Energy-Dependent Nucleolar Localization of P53 in Vitro Requires Two Discrete Regions Within the P53 Carboxyl Terminus

Oncogene (2007) 26, 3878–3891 & 2007 Nature Publishing Group All rights reserved 0950-9232/07 $30.00 www.nature.com/onc ORIGINAL ARTICLE Energy-dependent nucleolar localization of p53 in vitro requires two discrete regions within the p53 carboxyl terminus

O Karni-Schmidt1, A Friedler2,3, A Zupnick1, K McKinney1, M Mattia1, R Beckerman1, P Bouvet4, M Sheetz1, A Fersht2 and C Prives1

1Department of Biological Sciences, Columbia University, New York, USA; 2Department of MRC-CPE, Center for Protein Engineering, Medical Research Council, Cambridge, UK; 3Department of Organic Chemistry, The Hebrew University of Jerusalem, Givat Ram, Jerusalem, Israel; 4Laboratoire de Biologie Mole´culaire de la Cellule/UMR 5161, Ecole Normale Supe´rieure de Lyon 46, Alle´e d’Italie 69364 Lyon Cedex, France

The p53 tumor suppressor is a nucleocytoplasmic shuttling concentration, is presumably inactive as a transcription protein that is found predominantly in the nucleus of cells. factor and is diffusely distributed throughout the cell. In In addition to mutation, abnormal p53 cellular localiza- response to DNA damage or other forms of cellular tion is one of the mechanisms that inactivate p53 function. stress, p53 protein accumulates, frequently becomes To further understand features of p53 that contribute to extensively post-translationally modified and concen- the regulation of its trafficking within the cell, we analysed trates within the nucleus (Giaccia and Kastan, 1998; the subnuclear localization of wild-type and mutant p53 in Appella and Anderson, 2001; Liang and Clarke, 2001; human cells that were either permeabilized with detergent Prives and Manley, 2001). or treated with the proteasome inhibitor MG132. We, One property of the nucleus, analogous to cytoplas- here, showthat either endogenously expressed or exoge- mic organelles, is the existence of subnuclear domains. nously added p53 protein localizes to the nucleolus in These nuclear compartments concentrate specific sets detergent-permeabilized cells in a concentration- and ATP of functionally related proteins. Within the nucleus, hydrolysis-dependent manner. Two discrete regions within the nucleolus is the site of ribosomal RNA (rRNA) the carboxyl terminus of p53 are essential for nucleolar transcription, modification, maturation and ribosomal localization in permeabilized cells. Similarly, localization assembly (Carmo-Fonseca et al., 2000; Olson et al., of p53 to the nucleolus after proteasome inhibition in 2000; Leung et al., 2003). Although easily visualized unpermeabilized cells requires sequences within the as a discrete subnuclear entity, the nucleolus is not carboxyl terminus of p53. Interestingly, genotoxic stress separated from the surrounding nucleoplasm by markedly decreases the association of p53 with the membranes, and thus proteins localize to nucleoli by nucleolus, and phosphorylation of p53 at S392, a site processes that do not involve active membrane trans- that is modified by such stress, partially impairs its port. In recent years, multiple proteins involved in nucleolar localization. The possible significance of these cell-cycle progression were identified, whose regulation findings is discussed. involves sequestration within the nucleolus. In response Oncogene (2007) 26, 3878–3891. doi:10.1038/sj.onc.1210162; to oncogenic stress, the p14ARF protein may inactivate published online 22 January 2007 the mouse double minute (MDM)2 oncoprotein (an inhibitor of p53) by inducing nucleolar sequestration Keywords: p53; nucleolus; DNA damage; NoLS of MDM2 (Zhang et al., 1998; Tao and Levine, 1999; Weber et al., 1999; Ashcroft et al., 2000). The requirement of MDM2 localization to nucleoli for its inactivation by p14ARF, however, is not obliga- tory (Llanos et al., 2001). Among other cell cycle Introduction regulators and checkpoint factors that associate with the nucleolus are the Werner’s Syndrome Helicase (Marci- The human p53 protein contains 393 amino acids and is niak et al., 1998), the telomere maintenance enzyme comprised of several functional domains (Prives and (hTERT) (Wong et al., 2002) and Cdc14, a protein Hall, 1999). Appropriate subcellular localization is phosphatase in yeast crucial for promoting exit from crucial for regulating p53 function. Under unstressed mitosis Visintin et al., 1998). These and other findings conditions, the p53 protein is generally present at a low imply a novel role for the nucleolus’s participation in the regulation of cell cycle proteins (as reviewed in Visintin and Amon, 2000; Lamond and Sleeman, 2003; Leung Correspondence: Dr C Prives, Department of Biological Sciences, and Lamond, 2003; Mayer and Grummt, 2005; Olson Columbia University, 816 Fairchild, NY 10027 USA. E-mail: [email protected] and Dundr, 2005). Received 18 July 2006; revised 22 September 2006; accepted 23 October Within the nucleus, p53 has been reported to be 2006; published online 22 January 2007 associated with at least three subnuclear compartments ATP-dependent p53 nucleolar localization O Karni-Schmidt et al 3879 including the nucleolus, promyelocytic leukemia bodies loss of p53 upon permeabilization may reflect the release and cajal bodies (Mogaki et al., 1993; Benninghoff et al., of a subpopulation of p53 that is not bound to nuclear 1999; Rubbi and Milner, 2000; Horky et al., 2001; structures. Klibanov et al., 2001; Wesierska-Gadek et al., 2002; Parallel cultures to those in Figure 1a were either Young et al., 2002; Zimber et al., 2004). Approximately directly fixed with 4% paraformaldehyde (unpermeabi- half of nuclear p53 is soluble, whereas the remainder lized) or first permeabilized with PBS/1% Triton X-100 associates with nuclear structures (Zerrahn et al., 1992). for 2 min before fixation. Cells were then incubated p53 can interact with a number of nucleolar components with anti-p53 antibodies followed by incubation with such as L5 ribosomal protein (Marechal et al., 1994), secondary antibodies and analysis by laser scanning topoisomerase I (Gobert et al., 1996), nucleolin (Daniely microscopy and differential interference contrast (DIC). et al., 2002), nucleophosmin (Colombo et al., 2002) and As expected, without permeabilization wild-type and nucleostemin (Tsai and McKay, 2002). Intriguingly as mutant p53 proteins were detected in the nucleoplasm of well, the first cellular DNA-binding site of p53 to be unpermeabilized cells and appeared to be excluded from identified (RGC) is located within the ribosomal gene nucleoli (Figure 1b). Interestingly, after permeabiliza- cluster (Kern et al., 1991). Recently, it was reported that tion, nucleoplasmic staining was generally weaker in mutant p53 localizes to nucleoli in cells treated with a wild-type p53 cells, whereas in approximately 10–15% compound (PRIMA-1) that was reported to reactivate of cells expressing p53Q22/S23 and p53H175 almost exclusive mutant forms of p53 (Bykov et al., 2002a, b; Rokaeus nucleolar staining was observed (Figure 1c). By con- et al., 2006). trast, wild-type p53 was never observed in the nucleoli. Under non-stressed conditions, p53 was reported to Co-staining cells with anti-nucleolin antibody confirmed localize to the nucleolus in permeabilized Hep G2 cells the nucleolar localization of mutant p53. Our observa- (Rubbi and Milner, 2000). The nuclear and nucleolar tions suggest that some of the p53 in the nucleus is not association of p53 in such cells is RNA-dependent as tightly attached to nuclear components and can freely RNase A treatment of permeabilized cells abrogates diffuse outside the nucleus, whereas a subpopulation of detection of most of the anchored p53. In addition, p53 in some cells can be anchored by attaching to treatment of human fibroblasts with MG132, a protea- nuclear and nucleolar components as was reported some inhibitor, causes p53 to accumulate in the previously (Rubbi and Milner, 2000; Klibanov et al., nucleolus and nucleoplasm both in directly fixed cells 2001). and in permeabilized cells (Klibanov et al., 2001; Both mutant p53Q22/S23- and p53H175-expressing Pokrovskaja et al., 2001; Latonen et al., 2003). Although cells contain at least two-fold more p53 than those its functional significance is unknown, the nucleolar expressing wild-type p53 when normalized to cellular sequestration of p53 was reported to be one of the actin levels (Figure 1 and 2). We therefore tested mechanisms contributing to the reactivation of p53 in whether the level of p53 protein can determine HeLa cells after treatment with cisplatin (Horky et al., sequestration of the mutant p53 into the nucleolus 2001; Wesierska-Gadek et al., 2002). by varying the amount of tetracycline in the culture In the present study, we have established conditions medium of mutant p53-expressing cells (Chen for characterizing the association of endogenously et al., 1996). Indeed, when the level of mutant p53 was expressed or exogenously added p53 with nucleoli and lowered to that of fully expressed wild-type p53 neither have identified some of the requirements for this p53Q22/S23 nor p53H175 relocalized to the nucleolus process. (Figure 2). To determine whether endogenously expressed p53 in other cell lines could relocalize to nucleoli after permeabilization, we examined HT-29 cells RESULTS and T98G cells that each express high levels of mutant forms of p53. Note that the levels of endogenous The level of p53 within permeabilized H1299 cells mutant p53 in those cells are similar to those of p53H175 determines nucleolar sequestration that are expressed inducibly in H1299 cell line To investigate the localization of soluble and anchored Chen et al., 1996; Liu and Chen, 2002; Menendez p53 proteins within the nucleus, we used H1299 cells et al., 2006). In fact, after permeabilization of both cell expressing tetracycline-regulated wild-type p53, tran- lines, nucleolar staining of p53 was readily detectable scriptionally impaired p53Q22/S23 and a tumor-derived and predominant (Figure 1d and e). Although we mutant p53H175 (Figure 1a). Forty-eight hours after were concerned that only mutant forms of p53 can removal of tetracycline, cells were either immediately associate with nucleoli when at high concentrations lysed in electrophoresis sample buffer (unpermeabilized, in cells, transfected wild-type p53 protein, which is U) or first pre-permeabilized (P) in phosphate-buffered present at markedly higher levels than endogenously saline (PBS)/1% Triton X-100 for 2 min before lysis to expressed p53 protein, can also be found in nucleoli remove soluble proteins. The cells were then subjected to of permeabilized cells to roughly similar extents as sodium dodecyl sulfate polyacrylamide gel electropho- the mutant forms of p53 shown in Figures 1 and 2 (data resis (SDS–PAGE) and immunoblotting. Although not shown). We conclude from these data that the level levels of cellular proteins including p53 were reduced of p53 in permeabilized H1299 inducible cells is likely after permeabilization in the three clones, the remaining to be a critical feature in determining its intranuclear p53 protein was still readily detected (Figure 1a). The localization.

