Functional evidence for a novel human breast carcinoma metastasis suppressor, BRMS1, encoded at chromosome 11q13.

We previously showed that introduction of a normal, neomycin-tagged human chromosome 11 reduces the metastatic capacity of MDA-MB-435 (435) human breast carcinoma cells by 70-90% without affecting tumorigenicity, suggesting the presence of one or more metastasis suppressor genes encoded on human chromosome 11. To identify the gene(s) responsible, differential display comparing chromosome 11-containing (neo11/ 435) and parental, metastatic cells was done. We describe the isolation and functional characterization of a full-length cDNA for one of the novel genes, designated breast-cancer metastasis suppressor 1 (BRMS1), which maps to human chromosome 11q13.1-q13.2. Stably transfected MDA-MB-435 and MDA-MB-231 breast carcinoma cells still form progressively growing, locally invasive tumors when injected into mammary fat pads but are significantly less metastatic to lungs and regional lymph nodes. These data provide compelling functional evidence that breast-cancer metastasis suppressor 1 is a novel mediator of metastasis suppression in human breast carcinoma.

Breast cancer metastatic potential correlates with a breakdown in homospecific and heterospecific gap junctional intercellular communication.

Breast cancer progresses toward increasingly malignant behavior in tumorigenic and metastatic stages. In the series of events in the metastatic stage, tumor cells leave the primary tumor in breast and travel to distant sites where they establish secondary tumors, or metastases. In this report, we demonstrate that cell-cell communication via gap junctions is restored in the metastatic human breast carcinoma cell line MDA-MB-435 when it is transfected with breast metastasis suppressor 1 (BRMS1) cDNA. Furthermore, the expression profile of connexins (Cxs), the protein subunits of gap junctions, changes. Specifically, the expression of BRMS1 in MDA-MB-435 cells increases Cx43 expression and reduces Cx32 expression, resulting in a gap junction phenotype more similar to normal breast tissue. Taken together, these results suggest that gap junctional communication and the Cx expression profile may contribute to the metastatic potential of these breast cancer cells.

Molecular determination of perivesical and lymph node metastasis after radical cystectomy for urothelial carcinoma of the bladder.

PURPOSE: Current methods used to determine the pathological stage of the primary tumor and associated lymphatics after radical cystectomy are tedious, costly, and may lack the sensitivity afforded by molecular approaches such as reverse transcription-PCR (RT-PCR) for markers specific for urothelial tissue such as the uroplakin II (UPII) gene. Thus, we sought to evaluate an objective and sensitive molecular approach for the assessment of perivesical extension and lymph node status after radical cystectomy, based on the detection of UPII expression using RT-PCR and compare this assay to standard clinical and pathological examination. EXPERIMENTAL DESIGN: From November 1999 to September 2000, 27 patients with clinical T(a)-T(3)N(0)M(0) urothelial bladder cancer underwent radical cystectomy, 19 (70%) of which also had pelvic lymphadenectomy. At the completion of cystectomy, systematic biopsies of the external surface of the bladder specimen as well as from the largest palpable lymph node found at lymphadenectomy were obtained for molecular analysis. RT-PCR analysis for UPII mRNA was carried out on these biopsy specimens, and results were compared with data obtained from conventional pathological examination. RESULTS: Pathologically organ-confined tumors had a 42% (5 of 12) incidence of positive signals in the perivesical tissues and 17% (1 of 7) in the lymph nodes. Corresponding percentages for pT(3a)N(0) and pT(3b)-T(4)N(0) lesions were 67% (4 of 6)/25% (1 of 4) and 67% (4 of 6)/33% (2 of 6), respectively. Overall, pathologically node-negative cancers had a perivesical positivity rate of 54% (13 of 24) and a lymph node positivity rate of 25% (4 of 16). All patients with pathologically positive nodes had positive UPII signals in the lymph node sample. CONCLUSIONS: This molecular assay aimed at assessing perivesical extension and lymph node status after radical cystectomy appears to identify patients that may harbor residual disease not appreciated by conventional histology. Larger studies with 5-7-year follow-up will be required to determine the prognostic significance of such molecular information.

The relationship of BRMS1 and RhoGDI2 gene expression to metastatic potential in lineage related human bladder cancer cell lines.

