Abstract
Carnosol, a constituent of the herb, rosemary, has shown beneficial medicinal and antitumor effects. Using the C57BL/6J/Min/+ (Min/+) mouse, a model of colonic tumorigenesis, we found that dietary administration of 0.1% carnosol decreased intestinal tumor multiplicity by 46%. Previous studies showed that tumor formation in the Min/+ mouse was associated with alterations in the adherens junctions, including an increased expression of tyrosine-phosphorylated β-catenin, dissociation of β-catenin from E-cadherin, and strongly reduced amounts of E-cadherin located at lateral plasma membranes of histologically normal enterocytes. Here, we confirm these findings and show that treatment of Min/+ intestinal tissue with carnosol restored both E-cadherin and β-catenin to these enterocyte membranes, yielding a phenotype similar to that of the Apc+/+ wild-type (WT) littermate. Moreover, treatment of WT intestine with the phosphatase inhibitor, pervanadate, removed E-cadherin and β-catenin from the lateral membranes of enterocytes, mimicking the appearance of the Min/+ tissue. Pretreatment of WT tissue with carnosol inhibited the pervanadate-inducible expression of tyrosine-phosphorylated β-catenin. Thus, the ApcMin allele produces adhesion defects that involve up-regulated expression of tyrosine-phosphorylated proteins, including β-catenin. Moreover, these data suggest that carnosol prevents Apc-associated intestinal tumorigenesis, potentially via its ability to enhance E-cadherin-mediated adhesion and suppress β-catenin tyrosine phosphorylation.
- Gastrointestinal cancers: colorectal
- Animal models of cancer
- Biological and biochemical mechanisms in prevention
- Diet and cancer
Introduction
Phenolic diterpenes and triterpenes are active constituents of Rosmarinus officinalis (rosemary), a commonly used herbal flavoring agent for foods. Of these compounds, much interest has been focused on the phenolic diterpene, carnosol ( Fig. 1 ), because of its beneficial medicinal effects. For example, the products of arachidonic acid metabolism are implicated in tumorigenesis, and carnosol inhibited both lipoxygenase and cyclo-oxygenase activities (1). Rosemary extracts also exhibit potent antioxidant activities that reduced lipid peroxidation, production of reactive oxygen species, and inflammation (2–4) . Among constituents of rosemary extract, 90% of the total antioxidant activity were derived from carnosol and carnosic acid (3). Rosemary and carnosol inhibited potentially mutagenic DNA damage by suppressing the metabolic activation of procarcinogens. Thus, rosemary extract produced an antimutagenic effect in the Ames assay (5), and carnosol inhibited benzo[a]pyrene-DNA adduct formation in bronchial cells (6). A complementary means of reducing carcinogen-DNA adducts in tissues results from stimulation of detoxification pathways that promote in vivo elimination of genotoxicants. Exposure to agents in rosemary induced the expression of enzymes that conjugate carcinogens to glutathione or glucuronic acid and that limit levels of reactive quinones (7–9) . Additional studies suggest that the bioavailability of this compound in vivo is sufficient to produce antitumor effects on target tissues as shown in rodent skin (10) and mammary tumor models (11). Collectively, these data suggest that carnosol inhibits multistage carcinogenesis at initiation, promotion, and progression stages in vivo.
Structure of carnosol.
Inherited mutations in the tumor suppressor gene, APC, in humans are the cause of familial adenomatous polyposis coli, and somatic mutations in this gene are present in >80% of sporadic colon cancers (12). The product of the APC gene negatively regulates cytoplasmic levels of free β-catenin. The potential of β-catenin to induce intestinal neoplasia depends on its ability to activate the expression of growth-promoting genes when bound to Tcf-4 in the nucleus. In normal intestinal cells, APC forms a complex with axin and the serine/threonine kinase GSK3β to phosphorylate β-catenin, thereby targeting it for degradation. The C57BL/6J/Min/+ (Min/+) mouse bears a germ line mutation in the homologue APC gene and develops 30 to 50 grossly visible adenomas during its lifetime (13). Adenomas in the intestines of Min/+ mice consistently show loss of heterozygosity of the remaining Apc+ allele, a finding that is similarly observed in human familial and sporadic colon adenomas and carcinomas (14). For these reasons, the Min/+ mouse is a useful colon cancer model for screening the efficacy of potential chemopreventive agents in vivo.
