Oncogene (1998) 16, 3423 ± 3434  1998 Stockton Press All rights reserved 0950 ± 9232/98 $12.00 http://www.stockton-press.co.uk/onc Cytoplasmic retention of mutant tsp53 is dependent on an intermediate ®lament protein (Vimentin) sca€old

Oliver Klotzsche1,DoÈrte Etzrodt2, Heinz Hohenberg1, Wolfgang Bohn1 and Wolfgang Deppert1

1Heinrich-Pette-Institut fuÈr Experimentelle Virologie und Immunologie an der UniversitaÈt Hamburg, Martinistr. 52, D-20251 Hamburg, Germany

The temperature-sensitive mutant tsp53val135 accumulates these mutations do not simply inactivate the tumor in the cytoplasm of cells kept at the non-permissive suppressor functions of wild-type (wt) p53, but provide temperature (398C), but is rapidly transported into the a `gain of function' for the mutant (mt) p53 (Deppert, cell nucleus at the permissive temperature (308C). tsp53 1996; Dittmer et al., 1993). This hypothesis is thus may serve as a model for analysing cellular supported by a variety of biological and biochemical parameters in¯uencing the subcellular location of p53. observations, which strongly suggest that mt p53 exerts Here we provide evidence that retention of tsp53 in the properties of a dominant oncogene. cytoplasm at the non-permissive temperature is due to The multitude of functions ascribed to either wt or cytoskeletal anchorage of the p53 protein. Two sublines mt p53 requires that the functional properties of these of C6 rat glioma cells di€ering in their expression of the proteins are tightly and coordinatedly regulated. In intermediate ®lament protein vimentin (vimentin expres- addition to various posttranslational modi®cations and sing or vimentin negative cells) were stably transfected interactions with viral and cellular proteins (Deppert, with a vector encoding tsp53. Whereas cells of vimentin 1994a, b), the subcellular location of p53 seems to be expressing C6 subclones retained tsp53 in the cytoplasm an important determinant of its function (Shaulsky et at the non-permissive temperature, cells of vimentin al., 1991). The p53 molecule is endowed with three negative subclones exclusively harbored the tsp53 within nuclear translocation signals (NLS) (Shaulsky et al., their nuclei. Intermediate ®lament de®cient cells that had 1990b), but, in addition, also contains a nuclear export been reconstituted with a full length vimentin protein signal (Middeler et al., 1997), suggesting that p53 is again showed a cytoplasmic localization of tsp53, able to shuttle between the cytoplasmic and the nuclear whereas in cells expressing a C-terminally truncated compartment, depending on functional requirements. (tail-less) vimentin tsp53 localized to the nucleus. We Indeed, there is ample evidence for a function- conclude that cytoplasmic sequestration of tsp53 requires dependent localization of p53 to either cellular an intact intermediate ®lament system. compartment. When resting cells resume cycling, wt p53 remains in the cytoplasm during G1 phase, but is Keywords: nuclear exclusion; cytoplasmic anchorage; translocated into the nucleus at the onset of S-phase temperature sensitive mutant p53; vimentin; intermedi- (Shaulsky et al., 1990a), in accordance with a role of wt ate ®laments; C6 rat glioma cells p53 in translational control of its own mRNA in G1 (Mosner et al., 1995). During di€erentiation of neurons and oligodendrocytes explanted from embryonic rat brain, wt p53 is translocated into the nucleus and is Introduction relocated to the cytoplasm in di€erentiated cells (Eizenberg et al., 1996). Similarly, wt p53 is over- The tumor suppressor p53 is a multifunctional protein, expressed and located in the cytoplasm in normal exerting a variety of functions in cell cycle control, lactating breast tissue suggesting a role for nuclear DNA repair, DNA recombination, initiation of exclusion of p53 in the developing mammary gland apoptosis, and control of di€erentiation (Gottlieb and (Moll et al., 1992). Oren, 1996; Ko and Prives, 1996), rendering it the Variations in subcellular localization were also currently most prominent `gate-keeper of cell growth noticed in tumor cells expressing wt p53. In certain and di€erentiation' (Levine, 1997). In line with this mammary carcinomas and in undi€erentiated neuro- central role of p53 is the observation that inactivation blastomas wt p53 accumulates within the cytoplasm, of wild-type speci®c functions of p53 by mutations in whereas it localizes to the nucleus in di€erentiated the p53 gene is a very frequent event in human tumors neuroblastomas (Moll et al., 1992, 1995). (Hollstein et al., 1991; Hainhaut et al., 1997). Most of Like the subcellular localization of wt p53, also that these mutations are missense point mutations, leading of mt p53 is subject to considerable variation, ranging to the expression of a mutant p53 carrying a single from exclusively cytoplasmic to exclusively nuclear in amino acid substitution (Cho et al., 1994; Hollstein et tumors and in transformed cells (Porter et al., 1992; al., 1996; Medcalf et al., 1992). The unusual mutational Rotter et al., 1983; Takahashi and Suzuki, 1994). spectrum (490% missense point mutations, localized Importantly, the subcellular localization of mt p53 is within the p53 core domain) led to the hypothesis that not a function of speci®c mutations in the p53 gene, as in human glioblastomas a speci®c mt p53 protein (codon 273; Arg?Cys) localized to the nuclei in some Correspondence: W Bohn tumors, but to the cytoplasm in others, indicating that 2Current address: Klinik fuÈ r Innere Medizin an der UniversitaÈ t Rostock Ernst Heydemannstr. 6, 18055 Rostock, Germany the subcellular localization of mt p53 is not determined Received 9 March 1998; accepted 27 May 1998 by the mutation itself, but may be dependent on Cytoplasmic retention of tsp53 O Klotzsche et al 3424 environmental conditions in the cell (Ali et al., 1994), proteins can easily be addressed by the use of and/or its functional status. The latter has been inhibitors which speci®cally a€ect the structural observed with mt p53 in MethA tumor cells, which arrangement of these ®lament systems. In contrast, in cycling cells is predominantly nuclear, but becomes an involvement of intermediate ®laments is more transferred to the cytoplasm when the cells growth dicult to assay due to the stability of this system arrest at the restriction point in G1 (Steinmeyer and under physiological conditions, and due to the absence Deppert, 1988). All these data argue that the of speci®c inhibitors. Therefore we used a cell system subcellular localization of p53 is subject to certain which is based on C6 rat glioma cells. From these cells regulatory mechanisms within the cell, some of which subclones were isolated (RoÈ ser et al., 1991) that either might be connected with the status of di€erentiation. express or do not express vimentin, the most common However, the factors regulating the presence both of intermediate ®lament protein expressed in cultured cells mt and of wt p53 in di€erent subcellular locations are (Traub and Vorgias, 