Loss of Transgelin in Breast and Colon Tumors and in RIE-1 Cells by Ras Deregulation of Gene Expression through Raf-independent Pathways*

Activated Ras but not Raf can transform RIE-1 and other epithelial cells, indicating the critical importance of Raf-independent effector function in Ras transformation of epithelial cells. To elucidate the nature of these Raf-independent activities, we utilized representational difference analysis to identify genes aberrantly expressed by Ras through Raf-independent mechanisms in RIE-1 cells. We identified a total of 22 genes, both known and novel, whose expression was either activated (10) or abolished (12) by Ras but not Raf. The genes up-regulated encode proteins involved in protein or DNA synthesis, regulation of protease activity, or ligand binding, whereas those genes down-regulated encode actin cytoskeletal-, extracellular matrix-, and gap junction-associated proteins, and transmembrane receptor- or cytokine-like proteins. These results suggest that a key function of Raf-independent signaling involves deregulation of gene expression. We further characterized transgelin as a gene whose expression was abolished by Ras. Transgelin was identified previously as a protein whose expression was lost in virally transformed cell lines. We show that this loss is regulated at the level of gene expression and that both Raf-dependent and Raf-independent pathways are required to cause Ras down-regulation of transgelin in RIE-1 cells, whereas Raf alone is sufficient to cause its loss in NIH 3T3 fibroblasts. We also found that Ras-dependent and Ras-independent mechanisms can cause the down-regulation of transgelin in human breast and colon carcinoma cells lines and patient-derived tumor samples. We conclude that loss of transgelin gene expression may be an important early event in tumor progression and a diagnostic marker for breast and colon cancer development.

expression of dominant negative mutants of Ets-1, Ets-2 (18,19), c-Fos (20), or c-Jun (21) can block oncogenic Ras-mediated transformation of NIH 3T3 fibroblasts. c-jun null mouse embryo fibroblasts were found to be insensitive to Ras-mediated transformation (22). Thus, the deregulated function of these transcription factors is clearly important for Ras transformation. However, what remains poorly understood is what genes are the key targets of Ras important for transformation.
The majority of human cancers in which ras mutations are found are of epithelial or hematopoietic cell origin (1,2). Despite this, much of our understanding of Ras function is based on studies of Ras transformation of rodent fibroblast cell lines (23). Observations that Ras function exhibits significant cell type differences questions whether the delineation of the Ras signaling events important for transformation in rodent fibroblasts accurately reflect how oncogenic Ras promotes human oncogenesis. For example, we and others (24,25) 2 showed previously that although activated Ras and Raf can fully transform NIH 3T3 fibroblast cells, only activated Ras caused transformation of RIE-1, IEC-6, or MCF-10A and HEK human epithelial cells. Although activation of the Raf/MEK/ERK pathway is necessary for Ras transformation of RIE-1 cells, by itself, Raf is incapable of causing transformation of these rodent intestinal epithelial cells. Stanbridge and colleagues (26) found that activation of the Raf/MEK/ERK pathway was necessary for Ras-induced transformation of a human fibrosarcoma but not a human colon adenocarcinoma cell line. Gire et al. (27) determined that although activated Ras was capable of stimulating a proliferative response in human primary thyroid epithelial cells, activation of the Raf/MEK/ERK pathway alone was not able to mimic this proliferative response. Finally, Marshall and colleagues (28) found that up-regulation of the ERK MAPK cascade did not correlate with the presence of mutated ras in pancreatic carcinoma cell lines. Taken together, these studies demonstrate the importance of Raf-independent pathways in the transformation of epithelial cells.
To delineate the Raf-independent mechanisms important for Ras transformation of epithelial cells, we utilized the PCRbased cDNA subtraction library screening method of representational difference analysis (RDA) (29,30), and we identified genes whose expression was deregulated by activated Ras, but not Raf, in RIE-1 cells. Our analyses identified both known and novel genes that were either up-or down-regulated by Ras, but not Raf, in an essentially all-or-none fashion. These observations support the important role of Raf-independent effectors in causing changes in gene expression and the importance of gain as well as loss of gene expression in Ras-mediated transformation. Finally, we further characterized the regulation of transgelin, which was identified previously as a protein whose expression was lost in suspended or virally transformed rodent cell lines (31). We determined that both Raf-dependent and Raf-independent signalings are required for Ras to cause downregulation of transgelin gene expression in RIE-1 cells, whereas Raf activation alone was sufficient to cause downregulation in NIH 3T3 fibroblasts. Finally, we determined that Ras-dependent and Ras-independent mechanism can downregulate transgelin in breast and colon carcinoma cell lines and patient-derived tumors, supporting an important contribution of transgelin loss of function in human oncogenesis.

