The Y-box-binding Protein, YB1, Is a Potential Negative Regulator of the p53 Tumor Suppressor*

The p53 tumor suppressor plays a major role in preventing tumor development by transactivating genes to remove or repair potentially tumorigenic cells. Here we show that the Y-box-binding protein, YB1, acts as a negative regulator of p53. Using reporter assays we show that YB1 represses transcription of the p53 promoter in a sequence-specific manner. We also show that YB1 reduces endogenous levels of p53, which in turn reduces p53 activity. Conversely, inhibiting YB1 in a variety of tumor cell lines induces p53 activity, resulting in significant apoptosis via a p53-dependent pathway. These data suggest that YB1 may, in some situations, protect cells from p53-mediated apoptosis, indicating that YB1 may be a good target for the development of new therapeutics.

Apoptosis, or programmed cell death, is the mechanism by which damaged, modified or superfluous cells are removed from a complex organism (1). Another important role for apoptosis is preventing the development of cancer by removing cells with mutated or damaged DNA in order to preserve the integrity of the genome (2). The tumor suppressor protein p53 plays a pivotal part in this and consequently has been termed the "guardian of the genome" (2). In response to DNA damage, p53 is "activated" and initiates either growth arrest or apoptosis pathways (3). This allows DNA damage to be repaired or potential tumor precursor cells to be removed from tissues. Both pathways are initiated probably via the transcriptional activation of specific genes by p53. These genes include p21 WAF1/CIP1 (4), GADD45 (5), and 14-3-3 (6), which induce cell cycle arrest, and Bax (7), killer/DR5 (8), and PIG3 (9), which induce apoptosis.
Consistent with its important tumor suppressor role, more than 50% of human cancers contain mutations in the p53 gene, encoding a protein that is inactive for some or all of the functions of p53 (10). In the other 50% of cancers there is selection against other components of the p53 signaling pathway, suggesting that p53 may be functionally inactive in the vast majority of cancers. Thus, there appears to be a powerful selection against functional p53 during tumor development.
The Y-box-binding protein, YB1, belongs to the family of cold shock proteins, which is highly conserved from bacteria to man. YB1 is multifunctional and appears to regulate gene expression at both the transcriptional and translational levels (11,12). With regard to transcription, we noticed that YB1 regulates some of the same genes as wild type (wt) 1 p53 (13)(14)(15)(16)(17)(18)(19) but in an opposing manner. For example, the fas gene promoter is repressed by YB1 (13) and stimulated by wt p53 (19), whereas the multidrug resistance (mdr1) gene promoter is activated by YB1 (14) and repressed by wt p53 (16). These data suggest that YB1 might negatively regulate p53.
In this study we present evidence that YB1 represses the p53 promoter and down-regulates endogenous p53 expression. We then show that a reduction of YB1 in several tumor cell lines results in an induction of apoptosis via the activation of a p53 pathway. If this occurs in vivo, then these observations suggest that YB1 may play a role in the development of some tumors by protecting cells from apoptosis induced by p53.
Plasmids-Rat YB1 was cloned into the mammalian expression vector pcDNA3 as described previously (13). The promoterless reporter construct (pCAT3M) contains the CAT gene but no upstream eukaryotic promoter sequences (20). The 446-bp mouse p53 reporter construct contains Ϫ320 to ϩ116 of the promoter cloned into pCAT3M as described previously (20). The adenovirus-5 E3 promoter-CAT reporter construct (pKCAT23) was as described previously (23). The Y-box deletion construct was prepared by deleting the region Ϫ143 to Ϫ121 from the mouse p53 promoter using inverse PCR as described previously (24). The Ϫ93 to Ϫ51 deletion mouse p53 reporter construct has been described (24). The p21 reporter constructs contain the promoter of the p21 WAF1/CIP1 gene linked to the luciferase reporter and a variant promoter (p21⌬53) without the p53 binding site.
