Probing the free space within rat and chicken chromatin with active and passive probes

One approach to the problem of higher-order structure in chromatin is to attempt to describe the cavity system created by the structure and then to study the way various material elements are arranged with respect to that system.

Passive probes allow a description of some of the properties of nuclear free space but probes armed with some reactive grouping such as an enzymic active centre are needed to establish the relationship between material, structural elements and the various features of the cavity system.

As a partial-model for the structure of the resting chromatin of rat liver we have suggested (Burgoyne, Skinner & Marshall, 1978) that the resting chromatin may be usefully described as a packed mass with 2 main classes of cavity within it: one very large cavity system approximately 11–15 nm wide and a narrower system of cavities approximately 4 nm wide. Furthermore, we propose that there is some element of order in the structure that results in the internucleosomal bridges being encaged within the 4-nm cavities. To demonstrate feasibility we have proposed a slightly more detailed model as described in Burgoyne et al. (1978), although we do recognize that a great many other models may fit the basic proposition of an ordered structure with 2 classes of free space.

In this paper we wish to present further evidence indicating a biphasic character to nuclear free space and further evidence indicating that the intemucleosomal bridge DNA is selectively exposed to the small category of space but not the large category of space.

These conclusions, obtained in the rat liver nucleus, were then checked in the chicken erythrocyte nucleus.

Two basic approaches have been used and the results compared. The first method involves the passive, diffusive penetration of glycogen into rat and chicken erythrocyte nuclear space and the second involves the penetration of glycogen molecules that have been armed by the attachment of a nuclease.

We have previously assumed (Burgoyne, Mobbs & Marshall, 1976) that much of the well known differential selectivity of nucleases for different zones within chromatin simply reflected the differences in molecular diameters of the nuclease molecules. However, this assumption was open to the quite serious criticism that different nucleases have different active centres as well as different molecular diameters and so it became necessary to obtain a constant active centre on a molecular class with variable molecular diameters. We thus decided to attach the smallest nuclease available, micrococcal nuclease, to a large and continuous range of glycogen molecules. This ‘armed-glycogen’ gave a nucleolytic probe with a relatively constant active centre but continuously variable effective molecular diameter.

The theory and procedures are as previously discussed in Burgoyne et al. (1978). However the g forces used to carry out the final packing of the intact nuclei have been reduced from 70000 g for 25 min to 18000 g for 15 min. The analysing column in all experiments is LKB Ultrogel ACA-34, approx. 44 × 1·8 cm.

The normalized elution volume of a molecular species, or K_D, as defined by Gelotte (1960) and Akers (1964) can be approximately linearly related to the molecular diameter of the eluting species over restricted ranges (KT_D= V₆— V/V_t— V₀ where V₆ is the marker elution volume, V₀ is the void volume as determined with a substance that is completely excluded from the gel grains and V_t is the total accessible space in the column measured as the elution volume of mannitol or phenol red).

However, for various reasons no single index of effective molecular diameter is entirely satisfactory (e.g. see Nozaki, Schechter, Reynolds & Tanford, 1976, for discussion) and we have elected to use as our index of molecular diameter, the values obtained by simply assuming that all the mass of the standard proteins is packed into a simple sphere with a partial specific volume of 0·71.

Fig. 1 shows a plot of the normalized elution volume, K_D, against the molecular diameters (D) assigned to standard proteins from the above relationships. The results in Fig. 1 imply an approximately linear relationship with the form D ≃ (1·02 — K_D)/0·081 and this relationship is used throughout this paper to determine effective diameters from normalized elution volumes (K_D). It should be noted that free micrococcal nuclease showed clear evidence of a weak adsorption to the column material by giving smeared and slightly delayed peaks.

Female, hooded Wistar rats were used with a usual age of 6–8 weeks and weight of approximately 150 g. Labelled rats were not used until at least one week after intraperitoneal injection of 150 μCi [Me-³H]thymidine, 20 Ci/mmol per rat, 20–30 h after two-thirds hepatectomy.

Rat liver nuclei were prepared as previously (Wallace, Hewish, Venning & Burgoyne, 1971 ; Marshall & Burgoyne, 1976; Burgoyne & Mobbs, 1975). However, the final washing solution of buffer A with EDTA had the sucrose omitted if it was to be used for a glycogen-penetration study.

