Nuclei of in vitro cultured bovine liver cells, deprived of the membranes by Triton X-100, were treated with 2 M-NaCl and DNase. Changes in ultrastructure and protein composition were studied at successive steps during treatment. Electron micrographs of nuclei treated with 2 M-NaCl showed a peripheral lamina and an internal system of randomly coiled filaments embedded in a mass of DNA fibres. After partial removal of the DNA the filaments could be seen to serve as backbones for the DNA attachment. Artificial redistribution occurring during fixation with glutaraldehyde suggests that the salt-resistant filaments are not stably cross-bridged into a three-dimensional network. The existence of reversible cross-bridges in vivo cannot be excluded, however. From the available data it is inferred that the filaments represent a decondensed from of the chromosome scaffolds and play a basic role in the organization of the genome throughout the nuclear cycle.

The proper distribution of the replicated DNA during eucaryotic cell division involves the packing of DNA sister molecules into characteristically shaped chromatids in which the genes are arranged in a reproducibly linear order. How this structural organization can be obtained from the apparently chaotic mass of interphase chromatin is still not understood.

Cell fusion experiments have revealed that formation of chromosomes can take place at any time during interphase, be it in a somewhat disturbed manner during S-phase (Rao et al. 1977). One might envisage, therefore, that a decondensed form of the basic chromosomal organization persists throughout interphase. Such a view is supported by the observation that chromatids contain a central protein scaffold that is resistant to 2M-NaCl (Paulson & Laemmli, 1977), while a protein structure resistant to the same high salt concentration, the nuclear matrix, can be obtained from interphase nuclei (Berezney & Coffey, 1974, 1977; Mitchelsonet al. 1979; and many others, see Berezney, 1984).

Several data suggest a relationship between the two structures.

  1. Certain specific nuclear proteins are constituents of the nuclear matrix as well as the chromosomal scaffold (Pieck et al. 1985).

  2. DNA is attached to both structures; the distances between attachment sites are of the order of the lengths of replicons (Paulson & Laemmli, 1979; Vogelstein et al. 1980; Buongiomo-Nardelli et al. 1982).

  3. Origins of replicons are putative sites of attachment to the interphase nuclear matrix and the chromosomal scaffold (Wankaet al. 1982; Aelenet al. 1983; Van der Yelden et al. 1984; Dijkwel et al. 1986; Carnet al. 1986).

The internal nuclear matrix is usually described as a fibrogranular network (Berezney & Coffey, 1977; Shaper et al. 1979; Bekers et al. 1981; VanEekelenet al. 1981) but marked differences are observed when the preparation conditions are changed (Kaufmann et al. 1981; Kaufmann & Shaper, 1984; Galcheva-Gargova et al. 1982; Bouvier et al. 1985). The causes for these variations may be manifold and are only partly understood to date. We have studied the various steps of the matrix preparation routinely used in this laboratory for studies on DNA attachment (Wanka et al. 1977b-, Dijkwel et al. 1979; Van der Velden et al. 1984). Our results show that, depending on the isolation conditions, the internal protein matrix can appear as a number of randomly distributed protein filaments or as a threedimensional network. There is, in addition, ultrastructural evidence for the DNA attachment to the filaments, suggesting that they are involved in the structural organization of the DNA in interphase nuclei.

Growth of cells and isolation of nuclei

Bovine liver cells were grown in monolayer as described previously (Pieck et al. 1985). The cultures were rinsed twice with 0 ·9% NaCl and briefly with Triton/Tris (0 ·1 % Triton X-100 in 5mM-Tris buffer, pH 8). Cells were detached from the glass surface by shaking with 20 ml Triton/Tris per Roux bottle and collected by centrifugation for 2 min at 400 g. The sediment was resuspended in 5 ml Triton/Tris and cells were broken by thorough whirling on a Vortex mixer; 15 ml Triton/Tris were added and the nuclei were collected by centrifugation for 3 min at 450 g. The cytoplasmic proteins were precipitated from the supernatant by addition of 2 vol. of 96% ethanol and stored at −20°C. The nuclear pellet was washed once with 30 ml Tris/Triton and once with 5 mM-Tris buffer, pH 8.

Preparation of nuclear matrices

Nuclei from one Roux bottle were carefully suspended in 30 ml of 50 mM-Tris buffer, pH 8, and mixed with an equal volume of 4M-NaCl dissolved in the same buffer. The salt-treated nuclei were collected by centrifugation for 20 min at 16 000g. The sediment was washed twice with 2 M-NaCl in 50 mM-Tris buffer and once in plain buffer.

For the preparation of nuclear matrices the suspension of nuclei in 2 M-NaCl was digested with appropriate concentrations of DNase I (Sigma, electrophoretically pure) in the presence of 7 ·5 mM-MgCb. After incubation for 30 min at 37°C the residual material was collected and washed with buffer as described for salt-treated nuclei. Samples for electrophoresis were stored in 96% ethanol at −20°C.

