Resinless section immunogold electron microscopy of karyo-cytoskeletal frameworks of eukaryotic cells cultured in vitro: Absence of a salt-stable nuclear matrix from mouse plasmacytoma MPC-11 cells

Resin ultrathin section (Lawson, 1983; Wang et al. 1983) and whole-mount (Webster et al. 1978; Henderson and Weber, 1979, 1980; Heuser and Kirschner, 1980; De Mey et al. 1981; Granger and Lazarides, 1982; Birrell et al. 1987; Carmo-Fonseca et al. 1987; Katsuma et al. 1987; Murti and Goorha, 1989; Willingale-Theune et al. 1989) electron microscopy in combination with labeling of cellular proteins with gold-conjugated antibodies have contributed a great deal to the structural and functional characterization of cytoskeletal elements in eukaryotic cells. However, both techniques are handicapped by some inherent limitations, which have become more and more obvious as increased resolution and specification of cellular structures and constituents were required. One disadvantage of resin ultrathin section electron microscopy is, by definition, the thinness of individual sections, which does not allow the longitudinal extension of filamentous structures to be followed over longer distances unless the filaments are lying in the plane of section. Another unfavorable feature of this method is the destruction of the antigenicity of many cellular constituents caused by violent sample processing and masking of epitopes as a result of resin embedment. Of course, immunolabeling of cellular structures before embedment of cells or cell models into resin and ultrathin sectioning is possible, although not universally applicable. These problems were partially overcome when ultrathin cryosectioning was introduced to the field, but normally the preservation of cellular structures in cryosections is rather poor, so that only a limited number of investigations of the cytoskeleton, largely restricted to muscle (Tokuyasu, 1983; Tokuyasu et al. 1983a,b ; Langanger and De Mey, 1988), have been undertaken so far. In addition, the cryosection technique demands great experimental skilfulness from the individual researcher. On the other hand, the whole-mount technique applied not only to substratum-attached cells (Wolosewick and Porter, 1975, 1976; Brown et al. 1976; Osborn and Weber, 1977; Lenk et al. 1977; Small and Celis, 1978; Ip and Fishman, 1979; Pudney and Singer, 1979; Fulton et al. 1980; Schliwa and Van Blerkom, 1981; Schliwa, 1982; Ris, 1985) but to suspension-grown cells as well (Wang et al. 1989), has been developed to serve as an excellent substitute for the resin ultrathin section technique in analysing the cytoskeleton of eukaryotic cells. Nevertheless, although it is well suited to reveal the three-dimensional organization not only of cytobut also of peripheral karyoskeletal elements, it is inappropriate in the case of section analysis; for instance, when detailed information on the interior of the nucleus is sought.

With this background, the need for a technique that permits immunolabeling of the karyo-cytoskeleton in resinless sections is obvious. To date, two variations of resinless ultrathin sectioning have been established employing either polyethylene glycol (Wolosewick, 1980) or diethylene glycol distearate (DGD) (Capeo et al. 1984) as embedding material. With the use of DGD, it has been possible to obtain informative electron-microscopic pictures of both the karyo- and the cytoskeleton of cells cultured in. vitro (Capeo et al. 1984; Fey et al. 1986). This method has been improved by Jackson et al. (1988) by preembedding cells in agarose beads, following the procedure of Nilsson et al. (1983). The entrapment of cells in agarose prevents changes in cytoskeletal architecture that arise in response to mechanical stress inflicted upon the cells during their sequential extraction and repeated centrifugation as required by the original procedure of Capeo et al. (1984).

We have used minor modifications of a combination of the procedures of Capco et al. (1984) and Jackson et al. (1988) to perform an immunogold electron microscopic investigation of the karyo-cytoskeleton of mouse skin fibroblasts (MSF) and mouse plasmacytoma MPC-11 cells. MSF cells possess a well-developed cytoplasmic vimentin filament system and a nuclear lamina consisting of the three major subunit proteins, lamins A, B and C (Wang et al. 1989). On the other hand, MPC-11 cells are devoid of detectable amounts of vimentin and other cytoplasmic intermediate filament (IF) proteins and possess lamin B as the only nuclear lamina subunit protein (Paulin-Levasseur et al. 1988). However, they can be induced by the phorbol ester 12-O-tetradecanoylphorbol-13-acetate (TPA) to synthesize vimentin and to assemble it into IFs (Giese and Traub, 1986, 1988; Paulin-Levasseur et al. 1989). Our experimental results show that, in addition to the expected karyo- and cytoskeletal elements, MSF cells contain a well-defined intranuclear, fibrogranular network that is presumably responsible for the fixation of chromatin, whereas MPC-11 cells in both their normal and TPA-induced states are devoid of a salt-stable nuclear scaffold.

Diethylene glycol distearate (DGD) was obtained from Polysciences (Warrington, PA, USA). Agarose (type VII), poly-L-lysine (M_r 170000), fish gelatin and 12-O-tetradecanoylphorbol-13-acetate (TPA) were purchased from Sigma (St Louis, MO, USA). Collagen S (type I), bovine pancreatic DNase I (EC 3.1.21.1) and RNase A (EC 3.1.27.5) were from Boehringer-Mannheim (Mannheim, FRG), and grids, fixatives and Pioloform were from Plano (Marburg, FRG). Other biochemicals and all reagent grade chemicals were purchased from Merck (Darmstadt, FRG), Roth (Karlsruhe, FRG) and Serva (Heidelberg, FRG).

Embryonic mouse skin fibroblast (MSF) cells were grown as adherent monolayers in minimal essential Eagle’s medium (MEM; Flow Laboratories, Meckenheim, FRG) supplemented with 10 % fetal calf serum (FCS) (Biochrom, Berlin, FRG).

