the micrographs show that the molecules of protein and DNA are arranged so as to form 2 discrete regions or phases. When a thin section through a chromatin body is treated with uranyl-lead, known to combine preferentially with the DNA, the optical image is heterogeneous, a pattern of densely staining dots and dashes about 17·0 nm wide and spaced about 28·0 nm apart, with lesser staining regions in between. The regions, referred to as the e- and o-phases respectively (Fig. 1), contain relatively high and low concentrations of DNA. There are 2 possibilities regarding the phases. First, they may have the same molecular composition and differ only in the packing, or arrangement, of the molecules within them. For example, Zubay & Doty (1959) originally envisaged a nucleoprotein thread 3·0 nm wide, and such a thread could be tightly folded in the e-phase, less tightly packed in the o-phase to give the observed appearance. In bulk samples of long-chain polymers a differential packing is found, the chain being more tightly packed in ordered regions, separated by regions of lower concentration containing the same chain randomly arranged (Keller, 1968). Second, the 2 phases may differ in molecular composition, in protein-to-DNA ratio, as well as arrangement. Electron-microscope observations (Davies et al. 1974) on the binding of PTA to the protein component of chromatin supported this latter possibility, but there was uncertainty due to the anomalous binding of PTA, thought to result from steric hindrance. In this paper the use of KMnO₄ as a possible protein stain is explored. The molecular weight of permanganate is 119 compared with about 3000 for PTA and hence steric problems are likely to be less troublesome. The data reported here do not support models for the thread-like units in chromatin involving successive folding of a finer thread. Instead they support the hypothesis (Davies et al. 1974) of a different molecular composition in the 2 phases and hence, also, that the units which give rise to the phases, of diameter 28·0 nm, have an inner DNA-rich core 17·0 nm wide and an outer DNA-poor cylindrical shell in which the ratio of protein to DNA is relatively higher.

There are several new lines of work showing details of chromatin organization at level two. When chromatin is treated with a cellular Ca–Mg endonuclease, nuclease-specific sites are attacked, leaving protected nucleoprotein fragments of finite dimensions, indicating a regular substructure (Hewish & Burgoyne, 1973). When histones are gently isolated oligomers are obtained containing fixed numbers of the different histones (Kornberg & Thomas, 1974). Evidence for association of histones comes from many sources, for example Ilyin, Varshavsky, Mickelsaar & Georgiev (1971); D’Anna & Isenberg (1974). When chromosomes are treated with hypotonic solutions and prepared for electron microscopy by methods which have been very successful in revealing the transcription of DNA (Miller & Beatty, 1969) roughly spherical particles are observed, 6·0–10·0 nm diameter (Olins & Olins, 1973; Woodcock, 1973), which form beaded threads. How these particles relate to the in situ structure described here poses interesting problems. If these particles consist of a protein core covered with DNA (Kornberg, 1974; Van Holde, Saharasbudde & Shaw, 1974) then it is difficult to envisage how they could, as such, be assembled to form the 2 phases found in intact chromatin unless they are asymmetric.

Whole blood from 4-day chick and adult hen were processed for electron microscopy as described in detail elsewhere (Small & Davies, 1970). Briefly, after centrifugation in capillary tubes, fixation was in glutaraldehyde (3%), cacodylate buffer (0·1 M), with or without calcium chloride (3 mM), followed by washing; half the blocks were further fixed in 1 % OsO₄. Following dehydration, embedding was in Spurr resin (1969). Blocks of 17-day embryo erythrocytes were kindly supplied by Dr J. V. Small.

Adult hen erythrocytes were haemolysed in Triton X-100 (0·5 % (w/v) dissolved in SMTOG (sucrose, 0·4M; MgCl₂, 1 mM; Tris buffer 0·01 M, pH7·5; n-octanol, 4mM; gum arabic, 4 MM) and further washed in SMTOG and 0·075 M saline (Walmsley & Davies, 1975) to remove traces of haemoglobin. Haemoglobin is removed from cytoplasm, nuclear sap and chromatin bodies, leaving the plasma membrane collapsed on to the nucleus. The nuclei are referred to as washed-Triton-nuclei. Sections were stained with (Δ) 2% aqueous magnesium uranyl acetate for 15 min at 60 °C; (b) uranyl followed by lead citrate (Reynolds, 1963) for 30 s to 1 min; (c) 2 % PTA in 90% ethanol, 10% water for 30 min, at room temperature; or (d) 0·9 % potassium permanganate in 0·1 M phosphate buffer, pH 6·5, for 30 min (Soloff, 1973). A carbon coating was deposited on the surface of the sections which were examined in a Siemens Elmiskop 1 at too kV. Electron densities (ed) of stained areas, roughly 0·25 μm in diameter, were measured relative to clear plastic and expressed per too nm of section thickness (ed/t) as before (Davies et al. 1974). The value of the ed/t is approximately proportional to the product of the number of stain molecules per unit volume and their molecular weight. It has to be remembered that the values of ed are an average for the 2 phases.