EXPERIMENTAL PROCEDURES
Cell Lines-RIE-1 rat intestinal epithelial cells are a spontaneously immortalized, nontransformed, diploid, epidermal growth factor-responsive cell line (obtained from R. J. Coffey, Jr., Vanderbilt University) and maintained in Dulbecco's modified Eagle's medium (DMEM) supplemented with 5% fetal calf serum (FCS). Mass populations of RIE-1 cell lines stably expressing activated forms of K-Ras4B (K-Ras(12V)) and Raf-1 (Raf-22W) were established by transfection with pZIP-NeoSV(x)1 retrovirus vectors, where expression of the inserted gene is regulated by a Moloney long terminal repeat promoter, have been characterized previously, and are designated RIE(Ras) and RIE(Raf) cells, respectively (24). Mass populations of RIE-1 cells stably expressing constitutively activated forms of RhoA(63L) and Rac1(61L) (32) were generated in a similar fashion previously. 3 The retroviral vector pCTV3, where expression is controlled by the Moloney long terminal repeat promoter, was used to generate mass populations of RIE-1 cell lines stably expressing NH 2 -terminally deleted, constitutively activated forms of mouse Vav (pCTV3H-⌬N186-vav) and Dbl (pCTV3H-dblHA1) (33,34) similarly to those generated by retroviral infection with pZIP-NeoSV(x)1 previously. 3 Rat transgelin cDNA was PCR-amplified from a RIE-1 cDNA library, subcloned into a hemagglutinin (HA) epitopetagged pBabe-puro retroviral expression vector (designated pBabe-HAtransgelin), sequenced, and analyzed by comparison to rat transgelin sequence in GenBank TM (accession number X64422). pBabe-HA-transgelin plasmid DNA was then used to generate virus to infect RIE-1 cells to establish mass populations of RIE-1 cells stably expressing HAtransgelin, and protein expression was verified by Western blot analysis with protein lysates resolved over 15% SDS-PAGE gels, transferred onto Immobilon-P membranes (Millipore), and blotted with anti-HA serum (Babco).
The MCF-10A human breast epithelial cells, a spontaneously immortalized cell line that retains growth properties of normal breast epithelial cells, were obtained from M. Kinch (Purdue University) and maintained in DMEM/F-12 supplemented with 5% horse serum, 20 ng/ml epidermal growth factor, 0.5 g/ml hydrocortisone, and 10 g/ml insulin (35). The MCF-7 human breast ductal carcinoma cell line, provided by E. Stanbridge (University of California, Irvine), was maintained in DMEM supplemented with 10% FCS, 1 mM sodium pyruvate, and 8 g/ml insulin. The T47D human breast ductal carcinomas cell line, provided by P. Keely (University of Wisconsin), was maintained in RPMI supplemented with 10% FCS, 100 mM sodium pyruvate, and 4 mg/ml insulin. The DLD-1 human colon adenocarcinoma cell line contains one endogenous K-ras(13D) allele, and the derivative cell line of DLD-1 lacking the mutant K-ras allele (DKO-3) (36), provided by R. J. Coffey, Jr., was maintained in RPMI supplemented with 10% FCS. The reduced ability of the DKO-3 variant to form colonies in soft agar and tumors in athymic nude mice has been described previously (36,37). The HT1080 human fibrosarcoma cell line containing one endogenous mutant N-ras(61K) allele and the derivative cell line of HT1080 lacking the mutant ras allele (MCH 603c8) were provided by E. Stanbridge and maintained in DMEM supplemented with 10% FCS, 25 mM HEPES, and 1 mM sodium pyruvate, and additionally for the 603c8 cells, 1ϫ hypoxanthine/aminopterin/thymidine medium (Sigma) was also added. The human colon tumor cell lines, provided by L. G. Tillotson (University of North Carolina, Chapel Hill), were as follows: Colo320 HSR and Colo205 cells were maintained in RPMI 1640 supplemented with 10% FCS; HT29, HCT116, and CaCo2 cells were maintained in DMEM supplemented with 10% FCS; and SW480 cells were maintained in Leibovitz's medium supplemented with 10% FCS.
Representational Difference Analysis (RDA)-RDA was carried out essentially as described by Hubank and Schatz (29) and as we have described previously (30). Total RNA was isolated from RIE(Ras) and RIE(Raf) cell lines (24) that were cultured and harvested at the same time to minimize gene expression changes because of cell culture conditions. mRNA Maxi Kit from Qiagen was used to purify mRNA from total RNA and cDNA prepared with the Superscript Choice Kit (Invitrogen) as described by the manufacturers. Tester and driver amplicons were hybridized to each other at molar ratios of 1:100 (DP1), 1:400 (DP2), 1:80,000 (DP3), and 1:800,000 (DP4) for each successive round of RDA. After four rounds of hybridization and PCR, the final PCR products were size-fractionated over a 1.4% agarose gel, purified, cloned into the BamHI site of pBluescript (Stratagene), sequenced, and compared with the GenBank TM and EST data bases using BLAST algorithms.
Human Tumor Samples-Matched pairs of five colon adenocarcinomas and four breast invasive ductal carcinomas tumor samples along with surrounding marginal nontumor tissue samples were obtained from the Lineberger Comprehensive Cancer Center's Tissue Procurement Center at the University of North Carolina, Chapel Hill, and total RNA was extracted from the tissues as described below.
