Max-Planck-Institut fqr Zellbiologie, Rosenhof, 6802 Ladenburg, Federal Republic of Germany

Transcription

1 J. Cell Sci. 53, (1982) Printed in Great Britain Company of Biologist! Limited 1982 EFFECT OF THE IONIC ENVIRONMENT ON THE INCORPORATION OF THE INTERMEDIATE-SIZED FILAMENT PROTEIN VIMENTIN INTO RESIDUAL CELL STRUCTURES UPON TREATMENT OF EHRLICH ASCITES TUMOUR CELLS WITH TRITON X-100 I. BIOCHEMICAL ANALYSIS PETER TRAUB AND W. JAMES NELSON Max-Planck-Institut fqr Zellbiologie, Rosenhof, 6802 Ladenburg, Federal Republic of Germany SUMMARY Ehrlich ascites tumour cells, propagated in suspension culture, were extracted with Triton X-100 under different ionic conditions to study the sensitivity of intermediate-sized filaments formed in vivo to changes in the ionic environment in vitro. The following results were obtained: (1) in solution of low ionic strength and in the absence of di- and polyvalent cations, vimentin was quantitatively solubilized and recovered from the postnuclear supernatant by precipitation with (NH 4 )tso 4 (23 % saturation) or spermidine (2 IBM). (2) When the cells were extracted in low ionic strength buffer in the presence of 4 mm-mg-acetate or 1-2 mm-spermidine, vimentin was quantitatively incorporated into Triton X-ioo-resistant residual cell structures. It could be easily extracted from Mg l+ -stabilized cell residues with Mg l+ - and polyamine-free buffer of low salt concentration; spermidine-stabilized cell residues were very resistant to dissociation. (3) In a solution of high salt concentration and in the absence of di- and polyvalent cations, little or no vimentin was retained in residual cell structures. Better, though not quantitative, binding was observed in solution at physiological NaCl or KC1 concentrations. (4) A synergistic effect of mono- and di- or polyvalent cations on the association of vimentin with the nuclei was detected when the extraction of cells was performed at concentrations of these ions, which individually were very inefficient in the formation of residual cell structures. (5) Triton X-ioo-resistant cell residues formed under any ionic conditions were rather resistant to shearing forces. If release of vimentin occurred, it was due to perturbance of the dissociation equilibrium between bound and free vimentin by dilution or washing. (6) Preincubation of cells in buffers of varying ionic composition and in the absence of Triton X-100, followed by titration of the amount of vimentin still incorporable into residual cell structures by extraction of cells with Triton X-100 in solution of 4 mm-mg* +, showed that the incorporation of vimentin into cell residues was strictly dependent on the intracellular ionic conditions. (7) The intracellular dissociation of vimentin-containing intermediate-sized filaments, induced by preswelling of cells in low ionic strength buffer, could be reversed, to a considerable extent, by restoration of isotonicity. If the incorporation of vimentin into Triton X-ioo-resistant residual cell structures is taken as a measure of the intactness of the vimentin-rype intermediate-sized filaments, the experimental results demonstrate that the stability of the filaments is largely dependent on the maintenance of physiological concentrations of mono-, di- and polyvalent cations.

2 50 P. Traub and W. J. Nelson INTRODUCTION Immunofluorescence microscopy (see, e.g. Franke, Schmid, Osborn & Weber, 19786; Osborn, Franke & Weber, 1980) and electron microscopy (Henderson & Weber, 1980; Heuser & Kirschner, 1980; Small & Celis, 1978) have revealed the presence of intricate networks of intermediate-sized (io-nm) filaments in a variety of vertebrate tissues and derived cell lines (for references see Franke et al and Lazarides, 1980). A method commonly used to investigate intermediate-sized filaments and other protein fibrils is to treat cells with non-ionic detergents, which results in the removal of most of the cytoplasm and cytoplasmic organelles, leaving a detergent-resistant residual cell structure (the detergent-resistant ' cytoskeleton' enriched in intermediate-sized filaments, as named by Brown, Levinson & Spudich, 1976; and Franke et al. 1978c). Little, however, is known about the means by which this structure is stabilized and, in particular, what role the ionic environment plays. Morphological studies have shown that there appears to be a direct correspondence between the detergent-resistant 'cytoskeleton' and the filamentous network in intact cells (Ben-Ze'ev, Duerr, Solomon & Penman, 1979; Osborn, Born, Koitsch & Weber, 1978; Webster, Osborn & Weber, 1978). Thus detergent-extracted cells can be used as an in vitro model system, reflecting the in vivo situation while allowing controlled experimentation. The property of the intermediate-sized filament proteins of being insoluble in high ionic strength buffers containing non-ionic detergents (Franke, Schmid, Osborn & Weber, 1978a; Starger, Brown, Goldman & Goldman, 1978) has made possible their isolation and partial biochemical characterization. Employing immunological techniques and polyacrylamide gel electrophoresis, the subunit compositions of 5 groups of intermediate-sized filaments, which are characteristic of certain differentiated cell types, have been determined (for references see Franke et al and Lazarides, 1980). In primary cell cultures or permanent cell lines, however, more than one type of intermediate-sized filament can be expressed simultaneously (Borenfreund, Schmid, Bendich & Franke, 1980; Gard, Bell & Lazarides, 1979; Franke et al a; Osborn et al. 1980). It has also been found that the vimentin-type filaments, which in vivo are characteristic of cells of mesenchymal origin, are synthesized in all vertebrate cells proliferating in vitro (Franke et al. 1979c). Though there are certain molecular characteristics common to all subunit proteins of intermediate-sized filaments (Steinert, Zimmermann, Starger & Goldman, 1978; Steinert, Idler & Goldman, 1980), they were found to be divergent in their chemical and immunological properties (Bennett, Fellini & Holtzer, 1978a; Bennett et al ; Davison, Hong & Cooke, 1977; Fellini, Bennett, Toyama & Holtzer, 1978; Lazarides, 1980). So far, the study on intermediate-sized filaments has generally been focussed on cell biological aspects. Most of these problems are amenable to investigation using the techniques of immunology, electron microscopy and polyacrylamide gel electrophoresis at a more or less qualitative level. However, the elucidation of gene expression, for instance the regulation of transcription and translation, mrna and protein turnover and post-translational modification, eventually necessitates the quantification

3 Cations and stability of vimentinfilaments.i 51 of the filament proteins. One objective of the present work was, therefore, to develop technically simple and reproducibly operating methods for the quantitative determination of the filament protein vimentin in mammalian cells and their subcellular fractions, especially in regard to their usefulness in the simultaneous handling of a large number of samples. Among the variety of intermediate-sized filament subunit proteins, vimentin was selected for this investigation because it is probably expressed in all vertebrate cells cultured in vitro (Franke et al. 1979c). For this reason, the developed methods can be easily tested for their general applicability using a multitude of cell lines propagated in vitro under a variety of conditions. Ehrlich ascites tumour cells, adapted to cell culture conditions, were chosen for the research and as the source of the filament protein vimentin since they are obtainable in large quantities, easy to handle and rich in vimentin. MATERIALS AND METHODS Cell culture Ehrlich ascites tumour (EAT) cells were grown in suspension culture using minimum essential medium supplemented with Earle's salts and 5 % foetal calf serum as described previously (Egberts, Hackett & Traub, 1976). The cells were harvested at a density of 1-5 x 10' cells/ml. They were pelleted by centrifugation at 600 g for 5 min, washed with 50 vol. of cold io~' M-Tris-acetate (ph 7'5), 150 mm-nacl (Tris/saline) and repelleted by centrifugation as specified above. After resuspension in 5 vol. of Tris/saline, the suspension was evenly distributed among tightly fitting all-glass Dounce homogenizers, so that each homogenizer finally contained cells from 250 ml of suspension culture (3'75 x io 8 cells), and centrifuged at 600 g for 5 min. Unless otherwise stated, all following operations were carried out at 2 C C. Extraction of cells Direct extraction of EAT cells with 0-5 % Triton X-100 in buffers with various ionic compositions. The cells in each Dounce homogenizer were extracted 3 times with 7-ml portions of basal buffer (io~' M-Tris-acetate (ph 7-6), 1 mm-egta, 6 mm-2-mercaptoethanol) containing 0-5 % Triton X-100 and additional salts as described in the text and in the figures, applying 10 strokes per extraction. All buffers were freshly prepared before use. Between individual extractions, the Triton X-100-resistant residual cell structures were sedimented by centrifugation at 1500 for 5 min. Corresponding supernatants were combined and recentrifuged at g for 5 min. While the small pellets were combined with the major fractions of cell residues, the postnuclear supernatants were processed for recovery of solubhized vimentin (see below). As a control in most experiments, salts were added only after homogenization of cells in 7 ml basal buffer containing 0-5 % Triton X-100. After centrifugation of the homogenates at 1500 g for 5 min, the residual cell structures were washed twice with 7-ml portions of the buffers they were eventually suspended in before centrifugation. The combined supernatants were treated as described above. Preincubation of cells in buffers of different ionic composition and extraction with 05 % Triton X-100 in the presence of 4 mm-mg-acetate. In order to follow the kinetics of the intracellular dissociation of vimentin-containing intermediate-sized filaments, Tris/saline-washed EAT cells were gently resuspended in basal buffer at a cell density of 5 x io 7 cells/ml. After different times post resuspension, the cells of 7'5-ml aliquots of cell suspension were lysed by 10 strokes in Dounce homogenizers in the presence of 0-2 ml 20% Triton X-100, 155 mm-mg-acetate. The residual cell structures were sedimented by centrifugation at 1500 g for 5 min and rcextracted twice with 7-ml portions of basal buffer containing 0-5 % Triton X-100 and 4 mm-

