The Non-ankyrin C Terminus of IκBα Physically Interacts with p53 in Vivo and Dissociates in Response to Apoptotic Stress, Hypoxia, DNA Damage, and Transforming Growth Factor-β1-mediated Growth Suppression*

Transforming growth factor beta (TGF-β1) suppresses the growth of mink lung Mv1Lu epithelial cells, whereas testicular hyaluronidase abolishes the growth inhibition. Exposure of Mv1Lu cells to TGF-β1 rapidly resulted in down-regulation of cytosolic IκBα and hyaluronidase prevented this effect, suggesting a possible role of IκBα in the growth regulation. Ectopic expression of wild-type and dominant negative IκBα prevented TGF-β1-mediated growth suppression. Nonetheless, the blocking effect of IκBα is not related to regulation of NF-κB function by its N-terminal ankyrin-repeat region (amino acids 1–243). Removal of the PEST (proline-glutamic acid-serine-threonine) domain-containing C terminus (amino acids 244–314) abolished the IκBα function, and the C terminus alone blocked the TGF-β1 growth-inhibitory effect. Co-immunoprecipitation by anti-p53 antibody using Mv1Lu and other types of cells, as well as rat liver and spleen, revealed that a portion of cytosolic IκBα physically interacted with p53. In contrast, Mdm2, an inhibitor of p53, was barely detectable in the immunoprecipitates. The cytosolic p53·IκBα complex rapidly dissociated in response to apoptotic stress, etoposide- and UV-mediated DNA damage, hypoxia, and TGF-β1-mediated growth suppression. Also, a rapid increase in the formation of the nuclear p53·IκBα complex was observed during exposure to etoposide and UV. In contrast, TGF-β1-mediated promotion of fibroblast growth failed to mediate p53·IκBα dissociation. Mapping by yeast two-hybrid showed that the non-ankyrin C terminus of IκBα physically interacted with the proline-rich region and a phosphorylation site, serine 46, in p53. Deletion of serine 46 or alteration of serine 46 to glycine abolished the p53·IκBα interaction. Alteration to threonine retained the binding interaction, suggesting that serine 46 phosphorylation is involved in the p53·IκBα complex formation. Functionally, enhancement of p53 apoptosis was observed when p53 and IκBα were transiently co-expressed in cells. Together, the IκBα·p53 complex plays an important role in responses involving growth regulation, apoptosis, and hypoxic stress.

Transforming growth factor beta (TGF-␤) 1 plays an important role in controlling embryogenesis, immune cell functions, and extracellular matrix homeostasis (Refs. 1-4; for reviews). TGF-␤ promotes the growth of fibroblasts and other cell types, whereas it inhibits epithelial cell growth and induces apoptosis of hepatocytes and various cancer cells. The underlying mechanisms are unknown.
TGF-␤ signaling involves binding of TGF-␤ to the type II receptor, followed by recruiting the type I receptor, then inducing phosphorylation and hetero-dimerization of Smad2 and 3, and binding of Smad4 to the Smad2/3 complex. The Smad2/3/4 protein complex migrates to the nucleus and regulates the transcription of target genes (1,2).
Cross-talk of the TGF-␤/Smad pathway with other cellular signaling pathways has been demonstrated (5). For example, the Smad2/3/4 protein complex may recruit c-Jun, ATF-2, CBP/ p300, and STAT3 in the transcriptional regulation of a target gene (2). These recruited proteins are also transducers of the stress and STAT signaling pathways, respectively. NF-B p65 regulates Smad7 promoter activity (6). Smad7 is an inhibitor of the TGF-␤ signaling pathway. TGF-␤ activates Ras and three members of the mitogen-activated protein kinase (MAPK) superfamily in epithelial cells (7). However, TGF-␤-mediated activation of p44/42 MAPK (ERK) pathway is due to autocrine secretion of basic fibroblast growth factor in fibroblasts (8).
TGF-␤1 and its induced matrix protein(s) protect L929 fibroblasts from killing by tumor necrosis factor (TNF) (9 -11). This protective function is effectively abolished by testicular hyaluronidase (12,13). TGF-␤1 suppresses the proliferation of mink lung epithelial Mv1Lu cells and hyaluronidase inhibits the suppression (12,13). Here, it is demonstrated that TGF-␤1 rapidly downregulated IB␣ expression in Mv1Lu cells and hyaluronidase abolished the down-regulation, suggesting that IB␣ participates in the TGF-␤1-mediated growth suppression. Ectopic expression of IB␣ blocked TGF-␤1-mediated growth suppression of Mv1Lu cells, and the non-ankyrin C terminus of IB␣ was found to contribute to this regulatory activity. This region of IB␣ contains an acidic PEST (proline-glutamic acid-serine-threonine) domain and fails to inhibit NF-B function (14).