Reporter Gene Assays-Two ϫ 10 5 cells were transfected with 1 g of reporter plasmid and 1, 2, or 3 g of rat YB1 expression construct (13) using FuGENE 6 reagent (Roche Molecular Biochemicals). Control transfections were performed using the equivalent amount of either pUC19 or pCDNA3 empty vector (Invitrogen). Cells were harvested at 48 h post-transfection. Samples to be assayed for CAT activity were lysed in 0.25 M Tris-HCl, pH 7.5, by repeated freeze-thaw cycles, and cellular debris was removed by centrifugation. Supernatants were normalized for protein content using a BCA protein assay reagent kit (Pierce), and CAT and luciferase activities were determined. CAT activity was measured according to a previously described method (27), and luciferase activity was measured according to the manufacturer's instructions (Promega) and then standardized to cell number.
Electrophoretic Mobility Shift Assay-Nuclear extract preparation and electrophoretic mobility shift assays were performed as described previously (13).
Oligonucleotide Transfections-Cells were washed twice with serumfree medium prior to transfection with 160 pmol/ml each oligonucleotide and 6.7 l/ml Lipofectin (Invitrogen) except for HepG2, which was transfected with 10 l/ml Lipofectin. The cells were incubated for 5 h at 37°C, washed once with the appropriate medium supplemented with 10% fetal bovine serum, and recovered in fresh medium supplemented with 10% fetal bovine serum until harvested. The control oligonucleotide was used at an equimolar concentration. All transfections were performed in replicates of two or more, and all experiments were performed at least twice.
Apoptosis Analysis-After harvesting, cell pellets were resuspended in phosphate-buffered saline and 0.2% (w/v) trypan blue. Viable cells were determined by trypan blue staining and counted in quadruplicate on a hemocytometer. Cells undergoing apoptosis were analyzed by in situ TUNEL (terminal deoxynucleotidyl transferase-mediated dUTP nick end-labeling). Assays were performed using the In Situ Cell Death Detection Kit, Fluorescein (Roche Molecular Biochemicals) following the manufacturer's instructions, except that cell permeabilization was performed using a microwave as described previously (28). Cells were visualized by fluorescent microscopy and Nomarsky optics. DNA fragmentation of the transfected cells was analyzed by a previously described procedure (29).

YB1 Can Repress p53
Transcription-Analysis of a 446-bp region of the mouse p53 promoter (20) identified several potential Y-box sequences (30), one of which is between nucleotides Ϫ138 and Ϫ127 bp relative to the start site of transcription (Fig. 1A). The mouse and human p53 promoters are conserved within this region, although their transcriptional start sites differ. The putative YB1 site shows an 8 of 11 bp identity between the two promoter species (20). As YB1 is a singlestranded DNA-binding protein (31), single-stranded oligonucleotides spanning this region were tested in electrophoretic mobility shift assays with purified YB1 protein. These assays showed that YB1 binds this Y-box-like region of the p53 promoter although preferentially to the pyrimidine-rich upper strand (Fig. 1B). As two shifted complexes were observed, competition with a known YB1 binding site, the YB1 cis-element of the human fas promoter (13), was used to confirm the specificity of the interaction. The slower migrating complex was effectively competed out with increasing amounts of single-stranded competitor DNA (Fig. 1B). To further confirm specificity, oligonucleotides designed to other regions of the p53 promoter showed no YB1 binding (data not shown). These results demonstrate that YB1 specifically binds to certain sequences within the mouse p53 promoter in vitro.
Specific regulation of p53 transcription by YB1 in vitro was demonstrated using reporter gene assays. The human melanoma cell line SK Mel 5, was co-transfected with a 446-bp fragment of the p53 promoter linked to a CAT reporter (20) and a YB1 expression construct (13). YB1 was found to completely repress transcription from the p53 promoter (equivalent to the level of the promoterless control), whereas transcription from the adenovirus-5 E3 promoter (23) was not repressed by YB1 (Fig. 1C). In addition, the cytomegalovirus and proliferating cell nuclear antigen promoters also were not repressed by cotransfected YB1 (data not shown). This observation was extended to a further three cell lines of mouse and human origin, where between 75 and 95% of the p53 promoter activity was abolished following co-transfection with YB1 ( Fig 1D). Similar results were obtained when a human p53 reporter construct (32) was used in these same cells (data not shown). These data show that YB1 is capable of repressing transcription of both mouse and human p53 promoters in what appears to be a sequence-specific manner.
