An experimental model developed in our laboratory was used for this study, i.e. a minimum microbiota showing resistance to Clostridium difficile colonisation isolated from the gut of the hamster. Only trixenic mice (C. indolis, C. cocleatum, Eubacterium sp.) were able to eliminate the target strain C. difficile. With the 'Swiss rolls' technique and scanning electron microscopy, it was shown that at each step of colonisation the caecal barrier microbiota was observed in the mucus present on the surface of the caecal mucous membrane and at the openings of the intestinal crypts, but never in the mucus situated inside the crypts. Moreover, some differences in mucus colonisation could be seen related to each colonisation step. In C. difficile monoxenic mice, these bacteria colonised the mucus in the same fashion as did the bacteria responsible for colonisation resistance. The mechanisms of this barrier effect against C. difficile remain unknown, although it may involve competition between C. difficile and the barrier bacterial population either for an energy substrate or for a specific receptor localised on the oligosaccharide chains of the mucin.

WIF-B9 is a suitable model for in vitro studies of hepatocyte polarity. To better understand polarity establishment, we have localized key proteins of the adhesion system, cytoskeleton, and tight junctions soon after plating, when most cells are isolated or in doublets. In isolated attached cells, only cytoskeletal proteins (tubulin, cytokeratins) displayed a precise localization. As soon as two cells formed a doublet, E-cadherin, alpha-, beta-, and gamma-catenins, and p120 protein were present at the doublet contiguous membrane. Actin, ezrin, and zonula occludens-1 (ZO-1) colocalized at this membrane, but not in all doublets: ezrin was present only at contiguous membrane expressing ZO-1, and ZO-1 was present only at membrane expressing actin. In contrast, occludin was spread throughout the doublet cytoplasm. With time in culture, these proteins localized transiently, as in cells expressing simple epithelial polarity, and finally, as in hepatocytes. We conclude that during WIF-B9 early polarization, key proteins are settled according to a hierarchy, as has been shown for Madin-Darby canine kidney cells. Cytoplasmic complexes of E-cadherin-catenin were detected during the whole polarization process; they were more abundant in fully polarized cells.

WIF-B cells were generated previously to obtain a good in vitro model expressing the structural and functional polarity of hepatocytes, Here we tested the stability and the strength of the WIF-B polarized phenotype. WIF-B cells stayed polarized and formed functional bile canaliculi even after 3 months in culture or after injection in nude mice and culture of the resulting tumors. WIF-B was subcloned and 10,000 colonies were examined; all (except for 3) were composed of bile canaliculi forming cells, Some subclones were characterized; the polarized ones presented the same properties and karyotype as the WIF-B cells; the 3 unpolarized subclones had a lower level of E-cadherin and different karyotypes, WIF-B cells were fused with their nonpolarized hepatic parental cells. The polarity state of the resulting FWIF hybrids was studied from day 11 to day 38 after fusion, by immunolocalization of hepatocyte domain-specific plasma membrane proteins. Most FWIF colonies (>80%) were composed of polarized cells. Soon after fusion these cells were exclusively polarized as simple epithelial cells. The percent of colonies containing cells expressing the typical hepatocyte polarity increased with time and reached 80% at day 38. This result confirms the two-step polarization process previously described for WIF-B, Chromosomally complete FWIF hybrids were examined several months after fusion, As shown by the study of bile acid transport and by confocal analysis of the localization of membrane domain markers, FWIF cells expressed a functional and fully polarized hepatic phenotype, In conclusion, polarity is a stable and dominant trait of WIF-B.

Gap junction intercellular communication (GJIC) plays an essential role in the control of growth, differentiation, and functions of different tissues. The expression of connexins (Cxs), the structural proteins of gap junctions, is developmentally regulated and tissue-specific. In vivo hepatocytes express Cx32 and Cx26. Most currently available in vitro hepatic cell systems express Cx43 instead of the expected Cxs. This work analyzes the GJIC competence and Cx expression of the highly differentiated and polarized hepatoma-derived hybrid cell lines, WIF 12-1 and WIF-B. It shows (using two dye transfer assays) that both lines communicate efficiently and that the acquisition of GJIC competence precedes the formation of bile canaliculi. Interestingly, these cells communicate via Cx32 expression, whereas Cx26 and Cx43 are not expressed, as demonstrated by Western and Northern blotting, immunocytochemistry and confocal microscopy: The human fibroblast WI38 parent communicates via Cx43, whereas the rat hepatoma parent Fao and the subclone WIF 12-1 TGF, that has lost the human X chromosome, do not communicate, the expression of Cx32 being restricted to the mRNA in these two lines. The GJIC competence of WIF cells could thus result from the activation of the human X chromosome-linked Cx32 gene.