Notably, a portion of cytosolic IB␣ was found to interact with p53 in Mv1Lu and other unstimulated resting cells, as well as rat liver and spleen. Dissociation of the p53⅐IB␣ complex occurred in response to growth suppression by TGF-␤1, apoptotic stress, hypoxia, DNA damage, and UV radiation. The dissociation allows p53 nuclear translocation. In contrast, TGF-␤1-mediated promotion of fibroblast growth failed to mediate p53⅐IB␣ dissociation. Enhancement of p53 apoptosis was observed when various cells were co-transfected with p53 and IB␣ expression constructs. This study suggests that IB␣ regulates cytosolic p53 function in resting cells and participates in p53-mediated apoptosis in response to various stresses. The structural basis of the p53⅐IB␣ binding interaction was analyzed by the yeast two-hybrid system, and functional significance of this binding interaction was discussed.

EXPERIMENTAL PROCEDURES
Cell Lines-Mink lung Mv1Lu epithelial cells, human Molt-4 T cells, human prostate Du145 cells, monkey kidney Cos7 fibroblasts, and murine L929 fibroblasts were from American Type Culture Collections (ATCC, Manassas, VA) and cultured according to provider's instructions. Human SK-N-SH neuroblastoma cells, human ovarian HeLa cells, and human breast cancer MCF-7 cells were kindly provided by Jeff Mattison and Dr. John Noti of this Institute, respectively. A L929 cell line, which stably expressed GFP, was established as previously described (9).
Cell Proliferation Assays-TGF-␤1-mediated inhibition of Mv1Lu cell growth was performed as previously described (13). Briefly, Mv1Lu cells were cultured on 96-well plates overnight and treated with TGF-␤1 in the presence or absence of hyaluronidase or TNF-␣ for 24 and 48 h. The extent of cell proliferation was determined by both crystal violet staining and the MTS proliferation assay (Promega, Madison, WI) (13). Similar experiments were performed using L929 fibroblasts.
DNA Constructs-A murine full-length IB␣ cDNA was found in the universal EST data base (GenBank TM accession AA606238) and the clone was obtained from the Incyte Genosystems (St. Louis, MO). Three expression constructs were made with the pEGFP-C1 vector (cloning FIG. 1. Hyaluronidase counteracts TGF-␤1-mediated growth suppression of Mv1Lu cells. Mv1Lu cells were treated with TGF-␤1 (2 ng/ml), in the presence or absence hyaluronidase or TNF-␣ for 48 h. The cells were then stained with crystal violet to determine the extent of growth inhibition by TGF-␤1 (n ϭ 8). Hyaluronidase counteracted the growth inhibitory effect of TGF-␤1. Both TGF-␤1 and TNF-␣ suppressed the cell growth in an additive manner. Similar results were observed using the MTS proliferation assay (data not shown). Electroporation-Mv1Lu cells (3 million cells in 500 l of serum-free minimal essential medium) were electroporated in duplicates with 20 g of the above expression constructs (200 V and 50 ms using the BTX ECM 630 electroporator, Genetronics, San Diego, CA). The cells were then seeded onto 30-mm Petri dishes, grown overnight, and exposed to TGF-␤1 (2 ng/ml) for 48 h. The cells were then stained with crystal violet to determine the extent of cell growth as well as growth inhibition by TGF-␤1. In other experiments, Mv1Lu and other types of cells were co-transfected with p53-DsRed and the full-length IB␣ or IB␣243C (in pEGFP-C1) by electroporation. The cells were cultured for 48 h, and the extent of p53-mediated cell death was examined by crystal violet staining.
Yeast Two-hybrid Interactions-To determine whether p53 physically interacts with IB␣ in vivo, we utilized the CytoTrap yeast twohybrid system (Stratagene) as previously described (18). Binding of a cytosolic Sos-tagged bait protein to a cell membrane-anchored target protein (tagged with a myristoylation signal) results in activation of the FIG. 2. TGF-␤1 down-regulates IB␣, and hyaluronidase blocks the down-regulation in Mv1Lu cells. A, Mv1Lu cells were pretreated with or without hyaluronidase (100 units/ml) for 1 h, followed by exposure to TGF-␤1 (2 ng/ml) for 0.5-4 h. The expression of cytosolic p65 NF-B and IB␣ in these cells was examined by Western blotting. TGF-␤1 rapidly down-regulated IB␣, whereas hyaluronidase blocked the down-regulation. The cytosolic levels of p65 NF-B were not affected by TGF-␤1 and/or hyaluronidase. B, in comparison, L929 fibroblasts were pretreated with hyaluronidase (100 units/ml) for 1 h and then co-treated with TGF-␤1 (2 ng/ml) for various indicated times. The cytosolic level of IB␣ was not affected by TGF-␤1 and/or hyaluronidase. TGF-␤1 reduced the cytosolic level of p65 NF-B in a time-related manner. Hyaluronidase-pretreated cells had a reduced level of cytosolic p65 NF-B, whereas TGF-␤1 restored the NF-B level.