Blood was collected, by cardiac puncture, into a syringe containing heparin in approximately one tenth vol. of 100 mM Na-EDTA, pH 7·4. Unlabelled blood was collected from adult birds and labelled blood was collected from 2-week-old chicks, each labelled with 150 μCi of [Me-³H]thymidine, 20 Ci/mmol, given by intraperitoneal injection 7 days before collection of blood.

Chicken liver nuclei were prepared by a procedure that used essentially the same buffer system as the rat liver nuclei but the resistance of the erythrocyte outer membrane to mild detergents and mild mechanical treatment necessitated a number of modifications in the actual procedure. This difficulty has been previously studied and discussed by other groups, e.g. Harlow & Wells (1975). Our procedure used a mild detergent dissolved in a large excess of paraffin oil. This mixture breaks the membranes and adsorbs them to an oil phase that is much lighter than the nuclei. The oil is also expected to buffer the activity of the detergent at a low and constant level.

Sections from these nuclei have been studied by electron microscopy. They appear to be free of membranes and to be indistinguishable from the normal avian erythrocyte nuclei as reported by Davies, Murray & Walmsley (1974).

One volume of clot-free blood was chilled in ice for 30 min and then the cells were washed by one centrifugation through 2 layers of buffered sucrose at 2 °C, 20000 g for 10 min. The upper layer was 1 vol. of 0·34 M sucrose, 2 mM EDTA, 0·5 mM EGTA in buffer A and the bottom layer was 3 vol. of 1·37 M sucrose, 1 mM EDTA, 0.3mM EGTA in buffer A. Partial lysis was often observed during this wash.

The pellet was homogenized in 2 vol. of the same 1·37 M sucrose solution, 1 mM EDTA, 0·3 mM EGTA in buffer A, together with 0·8 vol. of paraffin oil containing 1% Nonidet P-40, for 3 min. Homogenization was carried out in an ordinary glass teflon laboratory homogenizer (tissue-grinder).

The emulsion/homogenate was layered onto a stack consisting of 2 vol. of 2·1 M sucrose and 2 vol. of 2·4 M sucrose, each buffered with buffer A, 0.1 mM EDTA, 0·1 mM EGTA. This stack was centrifuged for 30 min at 2 °C and 70000 g. The resultant pellet was then washed once in 5 ml of buffer A, o-i mM EDTA. This final wash may also have 0-34 M sucrose in it if the nuclei are not to be used for a penetration study of the nuclei. Excessive washing caused clumping.

All these solutions will keep indefinitely deep frozen, so long as the 15mM(0·1%) 2-mercaptoethanol has been omitted. In all preparations, both rat and chicken, the 2-mercapto-ethanol was always added freshly on the day of use and unused buffers discarded. It should be noted that it has long been suspected that thiol status may affect nuclear structure and there is published evidence that supports this, e.g. Bitny-Szlachto & Ochalska-Czepulis (1978).

Modification of glycogen molecular weight range and removal of protein or charged impurities was carried out as follows:

Oyster glycogen, 1·6 g, was hydrolysed in 40 ml of 0·1 M HC1 at 100 °C over a time range of 0.34 min by slowly pumping the contents of the digestion vessel into an iced receiver. The hydrolysate was made alkaline by the addition of an excess (5 mmol) of NaOH and reduced for 1 h at room temperature with an excess (300 mg) of Na borohydride. The excess borohydride was decomposed by addition of solid citric acid until foaming ceased and then the solution was extracted with an equal volume of 78% phenol. The aqueous phase was dialysed against 5 1. of 0·1% mannitol and 2 changes of 5.1. of 0.1 mM EDTA with changes at 24-h intervals. The solution was then passed through a pair of short columns, approximately 4 ml volume, of DEAE-acetate and Na⁺-carboxymethyl cellulose. Finally it was ethanol-precipitated and dried.