Previously, we associated the diminished localization of E-cadherin at the lateral membranes of Min/+ enterocytes with the increased expression of tyrosine-phosphorylated β-catenin (β-catenin-p-Y) in these cells when compared in parallel with the wild-type (WT) littermate (15). At the adherens junctions, homophilic ligation of E-cadherin in the plasma membranes of adjacent epithelial cells secures cell-cell adhesion (16). Binding of β-catenin to both E-cadherin and α-catenin strengthens adhesion by linking adherens junction constituent proteins to the actin cytoskeleton. Motile cells require that adhesion be dynamically alterable. Enterocytes proliferate in crypts then differentiate and migrate to villus tips or to the surface mucosa where they are exfoliated into the lumen. Tyrosine phosphorylation of β-catenin is one of several mechanisms that allow modulation of the adhesive function of E-cadherin. The epidermal growth factor receptor (EGFR) and c-Src family kinases phosphorylate β-catenin at distinct tyrosine residues, releasing it from associations with E-cadherin and α-catenin, respectively (17). Recently, we showed that Egfr activity is increased in histologically normal enterocytes and adenomas of Min/+ mice (18).
E-cadherin is a tumor and invasion suppressor in the gastrointestinal tract that negatively regulates β-catenin (19). In addition to its role in adhesion, E-cadherin engagement also evokes downstream signal transduction (19). Transgenic mice showed that E-cadherin-mediated cell-cell adhesion is required to regulate enterocyte differentiation and migration in the small intestine (20). We reported previously that Min/+ enterocytes display a significantly reduced migration rate within the crypt-villus unit and that tumor inhibition by effective chemoprevention regimens in Min/+ normalized enterocyte migration (21–23) . These findings in the Min/+ mouse add to the accumulating evidence suggesting that APC plays a role in positively regulating intercellular adhesion via effects on E-cadherin (24).
To develop strategies for dietary interventions that can be translated into regimens for use in humans, we have tested relatively nontoxic compounds at low doses for in vivo efficacy. Thus, in our previous studies, we showed that several plant-derived phenolic antioxidants were highly effective in preventing intestinal tumors in Min/+ mice (24–26) . In addition, we found that aspirin at a concentration equivalent to the 80 to 100 mg/d cardioprotective dose used in humans was effective in this model (27). In light of the potential impact of dietary chemoprevention on human health and the antitumor activities attributed to rosemary constituents, we investigated the effect of dietary inclusion of carnosol in the Min/+ model. We hypothesized that comparison of the relative amount of E-cadherin situated at lateral plasma membranes and the relative expression of β-catenin-p-Y in enterocytes of untreated and treated Min/+ tissues can serve as surrogate biomarkers for chemoprevention efficacy, because strong intercellular adhesion is inversely associated with tumor formation.
Materials and Methods
Materials. Five-week-old female Min/+ and Apc+/+ littermate mice were purchased from The Jackson Laboratory (Bar Harbor, ME). Carnosol was obtained from Kalsec (Kalamazoo, MI). This compound was pelleted into AIN-76A diet by Research Diets (New Brunswick, NJ) at a 0.1% concentration. All antibodies, reagents, and materials were as specified previously (15, 18) .
Dietary Treatments and Tissue Harvesting. On arrival, Min/+ mice and their WT littermates were evenly distributed into treatment groups of 10 animals each. They were then fed AIN-76A diet with and without 0.1% carnosol and given tap water to drink ad libitum. The procedures for animal care, euthanasia, tissue dissection, and tumor counting were as detailed (26). StatView was used to perform a two-tailed t test to determine statistical significance of tumor burden changes.