1983). Such cells thus either still largely unknown. contain or are de®cient of a cytoplasmic intermediate The temperature-sensitive mutant p53val135 in the past ®lament system. Ectopic expression of tsp53 in these has proven a useful system to analyse cellular cells demonstrated that tsp53 was strictly nuclear in parameters in¯uencing the subcellular location of both cells of vimentin-negative subclones, whereas cytoplas- mt and wt p53. At the non-permissive temperature mic sequestration of tsp53 was dependent on an intact (4378C), tsp53 is present in its mutant form and intermediate ®lament system. The system of vimentin- localizes predominantly to the cytoplasm. Upon shift expressing and vimentin-negative C6 subclones should to the permissive growth temperature (30 ± 328C), the provide a valuable experimental system for the tsp53 adopts its wild-type form, becomes translocated identi®cation of mechanisms regulating the subcellular to the cell nucleus and exerts wt p53 functions localization of p53. (Martinez et al., 1991; Michalovitz et al., 1990), i.e. induction of growth arrest or apoptosis, depending on cell type. However, there is neither a strict dependence Results for a cytoplasmic location of tsp53 when present in its mutant form, nor for a nuclear location when present tsp53 is bound to the cytoskeleton in IF (vimentin) in its wild-type form, indicating that cellular and/or positive cells functional parameters rather than its conformational state determine the subcellular location of tsp53. Thus Cytoskeletal anchorage of p53 was ®rst assayed in the cytoplasmic localization of tsp53 at the nonpermis- vimentin-expressing C6D8 cells stably transfected with sive temperature can be converted to a nuclear one by tsp53. C6D8 cells contain intermediate ®laments treating the cells with translational inhibitors, indicat- (vimentin) in addition to microtubules and actin ing that continuous protein synthesis is necessary to ®laments (RoÈ ser et al., 1991). By immunofluores- retain this mutant protein within the cytoplasm, cence, Western- and Northern-blotting, the parental possibly via short lived cytoplasmic anchor proteins cells do not express detectable levels of endogenous (Gannon and Lane, 1991). Conversely, tsp53 does not p53, which could interfere with or mask the function of localize to the nucleus in PC12 cells even at the an ectopic p53 (data not shown). By immunofluores- permissive temperature, and is redistributed to the cence and by Western-blotting the p53 antibodies used cytoplasm in clone 6 cells, when the cells are kept at did not cross-react with any other structural compo- the permissive temperature for a prolonged time. In nent in these cells. both instances, cytoplasmic localization of tsp53 at the Eleven out of 20 subclones isolated after transfection permissive temperature correlated with loss of wt p53 with vector pLTRcG9 expressed tsp53, eight of them function (Knippschild et al., 1996). showing a very strong cytoplasmic immuno¯uorescence Whereas the nuclear location of p53 is understood as staining of tsp53 at 398C (see Figure 1B for a a consequence of the speci®c transport of p53 to its representative example). In addition to the consistently major functional location, very little is known about strong cytoplasmic staining for p53 in these cells, also the mechanism retaining p53 in the cytoplasm. Cell some nuclear staining could be observed, varying fractionation studies showed that cytoplasmic tsp53 considerably in intensity from complete absence to resists extraction of cells with non-ionic detergents dot-like inclusions. Nuclear inclusions were present when applying conditions normally used to identify the independently of the way the cells were treated prior to association of proteins with the cytoskeleton. This not ®xation and therefore did not re¯ect an artefact due to only was found for cytoplasmic tsp53 both in its wild- improper handling of the probes. A comparative type as well as in its mutant form, but could be analysis of the arrangement of actin, tubulin, and extended to cytoplasmic wt p53 in other systems vimentin ®lament systems in tsp53 transfected versus (Gannon and Lane, 1991; Knippschild et al., 1996; non-transfected cells revealed no di€erences, indicating Moll et al., 1996; Rotter et al., 1983; Zerrahn et al., that ectopic expression of tsp53 did not alter the 1992), pointing to a general role of cytoskeletal normal arrangement of these ®lament systems. To structures for the cytoplasmic localization of p53. determine binding of tsp53 to the cytoskeleton the cells Using tsp53val135 as an experimental system we asked were lysed with a non-ionic detergent, applying in this study, whether any and which of the major conditions which were shown to produce properly cytoskeletal ®lament systems, namely actin ®laments, preserved cytoskeletons without remnants of mem- microtubules, or intermediate ®laments would serve as brane (Bohn et al., 1993; Nebe et al., 1997; Okabe and a backbone for cytoplasmic anchorage of p53 Hirokawa, 1988). The strong cytoplasmic ¯uorescence molecules. The question of an involvement of actin signal of tsp53 in extracted cells (Figure 1D) suggested or tubulin in the subcellular localization of cytoplasmic that the bulk of this protein was not lost during cell Cytoplasmic retention of tsp53 O Klotzsche et al 3425 lysis but was retained on the cytoskeleton. Correspond- tribution of actin into patches concomitant with ingly, the amount of tsp53 in the supernatant lysis changes in cell shape (compare FITC-labeling in bu€er (fraction S) was low in comparison to that in the Figure 2A and B). However, p53 (TRITC-labeling in detergent insoluble cytoskeletal fraction (fraction I) as Figure 2C) did not comigrate with actin into these determined by Western immunoblotting (Figure 1D: patches (compare Figure 2B and C) and the majority insert). of the tsp53 still remained bound to the cytoskeleton after extraction of these cells with a non-ionic detergent (Figure 2C). p53 relocalizes with vimentin but not with actin or To determine an involvement of microtubules in tubulin in drug treated cells subcellular localization of tsp53, we incubated the cells To evaluate which of the major ®lament systems might with vinblastin. This microtubule speci®c drug rear- be involved in cytoplasmic localization of tsp53 we ®rst ranged tubulin into paracrystalline aggregates (com- followed the subcellular distribution of the p53 protein pare FITC-labeling in Figure 3A and B) and in cells treated with ®lament speci®c drugs. Cytocha- cytoplasmic p53 (TRITC-labeling in Figure 3B) now lasin B was used to determine if the subcellular localized close to the nucleus. The vimentin network localization of cytoplasmic tsp53 could be a€ected by collapsed and was arranged around the nucleus in a disruption of the actin ®lament network. Cytochalasin whirl like manner (FITC-labeling in Figure 3C). tsp53 induced fragmentation of actin ®laments and redis- (TRITC-labeling in Figure 3D) remained bound to the extracted non extracted

