Ras-mediated Loss of the Pro-apoptotic Response Protein Par-4 Is Mediated by DNA Hypermethylation through Raf-independent and Raf-dependent Signaling Cascades in Epithelial Cells*

The apoptosis-promoting protein Par-4 has been shown to be down-regulated in Ras-transformed NIH 3T3 fibroblasts through the Raf/MEK/ERK MAPK pathway. Because mutations of the ras gene are most often found in tumors of epithelial origin, we explored the signaling pathways utilized by oncogenic Ras to down-regulate Par-4 in RIE-1 and rat ovarian surface epithelial (ROSE) cells. We determined that constitutive activation of the Raf, phosphatidylinositol 3-kinase, or Ral guanine nucleotide exchange factor effector pathway alone was not sufficient to down-regulate Par-4 in RIE-1 or ROSE cells. However, treatment of Ras-transformed RIE-1 or ROSE cells with the MEK inhibitors U0126 and PD98059 increased Par-4 protein expression. Thus, although oncogenic Ras utilizes the Raf/MEK/ERK pathway to down-regulate Par-4 in both fibroblasts and epithelial cells, Ras activation of an additional signaling pathway(s) is required to achieve the same outcome in epithelial cells. Methylation-specific PCR showed that the par-4 promoter is methylated in Ras-transformed cells through a MEK-dependent pathway and that treatment with the DNA methyltransferase inhibitor azadeoxycytidine restored Par-4 mRNA transcript and protein levels, suggesting that the mechanism for Ras-mediated down-regulation of Par-4 is by promoter methylation. Support for this possibility is provided by our observation that Ras transformation was associated with up-regulation of Dnmt1 and Dnmt3 DNA methyltransferase expression. Finally, ectopic Par-4 expression significantly reduced Ras-mediated growth in soft agar, but not morphological transformation, highlighting the importance of Par-4 down-regulation in specific aspects of Ras-mediated transformation of epithelial cells.

Proper development and tissue homeostasis are regulated in part through a complex balance between cell proliferation and cell death. Any disruption in the processes that control this balance can lead to disease states such as autoimmunity, neu-rodegenerative disorders, and cancer (1)(2)(3). Therefore, an understanding of the molecules involved in regulating cell death is crucial for prevention and treatment of these diseases.
The par-4 (prostate apoptosis response-4) gene was identified originally in a cDNA screen for genes up-regulated in prostate cancer cells that were induced to undergo apoptosis (4). Par-4 is expressed ubiquitously in a variety of cell lines and rat organs, suggesting a physiologically common function for Par-4 in most cells (5). However, although staining of Par-4 has been observed in terminally differentiated cells, consistent with its role in apoptosis, Par-4 staining has also been seen in areas enriched in stem cell populations, suggesting that Par-4 may possess additional functions that remain to be defined (5).
par-4 gene and protein expression has been found to be downregulated in Ras-transformed NIH 3T3 mouse fibroblasts and by transient expression of activated H-Ras in human fibroblasts and HeLa cervical carcinoma cells (6 -8). The ras gene family encodes small GTPases involved in regulation of cytoplasmic signaling pathways in response to diverse extracellular signals. ras gene point mutations of codons 12, 13, and 61 are found frequently in human cancers and result in persistent activation of Ras due to impaired GTPase activity (9,10). A consequence of constitutive Ras activation is downstream changes in gene expression, which in turn act to modulate a variety of processes, including proliferation, differentiation, angiogenesis, and apoptosis (11). Understanding how Ras subverts death is critical to understanding the role of Ras in cancer. Ras is known to cause a shift in the growth/apoptosis balance to favor growth over death and in doing so promotes cell proliferation (12). Ras achieves growth-favoring conditions in part by altering the expression of pro-apoptotic genes to promote tumorigenesis. Thus, oncogenic Ras protects cells from apoptosis by modulation of a variety of apoptosisrelated proteins, including down-regulation of Par-4 expression (6,7). Consistent with this possibility, transient Par-4 expression causes an apoptotic response in Ras-transformed NIH 3T3 cells (7).