Electron microscopy

Isolated nuclei, salt-treated nuclei and matrices were fixed with 3% glutaraldehyde in 04 M-sodium cacodylate (pH 7-2) for 30 min at 4°C and post-fixed with 1 % OsO4 in the same buffer for 60 min at 4°C. After a wash in the same buffer the material was dehydrated in a graded ethanol series and propylene oxide and embedded in Epon. Ultrathin sections were cut and stained with uranyl acetate and lead citrate. For whole-mount preparations the washed nuclear matrices were carefully suspended in a small volume of 50 mM-Tris buffer, pH 8, and 20-/J1 drops were placed on a flat piece of Teflon. The material was picked up from the surface onto Formvar/carbon-coated copper grids, dehydrated in a graded ethanol series and air-dried from a final step in methylbutane. Staining was for 5 min with 1 % phosphotungstic acid in the 50% ethanol step. Electron micrographs were taken with a Philips EM201 operating at 60 kV.

Gel electrophoresis

The ethanol-washed preparations were dried in the air and dissolved in Laemmli’s sample buffer supplemented with 6M-urea (Laemmli, 1970). Slab gel electrophoresis was carried out in the presence of sodium dodecyl sulphate (SDS) according to Laemmli using a gradient of 6 % to 18 % polyacrylamide.

For two-dimensional electrophoresis the samples were prepared according to Wilson et al. (1977) and separated in the polyacrylamide system of O’Farrell (1975). The Ampholine composition was of equal parts of pH 4—6, 6 ○8, 7 ○9, 9 ○11 and 3 ○10. NP40 was added according to Ames & Nikaido (1976). The proteins were electrofocused for 16h at 400 V and another hour at 700 V. SDS-polyacrylamide electrophoresis in the second direction was in gradient gels of 6% to 18% polyacrylamide. All gels were stained with Coomassie Blue.

More than 90% of the nuclei isolated in 50 mM-Tris buffer, pH 8, and 0 ·1% Triton X-100 had diameters between 10 and 16 μm. This was not significantly different from the diameter measured by phase-contrast microscopy in vivo. Electron micrographs of thin sections showed no differentiation in euchromatin and heterochromatin (Fig. 1). The homogeneously dispersed chromatin was contained by a typical lamina, from which the membranes had been removed by the detergent.

Fig. 1.

Electron micrograph of an isolated nucleus. Cross-section showing a homogeneous mass of chromatin surrounded by the lamina (arrow). Nuclear membranes have been removed by the presence of Triton in the isolation medium. Bar, 2μm.

Fig. 1.

Electron micrograph of an isolated nucleus. Cross-section showing a homogeneous mass of chromatin surrounded by the lamina (arrow). Nuclear membranes have been removed by the presence of Triton in the isolation medium. Bar, 2μm.

There was no visible contamination by cytoplasm. This was confirmed by electrophoretic analysis of the protein compositions of nuclei and cytoplasm. No cytoskeleton proteins were found in nuclei isolated in the absence of PMSF (phenyl-methylsulphonyl fluoride; Fig. 2, lane A). However, two polypeptides comigrating with cytoplasmic bands of 48 and 54(×103) Mr were invariably found when 1 mM-PMSF was included in the isolation buffer (Fig. 2, lanes B and C). They are likely to be contaminations by the cytoplasmic counterparts. There was no indication of any proteolytic activity in the absence of the inhibitor. This contrasts with the high proteolytic activities encountered in preparations of nuclear matrices from rat liver tissues (Berezney, 1979). Since previous studies have shown that the typical binding of the DNA to the nuclear matrix is not dependent on the presence of PMSF during the preparation (Wanka et al. 1977b; Dijkwel et al. 1979; Van der Velden et al. 1984), we have omitted PMSF from our routine isolation procedure.

Fig. 2.

Effect of PMSF on the protein composition of isolated nuclei. Electrophoretic separations of proteins of nuclei isolated in the absence (lane A) or presence (lane B) of PMSF. The amount of cytoplasmic proteins (lane C) applied was equivalent to 5 % of the amount of cells used for nuclear proteins. Marker proteins (lane M) and their molecular weights (×10−3) were: 68, bovine serum albumin; 45, ovalbumin; and 29, carboanhydrase. Arrows indicate polypeptides present in the cytoplasm and in nuclei isolated in the presence of PMSF.

Fig. 2.

Effect of PMSF on the protein composition of isolated nuclei. Electrophoretic separations of proteins of nuclei isolated in the absence (lane A) or presence (lane B) of PMSF. The amount of cytoplasmic proteins (lane C) applied was equivalent to 5 % of the amount of cells used for nuclear proteins. Marker proteins (lane M) and their molecular weights (×10−3) were: 68, bovine serum albumin; 45, ovalbumin; and 29, carboanhydrase. Arrows indicate polypeptides present in the cytoplasm and in nuclei isolated in the presence of PMSF.