FIG. 3. The non-ankyrin C terminus of IB␣ inhibits TGF-␤1mediated growth of Mv1Lu cells. Mv1Lu cells were transfected with various GFP-tagged IB␣ constructs by electroporation and cultured in 30-mm Petri dishes overnight, and expression of these GFP-tagged proteins was examined by fluorescence microscopy. The cells were than exposed to TGF-␤1 for 48 h, followed by staining with crystal violet to determine the extent of growth inhibition. Shown in the graph are representative data from three experiments. A schematic diagram of IB␣ is also shown. The N terminus (amino acids 1-68) contains two ubiquitination sites, two serine phosphorylation sites (Ser-32 and Ser-36) and five ankyrin repeats at the central region, followed by a PEST domain-containing C terminus (amino acids 244 -314). GFP, control pEGFP-C1 construct; GFP-IB␣, GFP-tagged wild-type IB␣; GFP-DN-IB␣, GFP-tagged dominant negative IB␣ (S32R and S36A); GFP-IB␣243N, GFP-tagged IB␣ N terminus (amino acids 1-243); GFP-IB␣243C, GFP-tagged IB␣ C terminus (amino acids 244 -314).

FIG. 4. p53 physically interacts with IB␣ in vivo.
A, both cytosolic and nuclear fractions were extracted from Mv1Lu cells. Immunoprecipitation of the lysates with anti-p53 IgG resulted in the presence of p53 in the cytosol and nucleus (top panel). IB␣ was co-precipitated with p53 in the cytosol but not in the nucleus (bottom panel), indicating p53 binds IB␣ in the cytosol. IP, immunoprecipitation; IgH, IgG heavy chain. B, similarly, transient expression of a FLAG-tagged p53 expression construct in L929 cells, followed by culturing overnight and processing immunoprecipitation, resulted in co-precipitation of IB␣ with FLAG-p53 in the cytosol but in the nucleus. C, precipitation of IB␣ from rat liver cytosolic extract with specific antibodies (immobilized on agarose beads) resulted in the co-presence of IB␣ and p53 in the precipitate (left panel). Similarly, the presence of both IB␣ and p53 in the precipitate was observed when anti-p53 IgG was used (right panel). D, co-transfection of Mv1Lu cells with p53-DsRed and GFP-IB␣ constructs resulted in colocalization of the expressed proteins in the perinuclear area. The nucleus was stained with DAPI.
Ras signaling pathway in yeast. This binding interaction allows mutant yeast cdc25H to grow in 37°C using a selective agarose plate containing galactose. Without binding, yeast cells fail to grow at 37°C. The aboveindicated IB␣ constructs were used as baits (in pSos vector; see Table  I for synthetic primers). The following p53 target constructs were made as previously described (18): (i) a full-length p53, (ii) a partial Nterminal p53 (amino acids 1-100), and (iii) the proline-rich (or growth regulatory) region of p53 (amino acids 66 -100). Additionally, a p53 construct without the proline-rich region was made (in pMyr vector). The original proline-rich region-deleted p53 construct (20) was a gift of Dr. U. Moll of the State University of New York at Stony Brook and Dr. A. J. Levine of the Rockefeller University. Self-interaction between pSos-MafB and pMyr-MafB as well as pMyr-p53 and pSos-WOX1 (WW domain oxidoreductase) (18) were tested as positive controls. Vector combinations for negative binding interactions were empty pSos versus empty pMyr and pSos-Lamin C versus pMyr-p53.
Immunoprecipitation-Immunoprecipitation was performed as previously described (18). Both cytosolic and nuclear proteins from Mv1Lu, L929, and other types of cells were extracted using the NE-PER Nuclear and Cytoplasmic Extraction reagents (Pierce, Rockford, IL). Endogenous p53 was precipitated by antibodies against p53 and protein Aagarose (Pierce), separated by SDS-PAGE, and detected in Western blotting. Co-precipitation of IB␣ with p53 was determined using anti-IB␣ antibody in duplicate precipitation experiments. In additional experiments, FLAG-tagged p53 (in pcDNA3.1) was expressed in cells and precipitated by anti-FLAG antibody (Santa Cruz Biotechnology) from the cell lysates. The presence of FLAG-p53 and the co-precipitated endogenous IB␣ was determined by antibodies against p53 and IB␣, respectively, in Western blotting. Where indicated, agarose bead-conjugated anti-p53 IgG was used in immunoprecipitation experiments. The immobilized anti-p53 IgG was less effective in binding to antigen than the solution-phase anti-p53 IgG.
Additionally, liver extracts from male Wistar rats were prepared by an ultrasonication-based Tissumizer (Tekmar, Cincinnati, OH), using a lysis buffer containing a mixture of protease inhibitors (11). The cytosolic lysate (ϳ5 mg) was incubated with agarose beads covalently conjugated with anti-IB␣ IgG or anti-p53 IgG (Santa Cruz Biotechnology) and rotated at 4°C for 4 h. Co-precipitation of IB␣ and p53 was then determined in Western blotting.
Transient Expression-Where indicated, Mv1Lu cells were cultured on coverslips overnight and co-transfected with IB␣-pEGFP-C1 and p53-DsRed constructs by FuGENE (Roche Molecular Biochemicals, Indianapolis, IN), a liposome-based reagent. Twenty-four hours later, the cells were fixed with 3.3% formaldehyde and permeabilized with 0.1% Triton X-100 (Sigma) and stained with the nuclear dye DAPI. (Calbiochem). The cells were examined under fluorescence microscopy.