4 52 P. Traub and W. J. Nelson Mg-acetate, applying 10 strokes each time. Corresponding supernatants were combined and recentrifuged at g for 5 min. The pellets were added to the main fractions of residual cell structures; the supernatants were kept for quantification of solubilized vimentin. When the cells were allowed to swell in basal buffer containing additional 175 mm and 3-5 mm-mgacetate, respectively, cell lysis was achieved in the presence of 0-2 ml 20% Triton X-100, 85 mm-mg-acetate and 20% Triton X-ioo, 20 mm-mg-acetate, respectively. For 2-fold reextraction, basal buffer with 0-5 % Triton X-100 and 4 mm-mg-acetate was used. The effect of different ionic compositions" of the cell suspension buffers on the intracellular stability of the vimentin-type intermediate-sized filaments was determined in a similar way. Pellets containing 375 x io e cells were gently suspended in 7-ml portions of basal buffer containing additional salts and allowed to stand for 30 min. The cells were sedimented by centrifugation at 600 g for 5 min. Since in buffers of low salt concentration the cells were considerably swollen, 1 M-Mg-acetate was added to adjust the volume of buffer taken up by the cells to 4 mm. If Mg-acetate-containing buffers were used for cell swelling, their Mg 1+ concentrations were taken into account. Immediately thereafter, the cells were lysed in basal buffer containing 0-5 % Triton X-100 and 4 mm-mg-acetate in a final volume of 7-5 ml by 10 strokes. Residual cell structures were pelleted by centrifugation at 1500 g for 5 min and reextracted twice with lysis buffer. The combined supernatants were processed as described above. Swelling of cells in basal buffer, restoration of isotonicity and extraction of cells with 0-5 % Triton X-100 in the presence of \ mm-mg-acetate. Tris/saline-washed EAT cells from 250-ml portions of suspension culture were gently suspended in 8 ml basal buffer and immediately centrifuged at 600 g for 5 min. They were resuspended in basal buffer in a final volume of 5-9 ml and allowed to swell for 15 min (total time of swelling: approx. 25 min). Increasing volumes of a solution of 2 M-KC1 in basal buffer were added with basal buffer to make up the final volume of each cell suspension to 8 ml. Twenty min later, the cells were again pelleted by centrifugation at 600 g for s min. Mg-acetate (1 M) was added to adjust the volume of buffer taken up by the cells during swelling to 4 mm. Finally, the cells were extracted 3 times with basal buffer containing 0-5 % Triton X-100 and 4 mm-mg-acetate in a final volume of 75 ml, applying 10 strokes per extraction. The homogenates were processed as described above. The cells of 2 further sets of pellets were treated in exactly the same way, except that the swelling buffers and the 2 M-KC1 solutions contained additional 1-5 mm and 4 mm-mg-acetate, respectively. The kinetics of reconstitution were followed using essentially the same technique. Tris/ saline-washed EAT cells from suspension culture were gently suspended in basal buffer, and the cell suspension was divided into three 74-ml portions. After standing at o C for 25 min, the Mg l+ concentrations of aliquots 2 and 3 were adjusted to 1-75 mm and 3-5 mm, respectively, by the addition of 1 M-Mg-acetate. The cells were allowed to incubate for a further 5 min, when portions were withdrawn from all 3 cell suspensions to be homogenized by 10 strokes in the presence of O'6 ml 2 M-KCI in basal buffer containing o, 1*75 and 3-5 mm-mg-acetate and ml 20 % Triton X-100 in 165, 90 and 20 mm-mg-acetate, respectively (final concentrations: 145 mm-kcl, 4 mm-mg-acetate) (time o.of reconstitution). At the same time (30 min after the onset of swelling), the KC1 concentrations of the remainder of the 3 cell suspensions were adjusted to 145 mm by the addition of 5-4 ml 2 M-KCI in basal buffer containing o, 17s and 3-5 mm-mg-acetate, respectively. At various time points after KC1 addition, 8-ml portions of each cell suspension were mixed with 0-2 ml 20 % Triton X-100 in 165, 90 and 20 mm-mg-acetate, respectively, and homogenized by 10 strokes. The homogenates were centrifuged simultaneously at 1500 g for 5 min, and the pellets were re-extracted twice with basal buffer containing 0-5 % Triton X-100 and 4 mm-mg-acetate. The supernatants of each sample were combined and processed as described above. Quantification of vimentin Extraction of protein from residual cell structures. The pellets of Triton X-ioo-resistant residual cell structures were suspended in distilled water (final vol.: 3-6 ml) by 20 strokes. After the addition of 0-4 ml 2-5 M-HC1, the suspensions were shaken for 2 h and then centrifuged at g for 5 min. Solubilized protein was precipitated from the supernatants by the addition of 40 ml acetone. The precipitates were collected by centrifugation at 1000 g for 5 min

5 Cations and stability of vimentin filaments. I 53 and-dissolved in 2 ml 6 M-urea, o-6% acetic acid, 10% 2-mercaptoethanol by brief sonication. A sample of 200 fil of each protein sample was diluted with 500 fil sample buffer, and 20 fil of each diluted solution was used for polyacrylamide gradient slab gel electrophoresis in urea/ acetic acid as buffer system. Recovery of vimentin from postnuclear supernatants (1) Precipitation of vimentin with (NH t )tso t. Usually, 21 ml postnuclear supernatant were obtained per extracted cell pellet. To precipitate vimentin, 2-8 g solid (NHJjSO, (23 % saturation) was added and the solution was shaken for 1 h. The precipitates were collected by centrifugation at g for 10 min and dissolved in 2 ml 6 M-urea, 3 mm-edta (ph 7-5) by sonication. In order to remove traces of RNA, the solutions were incubated at 37 C for 10 min in the presence of 10 fig/ml pancreatic RNase A and 1 mm-phenylmethane sulphonylfluoride. The protein was precipitated by the addition of 8 ml 10 % trichloroacetic acid, pelleted by centrifugation at 2000 g for 15 min, washed with 5 ml acetone andfinallydissolved in 2 ml 6 M-urea, o-6 % acetic acid, 10 % 2-mercaptoethanol by sonication. A sample of 200 fil of each protein sample was diluted with 500 fil of sample buffer, and 20 fil of each diluted solution was subjected to polyacrylamide gradient slab gel electrophoretic analysis in urea/acetic acid as buffer system. (2) Precipitation of vimentin with spermidine. Provided that the salt concentration of the postnuclear supernatant was low, vimentin could al9o be precipitated with spermidine. The spermidine concentration of the postnuclear supernatants (21 ml) was adjusted to 2 mm by the addition of 0-45 ml 100 mm-spermidine solution. After standing for 30 min, the precipitates were collected by centrifugation at g for 10 min. They were suspended in distilled water by several strokes in tightly fitting all-glass Dounce homogenizers. The volume was adjusted to 3-6 ml with water and the protein extracted by shaking the suspensions for 2 h in the presence of 0-4 ml 2-5 M-HC1. Insoluble material was removed by centrifugation at g for 5 min, and solubilized protein was precipitated from the supernatants by the addition of 40 ml acetone. The precipitates were processed exactly as the protein moieties obtained from residual cell structures (see above). Polyacrylamide gradient slab gel electrophoresis in urea/acetic acid and densitometric evaluation of the gel profiles All protein samples were analysed in duplicate employing the gel electrophoresis method described previously (Traub & Boeckmann, 1978). After staining with Coomassie Brilliant Blue, the gels were scanned at 590 nm in a Gilford S2400 spectrophotometer. The densities of the vimentin bands (plus degradation products) were related to the sum of the densities of the 3 histone H3 subunit bands, which served as an internal standard. These ratios were expressed as percentages of the ratio of total vimentin over histone H3 in a certain amount of cells (control). The control ratios were determined as described in Results. RESULTS Throughout the experiments described below, the quantitative estimation of vimentin has been based on the employment of high-resolution polyacrylamide gradient slab gel electrophoresis in urea/acetic acid as buffer system (Traub & Boeckmann, 1978). Fig. IA, B shows, for example, the electrophoretic separation of protein from residual cell structures and postnuclear supernatants, respectively, obtained by direct extraction of EAT cells with Triton X-100 in buffers of increasing KC1 or Mg 2+ concentration. To compensate for slight inaccuracies in the isolation of the intermediate-sized filament protein from residual cell structures, and in the course of polyacrylamide gel electrophoresis and staining, the co-extracted and wellseparated histone H3 subunit proteins were chosen as an internal standard. In the