By immunofluorescence and freeze fracture methods, we have studied the establishment of hepatic cell polarity in WIF-B9 cells, a subclone of the WIF-B rat hepatoma-derived hybrid cell line. As previously shown (Ihrke et al. (1993) J. Cell Biol. 123, 1761-1775; Shanks et al. (1994) J. Cell Sci. 107, 813-825), these cells are a suitable model for in vitro studies of various hepatic functions, particularly polarity: in confluent cultures, the majority of cells form bile canaliculus-like structures; membrane domains are settled, according to plasma membrane protein localization similar to rat hepatocytes in situ. We here report that the establishment of WIF-B9 cell polarity is a slow progressive biphasic phenomenon. During the first days of culture, the majority of cells do not make bile canaliculus-like structures. However, they display a polarity similar to that of simple epithelial cells: apical membrane proteins and villin are found at the cell apex; basolateral ones, excluded from this area, are expressed in the remaining membrane area; the tight junction-associated protein ZO-1 and actin are concentrated at the boundary of these two poles, whereas E-cadherin is present at the lateral pole just under the apex. With time in culture, the number of cells expressing this simple epithelial polarized phenotype decreases progressively and, after 10-15 days, depending on the plating density, nearly all the cells express the typical hepatic polarized phenotype. The expression of these two phenotypes is mutually exclusive. Freeze-fracture replicas of both types of polarized cells show either macula occludens, fascia occludens (simple epithelial polarity) or zonula occludens (hepatic polarity), associated with gap junctions. In this last case, two or three continuous strands are generally present all around the bile canaliculus-like structures.

An experimental model developed in our laboratory was used for this study, i.e. a minimum microbiota showing resistance to Clostridium difficile colonisation isolated from the gut of the hamster. Only trixenic mice (C indolis, C. cocleatum, Eubacterium sp.) were able to eliminate the target strain C. difficile. With the ‘Swiss rolls’ technique and scanning electron microscopy, it was shown that at each step of colonisation the caecal barrier microbiota was observed in the mucus present on the surface of the caecal mucous membrane and at the openings of the intestinal crypts, but never in the mucus situated inside the crypts. Moreover, some differences in mucus colonisation could be seen related to each colonisation step. In C. difficile monoxenic mice, these bacteria colonised the mucus in the same fashion as did the bacteria responsible for colonisation resistance. The mechanisms of this barrier effect against C. difficile remain unknown, although it may involve competition between C. difficile and the barrier bacterial population either for an energy substrate or for a specific receptor localised on the oligosaccharide chains of the mucin.

An experimental model developed in our laboratory was used for this study, i.e. a minimum microbiota showing resistance to Clostridium difficile colonisation isolated from the gut of the hamster. Only trixenic mice ( C indolis, C. cocleatum, Eubacterium sp.) were able to eliminate the target strain C. difficile . With the ‘Swiss rolls’ technique and scanning electron microscopy, it was shown that at each step of colonisation the caecal barrier microbiota was observed in the mucus present on the surface of the caecal mucous membrane and at the openings of the intestinal crypts, but never in the mucus situated inside the crypts. Moreover, some differences in mucus colonisation could be seen related to each colonisation step. In C. difficile monoxenic mice, these bacteria colonised the mucus in the same fashion as did the bacteria responsible for colonisation resistance. The mechanisms of this barrier effect against C. difficile remain unknown, although it may involve competition between C. difficile and the barrier bacterial population either for an energy substrate or for a specific receptor localised on the oligosaccharide chains of the mucin. Keywords - Clostridium difficile , Mucus, Barrier microbiota, Colonisation.