Mouse plasmacytoma cells MPC-11 (CCL 167, American Type Culture Collection, Rockville, MD, USA) were grown in suspension culture in Dulbecco’s modified Eagle’s medium (DMEM; Flow Laboratories) supplemented with 20% heat-inactivated horse serum (Boehringer, Mannheim). For the induction of vimentin synthesis, MPC-11 cells were treated at a starting density of 5 × 10⁵ cells ml^-1 with 10^-7 M TPA for 20 h (Giese and Traub, 1988) (for electron microscopy) or with 3 ×10^-8M TPA for 18 h (for biochemical analysis). Control cells were treated with the TPA solvent ethanol alone (0.1%, v/v, final concentration).

Cells were embedded in agarose following published methods (Nilsson et al. 1983; Cook, 1984; Jackson et al. 1988), with some modifications. MSF cells grown as a monolayer were washed twice with phosphate-buffered saline (PBS: 137 mM NaCl, 2.7 mM KC1, 1.5 mu KH₂PO₄, 8mM Na₂HPO₄), scraped off from the substratum at room temperature and pelleted by low-speed centrifugation; 2 ×10⁷ cells were resuspended in 100 –200 μl PBS at 37 °C. In the case of MPC-11 or TPA-treated MPC-11 cells, 3.5 ×10⁷ PBS-washed cells were suspended in the same volume of PBS. One third of the volume of the cell suspension of 2.5 % agarose in PBS, previously melted at 65 °C and then cooled down to 37 °C, was mixed into the cell suspension. The resulting mixture was spread out on parafilm to form a layer thinner than 1 mm and allowed to solidify at room temperature. It was then cut into small pieces approximately 2 mm × 2 mm in size. The agarose pieces with the entrapped cells were then carried through the extraction-digestion procedure described by Fey et al. (1984), using 10 ml buffer in each step. Briefly, the agarose pieces were incubated with cytoskeleton buffer I (CSK I: 100 mM NaCl, 3mM MgCl₂, 1mM EGTA, 10 mM Pipes, pH 6.8, 300 mM sucrose, 0.5% Triton X-100, 1 mM PMSF) for 10 min on ice, with CSK II buffer (same as CSK I, except that NaCl was replaced by 250 mM (NH₄)₂SO₄) for another 20 min on ice, with nuclease digestion buffer (same as CSK I but containing 50 mM NaCl) containing 100 μgml^-1 of both DNase I and RNase A for 40 min at room temperature and, finally, with CSK II buffer for 10 min again at room temperature. Centrifugation between the individual treatments as in the original procedures was unnecessary because the buffers could be quantitatively removed from the settled agarose pieces with the aid of a Pasteur pipette.

The agarose-embedded cytoskeletons were prefixed in 0.2% glutaraldehyde in nuclease digestion buffer lacking Triton X-100 for 5 min on ice. Excess glutaraldehyde was reduced with 0.05% NaBFL, in PBS for 10 min on ice. After three washes with TBS (20 mM Tris-HCl, pH 8.2, 0.5 M NaCl, 0.1% bovine serum albumin (BSA), 0.005% Tween 20, 0.001% streptomycin), the cytoskeletons were incubated with primary antibodies in TBS containing 5 % FCS for 1 h in an Eppendorf tube or on parafilm in a humid atmosphere. Excess primary antibody was removed by three washes with TBS. This was followed by incubation of the cytoskeletons with the appropriate gold-conjugated secondary antibodies or gold-conjugated protein A in TBS containing 0.5 % fish gelatin for 1h. After three washes with TBS, the cytoskeletons were finally fixed with 2% glutaraldehyde in 0.1M sodium cacodylate, pH 7.4, for 30 min and post-fixed with 1 % OsO₄ in the same buffer. All operations following the reduction of glutaraldehyde with NaBH₄ were carried out at room temperature. For controls, the primary antibodies were omitted.

The immunolabeled cytoskeletons were embedded in DGD as described by Capeo et al. (1984). For this purpose, they were dehydrated through an ascending ethanol series at room temperature and then carried through an ascending n-butanol series (ethanol:n-butanol=2:l, 1:2, v/v) terminating in pure n-butanol and an ascending DGD series (n-butanol:DGD=2:l, 1:2, v/v) terminating in pure DGD, both at 60 °C. After solidification of the DGD by cooling to room temperature, the hard DGD blocks were sectioned with an ultramicrotome. Nickel grids coated with 3% Pioloform in chloroform and then with 0.1 % poly-L-lysine in water were used to support the sections. The DGD was removed from the sections by a 3 h treatment at room temperature with n-butanol, which in turn was substituted for by ethanol. Following CO₂ critical-point drying with a Balzers CPD 020 instrument (Balzers Hochvakuum GmbH, Wiesbaden, FRG), the sections were viewed in a Philips 400 T electron microscope (Philips, Kassel, FRG) at 80 kV. Some sections were shadowed with tungsten/carbon in an Edwards Coating System E 306 A (Edwards, Crawley, England) in order to enhance the surface structure of karyo-cytoskeletal filaments.

MSF cells grown for 24 h on nickel grids coated with 3 % Pioloform and then with 0.03% collagen S in water were subjected to extraction and immunolabeling as described above and finally critical-point-dried in CO₂.