Vimentin p53

Figure 1 Retention of tsp53 on the cytoskeleton. Vimentin positive C6D8 cells stably transfected with tsp53 were either ®xed directly with acetone (A and B), or lysed with Triton X-100 and ®xed with formaldehyde (C and D) and labeled with antibodies to vimentin and p53; insert in (D): detection of tsp53 in the supernatant lysis bu€er (S) and the insoluble (I) cytoskeletal fraction of detergent treated cells by Western immunoblotting Cytoplasmic retention of tsp53 O Klotzsche et al 3426 eated untr eated Cytochalasin tr

Actin p53

Figure 2 E€ect of cytochalasin B on the distribution of actin and tsp53 in C6D8 cells at 398C. Immuno¯uorescence labeling with phalloidin-FITC (A and B) or p53 antibodies (TRITC) (C); (A): acetone ®xed cells; (B) and (C): lysed cells. Cytochalasin fragmented the actin stress ®bers and redistributed actin (compare A and B), but tsp53 did not comigrate with actin patches (compare B and C); tsp53 was still retained on the cytoskeleton (C)

cytoskeleton after cell lysis also under these conditions, immuno¯uorescence microscopy. Cells were grown and colocalized with the rearranged vimentin. on glass coverslips, and lysed with a non-ionic These data suggested that the distribution of tsp53 detergent as detailed in Materials and methods. p53 in the cytoplasm of C6D8 cells was neither was identi®ed by immunogold labeling with the determined by microtubules nor by the actin monoclonal p53 antibody PAb248. Cytoskeletons of ®lament network, but seemed to follow the tsp53 transfected cells exhibited a strong labeling distribution of the intermediate ®lament protein with the p53 antibody (Figure 4A). In controls only vimentin. a few gold particles were scattered throughout the network without any preferential association with ®laments (Figure 4B). The immunogold particles tsp53 complexes are attached to cytoskeletal ®laments at identifying p53 were not evenly distributed through- the EM level out the network, but were concentrated at complexes Our data obtained so far point to an association of attached to individual ®laments (Figure 4C and D). tsp53 with the cytoskeleton involving the vimentin In accordance with previous observations using this network. We next analysed the interaction of p53 cell line (Bohn et al., 1993; Foisner et al., 1995), with the cytoskeleton by immunoelectron micro- these ®laments were of intermediate size and thus scopy, as this approach allows a more precise represented vimentin ®laments. This could also be location of p53 to cytoskeletal elements than veri®ed by immunogold double-labeling showing that Cytoplasmic retention of tsp53 O Klotzsche et al 3427 p53 containing complexes identi®ed by 5 nm gold randomly selected showed expression of tsp53. particles were associated with ®laments that were Transfection of C6D10 cells did not lead to re- labeled with vimentin antibodies and 10 nm immu- expression of vimentin in these subclones as deter- nogold particles (Figure 4D). mined by immuno¯uorescence labeling with a poly- clonal vimentin antibody (Figure 5A). All cells exhibited a strong nuclear staining of tsp53 in IF-negative cells accumulate tsp53 within the nucleus immuno¯uorescence already at the non-permissive If vimentin indeed would be of functional importance temperature (Figure 5B), whereas no, or only very for the subcellular localization of tsp53, then cells little cytoplasmic staining was observed. Very low which are devoid of cytoplasmic IF proteins should amounts of tsp53 were present within the detergent show an altered subcellular localization of tsp53. soluble fraction as assayed by Western immunoblot- When the vimentin-de®cient C6D10 cells were stably ting, providing evidence that the nuclear tsp53 in these transfected with tsp53, eight out of 20 subclones cells was tightly bound (Figure 5B: insert).

untreated Vinblastin treated

Tubulin Tubulin/p53 eated inblastin tr V

Figure 3 E€ect of vinblastin on the distribution of tubulin, vimentin, and tsp53 in C6D8 cells at 398C. Immuno¯uorescence labeling with antibodies to tubulin (FITC) (A and B), vimentin (FITC) (C), and p53 (TRITC) (B and D): (A) and (B): acetone ®xed cells; (C) and (D): lysed cells; vinblastin rearranged tubulin into distinct paracrystalline aggregates (compare FITC-image of A and B) and vimentin collapsed into perinuclear whirls (C). tsp53 (TRITC-image) was concentrated around the nucleus (B), and colocalized with the rearranged vimentin (compare C and D) Cytoplasmic retention of tsp53 O Klotzsche et al 3428 vimentin sca€old had been reconstituted by ectopic Cells with a restored vimentin sca€old accumulate tsp53 expression. within the cytoplasm Vimentin-negative C6D10 cells were transfected with The presence of tsp53 within the cytoplasm of an expression vector encoding a full length mouse vimentin-positive cells, versus nuclear accumulation of vimentin protein which contained a single amino acid the protein in vimentin-negative cells suggested that the substitution (Val?Leu) in the rod domain. This presence of vimentin is a prerequisite for the substitution does not in¯uence ®lament formation by cytoplasmic anchorage of this protein. Therefore we the vimentin protein, but creates an epitope for the asked whether tsp53 would again localize to the monoclonal antibody VIM3B4, which otherwise does cytoplasm in originally IF-de®cient cells in which a not recognize rodent vimentin (Bohn et al., 1992). This

