|Home | About | Journals | Submit | Contact Us | Français|
Our recent study identified 2,261 dysregulated genes in the esophagi of rats that received a 1-week exposure to the carcinogen N-nitrosomethylbenzylamine (NMBA). We further reported that 1,323 of these genes were positively modulated to near-normal levels of expression in NMBA-treated animals that consumed dietary phenylethyl isothiocyanate (PEITC), a constituent of cruciferous vegetables. Herein, we report our results with companion animals that were fed a diet containing 5% freeze-dried black raspberries (BRB) instead of PEITC. We found that 462 of the 2,261 NMBA-dysregulated genes in rat esophagus were restored to near-normal levels of expression by BRB. Further, we have identified 53 NMBA-dysregulated genes that are positively modulated by both PEITC and BRB. These 53 common genes include genes involved in phase I and II metabolism, oxidative damage, and oncogenes and tumor suppressor genes that regulate apoptosis, cell cycling, and angiogenesis. Because both PEITC and BRB maintain near-normal levels of expression of these 53 genes, their dysregulation during the early phase of NMBA-induced esophageal cancer may be especially important in the genesis of the disease.
Worldwide, esophageal squamous cell carcinoma (ESCC) is the sixth most prevalent cancer in humans. Due to a lack of symptoms during the early stages of the disease, ESCC is rarely detected until it has metastasized (1). Tobacco products, alcohol, and inadequate diet are primary causes of ESCC; thus, in addition to promoting the lifestyle changes of smoking cessation and alcohol moderation, there is a need to identify foods and food constituents that inhibit or prolong the onset of clinical disease (2, 3). The F344 rat esophagus bioassay model (3), in which tumors can be induced by nitrosamine carcinogens such as N-nitrosomethylbenzylamine (NMBA), has long been used to identify foods and food constituents with anticarcinogenic activity. These studies have identified two dissimilar food groups, cruciferous vegetables and berries, to be potent inhibitors of rat esophageal tumors induced by NMBA (4, 5). The former contains numerous chemopreventive agents, including the isothiocyanates, and the latter contains high amounts of polyphenols and ellagitannins. The most extensively studied isothiocyanate is phenylethyl isothiocyanate (PEITC), which owes its prevention properties, at least in part, to inhibition of carcinogen-activating cytochrome P450 enzymes and induction of carcinogen-detoxifying phase II enzymes (6, 7). Regarding the polyphenols, we originally found that a diet supplemented with ellagic acid was preventative in the rat esophagus (8). Subsequently, we identified blackberries, raspberries, black raspberries (BRB), strawberries, and cranberries as being exceptionally rich in ellagic acid, with dry weight concentrations ranging from 630 to 1,500 µg/g (9). Further analyses showed that the ellagic acid was most abundant in the pulp and seeds of berries, with little in the juice. The pulp and seeds also contain many other known chemopreventive agents, including vitamins A, C, and E and folic acid, calcium and selenium, β-carotene, α-carotene, lutein, gallic acid, ferulic acid, p-coumaric acid, quercetin, several anthocyanins, β-sitosterol, stigmasterol, and kaempferol (4, 10, 11). Importantly, we also found that the active constituents can be concentrated 9- to 10-fold relative to the fresh berry by freeze drying (11). BRBs for the present study were obtained from a single Ohio farm, freeze dried, and ground into a powder as described (11). Typically, the berries are not grown organically; however, no pesticides or fungicides are applied to the berries within 10 days before harvest. This results in nondetectable levels of any pesticides or fungicides in the berry powder due to biodegradation of these compounds during the 10-day period before harvest.