While MSF cells were grown on coverslips to subconfluency, MPC-11 cells propagated in suspension culture were allowed to attach to coverslips coated with 0.1% poly-L-lysine for 10 min at room temperature. The cells were extracted with CSK I and CSK II buffer on ice, except that the latter contained 0.75 M KC1 instead of 0.25 M (NH₄)₂SO₄. After a 10 min fixation with 4% formaldehyde in PBS on ice, the cell residual structures were incubated with primary and secondary antibodies each for 1 h at room temperature and counterstained with Hoechst 33258. The specimens were viewed in a Zeiss ICM 405 microscope (Zeiss, Oberkochen, FRG) equipped for epifluorescence and phasecontrast and photographed on Ilford XP1-400 film (Ilford Ltd, Mobberly, Cheshire, UK).

A 3.5 g sample of MSF cells, which were grown to confluency on 242-liter roller bottles, scraped off from the substratum with the aid of a rubber policeman, pelleted by low-speed centrifugation and washed with PBS, were gently stirred with a spatula in 35 ml of CSK I buffer containing 2 mM vanadyl adenosine for 5 min at 2 °C and pelleted by a 5 min centrifugation at 1100 g_av. The pellet was resuspended in 35 ml of CSK II buffer containing 2 mM vanadyl adenosine by stirring at 2 °C, and, after 5 min, the suspension was centrifuged at 1100 g_av for 5 min. This was followed by a 30 min incubation of the resuspended sediment at room temperature in 35 ml of digestion buffer (same as CSK I but containing 50 mM NaCl) in the presence of 2mM vanadyl adenosine and 200 μg ml^-1 DNase I. After adjustment of the suspension to 250 mM (NH₄)₂SO₄, insoluble material was pelleted by centrifugation at 1100 g_av for 5 min, resuspended in 35 ml of CSK I buffer containing 40 μg ml^-1 RNase A, incubated at room temperature for 10 min and again centrifuged at 1100 g_av for 5 min. Finally, the residual material was extracted with 17.5 ml of 10 mM Pipes, pH 6.8, 3mM EDTA, 6mM 2-mercaptoethanol, 8 M urea in a loosely fitting Dounce homogenizer at room temperature. After 30 min, a small amount of insoluble material was pelleted by centrifugation at 37 000g_av for 10 min. It was dissolved in 2 ml SDS sample buffer by brief sonication and heating at 100°C for 2 min. In the case of normal and TPA-treated, PBS-washed MPC-11 cells (3.5 g), the extraction procedure was the same, except that after the first extraction with CSK I buffer all other extractions were carried out in a loosely fitting Dounce homogenizer until a homogeneous suspension of cellular material was obtained. Because of the high viscosity of the suspension of cell residues in (NH4)₂SO₄-containing CSK II buffer, the following centrifugation had to be performed at 37 000g_av for 30 min. All following centrifugations were performed at the same centrifugal force but for 10 min only.

While for SDS—polyacrylamide gel electrophoresis (SDS-PAGE) a portion of the first (CSK I) extract was mixed with 0.25 volume of 5 × SDS sample buffer, portions of all other extracts were mixed with the same volume of 40% TCA (trichloroacetic acid) and the precipitates were collected by centrifugation, washed with acetone and dissolved in SDS sample buffer. SDS-PAGE was earned out as described previously (Egberts et al. 1976). For protein identification, the electrophoretically separated proteins were either stained with Coomassie Brilliant Blue or transferred onto nitrocellulose foil (Millipore GmbH, Eschborn, FRG), following the method of Kyhse-Andersen (1984), and visualized by immunoblotting with various antibodies (see Antibodies) and ¹²⁵I-labeled protein A and fluorography (Paulin-Levasseur et al. 1989).

The extraction of control MPC-11 cells was also carried out in the absence of vanadyl adenosine and under slightly oxidizing conditions. In one case, the Triton cytoskeletons obtained by extraction of cells with CSK I buffer were treated for 10 min at 2°C with 1 mM CuCl₂ (Rzeszowska-Wolny et al. 1988) in CSK I buffer and then washed three times with the appropriate volume of CSK I buffer. All the following steps were the same as those described above. In another case, all extraction buffers contained 0.5 mM sodium tetrathionate (Bladon et al. 1988).

For immunogold labeling, affinity-purified, polyclonal goat antimouse vimentin antibody (Giese and Traub, 1986) and polyclonal rabbit anti-mouse lamin B antibody (Traub et al. 1988) were used at working dilutions of 1:10 in TBS/FCS (20 mM Tris-HCl, pH 8.2, 0.5 M NaCl, 0.1% BSA, 0.005% Tween 20, 0.001% streptomycin, supplemented with 5% FCS). The mouse monoclonal IgM antibody against all three lamins (Burke et al. 1983) was produced in suspension culture of the hybridoma cell line 41 CC4 kindly provided by Dr G. Warren (University of Dundee, Dundee, Scotland) and used at a working dilution of 1:10 in TBS/FCS. Rabbit anti-goat Ig- and goat anti-mouse IgM-5nm gold conjugates were purchased from Janssen Life Sciences Products (Bruggen, FRG). 18 nm gold-coryugated protein A was prepared according to a modification of the method described by Slot and Geuze (1985). All gold-conjugated secondary probes were used at working dilutions of 1:20 in TBS containing 0.5% fish gelatin (TBS/gelatm) (Behnke, 1981; Birrell et al. 1987).

For immunofluorescence microscopy of MPC-11 cells, rabbit anti-lamin B (1:50 in TBS/FCS) was used as primary antibody and FITC-conjugated swine anti-rabbit Ig (Dakopatts GmbH, Hamburg, FRG; 1:50 in TBS/gelatin) as secondary antibody. In the case of MSF cells, monoclonal mouse anti-lamin (1:50 in TBS/FCS) and TRITC-conjugated rabbit anti-mouse Ig (1:50 in TBS/gelatin) were employed as primary and secondary antibodies, respectively.