Ex vivo Carnosol Treatments, Lysate Preparation, Immunoprecipitation, and Immunoblot Analysis. All tissue treatment and enterocyte collection procedures were as detailed (18). Ex vivo treatments were done with 25 μmol/L carnosol dissolved in DMSO, 10 mmol/L pervanadate (18) in DMEM or this medium plus vehicle as the negative control. After treatments, the intestine was cut into proximal and distal halves and opened longitudinally. When dissecting Min/+ tissues, all visible adenomas were excised before cell collection. For total cell lysates, washed enterocytes were placed in dounces containing 1 to 2 mL lysis buffer [50 mmol/L Tris-HCl (pH 7.4), 150 mmol/L NaCl, 2 mmol/L MgCl2, 50 mmol/L NaF, 1 mmol/L NaVO4, 10 mmol/L Na4P2O7, 0.2 mmol/L phenylarsine oxide, 10 μg/mL leupeptin, 10 μg/mL aprotinin, 5 μg/mL pepstatin A, 1 mmol/L dichlorodiphenyltrichloroethane, 1 mmol/L phenylmethylsulfonyl fluoride, 10 mmol/L N-acetyl-leucyl-leucyl-norleucinal, 3 mmol/L H2O2, and 1% Triton X-100] and homogenized for 10 strokes. Lysates were clarified by centrifugation at 14,000 rpm for 10 minutes. Aliquots were removed for protein concentration determinations. Immunoprecipitations and immunoblotting were done as detailed (15). All immunoblots of immunoprecipitation samples included the corresponding input precleared total cell lysates as an internal control. All protein analysis experiments were repeated at least thrice using independently prepared lysates from different animals. Densitometric analysis was done using Un-Scan-It software. The ratio of the intensities of the corresponding heavy chain immunoglobulin and immunoprecipitation band in each of the gel lanes was used for quantitation.
Immunohistochemistry. Segments of proximal ileum from Min/+ and WT treated ex vivo with carnosol and pervanadate, respectively, together with the mock-treated control tissues were cut and formalin fixed immediately after treatments from sets of animals sacrificed for each experiment in parallel. Serial 4-μm sections of the paraffin-embedded tissues were deparaffinized and rehydrated. Antigen retrieval was done by placing slides in a pressure cooker for 2 minutes in 10 mmol/L citrate buffer (pH 6.0). Endogenous peroxidases were quenched with 3% H2O2 in methanol, and slides were then washed in PBS. The sections were reacted with rat anti-mouse E-cadherin (ECCD-2) antibody (1:250) for 2 hours at room temperature. Slides were washed in PBS and reacted with rabbit anti-rat IgG (1:5,000) for 30 minutes. After washing again, sections were reacted with Envision anti-rabbit-labeled polymer for 30 minutes. For β-catenin immunohistochemistry, the mouse tissue sections were blocked using the MOM reagent plus avidin D that were added for 1 hour followed by repeated washes in H2O. Monoclonal anti-β-catenin antibody (clone 14; 1:200) was used together with biotin block for 30 minutes at room temperature. After subsequent washes, secondary anti-mouse antibody was added for 30 minutes. Enzyme detection was done with 3,3′-diaminobenzidine chromogen. Sections were counterstained with hematoxylin, dehydrated, and coverslipped. Images were obtained using an Olympus BX40 microscope at 40× and 60× power. All immunohistochemistry data shown were from ex vivo–treated Min/+ and WT tissues subjected to the same immunohistochemistry procedures that were done at the same time.