The photographic densities of electron-microscope plates were determined in areas corresponding to 5·0 nm in the specimen by means of a Joyce–Loebl recording microdensitometer. The ratio of the relative electron scattering in the 2 phases (e/o ratio) was calculated making use of the approximately linear relationship between photographic density and electron exposure (Valentine, 1965). When measuring the ratio e/o it is desirable to work at the maximum of the binding curve (Fig. 2 and Results), when it can be assumed that all the available sites have been filled. At earlier times the number of stain molecules bound will depend partly on the rate of diffusion to the site, and this may vary depending on the concentration of matter in the different phases.

When a series of sections of approximately the same thickness are treated for varying times with KMnO₄ the electron density, shown in Fig. 2 for washed-Triton-nuclei (Fig. 10), increases to a maximum between 15 and 30 min and then remains constant up to 1 h; in another series of observations on adult hen erythrocytes the ed slowly decreased after 30 min. In one block of 4-day-chicken erythrocytes the cytoplasm had a variable stain uptake due, we assume, to variable haemolysis and/or volume change during preparation. Serial sections were stained with permanganate and ethanolic PTA, and micrographs of the same areas were compared. When cells were placed in order of increasing uptake of PTA, the same order was observed for uptake of permanganate. When repeated sections of the same block were stained with permanganate the values of ed/t were similar. These data indicate that potassium permanganate can be used with some confidence in dye-binding experiments. However, more detailed studies are needed of the relationship between material present and stain bound if precise quantitative studies, potentially very useful both here and elsewhere, are to be pursued. An example of variable stoichiometry is the binding of PTA to erythrocyte chromatin (Davies et al. 1974) which is further discussed below. However, even these departures are interesting for the light they throw on molecular organization.

In general the interaction of stains with nucleic acids and proteins can be studied in 2 ways: by means of model experiments using solutions or gels, or by studying the binding to the different regions of cells, chromosomes, polysomes, cytoplasm, etc., which are of known composition, previously determined by isolation and standard biochemical procedures. Erythrocytes are very suitable since it is known that their cytoplasm consists largely of the protein, haemoglobin, and that the condensed chromosomes contain roughly equal amounts of DNA and histone. Moreover, from interferometry of adult hen erythrocytes it is known that the overall concentration of substance in chromatin bodies is somewhat lower than in the cytoplasm.

Quantitative studies (Huxley & Zubay, 1961; Davies et al. 1974) with uranyl and uranyl-lead show the extent to which these molecules bind preferentially to the DNA; presumably the uranyl binds to the positively charged phosphate groups. An experiment with a block of adult hen erythrocytes fixed in glutaraldehyde–OsO₄ and used later to measure PTA and KMnO₄ binding gave the data shown in Fig. 3A. The ratio of binding in nucleus to cytoplasm (n/c) is 2·5 for uranyl-lead and 3·95 for uranyl only after correction for residual OsO₄-staining. In a second block fixed in glutaraldehyde alone (residual ed/t ∼ 0·01), the n/c ratios were 6·4 for uranyl-lead and 8·6 for uranyl alone. The higher ratios found after fixation in glutaraldehyde alone may merely reflect the different volume changes in nucleus and cytoplasm brought about by treatment with OsO₄. The data suggest that uranyl is somewhat more specific for nucleoprotein than is uranyl-lead. The ratios n/c would be higher if expressed in terms of equal concentration. The data can be compared with the ratio 2·5 for the uptake of uranyl by equal masses of nucleohistone and histone (Huxley & Zubay, 1961).