Figure 4 Immunoelectron microscopic localization of tsp53 on cytoskeletons of vimentin positive C6D8 cells. The 10 nm immunogold particles trace p53 containing complexes associated with the ®lament network (A); (B) control: labeling with normal mouse serum and 10 nm immunogold conjugates; (C) stereo image; p53 containing complexes are attached to ®laments corresponding in width to intermediate ®laments (large arrows); thin connecting ®laments (small arrow); ®lament corresponding in width to actin ®laments (arrowhead); (D) immunogold double labeling with antibodies to vimentin (10 nm immunogold particles) and p53 (5 nm immunogold particles); bar in A, B, and C=200 nm; bar in D=50 nm Cytoplasmic retention of tsp53 O Klotzsche et al 3429 enabled us to unequivocally distinguish the ectopic could be detected in these cells by Western immuno- vimentin from possibly induced endogenous vimentin, blotting with the Vim3B4 (Figure 6, lane 7), but not which can be detected with antibody V9. Stably with the V9 vimentin antibody (Figure 6, lane 3). These transfected cells exhibited a characteristic vimentin data veri®ed the presence of an ectopic mouse vimentin staining pattern with the VIM3B4 antibody in and the absence of endogenous vimentin. Stable immuno¯uorescence (Figure 5C), and a 56 kD protein transfection of cells re-expressing vimentin with tsp53 imentin negative V imentin full length V imentin uncated V tr

Vimentin p53

Figure 5 Subcellular localization of tsp53 in dependence of vimentin expression. Immuno¯uorescence double labeling of vimentin (A, C and E) and p53 (B, D, and f) in vimentin negative cells (A and B), vimentin negative cells with ectopic expression of a full length mouse vimentin protein (C and D), and in vimentin negative cells with ectopic expression of a tail-less mouse vimentin protein (E and F). Absence of vimentin or the presence of a tail-less vimentin protein correspond to nuclear accumulation of tsp53. Detection of tsp53 in the supernatant lysis bu€er (S) and the insoluble (I) cytoskeletal fraction of detergent treated cells by Western immunoblotting (B, insert) Cytoplasmic retention of tsp53 O Klotzsche et al 3430 again led to a cytoplasmic accumulation of tsp53 at the immunoblots a 50 kD vimentin protein could be non-permissive temperature of 398C, whereas only low detected with the VIM3B4 antibody (Figure 6, lane amounts of p53 were found within the nucleus (Figure 8), whereas no reaction was obtained with the V9 5D). Again, the cytoplasmic tsp53 was retained in the antibody (Figure 6, lane 4), again verifying the absence cytoskeletal fraction after lysis with non-ionic deter- of endogeneous vimentin. When these cells were gents. Thus complementation of cells of the IF- transfected with the tsp53 expression vector, 11 out de®cient C9 subclone with vimentin enabled binding of 20 subclones isolated showed expression of tsp53. of tsp53 to the cytoskeleton. All of them exhibited a strong nuclear, but no signi®cant cytoplasmic staining of tsp53 already at the non-permissive temperature (Figure 5F), suggesting Cells expressing a tail-less vimentin protein accumulate that the tail domain of vimentin is essential for the tsp53 within the nucleus cytoplasmic sequestration and cytoskeletal anchorage To further demonstrate the speci®city of the interaction of tsp53. of p53 with a vimentin cytoskeleton we analysed the subcellular localization of tsp53 in cells expressing a Disassembly of a preexisting vimentin sca€old truncated vimentin protein, which is still able to form a redistributes cytoplasmic tsp53 ®lament network. Whereas the aminoterminal domain, the head domain, is indispensable for assembly of We next wanted to know whether the breakdown of a vimentin proteins into ®laments, truncation of the preexisting vimentin ®lament network in cells in which carboxyterminal domain, the tail domain, does not tsp53 is anchored to the cytoskeleton would directly prevent this assembly process, but seems to a€ect the a€ect the subcellular localization of tsp53. The arrangement of these ®laments into a proper network assembly of vimentin protein into ®laments is within the cell (Eckelt et al., 1992; Herrmann et al., eciently prevented by mutations in the aminoterm- 1992; McCormick et al., 1993; Traub and Vorgias, inal head domain of vimentin (Herrmann et al., 1992; 1983). IF-de®cient C6D10 cells were transfected with Traub and Vorgias, 1983). Such mutant vimentin thus an expression vector coding for a vimentin protein can be used to disturb an existing vimentin system by lacking the carboxyterminal tail domain due to the ectopic expression. Therefore, we transiently trans- presence of a stop codon at the end of the rod domain fected tsp53 expressing, vimentin-positive C6D8 cells coding sequence. This tail-less vimentin protein also with a vimentin expression vector encoding a mutated contained the VIM3B4 epitope. Immuno¯uorescence vimentin protein lacking aa 43 ± 102 of the amino- staining of stably transfected cells with VIM3B4 terminal head domain. After transfection, the cells exhibited an irregular vimentin network, which was were ®xed with acetone and double-labeled with restricted to the cytoplasm (Figure 5E). In Western antibodies to vimentin and p53. The transfected cells showed dramatic rearrangements of vimentin into cytoplasmic aggregates (Figure 7A and C). tsp53 was redistributed within these cells and largely colocalized with these vimentin aggregates (Figure 7B). No enhanced nuclear accumulation of tsp53 was detect- able in these cells. In contrast to the drastic changes in subcellular distribution of tsp53, the distribution of actin was not a€ected by this vimentin rearrangement (Figure 7C and D), verifying that the redistribution of p53 induced by these conditions not merely re¯ected an unspeci®c rearrangement of cytoplasmic compo- nents.