Apoptosis, Hypoxia, DNA Damage, and UV Radiation-L929 and other indicated cells were treated with staurosporine (1 M) for 1 h to induce apoptosis, followed by preparation of cytosolic and nuclear fractions from the cells and immunoprecipitation using anti-p53 IgG beads (or anti-p53 IgG and protein A-agarose beads). Similarly, the cells were cultured under hypoxic conditions in the presence of deferoxamine (100 M) for 16 h prior to processing cell lysis and immunoprecipitation. Also, cells were treated with etoposide (50 g/ml) or exposed to UV light (120 mj/cm 2 ; using a UV cross-linker from Fisher Scientific). One hour later, the cells were subjected to lysis and co-immunoprecipitation. Where indicated, fresh spleen cells were isolated from the spleens of sacrificed male Wistar rats and treated similarly as above.

TGF-␤1-mediated Growth Inhibition of Epithelial Mv1Lu
Cells Is Blocked by Hyaluronidase and Is Related with Downregulation of IB␣-TGF-␤1 inhibits the proliferation of Mv1Lu epithelial cells (13). Exposure of Mv1Lu cells to TGF-␤1 for 48 h resulted in growth inhibition by ϳ40% (Fig. 1A). We have shown that bovine testicular hyaluronidase counteracts TGF-␤1-mediated protection of TNF-␣ cytotoxicity in L929 and LNCaP prostate cells (13). In agreement with previous observations (13), treatment of Mv1Lu cells with hyaluronidase resulted in increased cell proliferation and reduction of TGF-␤1mediated growth inhibition in a dose-dependent manner (Fig. 1A).
To determine the mechanisms for the functional antagonism between TGF-␤1 and hyaluronidase, Mv1Lu cells were treated with TGF-␤1 for various indicated times, followed by determining the expression of p65 NF-B and IB␣ by Western blotting. . L929-GFP cells were treated with TNF-␣ (50 ng/ml) for 1 h, followed by preparing cytosolic and nuclear fractions and immunoprecipitation using anti-p53 IgG agarose beads. One-tenth (ϳ20 g) of the total proteins were loaded for SDS-PAGE, and expression of p53, Mdm2, and IB␣ was examined using specific antibodies (right panel; see Pre-IP). IB␣, but not Mdm2, was co-precipitated with p53 in the cytosol and TNF-␣ mediated p53⅐IB␣ dissociation (right panel; see IP with anti-p53). B, under similar experimental conditions, L929 cells were treated with staurosporine (1 M) for 1 h, followed by determining the presence of p53⅐IB␣ complex by co-immunoprecipitation. The p53⅐IB␣ complex was found in the cytosol, and staurosporine mediated the complex dissociation. C, a similar experiment was performed using ovarian HeLa cells. Again, staurosporine mediated the dissociation of cytosolic p53⅐IB␣ complex, along with p53 nuclear translocation.
TGF-␤1 rapidly down-regulated the expression of cytosolic IB␣ in 30 min, whereas it had no effect on p65 NF-B ( Fig.  2A). Notably, pretreatment of Mv1Lu cells with hyaluronidase for 1 h, followed by exposure to TGF-␤1, prevented the downregulation of IB␣ ( Fig. 2A). These results suggest that suppression of IB␣ expression is necessary for TGF-␤1-mediated growth suppression of Mv1Lu cells.
Either TGF-␤1 or hyaluronidase increases L929 cell growth by 10 -30% in 48 h (11,13). However, no significant enhancement of cell growth (0 Ϯ 10%) was observed when L929 cells were exposed to both proteins. In control experiments, L929 fibroblasts were pretreated with hyaluronidase for 1 h and then exposed to TGF-␤1 for various indicated times. The cytosolic level of IB␣ was not affected by TGF-␤1 or by TGF-␤1 and hyaluronidase in combination (Fig. 2B). TGF-␤1 reduced the cytosolic level of p65 NF-B in a time-related manner (Fig. 2B). Hyaluronidasepretreated cells had a reduced level of cytosolic p65 NF-B, whereas TGF-␤1 restored the NF-B level (Fig. 2B).
The Non-ankyrin C Terminus of IB␣ Blocks TGF-␤1-mediated Growth Inhibition of Mv1Lu Cells-To determine whether IB␣ is involved in the TGF-␤1-mediated growth suppression, Mv1Lu cells were electroporated with the control GFP or the GFP-IB␣ construct. Following overnight culturing, the cells were treated with TGF-␤1 for 48 h. TGF-␤1 mediated 37.7% of the growth inhibition of GFP-expressing cells, whereas IB␣ reduced the growth inhibition down to 10.7% (Fig. 3). Dominant negative IB␣ also reduced the growth inhibitory effect of TGF-␤1 to 2.3% (Fig. 3).