Addition of NaCl at a final concentration of 2 M to nuclear suspensions resulted in a marginal increase in the nuclear diameters. A large proportion of the protein-deprived DNA penetrated the lamina and appeared as a fluorescent halo after staining with ethidium bromide. Such structures have already been shown by others under comparable conditions (Cookei al. 1976; Vogelsteinet al. 1980; Buongiorno-Nardelli et al. 1982; Dijkwel et al. 1986). In thin-section electron micrographs they were considerably shrunken due to fixation and dehydration effects. This is shown by the strong irregular folding of the 15 –20 nm thick lamina. The mass of DNA was evenly distributed over the nuclei and a zone around them (Fig. 3). Numerous, small electron-dense patches were visible in the entire intranuclear space. Upon examination of serial sections they turned out to be randomly cut segments of long irregularly wound filaments. Tracings of three consecutive sections of part of a nucleus are superimposed on each other in Fig. 4. The overlapping of the electron-dense areas of adjacent sections indicates that they represent segments of a continuous structure. Occasionally the filaments appeared to merge with the lamina. Their average diameter, as estimated from suitably oriented fragments at higher magnification was 25 nm (Fig. 5).

Fig. 3.

Electron micrograph of a salt-treated nucleus. Cross-section of an isolated nucleus extracted with 2M-NaCl, showing the lamina (arrow) and numerous segments of randomly oriented filaments (arrowheads). Bar, 2 μm.

Fig. 3.

Electron micrograph of a salt-treated nucleus. Cross-section of an isolated nucleus extracted with 2M-NaCl, showing the lamina (arrow) and numerous segments of randomly oriented filaments (arrowheads). Bar, 2 μm.

Fig. 4.

Superimposed tracings of serial sections of salt-treated nuclei. Tracings were made of the electron-dense areas of three consecutive sections (1, encircled; 2, vertical hatching; 3, horizontal hatching) of part of a salt-treated nucleus and superimposed photographically. Tracings were aligned by using suitably oriented segments of the lamina in sections 1 and 2 (arrowheads) and sections 2 and 3 (arrows). The original micrograph of tracing 2 is shown in B. Bar, 1 μm.

Fig. 4.

Superimposed tracings of serial sections of salt-treated nuclei. Tracings were made of the electron-dense areas of three consecutive sections (1, encircled; 2, vertical hatching; 3, horizontal hatching) of part of a salt-treated nucleus and superimposed photographically. Tracings were aligned by using suitably oriented segments of the lamina in sections 1 and 2 (arrowheads) and sections 2 and 3 (arrows). The original micrograph of tracing 2 is shown in B. Bar, 1 μm.

Fig. 5.

Higher magnification of part of a 2M-NaCl-treated nucleus. Short fragments of coiled filaments (arrowheads) oriented in the plain of sectioning show average diameters of 25 nm. Bar, 0 ·2μm.

Fig. 5.

Higher magnification of part of a 2M-NaCl-treated nucleus. Short fragments of coiled filaments (arrowheads) oriented in the plain of sectioning show average diameters of 25 nm. Bar, 0 ·2μm.

After partial digestion with DNase, residual DNA fragments remained closely associated with the electron-dense patches, suggesting that the DNA is attached to the filaments (Fig. 6). DNA fragments appeared also at the lamina, but it was not clear whether they originated from the lamina itself or from the internal protein structures frequently associated with it. When the digestion was carried out at higher DNase concentration, the internal electron-dense structures had increased in size but were less in number (Fig. 7A). After extensive removal of the DNA the electron-dense masses were confined to the peripheral region in about 50% of the cases (Fig. 7B).

Fig. 6.

Part of a 2M-NaCl-treated nucleus after limited DNase digestion. Nuclei were treated with 2M-NaCl and then digested with 2 units ml−1 DNase. Cross-sections show residual DNA fragments preferentially associated with the randomly sectioned protein filaments (arrowheads). Arrow indicates the lamina. Bar, 0 ·2 μm.

Fig. 6.

Part of a 2M-NaCl-treated nucleus after limited DNase digestion. Nuclei were treated with 2M-NaCl and then digested with 2 units ml−1 DNase. Cross-sections show residual DNA fragments preferentially associated with the randomly sectioned protein filaments (arrowheads). Arrow indicates the lamina. Bar, 0 ·2 μm.

Fig. 7.

Cross-sections of nuclear matrices. Nuclear matrices were prepared from 2 M-NaCl-treated nuclei by digestion with 4 units ml−1 (A) and 20 units ml−1 (B) DNase. The internal filaments have become aggregated (arrowheads) and frequently associated with the lamina (arrows). Bar, 2 μm.

Fig. 7.

Cross-sections of nuclear matrices. Nuclear matrices were prepared from 2 M-NaCl-treated nuclei by digestion with 4 units ml−1 (A) and 20 units ml−1 (B) DNase. The internal filaments have become aggregated (arrowheads) and frequently associated with the lamina (arrows). Bar, 2 μm.