Oncogene ATP-dependent p53 nucleolar localization O Karni-Schmidt et al 3880

a Wild type p53 p53H175 p53Q22/S23

P53– P53+P53– P53+ P53– P53+

U P UPUP UPUP U P 62 51 p53 38 actin

bc Wild type p53

p53H175

p53Q22/S23

DIC p53merge p53 nucleolin merge

d HT-29, p53H273 e T98G, p531237

p53 nucleolin merge p53 nucleolin merge Figure 1 Mutant p53 can associate with nucleoli in human cells. (a) H1299 cells expressing inducible wild-type p53, p53H175 or p53 Q22/ S23 were extensively washed and then maintained in tetracycline-free medium and serum. After 48 h, cells were either directly solubilized in SDS electrophoresis sample buffer (unpermeabilized, U) or first treated with 1 ml PBS/1% Triton X-100 for 2 min before suspension in SDS electrophoresis sample buffer (permeabilized, P). Samples were then resolved by SDS–PAGE and subjected to immunoblotting with anti-p53 antibody PAb 1801. As control blots were probed for actin. (b and c) H1299 cells expressing induced wild-type p53, p53H175 and p53Q22/S23 as in A were either directly paraformaldehyde fixed without permeabilization (b) or permeabilized with 1 ml PBS/ 1% Triton X-100 for 2 min before fixing (c). (d and e) HT-29 and T98G cells were permeabilized with PBS/1% Triton X-100 containing buffer and then fixed. Immunostaining for p53 was performed using PAb 1801 and Cy5 coupled goat anti-mouse IgG antibody. Permeabilized cells in (c–e) were also co-stained with nucleolin antibody to visualize the nucleolus. The cells were visualized by confocal laser scanning microscopy and DIC.

Exogenously added p53 proteins localize to nucleoli in derived hemagglutinin (HA)-tagged wild-type p53, or permeabilized cells baculovirus-derived untagged purified p53H175, (Supple- To gain further insight into the features of p53 that mentary Figure 1d–f), strong nucleolar localization of allow it to localize to nucleoli, we established an in vitro p53 was evident in approximately 20% of the cells with localization assay as reported previously for MDM2 about 60% showing nucleoplasmic localization. that entailed having purified p53 present in the buffer Further, p53 nucleolar association was detected by a used to permeabilize cells (Poyurovsky et al., 2003). variety of antibodies (PAb 1801, DO-1, PAb 421, anti- As seen with cellular p53, localization of exogenous GST, anti-p53 polyclonal antibody and anti-OMNI) p53 protein in permeabilized cells was followed by that were used to recognize p53 or p53 fusion proteins immunofluorence and DIC (Figure 3 and Supplemen- demonstrating that epitope tags or GST fusion did not tary Figure 1). The results showed that when H1299 cells affect p53 nucleolar localization (Figure 3 and Supple- were permeabilized in the presence of either bacterially mentary Figure 1). expressed glutathione S-transferase (GST)-tagged p53 Several controls were performed to support the (Figure 3c, panel 1) or his-tagged p53, baculovirus- validity of these results. First, we fixed the cells before

Oncogene ATP-dependent p53 nucleolar localization O Karni-Schmidt et al 3881

a WT p53 p53H175 p53Q22/S23

Tet (ng/ml) : + ––51020253040+– 5 1020253040+ 62

p53 51

actin 38

b 1. WT p53, 0 ng/ml tet 2. p53H175, 20 ng/ml tet 3. p53Q22/S23, 20 ng/ml tet

SYTOX GREEN p53 4. p53H175, 0 ng/ml tet 5. p53Q22/S23, 0 ng/ml tet

SYTOX GREEN p53 SYTOX GREEN p53 Figure 2 The level of p53 protein determines relocalization of p53 within nucleoli. (a) H1299 cells expressing wild-type p53, p53H175 and p53 Q22/S23 were maintained in 4500 ng/ml tetracycline ( þ ) or were switched to medium containing 40, 30, 25, 20, 10, 5 and 0 (À) ng/ ml of tetracycline for 48 h as indicated. Cells were lysed directly in electrophoresis sample buffer and subjected to immunoblotting with p53 monoclonal PAb 1801 and actin polyclonal antibody. Arrows indicate estimated equivalent levels of protein in mutant and wild- type p53 cell lines. (b) Parallel cultures of H1299 cells expressing wild-type p53 or p53H175 and p53 Q22/S23 as in (a) were placed in medium containing 20 and 0 ng/ml tetracycline as indicated for 48 h and then either directly ﬁxed, unpermeabilized (U) or ﬁrst permeabilized (P) as in Figure 1. Immunostaining for p53 was performed using PAb 1801 and Cy5 coupled goat anti-mouse IgG antibodies. Cells were co-stained with 10 nM SYTOX green nucleic acid stain to localize the nuclei.