Results
Dietary Carnosol Inhibited Tumor Formation in Min/+ Intestine. Following a 10-week administration of control diet (AIN-76A), Min/+ mice contained an average of 33.1 ± 14.1 intestinal adenomas, a result consistent with our previous studies (21–23, 25–27) , . No difference was noted in animal weight among treatment groups during the experiment. Animals treated for 10 weeks with AIN-76A diet supplemented with 0.1% carnosol showed a significantly reduced tumor multiplicity, with a mean of 18 ± 12.7 tumors per animal 46% decrease; Fig. 2 ). The distribution of tumors throughout the intestinal tract was also examined. From 82% to 87% of tumors were located in the small intestine, with no significant difference in this distribution between control and treatment groups. Significant treatment-associated reductions were observed for the duodenum and small intestine of the animals that received a 0.1% carnosol diet. Although no decrease in tumor number was detected for the colonic segment, the average number of colon tumors per mouse in the control group (0.2) was likely too low to allow adequate comparison.
Treatment with carnosol prevented tumor formation in the Min/+ mouse. Dietary inclusion of 0.1% carnosol decreased tumor multiplicity in female Min/+ mice by 44%. Tumor burden was significantly decreased (*, P = 0.021) in the entire bowel of Min/+ mice. Values for the tumor distribution in the untreated control Min/+ group for duodenum, small intestine, and colon were 4.8 ± 2.4, 28.2 ± 12.1, and 0.1 ± 0.3, respectively. Total tumor count for this group was 33.1 ± 14.1. Values for the tumor distribution in the carnosol-treated group for duodenum, small intestine, and colon were 2.9 ± 2.0, 14.9 ± 11.1, and 0.2 ± 0.6, respectively. Total tumor count for the treated group was 18.0 ± 12.7. Ps for the change induced by carnosol treatment for each segment of intestinal tissue were 0.068, 0.019, and 0.660, respectively. P for the total tumor numbers comparing treated and control groups was 0.021. These data indicate that tumor distribution remained the same in both control and carnosol-treated groups. Columns, mean (n = 10 in each treatment group); bars, SD.
Carnosol Treatment Enhanced E-Cadherin-Mediated Adhesion in Min/+ Enterocytes. Formerly, we showed that the tumor-prone small intestine of the Min/+ mouse is characterized by significant changes in cell-cell adhesion when compared with the WT littermate, including decreased E-cadherin localization at the lateral membranes of enterocytes, reduced association of E-cadherin with β-catenin, and increased expression of β-catenin-p-Y (15). However, adenomas of Min/+ showed strongly increased expression and association of β-catenin and E-cadherin, while levels of β-catenin-p-Y were minimal (15). We hypothesized that effective tumor prevention by carnosol would correct the adhesion defects in the histologically normal Min/+ mucosa. Therefore, this study focused on Min/+ enterocytes rather than on tumors. Enterocyte total cell lysates from WT, Min/+, and Min/+ mice treated with carnosol were subjected to immunoprecipitations using anti-β-catenin antibody. Subsequent immunoblotting with anti-E-cadherin antibody showed that Min/+ tissue exhibited decreased association of E-cadherin and β-catenin relative to WT ( Fig. 3A ). These samples were separately precipitated with anti-β-catenin antibody and immunoblot analysis using anti-phosphotyrosine antibody again showed that the overall level of β-catenin-p-Y was increased in Min/+ enterocytes. Ex vivo treatment of the intact small intestine of Min/+ with 25 μmol/L carnosol in DMEM for 30 minutes increased the association of E-cadherin and β-catenin to a level similar to that of the WT tissue ( Fig. 3A). This treatment also reduced expression of β-catenin-p-Y to a level resembling that of WT.