IB␣ Physically Interacts with p53 in Vivo-p53 is involved in the TGF-␤1-mediated growth suppression and apoptosis (21)(22)(23)(24). The possible binding interaction between p53 and IB␣ in vivo was examined. Immunoprecipitation of both cytosolic and nuclear p53 from Mv1Lu cells was performed using anti-p53 IgG antibody (against wild-type p53). Endogenous p53 was found in both cytosol and nucleus (Fig. 4A). IB␣ was co-precipitated with p53 in the cytosol but not in the nucleus (Fig. 4A). The p53⅐IB␣ interaction could not be abolished by treatment of cells with EDTA (5 mM), indicating that the binding is calcium-independent.
Similarly, transient overexpression of FLAG-tagged p53 in L929 cells, followed by overnight culturing and processing immunoprecipitation using anti-FLAG IgG, resulted in the presence of FLAG-p53 in the cytosol and nucleus (Fig. 4B). Endogenous IB␣ was co-precipitated with FLAG-p53 in the cytosol but not in the nucleus (Fig. 4B).
When rat liver cytosolic extract was incubated with immobilized anti-IB␣ IgG beads, p53 was found in the precipitate of IB␣ (Fig. 4C), again indicating that a portion of cytosolic p53 physically interacts with IB␣ in vivo. Similarly, precipitation of p53 using anti-p53 antibody also resulted in the co-presence of p53 and IB␣ in the precipitate using the rat liver cytosolic extract (Fig. 4C). In negative controls, when the above antibodies were heat-inactivated (75°C for 30 min) and used for immunoprecipitation, no precipitated target proteins were found (data not shown).
In parallel with the above observations, co-transfection of Mv1Lu cells with p53-DsRed and IB␣-pEGFP-C1 constructs was performed. The cells were cultured overnight and then examined under fluorescence microscopy. Both the expressed p53-DsRed and GFP-IB␣ were co-localized in the cytoplasm (Fig. 4D).
Dissociation of p53⅐IB␣ Complex in Response to Apoptotic Stress, Hypoxia, DNA Damage, and TGF-␤1-mediated Growth Suppression-To further elucidate the functional significance of p53⅐IB␣ binding in vivo, various cell lines were treated with TGF-␤1, TNF, staurosporine, and deferoxamine, then the dissociation of p53⅐IB␣ complex in the cytoplasm and nucleus was examined.
Exposure of L929 cells to TNF-␣ resulted in IB␣ degradation (Fig. 5A). Interestingly, the TNF-dependent IB␣ degradation was abolished in an established L929 cell line stably expressing GFP (Fig. 5A). This cell line is suitable for examining TNF-mediated p53⅐IB␣ complex formation and dissociation. The abundance of cytosolic IB␣ was greater than that of p53 ( Fig. 5A; see Pre-IP). Mdm2, a known inhibitor of p53, was barely detectable (Fig. 5A). Indeed, p53 level is normally low in unstimulated resting cells (Ref. 25, review). Precipitation of p53 with anti-p53 IgG beads showed the presence of cytosolic p53 and IB␣ in the precipitate, and TNF mediated p53⅐IB␣ dissociation (Fig. 5A). Mdm2 was barely detectable (Fig. 5A).
The results indicate that the cytosolic p53 complexes with IB␣, rather than with Mdm2, in unstimulated L929 cells. Also, only a small portion of cytosolic IB␣ physically interacts with p53.
Similarly, staurosporine-mediated apoptosis of L929 cells also resulted in p53⅐IB␣ dissociation in the cytosol (Fig. 5B). Immunoprecipitation was then performed with anti-p53 IgG beads. One tenth (ϳ20 g) of the total proteins were loaded for SDS-PAGE and Western blotting using anti-p53 and anti-IB␣ antibodies (see Pre-IP). The cytosolic levels of p53 and IB␣ were not altered by staurosporine and Bay11-7085. Staurosporine reduced the level of IB␣ in the nucleus, and Bay11-7085 failed to prevent the reduction. In contrast to L929 and HeLa cells (see Fig. 5), there was an increased amount of IB␣ binding to p53 in both cytosol and nucleus (see IP with anti-p53). Staurosporine mediated the dissociation of p53⅐IB␣ complex, and Bay11-7085 could not prevent the dissociation. B, to induce hypoxic conditions, SK-N-SH cells were treated with deferoxamine (DFO, 100 M) for 16 h, followed by examining the presence of p53⅐IB␣ complex. Dissociation of the p53⅐IB␣ complex occurred when the cells were cultured under hypoxic conditions.
In contrast to the above-indicated cells, there was an increased amount of IB␣ associated with p53 in the cytoplasm and nucleus of unstimulated neuronal SK-N-SH cells (Fig. 6A). Staurosporine mediated p53⅐IB␣ dissociation but could not induce IB␣ degradation (Fig. 6A). Bay11-7085 (16), inhibitor of IB␣ phosphorylation, failed to prevent the dissociation (Fig.  6A), indicating that p53⅐IB␣ dissociation is not due to IB␣ phosphorylation and degradation. Again, Mdm2 was barely detectable in the immunoprecipitates (Fig. 6A). Staurosporinemediated p53⅐IB␣ dissociation, as well as p53 nuclear translocation, was also found in other types of cells such as Molt-4 T cells, breast MCF-7 cells, and prostate Du145 cells (data not shown).