We have recently characterized a human bladder cancer cell line T24 and a more aggressive lineage related variant of it, T24T. To gain further insights, we have studied their metastatic ability in an in vivo model system. Results show that T24 forms significantly fewer [4/12 (1/11) mice had metastases with 1-2 lesions/mouse] metastasis in SCID/bg mice than T24T [14/14 (6/6) mice had metastases with a mean of 24-28 lesions/mouse]. To begin exploring the mechanisms underlying this difference, we evaluated the mRNA and protein expression levels of metastasis-suppressor genes, known to be important in the progression of other cancers, in our model of bladder cancer progression. A higher mRNA expression of BRMS1, a metastasis suppressor in breast cancer, was observed in T24 cells. In addition, RhoGDI2 mRNA expression was only observed in T24 when compared to T24T, suggesting that Rho activation might play a significant role in the metastatic cascade. However, a basal level mRNA expression of KISS1, described as metastasis suppressor in melanoma and breast, was observed in both the lines and had slightly higher expression in T24T. No difference of Nm23-H1, KAI1, MKK4/SEK1 and E-Cadherin protein levels were noted between these two lines. In summary, it appears that the T24/T24T paired cell lines constitute a useful model for the study of human bladder cancer metastasis that will allow both the discovery and mechanistic evaluation of genes potentially involved in this process.

Analysis of mechanisms underlying BRMS1 suppression of metastasis.

Introduction of normal, neomycin-tagged human chromosome 11 (neo11) reduces the metastatic capacity of MDA-MB-435 human breast carcinoma cells by 70-90% without affecting tumorigenicity. Differential display comparing MDA-MB-435 and neo11/435 led to the discovery of a human breast carcinoma metastasis suppressor gene, BRMS1, which maps to chromosome 11q13.1-q13.2. Stable transfectants of MDA-MB-435 and MDA-MB-231 breast carcinoma cells with BRMS1 cDNA still form progressively growing, locally invasive tumors when injected in mammary fat pads of athymic mice but exhibit significantly lower metastatic potential (50-90% inhibition) to lungs and regional lymph nodes. To begin elucidating the mechanism(s) of action, we measured the ability of BRMS1 to perturb individual steps of the metastatic cascade modeled in vitro. Consistent differences were not observed for adhesion to extracellular matrix components (laminin, fibronectin, type IV collagen, type I collagen, Matrigel); growth rates in vitro or in vivo; expression of matrix metalloproteinases, heparanase, or invasion. Likewise. BRMS1 expression did not up regulate expression of other metastasis suppressors, such as NM23, Kai1, KiSS1 or E-cadherin. Motility of BRMS1 transfectants was modestly inhibited (30-60%) compared to parental and vector-only transfectants. Ability to grow in soft agar was also decreased in MDA-MB-435 cells by 80-89%, but the decrease for MDA-MB-231 was less (13-15% reduction). Also, transfection and re-expression of BRMS1 restored the ability of human breast carcinoma cells to form functional homotypic gap junctions. Collectively, these data suggest that BRMS1 suppresses metastasis of human breast carcinoma by complex, atypical mechanisms.

Effects of dietary bile acids on formation of azoxymethane-induced aberrant crypt foci in F344 rats.

The present study has demonstrated the influence of bile acids (BAs) on the development and growth of azoxymethane (AOM)-induced aberrant crypt foci (ACF). Male F344 rats were treated with two doses of AOM (15 mg/kg) at 7 days apart and fed either basal MF or MF plus 0.4% of cholic (CA), deoxycholic (DCA), chenodeoxycholic (CDCA), lithocholic (LCA) and ursodeoxycholic (UDCA) acid mixed diets for 8 weeks after the first AOM dose. The mean number of ACF/colon of the rats fed CA, DCA, CDCA and LCA were higher than that of MF-fed group and the differences were statistically significant (P < 0.005). But the mean number of ACFs/colon was significantly (P < 0.005) lower in UDCA diet-fed rats compared to MF. UDCA-fed rats also showed a significant decrease in average crypt multiplicity (number of crypts/focus) of ACF compared to MF alone. The mean number of ACF with > or =5 crypts was about 2.5-3.7 times higher in case of CA, DCA, CDCA and LCA and about 8.2 times lower in UDCA compared to the control MF diet group. In a parallel study, feeding for 18 weeks of the same BAs mixed diets without AOM administration did not significantly induce ACF. Therefore, these data suggest that dietary BAs by themselves do not induce ACF in F344 rats but enhance or, in the case of UDCA, suppress the development and growth of AOM-induced ACF.