Discussion

Using the temperature-sensitive mutant tsp53Val-135 we show that in C6 rat glioma cells at the non-permissive temperature the subcellular localization of tsp53 is dependent on whether these cells express, or do not express an intact intermediate ®lament (vimentin) protein sca€old. In vimentin-positive C6 cells, tsp53Val-135 largely accumulated within the cytoplasm in accordance with its behaviour in other cell types (Gannon and Lane, 1991; Knippschild et al., 1996; Figure 6 Western immunoblot detection of vimentin. Cellular Martinez et al., 1991; Michalovitz et al., 1990). lysates of vimentin positive C6D8 cells, vimentin negative C6D10 cells, vimentin negative cells with ectopic expression of a full Vimentin-negative C6 cells, in contrast, exclusively length vimentin protein VimWT3B4, or a carboxyterminal harbored the tsp53 protein within the nucleus, whereas truncated vimentin protein VimMut3B4 were analysed for the same cells when reconstituted with vimentin again vimentin; lanes 1 ± 4 detection with antibody V9; lanes 5 ± 8 showed a cytoplasmic accumulation of tsp53. Thus, detection with antibody Vim 3B4. Ectopic vimentin is detected by depending on the intracellular environment of the cell, the VIM3B4 antibody (lanes 7 and 8), but not by the V9 antibody (lanes 3 and 4). The negative reaction with the V9 antibody in a given mutant p53 protein will behave di€erently with lanes 2 ± 4 indicates the absence of endogenous vimentin respect to its subcellular localization. Cytoplasmic retention of tsp53 O Klotzsche et al 3431 An association of cytoplasmic p53 with components the cytoskeleton could re¯ect a stabilizing e€ect of of the cytoskeleton has already been suggested vimentin on other cytoskeletal ®lament systems to previously (Knippschild et al., 1996; Rotter et al., which tsp53 would be bound. However, neither 1983). Here we demonstrate that this interaction is destruction of the actin ®lament system by cytochala- dependent on an intact intermediate ®lament (vimen- sin, nor the breakdown of the microtubule system by tin) system. Vimentin ®laments are arranged in close vinblastin displaced tsp53 from the cytoskeleton, nor interaction with other ®lament systems, and serve as a did these treatments induce translocation of tsp53 into backbone for the spatial organization of various the nucleus. After treatment with both agents the cellular organelles (Albers and Fuchs, 1992; Janmey majority of tsp53 was still retained on the cytoskeleton et al., 1991; Lazarides, 1980). In C6 cells the vimentin and followed the arrangement of vimentin. In contrast, ®laments were shown to be physically linked to disrupting the vimentin ®lament network in vivo by microtubules and to the actin ®lament system via thin coexpression of a trans-dominant interfering amino- connecting ®laments that are composed of plectin terminally truncated vimentin protein redistributed (Foisner et al., 1995). Therefore, one could argue that tsp53. These data render it extremely unlikely that the observed vimentin-dependent anchorage of tsp53 to tsp53 retained on extracted cells is simply trapped