RNA Isolation and Northern Blot Analyses-Total RNA from cultured cells and tumors was isolated by the guanidine thiocyanate, acid-phenol method (38). For Northern blot analyses, 25 g of total RNA was size-fractionated over 1.4% formaldehyde gels, transferred to Hybond-N nylon membrane (Amersham Biosciences), and hybridized to 32 P-labeled DNA probes. Control for equivalent RNA loading was determined by hybridization with a GAPDH cDNA probe. Hybridizations were with 2 ϫ 10 6 cpm/ml in a solution containing 5ϫ SSC, 50% formamide, 5ϫ Denhardt's, 0.5% SDS, 0.1% Nonidet P-40, 50 mM sodium phosphate, pH 7.0, and 0.05% sodium pyrophosphate at 42°C for 24 -48 h and washed in 1ϫ SSC with 0.1% SDS at 50°C followed by 0.2ϫ SSC with 0.1% SDS at 55°C.

K-Ras (12V) Transformation of Epithelial Cells Causes
Aberrant Gene Expression-Whereas both activated Ras and Raf can readily transform NIH 3T3 fibroblasts, only Ras can transform RIE-1 and other epithelial cell lines suggesting that Rafindependent pathways play a crucial role in transformation of epithelial cells (24). To assess further the role of Raf-independent effector function in promoting Ras transformation, we identified genes aberrantly expressed by oncogenic Ras through Raf-independent mechanisms in RIE-1 cells using the PCRbased cDNA subtraction cloning method of RDA (29,30). Because our goal was to identify gene expression changes that are important for the maintenance of oncogenic Ras transformation, we utilized RIE-1 cells stably expressing constitutively activated K-Ras4B(12V) or Raf-1 (Raf-22W) (designated RI-E(Ras) and RIE(Raf), respectively) (24). Thus, the gene expression changes identified may be caused directly by the activation of oncogenic Ras function or indirectly as consequence of Rasmediated transformation.
To isolate genes up-regulated by Ras, but not Raf, we subtracted cDNA from RIE(Ras) cells with cDNA from RIE(Raf) cells. By using this approach, genes up-regulated by the Raf/ MEK/ERK pathway were subtracted out leaving only those genes up-regulated by Ras that required activation of Rafindependent effector signaling pathways. To isolate genes down-regulated by Ras, but not Raf, a second subtraction was performed in the reverse direction. Here we subtracted cDNA from RIE(Raf) cells with cDNA from RIE(Ras) cells. In this direction, the absence of cDNA from those genes down-regulated by Ras through Raf-independent pathways allowed for the cloning of the corresponding cDNAs from the RIE(Raf) cells. Fig. 1 shows the difference products (DP) we obtained after four successive rounds of subtraction and amplification used to isolate the genes either up-regulated (Ras cDNA subtracted with Raf cDNA (Ras-Raf), lanes 2-5) or down-regulated (Raf cDNA subtracted with Ras cDNA (Raf-Ras), lanes 6 -9) in RI-E(Ras) cells. Sequential subtractions resulted in the enrichment of specific bands which represented various genes as we increased the stringency of each subtraction (compare the starting cDNA material, lanes 1 (Ras cDNA representation) and 10 (Raf cDNA representation) to the sequentially subtracted and amplified difference products (DP1 to DP4 for each subtraction, lanes 2-9)). For our analyses, we concentrated on the isolation and analyses of sequences from DP4 to identify those genes whose expression shows a large gain or loss of expression by Ras, but not Raf, activation.
We first confirmed that the clones isolated from the subtractions represented truly differentially expressed genes and compared relative expression levels by Northern blot analysis using total RNA from control, empty vector-transfected RIE-1 cells, RIE(Raf), and RIE(Ras) cells hybridized to 32 P-labeled probes of each clone (Fig. 2). We found that all the isolated clones did correspond to differentially expressed transcripts and most were highly differentially expressed in an essentially all-or-none fashion. Collectively, we identified 10 genes whose expression was up-regulated and 12 whose expression was down-regulated, by Ras but not by Raf or vector only expressing RIE-1 cells.
After verification of differential expression, the sequences of the isolated genes were used to search for related sequences in GenBank TM . As summarized in Tables I and II, our analyses identified both known and novel genes. Several genes encoding proteins involved with protein or DNA synthesis were upregulated by Ras, but not Raf, including DNA-directed RNA polymerase II, IMP dehydrogenase type II, and the L27a ribosomal protein subunit. In addition to four novel genes found to be up-regulated by Ras but not Raf (RDA clones 12, 28, 31, and 87), those encoding a putative ligand-binding protein (RY 2G5), a cartilage matrix proteoglycan (aggrecan), and a serine protease, nexin I (glial-derived nexin I), were also cloned.
One of the genes we identified that was strongly downregulated by Ras, but not Raf, was transgelin (Fig. 2B). Transgelin (also called SM22 (39)) is an actin-binding protein of unknown function and has been shown to cross-link actin filaments in vitro (40). Actin stress fibers are commonly reduced in transformed cells and in cells grown in suspension, and transgelin was shown to bind actin stress fibers in fibroblasts and to have reduced expression in SV40-and RSV-transformed mesenchymal cells and in untransformed 3T3 fibroblasts grown in suspension (31). Because current understanding of the mechanism by which transgelin protein expression is down-regulated and the role of this loss in transformation were limited, we centered our analyses on evaluating these issues.