Carnosol normalized the association β-catenin with E-cadherin by reducing the expression of β-catenin-p-Y. Immunoprecipitations of enterocyte total cell lysates (300 μg) following ex vivo carnosol (25 μmol/L) treatment of Min/+ small bowel used the anti-β-catenin antibody (clone 14; 4 μg). Immunoblots shown were probed with the anti-E-cadherin antibody (clone 36; 1:5,000) and with the anti-phosphotyrosine antibody (4G10; 1:2,000; A, left). Densitometry of the data from three independent experiments on each of the immunoprecipitations (A, right). In this and subsequent immunoprecipitation analyses, the entire amount of the each immunoprecipitated sample was resolved by gel electrophoresis. Immunoblot analysis was done using the input total cell lysates (TCL) as controls for the immunoprecipitations and to compare overall E-cadherin and β-catenin steady-state expression in Min/+ and WT enterocytes (B). All immunoblots include biotinylated size markers in the left-hand lanes. Molecular masses of E-cadherin and β-catenin are 120 and ∼94 kDa, respectively. Commercially obtained HeLa cell lysate was included as a positive control for both of these proteins. For the E-cadherin or β-catenin immunoblots, each gel lane was loaded with 10 μg protein. The bottom portion of the β-catenin membrane was cut and probed for β-actin as a loading control. Protein concentration uniformity of the three input lysates served as control for the immunoprecipitations.
Our previous studies showed that Min/+ intestine exhibited decreased membrane localization of E-cadherin, although overall E-cadherin protein expression remained similar to WT (15). We did immunoblot analysis to assess E-cadherin expression in enterocyte lysates from WT, Min/+, and carnosol-treated Min/+. This analysis used the same lysates prepared for the immunoprecipitations ( Fig. 3A). As shown in Fig. 3B, carnosol treatment of the Min/+ small intestine did not alter E-cadherin or β-catenin expression. Immunohistochemistry of proximal small bowel tissue from WT, Min/+, and carnosol-treated Min/+ mice was done to determine the location of E-cadherin within enterocytes. Figure 4 shows that the normal localization of E-cadherin at lateral membranes of enterocytes is observed in the WT tissue (top left), but its loss from these membranes is evident in the corresponding Min/+ tissue (top right). Consistent with results from the immunoprecipitation analyses ( Fig. 3), enterocytes from Min/+ mice treated with carnosol showed a normalized E-cadherin localization at lateral plasma membranes (bottom right). Because the localization of E-cadherin was changed but its overall expression was not ( Fig. 3B), these results suggest that its internalization from the plasma membrane is increased in Min/+ relative to WT enterocytes. The localization of E-cadherin in the lateral plasma membranes of Min/+ enterocytes treated with carnosol indicates that the adhesion deficiency of the tumor-prone tissue is at least partially reversible.
E-cadherin lateral membrane localization in enterocytes is increased by ex vivo carnosol treatment of Min/+ and inhibited by pervanadate treatment of WT intact small bowel tissues. Immunohistochemistry using rat anti-mouse E-cadherin antibody (ECCD-2) was done on WT and Min/+ proximal small intestine sections cut perpendicular to the crypt-villus axis. Negative control WT and Min/+ sections (top) and ex vivo treatments of WT tissue with pervanadate (10 mmol/L) and Min/+ tissue with carnosol (25 μmol/L; bottom) are at ×40 magnifications and show the integrity of crypt-villus units. Higher magnification images of the corresponding tissues were arranged such that all of the depicted enterocytes have the same orientation with the brush borders appearing above and the basement membranes below.
Pervanadate is a cell-permeable tyrosine phosphatase inhibitor that increases overall protein tyrosine phosphorylation. I.p. injection of pervanadate into mice stimulated expression of β-catenin-p-Y within minutes (28). We incubated intact WT small intestine with freshly prepared 10 mmol/L pervanadate in DMEM for 30 minutes. Figure 4 (bottom left) shows that this agent induced the loss of E-cadherin from the lateral membranes of WT enterocytes, mimicking the appearance of the untreated Min/+ specimen. Negative control tissue was incubated in DMEM for 30 minutes and subjected to immunohistochemistry in parallel (top left). These histologic examinations show that enterocytes with normal morphology are maintained on the basement membrane in intact villi. These images provide an important control for the ex vivo treatments, as they confirmed preservation of the structural integrity of the intestinal tissue. Furthermore, alteration of E-cadherin localization in WT enterocytes treated with pervanadate emphasize that tyrosine phosphorylation is an important regulator of enterocyte adhesion in vivo.