The conversion of a normal cell into a tumorigenic cell is driven by numerous molecular aberrations that arise over time that endow the ultimate cell with resistance to apoptosis, insensitivity to growth-inhibitory signals, limitless replicative potential, sustained angiogenesis, and tissue invasion/metastasis capabilities (12). The time line of these events in the rat esophagus model covers several months. We recently began delineating the genes dysregulated early in this carcinogenesis process and have reported that treatment of rats with three s.c. injections (0.5 mg/kg body weight) of NMBA over 1 week led to the dysregulation of 2,261 genes (13). We also reported that the expression of 1,323 of these genes was positively modulated if the animals were fed PEITC 2 weeks before and during the week of NMBA treatment. Herein, we have evaluated the esophagi of companion animals in the same protocol whose diet was supplemented with 5% freeze-dried BRBs, instead of PEITC, for 2 weeks before and during the week the animals were treated with NMBA. We found that 462 of the NMBA-dysregulated genes were restored to near-normal levels of expression by BRB. Interestingly, we found that 53of the 2,261 genes dysregulated by NMBA were expressed at control levels of transcription if the animals received either PEITC or BRB. These common genes include those involved in phase I and II metabolism, oxidative damage, and oncogenes and tumor suppressor genes that regulate apoptosis, cell cycling, and angiogenesis. Overall, our study has identified a unique collection of 53 genes whose early dysregulation may be especially important in the carcinogenesis of rat ESCC.
The procedures for determining the effects of dietary PEITC and BRB on gene expression in NMBA-treated rat esophagus have been described in detail in our companion report (13); thus, they are discussed only briefly here. Four- to 5-wk-old male Fischer F344 rats were randomized into three experimental groups of 18 animals each. Rats in group 1 received control AIN-76A diet. Those in group 2 were given AIN-76A + PEITC (5 µmol) and those in group 3wer e given AIN-76A + 5% BRB. After 2 wk, one half of the animals in group 1 received three s.c. injections of 20% DMSO/water (vehicle control) spaced every other day, and the other half received three s.c. injections of NMBA (0.5 mg/kg body weight in 20% DMSO/water). Similarly, one half of the animals in groups 2 and 3wer e divided into PEITC control and PEITC+NMBA subgroups and BRB control and BRB+NMBA subgroups, respectively. PEITC, at 5 µmol in the diet, reduces NMBAinduced tumors in the rat esophagus by ~ 100% and BRB, at 5% of the diet, causes an approximate 50% reduction in esophageal tumors (5). All rats were sacrificed 24 h after the last NMBA injection. The esophagus from each animal was excised, opened longitudinally, and cut into two parts. One part was fixed in 10% buffered formalin for routine histopathology, and the other was frozen in liquid nitrogen and stored at −80° C for analysis of transcription changes using rat microarrays (41,000 transcripts). Triplicate microarrays were completed for each of the six groups of nine rats for a total of 18 microarrays.
Normalized data for each microarray were imported into Rosetta Resolver (14) for analysis (version 22.214.171.124.23; Rossetta Biosoftware). ANOVA was performed on log ratios using the Rosetta Resolver error model and weighting, as described by Weng and colleagues (15). One-way ANOVA was used to identify genes with a significant treatment effect. A P value cutoff of 0.001 was used to determine statistical significance in each treatment comparison. Statistically significant genes were further filtered to identify those with a minimum 1.5-fold change (16, 17). Gene ontology and pathway analysis of differentially expressed genes was done using the DAVID program.4 Microarray results were confirmed by comparison with mRNA levels obtained by quantitative reverse transcription-PCR (RT-PCR) using selected gene-specific primer pairs for a total of eight genes as described in our companion report (13). For each of the eight genes evaluated, the RT-PCR results confirmed gene expression changes that were observed by microarray.
We previously reported that the cellular morphology of esophagi from rats treated with 5 µmol dietary PEITC alone or PEITC+NMBA was not significantly different from that of animals on control diet. In contrast, the esophagi of rats treated with NMBA only exhibited significant hyperplasia and low-grade dysplasia with evident cytotoxicity and marked infiltration of inflammatory cells. Essentially identical results were observed in the esophagi of rats treated with BRB and BRB+NMBA [i.e., the esophagi from both the BRB alone and BRB+NMBA animals were indistinguishable (P < 0.05) from the controls; Fig. 1].