Ras mediates its actions by interaction with multiple downstream effectors, with the best characterized being the Raf serine/threonine kinases (11,13). Ras binds to and activates Raf, which in turn phosphorylates and activates MEK1 1 and MEK2, which then activate the ERK1 and ERK2 MAPKs. The downstream consequences of this signaling pathway that are important in regulating cell survival are not clearly defined (14). The second best characterized effector of Ras is phosphatidylinositol 3-kinase (PI3K), a lipid kinase that facilitates the conversion of phosphatidylinositol 4,5-bisphosphate to phosphatidylinositol 3,4,5-triphosphate (15,16). In turn, phosphatidylinositol 3,4,5-triphosphate promotes activation of the Akt/ protein kinase B serine/threonine kinase pathway and a cell survival pathway (17). Phosphatidylinositol 3,4,5-triphosphate can also activate guanine nucleotide exchange factors (e.g. Sos and Vav) that activate the Rac small GTPase (18,19), and a prosurvival function for Rac has been described (20). Rac can activate the JNK and p38 MAPKs, which are also activated by apoptosis-inducing stimuli (21,22). Although Raf and PI3K represent the most studied effectors of Ras, other less well understood effectors such as the guanine nucleotide exchange factors (GEFs) for the Ral small GTPases (23) and the proapoptotic NORE1/RASSF1 family proteins (24) may also facilitate the ability of Ras to regulate cell survival and apoptosis.
Previous analyses of Ras-transformed NIH 3T3 cells have provided some indication of the mechanism of Par-4 downregulation by Ras (6 -8). Activated H-Ras was shown to downregulate Par-4 through the ERK MAPK pathway based in part on studies demonstrating that stable expression of activated Raf-1, MEK1, or ERK2 in NIH 3T3 cells causes down-regulation of Par-4 expression, whereas treatment with the MEK inhibitor PD98059 blocks Ras-mediated loss of Par-4. In contrast, activated PI3K alone does not cause down-regulation of Par-4 expression, and inhibition of PI3K does not restore Par-4 expression in Ras-transformed NIH 3T3 cells. In addition, ectopic expression of Par-4 was shown to block Ras-mediated growth transformation, to inhibit tumor development, and to induce apoptosis in transient assays by inhibiting NF-B (6)(7)(8). Thus, these observations support an important contribution of loss of Par-4 expression in Ras-mediated transformation and that Raf is the key effector utilized by Ras to mediate Par-4 down-regulation in rodent fibroblasts.
Although rodent fibroblast cell lines are transformed readily by activated Ras or Raf, we (25,26) and others (27)(28)(29) have shown that a variety of epithelial cell types (RIE-1, IEC-6, MCF-10A, and human embryonic kidney (HEK)) are transformed by Ras, but not by Raf. Instead, activation of Ral GEF alone is sufficient to mediate Ras transformation of human fibroblasts and epithelial cells (25). Thus, the signaling cascades that promote Ras transformation exhibit important cell type differences (11). Therefore, we initiated studies to determine the mechanism by which Ras causes downregulation of Par-4 and to assess the importance of Par-4 in Ras-mediated transformation of epithelial cells. We found that, although activation of the Raf/MEK/ERK pathway is necessary for par-4 down-regulation, activated Raf alone is not sufficient to mediate this process. We also show that the decrease in Par-4 transcript levels is due to promoter silencing through DNA hypermethylation. Importantly, Ras-mediated soft agar growth (but not morphological transformation) was significantly impaired upon ectopic Par-4 expression, demonstrating that down-regulation of Par-4 is a necessary prerequisite for specific aspects of Ras-mediated transformation of epithelial cells.
RNA Isolation and Northern Blot Analysis-Total RNA from cultured cells was isolated by the guanidine thiocyanate/acid/phenol method (36). For Northern blot analyses, 25 g of total RNA was size-fractionated on 1.4% formaldehyde gels, transferred to Hybond-N nylon membrane (Amersham Biosciences), and hybridized to 32 P-labeled DNA probes. The par-4 cDNA probe used was a fragment corresponding to the 1.4-kb EcoRI insert from pCB6 ϩ (described above). Rat plasmid cDNA clones for dnmt1, dnmt3a, and dnmt3b used for generating probes were kindly provided by Dr. E. Li (Harvard University). Verification of equivalent loading of RNA was done by hybridization with an oligonucleotide corresponding to the 28 S rRNA (Clontech). Hybridization was carried out by incubation with 1 ϫ 10 6 cpm/ml in PerfectHyb Plus hybridization buffer (Sigma) at 68°C for par-4 and dnmt1 and at 40°C for the 28 S rRNA oligonucleotide according to the manufacturer's recommendations.