The first reaction mix contained 120 mg/ml glycogen, 12 mg/ml Na periodate and was incubated under nitrogen for 18 h at 37 °C in the dark. The reaction was quenched with one tenth vol. ethylene glycol and the glycogen rapidly freed (approx. 30 min) of other reaction products on a very short column of Sephadex G-50 equilibrated with 25% ethylene glycol, 1 mM Na acetate, pH 5.5. The glycogen peak can be visually detected with the phenol sulphuric reaction (Ashwell, 1966; Dubois et al. 1956). The significant components of a 1·5-ml coupling reaction mix were approximately 400 μg micrococcal nuclease (free of amines or ammonia), 12 mg of oxidized glycogen, 0·04 μmol of 3′5′-thymidine diphosphate (to block the enzyme’s active centre), 35 μmol CaCl_t, 0·175 /*mol EDTA, 17·5 μmol K⁺ borate, 14% ethylene glycol, pH 8·5. Coupled for 10 h at 37 °C, under N₂ in the dark and then reduced with 7 mg Na borohydride for 1 h, the sample was then dialysed overnight against 25% ethylene glycol, buffer A (as used in nuclear preparations but without 2-mer-captoethanol), 0·1 mM EDTA, 1·0 mM CaCl₂. This was then loaded onto an Ultrogel ACA-34 column (as used for penetratograms) which had been equilibrated with the same buffer as the load was dialysed against. Fig. 2 shows the resultant size range of active products.

Figs. 3 and 4 show penetratograms for rat liver nuclei and chicken erythrocyte nuclei respectively. The penetratogram for rat liver nuclei is very similar to one published previously (Burgoyne et al. 1978) although the nuclei that gave this penetratogram were exposed to somewhat lower g forces during the final packing from the radioactive glycogen solution.

The curve from the chicken erythrocyte nuclei was obtained under conditions matching those used for the rat liver nuclei, and although it was grossly similar to the rat liver nuclei curve, there did appear to be repeatable differences. The high-molecular-diameter region of the penetratogram from chicken erythrocyte nuclei has an extensive section that is near linear, the corresponding section on the curve from rat liver nuclei being much less obviously linear. Fig. 7 (p. 95) is a difference curve that indicates the main region where the penetratograms from rat liver nuclei and chicken erythrocyte nuclei differ.

Both Figs. 3 and 4 are marked with solid lines indicating the approximately linear portion of the curves. These lines can be extended to the 2 axes, indicating a cavity system of 11—15 nm diameter (where the angle line cuts the lower horizontal axis) accounting for approximately 30–40% of the total free space (i.e. P = 0.3–0.4 where the angle line cuts the right-hand vertical axis, at K_D= 1.00). The diameteruncertainty (11 – 15 nm) of this large-cavity system is due to a number of technical limitations. However, the main limitation is the accurate measurement of the small amount of inter-nuclear void volume because this causes an uncertainty about the placing of the true base line.

Rat liver nuclei and chicken erythrocyte nuclei were exposed to nuclease-armed glycogen with a range of molecular diameters as shown in Figs. 5 and 6. Free, uncomplexed, micrococcal nuclease has a molecular diameter of approx. 3.4 nm and would make up almost all of the activity seen in this molecular diameter range (elution K_D∼> 0.75). As expected, the free enzyme attacked both trypsinized and intact nuclei. However the bound enzyme is responsible for the enzymic activity observed at the higher molecular diameters and this shows a very clear discrimination between intact nuclei and the trypsinized nuclei or naked DNA. After the overall molecular diameter of the nuclease-armed glycogen exceeds approx. 4.1–4.5 nm it completely fails to attack any significant proportion of the DNA in the native nuclei, although the control assays show that the nuclease active centre is still quite operational on naked DNA or on trypsinized nuclei.

Three different types of control are shown on Figs. 2, 5 and 6. Firstly, the free and bound nuclease species were assayed with naked DNA in the presence of intact nuclei, to check for any strange inhibitory interactions (Fig. 2). Secondly, the free and bound nuclease species were all assayed with trypsinized nuclei of exactly the same batch as the experimental nuclei (Figs. 5 and 6). It should be noted that trypsinization is known to cause partial unfolding of the nucleosomal structure (Lilley & Tatchell, 1977) as well as a visible collapse of the gross nuclear morphology. Finally, a plot (Figs. 5 and 6) is shown in which the results from trypsinized nuclei and the results from separate trials with free micrococcal nuclease (Figs. 5 A and 6 A) are combined to predict how much digestion of the intact nuclei should have occurred if the observed active centres were free instead of bound to glycogen.

Whichever control is used, the conclusions remain unchanged, i.e. raising the total molecular diameter above 4.1–4.5 nm results in the nuclease active centre being denied access to the bulk of the intemucleosomal DNA in both the rat and chicken erythrocyte nucleus.