For immunoblotting, the various antibodies were used at the following working dilutions (m 10 mM Tris-HCl, pH 7.6, 150 mM NaCl, 0.05 % Tween 20,0.1 % BSA, 1 % FCS): polyclonal goat anti-vimentin, 1:40; polyclonal rabbit anti-lamin B, 1:500; monoclonal anti-lamin, 1:10; mouse monoclonal anti-IFA directed against an epitope common to most IF proteins (Pruss et al. 1981) (produced in suspension culture of the hybridoma cell line TIB 131 from the American Type Culture Collection), 1:10. The secondary rabbit anti-mouse IgG (Dakopatts) was used at a working dilution of 1:100.

In order to demonstrate and differentiate between the specific advantages of whole-mount and resinless section immunogold electron microscopy in visualizing structural elements of the karyo-cytoskeletal frameworks of eukaryotic cells, MSF cells were used for comparison. Employing the procedure described by Fey et al. (1984), which entails sequential extraction of monolayer- or suspension-grown cells with non-ionic detergent and high ionic strength buffer followed by removal of chromatin and ribonucleoprotein material with DNase and RNase, whole-mount preparations of MSF cells were obtained in which the cytoskeleton could be clearly distinguished from the residual nuclear body. Most of the cytoplasmic, filamentous structures were identified as vimentin filaments, while the surface of the residual nucleus was densely decorated by a monoclonal anti-lamin antibody (not shown; for an example, see Wang et al. 1989). However, in comparison with the clear presentation of the cytoplasmic IF network, no structual details could be observed in the interior of the nucleus.

Principally, the same structural organization of karyo-cytoskeletal frameworks was seen when the resinless section technique was applied to MSF cells originally grown in monolayer culture and then scraped off from the substratum. Yet, in contrast to the whole-mount procedure the resinless section technique permitted a glimpse at the interior of the nucleus. Extraction of cells with Triton X-100-containing low-salt CSK I buffer alone revealed a dense lamina and an extended fibrogranular network as the most prominent substructures of the cell nuclei (Fig. 1A). A better presentation of the karyoskeletal structures was possible when the Triton cytoskeletons were additionally extracted with the high ionic strength buffer CSK II, and chromatin and ribonucleoprotein material were removed by nuclease digestion (Fig. 1B). Owing to the extrication of undefined and structureobscuring cellular material, not only the cytoskeleton but also the nuclear lamina and the internal nuclear matrix were more clearly visualized. Unexpectedly, the nuclear lamina could not be labeled with a polyclonal rabbit antilamin B antibody and 18 nm gold-conjugated protein A when the MSF cells were extracted with CSK I buffer alone (Fig. 1A). This could be achieved only after removal of histone Hl from chromatin with the high ionic strength buffer CSK II, digestion of the residual cell structures with DNase I and RNase A and washing of the digested cell models with CSK II buffer. The distribution of the 18 nm gold particles was restricted to the densely stained borderline between the cytoplasm and the nuclear interior (Fig. 1B).

Although the anti-vimentin and anti-lamin antibodies applied did not decorate structures in the interior of MSF cell nuclei, resinless section electron microscopy detected an intra-nuclear system of branching filaments that, in agreement with the data published by Jackson and Cook (1988) and He et al. (1990), were structurally very similar to cytoplasmic vimentin filaments. Fig. 2A shows a higher magnification of the nuclear interior of the DNase- and RNase-digested MSF Triton cytoskeletons depicted in Fig. 1A. Filaments (Fig. 2B) with a morphology almost identical to that of cytoplasmic vimentin filaments (Fig. 2C) were still covered with globular structures representing either residual nucleoprotein material or aggregated non-histone proteins (Fig. 2A). Short stretches of uncovered nuclear filaments even exhibited the same surface structure as cytoplasmic vimentin filaments.

The usefulness of the present technique of resinless section immunogold electron microscopy became particularly apparent when MPC-11 cells with their extremely labile and irregularly shaped nuclei were chosen as the subject of investigation. Extraction of suspension-grown, agarose-entrapped MPC-11 cells with CSK I buffer followed by treatment with polyclonal goat anti-vimentin and 5nm gold-conjugated anti-goat IgG antibody, and sectioning of the cell models, yielded specimens in which the majority of nuclei showed a high but heterogeneous chromatin density (Fig. 3). A substantial fraction of the nuclei was distinguished by the presence of much less dense chromatin (not shown). In addition, all cells harboring nuclei with largely unfolded chromatin (very likely proliferating cells), as well as most of the cells with highly condensed chromatin (Fig. 3A,B), were devoid of a cytoplasmically extended vimentin filament network. On the other hand, in approximately 5% of the cells of a proliferating population the nuclei were entangled in a cytoplasmic meshwork of vimentin filaments (Fig. 3C,D). All these cells possessed nuclei with highly condensed chromatin; they very likely represented spontaneously ‘differentiating’ cells that were no longer able to proliferate.

Further extraction of the Triton X-100-resistant residual cell structures with high ionic strength CSK II buffer and removal of chromatin and ribonucleoprotein material before immunolabeling of the karyo-cytoskeletal elements furnished cell models with an empty space at the original location of the nucleus (Figs 4B and 5A,B). Approximately 95 % of these cell models were vimentin-negative. In order to retrace the original nuclear surface, the residual cell structures were treated with the polyclonal anti-lamin B antibody and 18 nm gold-conjugated protein A. Whereas in cell residues that had not been washed with high-salt buffer and digested with nucleases, lamin B was not recognized by the anti-lamin B antibody (Fig. 4A), a situation identical to that seen with undigested karyo-cytoskeletal frameworks derived from MSF cells (Fig. 1A); in digested frameworks the nuclear lamina could be visualized as a continuous band by the immunospecific deposition of 18 nm gold particles (Fig. 4B).