Similar residual nuclear structures with empty central areas have been reported by other authors (Kaufmann et al. 1981; Kaufmann & Shaper, 1984; Galchewa-Gargova et al. 1982; Bouvier et al. 1985). The most obvious reason for this change is that, in the absence of intervening material, the fixation results in an aggregation of the filaments into large clusters and a partial association with the lamina. At higher magnification short DNA fragments can still be seen to emerge from the dense aggregates. Coincidence of DNA fragments with the lamina now appeared less frequently and this may be fortuitous (Fig. 8).

Fig. 8.

Higher magnification of part of a nuclear matrix. The matrix was prepared by digestion with lOunitsml-1 DNase. Short DNA fragments appear to emerge from the randomly sectioned protein filaments (arrowheads). Arrow indicates the lamina. Bar, 0 ·2 μm.

Fig. 8.

Higher magnification of part of a nuclear matrix. The matrix was prepared by digestion with lOunitsml-1 DNase. Short DNA fragments appear to emerge from the randomly sectioned protein filaments (arrowheads). Arrow indicates the lamina. Bar, 0 ·2 μm.

To check whether the disappearance of the central filamentous material is accompanied by a loss of proteins we have compared protein compositions before and after the DNase treatment. Fig. 9 shows the corresponding two-dimensional electrophoretic separations. Both patterns show lamins A, B and C in addition to several minor polypeptide spots as already described for nuclear matrices (Kaufmann & Shaper, 1984; Berezney, 1984). There was no significant loss of any polypeptide by the DNase treatment. This indicates that the DNA loops are essentially free of proteins after the 2M-NaCl treatment. It also supports the conclusion that the disappearance of the filaments from the central area is due to a redistribution during the subsequent fixation rather than a loss caused by the DNase treatment. Further evidence for this view was obtained by examining whole-mount preparations of unfixed material after extensive removal of the DNA. Spreading on the water surface caused a considerable expansion of the matrix. Average diameters were about 30 μm. The internal protein structure had collapsed, but owing to the speading its fibrous nature was clearly recognizable (Fig. 10). Whole-mount spreads were also prepared of matrices obtained by direct digestion of isolated nuclei with DNase I in the absence of Mg2 +Average diameters of the whole mounts were 15 μm, which is equal to the original nuclei. Consequently, the filaments remained coiled and partially aggregated, but could be clearly recognized at several points (Fig. 11).

Fig. 9.

Two-dimensional electrophoretic separations of nucleoid and nuclear matrix proteins. Proteins of 2M-NaCl extracted nuclei (a). Proteins of nuclear matrices prepared by digestion with 20 units ml−1 DNase (b). A, B and C indicate the respective lamins. Spots pointed at by arrows originate from the enzyme. Basic side is on the left.

Fig. 9.

Two-dimensional electrophoretic separations of nucleoid and nuclear matrix proteins. Proteins of 2M-NaCl extracted nuclei (a). Proteins of nuclear matrices prepared by digestion with 20 units ml−1 DNase (b). A, B and C indicate the respective lamins. Spots pointed at by arrows originate from the enzyme. Basic side is on the left.

Fig. 10.

Whole-mount preparation of a nuclear matrix. The matrix was prepared by digestion of salt-extracted nuclei with 50 units ml−1 DNase. The fibrous nature of the collapsed internal protein structure is clearly visible (arrowheads). Bar, 10 μm.

Fig. 10.

Whole-mount preparation of a nuclear matrix. The matrix was prepared by digestion of salt-extracted nuclei with 50 units ml−1 DNase. The fibrous nature of the collapsed internal protein structure is clearly visible (arrowheads). Bar, 10 μm.

Fig. 11.

Whole-mount preparation of a DNase-digested nucleus. Nuclear matrices were prepared by digesting isolated nuclei with 50 units ml−1 DNase I in the absence of Mg2+. Arrowheads indicate recognizable parts of the collapsed filaments. Bar, 5 μm.

Fig. 11.

Whole-mount preparation of a DNase-digested nucleus. Nuclear matrices were prepared by digesting isolated nuclei with 50 units ml−1 DNase I in the absence of Mg2+. Arrowheads indicate recognizable parts of the collapsed filaments. Bar, 5 μm.

Very different results were obtained when nuclei were digested with DNase in the presence of 7 mM-Mg2+ and subsequently extracted with 2M-NaCl. The average nuclear diameters decreased by about 50%, which is likely to be caused by the presence of MgCl2 (Bekers, 1982). They did not change noticeably during the subsequent treatment with 2M-NaCL In the final preparations the internal matrix had the appearance of a three-dimensional network containing residual nucleoli with a more homogeneously dense structure (Fig. 12). In cross-sections the network appeared to be made up of small spheres and closed tubules. Their size was variable but diameters of about 100 nm predominated. The walls of the spheres and tubules were about 25 nm thick. This picture contrasts with the usual description of the nuclear matrix as a fibrogranular network (Berezney, 1984).

Fig. 12.

Electron micrograph of a nuclear matrix obtained by DNase digestion prior to the 2M-NaCl treatment. Isolated nuclei were first digested with 50 units ml−1 DNase and then extracted twice with 2M-NaCl. The cross-section (A) shows the lamina (double arrow), a three-dimensional internal network (arrowhead) and a dense residual nucleolar structure (arrow). At higher magnification (B) the internal network appears to consist of spheres and short, closed tubules (arrowhead). Bars, 1 μm.