To induce hypoxia, SK-N-SH cells were treated with deferoxamine for 16 h. p53⅐IB␣ dissociation occurred at both cytosolic and nuclear levels under hypoxic conditions (Fig. 6B).
In contrast to the above effects, TGF-␤1 failed to induce p53⅐IB␣ dissociation in L929 cells when cultured in the presence of serum (Fig. 7A). TGF-␤1 promotes L929 cell growth by 20 -30% in 24 -48 h (11). Similar results were observed using SK-N-SH cells (Fig. 7B). TGF-␤1 was not growth inhibitory to SK-N-SH cells (data not shown). Nonetheless, when L929 cells had undergone serum starvation for 16 -24 h and were then treated with TGF-␤1 for 1 h, the expression of cytosolic IB␣ was suppressed by TGF-␤1 and the cytosolic p53⅐IB␣ complex was barely detectable (Fig. 7A).
Exposure of L929 cells to etoposide, to induce DNA damage, resulted in increased p53 and IB␣ accumulation in the nucleus (Fig. 8A). Etoposide mediated dissociation of the cytosolic p53⅐IB␣ complex but increased the complex formation in the nucleus during 1-h treatment (Fig. 8A). However, at a prolonged treatment time of 16 h, continued p53 accumulation was observed in the cytoplasm and nucleus, whereas the amount of nuclear IB␣ binding to p53 was reduced (Fig. 8A).
Exposure of Du145 cells to UV light, followed by culturing for 1 h, resulted in the p53⅐IB␣ dissociation in the cytosol, as well as p53 accumulation in the nucleus (Fig. 8B). Time course studies using L929 and SK-N-SH cells also showed that UVmediated nuclear p53⅐IB␣ complex formation reached a plateau in 1-2 h and was then decreased (data not shown). Similar results were also observed by testing Molt-4 T cells.
The Non-ankyrin C Terminus of IB␣ Physically Interacts with the Proline-rich Region and Serine 46 of p53-The Cytotrap yeast two-hybrid system was used to map the structural basis of p53 and IB␣ binding interactions. The non-ankyrin C-terminal IB␣ (IB␣243C) physically interacted with the full-length region, the N-terminal region (amino acids 1-100), and the proline-rich region (amino acids 66 -100) of p53 (Fig. 9). However, IB␣243C could not bind the proline-rich regiondeleted p53 (p53⌬pro) (Fig. 9). Also, the full-length IB␣ interacted with the full-length p53 but failed to bind p53⌬pro (Fig.  9). These results indicate that the non-ankyrin C terminus of IB␣ physically interacts with the proline-rich region of p53. In comparison, the N-terminal ankyrin region of IB␣ could not bind the above-indicated p53 proteins (Fig. 9).
In positive controls, the binding interaction between WWdomain oxidoreductase (WOX1) and p53 (18), as well as MafB/ MafB self-interaction, was demonstrated (Fig. 9). In negative controls, no binding interactions were observed for Lamin C and p53 or empty vector and empty vector (Fig. 9).
Transcriptional activation of the mitochondrial apoptosisinducer p53AIP1 requires serine 46 phosphorylation in p53 FIG. 7. TGF-␤1 could not mediate dissociation of p53⅐IB␣ complex in proliferating L929 cells. A, L929 cells were cultured overnight in fresh medium, supplemented with 10% fetal bovine serum, and treated with TGF-␤1 (2 ng/ml) for 1 h. The cytosolic and nuclear fractions were prepared and immunoprecipitated using anti-p53 IgG beads. One tenth (ϳ20 g) of the total proteins were loaded for SDS-PAGE, and expression of p53 and IB␣ was examined. Again, IB␣ was co-precipitated with p53 in the cytosol, and TGF-␤1 could not mediate dissociation of the p53⅐IB␣ complex (left panel). L929 cells were cultured under serum-free conditions overnight, exposed to TGF-␤1 (2 ng/ml) for 1 h, lysed, and immunoprecipitated. TGF-␤1 rapidly down-regulated IB␣, whereas p53 levels in these cells were very low (right panel). The p53⅐IB␣ complex was hardly detectable in these starved cells. B, under similar conditions, SK-N-SH cells were treated with TGF-␤1 (2 ng/ml) for 1 h. TGF-␤1 could not dissociate the p53⅐IB␣ complex. C, however, TGF-␤1 mediated p53⅐IB␣ dissociation at both cytosolic and nuclear levels in Molt-4 T cells during 16-h treatment. (26). Deletion of serine 46 (p53⌬46) abolished p53 interaction with the full-length IB␣ and IB␣243C (Fig. 9). Alteration of serine 46 to a non-phosphorylation glycine-46 also abolished the binding interaction (Fig. 9). However, mutation of serine 46 to threonine 46, a phosphorylation site, failed to abolish the binding (Fig. 9). The observations suggest that phosphorylation of serine 46 is necessary for p53 interaction with IB␣. Also, the proline-rich region and serine 46 in p53 appear to contribute equally to the binding interaction with IB␣.