adding the purified p53 protein. In this case, no and exogenously added p53 protein can translocate to p53 association with nuclei or nucleoli was and associate with the nucleolus in Triton-permeabilized observed (Supplementary Figure 1E). Second, several cells. truncated versions of p53 failed to localize to nucleoli The ability to demonstrate that exogenously added (Figure 3). Perhaps most relevantly, as the basic p53 purified wild-type p53 localizes to the nucleolus allowed C-terminus associates efficiently with nucleoli (Figure 3, us to identify potential region(s) of p53 responsible for panel 5), a highly basic deletion mutant of high- this property. To address this question, we included mobility-group (HMG)B1, HMG Box B domain (amino different versions of purified p53 protein in the acids 90–176, K96E) (Stros and Reich, 1998), failed to permeabilization buffer for H1299 cells and visualized localize to nucleoli even when added at higher concen- nuclei as above (Figure 3). A schematic diagram and tration than p53 protein (Supplementary Figure silver-stained gel of all purified forms of p53 proteins 1F). This result argues against a possible effect of used in these experiments are shown in Figure 3a and b. nonspecific charge on translocation of p53 to the Although the sources and tags of the various p53 nucleolus. proteins differed, we believe that this does not affect the Importantly, these results were not confined to interpretation of our data as different tagged versions of H1299 cells as, when we included purified p53 protein purified full-length p53 proteins localized similarly to in the permeabilization buffer to either IMR-90 nucleoli. cells that, when unstressed, express low levels of wild- Although, as mentioned above, GST-p53 was capable type p53 or to p53-null Saos-2 cells, it was possible to of associating with nucleoli (Figure 3c, panel 1), both detect p53 that was either exclusively or predominantly GST-p53 N-terminal domain (amino acids 1–82) and nucleolar (Supplementary Figure 2). Taken together, the his-tagged p53 core domain (amino acids 97–303) our results indicate that both endogenously expressed showed exclusively nucleoplasmic staining (Figure 3c,

Oncogene ATP-dependent p53 nucleolar localization O Karni-Schmidt et al 3882

a p53 Variants Nucleolar Localization c

TAD PRO DNA binding core L Tet B Wild type p53 N C + 1 20 60 90 100 300 325 355 363 393 1. GST-p53 arg→his p53 H175 + 175

his-p53C- terminus + 311 393

GST-p53N- - terminus 1 82 2. GST-p53N- his-p53 core - terminus 97 303

his-p53 core + + tetramerization 97 363

HA-p53∆C30 - 1 363

b us 3. his-p53 core minus ermin C30 ization∆ er H175

his-p53 core + HA-p53 his-p53 p53 his p53C-GST-p53 ter GST-p53N-t his-p53 core+tetramHA-p53 1 234 56789

84 4. his-p53 core + 62 tetramerization 51

26 5. his p53C- terminus

6. HA-p53 ∆30

p53 nucleolin merge Figure 3 The C-terminal portion of p53 can associate with nucleoli. Schematic diagram (a) and silver stained gel (b) of puriﬁed wild- type and mutant p53 proteins. (c) H1299 cells were permeabilized before ﬁxation for 2 min. The permeabilizing buffer was supplemented with (1) 0.4 mM full-length bacterial GST-p53, (2) 1.6 mM bacterially expressed GST-p53 N-terminus (amino acids 1–82), (3) 0.8 mM bacterially expressed his-p53 core domain (amino acids 97–303), (4) 0.4 mM bacterially expressed his-p53 core, linker and tetramerization domain (amino acids 97–363), (5) 0.4 mM baculovirally derived his-p53 C-terminus (amino acids 311–393) and (6) 0.4 mM baculovirally derived C-terminally deleted p53 (HA-p53DC30; amino acids 1–362). The p53 proteins were detected with PAb 1801 (1, 2 and 6), PAb 421 (5) or OMNI antibody (3 and 4).

panels 2 and 3, respectively). On the other hand, N- and somewhat paradoxical observation that the core/linker/ C-terminally deleted p53, comprising the central core, tetra construct that lacks the C-terminus was capable of linker and tetramerization domains (amino acids 97– concentrating in nucleoli, whereas HA-p53DC30 p53 363; core/linker/tetra), was detected both in the nucleo- could not. Possibly there are multiple determinants plasm and nucleoli (Figure 3c, panel 4). This was of p53 nucleolar association, one within the extreme consistent with our ﬁnding that the C-terminal portion C-terminal basic region, whereas the other(s) would of p53 (amino acids 311–393; linker/tetra/basic) loca- reside in the linker or tetramerization regions (because lized exclusively and very efﬁciently to the nucleolus the core alone did not localize to nucleoli). Data (Figure 3c, panel 5). These data suggested that one or supporting this likelihood are shown below. We spec- more determinant(s) present in the linker or tetrameri- ulate that the p53 N-terminus may be inhibitory in cis to zation regions are necessary for p53 to associate with the nucleolar localization determinant(s) within the nucleoli. It was therefore surprising that HA-p53DC30 linker or tetramerization region. Altogether, our data protein lacking the C-terminal basic domain (amino strongly suggest that the carboxyl terminus of p53 acids 1–362; N/core/linker/tetra) was not concentrated (amino acids 311–393) possesses sequences that are in nucleoli (Figure 3, panel 6). We cannot explain the essential for its nucleolar association.

Oncogene ATP-dependent p53 nucleolar localization O Karni-Schmidt et al 3883 Fluorescein-labeled peptides reveal that two discrete difference in gross morphology in cells that differed regions of p53 within its carboxyl terminus can in their ability to take up peptides. Both Fl-305–322 associate with nucleoli in both permeabilized and and Fl-367–393 peptides displayed strong nucleolar unpermeabilized cells staining, whereas Fl-1–42, Fl-325–355 and Fl-367– To focus more precisely on regions that are required 393–392P showed no nuclear or nucleolar staining for p53 nucleolar localization, fluorescein (Fl)-labeled (Figure 4c). Further, we also tested an NES mutant peptides were synthesized spanning the linker region version of Fl-325–355 (L348A/L350A) peptide, which (305–322), tetramerization domain (325–355) and basic showed readily detectable nucleoplasmic staining, but region (367–393), as well as the N-terminus (1–42). This did not associate with nucleoli, showing that not every approach has the added advantage that it is possible to peptide that enters the nucleus concentrates in nucleoli synthesize peptides fully phosphorylated or acetylated at (Figure 4c, panel 5). We cannot exclude that those residues within p53 known to be modified in vivo after peptides that failed to show nuclear or nucleolar loca- stress signaling. The subcellular localization of these lization were not taken up by cells or were degraded peptides in permeabilized cells was examined by more rapidly before fixation. Nevertheless, the results confocal microscopy (Figure 4). Consistent with the that peptides spanning 305–322 and 367–393 associate fact that GST-p53 N-terminus (1–82) fails to associate with nucleoli provide strong support that conditions with nucleoli, we observed no nucleolar localization used to permeabilize cells are not providing artifactual with the Fl-1–42 peptide (Figure 4a, panel 1). In fact, results. They also suggest the possibility that one or nucleolar localization of the Fl-1–42 peptide was not more modifications within these sequences may have an observed even when it was added at concentrations up impact on such localization. to 20 mM that is, five-fold greater than shown in Figure 4 (data not shown). Additionally, we did not observe nucleolar localization of the Fl-325–355 peptide, which Nucleolar localization of p53 after proteasome contains the tetramerization domain as well as a inhibition in unpermeabilized cells requires the p53 putative nuclear export signal (NES) (Figure 4a, panel carboxyl terminus 4). On the other hand, C-terminally derived peptides It was important to show that the C-terminus of p53 is from two discrete regions displayed very strong nucleo- required for nucleolar localization in unpermeabilized lar localization. The Fl-305–322 peptide was detected cells. Previous reports showed that p53 localizes to prominently in nucleoli as well as more weakly in the nucleoli after treatment of cells with the proteasome nucleoplasm (Figure 4a, panel 2). As the same peptide inhibitor MG132 (Klibanov et al., 2001; Latonen et al., when fully phosphorylated on serine 315, showed 2003). Here, we have confirmed and extended these essentially the same result, which indicates that the findings by identifying regions and sites that affect p53 modification of this residue is not likely to impact localization under these conditions. In H1299 p53-null nucleolar localization (Figure 4a, panels 2 and 3). cells transfected with wild-type p53, MG132 treatment Additionally peptides Fl-363–393, Fl-363–393 with led to p53 nucleolar localization in approximately 20% acetylated lysine 382 and Fl-363–393 with phosphory- of the cells (Figure 5b and c, panel 1). To gain lated S392 were clearly associated with nucleoli and information about which regions of p53 were involved, more faintly with nucleoplasm, although the relative we used a number of constructs in which either the intensity of their fluorescein signals was weaker than C-terminus was deleted (p53DC30) or lysine residues those seen with Fl-305–322 (Figure 4a, panels 5–7). within the C-terminal 30 amino acids affecting nuclear Interestingly, phosphorylation at serine 392 but not localization signal (NLS) sequences NLS II and NLS III acetylation at lysine 382 exerted a partial inhibitory were mutated to either arginine (p536KR) or glutamine effect on nucleolar localization, which was confirmed by (p536KQ), as well as constructs in which lysines in using Metamorph software (Dawningtown, PA, USA) both the linker region and the carboxyl terminus to quantify the intensity of nucleolar localization of 30 were mutated to either arginine (p538KR) or glutamine representative cells (Figure 4b). Our data thus support (p538KQ) (see Figure 5a). We also generated a construct the likelihood that p53 contains within its C-terminus p538KQ-SVNLS in which the SV40 NLS was fused to two discrete sequences that can facilitate its association the carboxyl terminus of this mutant (Figure 5a). In with nucleoli. transfected H1299 cells p536KR,p536KQ and p538KR Small peptides have the unique advantage that they localized to the nucleolus after MG132 treatment to can in some cases be taken up by intact cells (Issaeva roughly the same extent as wild-type p53 (Figure 5b). By et al., 2003). This allowed us to determine whether any contrast, neither transfected p53DC30 nor p538KQ were of the peptides could associate with the nucleoli in detected in nucleoli (Figure 5b). p538KQ was prima- cells without prior permeabilization. Indeed, when the rily cytoplasmic with weak nucleoplasmic staining peptides described above were added directly to cell (Figure 5c, panel 3), which could explain its failure to asso- cultures before fixation with paraformaldehyde, our ciate with nucleoli. When we transfected p538KQ-SVNLS results approximated what we had seen with permeabi- to the cells, however, p53 was detected exclusively lized cells. Note that, in contrast to permeabilized cells in the nucleoplasm demonstrating the effectiveness of where nucleolar peptides were detected in nearly all of the SV40 NLS. It was therefore striking that the accu- the cells, only about 5% of the intact cells allowed mulation of p538KQ-SVNLS to the nucleolus after pro- visualization of peptides. We did not observe any teasome inhibition was markedly decreased (Figure 5b