Fig. 12.

Electron micrograph of a nuclear matrix obtained by DNase digestion prior to the 2M-NaCl treatment. Isolated nuclei were first digested with 50 units ml−1 DNase and then extracted twice with 2M-NaCl. The cross-section (A) shows the lamina (double arrow), a three-dimensional internal network (arrowhead) and a dense residual nucleolar structure (arrow). At higher magnification (B) the internal network appears to consist of spheres and short, closed tubules (arrowhead). Bars, 1 μm.

Protein compositions of these preparations have been studied by SDS-poly-acrylamide gel electrophoresis. They show several additional polypeptides as compared to matrices obtained by the routine procedure (Fig. 13). How a simple change in the sequence of treatments can cause such a drastic effect is not clear. However, the strong condensation of the nuclei caused by the addition of Mg2+, possibly due to the formation of cross-bridges, may play an important role (Galcheva-Gargova et al. 1982; Bekers, 1982). We should mention that we have observed the same matrix structure as described by Berezney & Coffey (1977) and others, when we have used their preparation procedure. These results add to the great variability of structures reported for nuclear matrix preparations.

Fig. 13.

Effect of the preparation procedure on the protein compositions of the nuclear matrix. Electrophoretic separations of: lane A, total nuclear proteins; B, nuclear matrix proteins obtained by 2 M-NaCl treatment followed by digestion with 40 units ml−1 DNase; and C, nuclear matrix proteins obtained by digestion with 20 units ml−1 DNase followed by treatment with 2 M-NaCl. Standards for Mr (× 10−3) on the left (see legend to Fig. 2).

Fig. 13.

Effect of the preparation procedure on the protein compositions of the nuclear matrix. Electrophoretic separations of: lane A, total nuclear proteins; B, nuclear matrix proteins obtained by 2 M-NaCl treatment followed by digestion with 40 units ml−1 DNase; and C, nuclear matrix proteins obtained by digestion with 20 units ml−1 DNase followed by treatment with 2 M-NaCl. Standards for Mr (× 10−3) on the left (see legend to Fig. 2).

A thorough knowledge of the structure of the nuclear matrix in vivo is important for our understanding of its role in the spatial organization of the DNA during the cell cycle. Unfortunately, the elucidation of this nuclear component is rather difficult because of the great variability found in morphology and protein composition of the isolated structures reported by many investigators. Matrix preparations obtained from mammalian cell nuclei by digestion with DNase in the presence of MgCU prior to treatment with 2M-NaCl usually contain a fibrous network (Berezney & Coffey, 1977; Shaper et al. 1979; Galcheva-Gargova et al. 1982). On the other hand, matrices with large areas devoid of any structure, notably in the central region, are obtained when nuclei are first treated with 2M-NaCl in the absence of Mg2+ and subsequently digested with DNase as shown here (see also Bouvier et al. 1985).

An analysis of the various isolation steps shows that drastic changes in the residual protein structure can occur during the preparation. Residues obtained by treatment of nuclei with 2 M-NaCl reveal a filamentous protein structure embedded in a mass of DNA. After partial removal of the DNA it can be seen that the filaments serve as backbones for the attachment of DNA. The absence of filaments from the central matrix region in thin sections prepared after extensive DNase digestion must be ascribed to covalent cross-linking by fixation with glutaraldehyde. In the absence of intervening material, for example DNA or chromatin, cross-links will be formed increasingly between filaments as well as between lamina and filaments, resulting in the formation of dense aggregates, preferentially in the peripheral region. This conclusion is corroborated by the finding that no protein is lost as a consequence of DNase treatment.

The fact that the filaments aggregate into irregular masses during fixation excludes the possibility that they are stably cross-bridged into a three-dimensional network in the nuclei. However, the question of whether such a network exists in vivo cannot be answered on the basis of the available data. It is feasible that weak cross-bridges present m vivo are destroyed during the treatment with 2 M-NaCl. On the other hand cross-bridges could be formed artificially by improper preparation procedures. For example, cross-bridge formation might be the cause of the approximately 40% shrinkage of the nuclear diameters observed in the presence of MgCl2 (Bekers, 1982). It has also been shown that in some cases the presence of an internal network in matrix preparations is strongly dependent on native or artificially induced disulphide cross-binding (Dijkwel & Wenink, 1986; Kaufmann & Shaper, 1984).

Apparently conflicting results have also been reported for the attachment of transcriptionally active DNA to the matrix (Robinson et al. 1982; Basler et al. 1981 ; Ross et al. 1982). Recently, Razin and co-workers (Razin et al. 1985) have shown that the binding is lost by treatment of DNase-digested nuclei with Mg2+-free low ionic strength buffer. This suggests that under experimental conditions that lead to a less-compact matrix structure DNA binding is largely restricted to the origins of replicons and, during DNA synthesis, to sites close to the replication forks (Van der Velden et al. 1984; Dijkwel et al. 1979, 1986). Complete release of the DNA has only been possible by treatments that destroy the residual protein structure, such as proteolytic digestion (Wanka et al. 1977b), 8M-urea (Mullenders et al. 1982) and SDS (e.g. see Wanka et al. 1977a).