6 54 P. Traub and W. J. Nekon A P58 H3 B

7 Cations and stability of vimentinfilaments.i 55 following text, the staining density of the vimentin band is always expressed as a percentage of the ratio of the total amount of vimentin over that of histone H3 in any quantity of residual cell structures. The control ratio was determined in one of the following 2 ways: (1) EAT cells were extracted with Triton X-100 in a low ionic strength (basal) buffer of 4 mm-mg 2+ concentration or containing 1-2 mm-spermidine. Under these conditions, vimentin was incorporated quantitatively into Triton X-100- resistant cell residues (see Figs. 3, 5 A) and could be extracted, together with histones, in 0-25 M-HC1. (2) The vimentin not incorporated into cell residues was precipitated from the postnuclear supernatants with (NH 4 ) 2 SO 4 at 23 % saturation or, at low ionic strength, with 2 mm-spermidine. For each experimental series, which encompassed between 8 and 32 individual extractions, the staining densities of all vimentin bands, including those of the residual cell structures, were added up and related to the sum of the staining densities of all corresponding histone H3 subunit bands. However, in this case, vimentin had to be partially quantified without the internal standard and therefore, to obtain more reliable values, the first method was usually employed. In general though, both methods yielded results that were in fairly good agreement. With a high-resolution polyacrylamide gel electrophoresis system and an adequate quantification method at our disposal, the incorporation of vimentin into residual cell structures upon Triton X-100 treatment of EAT cells under various ionic conditions was investigated. Direct extraction ofea T cells ivith Triton X-100 in buffers of increasing salt concentration Since the preparation of ' cytoskeletons' enriched in intermediate-sized filaments is usually carried out by extraction of cells with high-salt buffer containing a non-ionic detergent (Franke et al. 1978a; O'Connor, Gard & Lazarides, 1981; Starger et al. 1978), the influence of increasing KCl concentration on the incorporation of vimentin into Triton X-100-resistant residual cell structures was tested first. Fig. 2 shows that at low ionic strength the cell residues contained very little vimentin. But with increasing KCl concentration an increasing amount of the intermediate-sized filament protein was incorporated into residual cell structures until a maximum (apprcx. 50 %) Fig. 1. Polyacrylamide gradient slab gel electrophoresis, in urea/acetic acid, of protein isolated from residual cell structures and their corresponding postnuclear supernatants. EAT cells were extracted 3 times with 0-5 % Triton X-100 in basal buffer of increasing KCl and Mg-acetate concentration, respectively. As described in detail in Materials and Methods, the protein of the residual cell structures was extracted with 0-25 M-HC1, and vimentin contained in the postnuclear supernatants was precipitated with (NH 4 ),SO 4 at 23 % saturation. The protein fractions were finally dissolved in equal volumes of 6 M-urea, 0'6 % acetic acid, 10 % 2-mercaptoethanol and analysed by polyacrylamide gradient slab gel electrophoresis. Slots 1-9: protein isolated from residual cell structures (A) and postnuclear supernatants (B), which were obtained by extraction of EAT cells with basal buffer supplemented with 0-5 % Triton X-100 and increasing amountsof KCl (o, io, 30, 50, 70, 90, no, 130, 150 mm) and, in the case of slots 10 to 16, with increasing amounts of Mg-acetate (05, 1, 1-5, 2 > 2 '5> 3. 3'S mm). The gels were stained with Coomassie Brilliant Blue.

8 P. Traub and W. J. Nelson mm-kci Fig. 2. The fraction of vimentin present in residual cell structures after extraction of EAT cells with 0-5 % Triton X-100 in buffer of increasing KCl concentration. The experimental design of the different extractions is described in Materials and Methods. (O, D) Direct extraction of cells with basal buffer containing 0-5 % Triton X-100 and increasing amounts of KCl; the curve is a composite of the results of 2 different experiments. ( ) Addition of KCl after homogenization of cells in basal buffer containing 0-5 % Triton X-100. After the first homogenization and low-speed centrifugation, the pellets were re-extracted twice with the same buffers in which the residual cell structures were eventually suspended before the first centrifugation. Protein was extracted from residual cell structures and postnuclear supernatants as described in Materials and Methods and equal amounts were subjected to polyacrylamide gradient slab gel electrophoretic analysis in urea/acetic acid. The Coomassie Brilliant Bluestained gels of the type presented in Fig. 1 were scanned at 590 nm. The densities of the vimentin (P58) bands (from residual cell structures) were related to the sum of the densities of the corresponding histone H3 subunit bands and expressed as percentages of the control P58/H3 ratio. The control value was obtained by relating the sum of the densities of all vimentin bands (from residual cell structures and postnuclear supernatants) to the sum of densities of all histone H3 bands. was reached under nearly physiological conditions. At KCl concentrations higher than 150 mm, more and more vimentin was solubilized; at 300 mm-kci, almost all the vimentin originally contained in EAT cells was recovered from the postnuclear supernatant. In order to reduce the possibility that the incorporation of vimentin into residual cell structures was only due to cosedimentation of intermediate-sized filament protein aggregates with actually naked nuclei formed after cell lysis, cells were disrupted in

9 Cations and stability of vimentinfilaments.i 57 mm-mg-acetate Fig. 3. The fraction of vimentin present in residual cell structures after extraction of EAT cells with 0-5 % Triton X-100 in buffer of increasing Mg-acetate concentration. Experimental details are given in the Materials and Methods section. (O) Direct extraction of cells with basal buffer containing 05 % Triton X-100 and increasing amounts of Mg-acetate; (D) addition of increasing amounts of Mg-acetate after homogenization of cells in basal buffer containing 0-5 % Triton X-100. Washing of cell residues, extraction of protein, polyacrylamide gel electrophoretic analysis and calculation of the P58/H3 ratios were done as described in the legend to Fig. 2. The figure also shows the amount of vimentin that remained associated with the nuclear fraction after Triton X-100-resistant residual cell structures, obtained in the presence of 4 mm-mg-acetate, were extracted 3 times with basal buffer of decreasing Mg 1+ concentration in the absence of Triton X-100 ( ). basal buffer containing 0-5 % Triton X-100, and only then KCl was added to raise the ionic strength. After low-speed centrifugation, only negligible amounts of vimentin could be isolated from the nuclear pellets (Fig. 2). In both experiments, the fraction of apparently insoluble vimentin was relatively high when the extraction with Triton X-ioowas carried out in basal buffer in the absence of any additional salts. This was due, at least in part, to insufficient washing of the residual cell structures by low-speed centrifugation because of considerable nuclear swelling. The remaining vimentin could be removed by more frequent washing of the cell residues with basal buffer. 3 CEL53