To further explore the sequence specificity of this repression, YB1 was transfected into SK Mel 5 cells along with two deletion constructs of the mouse p53 promoter, one of which is deleted for the putative Y-box binding site (Ϫ143 to Ϫ121 bp relative to the transcriptional start site; see Fig 1A) and another deleted between Ϫ93 and Ϫ51 bp relative to the start site. Results ( Fig 1E) show that YB1 was able to repress neither the Y-box deleted promoter nor a smaller deletion (Ϫ138 to Ϫ121 bp) within this region but was still able to repress transcription from the Ϫ93 to Ϫ51-bp construct. This latter result excludes structural effects as being responsible for loss of YB1 repression. We conclude that YB1 is capable of repressing transcription of the p53 promoter in vitro. YB1 Reduces p53 Expression and Activity in Transfected Cells-To determine whether YB1 also reduces expression of endogenous p53, A549 and SK Mel 5 cells, both expressing wild type p53 protein, were transfected with a YB1 expression construct, and p53 levels were measured. In the first part of this investigation, Western blotting was carried out. Results for A549 cells (Fig 2A) show that actin protein levels were essentially invariant after transfection of YB1, whereas by 18 h p53 levels were reduced to 75% (1 g of YB1 plasmid) and 65% (2 g of YB1 plasmid) of control and were further reduced to about 50% of control by 48 h after transfection. This was also evident at the same times after transfection of SK Mel 5 cells (Fig 2A) in which, for example, p53 levels were reduced to 50% of control by 18 h after transfection with 2 g of YB1 plasmid and actin levels were unchanged. A Northern blot was also done, and consistent with the protein experiments, YB1 reduced p53 mRNA levels (data not shown). Finally, the effect of transfec-tion of YB1 on p53 function was measured. A549 cells were transfected with YB1 along with a luciferase reporter construct linked to the p21 WAF1/CIP1 promoter, known to be transactivated by wt p53. In parallel, a similar experiment was carried out using a mutant p21 WAF1/CIP1 reporter construct without the p53 response element (p21⌬53). Results (Fig 2B) show that YB1 markedly reduces activity of the p21 WAF1/CIP1 promoter but has no effect on the activity of the p21⌬53 reporter. In other FIG. 1. YB1 binds to and represses the p53 promoter. A, schematic diagram showing the transcription factor binding sites located within a 446-bp region of the mouse p53 promoter. The transcriptional start site is shown as an arrow at ϩ1. The location of a Y-box sequence within the promoter is indicated as a solid box. B, glutathione S-transferase-tagged rat YB1 was incubated with single-stranded ␥-32 P end-labeled oligonucleotides containing the putative Y-box from the mouse p53 promoter. DNA complexes formed were analyzed by nondenaturing gel electrophoresis. Lanes A and B represent oligonucleotide probes corresponding to the lower (Strand 2) and upper strand (Strand 1), respectively. The specificity of complexes formed with Strand 1 was determined by competition with an excess of a single-stranded YB1 binding site from the human fas promoter (lanes C and D). Specific complexes formed are indicated by a single arrowhead and free probe by a double arrowhead. C, a 446-bp mouse p53 promoter-CAT reporter gene construct was co-transfected into SK Mel 5 cells with either control DNA (filled bars) or 1 g of a YB1 expression construct (hatched bars). The level of CAT reporter gene activity was determined for each transfection and is shown as 14 C-acetylated chloramphenicol. The adenovirus-5 E3 promoter and a promoterless control were also tested to determine specificity. Values are the mean and standard errors from four independent experiments. D, the 446-bp mouse p53 promoter-CAT reporter gene construct was co-transfected into mouse NIH3T3 or human HeLa or RKO cells with either 3 g of pCDNA3 control (filled bars) or 1 g (light hatched bars) or 3 g (dark hatched bars) of the YB1 expression construct. The level of CAT activity was determined for each transfection and is displayed as a percentage of the pCDNA3 control values. Values are the mean and standard errors from two independent experiments. E, the 446-bp mouse p53 promoter-CAT reporter construct lacking the putative Y-box sequence (Y-box Deln) or lacking the region Ϫ93 to Ϫ51 was co-transfected into HeLa cells with either 3 g of pCDNA3 control (filled bars) or 3 g of YB1 expression construct (hatched bars). Values are the mean and standard error from two independent experiments. experiments, YB1 does not inhibit p21 WAF1/CIP1 promoter activity in Skov3 and Saos2 cells, both of which are p53-deficient (data not shown). These data show that the inhibition of p21 WAF1/CIP1 promoter activity by exogenous YB1 parallels the decline in p53 expression and is not a direct effect of YB1. The experiments shown in both Figs. 1 and 2 are consistent with the interpretation that YB1 down-regulates p53 expression, which in turn leads to reduced p53 activity.