We also investigated whether the ‘differentiation’ of MPC-11 cells induced by the phorbol ester TPA had some influence on the structural organization of the nucleus. It should be noted that the vast majority of differentiated cells contained a nucleus with highly condensed chromatin and a cytoplasmically extended vimentin filament system (Fig. 5A,B), and that this state was indistinguishable from that observed in spontaneously differentiating MPC-11 cells (Fig. 3C,D). However, also in the case of TPA-treated MPC-11 cells, extraction with Triton X-100 and high-salt buffer, nuclease removal of chromatin and ribonucleoprotein material and immunolabeling of karyo-cytoskeletal structures left more or less empty spaces at the original locations of the nuclei. These empty spaces were also surrounded by a continuous, densely stained band identified as the nuclear lamina by labeling with anti-lamin B and 18 nm gold-conjugated protein A (not shown) and an interwoven network of vimentin filaments, which could be decorated with the anti-vimentin and corresponding 5nm gold-conjugated secondary antibody (Fig. 5A). The complete absence of gold particles from the position of the nuclear lamina in undigested karyo-cytoskeletal frameworks showed that also in TPA-treated MPC-11 cells the epitopes of lamin B were masked (not shown), whereas the absence of gold particles from the site of vimentin filaments, noted when the primary antibody was omitted, demonstrated the high specificity of the immunolabeling reaction (Fig. 5B). It is noteworthy that under the conditions of specimen preparation employing the present resinless section technique the original spatial extension of the nuclear lamina was maintained fairly well, although MPC-11 cells, in their untreated or TPA-treated state, were lacking a stabilizing intra-nuclear matrix. Even rather irregular nuclear shapes, like that of a horseshoe in Fig. 5A, could be traced throughout the section with a satisfactory degree of accuracy.

Parallel to the characterization of karyo-cytoskeletal frameworks by resinless section immunogold electron microscopy, the fate of cytoplasmic vimentin filaments and constituents of the nuclear matrix during extraction of cells with various buffers followed by digestion of nuclear DNA and RNA was studied by biochemical analysis. For this purpose, the supernatant of each extraction and digestion step was subjected to SDS-PAGE and immuno-blot analysis. When MSF cells grown in mass culture were extracted with CSK I buffer and then, in order to remove histone Hl from chromatin, with CSK II buffer, all cytoplasmic vimentin filaments and nuclear lamina material were retained in compact residual cell structures in which the nuclear chromatin remained tightly folded. In immunoblot analysis employing the monoclonal anti-IFA (Fig. 6B), polyclonal anti-vimentin (Fig. 6C), monoclonal anti-lamin (Fig. 6D) and polyclonal anti-lamin B (Fig. 6E) antibodies, the respective supernatants were free of vimentin and nuclear lamins. As shown by Coomassie Brilliant Blue staining (Fig. 6A, lanes 1 and 2), the core histones were also totally retained in the karyo-cytoskeletons, but were selectively released upon DNA digestion with DNase I (Fig. 6A, lane 3). Digestion of the residual cell structures with RNase A also did not solubilize vimentin and nuclear lamins (Fig. 6, lanes 4). Only the extraction of the remaining cell models with 8 M urea solubilized these cyto- and karyoskeletal proteins (Fig. 6, lanes 5). Small quantities, however, were left in the ureainsoluble material (Fig. 6, lanes 6) as well as small amounts of the core histones, with a preponderance of the arginine-rich histones H3 and H4. The poor resolution of the protein constituents of the urea-insoluble cell matrices suggests protein-protein and protein-DNA cross-linking.

In comparison to the results obtained with MSF cells, those obtained with MPC-11 cells were clearly different. The first significant difference was observed when Triton cytoskeletons derived from MPC-11 cells, irrespective of whether these had been treated with TPA (Fig. 8) or not (Fig. 7), were extracted with the high ionic strength buffer CSK II. In both cases, the Triton cytoskeletons could not be resuspended any more because of massive leakage of unfolded chromatin from the nuclei. To be able to handle the sticky material, it had to be gently homogenized in a loosely fitting Dounce homogenizer. This, however, led to breakage of the unfolded chromatin strands so that a substantial fraction of chromatin remained in the supernatant during the following centrifugation. Consequently, large amounts of core histones (and DNA) were detected in this supernatant although the residual cell structures had not been digested with DNase I (Figs 7 A and 8A, lanes 2).

In addition, immunoblotting with the monoclonal antilamin B antibody revealed the release of a considerable amount of lamin B from the karyo-cytoskeletons in the CSK II extraction step (Figs 7C and 8D, lanes 2). From MPC-11 cells treated with TPA, substantial quantities of the induced vimentin filaments were also solubilized as substantiated by immunoblotting with the monoclonal anti-IFA (Fig. 8B, lanes 2) and polyclonal anti-vimentin (Fig. 8C, lanes 2) antibodies. The remaining chromatin, here represented by core histones (Figs 7 A and 8A, lanes 3), and additional vimentin (Fig. 8B,C, lanes 3) were released from the karyo-cytoskeletons by digestion with DNase I. Digestion of the residues with RNase A did not bring about further solubilization of proteinaceous material (Figs 7 and 8, lanes 4), whereas extraction with 8 M urea effected the solubilization of the remaining fractions of lamin B (Figs 7C and 8D, lanes 5) and, in the case of TPA-treated MPC-11 cells, vimentin (Fig. 8B,C, lanes 5). In comparison to the situation seen with MSF cells, urea extraction of the cell residues also solubilized small amounts of chromatin with the preponderance of the core histones H2A and H2B (Figs 7A and 8A, lanes 5). Finally, SDS-PAGE analysis of the urea-insoluble material produced a complex, poorly resolved protein pattern (Figs 7 A and 8A, lanes 6) with a predominance of proteins that in immunoblotting with the monoclonal anti-IFA antibody gave rise to the production of strong fluorography signals (Figs 7B and 8B, lanes 6). Since the same reaction was observed with the secondary anti-mouse IgG antibody alone (not shown), the reactive polypeptides seem to be identical with or derived from the immunoglobulin chains synthesized by MPC-11 cells. They were also contained in the supernatants of the CSK I extraction steps (Figs 7B and 8B, lanes 1; only the reaction of the heavy chain is shown). Strikingly, the urea-insoluble material obtained from TPA-treated (Fig. 8A, lane 6) and untreated cells (Fig. 7A, lane 6) did not contain any residual chromatin as observed in the case of MSF cells (Fig. 6A, lane 6).