10 58 P. Traub and W. J. Nelson Beside buffers of high salt concentration, some authors also employed Mg 2+ - containing buffers for the preparation of ' cytoskeletons' enriched in intermediatesized filaments (Franke et al. 1978a, c\ Ben-Ze'ev et al. 1979; Fulton, Wan & Penman, 1980; Lenk, Ransom, Kaufmann & Penman, 1977; Lenk & Penman, 1979). This prompted a more systematic investigation of the effect of di- and polyvalent cations on the retention of vimentin in Triton X-100-resistant cell residues. As depicted in Fig. 3, Mg 2+ concentrations up to 2 mm were inefficient in retaining vimentin in Triton X-100-resistant residual cell structures. Maximal incorporation was achieved mm-kci Fig. 4. Vimentin content of residual cell structures prepared by extraction of EAT cells with 0-5 % Triton X-100 in basal buffer containing KC1 and Mg-acetate in various concentrations. The experimental details of the extractions are described in Materials and Methods. The stimulatory effect of 1-5 mm (O) and 4 mm ( ) Mgacetate, respectively, on the incorporation of vimentin into Triton X-ioo-resistant residual cell structures in basal buffer of increasing KC1 concentration was investigated. For comparison, the binding of vimentin in the presence of increasing amounts of KC1 alone (A) (taken from Fig. 2) is shown. In all 3 cases, the cells were extracted directly in the presence of 0-5 % Triton X-100 and re-extracted twice with the corresponding buffers of different ionic composition. In addition, the vimentin contents are presented of those cell residues that were obtained by homogenization of cells with 0-5 % Triton X-100 in basal buffer and subsequent adjustment of the homogenates to 4 mm-mg-acetate and increasing KC1 concentrations (#). In the latter case, the cell residues were processed as described in the legend to Fig. 2. The determination of the relative P58/H3 ratios is also described there.

11 Cations and stability of vimentinfilaments.i 59 in basal buffer containing 4 mm-mg-acetate and 0-5 % Triton X-100. The fraction of vimentin solubilized in this buffer was below 5 %, as determined by (NH^SC^ fractionation of the postnuclear supernatant or by precipitation with spermidine. The addition of increasing amounts of Mg-acetate after cell lysis in Triton X-100-containing basal buffer proved that vimentin, once solubilized, did not aggregate in response to addition of divalent cations in these concentrations and, therefore, did not cosediment with naked nuclei during low-speed centrifugation (Fig. 3). To learn more about the stability of the residual cell structures prepared in basal buffer containing 4 mm-mg-acetate and 0-5 % Triton X-100, they were extracted 3 times with basal buffer of decreasing Mg 2+ concentration, in the absence of Triton X-100. As illustrated in Fig. 3, the cell residues were resistant to dissociation down to a Mg 2+ concentration of 1-5 mm. However, below 1-5 mm-mg-acetate they fell apart, releasing vimentin into the postnuclear supernatant. The experimental results show that the Mg 2+ -dependent incorporation of vimentin into residual cell structures is reversible and that the binding and release of the filament protein can be described by a hysteresis loop. It should be noted that with increasing Mg*+ concentration of the Triton X-100-free extraction buffer an increasing number of extractions was needed to chase vimentin into the low-speed postnuclear supernatant. In the absence of Mg-acetate, one extraction resulted in approximately 70 % of the bound vimentin being released into the postnuclear supernatant (data not shown). The experimental results accumulated so far demonstrate that at low KC1 or Mgacetate concentrations almost no vimentin was incorporated into Triton X-100- resistant residual cell structures. As illustrated in Fig. 4, it could be shown, however, that low K+ and Mg 2+ concentrations had a synergistic effect on the retention of vimentin in detergent-resistant cell residues. Whereas in the presence of either 1-5 mm-mg-acetate or, for example, 50 mm-kcl nearly all vimentin was released into the postnuclear supernatant, the combination of both buffers resulted in the retention of 75 % of the total amount of vimentin in the Triton X-100-resistant residual cell structures. In this context, the binding potentials of various buffers of high Mg 2+ and K + concentration were of experimental interest since these buffers were employed in the reconstitution experiments described below or for the isolation and purification of vimentin (Nelson & Traub, unpublished data). As depicted in Fig. 4, increasing KC1 concentrations in basal buffer containing 4 mm-mg-acetate and 0-5 % Triton X-100 exerted only a slightly destabilizing effect on the residual cell structures. However, concentrations above 150 mm-kcl effected a relatively rapid decrease in the amount of vimentin retained in cell residues. It is a well-known experimental fact that the integrity of many cellular structures and the activities of many enzymes are Mg 2+ -dependent and that, at least in part, Mg 2+ can be substituted for polyamines. It was investigated, therefore, whether polyamines can replace Mg 2+ in the formation of vimentin-containing residual cell structures. Thus, EAT cells were extracted with 0-5 % Triton X-100 under standard conditions using basal buffer supplemented with increasing amounts of putrescine, spermidine and spermine, respectively. The results are summarized in Fig. 5 A. AS in 3-2

12 6o P. Traub and W. J. Nelson mm-spermidine, mm-spermine Fig. 5 A. For legend see opposite. the case of cell extraction in the presence of Mg 2+, the retention of vimentin in cell residues could be described by sigmoidal curves. As expected, less spermidine and spermine were needed on a molar basis than Mg-acetate to achieve the same binding effect. However, putrescine had a lower binding potential than Mg-acetate, with the result that at a concentration of 3-5 mm the binding efficiencies were 50 and 100 %, respectively. A comparison of the incorporation potentials of Mg 2+, spermidine and spermine revealed half-maximum binding of vimentin at 2-65, o-68 and 0-39 mm cation concentration, respectively. In addition, a synergistic effect was observed when the incorporation of vimentin into cell residues was followed, depending on the spermidine concentration, in basal buffer containing additional 1-5 mm-mg-acetate. Already at 0-08 mm-spermidine, half-maximum incorporation of vimentin was recorded. Since Mg-acetate was 4 times less efficient in the production of vimentincontaining cell residues, half-maximum binding should have been attained at only 0-3 mm-spermidine concentration if there had been a mere additive effect of Mgacetate and spermidine (Fig. 5 A). To exclude the possibility that aggregates of solubilized vimentin were cosediment-

13 Cations and stability of vimentin filaments. I 61 mm-putrescine mm-spermidine, mm-spermine FIG. 5 B. Fig. 5- A. Vimentin content of residual cell structures after extraction of EAT cells witho-5 % Triton X-ioo in the presence of increasing amounts of putrescine, spermidine and spermine, respectively. In addition to the direct extraction of cells with "5 % Triton X-ioo in basal buffer of increasing putrescine (#), spermidine (O) and spermine (D) concentration, in a further extraction series the effect of 1-5 mm-mgacetate in combination with increasing amounts of spermidine (A) on the incorporation of vimentin into residual cell structures was tested. The relative P58/H3 ratios were determined as described in the legend to Fig. 2. B. Precipitation of vimentin with polyamines after homogenization of EAT cells in basal buffer containing 0-5 % Triton X-ioo. EAT cells were homogenized with 0-5 % Triton X-ioo in basal buffer as described in Materials and Methods. Immediately after the addition of increasing amounts of putrescine ( ), spermidine (O) and spermine ( ) to the homogenates, the residual cell structures were pelleted by low-speed centrifugation, re-extracted twice with the corresponding final lysis buffers and extracted with 025 M-HC1 for polyacrylamide gradient slab gel electrophoresis of the residual proteins. The P58/H3 ratios were determined as described in the legend to Fig. 2. ing with residual cell structures, actually less rich in vimentin, EAT cells were lysed with Triton X-ioo before the addition of polyamines and then subjected to low-speed centrifugation. As illustrated in Fig. 5 B, only insignificant amounts of vimentin were pelleted at polyamine concentrations that, upon direct extraction of EAT cells,