Reducing YB1 Up-regulates p53-Because over-expression of YB1 leads to reduced p53 activity, the converse experiment, in which YB1 levels are reduced, may lead to up-regulation of p53. To test this possibility we used a YB1 cis-element from the human fas promoter (13), which should effectively function as a YB1 decoy to sequester YB1 from endogenous promoters. This approach has been used successfully for several transcription factors (33,34). An equimolar concentration of a random sequence oligonucleotide was also transfected as a negative control. First, the endogenous levels of p53 and the pro-apoptotic p53-regulated gene Bax (7) were determined by Western blotting. The results for transfected SK Mel 5 cells (Fig 3A) show that the levels of p53 increased 3-4-fold in cells receiving the decoy oligonucleotide, peaking at 4 h post-transfection. The levels of Bax also increased about 3-fold, peaking at 4 -6 h post-transfection, whereas the levels of ERK1 remained unchanged. In contrast, there was essentially no change in the levels of p53, Bax, or ERK1 after transfection with control oligonucleotides (Fig. 3A).
We next asked whether the YB1 decoy would increase a p53 response as measured by the p21 WAF1/CIP1 reporter constructs used in Fig 2A. A549 cells were transfected with the p21 WAF1/CIP1 reporter constructs along with decoy or control oligonucleotides or with a human p53 expression construct. The results (Fig 3B) show that both human p53 and YB1 decoy increased the activity of the p21 WAF1/CIP1 reporter, but this did not happen with control oligonucleotides. Furthermore, only the reporter with the p53 response element was up-regulated, showing that the effect of the oligonucleotides was p53-specific.
To confirm that the decoy oligonucleotide was specifically able to target YB1, SK Mel 5 cells were transfected with biotinylated decoy or control oligonucleotides. At 18 h posttransfection, oligonucleotides were removed by incubating lysates from transfected cells with streptavidin beads. The level of YB1 remaining in the supernatant of cells that had received the decoy oligonucleotide was 60% less than that found in control lysates (Fig 3C), but ERK1 levels were the same. This experiment therefore confirms the specificity of the decoy oligonucleotide.
Reduction of YB1 Induces Cells to Undergo Apoptosis-The results described above show that reducing YB1 causes an up-regulation of p53. We therefore asked whether reducing YB1 can cause cells to undergo apoptosis in a p53-dependent manner. At 18 h post-transfection of SK Mel 5 cells with either antisense YB1 or YB1 decoy oligonucleotides, about 30% of cells were in the late stages of apoptosis as assessed by in situ TUNEL assay (Fig 4A) and DNA laddering (Fig 4B). However, much less apoptosis was observed with the control oligonucleotide. To confirm that the antisense YB1 oligonucleotides did in fact reduce the level of YB1 protein, SK Mel 5 cell lysates were prepared from antisense and negative control oligonucleotide transfectants. Western blotting with a YB1 antibody showed that in cells transfected with the YB1 antisense oligonucleotide, 75% less YB1 protein was detected than from cells transfected with control oligonucleotide (Fig. 4C), whereas ERK1 levels were unchanged. This confirmed that the antisense YB1 oligonucleotides were able to target YB1 specifically.