Since this leakage of chromatin from nuclei of MPC-11 cells might be due to their exposure to mechanical stress during their suspension in high salt buffer, the cells were also analysed by employing the more gentle immunofluorescence microscopy technique, which largely avoids agitation of the cell population. However, when MPC-11 cells adhering to a polylysine-coated glass surface were treated with CSK II buffer (containing 0.75 M KC1 instead of 0.25 M (NH₄)₂SO₄) after their previous washing with CSK I buffer, all cell nuclei disintegrated (Fig. 9C) with fragmentation of the nuclear lamina (Fig. 9A) and release of chromatin (Fig. 9B). Even in very thin extensions of the lysate remote from the cell center, a codistribution of lamin B and DNA could be observed (Fig. 9A,B). A large amount of the lamin B material appeared to occur in the form of rod- or filament-like structures (Fig. 9A), very similar to the distribution of lamin B in whole-mount preparations of DNase- and RNase-digested Triton cytoskeletons (Fig. 9D). By contrast, the nuclei of MSF cells turned out to be absolutely stable under the same conditions of cell treatment (Fig. 9E to G) in that the chromatin remained compact (Fig. 9F) within an intact nuclear lamina (Fig. 9E).

The resinless section immunogold electron microscopy technique applied here is the result of the combination of two previously described methods (Capeo et al. 1984; Jackson et al. 1988), which were slightly modified. Its development was necessitated mainly by our previous observations that in some hemopoietic cell lines the cell nuclei are of irregular shape and extremely labile (Paulin-Levasseur et al. 1988; Wang et al. 1989). Because of this lability, the study of the karyoskeleton and its relationship with the cytomatrix turned out to be rather difficult. For instance, in whole-mount preparations of karyo-cytoskeletal frameworks of TPA-treated MPC-11 cells the nuclear lamina of nuclease-digested cell models was heavily fragmented, complicating the elucidation of its association with the surrounding vimentin filament system induced by TPA and the visualization of the structural elements of the internal nuclear matrix. In contrast to TPA-treated MPC-11 cells, MSF cells possessed a very stable karyo-cytoskeletal framework (Wang et al. 1989).

In order to protect these delicate structures from disintegration, which in part resulted from their exposure to mechanical stress during sample preparation, the cells to be examined were embedded in low-melting-point agarose and then sequentially extracted with non-ionic detergent-containing buffers of varying composition and subjected to digestion with DNase I and RNase A. In deviation from the original methods described by Nilsson et al. (1983) and Jackson et al. (1988), which provide the entrapment of cells in microbeads of agarose by mixing a cell suspension in melted agarose with paraffin oil, in the present study the cell suspension in agarose/PBS was spread out as a thin layer on parafilm and allowed to solidify. After cutting the solid agarose film into small pieces, the entrapped cells were still accessible to buffer ingredients, enzymes and (gold-conjugated) antibodies. Although the agarose blocks were relatively large in comparison with agarose microbeads in paraffin oil, they nonetheless allowed fragmented and solubilized cellular substructures and constituents to escape. Moreover, since the originally prescribed temperature of 40°C (Nilsson et al. 1983) might cause changes in the structure (and function) of cellular constituents due to heat shock (for reviews, see Craig, 1985; Lindquist, 1986; Subject and Shyy, 1986), the embedding temperature was lowered to 37 °C where the cell suspension in agarose type VII is still liquid. Such heat shock-induced changes in the architecture of MSF cells were occasionally observed when the embedment of the cells into agarose was carried out at 40°C (not shown).

In conjunction with the impregnation of the antibody-decorated residual cell structures with the organic solvent-extractable wax diethylene glycol distearate, this procedure guaranteed an optimal preservation of cellular fine structure during sample processing. It permitted the preparation of resinless sections of karyo-cytoskeletal frameworks and, in particular, a view of the internal nuclear matrix. Therefore, it is especially useful for the immunological detection of intranuclear antigens and their mutual structural relationships. Since it avoids the use of non-extractable resin as embedding material, it is complementary to the whole-mount technique. Principally, both techniques should visualize karyo-cytoskeletal elements with the same ultrastructural resolution and quality, except that the whole-mount technique provides a three-dimensional overview of whole cell residues and the resinless section technique only sections of these structures.