14 62 P. Traub and W. J. Nelson effected half-maximum binding of the filament protein into cell residues. However, at higher spermidine and spermine concentrations, nearly all previously solubilized vimentin was precipitated. When residual cell structures formed in the presence of i*2 mm-spermidine were extracted further with basal buffer in the absence of di- or polyvalent cations and No. of extractions Fig. 6. Vimentin content of residual cell structures after severalfold extraction of EAT cells with 0-5 % Triton X-100 in buffers with various ionic compositions. Cells were extracted at different times with 0-5 % Triton X-100 under standard conditions (see Materials and Methods) in basal buffer supplemented with 1 mm-mg-acetate ( ), 25 mm-mg-acetate ( ), 50 mm-kcl (O), 150 mm-kcl (#). The control P58/H3 ratio was determined by extracting one cell aliquot 3 times with basal buffer containing 0-5 % Triton X-100 and 4 mm-mg-acetate and gel electrophoretic analysis of the residual protein. Triton X-ioo, they proved to be extremely resistant to dissociation, quite in contrast to the behaviour of cell residues prepared in the presence of 4 mm-mg-acetate. Only the fifth extraction brought about approximately 30 % release of vimentin into the postnuclear supernatant (data not shown). Also in the case of spermidine, a synergistic effect of low concentrations of polyamine and KCl on the binding of vimentin into Triton X-i co-resistant cell residues could be demonstrated. Exposure of EAT cells to increasing KCl concentrations in the presence of Triton X-100 and 0-4 mm and 1-2 mm-spermidine, respectively,

15 Cations and stability of vimentinfilaments.i 63 yielded exactly the same results as presented in Fig. 4 for different Mg-acetate/KCl combinations. The identity of the results obtained for Mg-acetate and spermidine, at 2 different concentrations each time, plainly demonstrates the same mode of action of both cations in the stabilization of residual cell structures containing vimentin. So far, all direct extractions of EAT cells with Triton X-100 were carried out in tightly fitting all-glass Dounce homogenizers, applying 10 strokes per extraction step. Since, during homogenization, cells and derived cell structures were exposed to rather strong shearing forces, the question arose whether the different extents of vimentin solubilization observed in various ionic environments were not the consequences of different sensitivities of the delicate cytoskeletal structures to mechanical fragmentation. To answer this question, EAT cells were extracted for different times under standard conditions, applying 10 strokes per extraction step and using 4 buffers of different ionic composition (see legend to Fig. 6). Fig. 6 shows that repeated extraction caused a gradual release of vimentin into the postnuclear supernatants; the extent of solubilization depended on the buffer composition. That this gradual solubilization was due to perturbation of the equilibrium between 'cytoskeletal' and 'solubilized' vimentin and not a response to the exposure of cells or residual cell structures to increasing shearing forces is demonstrated by the experiment described in Fig. 7. Using the same 4 buffers as above, EAT cells were extracted only once, but with an increasing number of strokes. It is evident that already a small number of strokes were sufficient to solubilize vimentin, the extent of solubilization again being dependent on the buffer composition. Excessive homogenization did not significantly increase the amount of vimentin released from the residual cell structures. The perturbation of the equilibrium between aggregated and disaggregated vimentin could also be demonstrated by homogenizing the cells in different volumes of the same extraction buffer under otherwise constant (standard) conditions. Dilution of the homogenate led to a linear decrease of the amount of vimentin remaining in the residual cell structures when the relative P58/H3 ratios were plotted as a function of the logarithm of the extraction volume (data not shown). Prestvelling of EA T cells in buffers with various ionic compositions, removal of salts, and extraction of cells with Triton -Y-100 in a hypotonic buffer of high Mg i+ concentration In order to prevent continuous loss of apparently solubilized vimentin and perhaps other aggregate-stabilizing macromolecules into the extraction medium, as it was observed in the course of direct cell extraction with Triton X-100 (see Fig. 6), it was attempted to induce changes in the ionic conditions intracellularly, with preservation of the plasma membrane. EAT cells were allowed to swell for different time periods in basal buffer. They were then extracted in the presence of 0-5 % Triton X-100 and 4 mm-mg-acetate. Fig. 8 illustrates a steep decline of the amount of vimentin still incorporate into Triton X-100-resistant cell residues with the time of swelling. At 25 min after the onset of swelling, an abrupt change of the solubilization of vimentin was observed. At this time point, approximately 25 % of the total vimentin remained insoluble. Later on, further but slower disaggregation took place, with about 10% of the total vimentin being solubilized within 1-5 h. It should be noted here, that in

16 6 4 P. Traub and W. J. Nelson No. of strokes Fig. 7. The fraction of vimentin remaining in residual cell structures after one extraction of EAT cells with 0-5 % Triton X-100 in buffers with various ionic compositions and application of different shearing forces. Cells were extracted once with '5 % Triton X-100 under various ionic conditions by the application of 10, 20, 40 and 80 strokes, respectively, in rightly fitting all-glass Dounce homogenizers. The basal extraction buffer was supplemented with 1 mm-mg-acetate (D), 2'5 mm-mgacetate ( ), 50 mm-kcl (O), 150 mm-kcl ( ). The protein of the residual cell structures was extracted with 0-25 M-HC1 and electrophoresed on polyacrylamide gradient slab gels. The control P58/H3 ratio was determined by extraction of one cell aliquot with basal buffer containing 0-5 % Triton X-100 and 4 mm-mg-acetate and gel electrophoretic analysis of the residual protein. phase-contrast microscopy the cells looked swollen but intact (see accompanying paper), and that only an infinitesimal number of cells took up trypan blue in the course of swelling. When the cells were allowed to swell in basal buffer supplemented with 1-5 mm- Mg-acetate, a slight retardation of the first phase of the disaggregation process was noticed. But the overall shape of the curve was the same, again with an abrupt change at 25 min after the onset of swelling (Fig. 8). However, supplementation of the basal swelling buffer with 4 mm-mg-acetate brought about a significant change in the course of the intracellular disaggregation

17 66 P. Traub and W. J. Nelson The above results indicate that Triton X-ioo-containing basal buffer of high Mg 2+ concentration can be used for the titration of insoluble, filamentous vimentin. Moreover, they show that, upon changes in the extracellular ionic milieu, the ionic equilibrium between the extracellular and intracellular space is apparently stabilized 25 min after the changes have been set O /a o _^C^ o rol) c p u 'o :* P58/H3 (' 70 I L / ) / A /» / 20 - / 10' / mm-kci Fig. 9. The fraction of vimentin found in residual cell structures after preincubation of EAT cells in buffers with various Mg-acetate and KC1 concentrations and extraction with 0-5 % Triton X-100 in basal buffer supplemented with 4 mm-mg-acetate. Cells were preincubated for 30 min in basal buffer (#) supplemented with 1-5 mm ( ) and 4 mm (O) Mg-acetate and increasing amounts of KC1. After low-speed centrifugation, the cells were extracted 3 times with basal buffer containing 0-5 % Triton X-100 and 4 mm-mg-acetate. To determine the control P58/H3 ratio, one original cell aliquot was extracted in the same way. The protein of the cell residues was extracted with 0^25 M-HC1 and analysed by polyacrylamide gradient slab gel electrophoresis. In addition, the control P58/H3 ratio was also calculated as described in the legend to Fig. 2. Both control values were in good agreement. 1 With this information available, the influence of changes in the monovalent cation concentration of the suspension buffer on the intracellular aggregation state of vimentin was studied. EAT cells were suspended in basal buffer of increasing KC1 concentration and incubated for 30 min. The cells were then pelleted by low-speed

18 Cations and stability of vimentinfilaments.i 67 centrifugation and extracted with basal buffer containing 4 mm-mg-acetate and 0-5 % Triton X-100 under standard conditions. There was only little incorporation of vimentin into residual cell structures in the absence of KCl (see also Fig. 8). However, with increasing KCl concentration of the suspension medium, more and more vimentin was bound into Triton X-100-resistant cell residues until a plateau value of approximately 85 % was reached between 150 and 200 mm-kcl. When the swelling buffers contained 1-5 mm and 4 mm-mg-acetate, respectively, similar cell responses were 60 mm-spermine mm-spermidine mm-mg-acetate Fig. 10. Vimentin content of residual cell structures after swelling of EAT cells in basal buffer of increasing Mg-acetate (O), spermidine (#) and spermine ( ) concentration and extraction with 05 % Triton X-100 in basal buffer supplemented with 4 mm-mg-acetate. After swelling for 30 min, the cells were pelleted by low-speed centrifugation and extracted 3 times with 0-5 % Triton X-100 in basal buffer containing 4 mm-mg-acetate. To determine the control P58/H3 ratio, one aliquot of original cells was extracted in the same way. The protein of the cell residues was extracted with 0-25 M-HC1 and analysed by polyacrylamide gradient slab gel electrophoresis. observed. Since, however, in the presence of Mg 2+ the disaggregation of vimentincontaining intermediate-sized filaments, on average, was retarded (see Fig. 8), higher relative P58/H3 ratios were obtained with a plateau value of approximately 95 %. The direct extraction of EAT cells with Triton X-100 in basal buffer of increasing Mg 2+ or polyamine concentration has shown that di- and polyvalent cations have a stabilizing effect on residual cell structures, even at low ionic strength (see Figs. 3, 5 A). Thus, it was of interest to find out whether these cations are also able to preserve intermediate-sized filaments from disassembly when the plasma membrane is still intact. EAT cells were again suspended in basal buffer of increasing Mg 2+, spermidine or spermine concentration and, after 30 min, pelleted for extraction with basal buffer supplemented with 4 mm-mg-acetate and 0-5 % Triton X-100. As one can see from