Oncogene ATP-dependent p53 nucleolar localization O Karni-Schmidt et al 3884 a 1) FI- 1-42 5) FI- 367-393

2) FI- 305-322 6) FI- 367-393 382Ac

3) FI- 305-322 7) FI- 367-393 315P 392P

DIC p53 4) FI- 325-355

DIC p53

18000 b 14903 16000 14254 1-42 14000 305-322 12000 305-322, 315P 10000 6801 6295 325-355 8000 367-393 6000 3305 3178 2879 367-393, 382Ac Fluorescent signal 4000 367-393, 392P 2000 0

c 1) FI- 1-42 5) FI-325-355 A348/A350

2) FI- 305-322 6) FI- 367-393

3) FI- 305-322 7) FI- 367-393 315P 382Ac

4) FI- 325-355 7) FI- 367-393 392P

DIC p53 DIC p53

Figure 4 Subcellular localization of fluorescein-labeled peptides in H1299 p53-null cells. (a) Fluorescein-labeled 4 mM peptides: 1–42, 305–322, 305–322 (315P), 325–355, 367–393, 367–393 (382Ac) and 367–393 (392P) were added to H1299 cells in Triton X-100/PBS buffer and then cells were subjected to immunofluorescence analysis. (b) Fluorescent signal intensity of the peptides using the Metamorph software program. (c) Fluorescein-labeled 20 mM p53 peptides (1–42, 305–322, 305–322 (315P), 325–355, 325–355 (L348A/ L350A), 367–393, 367–393 (382Ac) and 367–393 (392P)) in PBS buffer were incubated for 3 min with H1299 cells followed by directly fixing cells and immunofluorescence analysis as described in Figure 1.

and c, panel 4). This supports the likelihood that p53 has We also compared Flag-tagged versions of wild- bona ﬁde nucleolar localization signal sequences rather type p53, a neutral S392 substitution mutant p53S392A than its nucleolar association being an indirect result of and a phosphorylation mimicking mutant p53S392D. its being present in the nucleus. Importantly, after treatment of cells with MG132,

Oncogene ATP-dependent p53 nucleolar localization O Karni-Schmidt et al 3885 a p53 Variants TAD PRO DNA binding core L Tet B WT p53 N C 1 20 60 90 100 300 325 355 363 393 p53∆C30 N 363

6KR or 6KQ N C

8KR, 8KQ N

8KQ-SVNLS N

S392A/S392D N

b 35 30

5 % p53 Nucleolar Localization 0 WTp53p53∆C30 6KR 6KQ8KR 8KQ8KQ-SVNLS S392A S392D c

WT 1.p53 2.p53 8KR

+MG 132 +MG 132

3.p53 8KQ 4. p53 8KQ-SVNLS

+MG 132 +MG 132

p53 nucleolin merge p53 nucleolin merge

Figure 5 Nucleolar localization of p53 after proteasome inhibition requires sequences within the p53 carboxyl terminus. (a) Diagram of p53 constructs used in the transfection experiments. p536KR and p538KR refer to substitution mutants in which the indicated lysine residues were changed to arginine; p536KQ or p538KQ refer to mutants in which these lysines were changed to glutamine. p538KQ-SVNLS refer to the substitution mutant 8KQ in which the SV-40 NLS was added to its carboxyl terminus. p53S392A/S392D refer to mutants in which the amino acid serine was substituted to either alanine or aspartate, respectively. In each case H1299 p53-null cells were transfected with different versions of p53 (1 mg of plasmid in each case) and treated with 10 mM MG132 for 8 h as described in Materials and methods. Cells were directly ﬁxed and analysed as detailed in Figure 1. (b) Percent of cells with nucleolar p53 was estimated by counting at least 200 cells in two separate experiments. (c) Representative images of cells transfected with constructs described in (a) with or without treatment with MG132. p53S392A localized to the nucleolus with similar efﬁciency Taken together, these data (1) validate our to wild-type p53 (Figure 5b), whereas there was a 40– experiments with p53 proteins’ and peptides’ localiza- 50% drop in the number of cells with nucleolar p53S392D tion to nucleoli in permeabilized cells; (2) show that (Figure 5b). This is consistent with our observation of the carboxyl terminus of p53 plays an important an approximately 50% reduction in the accumulation of role in redirecting p53 to the nucleolus after protea- the peptide with phosphorylated S392. some inhibition and (3) provide impetus for our

Oncogene ATP-dependent p53 nucleolar localization O Karni-Schmidt et al 3886 subsequent experiments described below examining the to substantially reduce the levels of ATP in the cells effect of genotoxic stress on p53 nucleolar localization. (Figure 6). When cells were first treated with 2- deoxyglucose and sodium azide (Az/DOG), almost no p53 could be detected in the nucleolus although it was ATP hydrolysis is required for p53 localization to readily detected in the nucleoplasm. Importantly, the the nucleolus effect of prior Az/DOG treatment was reversible, as We considered the possibility that p53 nucleolar adding ATP to the permeabilization buffer along localization might be energy-dependent. To determine with the purified p53 protein resulted in a significant whether there is a requirement for cellular adenosine (approximately six-fold) increase in the percent of triphosphate (ATP) for p53 nucleolar localization before cells in which p53 protein localized to the nucleolus permeabilization, cells were pre-treated with a combina- (Figure 6a, panel 3 and see graph at right). Weak tion of sodium azide (which inhibits cytochrome nucleoplasmic staining was also always observed in oxidase) and 2-deoxyglucose (which inhibits glycolysis) this case.

a 21.75 1.