On the basis of these data the putative role of the salt-resistant, intranuclear protein structure in the organization of the nuclear DNA can be envisaged as follows.

For each chromosome the nucleus contains a protein filament that serves as a backbone for the anchorage of the chromatin fibre. Each filament is attached to the lamina by its ends. This is suggested by the observations that selectively labelled telomeres become located at particular regions of the nuclear envelope when this is re-formed in telophase, and that the label remains in the peripheral region during the subsequent interphase (Fussel, 1975). It is further supported by the recent finding that heterochromatic parts, i.e. telomeres and centromeres, of Drosophila salivary gland chromosomes are in stable contact with the nuclear envelope (Hochstrasser et al. 1986). The observed attachment of chromosomes to the nuclear envelope by the telomeres during the meiotic prophase (von Wettstein et al. 1984) might simply be a continuation of the proposed interphase attachment.

The chromosomal chromatin fibre is considered to be repeatedly bound to the protein backbone by consecutive origins of replicons. Permanent attachment of replication origins to the salt-resistant residual protein structure has been demonstrated in nuclei of Physarum polycephalum (Wanka et al. 1982; Aelen et al. 1983) and of in vitro cultured cells of Xenopus laevis (Carri et al. 1986) as well as mammalian cells (Van der Velden et al. 1984; Dijkwel et al. 1986). It is consistent with the observation that lengths of DNA loops between attachment sites in salt-treated nuclei and chromosomes are in the range of replicon lengths (Paulson & Laemmli, 1979; Vogelstein et al. 1980; Buongiorno-Nardelli et al. 1982).

We therefore propose that the 2M-NaCl-resistant protein filaments represent a decondensed equivalent of the chromosome scaffold. Its duplication must be assumed to coincide with DNA replication, since premature chromosome condensation shows an increasing proportion of binemic chromosomes in late S-phase (Hameister & Sperling, 1984). Formation of the solenoid-like structure of the chromosome scaffold (Marsden & Laemmli, 1979) presumably results from an orderly folding up the randomly coiled interphase filaments. Such a solenoid-like scaffold is compatible with elastic properties indicated by the reversible stretching of chromosomes by in vivo micromanipulation (Nicklas & Staehli, 1967). The strong variations of the chromosome length during meiotic prophase might be due to different degrees of folding.

Persistence of a decondensed chromosome scaffold during interphase is further supported by the finding that certain proteins of the chromosomal scaffold are also present in nuclear matrix preparations (Pieck et al. 1985). Another line of evidence for the persistence of the 2M-NaCl-resistant protein backbones has been obtained in P. polycephalum. Owing to the closed mitosis in this organism it has been possible to follow the gradual transformation and translocation of the filamentous interphase matrix throughout mitosis (Bekers et al. 1981).

The ordered and permanent binding of the chromatin fibre to a persistent protein filament has some interesting implications. First of all it provides a simple means by which the linear order of the genes along the chromosome can be maintained. Second, it is an appropriate prerequisite for the proper unwinding of the double helix during DNA replication and the segregation of the daughter molecules during mitosis. It may also be the basis for the rapid chromosome condensation at any time during interphase when induced by fusion with mitotic cells. Finally, we would like to direct attention to a particular role in the regulation of DNA replication. As the eucaryotic genome is subdivided into thousands of replicons it has to be ensured that no part can be replicated more than once during each cell cycle. To exclude supernumerary duplication of replicons the origin sites have to be altered by the replication in such a way that they become inaccessible to further action by initiation factors. This block must be reversed in telophase because new rounds of replications can generally be initiated in early interphase. It seems a provocative idea that origin binding sites and proteins associated with them may play a crucial role in this regulation process.