Previously, the cytoskeletal proteins actin (Scheer et al. 1984; Valkov et al. 1989) and tubulin (Douvas et al. 1975; Scheuermann et al. 1988) have been detected in the nuclei of eukaryotic cells. Since the third type of cytoskeletal proteins, the IF proteins, show a high affinity in vitro for nuclear constituents (Traub et al. 1987; Kühn et al. 1987; Shoeman et al. 1988), it is reasonable to assume that they are also associated with intranuclear substructures in order to perform nuclear functions (Traub et al. 1987a). This notion is getting increasing support from immunoelectron-microscopical studies (Paulus and Roggendorf, 1988), from the isolation of covalently cross-linked vimentin-DNA adducts from cultured mammalian cells (Cress and Kurath, 1988) and from the detection of IF-like filaments with 23 nm axial repeats in the interior of cell nuclei by resinless section electron microscopy (Jackson and Cook, 1988; He et al. 1990). It was therefore one of the goals of this study to localize, by means of resinless section immunogold electron microscopy, the IF protein vimentin in the nuclei of MSF cells. Although the nuclei of these cells were filled with an intricate network of filaments of heterogeneous appearance, some having a morphology very similar to that of cytoplasmic vimentin filaments, no gold particles could be immunospecifically deposited on these filaments. It is possible that the vimentin antigen, if it did indeed occur in the nuclei of MSF cells, was still masked after this extraction step and then removed together with chromatin constituents after nuclease digestion of DNA. However, it is equally plausible that the IF-type fibrous structures seen in nuclease-digested MSF cell nuclei consist of distinct subunit proteins different from vimentin and other cytoplasmic IF proteins and nuclear lamins.

As indicated above, another concern of the present investigation was to obtain some information on the architecture of the nuclear matrix of MPC-11 cells. MPC-11 cells possess abnormally shaped and structurally labile nuclei that are not spatially fixed in the cytoplasm by an associated, supporting network of vimentin filaments (Giese and Traub, 1986; Wang et al. 1989). In addition, their nuclear lamina, which very likely also contributes to the stabilization of the nuclear architecture, contains lamin B as the only lamin subspecies (Paulin-Levasseur et al. 1988) and might therefore be structurally labile. TPA-induced formation of vimentin filaments and their association with the nucleus did not provide any improvement with respect to the stability of the nuclear structure (Wang et al. 1989). In extension of these previous findings, the present application of resinless section immunogold electron microscopy to MPC-11 cells demonstrated the essentially total absence from the interior of cell nuclei of a salt-stable, filamentous network as was seen in the nuclei of MSF cells. After removal of chromatin and ribonucleoprotein material by digestion of DNA and RNA with nucleases, the only remainder of the nuclear matrix was the nuclear lamina. Exploiting the advantages of agarose embedment of these cells, this lamina was shown to consist of a continuous shell of proteinaceous material surrounding the chromatin. In previous experiments employing whole-mount immunogold electron microscopy, the lamina was found to be highly fragmented (Wang et al. 1989) (see also Fig. 9D). This aspect of nuclear structure was of interest because it was not previously clear whether lamin B can form a continuous shell around the whole nucleus in the absence of lamins A and C.

Interestingly, the nuclear lamina of both MPC-11 cells and MSF cells could not be labeled with anti-lamin B antibody when the respective nuclei still contained chromatin. Only after removal of the nuclear contents by digestion of the residual cell structures with nucleases and washing of the residues with high ionic strength buffer could this be achieved, indicating masking of the lamin B epitopes by chromatin or chromatin-associated nuclear constituents. In this context, it is pertinent to mention that in vitro the nuclear lamins A and C show a ten times higher affinity for DNA, e.g. telomere sequences (Shoeman and Traub, 1990), than lamin B. Moreover, using lysates of mitotic Chinese hamster ovary cells for the study of nuclear assembly, Burke and Gerace (1986) found that, in extracts depleted of lamin B, lamins A and C bind to chromosomes, whereas, in extracts depleted of lamins A and C, the binding of lamin B to chromosomes is significantly reduced. If such lamin-DNA (chromatin) associations should indeed contribute to the stabilization of the nuclear structure, the lower affinity of lamin B for DNA would provide an additional explanation for the structural instability and abnormal shape of MPC-11 cell nuclei.

Biochemical analysis at the preparative scale confirmed the differences existing between MSF and MPC-11 cells regarding their nuclear matrices. When equal amounts of both types of cells were carried through the extraction and nuclease digestion procedure described by Fey etal. (1984), the MSF cells remained compact throughout whereas the MPC-11 cells underwent disintegration in high ionic strength CSK II buffer, characterized by a burst of chromatin release. During gentle homogenization of the lysed karyo-cytoskeletal frameworks, a substantial amount of lamin B was solubilized. In addition, after induction in MPC-11 cells of vimentin synthesis by TPA and formation of cytoplasmic LFs, a significant quantity of the IF protein was also solubilized. Similar observations have been made with control and TPA-treated human promyelocytic leukemia HL-60 cells, whereas Ehrlich ascites tumor cells and SV40-transformed human skin fibroblasts behaved like MSF cells (not shown). Together with the results of indirect immunofluorescence and resinless section immunogold electron microscopy, these findings suggest the absence of a salt-stable karyoskeletal framework from the interior of MPC-11 or HL-60 cell nuclei to which chromatin is stably anchored during interphase. Furthermore, 8 M urea extraction of chromatin-depleted karyo-cytoskeletal frameworks left some residual cellular material of heterogeneous protein composition, which when derived from MSF cells was substantially larger in quantity than when it was obtained from control MPC-11 cells. In the former case, it still contained core histones, indicating the presence of some nuclease-resistant chromatin, whereas in the latter case such residual chromatin material was completely absent. This observation again suggests that in karyoskeletons derived from MSF cells a small fraction of chromatin was tightly associated with structural elements of the nuclear matrix and thus protected from DNA digestion. These results and considerations are consistent with the view that in histone-depleted interphase nuclei DNA is organized into topologically constrained, supercoiled loops each nonran-domly anchored at its base to elements of the nuclear scaffold (for a short review, see Zehnbauer and Vogelstein, 1985). In the nuclei or nuclear residues of MPC-11 cells, however, no such protection or retention of chromatin could occur because of the lack of a salt-stable nuclear matrix. Our experimental results also show that treatment of MPC-11 cells with TPA did not induce the formation of a salt-stable, intranuclear framework for the anchorage of chromatin.