DNA Methylation Analysis-To assess the role of DNA methylation in Ras-mediated down-regulation of Par-4, cells were incubated in growth medium supplemented with 2 M 5-aza-2Ј-deoxycytidine (Sigma) or Me 2 SO for 6 days. Azadeoxycytidine is an inhibitor of cytosine methyltransferase that methylates CpG islands in genomic DNA (37). Methylation patterns in the CpG island of par-4 were assessed by chemical bisulfite modification as described previously (3). In this procedure, unmethylated (but not methylated) cytosines are converted to uracil. Methylation-specific PCR was performed with primers specific for either methylated or bisulfite-modified unmethylated DNA. DNA methylation patterns in the CpG island of the par-4 gene were determined using multiple sets of primers. The primers specific for methylated DNA, which were designed based on DNA sequence information (GenBank TM accession number AF503628) to target a CpG-rich region Ϫ4466 bp upstream of the transcription start site (5), were as follows: GTTGGCGTAGGGTAGGTTGTAGC (forward) and GAAACACGAAC-GAAAAAAACCGA (reverse). Those specific for unmethylated DNA were as follows: GGTTGGTGTAGGGTAGGTTGTAGTG (foward) and CCAAAACACAAACAAAAAAAACCAAA (reverse). PCR was optimized for each primer pair. DNA treated in vitro with SssI methyltransferase (New England Biolabs Inc.) was used as positive control for methylated DNA and that from normal lymphocytes was used as a negative control for unmethylated DNA in PCR with the methylation-specific par-4 primers.
For sample preparation, 1 g of DNA isolated from HEK cells stably expressing vector or activated Ras was denatured by NaOH and modified by sodium bisulfite. DNA samples were then purified using Wizard DNA purification resin (Promega Corp.), again treated with NaOH, precipitated with ethanol, and resuspended in water. PCRs without DNA were performed as a control. Each PCR product (10 -12 l) was directly loaded onto nondenaturing 6% polyacrylamide gels or 4% agarose gels consisting of 2% ultrapure agarose (Invitrogen) plus 2% Sea-Plaque low melting agarose (BMA Products), stained with ethidium bromide, and visualized under UV illumination.
Focus Formation and Soft Agar Colony Formation Assays-The ability of Par-4 to inhibit K-Ras(12V)-mediated focus formation was determined by cotransfection using Lipofectamine Plus (Invitrogen) and 50 ng of pZIP-K-ras(12V) with 0.5 ng of pBabe-Par-4 or empty vector, and the appearance of transformed foci was quantified after 21 days in culture. For photomicrographs, cells were fixed in 10% acetic acid and 10% methanol and stained in 0.4% crystal violet in 20% ethanol.

RESULTS
Previous studies of the relationship between oncogenic Ras transformation and Par-4 expression have focused primarily on analyses in mouse fibroblasts. In these studies, expression of oncogenic Ras resulted in nearly complete loss of detectable Par-4 protein expression (6 -8). To assess the relationship between Ras and Par-4 in epithelial cells, we examined Par-4 protein levels in several Ras-transformed epithelial cell lines. As described previously for NIH 3T3 mouse fibroblasts (6 -8), we observed that 208F rat fibroblasts expressing constitutively active K-Ras(12V) also showed down-regulated Par-4 protein levels (Fig. 1). Similarly, RIE-1, ROSE, HEK, and human mammary epithelial cells expressing either activated H-Ras(12V) or K-Ras(12V) exhibited a decrease in Par-4 protein expression (Fig. 1). Thus, Ras-mediated transformation of a variety of epithelial cell types as well as fibroblasts resulted in down-regulation of Par-4. Interestingly, fibroblasts expressed less endogenous Par-4 compared with the epithelial cells, suggesting that endogenous levels of Par-4 expression are cell type-dependent.