Transgelin Down-regulation Represents an Early Event in
Ras-mediated Transformation of RIE-1 Cells-We first wanted to establish if the down-regulation of transgelin expression is a direct consequence of Ras activation or, alternatively, is a consequence of Ras-mediated transformation. To determine this, we utilized a RIE-1 cell line that harbors an isopropylthiogalactosidase (IPTG)-inducible expression vector encoding Ha-Ras(12V) (41). When cultured in the absence of IPTG, no significant expression of Ha-Ras(12V) protein was seen, and the cells exhibited the nonrefractile, cobblestone-like cell morphology and anchorage-dependent growth properties characteristic of untransformed parental RIE-1 cells (data not shown). 3 After exposure to IPTG, expression of Ha-Ras(12V) protein was detected within 6 h (Fig. 3). However, morphologic transformation was not observed until after 48 h of IPTG treatment, and complete morphologic transformation was not achieved until after 72 h (data not shown), 3 Thus, this cell line allowed us to assess whether the loss of expression of transgelin is associated with Ha-Ras(12V) protein expression or Ha-Ras(12V)-induced transformation.
We performed Northern blot (transgelin) analyses on cell lysates from the Ras-inducible RIE-1 cells treated with IPTG for 0 -96 h (Fig. 3). Transgelin mRNA showed a progressive time-dependent decrease beginning at 6 h and was completely lost after 24 h. These results suggest that down-regulation of transgelin expression is caused directly by Ras activation and is not a consequence of the transformed state induced by Ras. Thus, the loss of function of transgelin may be an important event in facilitating Ras transformation.
Ras Down-regulation of Transgelin Shows Cell Type Differences in the Requirement for Raf-independent Effector Activation-Because Raf activation alone can cause transformation of NIH 3T3 mouse fibroblasts, we also determined if Raf-independent pathways were also required for Ras-mediated downregulation of transgelin in NIH 3T3 cells. Northern blot analyses (Fig. 4A) showed that transgelin was expressed in control and empty vector-transfected NIH 3T3 cells but lost in NIH 3T3 cells stably expressing either activated Ras or Raf. Thus, Ras-mediated down-regulation of transgelin does not require Raf-independent signaling in NIH 3T3 cells. To determine whether the Raf/ERK effector pathway is involved in Rasmediated down-regulation of transgelin in RIE-1 cells, we assessed whether treatment of RIE(Ras) cells with the U0126 MEK inhibitor to block ERK activation would cause any restoration of transgelin expression. U0126-treated RIE(Ras) cells showed the reappearance of transgelin expression (Fig. 4B). Thus, although activation of the Raf/ERK pathway alone is not sufficient to down-regulate transgelin expression in RIE-1 cells, it is necessary for full down-regulation by Ras.
Transgelin Is Deregulated in Human Tumor Cell Lines by Ras-dependent and Ras-independent Mechanisms-Our analyses showed that Ras transformation of rodent fibroblasts or epithelial cells is associated with the loss of transgelin expression. We next determined whether transgelin expression is also down-regulated by oncogenic Ras in human tumor cells. For these analyses, we utilized two human tumor cell lines where it has been shown that oncogenic Ras expression is critical for promoting their transformed and tumorigenic growth properties (36,37). Additionally, genetic variants of each have been isolated that have lost the mutated ras allele. The DLD-1 human colon adenocarcinoma cell line contains a K-ras(13D) mutant allele, and DKO3 is a variant cell line of DLD-1 where the mutant ras allele was disrupted by homologous recombination. The HT1080 human fibrosarcoma cell line harbors a mutant N-ras(61K) allele, and the MCH 603c8 cell line is a variant of HT1080 that has lost the mutated N-ras allele. For both cell lines, loss of mutant Ras function corresponded to impaired ability for anchorage-independent growth in soft agar and tumor formation in nude mice.
Northern blot analyses (Fig. 5) showed that transgelin expression was down-regulated in the ras mutation-positive parental HT1080 and DLD-1 cell lines and was elevated in the MCH 603c8, but not DKO-3, ras mutation-deficient cell line. These results show that loss of transgelin expression in the HT1080 fibrosarcoma cell line, like that observed in the RIE-1 cell line, is Ras-dependent and in the DLD-1 colon adenocarcinoma cell line is Ras-independent.
We further evaluated the role of Ras activation in downregulation of transgelin expression in additional human tumor cell lines. First, we evaluated six colon carcinoma cell lines that either harbor activating K-ras mutations (HCT116, SW480, and HT29) or are wild type for ras (ColoHSR, Colo205, and CaCo2). All of the human colon carcinoma cell lines also showed a loss of expression of transgelin that was independent of ras mutation status (Fig. 6). Similarly, whereas transgelin expression was detected in the normal MCF-10A human breast epithelial cell line, expression was lost in two human breast carcinoma cell lines (MCF-7 and T47D) that also do not possess mutated ras alleles. Thus, although aberrant activation of Ras can cause down-regulating transgelin expression, Ras-independent mechanisms must also exist in some tumor cells that down-regulate its expression.