p53 and IB␣ Synergistically Mediates Apoptosis-Finally, whether IB␣ enhanced p53 apoptosis was examined. Transient expression of p53 and the full-length IB␣ in L929 cells resulted in enhancement of p53-mediated cell death (Fig. 10A). The cells were electroporated with a non-cytotoxic dose of p53 FIG. 8. DNA damage-mediated dissociation of p53⅐IB␣ complex in the cytosol but formation in the nucleus. A, L929 cells were treated with etoposide (50 g/ml) for 1 and 16 h. Immunoprecipitation was performed from the cytosolic and nuclear fractions using anti-p53 IgG beads (left panel). One tenth (ϳ20 g) of the total proteins were loaded for SDS-PAGE, and expression of p53 and IB␣ was examined. During a 1-h treatment, etoposide mediated dissociation of the cytosolic p53⅐IB␣ complex. However, etoposide increased the nuclear p53⅐IB␣ complex formation (left panel). During a 16-h treatment, etoposide increased p53 expression, whereas the amount of nuclear IB␣ binding to p53 was reduced (right panel). In this experiment, immunoprecipitation was performed using anti-p53 IgG antibody and protein A-agarose beads. B, Du145 cells were exposed to UV light and then cultured for 1 h. Immunoprecipitation showed that the cytosolic p53⅐IB␣ complex dissociated, whereas the complex formed in the nucleus.
FIG. 9. Serine 46 and the proline-rich region (amino acids 66 -100) of p53 physically interact with the non-ankyrin C terminus (amino acids 244 -317) of IB␣ (designated IB␣243C). Binding of bait protein to target protein in the cytosol results in the activation of the Ras signaling pathway and thus allows mutant cdc25 yeast to grow at 37°C in agarose containing galactose (Cytotrap yeast two-hybrid system). A, the full-length IB␣ physically interacted with the full-length p53, but not with p53 devoid of the proline-rich region (p53⌬pro) or serine 46 (p53⌬46). Alteration of serine 46 to glycine (p53S46G) abolished the binding. However, alteration of serine 46 to another phosphorylation residue, threonine 46, failed to abolish the binding. No interaction was observed between a dual mutation of p53 (p53⌬proS46T) and IB␣. B, the C terminus of IB␣ (IB␣243C) interacted with the N-terminal proline-rich region of p53, but not with p53⌬pro and p53⌬46. C, the N-terminal ankyrin region of IB␣ (IB␣243N; amino acids 1-243) failed to interact with the above-indicated p53 proteins. D, in positive controls, WOX1⅐p53 (18) and MafB/MafB self-binding interactions were tested. E, in negative controls, Lamin C⅐p53 and empty vector (pSos)/empty vector (pMyr) interactions were examined. and IB␣. Additionally, co-expression of p53 with the fulllength IB␣ or IB␣243C in COS7 cells also resulted in enhancement of p53-mediated cell death (Fig. 10B). Similar results were observed by testing Mv1Lu, HeLa, and Du145 cells (data not shown). DISCUSSION Functionally, IB␣ and other IB family proteins prevent NF-B activation by binding and sequestering NF-B in the cytoplasm (14,27,28). In addition, IB␣ binds the hepatitis B virus X protein and mediates nuclear import of this protein (29). Here, it is demonstrated that a portion of cytosolic IB␣ interacts with p53 in vivo, as determined using various types of cells and rat liver and spleen. Notably, the cytosolic p53⅐IB␣ complex rapidly undergoes dissociation in response to apoptotic stress, DNA damage, UV radiation, hypoxia, and TGF-␤1-mediated growth suppression, indicating a regulatory control of cytosolic p53 by IB␣ in resting cells. That is, IB␣ plays a role in sequestering p53 in the cytoplasm and preventing p53 nuclear translocation. In response to stress stimuli, dissociation of p53⅐IB␣ complex occurs and p53 translocates to the nucleus. Apparently, there is an IB␣-regulated p53 pool in resting cells and this p53 readily translocates to the nucleus in response to exogenous stress.
Unlike Mv1Lu and lymphoid cells, TGF-␤1 could not inhibit the growth of L929 fibroblasts and neuronal SK-N-SH cells.
TGF-␤1 failed to mediate the dissociation of the cytosolic p53⅐IB␣ complex. Indeed, the amount of cytosolic p53⅐IB␣ complex was increased by TGF-␤1 in L929 cells. Presumably, during growth response, a strict control of cytosolic p53 by IB␣ is needed for preventing p53 nuclear translocation and its induction of p21 for growth suppression (25).
As a tumor suppressor protein, p53 controls cell growth, repairs damaged DNA, maintains mitochondrial biogenesis, and mediates apoptosis (25). In unstimulated resting cells, the cytosolic and nuclear p53 levels are very low. A proposed model for p53 turnover in resting cells is that a portion of cytosolic p53 complexes with IB␣ and the complex may constantly dissociate to allow p53 nuclear translocation for maintaining optimal cellular function. The nuclear p53 then complexes with Mdm2 and is exported to the cytoplasm for ubiquitination and degradation (25). Newly synthesized p53 again complexes with IB␣ for serving as a new cytosolic stock.