19 68 P. Traub and W. J. Nelson Fig. 10, all 3 cations stabilized the vimentin-type intermediate-sized filaments, although only to a certain extent since the plateau was reached already at a relative P58/H3 value of approximately 60%. 100 mm-kci Fig. 11. Reconstitution of vimentin-containing cytoplasmic structures incorporable into Triton X-100-resistant cell residues by the addition of increasing amounts of KC1 to EAT cells preincubated in basal buffer of different Mg'+ concentration. Cells were allowed to swell for 30 min in basal buffer ( ) supplemented with 1-5 mm ( ) and 4 mm (O) Mg-acetate. Increasing amounts of KC1 were added and the incubation was continued for further 30 min. After low-speed centrifugation, the cells and one untreated control pellet were extracted 3 times with basal buffer supplemented with O'5 % Triton X<-ioo and 4 mm-mg-acetate. The protein of the cell residues was solubilized with 0-25 M-HC1 and electrophoresed on polyacrylamide gradient slab gels. 400 Szvelling of EAT cells in hypotonic buffer, restoration of isotonicity, and extraction of cells with Triton X-100 in hypotonic buffer of high Mg i+ concentration Preliminary experiments have indicated that the incorporability of vimentin into residual cell structures, lost during swelling of EAT cells in basal buffer, could be partially restored if the swollen cells were incubated for some time in solution of physiological salt concentration before extraction with Triton X-100. With a better analytical method available, a clearer demonstration of the intracellular disassemblyreassembly of vimentin-containing intermediate-sized filaments was attempted, which

20 Cations and stability of vimentinfilaments.i 69 is expected when cells pass through a low ionic strength-high ionic strength buffer incubation cycle. To induce disaggregation of vimentin-containing intermediate-sized filaments, EAT cells were incubated in basal buffer for 30 min. The ionic strength of the cell suspension was then gradually raised by the addition of KC1. After standing for a further Time (min) Fig. 12. Kinetics of the reconstitution of vimentin-containing cytoplasmic structures incorporable into Triton X-100-resistant cell residues by restoration of isotonicity in hypotonic cell suspensions. EAT cells were allowed to swell for 30 min in basal buffer (O) which, 5 min prior to reconstitution, was supplemented with 17s mm (#) and 3-5 mm ( ) Mg-acetate. After the addition of KC1 to afinalconcentration of 150 mm, aliquots of cell suspension were withdrawn at different times and homogenized in the presence of 0-5 % Triton X-100 and 4 mm-mg-acetate. The cells were re-extracted twice with basal buffer supplemented with 0-5 % Triton X-100 and 4 mm-mg-acetate. To determine the control P58/H3 ratio, one aliquot of original cells was extracted 3 times with the same buffer. The protein of the residual cell structures was extracted with 0-25 M-HC1 and subjected to polyacrylamide gradient slab gel electrophoretic analysis. 20 min, the cells were pelleted and extracted with basal buffer containing 4 mm-mgacetate and 0-5 % Triton X-100. Fig. 11 clearly shows that raising the ionic strength caused reassembly of solubilized vimentin into structures that eventually could be incorporated into cell residues. Maximal reconstitution, with approximately 55 % of the total amount of vimentin detectable in cell residues, was achieved between 150 and 200 mm-kcl. Further increase of the KC1 concentration of the cell suspension

21 70 P. Traub and W. J. Nelson gave rise to a slight decline in the efficiency of reassembly. Identical experiments were conducted with 1-5 mm and 4 mm-mg-acetate in the basal swelling buffer. As is evident from Fig. 11, the same results were obtained, with the exception that the reconstitution curves were shifted in parallel to higher relative P58/H3 ratios. Since it was conceivable that the deviation of the optimal yield of reconstitution from 100 % was due to a too early interruption of the reassembly process (after 25 min incubation of swollen cells in buffers of increasing ionic strength), the kinetics of reconstitution were followed. The experimental design was as described for the last experiment, except that the pelleting of the cells by low-speed centrifugation was omitted following adjustment of the KC1 concentration of the hypotonic cell suspension to 150 mm. Instead, the extraction of the cells was carried out in solution of physiological salt concentration by simultaneous addition of Triton X-100 and Mgacetate to final concentrations of 0-5 % and 4 mm, respectively. Fig. 12 shows that the reconstitution process was terminated after 20 to 25 min following the addition of KC1 to the suspension of swollen cells. In 2 parallel experiments, 20 min after the onset of swelling in basal buffer, the Mg 2+ concentration of the cell suspension wa9 adjusted to 1*75 and 3-5 mm, respectively. The Mg ions were allowed to react for 5 min, and only then the KC1 concentration was raised to 150 mm. However, as illustrated in Fig. 12, the presence of Mg 2+ in the reconstitution buffer had only a slightly stimulatory effect on the eventual incorporation of vimentin into Triton X-100-resistant residual cell structures. DISCUSSION From the results described (see Figs. 6, 7), it is apparent that the release of vimentin from supramolecular structures is governed by the dissociation equilibrium between bound and free vimentin. Therefore, the technique of severalfold direct extraction of EAT cells with Triton X-100 proved to be useful in determining the stability of residual cell structures and filaments, respectively, in various ionic environments. Thus, for example, in the absence of di- or polyvalent cations, only very little filament protein was incorporated into Triton X-100-resistant cell residues in solutions of low and high ionic strength. Considerable stabilization of vimentin-containing residual cell structures was achieved only in solution of physiological salt concentration. The wide solubilization of vimentin in Triton X-100-containing buffers of high salt concentration is not in agreement with reports by Franke et al. (1978a), O'Connor et al. (1981) and Starger et al. (1978) on the preparation of detergent-resistant 'cytoskeletons' enriched in intermediate-sized filaments by combined extraction of cells in low- and high-salt buffers. Although the nearly quantitative incorporation of vimentin into Triton X-100-resistant residual cell structures by one extraction in solution of physiological salt concentration in the absence of divalent cations could be confirmed (Figs. 6, 7), these structures proved to be very unstable upon further washing, particularly with low ionic strength buffer free of divalent cations. The situation was not improved when the residual cell structures were exposed to 1-5 M-KC1 before washing. The reason for this discrepancy is not known. It is possible that the low

22 Cations and stability of vimentinfilaments.i 71 stability of the vimentin-containing cell residues in solution of high salt concentration is observed only in the case of EAT cells grown in suspension culture. In comparison with monovalent cations, di- and polyvalent cations were much more efficient in the retention of vimentin in Triton X-100-resistant residual cell structures (Figs. 3, 5 A). In solution of low ionic strength and in the presence of 0-5 % Triton X-ioo, 3-5 mm-mg-acetate, 0-9 mm-spermidine or 0-5 mm-spermine were sufficient to retain virtually all vimentin originally present in EAT cells in the detergentresistant residual cell structures. The binding took place within narrow, physiological concentration ranges. In this context, a comparison of the present experimental data with results of the in vitro assembly of intermediate-sized filaments is pertinent. Zackroff & Goldman, studying the properties of decamin from baby hamster kidney (BHK-21) cells (1979) and the subunit proteins of neurofilaments from squid brain (1980), have shown that the equilibrium between assembled and disassembled filament protein is sensitive to changes in the ionic environment. While in the latter case intermediate-sized filaments disaggregated when squid brain tissue was exposed to very high ionic strength, filaments from BHK-21 cells seemed to be stable in solution of high salt concentration and to fall apart at low ionic strength. In both cases, the disassembly resulted in the formation of soluble protofilaments that, upon restoration of physiological conditions, associated end-to-end to form intermediate-sized filaments. Furthermore, Rueger, Huston, Dahl & Bignami (1979) have shown that glial fibrillary acidic protein from bovine brain was soluble in low ionic strength buffer and showed strong tendency to form curvilinear 10-nm filaments when the salt concentration was raised to physiological values, although an apparent specificity for imidazole-hcl buffers was noted. It should be mentioned here also that, upon Triton X-100 treatment of EAT cells under certain ionic conditions, vimentin was probably released into the extraction medium as protofilaments since it sedimented very slowly at high centrifugal forces but was completely excluded from Sephacryl S-300 (data not shown). Thus, in conjunction with the experimental data presented here (Fig. 2), these results point at an optimal stability of intermediate-sized filaments in solution of physiological monovalent cation concentration, in the absence of di- and polyvalent cations. This common property of filaments originating from different cell types can be explained on the basis of the observation that their subunits are structurally and chemically related and appear to be highly a-helical proteins of the K-m-e-f-class (Steinert et al. 1978; Steinert et al. 1980). Concerning the effect of divalent cations on the in vitro formation of intermediatesized filaments, the results described by Fukuyama, Murozuka, Caldwell & Epstein (1978) and Sakamoto, Tzeng, Fukuyama & Epstein (1980) are in remarkably good agreement with those of the present investigation. Employing electron microscopy, these authors have shown that upon dialysis of a solution of denatured keratin (obtained from cornified cells of newborn rats) against a low ionic strength buffer, a mesh-like network of curvilinear 7 to 8 nm wide filaments formed. When the dialysis buffer contained MgCl 2 or CaCl a, the filaments assembled side by side and formed thick cables which, in the polarizing microscope, exhibited strong birefringence