In our companion report (13), we noted that the expression of 2,261 genes was either up-regulated or down-regulated more than ≥1.5-fold (P ≤ 0.001) in the esophagi of rats treated for 1 week with NMBA, and PEITC modulated 1,323 of these genes to near-normal levels of expression. In contrast, only 462 of the NMBA-affected genes were modulated by BRB. Of these, 203exhibited partially restored expression levels and 259 were expressed at control levels (±33%; see Fig. 2 and Supplementary Table S1). Two hundred and fourteen of the 259 genes were up-regulated by NMBA and 45 were down-regulated. We hypothesize that an important mechanism of chemoprevention by BRB is the correction of NMBA-dysregulated genes to homeostatic levels of expression; thus, we have focused on understanding the possible roles of the 259 genes returned to near-normal levels of transcription by the BRB diet. Not all of these genes have known functions, as denoted by the DAVID program; however, known among them were those involved in signal transduction, cell proliferation, cell cycle progression, chromosome partitioning, inflammation, differentiation, cell junctions, cytoskeleton, apoptosis, and angiogenesis (Table 1). As in our companion report (13), we also observed effects of BRB on the expression of genes involved in carcinogen metabolism and DNA adduct formation. These and other effects of BRB on gene expression in NMBA-treated rat esophagus are described below, in brief.
Exposure of rats to NMBA causes an elevation in O6-methylguanine adducts in esophageal DNA (3, 11). Formation of O6-methylguanine adducts is considered to be an important early event in NMBA-induced esophageal carcinogenesis because these adducts produce GC→AT transition mutations in the second base of codon 12 of the Hras1 oncogene, a gene that is activated early in esophageal tumorigenesis (18). Both BRB and PEITC have been shown to reduce O6-methylguanine adduct levels in the esophagi of NMBA-treated rats (11, 19). In our companion report (13), we noted that the phase I enzymes, CYP2a2 and CYP3a13, were overexpressed in NMBA-treated rat esophagus, whereas their expression was normal in the esophagus of NMBA+PEITC rats. In the present study, CYP2a2 expression was also found to be near normal in BRB+NMBA–treated rat esophagus; however, CYP3a13 remained up-regulated. Thus, our data suggest that CYP2a2 may be important in the bioactivation of NMBA in rat esophagus and that down-regulation of CYP2a2 by both BRB and PEITC may be responsible for the observed ability of these chemopreventives to inhibit O6-methylguanine adduct formation in NMBA-treated esophagus.
Phase II enzymes lower the level of DNA adducts because they inactivate carcinogens (and/or their metabolites) by conjugating them to acceptor molecules (e.g., glutathione, glucuronic acid, and sulfates; ref. 6). The specifics of phase II enzymes involved in rat esophageal metabolism of NMBA have not been fully elucidated, although the formation of glucuronide conjugates has been reported (20). As we previously reported (13), PEITC corrected the NMBA-reduced expression levels of some glutathione S-transferases (i.e., Gsta1, Gsta2, and Gstt2). In contrast, expression of these genes remained low in the esophagi of BRB+NMBA rats, suggesting that overexpression of these phase II enzymes may not be sentinel for the inactivation of NMBA in rat esophagus. In keeping with this interpretation is the finding that BRB also antagonized NMBA-induced up-regulation of Taldo1, a key transaldolase in the pentose phosphate pathway that provides the NADPH needed for reducing glutathione (21).
Inducible nitric oxide synthase (iNOS) is overexpressed in many human cancers, including ESCC (22). We have reported that this radical generating enzyme is elevated in the esophagi of rats 15 and 25 weeks after NMBA treatment, whereas companion rats that also received the BRB diet had normal levels of this enzyme (23). In contrast, iNOS was not elevated in the esophagi of rats in the present study. Cyclooxygenase-2 (COX-2) expression and enzyme activity are also up-regulated in human ESCC (24), with a corresponding increased level of prostaglandin E2 and associated inflammatory processes. We have found that COX-2 was overexpressed in dysplastic lesions 15 and 25 weeks after NMBA treatment (23), but the expression level of COX-2 in the esophagi recovered from corresponding NMBA-treated animals in the present study was not elevated. Thus, iNOS and COX-2 dysregulation does not seem to be involved with the early phase of NMBA-caused esophageal carcinogenesis and the chemopreventive effects of BRB regarding the regulation of these genes are seemingly restricted to later stages of tumor development.