Vimentin p53

Vimentin Actin

Figure 7 Redistribution of vimentin and tsp53 in C6D8 cells transiently expressing a vimentin protein lacking part of the aminoterminal head domain. Double immuno¯uorescence labeling of either vimentin (A) and p53 (B), or vimentin (C) and actin (D); tsp53 (B) but not actin (D) colocalizes with vimentin aggregates (A and C) Cytoplasmic retention of tsp53 O Klotzsche et al 3432 unspeci®cally by the cytoskeleton, but rather point to a addition, hsc70 stayed in complex with tsp53, when the speci®c association of the tsp53 with the vimentin cytoplasmic localization of tsp53 was lost in the sca€old. The speci®city of this interaction is further presence of cycloheximide (Gannon and Lane, 1991), substantiated by our observation that reconstitution of arguing against a role of hsc70 in cytoplasmic a vimentin system using a tail-less vimentin failed to sequestration. retain tsp53 in the cytoplasm. Tail-less vimentin in It is tempting to speculate that the mechanisms, these cells formed an irregular network which resisted which are responsible for the vimentin-dependent extraction with nonionic detergents, indicating that the subcellular localization of tsp53 in C6 rat glioma cells truncated vimentin was not soluble, but still was able may be of general relevance for the cytoplasmic to form stable structures within the cell. These data sequestration of either mutant or wild type p53 in argue that either a properly organized vimentin tumor cells. This speculation is supported by the ®lament network or the vimentin tail domain itself is observation that in clone 6 cells tsp53 in its wild-type necessary to retain tsp53 on the cytoskeleton. While the form was excluded eciently from the nucleus upon function of the head domain of vimentin has well been prolonged cultivation of the cells at the permissive described, that of the tail domain is still not completely temperature (Knippschild et al., 1996). Again, this understood (Ciesielski-Treska et al., 1991; Herrmann et cytoplasmic p53 resisted extraction with non-ionic al., 1992; Inagaki et al., 1990; Ogawara et al., 1995; detergents. However, so far no detailed studies have Traub and Vorgias, 1983). Probably, the tail controls been conducted on a possible correlation between the the lateral association of IF-protein tetramers and subcellular localization of p53 and the IF-expression thereby stabilizes the ®lament structure, which may be patterns in tumors or in cultured cells derived thereof. essential to establish a proper ®lament network (Heins The experimental system established here should o€er a and Aebi, 1994). In addition the tail domain is exposed reasonable approach to obtain more insight into the at the ®lament surface (Troncoso et al., 1990), and thus mechanisms regulating the subcellular localization of is easily accessible to interaction with proteins. The tail p53, expecially its speci®c retention in the cytoplasm, domain therefore may be required for an interaction and should allow the identi®cation of additional with cellular components (Bader et al., 1991; Eckelt et proteins involved in this process. al., 1992), a suggestion which would correspond to our observations shown here. Despite the documented importance of the vimentin Materials and methods sca€old for the cytoplasmic anchorage of tsp53, there is at least indirect evidence that this anchorage is not Cell culture mediated via a direct association of the p53 protein C6 rat glioma cells were grown in Dulbecco's Modi®ed with vimentin molecules. Firstly, our immunoelectron Eagle's Medium (DMEM) supplemented with 10% fetal microscopic images clearly showed that tsp53 was not calf serum (FCS). Media used for selection of transfected distributed along vimentin ®laments, but rather was cells either contained Geneticin (400 mg/ml) or both found in association with so far unidenti®ed protein Geneticin (400 mg/ml) and Hygromycin B (100 mg/ml). complexes attached to these ®laments. Also the ®nding that the association of tsp53 with the intermediate Vimentin and tsp53 expression vectors ®lament sca€old is sensitive to a short treatment of the cells with inhibitors of protein biosynthesis argues A full length mouse vimentin cDNA (Hennekes et al., against a direct interaction of tsp53 with vimentin, as 1990) was subcloned into the eukaryotic expression vector pSG5 (Stratagene, La Jolla) which drives expression of the the vimentin ®lament system as such is not noticeably inserted cDNA by the SV40 early promoter. At position disturbed under such conditions (unpublished observa- 353 of this vimentin protein valine was substituted for tion). We therefore favour the hypothesis that tsp53 is leucine by in vitro mutagenesis (Bohn et al., 1992) to not directly anchored to the ®lament network, but is facilitate speci®c recognition of this protein by the indirectly bound via a linker protein which has an VIM3B4 antibody. This plasmid was designated anity for vimentin (intermediate) ®laments and is pMBVim2. The vimentin expression vector pVimEco is subject to regulatory mechanisms. This would also be identical to pMBVim2 but contains a stop codon compatible with the previous suggestion that tsp53 is immediately after the sequence coding for helix 2B. This retained in the cytoplasm by a short lived anchor vimentin protein thus lacked the complete tail domain. The protein(s) (Gannon and Lane, 1991; Knippschild et al., expression vector D61 codes for a vimentin protein lacking AS 50 ± 104 of the head domain. Expression of the murine 1996). tsp53Val135 was based on the eukaryotic expression vector Nothing, so far is known about the postulated pLTRcG9 (Eliyahu et al., 1985). bridging or anchor protein(s) mediating the association of p53 with the vimentin sca€old. The constitutively expressed heat shock protein hsc70 which preferentially Stable transfection binds to tsp53 in its mutant conformation (Gannon Transfection of C6 subclones was carried out by and Lane, 1991; Hinds et al., 1987; Pinhasi-Kimhi et electroporation. The cells were grown to 70% con¯uence, al., 1986) in theory could represent such an anchor trypsinized, and suspended in serum free medium at 107 protein. Chaperones themselves have an anity for cells/ml. Four hundred ml of this suspension were mixed intermediate ®laments (Cheng and Lai, 1994; Liao et with 40 mg DNA in an electroporation cuvette. Vimentin and tsp53 expression vectors were linearized and cotrans- al., 1995) and therefore could link p53 to the fected with the pM5neo (Laker et al., 1987) and the pHyg- cytoskeleton. However, tsp53 can be retained in the vector (Sugden et al., 1985) respectively. The cells were cytoplasm of IF-positive cells even when present in its exposed to one electric pulse of 260 V and 1500 mF in an properly folded wild-type conformation, to which Easyject plus genepulser (EuroGentec; Seraig, Belgium). hsc70 does not bind (Knippschild et al., 1996). In After electroporation the cells were immediately diluted in Cytoplasmic retention of tsp53 O Klotzsche et al 3433 prewarmed medium. Viable cells were seeded in culture Cytoskeletons were blocked with normal goat serum for dishes (1 10 cm) at 103, 104 and 105 per plate. The viability 20 min and labeled with antibodies to vimentin (V9) or p53 of the cells was determined by staining with trypan blue. (either mouse monoclonal antibody 248 or rabbit anti- The selection procedure was started 48 h after electropora- serum) for 20 min at room temperature. Goat anti-mouse tion by adding Geneticin (400 mg/ml) or Hygromycin B IgG conjugated with 10 nm gold particles (Biocell, UK) (100 mg/ml). Subclones isolated were tested for the and goat anti-rabbit IgG conjugated with 5 nm gold expression of vimentin and tsp53 by indirect immuno- particles (Biocell, UK) were used as secondary antibodies. ¯uorescence. The cells were post®xed with 0.5% glutaraldehyde in PBS for 1 h, dehydrated through a graded series of ethanol and critical point dried with CO . The dried cytoskeletons were Transient transfection 2 shadowed with pure carbon at a 908 angle at 1077 mbar in Cells grown to 50% con¯uence on glass coverslips were a Biotech 2000 (Leybold-Haereus, Germany). They were transiently transfected with the expression vector D61 ¯oated on 35% hydro¯uoric acid, washed on distilled coding for an aminoterminal mutated vimentin protein. water and mounted on copper grids without a supporting Two mg of plasmid DNA were diluted in 100 ml DMEM ®lm. Micrographs were taken in a Philips CM 120 without FCS and incubated with 10 ml Superfect (Qiagen, transmission microscope. Germany). Cells grown on cover slips were incubated for 2 h with the DNA-Superfect complexes in 700 ml DMEM. Fractionation of cells The transfection medium was removed and the cells were allowed to express the aminoterminal mutated vimentin for Subcon¯uent cells were washed twice with PBS and lysed 24 h. Vimentin and p53 were localized in the ®xed cells by in 2 ml lysis bu€er I (10 mM HEPES pH 6.6; 10% indirect immuno¯uorescence. Glycerol; 5 mM MgCl2; 1% NP40; 1 mM EGTA; 1% aprotinin, 50 mM Leupeptin). The soluble fraction was removed and the insoluble material was sonicated in 2 ml Preparation of cytoskeletons lysis bu€er. The proteins were precipitated with 10% TCA, The cells were grown on glass coverslips to 50% washed with acetone and dissolved in sample-bu€er con¯uence. The coverslips were rinsed brie¯y with PBS (Laemmli, 1970). They were separated on a 10% and lysed with Triton X-100 in a PIPES bu€ered salt polyacrylamide gel, transferred to nitrocellulose and solution (80 mM PIPES pH 6.8; 4% PEG 6000; 1 mM detected by Western immunoblotting.