13.25

2.azide+ deoxy 1.75 glucose % p53 Nucleolar Localization

3. azide+ + + azide + deoxy deoxyglucose glucose +ATP + ATP

nucleolin p53 merge

b 25.25

1. 20

4.75 5.75 2. P 1.5 % p53 Nucleolar Localization

+ p53 3. P + P, p53 +ATP + P+ ATP, p53 + P+ ADP, p53 + P+ ATPS, p53

4. P +ADP

5. P +ATPS

nucleolin p53 merge Figure 6 ATP hydrolysis is required for nucleolar localization of p53. (a) H1299 p53 null cells were either untreated or pre-treated with 5 mM sodium azide and 1 mM 2-deoxyglucose for 45 min (upper left panels) before permeabilization with PBS/1% Triton X-100 containing purified bacterial wild-type his-p53 protein with or without addition of 4 mM ATP. Cells were then fixed and stained as in Figure 3. Bar graph on right shows percentage of cells with nucleolar localization of his-p53 protein after the indicated conditions. The graph represents two independent experiments where at least 200 cells were counted for each slide. (b) Purified bacterial wild-type his- p53 protein was either present in the PBS/1% Triton X-100 permeabilization buffer (panels 1), or added 3 min after the cells were incubated with PBS/1% Triton X-100 supplemented with 4 mM ATP (panels 3), 4 mM ADP (panels 4) or 4 mM ATPgS (panels 5). Cells were then fixed and stained as in Figure 2. The graph at right shows the percentage of cells with nucleolar localization of exogenous purified p53 protein after the different treatments. The graph was prepared as in (a). Immunostaining for p53 was performed as described in Figure 1.

Oncogene ATP-dependent p53 nucleolar localization O Karni-Schmidt et al 3887

To test whether ATP hydrolysis is a requirement for 30 Control 20.3 nucleolar localization, cells were pre-permeabilized (P) 25 Daunorubicin with PBS/1% Triton X-100 for 3 min before addition of UV 20 p53 protein (0.4 mM) in the same buffer for an additional Actinomycin D 3 min (Figure 6b). Without added ATP p53 was 15 9.25 primarily detected in the nucleoplasm, whereas, as 5.93 10 5 before, when cells were permeabilized directly with 2.83 5 0.5 1 buffer containing p53 protein, nucleolar staining was 0.16 0.3 0.16 0.3 0 evident (Figure 6b, panels 1 and 2). Similar experiments 0 % p53 Nucleolar Localization were performed in which the cells were incubated with –5 Exogenous H1299 T98G the pre-permeabilization buffer containing 4 mM adeno- wild type p53 p53H175 p531237 sine diphosphate (ADP) and ATPgS for 3 min followed by addition of buffer containing purified his-p53 Figure 7 Nucleolar association of p53 is reduced after genotoxic protein. Nucleolar and nucleoplasmic staining of p53 stress. The subnuclear localization of p53 in H1299 cells was detected when the cells were incubated with ATP permeabilized with exogenously added purified bacterially expressed wild-type his-p53 protein (as in Figure 2) or permeabilized before addition of the purified p53 protein. However, no H1299 expressing inducible p53H175 and T98G cells expressing nucleolar staining of p53 was observed in cells that were endogenous mutant p53 (as in Figure 1) was determined. Cells were pre-permeabliized with ADP or ATPgS (Figure 6b, examined after treatment with 0.22 mM daunorubicin, 0.4 mM panels 4 and 5). This result indicates that ATP actinomycin D and UV irradiation of 25 J/M2 for 16 h. Graphic hydrolysis facilitates movement of p53 to the nucleolus. representation shows the percentage of different cells with nucleolar p53 after different treatments. The graph was prepared It should be mentioned that localization of small as in Figure 6a. peptides to nucleoli occurred in an ATP-independent manner. Although we do not as yet understand the basis for these differences, our data nevertheless indicate that nucleolar localization of full-length p53 can be an mimetic mutant p53S392D were somewhat impaired in actively regulated energy-dependent process. nucleolar localization suggesting that modification of the p53 protein itself is a determinant of p53 trafficking Genotoxic stress decreases p53 association with within the nucleus. the nucleolus It was interesting to determine whether forms of cellular stress that impact p53 could affect its association with Discussion nucleoli. We examined both exogenous and endogenous p53 nucleolar localization in cells treated with daunor- We show herein that both endogenously expressed ubicin, a topoisomerase II inhibitor and an intercalating mutant p53 protein and exogenously added wild-type agent, ultraviolet (UV) irradiation or actinomycin D, an and mutant p53 can localize to the nucleolus in a inhibitor of RNA transcription (Figure 7). At 16 h after concentration- and ATP-dependent manner. We have treatment with these agents, cells were permeabilized also identified two discrete regions within the carboxyl with or without exogenous p53 protein as indicated and terminus of p53 that are essential for nucleolar localiza- then fixed and treated as before. Importantly, each of tion: amino acids 305–322 (NLS I) and the C-terminal these treatments had a dramatic inhibitory effect on the basic domain (amino acids 363–393). Furthermore, we ability of both exogenous and endogenous p53 to found that the treatment of the cells with DNA- associate with nucleoli such that p53 was restricted damaging agents decreases the association of p53 with to the nucleoplasm. Subnuclear localization of p53 the nucleolus. and another nucleolar protein, nucleophosmin, were Use of fluorescein labeled peptides provided us with examined by immunofluorescence using their respective important information. First, it allowed us to defini- antibodies. Note that after actinomycin D treatment tively identify two discrete regions on p53 that can nucleophosmin localized mainly to the nucleoplasm associate with nucleoli. Second, the fact that these same indicating partial disruption of nucleolar structure as peptides when taken up by intact cells could also be described previously (Zatsepina et al., 1989; Chan et al., found associated predominantly with nucleoli lends 1996; Haaf and Ward, 1996). considerable support for our results obtained by pre- We cannot rule out that these treatments have an permeabilization of cells. Third, our results with effect on nucleolar integrity, which prevents p53 from modified peptides suggest that phosphorylation of p53 associating with this organelle. In fact, we have extended at select sites might negatively regulate p53 association previous observations implicating a role for RNA in p53 with nucleoli, consistent with the observation that DNA nucleolar association by showing that RNase A treat- damage prevents this association. This was nicely ment prevents both endogenous and recombinant p53 supported by the observation that the substitution from accumulating in nucleoli of permeabilized cells mutation in p53S392D showed a decrease in p53 nucleolar (Supplementary Figure 3). Nevertheless, p53 is phos- accumulation after proteasome inhibitor treatment. phorylated at S392 after some forms of DNA damage Although our results confirm and extend other studies (Appella and Anderson, 2001), and we found that both demonstrating that p53 can associate with nucleoli, they a peptide phosphorylated at this site and a phospho- pose a number of questions.