Among the DAVID clustering of the genes dysregulated by NMBA, listed in Table 1, are 37 genes involved in signal transduction. Of interest is the observation that several of these genes are associated with ras activity. Among these are Map2k3 and Apc1, whose dysregulation affects Hras1 activity. As noted above, ~ 100% of rat esophageal tumors induced by NMBA have a GC → AT transition mutation in the second base of codon 12 of the Hras1 gene (18). Map2k3is a mitogen-activated protein kinase (MAPK) kinase family member that is activated by stress to phosphorylate MAPK14/p38-MAPK, and elevated expression of the Hras1 oncogene results in the accumulation of the active form of MAPK14/p38-MAPK in breast cancer cells (25). Apc1, which is transcribed at increased levels in breast and colon cancers, has transforming capability, possibly through tyrosine dephosphorylation of the EphA2 receptor (26). Apc1 also interacts with several other receptor tyrosine kinases and docking proteins, including platelet-derived growth factor receptor and β-catenin. It is considered to be a negative regulator of growth factor–induced cell proliferation, but in Hras1-transformed cells, its overexpression increases cell proliferation (27). Thus, reducing NMBA up-regulated transcription of Map2k3 and Apc1 by BRB would antagonize the pro-oncogenic activities of activated Hras1. Other ras-related NMBA-dysregulated genes are Rab1 and H2A histone family member Z (H2afz). Rab1 is a member of the Ras oncogene superfamily and is up-regulated in tongue SCC (28). H2afz is interesting because it is an adaptor protein that interacts preferentially with the active form of Hras1 and augments cell growth. Thus, several genes associated with the ras signal transduction pathway were returned to homeostasis by the BRB diet.
Eighteen transcripts were classified as cellular differentiation and morphogenesis genes (Table 1). Two notable ones were Grem1 and the 14-3-3 family chaperone protein Ywhah. Grem1, which interacts with the Ywhah protein, encodes a secreted antagonist of the bone morphogenetic protein pathway, which in turn plays a crucial role in regulating the balance between expansion and cell differentiation (29). In addition to its interaction with Grem1, Ywhah mediates signal transduction via activation of protein kinase C and calcium/calmodulin-dependent protein kinase II. Loss of the chaperone activity of Ywhah may also play a role in oxidative signaling underlying oxidative damage (29). Another interesting transcript is sciellin, which encodes a precursor to the cornified envelope of terminally differentiated cells (30). Its down-regulation by NMBA could disrupt the normal differentiation program of esophageal squamous cells and promote cell transformation. Thus, the observations that BRB restore near-normal expression of Grem1, Ywhah, and sciellin could be mechanisms by which the berries impede esophageal cell transformation.
Twenty transcripts were classified by the DAVID program as cell junction, adhesion, or motility genes (Table 1). Among these is CD44 antigen, which encodes a cell surface glycoprotein involved in cell-cell interactions, cell adhesion, and migration. The v6 alternate transcript of CD44 is up-regulated in human ESCC (31). Another is Actn4, which encodes a nonmuscle actinin and is localized to moving structures and is significantly elevated in cells exhibiting enhanced motility. The level of this protein progressively increases from early- to late-stage ESSC (32). Another NMBA-caused overexpression normalized by BRB was moesin. Elevated moesin has been associated with oral squamous cell carcinomas (33). The protein is a membrane-cytoskeleton linker in microvilli, ruffles, and cleavage furrows and thus plays a key role in cell morphology, adhesion, and motility. Lastly, the up-regulation of Cdc42 by NMBA was also normalized by BRB. Thus, another possible chemopreventive mechanism of BRB may be to interfere with the formation of the Cdc42/Rac1 complex (34), which promotes cell migration. Overall, the data support the conclusion that an important mechanism of BRB chemoprevention is correcting expression of NMBA-dysregulated cell junction, adhesion, and motility genes.