Data on the binding of PTA and KMnO₄ to sections from the same block, fixed in glutaraldehyde–OsO₄ are shown in Fig. 3A; see also Figs. 4, 5. Silverman & Glick (1969) in model experiments showed that PTA does not bind to/DNA but does attach to proteins. Consistent with this and their known chemical composition, the chromatin bodies were found to stain less than the cytoplasm, with n/c about 0·62. The value of n/c for permanganate is closely similar, about 0·54, which indicates that this stain binds preferentially to protein, much like PTA. Similar data were obtained on a preparation of erythrocytes fixed in glutaraldehyde alone. This shows that neither PTA nor KMnO₄ interacts strongly with the OsO₄ which is bound in the fixed preparations; nor does OsO₄ greatly alter the number of binding sites on protein.

A clear and well known indication of the preferential binding of uranyl-lead to nucleic acids is seen in Fig. 7. In these reticulocytes, the polysomes, consisting of roughly equal proportions of nucleic acid and protein, stand out against the lesser-stained haemoglobin. However, when reticulocytes are stained with PTA (Davies et al. 1974) the polysomes are difficult to see because the uptake by the haemoglobin is roughly the same as that by the polysome, presumably its protein moiety. Similarly, polysomes are difficult to distinguish in reticulocytes treated with KMnO₄ (compare Figs. 7, 8).

In a model experiment a few drops of the KMnO₄ fixing solution were added to a 0·4% aqueous gel of DNA and a solution of histones at a similar concentration. KMnO₄ reacted quickly with the histones to form a dark brown precipitate, but the DNA gel remained clear and there was a very slow colour change over a period of 1 h. These simple observations support the conclusions from electron microscopy that KMnO₄ reacts preferentially with proteins. Clearly further quantitative experiments are needed to see if KMnO₄ like PTA, does not react with DNA.

When PTA binds to nuclei the staining patterns are not what would have been predicted from simple considerations of the uranyl or uranyl-lead patterns. This anomalous behaviour is also shown by KMnO₄ but to a lesser extent, and it can be recognized by 2 related effects, edge effects and bulk effects. These effects, which have been discussed elsewhere (Davies et al. 1974) enable certain suggestions to be made about the packing arrangements in chromatin, but clearly more needs to be learned about the detailed molecular structure by other methods before the phenomena involved in dye-binding can be fully understood and the predictions confirmed.

In adult hen erythrocytes treated with PTA there is a zone of high uptake compared with the rest of the chromatin body, 10–20 nm wide, occupying the position of band b1 and part of b2 (Fig. 4). Such a dense zone is absent from erythrocytes stained in KMnO₄ (Fig. 5) although b2 is sometimes seen to be marginally denser owing to inherent closer packing and alignment of the units at the cell surface as in Fig. 7.

In a section through a preparation of erythrocytes from a 17-day embryo treated with PTA 3 regions were found, designated α, β, γ, which symbols can also be used to describe the type of staining pattern found in the chromatin (figs. 2, 3–8 in Davies et al. 1974). In region α the nuclei stained like the one shown in Fig. 4. In region β each chromatin body had around its entire periphery a denser zone, as in the washed-Triton-nucleus in Fig. 6. In region γ cells contained nuclei which did not show these edge effects, and the average electron density throughout the chromatin bodies was equal to, or somewhat greater than, that of the cytoplasm, not less as in Fig. 4. The values of ed/t were obtained for sections through the same nuclei treated with uranyllead. In going from region γ to α there was a large percentage decrease in PTA uptake in the bulk of the chromatin body, but a small percentage increase in uranyl-lead binding. We supposed that as a result of the preparation methods there were small volume changes produced in the cell population, which changes varied in the different regions of the spun-down block. And further, that an increase in concentration of nucleoprotein resulted in a closer proximity of binding sites, still accessible to uranyllead but now increasingly inaccessible to the larger molecules of PTA. When a section from the same block of 17-day-embryo erythrocytes was treated with KMnO₄ edge effects were completely absent. Inspection of a control serial section stained with PTA showed that in some grid areas the nucleus, ed/t about 0·27, stained more than the cytoplasm, whereas in another area the value of ed/t was much less, about 0·12, and the nucleus was also less dense than the cytoplasm. When the corresponding areas in the KMnO₄ sections were examined all the nuclei were seen to stain to about 296 the same extent as the cytoplasm and the fluctuations in ed/t, average value about 0·14, were smafl. Measurements with the 3 stains on the same cells were not made but it was quite clear that anomalous bulk effects, also, were largely absent after KMnO₄ treatment.