Oncogene ATP-dependent p53 nucleolar localization O Karni-Schmidt et al 3888 What is the mechanism by which p53 associates after detergent permeabilization. In MG132-treated with nucleoli? cells, p53 and MDM2 proteins co-localize in the It was reported that macromolecular transport nucleolus following accumulation of high levels of inside the nucleolus is controlled by a combination of p53 protein (Klibanov et al., 2001). In addition, passive diffusion and ATP-dependent processes both p53 and MDM2 have been reported to be degraded (Carmo-Fonseca et al., 2002). In line with this, our in the nucleus under certain treatments (Xirodimas results show that nucleolar localization of p53 is et al., 2001; Shirangi et al., 2002). Another possibility facilitated by active ATP hydrolysis. Identification of is that the nucleolus may serve as a transit station the energy-requiring machinery that facilitates p53 for p53 nuclear export. The model for this is the HIV-1 association would provide useful insight into the Revprotein (Dundr et al., 1995; Zolotukhin and mechanism of this process. Felber, 1999). Importantly, both p53 and Revutilize CRM1 for their export (Thomas and Kutay, 2003). Finally, the nucleolus is the site of ribosomal RNA What are the nucleolar localization signals on p53? synthesis, and transcription by RNA polymerase I Nucleolar localization signal sequences have been occurs at a high rate in this region. That p53 can identified in several proteins notably ribosomal proteins, repress transcription by RNA polymerase I in vitro and and nucleolar components such as nucleolin and in vivo may therefore be relevant to our findings (Budde nucleophosmin. With respect to p53, the one common and Grummt, 1999). Perhaps a subset of p53 is always functional feature of the two nucleolar-associating found in nucleoli as speculated previously (Rubbi and regions of p53 is their possession of nuclear-localization Milner, 2000) and what we have discovered is the basis sequences. It can be speculated that as nucleoli do not of a mechanism for getting it there. The detailed possess defined membrane barriers and thus may not mechanism of how p53 is targeted to nucleoli and the have a system akin to nuclear transport machinery (such physiological relevance of this process remains to be as importins, nuclear pore complex and exportins), elucidated. proteins that localize to nucleoli might do so as a result of specific interactions with binding sites within nucleolar components. Identification of the nucleolar compo- What keeps p53 away from the nucleolus? nents that facilitate p53 nucleolar localization would We have identified in vitro conditions where the p53 greatly illuminate this question. C-terminus and discrete portions of this region can Results with p53 C-terminal lysine substitu- localize rather efficiently in vitro to nucleoli. Yet tion mutants should also help clarify the process. Those nucleolar localization of endogenously expressed full- that allowed nucleolar localization (p536KR, p536KQ length p53 is only rarely documented (Young et al., or p538KR) had positively charged residues in the 2002) and, instead, this phenomenon is found under linker region containing NLS I. Nevertheless, the fact only specialized circumstances, such as pre-permeabili- that p53DC30 that has this region failed to display such zation or treatment of cells with a proteasome inhibitor. localization indicates that the situation is more It is therefore likely that there are features of the complex. A version of p53 with its own nucleolar/NLS full-length protein that preclude its efficient association sequences mutated but with an unrelated NLS (SV40) with this subnuclear structure. Our result that can enter the nucleus but is severely impaired in p53 residues 97–363 (core/linker/tet) can concentrate nucleolar association. This argues that the association within nucleoli, whereas p53 residues 1–363 (N/core/ of p53 with nucleoli is meaningful, although, because the linker/tet) cannot (Figure 3) suggests that the N- motifs that are crucial for nucleolar localization overlap terminus can somehow prevent the functioning of the p53 NLSs, it will be challenging to perform a more the p53 nucleolar localization sequence within its linker refined analysis of the putative amino acids participating region. As the functioning of the other putative NoLS in localizing p53 to the nucleolus. However, in agree- at the extreme C-terminus was partially inhibited by ment with our experiments using exogenously added phosphorylation at S392 (Figure 4) and the association purified proteins and peptides, we conclude that the of the transfected mutant p53S392D with nucleoli was carboxyl terminus of p53 plays an important role in substantially lower after MG132 treatment perhaps this redirecting p53 to the nucleolus after proteasome and other modification(s) may also contribute to exclude inhibition. p53 from this structure. This cannot be the sole expla- nation, however, as we found that bacterially expressed Why sequester p53 in the nucleolus? p53 could localize to nuclei in unstressed cells but A number of different scenarios can be considered. not in cells subjected to DNA damage, the recom- Studies have indicated that nucleolar sequesteration binant p53 protein is unlikely to become efficiently can be a mechanism for inhibiting protein function modified in the permeabilized cells. We cannot exclude (Visintin and Amon, 2000; Zimber et al., 2004). that one or more components of the machinery that Perhaps inappropriate overexpression of p53 results allows p53 to associate with nucleoli may be regulated in its relocalization to nucleoli where, when required, by DNA damage. it may more rapidly reenter the nucleoplasm. This is The establishment of an in vitro system for studying consistent with our observation that only high levels p53 nucleolar localization may eventually provide of endogenous p53 permit its association with nucleoli answers to these and other questions

Oncogene ATP-dependent p53 nucleolar localization O Karni-Schmidt et al 3889 Materials and methods For experiments in which cells were either directly fixed or pre-permeabilized with PBS/1% Triton X-100, immunofluo- Cell culture rescence was performed by adding 50 ml of diluted primary H1299 cells expressing tetracycline-regulated wild-type p53 antibody solution to coverslips. After 1 h of incubation at (p53–14), the transactivation domain mutant, p53Q22/S23 and the room temperature, coverslips were washed and then incubated specific DNA-binding domain mutant, p53H175 were generated for 1 h with 50 ml of diluted (1:100) secondary Cy5 goat anti- as described previously (Resnitzky et al., 1994; Chen et al., mouse IgG antibody (Jackson Immunoresearch), Alexa fluor 1996). Cells were grown and maintained in Rosewell’s Park 488 goat anti-rabbit or Alexa fluor 594 donkey anti-goat Memorial Institute (RPMI) medium supplemented with 10% antibodies (Molecular Probes, Eugene, OR, USA). Coverslips fetal bovine serum (FBS), puromycin (2 mg/ml; Sigma, St were washed three times with PBS after each of the above Louis, MO, USA), G418 (300 mg/ml; Invitrogen, Carlsbad, steps, mounted on a microscope slide with 10 ml of cold 50% CA, USA) and tetracycline (4.5 mg/ml). To induce p53, glycerol, and images were analysed by confocal laser scanning medium was removed from cultures, which were washed with microscopy (Olympus model 1X70) using Fluoview software PBS before plating in the above medium lacking tetracycline (Canter Valley, PA, USA). DIC imaging was used in parallel for 48 h. HT-29 cells were grown in McCoy medium to directly visualize nuclei and nucleoli. In general, localization supplemented with 10% FBS, whereas T98G cells and IMR- of p53 proteins and peptides were determined by visually 90 were grown in minimal essential medium with 2 mML- screening fields of at least 200 cells and choosing representative glutamine supplemented with 10% FBS. Saos-2 cells were images. maintained in Dulbecco’s modified Eagle’s, and H1299 cells DNA staining was performed using 10 nM SYTOX Green were maintained in RPMI medium. All above medium were nucleic acid dye (Molecular Probes) for 10 min. To inhibit supplemented with 10% FBS. proteasome function, cells were incubated with MG132 (10 mM) for 8 h. For ATP-depletion experiments, cells were washed twice with PBS and then were incubated for 45 min at Immunoblot analysis room temprature with 5 mM sodium azide (NaN3) and 1 mM p53 in H1299 cells was induced for 48 h at full or partial levels 2-deoxyglucose. To induce genotoxic stress, cells were treated by varying the amount of tetracycline in the medium. Cells with 0.22 mM daunorubicin (Sigma), 0.4 mM actinomycin D were washed twice with PBS and then collected by scraping in (Calbiochem, San Diego, CA, USA) or 25 J/m2 UV light in a cold extraction buffer (10 mM Tris (pH 7.5), 1 mM ethylene- UV Crosslinker (Fisher Scientific, Pittsburgh, PA, USA) 16 h diaminetetraacetate, 400 mM NaCl, 10% glycerol, 0.5% NP40, before detection of exogenously added or endogenously 5mM NaF, 1 mM dithiothreitol and 0.1 mM phenylmethyl- expressed p53. For RNase treatment, 200 U/ml RNase A sulfonyl fluoride (PMSF)) followed by centrifugation at (Sigma) as added to the permeabilization buffer. 13 000 r.p.m. for 10 min. The supernatants were resuspended in 3 Â protein sample buffer and boiled for 10 min. Proteins were separated by SDS–PAGE using 10% polyacrylamide gels In vitro localization assay and transferred to nitrocellulose for 3 h. The monoclonal H1299 cells were plated on cover slips in 35-mm dishes and antibody PAb 1801 was used to detect p53 and an antiactin grown to 80% confluence. The cells were washed twice with antibody (Sigma) was used as a loading control. PBS and permeabilized with 0.2 ml PBS/1% Triton X-100 supplemented with the purified protein (0.4 mM) or fluorescein- labeled peptides (4–16 mM) for 2 min followed by fixation with Immunofluorescence, cell permeabilization assays 4% paraformaldehyde (Poyurovsky et al., 2003). Immuno- Depending on the experiment, several different antibodies fluorescence was performed as described above. were used in these experiments. Monoclonal anti-p53 antibodies included PAb 1801, PAb DO1 and PAb 421. Expression and purification of recombinant p53 proteins Additionally anti-GST, anti-p53 polyclonal antibody, anti- H175 OMNI and antinucleophosmin polyclonal antibodies were Baculoviruses that express mutant p53 , p53 lacking amino used and purchased from Santa-Cruz Biotechnology Inc. acids 363–393 (HA-p53DC30), his-p53C-terminus domain of (Santa Cruz, CA, USA). The antinucleolin antibody was p53 (amino acids 311–393) and wild-type HA-p53 were described previously (Ghisolfi-Nieto et al., 1996). To detect previously described (Jayaraman and Prives, 1995; Jayaraman total p53 in cells without prior permeabilization (i.e., by direct et al., 1997). Infection and purification of p53 proteins from fixation), cultures were grown on coverslips in 35 mm plates insect cells was performed as before (Jayaraman et al., 1997). (with or without 4 mg/ml tetracycline). Twenty-four hours later, where applicable, medium was removed from cultures Transfection and plasmids and replaced with fresh tetracycline-free medium. Forty-eight Transfections were performed using Lipofectamine 2000 hours after plating the cells were washed twice with PBS and reagent (Invitrogen), according to the manufacturer’s proto- then fixed for 15 min with 4% paraformaldehyde (Sigma) cols. Expression vectors derived from pcDNA3 were pc53 followed by three washes in PBS. Cells were then treated with expressing full-length p53 cDNA, pc53DC30 expressing p53 PBS/0.5% Triton X-100 for 1.5 min followed by blocking with lacking the last 30 amino acids, Flag-tagged wild-type p53 and 0.5% bovine serum albumin (Sigma) for 30 min at room Flag-tagged substitution mutant p53S392A and Flag-p53S392D temperature before preparation for immunofluorescence. in which serine 392 was substituted with either alanine To detect p53 in pre-permeablized cells, 60 mm culture or aspartate, respectively. Substitution mutants p536KR and dishes were incubated with 0.2 ml PBS/1% Triton X-100 p536KQ (amino acids 370, 372, 373, 380, 381 and 386 supplemented with protease inhibitors (100 mM sodium substituted with arginine and glutamine, respectively), and orthovanadate, 50 mM PMSF and 10 mg/ml chymostatin, p538KR and p538KQ (amino acids 305, 320, 370, 372, 373, 380, leupeptin, antipain and pepstatin) for approximately 2 min, 381 and 386 substituted with arginine and glutamine, that is, until only nuclei without cytoplasm were visible in at respectively) were made by using the QuickChange II Site- least 80% of the cells. Nuclei were then fixed immediately with Directed Mutagenesis Kit (Stratagene, La Jolla, CA, USA) on 4% paraformaldehyde for 15 min followed by preparation for pTRE2-hyg (Invitrogen) with p53 cDNA ligated between visualization by immunofluorescence as described above. Nhe1 and EcoRV sites. In order to make the p538KQ-SVNLS