Effect of bile acids on formation of azoxymethane-induced aberrant crypt foci in colostomized F344 rat colon.

The effect of bile acids on the formation of azoxymethane induced aberrant crypt foci (ACF) was investigated using the fecal stream-excluded colons of colostomized F344 rats. The excluded colon was irrigated with saline or bile acids (1 mg/0.5 ml per day, 5 days/week) for 4 weeks. The mean numbers of ACF per colon in rats given cholic acid, deoxycholic acid (DCA), chenodeoxycholic acid (CDCA), lithocholic acid, and ursodeoxycholic acid (UDCA) were 160.8, 118.2, 227.8, 150.7 and 87.3, respectively, while that of the control was 174.0. The number of ACF was significantly larger in CDCA, but smaller in UDCA and DCA-treated rats than the control (P<0.01). DCA did not induce apoptosis in the colon under the present conditions.

DNA adduct formation by hormonal steroids in vitro.

We examined the binding of various steroid hormones to DNA in vitro by means of 32P-postlabeling. Seventeen steroid hormones and cholesterol (CS) were incubated with human liver DNA at 37 degrees C for 1 h under aerobic conditions in the absence of catalysis. The reaction mixtures were analyzed by the nuclease P-1 version of 32P-postlabeling. The results showed that cortexolone (CX), prednisolone (PS), cortisone (CN), cortisol (CL), tetrahydrocortisol (TC), corticosterone (CC), 11-deoxycorticosterone (DC), dexamethasone (DX), dihydrocortisol (DL), and aldosterone (AL) covalently bound with DNA. However, progesterone (PG), 17 alpha-hydroxyprogesterone (HG), estrone (E1), estradiol (E2), estriol (E3), testosterone (TS), cortol (CR) and the original compound for biosynthesis, CS, did not form adducts. In absence of DNA, the steroids themselves did not give rise to any spot on TLC under the same conditions. The dose-responses of DNA binding by DC, DL, CC, CL and CN were linear. The relative adduct labeling of reactive steroids at a concentration of 2 mM were as follows: 68.8 (CX), 53.2 (PS), 39.6 (CN), 29.9 (CL), 20.9 (TC), 12.9 (CC), 12.3 (DC), 7.5 (DX), 4.7 (DL), 1.2 (AL) adducts per 10(8) nucleotides. Reactive and nonreactive steroids were distinguishable by the presence or absence of the carbonyl group (-CO-CH2OH) at carbon seventeen (C17) of the cholesterol skeleton. This implies that the electrophilic carbonyl or a neighboring group perhaps involved in the formation of covalent bond with DNA. To investigate the nature of target base(s) of these DNA reactive steroids, mononucleotides of all four bases of DNA were reacted with CN, CL, CC and cochromatographed with the obtained spots of DNA reactions. The results of which stated that these steroids and guanine reaction gave the same spots as observed in DNA reaction, indicating guanine is the main target of these DNA reactive steroids. Hep G2 human hepatocellular carcinoma cells were used as an alternative model. Although nine steroids (CL, DL, TC, PS, DX, PG, E2, TX, CR) did not react with intracellular DNA under our experimental conditions, our findings suggested that some hormonal steroids can form covalent DNA adducts in vivo.

In vitro formation of DNA adducts with bile acids.