An increased p53⅐IB␣ complex formation was observed in the nucleus during exposure of cells to etoposide and UV for 1 h. However, this event did not occur when cells were treated with staurosporine, indicating that specific exogenous signals are needed for the nuclear p53⅐IB␣ complex formation. When cells were treated for a prolonged time with etoposide, dissociation of the nuclear p53⅐IB␣ complex occurred, along with simultaneous p53 accumulation in the nucleus. The event is apparently necessary for p53-mediated apoptosis.
p53 and IB␣ synergistically enhance cell death during transient co-expression. To elucidate the underlying mechanism, it is necessary to determine the kinetics of complex formation of ectopic IB␣ and p53 in both cytoplasm and nucleus before and at the time of cell death. Also, inhibition of NF-B, an antiapoptotic factor (12), by ectopic IB␣ is likely. p53 apoptotic function may be enhanced due to suppression of NF-B function by IB␣. Nonetheless, the notion is not supported by a recent study showing that NF-B is a co-factor of p53 in mediating cell death (30).
Dissociation of the cytosolic p53⅐IB␣ complex is not due to IB␣ phosphorylation and subsequent ubiquitination for degradation. Inhibition of IB␣ phosphorylation by Bay11-7085 could not prevent staurosporine-mediated p53⅐IB␣ dissociation. Additionally, exposure of cells to EDTA failed to dissociate the cytosolic p53⅐IB␣ complex, indicating that the p53⅐IB␣ interaction is calcium-independent. Nonetheless, the experiments from the yeast two-hybrid interaction system showed that phosphorylation of serine 46 in p53 may be involved in the p53⅐IB␣ complex formation. Dissociation of this complex may be due to de-phosphorylation of serine 46 in p53, when cells are exposed to exogenous stress stimuli.
Structurally, IB␣ has a surface-exposed N terminus, a central, protease-resistant domain containing five ankyrin repeats, and a compact, highly acidic PEST domain-containing C terminus (27). The ankyrin repeats are responsible for the binding of IB␣ with NF-B/Rel proteins. IB␣ nuclear localization is independent of NF-kB/Rel proteins, and the second ankyrin repeat of IB␣ has been shown to be responsible for nuclear localization (28). Additionally, importins ␣ and ␤, the small GTPase Ran, and unidentified proteins that interact with the ankyrin repeats are involved in nuclear transport of IB␣ (27). Nuclear IB␣ inhibits the interaction of NF-B with target DNA and promotes the export of NF-B from the nucleus to the cytoplasm (27). The first ankyrin repeat of IB␣ has the strongest inhibitory effect on NF-B activation (14).
Two pathways are involved in IB␣ degradation. One pathway is the TNF-mediated activation of IKK kinases, which phosphorylates IB␣ (31). The phosphorylated IB␣ dissociates from the NF-B⅐IB␣ complex and is degraded by the protea- some/ubiquitin pathway. A parallel pathway is the TNF activation of cytosolic calpains, which degrades IB␣ and activates NF-B independently of the ubiquitin/proteasome pathway (32,33). In contrast to these observations, constitutive nuclear translocation of NF-B in B cells fails to result in degradation IB proteins (34). Unlike inflammatory cytokines, hypoxia, reoxygenation, and the tyrosine phosphatase inhibitor pervanadate activate NF-B and tyrosine phosphorylation of IB␣, whereas this event could not induce degradation of IB␣ by the proteasome/ubiquitin pathway (34 -36).
Whether TGF-␤1-mediated down-regulation of cytosolic IB␣ in Mv1Lu cells is due to degradation by the proteasome/ ubiquitin pathway or due to inhibition of gene transcription by the Smad protein complex remains to be determined. Previously we have shown that TIAF1, TGF-␤-induced antiapoptotic factor, inhibits IB␣ expression when overexpressed in L929 cells (9). TIAF1 suppresses gene expression of IB␣. 2 TGF-␤1 is not an activator of NF-B, and both proteins may counteract each other's functions (6,12,(37)(38)(39). Thus, inhibition of IB␣ expression by TGF-␤1 in Mv1Lu cells is likely due to TIAF1suppressed gene expression rather than IB␣ degradation by the proteasome/ubiquitin system. Suppression of IB␣ expression by TGF-␤1 allows p53 nuclear translocation for exerting growth arrest and apoptosis.
The functional significance of IB␣ interaction with the proline-rich or growth regulatory region of p53 is unknown. Berger et al. (20) showed that p53 without the proline-rich region becomes more susceptible to ubiquitination, nuclear export, and Mdm2-mediated degradation, indicating that this region is necessary for p53 protein stability. We demonstrated that WOX1 interacts with p53 at the proline-rich or growth regulatory region (18). Most recently, we found that WOX1 may interact with p53 at other regions. 2 WOX1 mediates apoptosis synergistically with p53. Suppression of WOX1 expression by antisense mRNA blocks p53 apoptosis, suggesting that WOX1 is a potential partner of p53 in mediating cell death (18). Whether WOX1 competes with IB␣ in binding to p53 is unknown. Nonetheless, the functional relationship among p53, WOX1, and IB␣ is intriguing and remains to be determined.