Transgelin Is Down-regulated in Human Breast and Colon Tumors-Whereas our observations above and those of others (31) have shown that transgelin expression is lost in transformed and tumor cell lines, to date there have been no reports showing aberrant expression of transgelin in patient-derived human tumor tissue. Thus, to address this possibility, we performed Northern blot analyses and compared the expression of transgelin in patient-derived colon and breast tumor tissue with matched adjacent nontumor histologically normal tissue from the same patient. We found that transgelin expression was greatly reduced or lost in all five colon and all four breast tumor samples (Fig. 7). Furthermore, reduced transgelin expression was also seen in some nontumor colon (samples 4 and 5) and breast (samples 2 and 4). Because these samples may represent benign tumor rather than normal tissue, it suggests that loss of transgelin expression may be an early event in the development of these cancers. These results demonstrate that loss of transgelin expression is not restricted to cells in culture, that transgelin expression is deregulated frequently in human colon and breast tumors, and may represent an early event in carcinogenesis.

Transgelin Is Down-regulated by Other Ras Family Small
GTPases-Transgelin is an actin-binding protein whose expression is decreased in association with the loss of stress fibers (31). Rho family small GTPases, such as Rac1, RhoA, and Cdc42, are regulators of actin cytoskeletal organization (42,43). Furthermore, Rho family GTPases can be activated by oncogenic Ras, and their functions are required by oncogenic Ras to transform rodent fibroblasts (5). Therefore, we determined if activated RhoA, Rac1, or Cdc42 alone could also cause down-regulation of transgelin gene expression. In addition, we also determined if the down-regulation of transgelin by Ras was Ras isoform-specific. For these analyses, we utilized RIE-1 cells stably transfected with expression vectors encoding constitutively activated K-Ras(12V), Ha-Ras(61L), N-Ras(12D), Rho(63L), and Rac1(61L) mutant proteins. In addition, cells expressing constitutively activated Vav (an exchange factor and activator of RhoA, Rac1, and Cdc42) and Dbl (an exchange factor and activator of RhoA and Cdc42) were also utilized. 3 As shown in Fig. 8, relative to vector only transfected RIE-1 cells, transgelin expression was down-regulated by K-, Ha-, and N-Ras, partially down-regulated by Rac1 and Vav, but not by Raf, RhoA, or Dbl. These results show that in addition to K-, Ha-, and N-Ras, Rac and Vav are also capable of down-regulating the expression of transgelin in RIE-1 cells. From these data, we conclude that Vav is probably down-regulating transgelin through Rac1 because Vav activates Rac, Rho, and Cdc42, whereas Dbl, which activates Rho and Cdc42 (44), did not down-regulate transgelin. Whereas these data might suggest that Ras may mediate down-regulation of transgelin, in part through activation of Rac1, we recently found that Rac-GTP  (31,45). Our results above showed that transgelin is downregulated at a high frequency in a variety of human tumor cell lines and patient-derived tumor tissues suggesting that the mechanism by which transgelin is down-regulated is one that occurs frequently in carcinogenesis. One common mechanism for suppression of gene expression in human carcinomas involves DNA methylation of promoters of genes to block transcription (46). For example, the expression of various tumor suppressor genes, including BRCA1, E-cad, hMLH1, p16, VHL, and Rb, are frequently inhibited due to hypermethylation in human cancers. Additionally, promoter methylation has been described as a mechanism to repress the transcriptional activity of transgelin in smooth muscle cells (47). Thus, we evaluated the possibility that the loss of transgelin expression in Ras-mediated transformation was due to DNA hypermethylation.
To determine whether transgelin expression was lost due to DNA methylation, we treated RIE(Ras) cells with the demethylating agent azadeoxyactidine and then examined, by Northern blot analyses, the level of mRNA in treated cells compared with vehicle only (Me 2 SO) treated cells. As shown in Fig. 9, transgelin expression was not restored after 6 days of treatment with azadeoxycytidine when compared with vehicle only treated RIE(Ras) cells. In contrast, the expression of tropomyosin, a gene we showed previously to be down-

FIG. 3. Down-regulation of transgelin correlates directly with the temporal induction of Ha-Ras(61L) protein expression in RIE-1 cells and not morphological transformation.
Ha-Ras(61L)inducible RIE-1 cells were incubated in growth medium supplemented with 1 mM IPTG for the times indicated and refed with growth medium supplemented with fresh IPTG at 48 h, and duplicate cultures at each time point were used for protein lysates or for isolation of total RNA. Cells cultured without IPTG exhibited a flat, cobblestone appearance typical of nontransformed RIE-1 cells and became highly morphologically transformed by the appearance of highly refractile, non-clustering, spindly shaped cells after 96 h of IPTG treatment (A). Protein lysates (30 g) were resolved by separation on 15% SDS-PAGE, transferred to Immobilon-P, blotted with anti-Ha-Ras serum (146, Quality Biotechnologies) followed by horseradish peroxidase-conjugated secondary antibody, and visualized by ECL. Ha-Ras expression was detected by 6 h after the addition of IPTG. Note that the Ha-Ras protein present at time 0 represents expression of endogenous Ras, and the reappearance of unprocessed Ha-Ras at 72 h is due to the re-addition of IPTG (B). Total RNA (25 g) was size-fractionated over formaldehyde-agarose gels, transferred to Hybond-N (Amersham Biosciences), and hybridized to 32 P-labeled cDNA probes of transgelin (C) or GAPDH (D), with the latter used as a control for loading.