The in vitro experiment was carried out following 32P-postlabeling analysis to determine the DNA-reactive bile acids present in the human body. The unconjugated and conjugated forms of cholic (CA), chenodeoxycholic (CDCA), deoxycholic (DCA) and lithocholic acid (LCA) were added to calf thymus DNA followed by 1 h of incubation at 37 degrees C. After the incubation the mixture was analyzed by the nuclease P1 modification of 32P-postlabeling. Among the 12 bile acids tested, our results showed that unconjugated CDCA and LCA and the glycine and taurine conjugates of LCA (g-LCA, t-LCA) were able to bind covalently with naked DNA in vitro without intervention of any catalyst. It was also shown that bile acid alone did not give any spot on TLC. These binding reactions depended on the bile acid concentration in a linear manner. The data on the extent of binding at a concentration of 0.1 mg/ml showed values of 28.5 (t-LCA), 23.7 (g-LCA), 3.47 (LCA) and 1.32 (CDCA) adducts per 10(8) nucleotides. These reactive bile acids were also incubated with COLO 205 human colon carcinoma cells and Hep G2 human hepatocellular carcinoma cells for 24 h, but no specific DNA adduct was formed. Further, when LCA or CDCA was administered into male Fischer 344 rats by gavage at a dose of 10 mg/rat every 8 h for 3 days, no specific DNA adduct was detected in their liver or colon. Covalent DNA adducts are believed to cause alteration of the primary structure of genes, which is potentially linked to carcinogenesis. Though our preliminary data failed to detect any bile acid-related DNA adducts in the cultured cells or in rats, the results may provide a basis for assuming some of these bile acids to be potential initiators of colon cancer.

Mucosa-specific DNA adducts in human small intestine: a comparison with the colon.

The presence of several covalent DNA adducts in the human colon was demonstrated by 32P-postlabeling in a previous study. We demonstrated that DNA of all the colonic mucosa tested were selectively adducted by a single genotoxic agent and this modification was completely absent in the DNA of muscular layers. In this study, the DNA adducts of the small intestine are compared with those of the colon to understand the role of mucosa-specific DNA adduct (MSA) in intestinal carcinogenesis. The mucosal DNA of the small intestine from 19 adults undergoing surgery due to gastric carcinoma (seven cases), pancreatic carcinoma (four cases), colon carcinoma (four cases), small intestinal tumor (two cases), intestinal trauma (one case) and ectopic pancreas (one cases) were analyzed. The mucosa-adjacent muscular layer DNA of the corresponding samples was examined as a control. Several common DNA adducts were observed in both mucosal and muscular layers of all the adults. Besides these common background DNA adducts, two MSAs (Si1, Si2) were detected in most of the adults as in colon cases. Si1 was present in all adults examined (19/19 cases) at a level of 0.04-0.22 adducts/10(8) nucleotides (average 0.09) and Si2 was found in 13/19 patients at a level of 0.03-0.08 adducts/10(8) nucleotides (average 0.05). Si2 was the same adduct detected in the adult colonic mucosa. However, they were absent in the adjacent muscular layer as well as in the neonatal intestine tested as a control. The total level of the mucosa-specific DNA adducts of the small intestine was approximately 28-fold lower than that of the colon. Considering that the incidence of cancer in the small intestine is rather lower than that in the colon, these results may be relevant to the development of human intestinal cancer.

Presence of mucosa-specific DNA adduct in human colon: possible implication for colorectal cancer.

DNA of normal mucosa and the adjacent muscular layer from 18 adults suffering from colorectal neoplasms was examined by 32P-post-labeling analysis in order to estimate the exposure of the human colon and rectum to environmental carcinogens. Colorectal DNA samples obtained from six newborns were also examined as a normal control because they were presumed to have been minimally exposed to environmental carcinogens. One common mucosa-specific DNA adduct was found in the normal colorectal wall in all adults at the level of 0.10-34.13 adducts/10(8) nucleotides (mean +/- SD: 3.64 +/- 7.92 adducts/10(8) nucleotides), however, these were absent from the newborns' colons. Although several common spots were present in the mucosa, muscular layer and newborn tissues, there was no muscular layer-specific DNA adduct. The relationship between the levels of the mucosa-specific DNA adduct in the non-cancerous part and the histological degree of malignancy was not significant. The presence of this mucosa-specific DNA adduct in adult colon suggests that the human colon is commonly exposed mainly to one environmental carcinogen. This carcinogen is supposed to originate from foods, because the incidence of colorectal carcinoma is closely linked to dietary habits and the mucosa-specific DNA adduct was not present in newborns who had never ingested food. The incidence of adult colonic cancer originating from its mucosa is high, while cases of muscular origin or in newborn colon are rare. Therefore, the mucosa-specific DNA adduct is presumably responsible for the development of colonic cancer of epithelial origin.