Previously, serial sections from a block of cells from 4-day chicken containing reticulocytes and erythrocytes were stained with PTA and uranyl-lead (Davies et al. 1974). A relatively denser-staining zone after PTA treatment near the edge of the nucleus was present in both reticulocytes and erythrocytes. The change from reticulocyte to erythrocyte is accompanied by an increase in the value of ed/t for the bulk of the chromatin after treatment with uranyl-lead and a decrease after PTA. Since maturation is known to be accompanied by a decrease in volume, an increase in uranyl-lead binding is to be expected, and this is why the behaviour of PTA is described as anomalous. When a section from the same block was stained with KMnO₄ there were no denser zones near the envelope. When the values of ed/t for chromatin bodies of reticulocytes and mature cells which lay side by side in the section were compared, the chromatin was about 7 % higher in mature cells. The average value of ed/t was about 0-14. Chromatin in mature cells stained about 8 % more after uranyl-lead (calculation from data in Davies et al. 1974). Therefore anomalous effects are absent in these preparations also. These further studies have shown up a discrepancy in previous work (Davies et al. 1974). The data there on this block of 4-day-chick cells, essentially correct in showing anomalous binding of PTA, were taken from a section with a low PTA uptake and caused us to assume that PTA uptake in the chromatin of mature adult cells was also small. Measurements, for example those in Fig. 3 A, show that this is not so. Further data (Fig. 3B) from another experiment with 4-day chick also show that both KMnO₄ and uranyl-lead binding increase in going from reticulocyte to erythrocyte and that PTA binding decreases as before but with an appreciably higher absolute value. The uptake of the protein stains varies somewhat, depending on the quality of fixation and volume changes no doubt associated with it, being higher in erythrocyte nuclei which show the dot-dash pattern at an unusually high visibility after uranyl-lead. This is a complicating factor but quite understandable in terms of findings discussed later. Well fixed blocks are defined as those where the reticulocytes and erythrocytes retain their characteristic difference in nuclear morphology (Everid, Small & Davies, 1970), the gap between the membranes comprising the nuclear envelope is not enlarged, etc.

Washed-Triton-nuclei also show the β-type edge effects after treatment with PTA (Walmsley & Davies, 1975) but they are absent after KMnO₄; compare Figs. 6, 10.

The above data show that, after staining with KMnO₄, there are no anomalous edge effects and that the bulk uptake also does not exhibit the same behaviour as with PTA. In any one preparation of 4-day chick, which contains both reticulocytes and mature cells, the uranyl-lead and KMnO₄ uptake are both greater in the mature cells, whereas the PTA uptake is less in mature cells than in the reticulocytes. In the cells from 17-day embryo there were large fluctuations in the PTA uptake associated with the α, β, γ, appearances but KMnO₄, like uranyl-lead, showed smaller variations. However, in an important respect KMnO₄ also behaves in a manner which is anomalous with respect to uranyl-lead. Both protein-binding molecules, KMnO₄ and PTA, behave in this way. The effect is shown in 2 sets of data. First, in the preparations of adult hen erythrocytes a small fraction of the cells appear haemolysed. This chance haemolysis presumably is due to some cells not having withstood the handling procedures and there is no reason to suppose that cells with a more fragile plasma membrane have atypical chromosomes. Comparative data on ed/t in the chromatin of intact and chance-haemolysed cells, side-by-side in a section are very convincing, not being subject to any uncertainty in measuring section thickness, etc. Fig. 3A shows that whereas ed/t for uranyl-lead has decreased slightly, consistent with a slight volume increase on haemolysis, binding of both KMnO₄ and PTA have appreciably increased. Chance haemolysis is accompanied by a change in the appearance of the nucleus. In the intact nuclei of adult hen erythrocytes the 2 phases are difficult to distinguish, probably due to a closer packing of the molecules in the outer shells of the interacting units, the o-phase, which accompanies maturation. In the haemolysed nuclei the dotdash pattern after uranyl-lead is very distinct, similar to that seen in reticulocytes from 17-day embryos and 4-day chick.

The second set of data is derived from a comparison of the values of ed/t for washed-Triton-nuclei of adult hen (Fig. 3c) with the nuclei in intact adult hen cells (Fig. 3A). The PTA and KMnO₄ uptakes are both higher than in the intact cell, whereas the uranyl-lead binding is appreciably decreased. The amount of PTA and KMnO₄ bound to the washed-Triton-nuclei is also greater than that bound to chromatin in intact cells from 17-day embryos and 4-day chicks, whereas there are similar amounts of uranyl-lead bound to all these nuclei (data in Fig. 3B, c; this paper and Davies et al. 1974)·