Oncogene ATP-dependent p53 nucleolar localization O Karni-Schmidt et al 3890 construct, the SV-40 NLS sequence (PKKKRK) was added to specifically at its N-terminus at 100% yield. All other amino the carboxyl terminus of the p538KQ using the same vector as acids were purchased from NOVAbiochem, La Jolla, CA, described above. All substitution mutants were confirmed by USA. The peptides were purified on a Waters HPLC using a sequencing. reverse-phase C8 preparative column (Vydac, Hesperia, CA, USA). Peptide synthesis Peptides used in these studies were synthesized on a Pioneer peptide synthesizer (Applied Biosystems, Foster City, Acknowledgements CA, USA) using standard Fmoc chemistry (Fernandez- Fernandez et al., 2005). Double couplings were used when We are particularly grateful to V Gottifredi and K De Vos for needed. The peptides were labeled with fluorescein at the guidance during the earlier stages of this work. Thanks to N-terminus while still on the resin on the Pioneer peptide J Ahn, N Baptiste, JC Bulinsky, M Poyurovsky, RPT Tanaka synthesizer using a four-fold excess of fluorescein-OSu and M Urist for advice and help and to E Freulich who (Molecular Probes) and a four-fold excess of hydroxybenzo- provided expert technical assistance. This work was supported triazole. This procedure ensures that the peptide is labeled by NIH Grants CA58316 and CA87497.

References

Appella E, Anderson CW. (2001). Post-translational modifica- Ghisolfi-Nieto L, Joseph G, Puvion-Dutilleul F, Amalric F, tions and activation of p53 by genotoxic stresses. Eur J Bouvet P. (1996). Nucleolin is a sequence-specific RNA- Biochem 268: 2764–2772. binding protein: characterization of targets on pre-ribosomal Ashcroft M, Taya Y, Vousden KH. (2000). Stress signals RNA. J Mol Biol 260: 34–53. utilize multiple pathways to stabilize p53. Mol Cell Biol 20: Giaccia AJ, Kastan MB. (1998). The complexity of p53 3224–3233. modulation: emerging patterns from divergent signals. Genes Benninghoff J, Kartarius S, Teleb Z, Selter H, Unteregger G, Dev 12: 2973–2983. Zwergel T et al. (1999). Two different forms of p53 localized Gobert C, Bracco L, Rossi F, Olivier M, Tazi J, Lavelle F et al. differently within cells of urogenital tumours. Cancer Lett (1996). Modulation of DNA topoisomerase I activity by 144: 55–64. p53. Biochemistry 35: 5778–5786. Budde A, Grummt I. (1999). p53 represses ribosomal gene Haaf T, Ward DC. (1996). Inhibition of RNA polymerase II transcription. Oncogene 18: 1119–1124. transcription causes chromatin decondensation, loss of Bykov VJ, Issaeva N, Selivanova G, Wiman KG. (2002a). nucleolar structure, and dispersion of chromosomal Mutant p53-dependent growth suppression distinguishes domains. Exp Cell Res 224: 163–173. PRIMA-1 from known anticancer drugs: a statistical Horky M, Wurzer G, Kotala V, Anton M, Vojtesek B, analysis of information in the National Cancer Institute Vacha J et al. (2001). Segregation of nucleolar components database. Carcinogenesis 23: 2011–2018. coincides with caspase-3 activation in cisplatin-treated HeLa Bykov VJ, Issaeva N, Shilov A, Hultcrantz M, Pugacheva E, cells. J Cell Sci 114: 663–670. ChumakovP et al. (2002b). Restoration of the tumor Issaeva N, Friedler A, Bozko P, Wiman KG, Fersht AR, suppressor function to mutant p53 by a low-molecular- Selivanova G. (2003). Rescue of mutants of the tumor weight compound. Nat Med 8: 282–288. suppressor p53 in cancer cells by a designed peptide. Proc Carmo-Fonseca M, Mendes-Soares L, Campos I. (2000). Natl Acad Sci USA 100: 13303–13307. To be or not to be in the nucleolus. Nat Cell Biol 2: Jayaraman J, Prives C. (1995). Activation of p53 sequence- E107–12. specific DNA binding by short single strands of DNA Carmo-Fonseca M, Platani M, Swedlow JR. (2002). Macro- requires the p53 C-terminus. Cell 81: 1021–1029. molecular mobility inside the cell nucleus. Trends Cell Biol Jayaraman L, Freulich E, Prives C. (1997). Functional 12: 491–495. dissection of p53 tumor suppressor protein. Methods Chan PK, Qi Y, Amley J, Koller CA. (1996). Quantitation of Enzymol 283: 245–256. the nucleophosmin/B23-translocation using imaging analy- Kern SE, Kinzler KW, Bruskin A, Jarosz D, Friedman P, sis. Cancer Lett 100: 191–197. Prives C et al. (1991). Identification of p53 as a sequence- Chen X, Ko LJ, Jayaraman L, Prives C. (1996). p53 levels, specific DNA-binding protein. Science 252: 1708–1711. functional domains, and DNA damage determine the extent KlibanovSA, O’Hagan HM, Ljungman M. (2001). of the apoptotic response of tumor cells. Genes Dev 10: Accumulation of soluble and nucleolar-associated 2438–2451. p53 proteins following cellular stress. J Cell Sci 114: Colombo E, Marine JC, Danovi D, Falini B, Pelicci PG. 1867–1873. (2002). Nucleophosmin regulates the stability and transcrip- Lamond AI, Sleeman JE. (2003). Nuclear substructure and tional activity of p53. Nat Cell Biol 4: 529–533. dynamics. Curr Biol 13: R825–R825. Daniely Y, Dimitrova DD, Borowiec JA. (2002). Stress- Latonen L, Kurki S, Pitkanen K, Laiho M. (2003). p53 and dependent nucleolin mobilization mediated by p53-nucleolin MDM2 are regulated by PI-3-kinases on multiple levels complex formation. Mol Cell Biol 22: 6014–6022. under stress induced by UV radiation and proteasome Dundr M, Leno GH, Hammarskjold ML, Rekosh D, Helga- dysfunction. Cell Signal 15: 95–102. Maria C, Olson MO. (1995). The roles of nucleolar structure Leung AK, Andersen JS, Mann M, Lamond AI. and function in the subcellular location of the HIV-1 Rev (2003). Bioinformatic analysis of the nucleolus. Biochem J protein. J Cell Sci 108(Part 8): 2811–2823. 376: 553–569. Fernandez-Fernandez MR, VeprintsevDB, Fersht AR. (2005). Leung AK, Lamond AI. (2003). The dynamics of the Proteins of the S100 family regulate the oligomerization nucleolus. Crit Rev Eukaryot Gene Expr 13: 39–54. of p53 tumor suppressor. Proc Natl Acad Sci USA 102: Liang SH, Clarke MF. (2001). Regulation of p53 localization. 4735–4740. Eur J Biochem 268: 2779–2783.