The reduced expression of Par-4 observed in response to the forced expression of mutant Ras led us to investigate whether a similar loss of Par-4 would occur in tumor cell systems with endogenous aberrant Ras activity and thus more closely reveal in vivo physiological oncogenic Ras conditions. To address this question, we evaluated Par-4 protein levels in a human colon tumor cell system in which oncogenic Ras was shown previously to be critical for tumorigenic transformation. Shirasawa et al. (34) used homologous recombination to knock out the single mutant K-ras(13D) allele present in DLD-1 human colon carcinoma cells and determined that loss of oncogenic Ras function results in a drastic impairment of their transformed phenotype both in vitro and in vivo. As shown in Fig. 1B, the DLD-1 colon tumor cell line, which harbors a naturally occurring mutation in K-ras(13D), showed lower levels of Par-4 compared with its genetic variant (DKO-3) in which the mutant K-ras allele was deleted. Thus, these data demonstrate that decreased levels of Par-4 are associated with naturally occurring aberrant Ras activity, suggesting that loss of Par-4 is an important event in the tumorigenesis of cells that display abnormal Ras function.
Because we observed that down-regulation of Par-4 protein levels is correlated with Ras transformation of multiple cell types, we next investigated whether this reduction is biologically important to Ras-mediated transformation. Loss of anchorage and density dependence growth is a key growth characteristic of tumor cells that distinguishes them from normal cells. The ability of tumor cells to continue to proliferate under conditions of confluency (density dependence) and in the absence of substratum attachment (soft agar growth) distinguishes tumor cells from normal cells (39). First, we examined the ability of Par-4 to inhibit Ras-mediated focus forming (density dependence) activity. In these experiments, we transfected RIE-1 cells with an expression vector encoding activated K-Ras(12V) either alone or together with Par-4 and evaluated the consequences of Par-4 expression on Ras focus forming activity. As shown in Fig. 2  To further evaluate how Par-4 antagonizes Ras transformation, we generated stable cell lines that ectopically express Par-4 to determine whether re-expression of Par-4 could inhibit or reverse Ras-mediated transformation. As shown in Fig. 3A, forced ectopic Par-4 protein expression was restored to levels similar to those seen in the control empty vector-transfected cells, but there was no change or reversion in the morphological transformation (Fig. 3B) compared with cells transfected with the empty vector. However, when we compared their ability to grow under anchorage-independent conditions, we found that restoration of Par-4 expression reduced the ability of Rastransformed RIE-1 cells to grow in semisolid agar by 55% (Fig.  3, C and D). In addition, re-expression of Par-4 reduced not only the number of colonies, but also the size of the colonies (Fig.  3E). Because Ras (but not Raf) results in down-regulation of Par-4, we also examined whether forced expression of dominant-negative Par-4 (40) in the Raf-expressing cells would render these cells transformed. However, we observed no change in morphology or ability to grow in soft agar (data not shown). Thus, because forced re-expression of Par-4 inhibited the soft agar growth of the Ras-transformed cells, we conclude that the reduced expression of Par-4 observed in Ras-transformed cells is biologically relevant and represents an important mechanism by which Ras mediates transformation.
To investigate the mechanism by which Ras down-regulates Par-4 in epithelial cells, we next evaluated the potential involvement of several signaling pathways commonly activated by Ras. Although the Raf/MEK/ERK MAPK pathway represents the most widely studied Ras effector pathway, other pathways such as the PI3K/Akt and Ral GEF effector pathways have also been shown to be important for Ras transformation (11,13). To determine whether activation of a specific effector pathway alone is sufficient to down-regulate Par-4 expression, we previously established ROSE cells stably expressing constitutively active forms of Raf-1 (membrane-targeted Raf-CAAX and N-terminally truncated Raf-22W), PI3K (p110-CAAX), and Ral GEF (Rlf-CAAX) and verified that they cause activation of ERK, Akt, and Ral-GTP, respectively (31). Activated Raf alone is sufficient to cause partial growth transformation of ROSE cells. In contrast to the reduced levels of Par-4 protein seen in H-Ras(12V)-expressing ROSE cells (Fig. 4A), there were no detectable reductions in Par-4 expression in ROSE cells expressing any one activated effector alone. Finally, similar to our results in ROSE cells, activated Raf alone was insufficient to cause loss of Par-4 expression in RIE-1 cells (Fig. 4A). Therefore, our results in ROSE and RIE-1 cells are in contrast to those described in fibroblasts, where Raf activation alone is sufficient to down-regulate Par-4 expression (6 -8). Thus, the mechanism by which Ras down-regulates Par-4 in epithelial cells requires activation of Raf-independent effector pathways.