Previous studies have shown that the BRB diet increased the level of apoptosis/cell death in tumors and tumor cell lines (4). The DAVID program analysis identified 14 genes (Table 1) that were expressed at normal levels in the NMBA+BRB rats compared with the NMBA animals. One is Hmox1; its increased expression is associated with resistance to induction of apoptosis by oxidative stress caused by a wide range of chemical injuries (35). Another is Mcl1, a member of the Bcl-2 family, which was regulated to near-normal levels by the BRB diet. Mcl1 protects mitochondrial integrity though suppression of cytochrome c release (36). Therefore, correction of Mcl1 transcription by BRB should augment apoptotic activity in NMBA-initiated cells. Expression of the gene Pip5k1a (predicted) was also corrected by the BRB diet. Overexpression of this gene can rescue cells from stress-induced apoptosis mediated by the activation of extracellular signal-regulated kinase (ERK) 1/2 signaling (37). Thus, appropriate regulation of apoptosis seems to be a mechanism of BRB chemoprevention.
The BRB diet also normalized the expression of several genes the DAVID program clustered with angiogenesis activities that were dysregulated by NMBA (Table 1). Among these was Klf5, which was down-regulated by NMBA. It is a Kruppel-like zinc finger transcription factor that modulates cell proliferation, differentiation, cell cycle, apoptosis, and angiogenesis. Klf5 seems to be a tumor suppressor for breast cancer (38). Thus, one mechanism of BRB may be to restore expression of this tumor suppressor factor. Fgfbp1 was up-regulated by NMBA; it is a secreted protein thought to enhance fibroblast growth factor activity and drive tumor angiogenesis. Fgfbp1 is up-regulated early during wound healing of mouse and human skin as well as during the initiation of skin neoplasia by chemical carcinogens (39). Thus, its normalization by BRB should depress angiogenesis. Interestingly, previous studies have shown that a 5% BRB diet down-regulates the expression levels of vascular endothelial growth factor-1 (VEGF-1) in the esophagus of rats when given for several weeks after treatment of rats with NMBA (40). VEGF-1 was not overly expressed in the present study, indicating that this gene, along with iNOS and COX-2, does not seem to be involved in the early stages of tumor initiation in the rat esophagus.
Another purpose of our study was to identify gene expression changes in NMBA-treated rat esophagus that were returned to homeostasis by both PEITC and BRB. We identified 53genes modulated by both treatments (Table 2). These genes are of special interest because they may be sentinel genes of NMBA-induced carcinogenesis and, as such, primary targets for chemoprevention. One of the most interesting among these genes is Pls3. Its overexpression increases cell proliferation and invasion of tumor cells, probably due to suppression of E-cadherin. Expression of this gene is increased in the lung tissue of mice treated with carcinogens, and its dysregulation is corrected by a diet containing indole-3-carbinol (41). Thus, Pls3 overexpression is antagonized by three different forms of chemopreventive diets (i.e., BRB, PEITC, and indole-3-carbinol) in both esophagus and lung rodent cancer models.
Other up-regulated genes that have been associated with cancer are Bub1, Crk, Map2k3, Psck9, PVR, Rab1, PDCD10, H2afz, and Cyp2a2. Map2k3, PVR, Rab1, H2afz, and Cyp2a2 are discussed above. Mutations in spindle checkpoint function kinase gene Bub1 have been associated with aneuploidy (42). The oncogene Crk is increased in several human cancers and its overexpression in cultured epithelial cells causes them to exhibit an altered morphology, to proliferate in soft agar, and to grow as massive tumors in nude mice (43). Psck9 is a Ca2+-dependent apoptosis-regulated convertase that stimulates tumor cell proliferation, motility, and invasiveness (44). PDCD10 encodes a protein with similarity to proteins that participate in apoptosis. It interacts with MST4 kinase to promote cell proliferation via modulation of the ERK pathway when overexpressed (45). Thus, returning all of these genes to near-normal levels of expression might be mechanisms by which PEITC and BRB impede esophageal cell transformation.