Aelen
,
J. M. A.
,
Opstelten
,
R. J. G.
&
Wanka
,
F.
(
1983
).
Organization of the DNA replication in Physarum. Attachment of origins of replicons and replication forks to the nuclear matrix
.
Nucl. Acids Res
.
11
,
1181
1195
.
Ames
,
G. F.-L.
&
Nikaido
,
K.
(
1976
).
Two-dimensional gel electrophoresis of membrane proteins
.
Biochemistry
15
,
616
623
.
Basler
,
J.
,
Hastie
,
N. D.
,
Pietras
,
D.
,
Matsui
,
S.-J.
,
Sandberg
,
A. A.
&
Berezney
,
R.
(
1981
).
Hybridization of nuclear matrix attached deoxyribonucleic acid fragments
.
Biochemistry
20
,
6921
-
6929
. ‘
Bekers
,
A. G. M.
(
1982
).
Structural components of the cell nucleus and the organization of chromosomal DNA. Ph.D. thesis, University of Nijmegen
.
Bekers
,
A. G. M.
,
Gijzen
,
H. J.
,
Taalman
,
R. D. F. M.
&
Wanka
,
F.
(
1981
).
Ultrastructure of the nuclear matrix from Physarum polycephalum during the mitotic cycle
.
J. Ultrastruct. Res
.
75
,
352
-
362
. ‘
Berezney
,
R.
(
1979
).
Effect of protease inhibitors on matrix proteins and the association of replicating DNA
.
Expl Cell Res
.
123
,
411
414
.
Berezney
,
R.
(
1984
).
Organization and functions of the nuclear matrix
. In
Chromosomal Nonhistone Proteins, vol. IV
(ed.
L. S.
Hnilica
), pp.
119
180
.
Boca Raton
:
CRC Press
.
Berezney
,
R.
&
Coffey
,
D. S.
(
1974
).
Identification of a nuclear protein matrix
.
Biochem. biophys. Res. Commun
.
60
,
1410
1417
.
Berezney
,
R.
&
Coffey
,
D. S.
(
1977
).
Nuclear matrix. Isolation and characterization of a framework structure from rat liver nuclei
.
J. Cell Biol
.
73
,
616
637
.
Bouvier
,
D.
,
Hubert
,
J.
,
Seve
,
A.-P.
&
Bouteille
,
M.
(
1985
).
Characterization of laminabound chromatin in the nuclear shell isolated from HeLa cells
.
Expl Cell Res
.
156
,
500
512
.
Buongiorno-Nardelu
,
M.
,
Micheli
,
G.
,
Carri
,
M. T.
&
Marilly
,
M.
(
1982
).
A relation between replicon size and supercoiled loop domains in the eukaryotic genome
.
Nature, Land
.
298
,
100
102
.
Carri
,
M. T.
,
Micheli
,
G.
,
Graziano
,
E.
,
Pace
,
T.
&
Buongiorno-Nardelu
,
M.
(
1986
).
The relationship between chromosomal origins of replication and the nuclear matrix during the cell cycle
.
Expl Cell Res
.
164
,
426
436
.
Cook
,
P. R.
,
Brazell
,
I. A.
&
Jost
,
E.
(
1976
).
Characterization of nuclear structures containing superhelical DNA
.
J. Cell Sa
.
22
,
303
324
.
Dukwel
,
P. A.
,
Mullenders
,
L. H. F.
&
Wanka
,
F.
(
1979
).
Analysis of the attachment of replicating DNA to the nuclear matrix in mammalian interphase nuclei
.
Nucl. Acids Res
.
6
,
219
230
.
Dukwel
,
P. A.
&
Wenink
,
P. W.
(
1986
).
Structural integrity of the nuclear matrix. Differential effects of thiol-agents and metal-chelators
.
J. Cell. Sei
.
84
,
53
67
.
Dukwel
,
P. A.
,
Wenink
,
P. W.
&
Poddighe
,
J.
(
1986
).
Permanent attachment of replication origins to the nuclear matrix in BHK cells
.
Nucl. Acids Res
.
14
,
3241
3249
.
Fussel
,
C. P.
(
1975
).
The position of the interphase chromosomes and late replicating DNA in centromere and telomere regions of Allium cepa L
.
Chromosoma
50
,
201
210
.
Galcheva-Gargova
,
Z.
,
Petrov
,
P.
&
Dessev
,
G.
(
1982
).
Effect of chromatin decondensation on the intranuclear matrix
.
Eur.J. Cell Biol
.
28
,
155
159
.
Hameister
,
H.
&
Sperling
,
K.
(
1984
).
Description of a chromosome replication unit in individual prematurely condensed human S-phase chromosomes
.
Chromosoma
90
,
389
393
.
Hochstrasser
,
M.
,
Mathog
,
D.
,
Gruenbaum
,
Y.
,
Saumweber
,
H.
&
Sedat
,
J. W.
(
1986
).
Spatial organization of chromosomes in salivary gland nuclei of Drosophila melanogaster. J. Cell Biol
.
102
,
112
123
.
Kaufmann
,
S. H.
,
Coffey
,
D. S.
&
Shaper
,
J. H.
(
1981
).
Considerations in the isolation of rat liver nuclear matrix, nuclear envelope and pore complex lamina
.
Expl Cell Res
.
132
,
105
132
.
Kaufmann
,
S. H.
&
Shaper
,
J. H.
(
1984
).