To determine whether activation of a particular effector pathway is necessary to cause down-regulation of Par-4, we utilized several pharmacological inhibitors of MEK and PI3K to assess the importance of these pathways in Ras down-regulation of Par-4. Interestingly, we found that inhibition of ERK activation in both Ras-expressing RIE-1 and ROSE cells by treatment with the MEK inhibitors PD98059 and U0126 restored the level of Par-4 expression to that seen in control untransformed RIE-1 and ROSE cells (Fig. 4B). Inhibition of MEK also reverses Ras-induced growth transformation of RIE-1 (41) and ROSE (31) cells. In contrast, we found that the p38 MAPK inhibitor SB203580, the PI3K inhibitor LY294002, and the JNK inhibitor SP600125 had no effect on Par-4 expression. Thus, we conclude that, although activation of the ERK pathway alone is not sufficient to recapitulate the signaling cascade mediated by Ras to down-regulate Par-4, activation of ERK by Ras does appear to be a necessary component.
Other signaling pathways utilized by Ras include those activated by the Rho family of small GTPases, such as Rac1, RhoA, and Cdc42, which can be activated by Ras. Ras-mediated transformation of rodent fibroblasts has been shown to require the function of these GTPases (42). Because Rho family GT-Pases can either trigger cell death or promote cell survival depending on the cellular context (43) and because Par-4 is a known pro-apoptotic response protein, we sought to determine whether Rac1 or other small GTPases such as RhoA and Cdc42 could also down-regulate Par-4. For these analyses, we utilized mass populations of RIE-1 cells stably expressing constitutively active Rac1(61L) or RhoA(63L) or activated forms of the Dbl family proteins Vav and Dbl, which are GEFs and activators of Rho family GTPases (44). Vav is an activator of RhoA, Rac1, and Cdc42, whereas Dbl is an activator of RhoA and Cdc42 (45,46). Although activated Ras down-regulated Par-4, no change in Par-4 levels was observed with RhoA, Rac1, Vav, or Dbl (data not shown), suggesting that Ras does not mediate down-regulation of Par-4 through RhoA, Rac1, or Cdc42 in RIE-1 cells.
We recently demonstrated that Ras-mediated transformation causes down-regulation of ␣-tropomyosin gene expression and that the mechanism is through promoter methylation (44). Promoter methylation represents one common mechanism of epigenetic gene silencing in human carcinomas (47). For example, the expression of various tumor suppressor genes, including BRCA1, E-cadherin, MLH1, p16, VHL, and RB1, is inactivated by hypermethylation in many human cancers (48). Therefore, we assessed the possibility that promoter methylation is involved in oncogenic Ras suppression of Par-4 expression. To determine whether loss of Par-4 expression is due to DNA methylation, we first treated Ras-transformed RIE-1 cells with the azadeoxycytidine demethylating agent and determined whether Par-4 expression was restored. Azadeoxycytidine is a specific inhibitor of DNA methyltransferase (49,50) and has been used widely to assess the role of methylation in promoter silencing. As shown in Fig. 5A, Par-4 mRNA and protein expression was increased after treatment with azadeoxycytidine compared with that in untreated Ras-expressing RIE-1 cells. To extend these analyses, we also examined DNA methylation of Par-4 in azadeoxycytidine-treated ROSE cells overexpressing mutant Ras and in DLD-1 cells, which harbor an endogenous mutant ras allele. As shown in Fig. 5B, azadeoxycytidine treatment also caused an increase in Par-4 expression in both ROSE and DLD-1 cells. Thus, these results demonstrate that promoter methylation may be a general mechanism of Rasmediated down-regulation of Par-4.
Interestingly, azadeoxycytidine treatment also causes a partial reversion of the morphological transformation of both Rastransformed RIE-1 (44) and ROSE (data not shown) cells, supporting the importance of DNA methylation in Ras-mediated transformation. To directly determine the methylation status of the par-4 promoter, we analyzed DNA obtained from Rasexpressing or vector only-transfected HEK cells by methyla- tion-specific PCR (Fig. 5C). HEK cells were chosen because more promoter DNA sequence is available for the human gene. The region chosen for methylation-specific PCR analysis spans an area of high CpG density Ϫ4466 bp upstream of the transcription start site (5). As shown in Fig. 5C, par-4 DNA was present in the unmethylated form in both vector only-and Ras-expressing HEK cells. Importantly, the methylated form of par-4 DNA was observed only in the Ras-expressing cells. These data are consistent with our observed partial but not complete down-regulation of Par-4 protein expression in the Ras-expressing HEK cells (Figs. 1 and 4A). Because treatment of the Ras-expressing HEK cells with MEK inhibitors restored Par-4 protein expression, we next investigated whether the mechanism for restored Par-4 expression is through a MEKdependent pathway, resulting in loss of DNA methylation of the par-4 DNA promoter. As shown in Fig. 5C, the methylated form of par-4 DNA was abolished in the Ras-expressing HEK cells treated with U0126 compared with control vehicle-treated cells. Taken together, these data show directly that expression of mutant Ras results in MEK-dependent down-regulation of Par-4 through CpG methylation of the par-4 DNA promoter.