These data show that chance haemolysis of adult hen nuclei, and haemolysis plus salt-washing both lead to an increase in the concentration of available binding sites for PTA and KMnO₄, with little change or a reduction in the number of uranyl-lead binding sites. Quite likely the reduction in the concentration of binding sites on DNA is due to volume increases and a useful measure of the proportion of protein available in the different nuclei is the ratio of the electron density of KMnO₄, or PTA, to uranyl-lead. The approximate values of the ratio KMnO₄/uranyl-lead are: adult hen, 0·27; chance-haemolysed nuclei, 0·5; washed-Triton-nuclei, 1·0; 17-day embryo, 0·5; 4-day chick, 0·5–0·8 in the differently fixed blocks. These numbers indicate, e.g. that there are about 4 times as many available sites in washed-Triton-nuclei, per molecule, as in intact adult hen.

sWhen washed-Triton-nuclei are stained with uranyl-lead, the chromatin bodies show a dot-dash pattern, the e-phase, separated by regions of lower staining, the o-phase (Figs, 1, 9). When a section is treated with PTA the chromatin bodies take up a large amount of stain (Fig. 3c) and the bulk of their area is homogeneous, apart from a denser zone around each body (Fig. 6). This different appearance (compare Figs. 6, 9) was previously described by Walmsley & Davies (1974) and provided very convincing evidence that chromatin does indeed consist of 2 phases: the lower-staining regions after uranyl-lead do nor arise from geometric overlap effects but actually contain chemical substance which binds PTA to the same extent as the regions which preferentially bind uranyl-lead. When a section is treated with KMnO₄ the resulting ed/t is also high, comparable with that for uranyl-lead and PTA (Fig. 3c). When examined on the fluorescent screen of the electron microscope the dot-dash pattern is dramatiscally clear after uranyl-lead but can only just be distinguished after KMnO₄. Figs. 9 and 10 are serial sections through the same nucleus treated with uranyl-lead and KMnO₄ respectively. In the latter micrograph the dot-dash pattern can be more easily distinguished than on the fluorescent screen, as is the common experience in examining detail of low visibility, but is at much lower contrast than after uranyl-lead. Densitometry of the photographic plates showed that the ratio of stain uptake in the 2 phases, e/o, is 1·34±0·13 after KMnO₄ and 2·20±0·17 after uranyl-lead. As far as possible end-on views of the units were measured, but these data probably do not accurately represent the absolute values for uptake in the 2 phases, due to thread curvature. After fixation of washed-Triton-nuclei in glutaraldehyde alone, similar effects were observed. Intact nuclei of adult hen erythrocytes do not provide a system for checking on the possible differences between the molecular composition of the 2 phases, because of the low, or negligible, visibility of the dot-dash pattern after uranyl-lead. Chance-haemolysed nuclei, intact reticulocytes from 4-day chick and 17-day embryos are suitable for making the comparison because the two phases are clearly visible after uranyl-lead treatment. In all these cells the ratio of KMnO₄ to uranyl-lead binding is similar, about o·5. However, a visual comparison of sections treated with uranyl-lead and KMnO₄ is less convincing than it is for the isolated nuclei because of the lower uptake of KMnO₄ in well fixed intact cells. Experience has shown that it is easy to mistake a change in scale of an electron-density profile for a change in shape. To make visual comparison easier it is possible to reduce the time of post-staining in lead so as to give an overall uptake the same as that found with KMnO₄. Densitometry of the 2 phases in reticulocytes from 17-day embryo (Figs. 7, 8) gave e/o ratios of 1·73 ± 0·13 after KMnO₄ and 2·39 + 0·19 after uranyl-lead. The somewhat lower ratio for KMnO₄ in washed-Triton-nuclei is consistent with the visual impression that, despite the higher uptake, the dot-dash pattern is more difficult to distinguish in it than in the chromatin bodies of intact reticulocytes.

Potassium permanganate is a strong oxidizing agent used in volumetric analysis. It is not much used (Lawn, 1960; Sutton, 1968; Soloff, 1973) as an electron stain, presumably because it does not give interesting micrographs in which polysomes, etc., can be clearly visualized and also partly because it has been reported to give dirty or contaminated sections. Following Soloff’s procedure clean sections could be obtained. It might be assumed that KMnO₄ uptake is a measure of the number of oxidizable sites, but it seems to depend on the fact that permanganate, like PTA, is anionic, since the stain uptake by nuclei acid and protein-containing regions of cells has been shown to resemble closely that for PTA. Evidently, KMnO₄ preferentially stains protein. Compared to uranyl-lead, KMnO₄, like PTA, is also a good membrane stain (Fig. 10).