A subset of non-histone nuclear proteins reversibly stabilized by the sulfhydryl cross-linking reagent tetrathionate
.
Expl Cell Res
.
155
,
477
495
.
Laemmli
,
U. K.
(
1970
).
Cleavage of structural proteins during the assembly of the head of bacteriophage T4
.
Nature, Land
.
277
,
680
685
.
Marsden
,
M. P. F.
&
Laemmli
,
U. K.
(
1979
).
Metaphase chromosome structure: Evidence for a radial loop model
.
Cell
17
,
849
858
.
Mitchelson
,
K. R.
,
Bekers
,
A. G. M.
&
Wanka
,
F.
(
1979
).
Isolation of a residual protein structure from nuclei of the myxomycete Physarum polycephalum. J. Cell Sci
.
39
,
247
256
.
Mullenders
,
L. H. F.
,
Eygensteyn
,
J.
,
Broen
,
A.
&
Wanka
,
F.
(
1982
).
Composition and DNA binding properties of the nuclear matrix proteins from mammalian cell nuclei
.
Biochem. biophys. Acta
598
,
70
77
.
Nicklas
,
R. B.
&
Staehli
,
C. A.
(
1967
).
Chromosome micromanipulation. I. The mechanics of chromosome attachment to the spindle
.
Chromosoma
21
,
1
16
.
O’farell
,
P. H.
(
1975
).
High resolution two-dimensional electrophoresis of proteins
.
J. biol. Chem
.
250
,
4007
4021
.
Paulson
,
J. R.
&
Laemmli
,
U. K.
(
1977
).
The structure of histone depleated chromosomes
.
Cell
12
,
817
828
.
Pieck
,
A. C. M.
,
Van Der Velden
,
H. M. W.
,
Ruken
,
A. A. M.
,
Neis
,
J. M.
&
Wanka
,
F.
(
1985
).
Protein composition of the chromosomal scaffold and interphase nuclear matrix
.
Chromosoma
91
,
137
144
.
Rao
,
P. N.
,
Wilson
,
B.
&
Puck
,
T. T.
(
1977
).
Premature chromosome condensation and cell cycle analysis
.
J. cell. Physiol
.
91
,
131
141
.
Razin
,
S. V.
,
Yarovaya
,
O. V.
&
Georgiev
,
G. P.
(
1985
).
Low ionic strength extraction of nuclease-treated nuclei destroys the attachment of transcriptionally active DNA to the nuclear skeleton
.
Nucl. Acids Res
.
13
,
8428
8444
.
Robinson
,
S. I.
,
Nelkin
,
B. D.
&
Vogelstein
,
B.
(
1982
).
The ovalbumin gene is associated with the nuclear matrix of chicken oviduct cells
.
Cell
28
,
99
106
.
Ross
,
D. A.
,
Yen
,
R. W.
&
Chae
,
C. B.
(
1982
).
Association of globin ribonucleic acid and its precursors with the chicken erythroblast nuclear matrix
.
Biochemistry
21
,
764
771
.
Shaper
,
J. H.
,
Pardoll
,
D. M.
,
Kaufmann
,
S. H.
,
Barrack
,
E. R.
,
Vogelstein
,
B.
&
Coffey
,
D. S.
(
1979
).
The relationship of the nuclear matrix to cellular structure and function
.
Adv. Enzyme Regul
.
17
,
213
248
.
Van Der Velden
,
H. M. W.
,
Van Willigen
,
G.
,
Wetzels
,
R. H. W.
&
Wanka
,
F.
(
1984
).
Attachment of origins of replication to the nuclear matrix and chromosomal scaffold
.
FEBS Lett
.
171
,
13
16
.
Van Eekelen
,
C. A. G.
&
Van Venrooy
,
W. J.
(
1981
).
hnRNA and its attachment to a nuclear protein matrix
.
J. Cell Biol
.
88
,
554
563
.
Vogelstein
,
B.
,
Pardoll
,
D. M.
&
Coffey
,
D. S.
(
1980
).
Supercoiled loops and eucaryotic DNA replication
.
Cell
22
,
79
85
.
Von Wettstein
,
D.
,
Rasmussen
,
S. W.
&
Holm
,
D. B.
(
1984
).
The synaptonemal complex in genetic segregation. A
.
Rev. Genet
.
18
,
331
413
.
Wanka
,
F.
,
Brouns
,
M. G. M. E.
,
Aelen
,
J. M. A.
&
Eygensteyn
,
J.
(
1977a
).
The origin of nascent single-stranded DNA extracted from mammalian cells
.
Nucl. Acids Res
.
4
,
2083
2097
.
Wanka
,
F.
,
Mullenders
,
L. H. F.
,
Bekers
,
A. G. M.
,
Pennings
,
L. J.
,
Aelen
,
J. M. A.
&
Eygensteyn
,
J.
(
1977b
).
Association of nuclear DNA with a rapidly sedimenting structure
.
Biochem. biophys. Res. Commun
.
74
,
739
747
.
Wanka
,
F.
,
Pieck
,
A. C. M.
,
Bekers
,
A. G. M.
&
Mullenders
,
L. H. F.
(
1982
).
The attachment of replicating DNA to the nuclear matrix
. In
The Nuclear Envelope and the Nuclear Matrix
(ed.
G. G.
Maul
), pp.
199
211
.
New York
:
Alan R. Liss Inc
.
Wilson
,
D. L.
,
Hall
,
M. E.
,
Stone
,
G. C.
&
Rubin
,
R. W.
(
1977
).
Some improvements in twodimensional gel electrophoresis of proteins. Protein mapping of eucaryotic tissue extracts
.
Analyt. Biochem
.
83
,
33
44
.