FIG. 4. Role of Raf in down-regulation of transgelin in RIE-1 and NIH 3T3 cells. A, activated Ras or
Raf is sufficient to cause down-regulation of transgelin in NIH 3T3 mouse fibroblasts. Total RNA (25 g) was isolated from NIH 3T3 cells stably transfected with the empty pZIP-NeoSV(x)1 vector or encoding Raf-22W or Ha-Ras(12V), size-fractionated by electrophoresis over formaldehyde-agarose gels, and analyzed by Northern blot with 32 P-labeled transgelin cDNA probe. B, inhibition of ERK activation partially restores transgelin expression in RIE(Ras) cells. RIE(Ras) cells were treated for 24 or 48 h with Me 2 SO (DMSO) vehicle control, with 10 M U0126, SB203580, or LY294002, or with 50 M PD153035, and total RNA (25 g) was isolated, size-fractionated by electrophoresis over formaldehyde-agarose gels, and analyzed by Northern blot with 32 P-labeled DNA probes of transgelin or GAPDH.

FIG. 5. Transgelin down-regulation in human epithelial cells is mediated by Ras-dependent and Ras-independent mechanisms.
Total RNA (25 g) from DLD-1, DKS-3 (derivative of DLD-1 with mutant K-ras(13D) allele deleted), HT1080 and MCH 603c8 (derivative of HT1080 with mutant N-ras(61K) allele deleted) was size-fractionated over formaldehyde-agarose gels and analyzed by Northern blot with 32 P-labeled DNA probes for transgelin or GAPDH. regulated by Ras by DNA methylation, was restored after 6 days of treatment with azadeoxycytidine. 3 These results suggest that the mechanism by which transgelin expression is down-regulated in Ras-transformed RIE-1 cells is not by DNA methylation.

Forced Re-expression of Transgelin Alone Is Not Capable of Reverting Ras-mediated Transformation of RIE-1 Cells-Our
determination that loss of transgelin expression is associated with oncogenesis in a wide variety of cell types suggests that transgelin may function to antagonize growth transformation. Thus, we sought to determine whether forced re-expression of transgelin could revert the transformed phenotype of RIE(Ras) cells. We established mass populations of RIE(Ras) cells stably transfected with an expression vector encoding HA epitopetagged transgelin, and we verified expression of HA-transgelin by Western blot analyses using anti-HA serum (data not shown). When compared with RIE(Ras) cells transfected with the empty vector, those expressing HA-transgelin showed no detectable change in either cell morphology or ability to grow in soft agar (data not shown).
Because transgelin is an actin-binding protein, we also examined the possible role for the loss of transgelin in promoting tumor cell invasion and metastasis. To address this question, we utilized a human breast tumor cell line, MDA-MB-231, that we showed does not express transgelin and that exhibits an invasive phenotype in vitro (data not shown). Mass populations of MDA-MB-231 cells stably overexpressing HA-tagged transgelin were established similarly to those generated in the RIE-1 cells, and expression of HA-transgelin protein was confirmed by Western blot analysis with anti-HA serum (data not shown). By using two in vitro invasion assays, we next compared the invasive phenotype of the parental MDA-MB-231 cells transfected with vector only (pBabe-HA) with that of the cells expressing HA-transgelin using the Matrigel (48) and G8 myoblast (49,50) invasion assays, and we found no difference between the cells that did not and those that did express transgelin (data not shown). Thus, these data show that forced re-expression of transgelin alone is not sufficient to revert the Ras-induced morphologic and growth transformation of RIE-1 cells or the invasive properties of MDA-MB-231 cells.  9. The mechanism by which transgelin is down-regulated by Ras is not mediated by DNA methylation. RIE-1 cells stably expressing K-Ras(12V) were treated with the demethylation agent 5-aza-2Јdeoxycytidine (Sigma) at 1 M for 3 or 6 days or with vehicle (Me 2 SO) alone for 6 days, and total RNA was isolated, size-fractionated by electrophoresis over formaldehyde-agarose gels, and analyzed by Northern blot with 32 P-labeled DNA probes of transgelin or GAPDH as a control for loading. Untreated RIE-1 cells expressing the empty vector (pZIP-NeoSV(x)1) were used as a positive control for detection of transgelin transcript.