The 660 epitope was defined by a monoclonal antibody raised against rat gastric surface epithelium scrapings. This epitope, a marker of goblet cell differentiation, shows oncofetal behaviour in the colonic mucosa. We found that it co-purified with gastric mucin glycoproteins. We isolated rat gastric mucus glycoproteins using standard techniques: gastric scrapings in PBS were submitted to isopycnic density gradient centrifugation in CsCl in the presence of proteinase inhibitors. Fractions of relative density 1.4-1.45 with a high neutral sugar/protein ratio were chromatographed on an Ultrogel A4 column. According to the usual criteria, the high-molecular mass glycoproteins recovered in the excluded volume were purified mucins; when stained with periodic acid/Schiff reagent, they showed little migration on 4-15% gradient gel acrylamide electrophoresis. Serine+threonine+proline residues accounted for 35% of the total amino acids; the carbohydrate composition consisted of galactose, fucose, N-acetylgalactosamine and N-acetylglucosamine. These mucus glycoproteins carried the 660 epitope. After disulphide bond reduction, the remaining high-molecular-mass subunits were retained by the Ultrogel A4 column; amino acid and saccharide compositions were generally similar to those of the unreduced fraction. Trypsin digestion of the 660 epitope glycoprotein carrier did not modify its chromatographic and electrophoretic patterns, nor its chemical composition. The 660 epitope was still present after these treatments. However, trypsin digestion of subunits gave rise to smaller components that were retained by an Ultrogel A4 column. The saccharide composition of these fragments was unchanged, but the proportion of serine+threonine+proline residues rose to 46% of the total. These digested subunits had lost nearly all reactivity with monoclonal antibody 660. Our results fit well with the macromolecular model of Carlstedt, Lindgren and Sheehan [(1983) Biochem. J. 213, 427-435]: mucin glycoproteins are homopolymers of subunits assembled end-to-end via disulphide bonds into very large linear macromolecules. After disulphide bond reduction, proteolytic attack sites are uncovered and trypsin digestion results in glycopeptides bearing the typical oligosaccharidic units and with enhanced amounts of serine, threonine and proline, the characteristic amino acids of this hyperglycosylated region of the peptide core. These digested subunits have lost virtually all 660 epitope reactivity. We thus show that the 660 epitope, a determinant of a mucin molecule, is probably associated with the peptide core of the glycoprotein.

Periodate pretreatment of paraffin sections of ethanol-fixed gastrointestinal mucosae was used to characterize the carbohydrate or peptidic nature of mucin epitopes by immunoperoxidase. Immunoreactivity of monoclonal antibodies (MAbs) against histo-blood group related carbohydrate epitopes such as A, Lea, Lec, Sialosyl Lea, H type 2, I, T, Tn and sialosyl Tn dramatically decreased or even disappeared after periodate pretreatment of deparaffinized sections. In contrast, the immunoreactivity of MAbs against peptide mucin epitopes such as the gastric M1 mucin epitopes was almost unaffected by this treatment. Moreover, periodate treatment revealed cryptic peptide M1 mucin epitopes and the peptide MUSE11 epitope associated with the 20 amino acid tandem repeat (PDTRPAPGSTAPPAHGVTSA). An increase of cross-reactions of anti-human M1 MAbs with gastric epithelium of different vertebrate species was detected with periodate treatment. Our results suggest that this method can be useful for preliminary characterization of the biochemical nature of mucin epitopes (peptidic or saccharidic) and for demasking peptidic tumour markers which are hidden by saccharide molecules in normal tissues.

The gastric mucin M1 antigens, markers associated with colonic carcinogenesis, have been characterized by new anti‐mucin monoclonal antibodies (MAbs). These MAbs, obtained against mucins isolated from a human ovarian mucinous cyst (MAbs 19M1, 21M1 and 45M1) and from a pancreatic adenocarcinoma (MAb 96RA), were compared with 5 other anti‐M1 mucin MAbs described previously, which characterized the a , b, c, d and e mucin M1 epitopes. Using immunoperoxidase, these new MAbs exclusively stained the surface gastric epithelium of normal human gastro‐intestinal tract and reacted with fetal, precancerous and cancerous colonic mucosa, but not with normal colon. Immunoradiofixation studies showed that these new MAbs are directed against 3 epitopes ( f , g and h ) which are different from the a , b , c , d and e mucin M1 epitopes, though present on the same a immunoreactive high‐molecular‐weight components (> 1,000 kDa) with a density of 1.4 by CsCl‐density‐gradient ultracentrifugation. M1 antigenicity is characterized by a family of 8 different M1 epitopes which were destroyed with β‐mercaptoethanol (except for the f epitope), sensitive to a 5 hr trypsin treatment and resistant to 5 mM periodate (except for the h epitope). Some epitopes ( b , c and d ) showed increasing immunoreactivity after 20 mM periodate treatment, suggesting cryptic location. In rat‐colon adenocarcinomas, M1 mucin epitopes were masked but could be decrypted using high periodate treatment, similar to normal rat gastric mucosa, thus suggesting the absence of drastic changes in the saccharide coat of the peptide mucin portion bearing M1 epitopes. Cryptic location, periodate resistance, sensitivity to protease and conformational behavior strongly suggest that the peptidic core of gastric (or fetal colonic) mucin plays a role in M1 immunoreactivity. Indeed, the resurgence of M1 antigens during colonic carcinogenesis is due to re‐expression of the peptide core of gastric (or fetal colonic) mucins.