It might be argued that in MPC-11 and related cells the nuclear scaffold is particularly sensitive to the slightly reducing conditions employed in this study for cell extraction (the extraction buffers CSK I and CSK II contained vanadyl ribonucleoside complexes to inhibit RNases during the first two extraction steps). According to Kaufman et al. (1986), reducing agents exert a destabilizing effect on the internal nuclear matrix. Replacement of the vanadyl ribonucleoside complexes by tetrathionate (Bladon etal., 1988), however, did not bring about any improvement (not shown). Moreover, since copper ions must be considered as an essential structural element of the nuclear matrix (Lebkowski and Laemmli, 1982), the MPC-11 cells were pretreated with 1 mM CuCl₂ according to Rzeszowska-Wolny etal. (1988) before sequential extraction. No changes were observed (not shown). Finally, TPA treatment of MPC-11 cells should stabilize a preexisting nuclear scaffold through the formation of H₂O₂ and oxygen radicals, and cross-linking of scaffold proteins and proteins with DNA, on the basis of a free radical mechanism. This seems possible, since it is known that cells, e.g. leukocytes, suffer DNA strand breakage along a free radical pathway When exposed to TPA (Birnboim and Kanabus-Kaminska, 1985). We found that 8 M urea-insoluble residues of extracted and nuclease-digested MPC-11 cells previously exposed to TPA contained substantially more, and in part cross-linked, proteins in comparison with the corresponding material isolated from control cells. In particular, light and heavy chain immunoglobulins synthesized by MPC-11 cells were enriched in this fraction. Nevertheless, the nuclei of TPA-treated MPC-11 cells were extremely labile and responded with massive chromatin release to treatment with high ionic strength buffer, in agreement with the results of resinless section electron microscopy. We must therefore conclude from the microscopy and biochemical data that the nuclei of MPC-11 cells indeed lack a salt-stable, internal nuclear matrix. Of course, this does not exclude the possible existence of such a matrix that, however, does not survive the experimental manipulations employed here and by others (Fey et al. 1984).

This conclusion, of course, raises the question of how cells without a cytoplasmic IF network, with a nuclear lamina consisting of lamin B as the only lamin subspecies and without a salt-stable intra-nuclear framework can maintain an ordered cycle of cellular activities such as DNA replication, gene expression etc. In the absence of stabilizing structural elements, chromatin organization must obviously and to a large extent be an autonomous process in which ribonucleoproteins probably play an important role. Such a view has been put forward previously by Nickerson et al. (1989) on the basis of results of resinless section electron microscopy studies. Undoubtedly, MPC-11, HL-60 and other hemopoietic cells possess a wealth of nuclear non-histone proteins, which, however, are apparently not able to assemble into salt-stable filamentous structures and to build up a three-dimensional network as is seen in the chromatin-depleted nuclei of MSF or Ehrlich ascites tumor cells. On the assumption that the filamentous structural elements of the internal nuclear matrix are similar or even identical in morphology to cytoplasmic and lamin IFs (Jackson and Cook, 1988; He et al. 1990), MPC-11 or HL-60 cells may be considered as cells with an extremely low, general potential for the synthesis of IF and related proteins. Obviously, the only exception is lamin B, which appears to be constitutively expressed in all eukaryotic cells. Because of these features, MPC-11, HL-60 and related hemopoietic cells provide excellent experimental systems with which to study the synthesis, regulation and cellular functions of individual karyo- and cytoskeletal elements.

Finally, we should refer to two recent publications from Penman’s group that appeared in the literature when the experimental part of this investigation had essentially been completed. Both papers report on results that are very similar to those described here. While He et al. (1990) performed a detailed study on the distribution, morphology and composition of nuclear core filaments, Nickerson et al. (1990) applied resinless section immunogold electron microscopy to localize cytoskeletal and nuclear matrix proteins in mammalian cells. A comparison of the results obtained by Jackson and Cook (1988), He et al. (1990) and ourselves shows that the nuclear filaments detected and analysed are very likely of the same type. Slight differences between the different preparations with respect to their morphology are probably due to differences in the experimental procedures employed. Although He et al. (1990) consider actin and nuclear RNA as essential structural constituents of the nuclear core filaments on the basis of immunological and RNase digestion experiments, the establishment of the structure of these filaments has to await the biochemical characterization of their components. It is striking that the nuclear filaments are morphologically nearly identical to cytoplasmic IFs (Jackson and Cook, 1988) and nuclear lamin filaments, suggesting that they are also of the IF type. If this notion could be substantiated, a continuum of IFs would exist between the plasma membrane and the center of the nucleus. Since both the nuclear (He et al. 1990) and cytoplasmic (Katsuma et al. 1987) IFs appear to be in physical contact or even spatially overlapping with the nuclear lamina, such a continuum of structurally very similar filaments would provide the basis for extensive communication between the cell periphery and the cytoplasm on the one hand and the interior of the nucleus on the other.

We are grateful to Mrs Ulrike Traub and Mrs Margot Bialdiga for the provision of cultured cells, to Miss Annemarie Scherbarth for excellent technical assistance, to Mrs Heidi Klempp for secretarial work and to Dr R. L. Shoeman for critical reading of the manuscript.