Simple inspection of the penetratograms from the passive penetration studies indicates that external membranes are not responsible for these patterns. An unbroken membrane should generate a step-curve with a sharp cut-off at some particular molecular diameter or else the patterns should be highly unstable with time. There is no sign of such a step-curve and experiments carried out at other penetration times have indicated that there is no significant instability of the patterns.

It is possible that the membranes are always torn during removal, or if not, the pores in them must be larger than the largest intranuclear cavity class with diameters of 11–15 nm.

However, whatever the explanation, it seems that the nuclear membrane is not involved in the explanation of our results.

It was previously deduced that the penetratogram for rat liver nuclei was probably biphasic (Burgoyne et al. 1978) but this deduction was based on a comparison of DNAase-I and micrococcal-nuclease digestion as well as passive penetration studies. In the case of the chicken erythrocyte nuclei, however, this biphasicness is much more pronounced and is clearly indicated from the passive penetration studies alone. Moreover, the nuclease-armed glycogen loses its ability to attack the nucleo-somal bridge at almost exactly the same molecular diameter threshold in both systems.

Thus both the rat liver nuclei and the chicken erythrocyte nuclei show surprisingly similar types of space categories and patterns of susceptibility to conjugated micrococcal nuclease. They are also quantitatively similar in their pattern of space distribution with, near 30% of nuclear space as 11–15-nm holes, 40–50% 4–5-nm holes and near 30% nuclear space impenetrable. The near 30% impenetrable space is presumably occupied by DNA and histones. This seems to indicate some strong similarity in their higher-order structure, despite their differences in origin, histone composition and their apparent differences in potential activity.

These observations actually have 2 quite different types of explanation which are not mutually exclusive. The simplest explanation and the one most immediately useful is that the 2 different nuclei have the bulk of their chromatin or very large zones of chromatin with similar higher-order structure.

These large zones of chromatin would have to make up a very high proportion of the chromatin in both types of nuclei which would seem to suggest that ‘inactive’ chromatin provides the bulk of both these categories of space.

The other explanation that must be considered is the possibility that a process like convergent evolution has resulted in the selection of structures that have similar cavity systems; however, the structures may vary in other respects. This latter explanation is not a strict alternative to the first explanation because both may be partially true. The basis of such selective forces could be the size and shape of the molecules that initiate processes like transcription, replication and repair.

The work presented here indicates that a change in probe diameter from approximately 3.4 to 4.5 nm results in the probe losing all access to the bridge DNA even though all probes have the same active centre and are still quite capable of attacking naked DNA or trypsinized chromatin. This clearly indicates that much of the protection of the bridge is due to structures that are near the bridge but not actually bound to it, i.e. the bridge DNA is in a small cavity like the approx. 4·0-nm cavities that have been reported in reprecipitated chromatin (Burgoyne et al. 1978).

This work is highly compatible with the view that much of the self-protective structure of the nuclei is a secondary consequence of structure-inducing action of proteins like the Hi-type histones (Sluyser, 1970). That is to say, the Hi-type histones induce a nucleosomal packing pattern which is highly self-protective but the direct Steric hindrance of nuclease approach, caused by Hi, may not be as important as this indirect effect (see Noll, 1976, and Van Holde & Weischet, 1978, for related reviews).

To a first approximation the chicken erythrocyte nuclear penetratograms and rat liver nuclei penetratograms are similar, as they both indicate the presence of a class of large cavities that generates the low slope on the left-hand sides of the curves and a class of smaller cavities that generates the sharper slope on the right-hand sides of the curves (Figs. 3 and 4). Moreover, both nuclei have a similar amount of total free space as measured by mannitol, i.e. near 72–73% of total nuclear space. Differences may exist in this ‘total free space’ but they appear to be less than the errors inherent in the procedure for measuring total free space in these pellets.

However, the rat nuclear penetratograms do seem to show a zone of repeatable difference from the chicken erythrocytes nuclear penetratograms and this begins somewhere near 7 nm (Fig. 7). As this small category of space is present in nuclei that are known to be transcriptionally active and is absent in erythrocyte nuclei that are thought to be transcriptionally inactive, it is possible that this may represent transcriptionally active chromatin.

This work was supported by the Australian Research Grants Committee and the Flinders University Research Budget.