Also among the 53 genes are some that were not associated with a DAVID-identified cancer ontology. On closer review of the literature, however, they are of interest because they have activities that, if dysregulated, could have carcinogenic effects. Several of these genes were up-regulated by NMBA. Among these is Adamts6_predicted, a member of a family of 20 genes involved in tissue organization during embryogenesis and angiogenesis. Some of the proteins encoded by members of this family have matrix-degrading activity and may be involved in cell invasion (46). The gene Ddx5 is a transcriptional coactivator and/or corepressor, depending on the context of the promoter and the transcriptional complex in which it associates (47). Thus, its dysregulation may cause inappropriate expression of genes with oncogenic or tumor suppressor activities. The transcript NS5A (hepatitis C virus) transactivated protein 9 (Ns5atp9) is the rat homologue of human KIAA0101 whose function is unknown. However, it binds with proliferating cell nuclear antigen and is overexpressed in hepatocellular carcinomas (48). Transfection of this gene enhanced cancer cell growth and transformed NIH3T3 cells, whereas its inhibition caused attenuation of proliferation.
Both BRB and PEITC diets normalized the expression of seven genes that were down-regulated by NMBA. Of special note is Rbbp6, which encodes a 250-kDa ring finger-containing protein that is frequently up-regulated in human ESCC. This protein binds to underphosphorylated but not phosphorylated retinoblastoma protein (Rb). The phosphorylated Rb gene product binds the nuclear transcription factor E2F and prevents its ability to function in the S phase of the cell cycle (49). Thus, Rbbp6 competes with E2F for binding to the underphosphorylated form of Rb and high concentrations of the protein, as are found in human ESCC cells, would free E2F to stimulate cell proliferation. Rbbp6 also binds to p53, thereby enhancing Mdm2-mediated ubiquitination and degradation of p53, leading to decreased apoptosis. Consequently, down-regulation of Rbbp6 by BRB or PEITC should slow cell growth and increase the rate of apoptosis.
Another NMBA-dysregulated gene restored to normal levels of expression by BRB and PEITC was homologue of zebra fish ES1 (RGD1303003), which is the rat homologue of human C21orf33. This gene is required for the growth-inhibitory effect of all-trans retinoic acid on MCF-7 breast cancer cells (50). Its down-regulation by NMBA suggests a mechanism whereby the carcinogen promotes cancer by repressing terminal differentiation pathways. Mss4 protein (Mss4) binds to the membrane proximal conserved region of α-integrin chains and regulates the activation of inactive pro– matrix metalloproteins. The active forms of metalloproteins are important in tumor invasion and metastasis (51). Thus, it will be interesting to determine if and how down-regulation of Mss4 by NMBA contributes to esophageal carcinogenesis.
In summary, in this and our companion report (13), we have found that a short exposure of the rat esophagus to NMBA causes dysregulation of 2,261 transcripts that affect a multitude of cellular functions. Diets containing 5 µmol PEITC or 5% BRB restored 1,323 or 462 of these 2,261 transcripts to near-normal levels of expression, respectively. Thus, as might be expected, there is a direct correlation between the number of transcripts modulated to near-normal levels of expression and the inhibitory potential of PEITC and BRB in this model because PEITC is a more potent inhibitor of NMBA-induced rat esophageal tumorigenesis than BRB (3). In the present study, we discuss 53 transcripts that are dysregulated by NMBA and restored to homeostatic levels of expression by both BRB and PEITC. We speculate that these 53 genes are sentinel early changes in the process of NMBA-induced tumorigenesis in the rat esophagus, and future investigations will focus on elucidating their mechanistic roles.
Grant support: NIH grants RO1 CA103180 and R01CA96130 (G.D. Stoner). The microarray and bioinformatics work was facilitated by the Microarray and Bioinformatics Facility Core of the Environmental Health Sciences Center at Wayne State University (National Institute of Environmental Health Sciences Center grant P30 ES06639).
We thank Ronald Nines for his excellent technical assistance.