A main objective of this study was to try to obtain further information about the nature of the 2 phases which comprise condensed chromatin. It is well known that the interpretation of dye-binding experiments is attended by problems and uncertainties, but there seem to be few, if any, other ways of attempting to explore the spatial distribution of DNA and protein in the chromatin bodies. Previously (Davies et al. 1974) it was not possible to exclude the possibility that the same nucleoprotein thread was more tightly packed in the e-phase, thereby lowering the binding of PTA to equal, by chance, that of the o-phase. The present experiments on Triton-nuclei, which show the even-staining pattern after PTA, reveal that the e/o ratio for KMnO₄ is somewhat greater than for PTA, ∼ 1·3 compared with ∼1·0. This suggests some steric hindrance to PTA binding if it is assumed that both heavy molecules bind to the protein rather than the DNA. However, the e/o ratio for KMnO₄ is significantly lower than that for uranyl-lead, this being one of the main findings. The above thread can be envisaged as a DNA molecule with histones wound along it. If KMnO₄ binding in the e-phase was inhibited for steric reasons due to a higher concentration then it is difficult to see why uranyl-lead with roughly similar molecular weight and presumably size is not similarly inhibited. This would lead to equal values of e/o for the 2 stains, and as this is not found, differential folding is excluded.

The obvious conclusion is that the numerical data, coupled with the less anomalous behaviour of KMnO₄, support the hypothesis that the e- and o-phases differ in their molecular composition, not merely in concentration. The measured values of e/o show that there is a higher concentration of DNA in the e-phase. They also indicate that the protein-to-DNA ratio is relatively higher in the o-phase: unequivocal proof of this is lacking because of the possibility that, even in Triton-nuclei where the protein stain uptake has reached its maximum value, there are still sites in the interacting protein postulated below, which are preferentially unavailable in the e-phase, although this seems unlikely.

A further finding which needs discussion and explanation is the substantial increase in binding of both KMnO₄ and PTA to protein without much change or a decrease in the concentration of sites for uranyl-lead, when the various states of chromatin bodies are compared. This leads to the suggestion that there is interaction between the protein molecules in both phases, and that changes in this interaction alter binding to proteins. Interaction of histones is in line with current thinking about the structure of the p-bodies derived from chromatin (Olins & Olins 1973). We suppose that in the intact cell part of the DNA has its negative charge neutralized by part of the histone positive charge and that the remainder of the DNA is neutralized by divalent cations; and further that all the DNA charge is available for binding uranyl-lead. Calculations show that in the Huxley & Zubay (1961) model experiments, the binding of uranyl to DNA was not greatly affected by the presence of histones. The remainder of the histone positive charge not bound to DNA is not available for KMnO₄ and PTA binding due, we suppose, to compact folding of the polypeptide chain or protein–protein interaction; but the groups become available upon nuclear haemolysis or isolation. This hypothesis is consistent with the well known requirement of divalentions for keeping chromatin bodies condensed (e.g. Davies & Spencer, 1962), and it appears also to conform with experiments by Clark & Felsenfeld (1971) and Axel, Melchior, Sollner-Webb & Felsenfeld (1974), who showed by titration and digestion techniques that part (about half) of the DNA in calf thymus is unprotected by proteins. The numerical data on the ratios of KMnO₄ to uranyl-lead suggest that haemolysis of adult hen erythrocytes without or with salt-washing leads to diminished protein interaction and, further, that the processes involved in maturation from reticulocytes and young erythrocytes to older erythrocytes in adult hen also involve protein interaction.

Equations can be deduced for calculating the protein-to-DNA ratios in the 2 phases from the measured e/o ratios. However, there are several uncertainties, especially the question of variable stoichiometry, the effect of the arrangement of charged groups along the histones and whether the uranyl-binding data of Huxley & Zubay (1961) apply in situ. Further work is therefore needed to allow estimates of the relative amounts of DNA and protein in the two phases to be made and to relate our observations on the in situ structure to those being obtained by others (Hewish & Burgoyne, 1973; Olins, Carlson & Olins, 1975; Kornberg & Thomas, 1974).

I thank Professor M. H. F. Wilkins, F.R.S., for encouragement, Miss A. B. Murray, Mr W. Richardson for help and discussion, Mr D. Back for technical help and Mr Z. Gabor for preparing the plates.