Truncated APC proteins, such as those that occur in colon cancer, failed to associate with the plasma membrane and perturbed adhesion in epithelial cells via effects on the membrane localization of β-catenin (23). Because Min/+ enterocytes contain a truncated Apc protein and display defective intercellular adhesion,we anticipated that changes in the membrane localization of β-catenin would be detectable in Min/+ relative to WT. Using our tissue specimens, we examined the location of β-catenin by immunohistochemistry in WT and Min/+ as well as the treated samples. In the paired tissue sections of Fig. 5A , strong positively stained margins between untreated WT enterocytes are apparent (top left and bottom left). On the other hand, β-catenin staining of the normal Min/+ enterocytes showed in Fig. 5A a reduced amount of this protein at lateral membranes (top right and bottom right). This result is consistent with reported data of others (22). Immunohistochemistry of Min/+ tissues treated with and without carnosol ( Fig. 5B) showed that the lateral membrane localization ofβ-catenin was increased following exposure to carnosol. These results support the conclusion that the association of E-cadherin with β-catenin is reduced in the plasma membranes of Min/+ enterocytes from the normal condition in WT cells (15) and extends the data of Fig. 3. The ability of carnosol to correct the defect in β-catenin localization in the Min/+ tissue is consistent with its inducible effect on E-cadherin localization ( Fig. 4).
β-catenin immunohistochemistry in WT and Min/+ small intestines untreated and treated with carnosol and pervanadate, respectively. Paired WT and Min/+ small bowel sections were reacted with anti-β-catenin antibody (clone 14; A). Immunohistochemistry for β-catenin using paired sections of Min/+ small bowel treated with or without carnosol (25 μmol/L; B). Immunohistochemistry for β-catenin using paired sections of WT small bowel treated with or without carnosol (25 μmol/L) and pervanadate (10 mmol/L; C). Tissues, sectioning, and orientation were the same as in Fig. 4. Magnification, ×40. B and C, villi cut obliquely (right), an angle showing the circumferential staining at the periphery of cells.
Carnosol Inhibited Pervanadate-Inducible Effects in the Murine Small Intestine. Tyrosine phosphorylation is a post-translational modification crucial in signal transduction pathways governing cell growth, migration, and differentiation. Moreover, impaired down-regulation of tyrosine kinase activity is associated with cancer. Incubation of Madin-Darby canine kidney cells with pervanadate increased adherens junction–associated phosphotyrosine, whereas treatment of Rous sarcoma virus–transformed chicken cells with tyrophostin inhibitors of tyrosine kinases induced adherens junction reassembly (29). The suppression of β-catenin tyrosine phosphorylation induced by carnosol in the Min/+ intestine suggested that this compound inhibits tyrosine kinases and/or stimulates their negative regulators. To gain support for this idea, we treated WT tissue with 10 mmol/L pervanadate as above and did immunohistochemistry for β-catenin using control and pervanadate-treated WT small bowel tissues. In addition, we examined the WT tissue pretreated with 25 μmol/L carnosol for 15 minutes before the pervanadate exposure ( Fig. 5C). β-Catenin staining of lateral membranes was greater in the untreated WT enterocytes relative to the pervanadate-treated WT ones. Again, the appearance of the WT tissue after pervanadate treatment resembled that of untreated Min/+ ( Fig. 5B). Carnosol pretreatment of the WT mucosa blocked the loss of β-catenin from lateral plasma membranes ( Fig. 5C). To confirm these results, lysates from enterocytes that were collected in parallel following these treatments were used to perform immunoprecipitations with anti-phosphotyrosine antibody ( Fig. 6 ). Subsequent immunoblotting for β-catenin showed that inhibition of tyrosine phosphatase activity, as expected, increased the tissue levels of β-catenin-p-Y, yielding a phenotype similar to that of untreated Min/+ enterocytes. Pretreatment of WT mucosa with 25 μmol/L carnosol before pervanadate inhibited β-catenin-p-Y expression. Thus, these data suggest that carnosol affects adherens junction–associated protein phosphorylation by suppressing tyrosine kinase activity and/or stimulating tyrosine phosphatase activity in the mouse small bowel.