Because Par-4 transcript levels were restored upon treatment with azadeoxycytidine, we assessed whether Ras causes a change in expression of three DNA methyltransferases involved in CpG hypermethylation. Dnmt1 methylates hemimethylated DNA and is therefore believed to be responsible for maintenance of methylation patterns. On the other hand, Dnmt3a and Dnmt3b are thought to be responsible for de novo methylation of previously unmethylated DNA. Thus, Dnmt3a and Dnmt3b may function as initiators of changes in DNA methylation patterns (51), although the exact function of these enzymes is presently not clear. To determine whether Ras up-regulates the methyltransferases, we performed semiquantitative reverse transcription-PCR on mRNA isolated from RIE-1 cells expressing Ras or the vector only. As shown in Fig.  6A, Ras-transformed RIE-1 cells showed an increase in both dnmt1 and dnmt3a, but not dnmt3b. To further support this finding, we performed Northern blot analyses and found that dnmt1 was indeed increased in Ras-expressing cells (Fig. 6B). We also tried to detect dnmt3a and dnmt3b; however, these genes were below the level of detection. These results suggest that Ras activation alone is sufficient to cause an increase in the enzymes responsible for genomic DNA hypermethylation, resulting in gene silencing of many genes, including par-4.

DISCUSSION
Emerging complexities of Ras signaling include the involvement of Raf-independent effectors in mediating Ras transformation, cell type differences in Ras signaling, and the complexity of genes deregulated in Ras-transformed cells (11). par-4 has been identified as a gene down-regulated by activation of the Raf/ERK MAPK cascade in Ras-transformed fibroblasts (6 -8). In light of differences in the signaling pathways involved in Ras transformation of fibroblasts versus epithelial cells, we evaluated the mechanism and role of Par-4 down-regulation in Ras transformation of epithelial cells. We found that Ras transformation of a variety of epithelial cells was associated with down-regulated Par-4 protein expression; that the Raf/ERK pathway was necessary but not sufficient for Ras down-regulation of Par-4 in rodent epithelial cells; that loss of Par-4 contributed to soft agar growth, but not morphological transformation by Ras; and finally, that CpG DNA methylation through a MEK-dependent pathway, possibly via up-regulation of the Dnmt1 and Dnmt3a DNA methyltransferases, was involved in suppression of par-4 gene expression.
Activated Ras and Raf show equivalent abilities to cause transformation of NIH 3T3 and other rodent fibroblast cell lines (26). In contrast, activated Ras (but not Raf) causes transformation of RIE-1 rat intestinal and other epithelial cells (25)(26)(27)(28)(29), thus demonstrating that the signaling cascades in fibroblasts and epithelial cells, while similar, exhibit important differences in signaling strategies and biological outcome (11). Recent reports utilizing NIH 3T3 cells demonstrated that the mechanism by which activated Ras causes down-regulation of Par-4 is by signaling through the Raf/MEK/ERK MAPK pathway (6 -8). Therefore, we assessed the contribution of the ERK MAPK and other signaling pathways to mediating Par-4 down- were maintained in growth medium supplemented with 2 M azadeoxycytidine for 6 days. Total RNA (25 g) from treated cells was sizefractionated on formaldehyde-agarose gels and analyzed by Northern blot analysis with 32 P-labeled DNA probes for Par-4 or the 28 S ribosomal RNA as a loading control. Cell lysates from companion treated cultures were used for Western blot analysis with anti-Par-4 antiserum. To control for loading, the anti-Par-4 antibody was stripped, and the membrane was reblotted with anti-actin serum. In