23 72 P. Traub and W. J. Nelson (Fukuyama et al. 1978). The quantitative investigation of the assembly process revealed a characteristic relationship between the light-scattering intensity of the reconstitution mixtures and the divalent cation concentration. Sigmoidal curves were recorded of the same type as shown in Fig. 3. Interestingly, the light-scattering intensity increased in the same concentration range in which a steep augmentation of the incorporation of vimentin into Triton X-100-resistant residual cell structures was observed. In addition, the divalent cation concentration for half-maximum intensity was shifted to higher values with increasing ionic strength of the reaction mixtures (Sakamoto et al. 1980), suggesting a competition between mono- and divalent cations for critical binding sites on the macromolecules to be crosslinked. Whereas in the case of direct extraction of cells with Triton X-100 in basal buffer vimentin was almost quantitatively solubilized, in intact cells incubated in the same but Triton X-100-free buffer approximately a quarter of the total amount of vimentin was relatively resistant to further disaggregation. It is not known whether, after a 25 min incubation period, a filamentous backbone structure remained or whether the intracellular concentration of solubilized vimentin and other macromolecules participating in the construction of cytoplasmic supramolecular structures was so high that, upon lysis of cells with Triton X-100 in the presence of 4 mm-mg-acetate, a spontaneous, Mg 2+ -dependent reaggregation of 25 % of the filament protein took place. The first possibility, on the other hand, might be due to the fact that the dissociation of the vimentin-type intermediate-sized filaments is governed by an equilibrium and that the disassembly of the cytoplasmic supramolecular structures is limited because of high reactant concentration in the small volume of the swollen but intact cells. In some experiments (Figs. 9, 10, 11), the fraction of filament protein remaining insoluble after swelling was only between 5 and 15%. However, in these cases, a low-speed centrifugation step was included after swelling and the cells were resuspended in fresh basal buffer before extraction with Triton X-100 in solution of high Mg 2 " 1 " concentration. This additional manipulation might well have caused further dissociation of vimentin-containing aggregates. When the swelling buffer contained additional 4 mm-mg-acetate, % of the filament protein was still insoluble after a 25 min swelling period. Since in phasecontrast microscopy about two-thirds of the swollen cells possessed shrunken nuclei with clearly visible chromatin structure (see the accompanying paper), it must be concluded that apparently % of the cells excluded Mg 24 - from their cytoplasm with the result that their vimentin-containing intermediate-sized filaments fell apart. The reason for the existence of such a Mg 2+ barrier in only a fraction of EAT cells is unknown. A similar situation was observed when the cells were allowed to swell in basal buffer supplemented with optimal amounts of spermidine or spermine. Only % f tne total amount of vimentin was eventually recovered from Triton X-100-resistant residual cell structures indicating that 35-40% of the swollen cells were unable to take up the polyamines from the environment. Furthermore, higher concentrations of spermidine and spermine had to be used, in comparison with the results obtained in the case of direct cell extraction, in order to achieve incorporation of the same quantity

24 Cations and stability of vimentinfilaments.i 73 of vimentin into cell residues. Only in the case of Mg 2+, were the concentrations for half-maximum binding nearly identical (2-5 mil) under both extraction conditions. From these results it is concluded that the plasma membrane barrier for Mg 2+ was rather low in that fraction of swollen cells that were able to take up Mg 2+ from the extracellular space, and that the barrier increased in height with the positive net charge of the cations passing the plasma membrane. Concerning the response of EAT cells to cyclic alterations of the extracellular ionic strength, substantial quantities of vimentin could be incorporated into Triton X-100- resistant residual cell structures after swelling of EAT cells in basal, low ionic strength buffer, followed by restoration of isotonicity in the cell suspension and, finally, extraction of cells with Triton X-100 in the presence of 4 mm-mg-acetate. The yield of reconstitution of intermediate-sized filaments was somewhat higher when the reassembly was conducted in the presence of Mg 2+ (Fig. 12). Furthermore, when KC1 was substituted for NaCl, an improvement of the reassembly by approximately 10% was observed. Finally, the reconstitution was performed in minimum essential medium supplemented with Earle's salts, in the absence and the presence of foetal calf serum. Also under these optimal, physiological conditions, the relative yield of reconstitution could not be raised above 70%. But it should be pointed out that in some experiments (see, for instance, Fig. 12) the salt was not removed by low-speed centrifugation before extraction of cells with Triton X-100 in solution of high Mg 2+ concentration. However, under these conditions, as shown in Fig. 4, the amount of vimentin incorporated into residual cell structures was approximately 15 % lower than that determined after cell extraction in buffer of low salt concentration but otherwise identical ionic composition. The incomplete reassembly of vimentin-containing intermediate-sized filaments, of course, was not unexpected. Once solubilized under hypotonic conditions, vimentin and other macromolecules eventually making up Triton X-100-resistant cell residues probably interacted unspecifically and partially irreversibly with cytoplasmic constituents, which upon cell extraction with Triton X-100 in solution of high Mg 2+ concentration are normally not incorporated into cell residues. Alternatively, the growing ends of the vimentin filaments might have been blocked, again by nonspecific interaction with other cytoplasmic constituents. In both cases, restoration of physiological ionic strength before cell extraction with Triton X-100 might not have completely abolished non-specific interactions with other cytoplasmic constituents. Such an interference by non-filament proteins with the in vitro assembly of bovine epidermal keratin filaments has actually been demonstrated by Steinert & Idler (1976). The effect of higher temperature on the reversibility of the disassembly process has not yet been investigated. In summary, the experimental data show that the stability of vimentin-containing intermediate-sized filaments in Triton X-100-resistant cell residues as well as in cells with widely intact membrane systems depends, to a great extent, on the ionic composition of the suspension buffer and cytoplasm, respectively. Mono-, di- and polyvalent cations are active in this respect, and they are most efficient at physiological concentrations. If the extent of incorporation of vimentin into Triton X-100-resistant