Carnosol pretreatment inhibited pervanadate-inducible tyrosine phosphorylation of β-catenin in WT enterocytes. Pretreatment of WT tissue with carnosol (25 μmol/L) for 15 minutes suppressed the expression of β-catenin-p-Y inducible by 10 mmol/L pervanadate. Immunoprecipitations and immunoblots were done as described in Fig. 3.
Discussion
Epidemiologic studies suggest that a diet rich in plant products reduces colon cancer risk in humans (30). Here, we showed that carnosol, an important constituent of rosemary, significantly reduced tumor multiplicity in the Min/+ mouse at a relatively low concentration (0.1%; Fig. 2). This work adds carnosol to a list of phenolic compounds that inhibit intestinal tumor formation in rodent models of colon cancer. Effective plant-derived compounds tested in the Min/+ mouse include (+)-catechin (25), mixed polyphenols in tea (31), curcumin (24, 32) , caffeic acid phenethyl ester (24), and coumesterol (23). Interestingly, several promising compounds, including resveratrol, epigallocatechin-3-gallate, and genistein, constituents of grapes, tea, and legumes, were ineffective in this model presumably because of poor bioavailability (23, 33, 34) .
Our data showed that carnosol directly promoted the lateral plasma membrane localization of E-cadherin and β-catenin in Min/+ enterocytes such that they resembled those of WT littermate ( Fig. 4). This change is inferred to enhance the adhesive function of E-cadherin. The immunohistochemistry for E-cadherin of untreated Min/+ tissue ( Fig. 4) may reflect various fates that allow steady-state E-cadherin expression to be minimally altered ( Fig. 3B). For instance, membrane trafficking dynamically regulates E-cadherin function (34). Data of Fig. 4 suggest that the loss of E-cadherin from the plasma membrane of Min/+ enterocytes may indicate its increased internalization by endocytosis. This may explain the reduced amounts of β-catenin localized to the plasma membrane of these cells as well ( Fig. 5). E-cadherin can be endocytosed in a complex with β-catenin, and a pool of this oncoprotein may also leave the plasma membrane of Min/+ enterocytes in this manner (35). Interactions with certain cytoplasmic proteins inhibit the adhesive function of E-cadherin but have not been associated with increasing its turnover (36, 37) . If such associations are more prevalent in Min/+ relative to WT enterocytes, they would provide alternative or additional explanations for the adhesion deficiency observed in the Min/+ cells.
We also showed in Fig. 3A that carnosol treatment of the Min/+ small intestine reduced the expression of β-catenin-p-Y, a modification that abrogates its binding to E-cadherin (17). Migrating cells maintain dynamic cell-cell adhesions via production of β-catenin-p-Y, and this modification expands the pool of free cytoplasmic β-catenin that is resistant to degradation (38). Tyrosine phosphorylation of β-catenin occurs due to interactions with several membrane receptor and cytoplasmic tyrosine kinases, including EGFR (17, 39–41) , c-Met (42), and c-Src family kinases (17, 43) . Certain plant phenolic compounds that are potent antioxidants were highly effective in preventing tumor formation in Min/+ (24, 25) , suggesting that genetic and dietary factors cooperate to cause oxidative stress on the intestinal mucosa of this animal. Oxidative processes play an important role in regulating theactivities of both tyrosine kinases and their negative regulators, protein tyrosine phosphatases (PTPase). For instance, H2O2 stimulates the signaling potential of ligand-activated EGFR and other growth factor receptors (44–47) . In separate genetic and biochemical studies, Roberts et al. and our laboratory showed that Egfr activity promotes tumorigenesis in Min/+ (18, 48) . H2O2 also increases the duration and affects the location of EGFR signaling by inhibiting receptor down-regulation (49, 50) . Interestingly, several of the plant phenolic compounds that were effective inhibitors of tumor formation in the Min/+ mouse also were found to inhibit EGFR activation (51–53) . Regarding c-Src family kinases, overexpression of these kinases occurs in polyps and in most colon cancers, and their up-regulated activities inhibit E-cadherin-dependent cell-cell adhesion (54, 55) .