parallel with the restoration of Par-4 mRNA expression (upper panels), Par-4 protein expression was restored after azadeoxycytidine treatment to levels comparable to those seen in control vector only-expressing cells (lower panels). B, azadeoxycytidine treatment of Ras-expressing ROSE and DLD-1 cells enhances Par-4 expression. ROSE cells stably expressing K-Ras(12V) and DLD-1 human colon tumor cells (which harbor an endogenous mutant ras allele, K-ras(13D)) were treated with 2 M azadeoxycytidine for 6 days, and protein lysates were used for Western blot analysis with anti-Par-4 antiserum. To control for loading, the anti-Par-4 antibody was stripped, and the membrane was reblotted with anti-actin serum. C, par-4 promoter DNA is methylated in Rastransformed HEK cells. Primers specific for unmethylated (U) or methylated (M) DNA were used to amplify DNA from HEK cells expressing the empty vector or H-Ras(12V) (left panel) or from Ras-expressing HEK cells treated with vehicle alone (dimethyl sulfoxide (DMSO)) or U0126 (30 M) for 6 days (right panel). regulation in Ras-transformed epithelial cells. We found that the expression alone of constitutively active forms of three Ras effectors (Raf, Ral GEF, and PI3K) was not sufficient to mimic Ras-mediated down-regulation of Par-4 in epithelial cells. These data suggest that, unlike in Ras-transformed rodent fibroblasts, in which Par-4 is down-regulated via the Raf/ERK pathway, in epithelial cells, activation of this pathway is not sufficient to down-regulate Par-4. We recently found that two other genes deregulated in Ras-transformed RIE-1 cells are aberrantly expressed as a result of the combined concomitant effects of two pathways. Expression of the actin-binding protein tropomyosin is down-regulated in Ras-transformed RIE-1 cells through a mechanism requiring activation of the Raf/MAPK pathway along with inhibition of the p38 MAPK pathway (44). Similarly, we also showed that the cell cycle regulatory protein cyclin D 1 is up-regulated in Ras-transformed cells through a mechanism requiring activation of the Raf/ERK pathway together with inhibition of the p38 MAPK pathway (52). Thus, it appears that, in epithelial cells, it is not uncommon to find Ras utilizing a multiple-pathway approach to elicit a change in gene expression.
Interestingly, however, although activation of the Raf/MAPK pathway alone was not sufficient to down-regulate Par-4, inhibition of MEK with the pharmacological inhibitors restored Par-4 expression in Ras-transformed RIE-1 and ROSE cells. This suggests that the Raf/ERK pathway represents one branch in the pathway that Ras utilizes to down-regulate Par-4 and that inactivation of this branch alone is sufficient to disrupt Ras-mediated down-regulation of Par-4. Thus, the clinical utility of pharmacological drugs that target the Raf/MEK/ERK pathway may inhibit tumor growth in part by restoring Par-4 expression and inhibition of transformation.
We determined that ectopic restoration of Par-4 expression to levels found in untransformed cells caused a partial reversion of Ras-mediated focus formation and soft agar growth of RIE-1 cells. This observation is consistent with the previous observation that Par-4 down-regulation contributes to growth transformation of Ras-transformed NIH 3T3 rodent fibroblasts (6 -8). Similar to data observed in fibroblasts (6), we found that restoration of Par-4 expression did not reverse the morphological transformation of Ras-transformed RIE-1 cells. That Par-4 affects anchorage dependence and loss of density-dependent growth (but not morphological transformation) is not unexpected and is consistent with the role of Par-4 as a regulator of cell survival rather than a regulator of actin organization or cell adhesion.