DISCUSSION
We have shown that whereas activated Ras and Raf can fully transform NIH 3T3 fibroblast cells, only activated Ras can cause transformation of RIE-1 and other epithelial cells (24). Consequently, Ras activation of Raf-independent effector signaling pathways is critical for transformation of these and other epithelial cells. To elucidate further the nature of Rafindependent signaling important for Ras transforming activity, we used RDA and identified those genes whose expression was deregulated by Ras transformation through Raf-independent pathways in RIE-1 intestinal epithelial cells. We identified 10 genes whose expression was up-regulated and 12 genes that were down-regulated by Ras-, but not Raf-, expressing RIE-1 cells. Both known (15) and novel (6) genes, as well as one EST sequence, were identified. These observations support the important role of Raf-independent effectors in causing changes in gene expression and the importance of gain as well as loss of gene expression in Ras-mediated transformation. Finally, we identified transgelin as a gene down-regulated by Ras in RIE-1 cells, and we determined that a reduction in transgelin gene expression is a common event associated with human breast and colon carcinogenesis. However, forced re-expression of transgelin did not reverse Ras transformation of RIE-1 or NIH 3T3 (data not shown) cells or impair the invasive properties of breast carcinoma cell lines in vitro. In light of the multitude of gene expression changes seen in Ras-transformed cells, it is not surprising that the correction of one gene expression defect alone will not be sufficient to reverse Ras transformation.
In an analysis of the genes we isolated by RDA, we found that all genes identified were up-or down-regulated by Ras, but not Raf, in RIE-1 cells. The only exception was connexin-26, which was down-regulated by Ras but up-regulated by Raf. This suggests that perhaps Raf-dependent signaling positively regulates connexin-26 expression, whereas Raf-independent pathways negatively regulate connexin-26 expression. Thus, the Raf-independent pathways activated by Ras that inhibit connexin-26 transcription may override a positive signal transduced from the Raf/MEK/ERK pathway and result in transcriptional repression of connexin-26.
Our observation of significant changes in gene expression in RIE(Ras) versus RIE(Raf) cells indicates that an important consequence of Raf-independent signaling is the changes in gene expression. The observations that we obtained using RDA are similar to those made by Schä fer and colleagues (51), where they utilized subtractive suppression hybridization, a similar PCR-based cDNA subtraction technique, and identified 393 known or novel genes whose expression was up-regulated or down-regulated by Ras transformation of 208F rat fibroblasts. Of these genes, the expression of only 61 genes was reversed by treatment of Ras-transformed cells with pharmacological inhibition of ERK activation, indicating that the majority of gene expression changes caused by Ras are mediated by activation of Raf-independent signaling pathways.
In another study using microarray analyses of over 6,000 human genes, Downward and colleagues (25) identified 124 genes that were differentially expressed upon Raf activation in MCF-10A human breast epithelial cells. As with RIE-1 cells, Raf activation alone is not sufficient to cause transformation of MCF-10A cells (24,25). Interestingly, the vast majority (85%) of these genes were up-regulated rather than down-regulated in expression. This contrasts with our observation where we found an equivalent number of genes up-regulated and downregulated by oncogenic Ras. Thus, it appears that Raf activation involves primarily activation of gene expression, whereas a significant consequence of Raf-independent effector activation involves down-regulation of gene expression.
Interestingly, a number of the genes up-regulated by Ras in RIE-1 cells have been shown previously to be associated with oncogenesis. For example, we identified IMP dehydrogenase II (IMPDH II) as a gene up-regulated by Ras. IMPDH II is the rate-limiting enzyme in GTP biosynthesis, and its activity has been found to be up-regulated in many cancer cells, although no association with Ras activation has been described previously (52). Inhibitors of IMPDH II have been evaluated for use in cancer chemotherapy, and we are currently assessing the consequence of IMPDH II inhibition on the transformed growth properties of RIE(Ras) cells. Similarly, we found that glialderived nexin I was up-regulated by Ras, and glial-derived nexin I activity was shown to be increased in 9L rat brain gliosarcoma cells (53).
We also found that genes down-regulated by Ras have been described previously as genes whose products may be antagonistic to epithelial cell growth. For example, we found that the gene expression of thymic lymphoma cell stimulating factor-␤ (TLSF-␤)/stromal cell-derived factor-␤ (SDF-1␤) was down-regulated by Ras. SDF-1 is a cytokine for the chemokine receptor CXCR-4 that is thought to be responsible for B-cell lymphopoiesis and bone marrow myelopoiesis (54). Although TLSF-␤/ SDF-1␤ stimulates the growth of lymphocytes, it may inhibit the growth of epithelial cells and, consequently, function similarly to transforming growth factor-␤ in its opposing consequences on the growth of mesenchymal versus epithelial cells (55). In addition, expression of the closely related gene SDF-1␣ was greatly reduced in a majority of gastrointestinal tumors (56). When taken together, these observations suggest that the loss of expression of TLSF-␤/SDF-1␤ may contribute to Rasmediated transformation.