PTPases dephosphorylate β-catenin-p-Y, and their activities are inhibited by oxidative stress. Numerous receptor and cytosolic tyrosine kinases dephosphorylate β-catenin and promote cell-cell adhesion (36, 56, 57) . All PTPases contain conserved catalytic motifs with an essential cysteine residue in the active site, and H2O2 induces reversible PTPase inactivation (58, 59) . The association of β-catenin to receptor PTPases β/ζ occurred within the active site of this receptor (60). We showed that Min/+ enterocytes contained a reduced association of β-catenin with receptor PTPases β/ζ when compared with WT (15). It is not known whether oxidation of PTPases precludes β-catenin binding, but this is an attractive possibility. Thus, our data ( Figs. 5 and 6) suggest that inhibition of tumorigenesis in the gut by phenolic antioxidants, including carnosol, suppresses tyrosine kinase activities, stimulates PTPase activities, and promotes E-cadherin-mediated adhesion in part by reducing the expression of β-catenin-p-Y in intestinal cells.
The lumen of the intestine is a monolayer of polarized epithelial cells that are adapted for absorption. We anticipated that ex vivo treatment of the intact small bowel would allow certain rapid drug-inducible changes to be evaluated in a manner similar to cultured cells but with several advantages. We are interested in the physiology of initiated but histologically normal enterocytes in their tissue-specific stromal environment. Thus, our approach may obviate certain biases resulting from the use of nonpolarized, immortalized, and/or transformed cells in culture. In addition, it is probable that lower, more physiologically relevant, concentrations of drugs can be used in the intact tissue to illicit more specific responses than in cultured cells. Moreover, enterocytes removed from their basement membrane rapidly undergo apoptosis (61). Others and we showed the histology of scraped intestinal villi (15, 61) . Enterocytes evaluated after this collection technique suffer morphologic changes that preclude assessment of inducible changes in the localization of proteins, such as β-catenin and E-cadherin. In contrast, the techniques described here using the intact small intestine showed such changes well ( Figs. 4 and 5). Protein tyrosine phosphorylation is a rapidly inducible response to changes in the extracellular environment (28, 29) . This modification of β-catenin disassembles adherens junctions and augments a stabilized pool of this oncoprotein (37). In this study, we associated inducible changes in β-catenin-p-Y expression with chemoprevention efficacy in a tumor-prone tissue. This ex vivo treatment approach should be useful in evaluating other agents with the potential to suppress tumorigenesis in vivo. In addition, treatments of freshly harvested intestinal tissues with rapid-acting, selective inhibitors can test the tissue-specific relevance of particular signaling pathways to mechanisms of carcinogenesis. Finally, this assay design can be adapted for use on human colon biopsy specimens.
Acknowledgments
Grant support: National Cancer Institute, NIH grant R29CA74162 (M.M. Bertagnolli), Irving Weinstein Foundation (A.M. Carothers), and NIH Surgical Oncology research training grant T32-CA68971 (M.J. Weyant).
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in ccordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Footnotes
-
Note: A.M. Carothers is currently at the Department of Surgery, Brigham and Women's Hospital, 75 Francis Street, Carrie Hall, Room 116, Boston, MA 02115.
- Received May 13, 2004.
- Revision received October 5, 2004.
- Accepted November 18, 2004.
- ©2005 American Association for Cancer Research.