MgCl2; 0.5% Triton X-100; 4 mM EGTA; 1% aprotinin (Trasylol: Bayer; Leverkusen, Germany); 50 mM Leupep- Western immunoblotting tin) (Okabe and Hirokawa, 1988). The cytoskeletons were washed with 25 mM HEPES pH 7.4; 29 mM NaCl; 100 mM About 26106 cells of a subcon¯uent culture were lysed

KCl; 0.96 mM NaH2PO4; 5 mM MgCl2 for 5 min and ®xed with 1 ml sample bu€er. 20 ml of each probe were run on a for 20 min with 2% paraformaldehyde. All steps were 10% polyacrylamide gel and blotted onto nitrocellulose as carried out at 378C in a volume of 100 ml. previously described (Staufenbiel and Deppert, 1983). The nitrocellulose sheets were blocked with 2% BSA in Tris- bu€ered salt solution (20 mM Tris/HCl pH 7.6; 150 mM Drug treatment of cells and immuno¯uorescence NaCl) containing 0.05% Tween. For detection of vimentin Cells were grown on glass coverslips to 50% con¯uence, the sheets were incubated with either the mouse anti- incubated with cytochalasin B (10 mg/ml) (Sigma, Munich, vimentin antibodies V9 or VIM3B4, and subsequently with Germany) for 15 min, or with vinblastin (Sigma) for 2 h, peroxidase conjugated donkey anti-mouse IgG (Biorad, or were left untreated. They were then rinsed with PBS, Munich, Germany). For detection of p53 the sheets were extracted, and ®xed with 1% paraformaldehyde for 15 min. incubated with the monoclonal antibody pAb421, secondly Alternatively they were immediately ®xed with acetone with a rabbit anti-mouse IgG (DAKO, Hamburg, (7208C) for 20 min. The ®xed cells were washed with PBS Germany), and then with a peroxidase conjugated mouse for 10 min. To minimize unspeci®c binding the samples antibody (DAKO, Hamburg, Germany). Bound peroxidase were saturated with 1% normal goat serum prior to was detected using the ECL Western-blot detection kit antibody incubation. The following probes were used for (Amersham; Braunschweig, Germany) and chemilumines- localizing cytoskeletal proteins and p53: a rabbit anti-p53 cence was visualized by exposing the membrane to X-ray antiserum, produced in our laboratory against puri®ed, ®lms. SDS denatured mouse wt p53 protein, the mouse monoclonal anti-p53 antibody pAb 248, the mouse monoclonal anti-vimentin antibodies V9 and VIM3B4 (Boehringer Mannheim, Mannheim, Germany), a mono- Acknowledgements clonal antibody to b-tubulin (Boehringer Mannheim), and The authors are grateful to M KuÈ hn and P Traub, Max- FITC-conjugated phalloidin (Sigma). Goat anti-mouse IgG Planck-Institut fuÈ r Zellbiologie, Ladenburg, for providing and goat anti-rabbit IgG (Dianova, Hamburg, Germany) the mouse vimentin expression vectors. This work was conjugated with FITC or Texas-Red were used as second supported by a grant from the Deutsche Krebshilfe, Dr antibodies. Incubation with antibodies was done for Mildred Scheel Stiftung, to WD, by ®nancial support from 30 min each, followed by washing with PBS for 20 min. Boehringer Mannheim GmbH, Germany, to OK, and the Photographs were taken in a Leica CLSM. Fonds der Chemischen Industrie. DE was a recipient of a fellowship from the GemeinnuÈ tzige Hertie-Stiftung, Frank- furt/Main. The Heinrich-Pette-Institut is ®nancially sup- Immunoelectron microscopy of cytoskeletons ported by Freie und Hansestadt Hamburg and Cells grown on glass coverslips were lysed with a nonionic Bundesministerium fuÈ r Gesundheit. This work is part of detergent as described above and ®xed with 1% the PhD thesis of OK, Fachbereich Biologie at the paraformaldehyde for 20 min at room temperature. University of Hamburg. Cytoplasmic retention of tsp53 O Klotzsche et al 3434 References