25 74 P. Traub and W. J. Nelson residual cell structures is taken as a measure of the intactness of intermediate-sized filaments, they are stable in solution of physiological salt concentration and labile at low and high ionic strength, in the absence of di- and polyvalent cations. Mg 2+, spermidine, spermine and possibly other di- and polyvalent cations of low and high molecular weight are themselves able to stabilize vimentin-containing intermediatesized filaments, but within the cell they are very likely to act in combination with monovalent cations; although, such a synergistic effect has been observed only in Triton X-ioo-resistant residual cell structures depleted of a large fraction of cytoplasmic constituents and virtually all membrane systems (Fig. 4). On the basis of the experimental findings presented here, several methods for the isolation, partial purification and quantification of vimentin from mammalian cells have been worked out. In addition, the techniques employed are very suitable for the screening of eukaryotic cells for the presence of vimentin. In this laboratory, in the meantime approximately 70 different cell lines cultured in vitro have been found to contain the intermediate-sized filament protein. The results of these investigations will be published elsewhere. We are very grateful to Mrs Ulrike Traub and Mrs Margot Bialdiga for their invaluable help in culturing the cells and to Miss Brigitte Geisel, Miss Gabi Haun and Miss Annemarie Scherbarth for excellent technical assistance in the performance of polyacrylamide gradient slab gel electrophoresis. REFERENCES BENNETT, G. S., FELLINI, S. A., CROOP, J. M., OTTO, J. J., BRYAN, J. & HOLTZER, H. (19786). Differences among 100 A filament subunits from different cell types. Proc. natn. Acad. Sci. U.S.A. 75, BENNETT, G. S., FELLINI, S. A. & HOLTZER, H. (1978a). Immunofluorescent visualization of 100 A filaments in different cultured chick embryo cell types. Differentiation 12, BEN-ZE'EV, A., DUERR, A., SOLOMON, F. & PENMAN, S. (1979). The outer boundary of the cytoskeleton: a lamina derived from plasma membrane proteins. Cell 17, BORENFREUND, E., SCHMID, E., BENDICH, A. & FRANKE, W. W. (1980). Constitutive aggregates of intermediate-sized filaments of the vimentin- and cytokeratin-rype in cultured hepatoma cells and their dispersal by butyrate. Expl Cell Res. 127, BROWN, S., LEVINSON, W. & SPUDICH, J. A. (1976). Cytoskeletal elements of chick embryo fibroblasts revealed by detergent extraction. _?. supramolec. Struct. 5, DAVISON, P. F., HONG, B.-S. & COOKE, P. (1977). Classes of distinguishable 10 nm cytoplasmic filaments. Exptl Cell Res. 109, EGBERTS, E., HACKETT, P. B. & TRAUB, P. (1976). Protein synthesis in postnuclear supernatants from mengovirus-infected Ehrlich ascites tumor cells. Hoppe Seyler's Z. physiol. Chan. 357, FELLINI, S. A., BENNETT, G. S., TOYAMA, Y. & HOLTZER, H. (1978). Biochemical and immunological heterogeneity of 100 A filament subunits from different chick cell types. Differentiation 12, FRANKE, W. W., SCHMID, E., BREITKREUZ, D., LODER, M., BOUKAMP, P., FUSENIG, N. E., OSBORN, M. & WEBER, K. (1979a). Simultaneous expression of two different types of intermediate-sized filaments in mouse keratinocytes proliferating in vitro. Differentiation 14, FRANKE, W. W., SCHMID, E., KARTENBECK, J., MAYER, D., HACKER, H.-J., BANNASH, P., OSBORN, M., WEBER, K., DENK, H., WANSON, J.-C. & DROCHMANS, P. (19796). Characterization of the intermediate-sized filaments in liver cells by immunofluorescence and electron microscopy. Biol. cellulaire 34,

26 Cations and stability of vimentin filaments. I 75 FRANKE, W. W., SCHMID, E., OSBORN, M. & WEBER, K. (1978a). The intermediate-sized filaments in rat kangaroo PtK, cells. II. Structure and composition of isolated filaments. Cytobiologie 17, FRANKE, W. W., SCHMID, E., OSBORN, M. & WEBER, K. (19786). Different intermediate-sized filaments distinguished by immunofluorescence microscopy. Proc. natn. Acad. Sci. U.S.A. 75, S034-S038. FRANKE, W. W., SCHMID, E., WINTER, S., OSBORN, M. & WEBER, K. (1979c). Widespread occurrence of intermediate-sized filaments of the vimentin-type in cultured cells from diverse vertebrates. Expl Cell Res. 123, FRANKE, W. W., WEBER, K., OSBORN, M., SCHMID, E. & FREUDENSTEIN, C. (1978c). Antibody to prekeratin. Decoration of tonofilament-like arrays in various cells of epithelial character. Expl Cell Res. 116, FUKUYAMA, K., MUROZUKA, T., CALDWELL, R. & EPSTEIN, W. L. (1978). Divalent cation stimulation of in vitro fibre assembly from epidermal keratin protein. J. Cell Sci. 33, FULTON, A. B., WAN, K. M. & PENMAN, S. (1980). The spatial distribution of polyribosomes in 3T3 cells and the associated assembly of proteins into the skeletal framework. Cell 20, GARD, D. L., BELL, B. P. & LAZARIDES, E. (1979). Coexistence of desmin and the fibroblastic intermediate filament subunit in muscle and nonmuscle cells: Identification and comparative peptide analysis. Proc. natn. Acad. Sci. U.S.A. 76, HENDERSON, D. & WEBER, K. (1980). Immunoelectron microscopic studies of intermediate filaments in cultured cells. Expl Cell Res. 129, HEUSER, J. E. & KIRSCHNER, M. W. (1980). Filament organization revealed in platinum replicas of freeze-dried cytoskeletons. J. Cell Biol. 86, LAZARIDES, E. (1080). Intermediate filaments as mechanical integrators of cellular space. Nature, Land. 283, LENK, R. & PENMAN, S. (1979). The cytoskeletal framework and poliovirus metabolism. Cell 16, LENK, R., RANSOM, L., KAUFMANN, Y. & PENMAN, S. (1977). A cytoskeletal structure with associated polyribosomes obtained from HeLa cells. Cell 10, O'CONNOR, C. M., GARD, D. L. & LAZARIDES, E. (1981). Phosphorylation of intermediate filament proteins by camp-dependent protein kinases. Cell 23, OSBORN, M., BORN, T., KOITSCH, H.-J. & WEBER, K. (1978). Stereo immuno-fluorescence microscopy: I. Three-dimensional arrangement of microfilaments, microtubules and tonofilaments. Cell 14, OSBORN, M., FRANKE, W. W. & WEBER, K. (1980). Direct demonstration of the presence of two immunologically distinct intermediate-sized filament systems in the same cell by double immunofluorescence microscopy. Expl Cell Res. 135, RUECER, D. C, HUSTON, J. S., DAHL, D. & BIGNAMI, A. (1979). Formation of 100 A filaments from purified glial fibrillary acidic protein in vitro. J. molec. Biol. 135, SAKAMOTO, M., TZENG, S., FUKUYAMA, K. & EPSTEIN, W. L. (1980). Light-scattering studies of cation-stimulated filament assembly of newborn rat epidermal keratin. Biochim. biophys. Ada 624, SMALL, J. V. & CELIS, J. E. (1978). Direct visualization of the 10-nm (100 A) filament network in whole and enucleate cultured cells. J. Cell Sci. 31, STARGER, J. M., BROWN, W. E., GOLDMAN, A. E. & GOLDMAN, R. D. (1978). Biochemical and immunological analysis of rapidly purified 10 nm-filaments from baby hamster kidney (BHK-21) cells. J. Cell Biol. 78, STEINERT, P. M. & IDLER, W. W. (1976). Self-assembly of bovine epidermal keratin filaments in vitro. J. molec. Biol. 108, STEINERT, P. M., IDLER, W. W. & GOLDMAN, R. D. (1980). Intermediate filaments of baby hamster kidney (BHK-21) cells and bovine epidermal keratinocytes have similar ultrastructures and subunit domain structures. Proc. natn. Acad. Sci. U.S.A. 77, STEINERT, P. M., ZIMMERMANN, S. B., STARCER, J. M. & GOLDMAN, R. D. (1978). Ten nanometer filaments of hamster BHK-21 cells and epidermal keratin filaments have similar structures. Proc. natn. Acad. Sci. U.S.A. 75, TRAUB, P. & BOECKMANN, G. (1978). High-resolution polyacrylamide gradient slab gel electrophoresis of histones Hi, H3 and H4. Hoppe Seyler's Z. physiol. Chem. 359,

27 76 P. Traub and W. J. Nelson WEBSTER, R. E., OSBORN, M. & WEBER, K. (1978). Visualization of the same PtK, cytoskeletons by both immunofluorescence microscopy and low power electron microscopy. Expl Cell Res. 117, ZACKROFF, R. V. & GOLDMAN, R. D. (1979). In vitro assembly of intermediate filaments from baby hamster kidney (BHK-21) cells. Proc. natn. Acad. Sci. U.S.A. 76, ZACKROFF, R. V. & GOLDMAN, R. D. (1980). In vitro reassembly of squid brain intermediate filaments (neurofilaments): Purification by assembly-disassembly. Science, N.Y. 208, (Received 28 May 1981)