To determine whether wt p53 is required for the apoptosis observed when YB1 is reduced by the antisense or decoy oligonucleotides, RKO and RKO p53.13 cells were used. RKO are colorectal cells expressing wt p53, and RKO p53.13 are a variant cell line containing a dominant-negative mouse p53 mutant (25). Thus RKO p53.13 are functionally p53 null. Both cell lines were transfected with the YB1 decoy or negative control oligonucleotides. Results (Fig 4D) show that at 18 h posttransfection there was a 53% decline in the viability of RKO cells when transfected with the YB1 decoy compared with the control oligonucleotide, however, there was no reduction in viability of RKO p53.13 cells. These results provide evidence that the induction of apoptosis caused by reduction of YB1 is via a p53-mediated pathway. This experiment was repeated in a panel of human and mouse tumor cell lines that differed in their p53 status. Apoptosis/cell death was measured by TUNEL assay and/or by trypan blue exclusion. Some of the TUNEL staining results are shown in Fig 4E, and a summary of these and other results is shown in Table I. Collectively these experiments show that at 18 h post-transfection, Ͼ30% apoptosis occurred after the addition of decoy oligonucleotide to cells expressing wt p53, but no apoptosis was observed in cells with a mutant p53 (e.g. HaCAT, Fig 4E, Table I) or in cells expressing no p53 (e.g. HL60, Table I). Additionally, we were also able to prevent YB1 decoy-induced apoptosis by co-transfecting a p53 decoy (a double-stranded p53 oligonucleotide) from the human GADD45 promoter (see "Experimental Procedures" for the sequence) but not with the control oligonucleotide (data not shown).
The effect of transfection of antisense and decoy oligonucleotides was also examined at later times. We generally found that more extensive cell death occurred by 42 h post-transfection (about 50%), but was no greater beyond this time (data not shown). This may be because of oligonucleotide instability, but the reasons for this are not clear. This qualification notwithstanding, the above experiments show that apoptosis/cell death is induced by inhibiting YB1 and that this appears to require wt p53. DISCUSSION There is growing evidence that the Y-box-binding protein, YB1, may be important in tumor biology. YB1 is activated in response to genotoxic stress (35,36) and is associated with drug resistance (35), and both YB1 protein level (37) and nuclear localization (38) appear to be prognostic for some human cancers. In addition, YB1 regulates expression of several tumor-associated genes. These include epidermal growth factor receptor (EGFR or c-ErbB1), matrix metalloproteinase-2 (MMP-2), fas, mdr1, DNA topoisomerase II␣ and MHC Class II (13-15, 39, 40). As an activator of MMP-2, elevated levels of YB1 would facilitate tumor cell invasion and metastasis, and as a promoter of EGFR and mdr1 gene expression, YB1 would enhance cell growth and resistance to chemotherapeutic agents. The p53 tumor suppressor regulates a similar cluster of genes (16 -19, 41, 42), but unlike YB1, it represses the expression of MMP-2, EGFR, and mdr1 and promotes the expression of fas. However, as p53 regulates these genes to (presumably) protect against tumor development, we postulated that YB1 may function as a dominant-negative regulator of p53. This is the subject of the present study, for which we provide some evidence.