Albers K and Fuchs E. (1992). Int. Rev. Cytol., 134, 243 ± Lazarides E. (1980). Nature, 283, 249 ± 256. 279. Levine AJ. (1997). Cell, 88, 323 ± 331. Ali IU, Schweitzer JB, Ikejiri B, Saxena A, Robertson JT and Liao J, Lowthert LA, Ghori N and Omary MB. (1995). J. Old®eld EH. (1994). Cancer Res., 54, 1 ± 5. Biol. Chem., 270, 915 ± 922. Bader BL, Magin TM, Freudenmann M, Stumpp S and Martinez J, Georgo€ I, Martinez J and Levine AJ. (1991). Franke WW. (1991). J. Cell. Biol., 115, 1293 ± 1307. Genes Dev., 5, 151 ± 159. Bohn W, Roser K, Hohenberg H, Mannweiler K and Traub McCormick MB, Kouklis P, Syder A and Fuchs E. (1993). J. P. (1993). J. Struct. Biol., 111, 48 ± 58. Cell. Biol., 122, 395 ± 407. Bohn W, Wiegers W, Beuttenmuller M and Traub P. (1992). Medcalf EA, Takahashi T, Chiba I, Minna J and Milner J. Exp. Cell. Res., 201, 1 ± 7. (1992). Oncogene, 7, 71 ± 76. Cheng TJ and Lai YK. (1994). J. Cell. Biochem., 54, 100 ± Michalovitz D, Halevy O and Oren M. (1990). Cell, 62, 671 ± 109. 680. Cho Y, Gorina S, Je€rey PD and Pavletich NP. (1994). Middeler G, Zerf K, Jenovai S, Thulig A, Tschodrich-Rotter Science, 265, 346 ± 355. M, Kubitscheck U and Peters R. (1997). Oncogene, 14, Ciesielski-Treska J, Ulrich G and Aunis D. (1991). J. 1407 ± 1417. Neurosci. Res., 29, 362 ± 378. Moll UM, LaQuaglia M, Be nard J and Riou G. (1995). Proc. Deppert W. (1994a). Prospects in Diagnosis and Treatment of Natl. Acad. Sci. USA, 92, 4407 ± 4411. Breast Cancer. Schmidt N et al. (eds). Elsevier Publishers, Moll UM, Ostermeyer AG, Haladay R, Wink®eld B, Frazier pp. 11 ± 22. M and Zambetti G. (1996). Mol. Cell. Biol., 16, 1126 ± Deppert W. (1994b). Semin. Cancer Biol., 5, 187 ± 202. 1137. Deppert W. (1996). J. Cell. Biochem., 62, 172 ± 180. Moll UM, Riou G and Levine AJ. (1992). Proc. Natl. Acad. Dittmer D, Pati S, Zambetti G, Chu S, Teresky AK, Moore Sci. USA, 89, 7262 ± 7266. M, Finlay C and Levine AJ. (1993). Nat. Genet., 4, 42 ± 46. Mosner J, Mummenbrauer T, Bauer C, Sczakiel G, Grosse F Eckelt A, Herrmann H and Franke WW. (1992). Eur. J. Cell. and Deppert W. (1995). EMBO J., 14, 4442 ± 4449. Biol., 58, 319 ± 330. Nebe B, Bohn W, Pommerenke H and Rychly J. (1997). Eizenberg O, Faber-Elman A, Gottlieb E, Oren M, Rotter V Cytometry, 28, 66 ± 73. and Schwartz M. (1996). Mol. Cell. Biol., 16, 5178 ± 5185. Ogawara M, Inagaki N, Tsujimura K, Takai Y, Sekimata M, Eliyahu D, Michalovitz D and Oren M. (1985). Nature, 316, Ha MH, Imajoh-Ohmi S, Hirai S, Ohno S, et al. (1995). J. 158 ± 160. Cell. Biol., 131, 1055 ± 1066. Foisner R, Bohn W, Mannweiler K and Wiche G. (1995). J. Okabe S and Hirokawa N. (1988). J. Cell. Biol., 107, 651 ± Struct. Biol., 115, 304 ± 317. 664. Gannon JV and Lane DP. (1991). Nature, 349, 802 ± 806. Pinhasi-Kimhi O, Michalovitz D, Ben-Zeev A and Oren M. Gottlieb TM and Oren M. (1996). Biochim. Biophys. Acta, (1986). Nature, 320, 182 ± 184. 1287, 77 ± 102. Porter PL, Gown AM, Kramp SG and Coltrera MD. (1992). Hainhaut P, Soussi T, Shomer B, Hollstein M, Greenblatt M, Am. J. Pathol., 140, 145 ± 153. Hovig E, Harris CC and Montesano R. (1997). Nucleic RoÈ ser K, Bohn W, Giese G and Mannweiler K. (1991). Exp. Acids Res., 25, 151 ± 157. Cell. Res., 197, 200 ± 206. Heins S and Aebi U. (1994). Curr. Opin. Cell. Biol., 6, 25 ± 33. Rotter V, Abutbul H and Ben-Ze'ev A. (1983). EMBO J., 2, Hennekes H, Kuhn S and Traub P. (1990). Mol. Gen. Genet., 1041 ± 1047. 221, 33 ± 36. Shaulsky G, Ben-Ze'ev A and Rotter V. (1990a). Oncogene, Herrmann H, Hofmann I and Franke WW. (1992). J. Mol. 5, 1707 ± 1711. Biol., 223, 637 ± 650. Shaulsky G, Gold®nger N, Ben-Ze'ev A and Rotter V. Hinds PW, Finlay CA, Frey AB and Levine AJ. (1987). Mol. (1990b). Mol. Cell. Biol., 10, 6565 ± 6577. Cell. Biol., 7, 2863 ± 2869. Shaulsky G, Gold®nger N, Tosky MS, Levine AJ and Rotter Hollstein M, Shomer B, Greenblatt M, Soussi T, Hovig E, V. (1991). Oncogene, 6, 2055 ± 2065. Montesano R and Harris CC. (1996). Nucleic Acids Res., Staufenbiel M and Deppert W. (1983). Eur. J. Cell. Biol., 31, 24, 141 ± 146. 341 ± 348. Hollstein M, Sidransky D, Vogelstein B and Harris CC. Steinmeyer K and Deppert W. (1988). Oncogene, 3, 501 ± (1991). Science, 253, 49 ± 53. 507. Inagaki M, Gonda Y, Nishizawa K, Kitamura S, Sato C, Sugden B, Marsh K and Yates J. (1985). Mol. Cell. Biol., 5, Ando S, Tanabe K, Kikuchi K, Tsuiki S and Nishi Y. 410 ± 413. (1990). J. Biol. Chem., 265, 4722 ± 4729. Takahashi K and Suzuki Y. (1994). Oncogene, 9, 183 ± 188. Janmey PA, Euteneuer U, Traub P and Schliwa M. (1991). J. Traub P and Vorgias CE. (1983). J. Cell. Sci., 63, 43 ± 67. Cell. Biol., 113, 155 ± 160. Tronsco JC, March JL, Haner M and Aebi U. (1990). J. Knippschild U, Oren M and Deppert W. (1996). Oncogene, Struct. Biol., 103, 2 ± 12. 12, 1755 ± 1765. Zerrahn J, Deppert W, Weidemann D, Patschinsky T, Ko LJ and Prives C. (1996). Genes Dev., 10, 1054 ± 1072. Richards F and Milner J. (1992). Oncogene, 7, 1371 ± 1381. Laemlli UK. (1970). Nature, 227, 680 ± 685. Laker C, Stocking C, Bergholz U, Hess N, De Lamarter JF and Ostertag W. (1987). Proc. Natl. Acad. Sci. USA, 84, 8458 ± 8462.