Promoter methylation represents a common mechanism utilized by cancer cells to silence the expression of a variety of genes (47). We showed recently that Ras down-regulates tropomyosin through cytosine methylation of the promoter (44). In addition, other studies have shown that oncogenic Ras-mediated down-regulation of the genes encoding lysyl oxidase and the Fas ligand is mediated by DNA methylation (53,54). Similarly, we have shown here that treatment of Ras-transformed RIE-1, ROSE, and DLD-1 cells with azadeoxycytidine, a known inhibitor of CpG methylation (37), restored Par-4 expression. Interestingly, Barradas et al. (6) showed that, in NIH 3T3 cells, Par-4 protein expression is not restored after treatment with azadeoxycytidine. This discrepancy may be explained in part by cell type differences. Barradas et al. used NIH 3T3 fibroblasts, whereas in this study, we utilized epithelial cells. In addition, Barradas et al. treated the cells for only 2 days, whereas we used a more extensive treatment time of 6 days. In combination with our direct demonstration of promoter methylation using methylation-specific PCR, our data show that the long-term mechanism for Ras-mediated loss of Par-4 expression involves CpG DNA hypermethylation, resulting in gene silencing through aberrant cytosine methylation in epithelial cells. In addition, the loss of methylation upon U0126 treatment suggests that Ras utilizes a MEK-dependent pathway leading to par-4 methylation. Finally, we observed that DNA demethylation also caused a partial reversion of the morphological transformation in both Ras-expressing RIE-1 and ROSE cells. Thus, promoter methylation and gene silencing of a variety of tumor cell growth inhibitory genes, including par-4, may be an important mechanism of Ras-mediated transformation for RIE-1, ROSE, and other epithelial cells.
The association of Ras and DNA methylation is highly intertwined. The DNA repair enzyme 6-O-methylguanine-DNA methyltransferase functions to remove mutagenic and cytotoxic adducts from the position of guanine. If adducts such as 6-O-methylguanine are not removed, mispairing between the 6-O-methylguanine and thymine can occur during replication, resulting in G-to-A mutations (55). In both colorectal and pancreatic carcinomas, the most prevalent ras mutations involve FIG. 6. Dnmt1 and Dnmt3a DNA methyltransferase mRNA expression is up-regulated in Ras-transformed RIE-1 cells. Total RNA from RIE-1 cells stably expressing K-Ras(12V) or the vector only (A) along with cells expressing Raf-22W (B) was used for reverse transcription-PCR analyses of Dnmt1, Dnmt3a, Dnmt3b, and ␤-actin mRNA levels (A) or for Northern blot analysis with 32 P-labeled DNA probes for Dnmt1 and the 18 S ribosomal RNA as a loading control (B). In A, control PCRs (Mock) were performed without template cDNA to show specificity of PCR products. G-to-A mutations at codons 12 and 13 (56). Esteller et al. (55) showed that 71% of colorectal tumors exhibiting G-to-A mutations in K-ras show loss of 6-O-methylguanine-DNA methyltransferase expression. In addition, their study found that loss of 6-O-methylguanine-DNA methyltransferase occurs in small adenomas that do not yet harbor K-ras mutations, further supporting the idea that loss of the DNA repair function of 6-O-methylguanine-DNA methyltransferase results in the accumulation of K-ras mutations. In turn, Ras activation has been shown in several studies to cause up-regulation of the DNA methyltransferase Dnmt1. In one study, Ras-transformed IEC-18 rat intestinal epithelial cells showed increased DNA methyltransferase activity and methylation of the p16 tumor suppressor gene, and treatment with azadeoxycytidine resulted in the restoration of p16 expression (57). In another study, inhibition of Ras activity in the Y1 mouse adrenocortical tumor cell line, which overexpresses wild-type K-Ras by 30fold, resulted in a 50-fold decrease in DNA methyltransferase enzyme activity and mRNA expression (58). In addition, injection of dnmt1 antisense oligodeoxynucleotides into Y1 mouse tumors inhibits tumor growth (49). Our studies now show that, in addition to up-regulating the Dnmt1 DNA methylation maintenance enzyme, Ras also up-regulates the Dnmt3a de novo DNA methylation enzyme. Taken together, these studies suggest a dependence of Ras-mediated tumor formation on gene silencing through DNA methylation by a mechanism that involves up-regulation of Dnmt1 and Dnmt3a. Additional studies to determine the mechanism by which Ras initiates upregulation of Dnmt1 and Dnmt3a will further our understanding of how Ras causes down-regulation of certain genes (such as par-4) that function to inhibit tumor growth.
In summary, our analyses of Ras-mediated down-regulation of Par-4 support a model in which, in epithelial cells, oncogenic Ras activation of ERK along with a yet unidentified pathway(s) results in down-regulation of Par-4 by causing an increase in the expression of CpG methyltransferases. The methylation of promoters of key genes such as par-4 may then result in downregulation of their expression and thus contribute to the process of transformation. The altered expression of many genes is observed in Ras-transformed cells, and an important goal of future studies will be to identify those that facilitate, rather than simply correlate with, Ras transformation and those that are mediated by changes in DNA methylation.