The protein products of several genes down-regulated by Ras have been implicated in the regulation of cell-cell and/or cellmatrix interactions. Nidogen/entactin is a component of basement membranes and has been shown to promote cell attachment (57) and differentiation (58) of mouse mammary tumor cells. Another down-regulated gene encodes connexin-26, a gap junctional intercellular communication protein. Decreased expression of connexins is associated with the down-regulated loss of gap junctional intercellular communication protein, which has been associated with uncontrolled cell growth and neoplasia (59) and is thought to be an early event in tumor progression (60). In addition, although normal mammary epithelial cells express predominantly connexin-26, little to none is detected in breast cancers (61). We have found that gap junctions are disrupted in RIE(Ras) cells, and this loss may represent an important event in Ras transformation. 4 Finally, our identification of tropomyosin as a gene down-regulated in RIE(Ras) cells is consistent with previous studies (62)(63)(64) where loss of its RNA and protein expression has been associated with Ras transformation of a variety of fibroblast and epithelial cells. Decreased expression of tropomyosin has also been observed in human prostate, breast, and ovarian tumors and in squamous cell carcinoma tissues (65)(66)(67). Tropomyosin is an actin-binding protein of unknown function, and its forced re-expression in Ras-transformed fibroblasts has been shown to reverse morphologic and growth transformation. However, our recent studies 3 determined that forced re-expression of tropomyosin in Ras-transformed RIE-1 cells did not cause any significant reversal of transformation.
Finally, we found that transcripts for bone morphogenetic protein (BMP)-4 were down-regulated by Ras in the RIE-1 cells. BMP-4 is a member of the transforming growth factor-␤ family (68), and BMP-4 has been shown to induce the cyclindependent kinase inhibitor p21 (69) and apoptosis (68,70,71). Thus, the loss of expression of BMP-4 may promote the growth and survival of Ras-transformed RIE-1 cells. This observation, when taken together with the fact that a number of the genes that we have identified have been linked previously to oncogenesis, supports the possibility that these gene expression changes may indeed be important in mediating Ras transformation via activation of Raf-independent signaling pathways. Thus, the long term goal of our studies will be to determine whether genes down-regulated by Ras but not Raf, when reexpressed in RIE(Ras) cells, cause a reversion of Ras transformation. Conversely, we will determine whether genes up-regulated by Ras but not Raf cause growth transformation when forced-overexpressed in RIE(Raf) cells.
In the present study, we extended our analyses to evaluate the regulation of transgelin expression by Ras and its role in Ras-mediated transformation. Transgelin was identified previously as a transformation and shape change-sensitive actingelling protein (40). However, only limited analyses have been done to evaluate the mechanism of down-regulation. Furthermore, no evaluation of transgelin expression in human cancers had been done. First, we showed that the expression of transgelin is abolished at the level of transcription by oncogenic Ras in NIH 3T3 mouse fibroblasts, RIE-1 rat intestinal epithelial cells, and HT1080 human fibrosarcoma cells. Second, we showed that transgelin expression was down-regulated in a variety of human breast, colon, and fibrosarcoma tumor cell lines. However, the absence of transgelin expression in breast and colon carcinoma cell lines that lack ras mutations indicates that Ras-independent mechanisms of down-regulation also exist. In addition, it remains possible that in those cell lines that harbor a mutant ras allele, the mechanism by which transgelin is down-regulated may also be independent of Ras activity. However, because our analyses of RIE-1 and HT1080 cells showed that transgelin down-regulation is dependent on Ras expression in these cell types, we would expect that downregulation of transgelin in some tumor cell lines can be through a Ras-dependent mechanism. Finally, we showed that transgelin expression was lost in a large number of human colon and breast tumor samples. Thus, this is the first report showing transgelin to be down-regulated in vivo.
We observed that down-regulation of transgelin was associated with the temporal expression of oncogenic Ras, rather than with the onset of the transformed phenotype. This supports a possible contribution of the loss of transgelin expression in mediating Ras transformation. However, when we forced re-expressed transgelin in Ras-transformed RIE-1 cells, we saw no change in morphology or ability to grow in soft agar. In addition, we also found that coexpression of transgelin was not capable of inhibiting Ras-mediated transformation of NIH 3T3 cells (data not shown). We also found that re-expression of transgelin did not reduce the invasive properties of the MDA-MD-231 breast tumor cells. These results do not necessarily mean that the loss of transgelin is not involved in Ras-mediated transformation because it is likely that the correction of one gene defect alone will not be sufficient to reverse Ras transformation. For example, we also found that the expression of another actin-binding protein, tropomyosin, is lost in Rastransformed cells. Perhaps the re-expression of transgelin, together with tropomyosin, will be required to reverse the morphologic transformation seen with Ras-transformed cells. Finally, it is also possible that the addition of an NH 2 -terminal HA epitope tag onto transgelin impaired its function or that ectopic overexpressed transgelin will not be incorporated prop-erly into actin. In the absence of clear functional assays for transgelin function, these possibilities cannot be excluded.
In summary, we have identified a collection of genes aberrantly expressed as a result of Ras transformation through Raf-independent pathways in RIE-1 colonic epithelial cells. Our observations, when considered together with those made by Schä fer and colleagues (51), indicate that changes in gene expression are clearly associated with Ras activation of Rafindependent effectors. With the increasing application of gene array expression analyses, the roster of Ras-deregulated genes is certain to expand significantly. Thus, the task at hand will be to identify those changes in gene expression that are important mediators of oncogenic Ras transformation. Such genes may identify novel targets for anti-Ras drug development as well as markers for the early detection of ras mutation-positive cancers.