We investigated seven different procedures for chemical induction of rat colonic carcinogenesis using: repeated high doses (i) weekly 10 x 15 mg/kg, (ii) quarterly 8 x 15 mg/kg; repeated low doses (iii) weekly 27 x 1.5 mg/kg, (iv) quarterly 8 x 5 mg/kg, (v) quarterly 8 x 1 mg/kg; and single injections of 1,2-dimethylhydrazine (DMH) at (vi) 1 x 40 mg/kg or (vii) 1 x 20 mg/kg. Rats had typical histological precancerous lesions of the colon (dysplasia) and intestinal carcinomas in the six groups receiving a total dose of more than 8 mg/kg DMH. With doses of greater than 120 mg/kg, rats had more cancers, particularly intestinal carcinomas and significantly reduced survival. Rats receiving a single injection of 20 or 40 mg/kg also had histological lesions and colonic carcinomas. The group receiving a 40 mg/kg total dose in a single injection had a significantly higher frequency of colonic lesions per rat and a higher incidence of rats with colonic lesions than groups receiving the same total dose but fractionated. Control rats and groups injected with an 8 mg/kg total dose had no cancers, nor pre-cancerous histological lesions. Thus the carcinogenesis is dose-related, but for the same final dose a single injection gives more histological colonic lesions than several recurrent injections; however, the number of colonic carcinomas and the survival did not vary. This lack of correlation between the number of dysplasia and the number of adenocarcinomas may indicate that dysplasia is not always the precursor lesion of adenocarcinoma. With repeated low doses, we obtained colonic adenocarcinomas after a long period of latency in old rats, thus producing an experimental model with characteristics similar to those found in humans.

We report the characterization of an IgG2a monoclonal antibody, (MAb) 660, prepared against rat gastric high molecular weight glycoproteins. By immunoperoxidase staining, MAb 660 reacted only with the mucous cells of surface gastric epithelium and with a few duodenal goblet cells close to the pylorus in normal adult rats. In fetuses, it reacted with intestinal and colonic goblet cells. The adult colon was always negative. The MAb 660 stained 100% (30 of 30) of chemically induced colonic carcinomas and 100% (7 of 7) of duodenal carcinomas. Several weeks before the appearance of tumors, histologically normal glands, then hyperplasia and dysplasia were precociously stained with MAb 660. The tissue distribution was different from that of blood group related antigens and M1 fucomucins. The recognized antigen was not sensitive to neuraminidase treatment. After electrophoresis in polyacrylamide gel, staining with periodic acid-Schiff reagent and Western blotting showed that the MAb 660 recognized an epitope associated with high molecular weight glycoproteins. This epitope was unaffected by beta-mercaptoethanol reduction-periodate treatment and neuraminidase and trypsin digestion. However, trypsin digestion performed after beta-mercaptoethanol reduction destroyed the 660 epitope. These data suggest that the antibody could recognize the peptide moiety of the mucin rather than its carbohydrate moiety. Thus, the new antigen identified by MAb 660 is a mucin-type glycoprotein with an oncofetal behavior in the rat colon and is precociously expressed by precancerous colonic mucosa.

A group of 44 rats underwent the equivalent of a ureterosigmoidostomy (US), while a second group of 18 rats underwent a pediculated graft (PG) of urothelial tissue in the sigmoid wall. Histological lesions were observed in the colon near the bladder colon junction in US rats exclusively. These lesions included dysplasias (5/23), cystic glands (4/23) and 10 neoplasms (9/23), three of which were adenomas, showing elements of juvenile polyp and tubular adenoma in one case. The seven other tumors showed typical histological features of colonic adenocarcinomas, but no frank evidence of parietal tumoral invasion was observed and their cancerous nature was questionable. It is probably a true carcinogenesis since we induced the same histological changes as those in the mucosae adjacent to colonic adenocarcinomas after human US surgery. Moreover, by immunoperoxidase using antibodies against mucus associated antigens (Ml and M3C antigens) we demonstrated that US rat carcinogenesis differs from dimethylhydrazine (DMH) rat carcinogenesis. Furthermore, our results suggest that urine may be an important factor in inducing this type of US carcinogenesis.