Oncogene ATP-dependent p53 nucleolar localization O Karni-Schmidt et al 3891 Liu G, Chen X. (2002). The ferredoxin reductase gene is Stros M, Reich J. (1998). Formation of large nucleoprotein regulated by the p53 family and sensitizes cells to oxidative complexes upon binding of the high-mobility-group (HMG) stress-induced apoptosis. Oncogene 21: 7195–7204. box B-domain of HMG1 protein to supercoiled DNA. Eur J Llanos S, Clark PA, Rowe J, Peters G. (2001). Stabilization of Biochem 251: 427–434. p53 by p14ARF without relocation of MDM2 to the Tao W, Levine AJ. (1999). P19(ARF) stabilizes p53 by nucleolus. Nat Cell Biol 3: 445–452. blocking nucleo-cytoplasmic shuttling of Mdm2. Proc Natl Marciniak RA, Lombard DB, Johnson FB, Guarente L. Acad Sci USA 96: 6937–6941. (1998). Nucleolar localization of the Werner syndrome Thomas F, Kutay U. (2003). Biogenesis and nuclear export of protein in human cells. Proc Natl Acad Sci USA 95: ribosomal subunits in higher eukaryotes depend on the 6887–6892. CRM1 export pathway. J Cell Sci 116: 2409–2419. Marechal V, Elenbaas B, Piette J, Nicolas JC, Levine AJ. Tsai RY, McKay RD. (2002). A nucleolar mechanism (1994). The ribosomal L5 protein is associated with mdm-2 controlling cell proliferation in stem cells and cancer cells. and mdm-2-p53 complexes. Mol Cell Biol 14: 7414–7420. Genes Dev 16: 2991–3003. Mayer C, Grummt I. (2005). Cellular stress and nucleolar Visintin R, Amon A. (2000). The nucleolus: the magician’s hat function. Cell Cycle 4: 1036–1038. for cell cycle tricks. Curr Opin Cell Biol 12: 752. Menendez D, Inga A, Resnick MA. (2006). The biological Visintin R, Craig K, Hwang ES, Prinz S, Tyers M, Amon A. impact of the human master regulator p53 can be altered by (1998). The phosphatase Cdc14 triggers mitotic exit by mutations that change the spectrum and expression of its reversal of Cdk- dependent phosphorylation. Mol Cell 2: target genes. Mol Cell Biol 26: 2297–2308. 709–718. Mogaki M, Hirota M, Chaney WG, Pour PM. (1993). Weber JD, Taylor LJ, Roussel MF, Sherr CJ, Bar-Sagi D. Comparison of p53 protein expression and cellular localiza- (1999). Nucleolar Arf sequesters Mdm2 and activates p53. tion in human and hamster pancreatic cancer cell lines. Nat Cell Biol 1: 20–26. Carcinogenesis 14: 2589–2594. Wesierska-Gadek J, Schloffer D, Kotala V, Horky M. (2002). Olson MO, Dundr M. (2005). The moving parts of the Escape of p53 protein from E6-mediated degradation in nucleolus. Histochem Cell Biol 123: 203–216. HeLa cells after cisplatin therapy. Int J Cancer 101: 128–136. Olson MO, Dundr M, Szebeni A. (2000). The nucleolus: an old Wong JM, Kusdra L, Collins K. (2002). Subnuclear shuttling factory with unexpected capabilities. Trends Cell Biol 10: of human telomerase induced by transformation and DNA 189–196. damage. Nat Cell Biol 4: 731–736. Pokrovskaja K, Mattsson K, Kashuba E, Klein G, Szekely L. Xirodimas DP, Stephen CW, Lane DP. (2001). Cocompart- (2001). Proteasome inhibitor induces nucleolar translocation mentalization of p53 and Mdm2 is a major determinant for of Epstein–Barr virus-encoded EBNA-5. J Gen Virol 82: Mdm2-mediated degradation of p53. Exp Cell Res 270: 345–358. 66–77. Poyurovsky MV, Jacq X, Ma C, Karni-Schmidt O, Parker PJ, Young PJ, Day PM, Zhou J, Androphy EJ, Morris GE, Chalfie M et al. (2003). Nucleotide binding by the Mdm2 Lorson CL. (2002). A direct interaction between the survival RING domain facilitates Arf-independent Mdm2 nucleolar motor neuron protein and p53 and its relationship to spinal localization. Mol Cell 12: 875–887. muscular atrophy. J Biol Chem 277: 2852–2859. Prives C, Hall PA. (1999). The p53 pathway. J Pathol 187: Zatsepina OV, Voronkova LN, Sakharov VN, Chentsov YS. 112–126. (1989). Ultrastructural changes in nucleoli and fibrillar centers Prives C, Manley JL. (2001). Why is p53 acetylated? Cell 107: under the effect of local ultraviolet microbeam irradiation of 815–818. interphase culture cells. Exp Cell Res 181: 94–104. Resnitzky D, Gossen M, Bujard H, Reed SI. (1994). Zerrahn J, Deppert W, Weidemann D, Patschinsky T, Acceleration of the G1/S phase transition by expression of Richards F, Milner J. (1992). Correlation between the cyclins D1 and E with an inducible system. Mol Cell Biol 14: conformational phenotype of p53 and its subcellular 1669–1679. location. Oncogene 7: 1371–1381. Rokaeus N, Klein G, Wiman KG, Szekely L, Mattsson K. Zhang Y, Xiong Y, Yarbrough WG. (1998). ARF promotes (2006). PRIMA-1(MET) induces nucleolar accumulation of MDM2 degradation and stabilizes p53: ARF-INK4a locus mutant p53 and PML nuclear body-associated proteins. deletion impairs both the Rb and p53 tumor suppression Oncogene [Epub ahead of print]. pathways. Cell 92: 725–734. Rubbi CP, Milner J. (2000). Non-activated p53 co-localizes Zimber A, Nguyen QD, Gespach C. (2004). Nuclear bodies with sites of transcription within both the nucleoplasm and and compartments: functional roles and cellular signalling in the nucleolus. Oncogene 19: 85–96. health and disease. Cell Signal 16: 1085–1104. Shirangi TR, Zaika A, Moll UM. (2002). Nuclear degradation Zolotukhin AS, Felber BK. (1999). Nucleoporins nup98 and of p53 occurs during down-regulation of the p53 response nup214 participate in nuclear export of human immunode- after DNA damage. FASEB J 16: 420–422. ficiency virus type 1 Rev. J Virol 73: 120–127.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc).

Oncogene