In this paper we have shown that YB1 represses transcription of the promoter of the p53 gene and that this occurs in a sequence-specific manner (Fig. 1). We also have shown that YB1 bound preferentially to the upper pyrimidine-rich strand, consistent with other data showing it to be a single-stranded DNA-binding protein (13,30,31). Despite binding singlestranded DNA, our data suggesting that YB1 transcriptionally regulates the p53 promoter are not unique, as YB1 has been reported to transcriptionally regulate a number of genes (13, 21, 30, 31, 43) as do other single strand binding proteins such as Pur␣, Pur␤ (44,45), and the polypyrimidine tract-binding protein (46). In addition to the reporter experiment, we also showed that transfection of YB1 caused a down-regulation of endogenous p53 and a concomitant reduction in the ability of p53 to transactivate the p21 WAF1/CIP1 promoter (Fig. 2). These data suggest that YB1, among other genes it regulates, might also be a negative regulator of p53, consistent with our hypothesis. In other experiments, the reduction of YB1 levels using antisense oligonucleotides, or an inhibition of YB1 activity using decoy oligonucleotides, resulted in an up-regulation of p53 (Fig. 3) and induction of apoptosis in a range of tumor cell lines that appeared to be p53-dependent (Fig. 4, Table I). A recent publication has shown that YB1 and p53 form a stable protein complex and that this results in increased binding of p53 to its response element in the p21 WAF1/CIP1 promoter (47). It was also shown that transfection of an antisense YB1 expression construct led to a reduction in p53-dependent promoter activity (47), completely the opposite of what we report here. However, our experiments have been repeated several times in different cells and with other p53-dependent promoters (48). We have never seen an up-regulation of p53 activity with YB1 in any of our experiments, although our cells were not treated with cisplatin as was used in the above report (47), which may well be an important difference. Nonetheless, a priori it is possible that the down-regulation of the p53 promoter and endogenous p53 levels ( Figs. 1 and 2) is due to a p53-YB1 complex. However, this seems unlikely, as the p53 promoter repression by YB1 also occurs in the p53-defective cells (HeLa (Fig. 1), RKO p53.13, Saos2, and Skov3 (data not shown)). Nevertheless, there are reports that YB1 and p53 regulate expression from the same site in the mdr1 promoter (14,47,49) and also interact with a third transcription factor, AP2␣, to transactivate the MMP-2 promoter (49). Thus, although YB1 and p53 may not interact to regulate expression of p53 itself, YB1 may bind p53 to regulate expression of other genes.
Although the down-regulation of p53 as we have reported here appears to be best explained by a transcriptional mechanism, consistent with other reports implicating a role for YB1 in transcriptional regulation (14,47,49), such an interpreta-tion raises a paradox. Because YB1 is located in the cytoplasm (50,51), how can it down-regulate gene expression by a transcriptional mechanism? At this time, one can only speculate on the answer. However, although the majority of YB1 is in the cytoplasm at any point in time, as it is in our experiments using immunofluorescent microscopy (data not shown), recent evidence has shown that YB1 can shuttle between the nucleus and the cytoplasm depending on the functional status of the splicing factor SRp30c, which binds YB1 (52), and on p53 (48). Thus, there is always likely to be a proportion of YB1 in the nucleus that could account for our observations and those of others, even if it is not easily detected. Indeed, this must be the case, as we have shown that nuclear localization of YB1 is linked to an inhibition of p53 activity (48) and that up-regulation of mdr1 by YB1 is observed only when YB1 is in the nucleus (53). However, this issue needs to be explored further.
The observations that YB1 binds p53 and may regulate p53 transactivation ability, as well as its role in cooperating with p53 and AP2␣ to regulate MMP-2 transcription (49) together with the data presented here and in separate publication (48), serve to emphasize that interactions between p53 and YB1 may have important consequences for the functioning of both proteins. These interactions may also have some impact on tumor development. In this context, one implication of our findings is that in a proportion of tumors that retain a functional wt p53, there would be selection in favor of increased YB1 expression in order to neutralize a p53 response. To this end we examined a small panel of tumors and cell lines to determine whether this was the case. We found over-expression of YB1 in five of seven randomly selected tumors compared with corresponding normal tissue. In addition, in three of six wt p53 tumor cell lines we found a reciprocal relationship between p53 and YB1 expression levels (data not shown). Such data are consistent with our hypothesis, but clearly there are exceptions, and a large panel of well characterized cell lines and tumors need to be examined before a firm conclusion can be reached.
Inactivation of p53 is thought to be a common and necessary step in the development of most common human malignancies. About 50% of all tumors have mutant p53 genes (10), and a significant proportion of the remainder has other alterations that lead to functional inactivation of the wt p53 protein (54,55). In addition to these, recent evidence suggests that the dysregulation of p53 gene transcription by HoxA5, Pax, and Maf proteins might also be important (56 -58) in mediating a p53 response, which may be relevant to tumor development.
Our data suggest that YB1 should join this list and, in addition, that it may provide a suitable target for new therapeutic development.