The method developed to isolate cell material from the bovine jejunum and spiraled colon led to the production of suspension enriched in undissociated cell aggregates (or organoids; Figure 1A) adhering to collagen coated culture flasks. They give rise to circular proliferating foci (Figure 1B). Within 5 days the initial foci enlargement had led to the fusion of cell plaques in one confluent layer (Figure 1C). Confluent monolayers arising from the initial seeding of cultures presented a heterogeneous aspect (Figure 1C) due to the presence of residual multicellular organoids dispersed in the newly formed monolayer. With the first passage jejunocyte and colonocyte cultures acquired a homogeneous pavement-like aspect typical of epithelial sheets. We generally carried out 7 to 9 passages (with split ratio 1:2 or 1:3) over 4 to 5 weeks in both cell series (jejunocytes and colonocytes) without major changes of cell morphology. During early passages (1–3), the confluence recovery was achieved within 2 to 3 days and then a progressive slowdown in the cell proliferation rate was observed until cessation. Confluent monolayers of both cell types exhibited domes, appearing as round blurred arrays corresponding to fluid entrapment between the flask wall and the basolateral side of the monolayer, resulting from ionic regulations by functional epithelial cells (Figure 1D). This observation is in agreement with dome formation in cultures of the human intestinal cell lineage Caco-2 28.

Ultrastructural analyses of cultured cells revealed apical tight junctions (Figure 2A) between cells forming the monolayer of both cell types. Cultured jejunocytes and colonocytes also exhibited a few apical microvilli (Figure 2A, B).

To ascertain the identity and functionality of cultured monolayers, cells components from various culture passages were submitted to immunolabelling (immunocytochemistry and Western blot). These data were compared to results from gene expression studies (RT-PCR) and measurements of specific activities of brush border associated enzymes.

Antibodies directed against cytokeratins (intermediary type II filaments specifically expressed in epithelia) and vimentin (intermediary type III filaments) were used to distinguish cells of epithelial origin from contaminating fibroblasts. Double immunolabeling of cryosections of the bowel wall showed the presence of the cytokeratin positive cells all over the lining epithelium and vimentin positive cells in the submucosa (Figure 3A). Cytokeratins showing the characteristic disposition of intermediate filaments were also recognized in obtained monolayers. Surprisingly, double staining vimentin-cytokeratin (Figure 3B) clearly showed that cultured cells expressed both epithelial and mesenchymal markers. However, the double staining cytokeratin-α-actin (Figure 3C) distinguished separate cells populations. The few α-actin positive cells, probably contaminant myofibroblasts (and/or smooth muscle cells) were released from the gut wall during the dissociation procedure and not completely eliminated by the sorbitol centrifugation or the selective attachment. Indirect immunofluorescence revealed also the membrane distribution of the epithelial specific antigen (ESA, a cell surface glycoprotein) in cultured enterocytes (Figure 4A) and intracellular distribution of the cytokeratin 18 (Figure 4B). E-Cadherin representing one of the proteins located at the adherence junctions was also detected, but a mis-localization was observed. The cells did not present the characteristic staining pattern to the cell periphery, but a granular distribution into the cytoplasm (data not shown).

Western blot analyses also confirmed the presence of epithelial characters in cultured cells. Pan-anti-cytokeratin and anti-cytokeratin 18 antibodies recognised the same major 45 KDa protein (Figure 5A, B) in homogenates from: (1) freshly scrapped intestinal epithelium, (2) organoid suspensions used to seed cultures, (3) intestinal cell cultures of both types, and in the (4) positive control Caco-2 cells. No cytokeratin staining was noticeable in 3T3 fibroblasts. The size uniformity of stained products in all tested epithelial cells was also confirmed using the pan-cytokeratin antibody (major signal at 45 KDa, Figure 5C, D). In addition, blots were also positive for another epithelial cell marker, the E-cadherin (around 35 KDa; Figure 5G, H). Compared to the expected size (120 KDa), the anti-E cadherin target likely presented the size of a protein domain (35 KDa, Figure 5H). The antibody used was directed against the intracellular domain of this protein.

In agreement with immunocytochemistry results, vimentin and α-actin presented distinct distribution in sample homogenates. Indeed, the anti-vimentin targeted the same proteins in all bovine intestinal cell cultures and in the control 3T3 fibroblasts (Figure 5I, J), while α-actin was scarcely detected in more advanced passages (Figure 5E, F). Taken together, western blot and immunocytochemistry results strongly suggested that vimentin is expressed in primary cultures of epithelial intestinal cells from bovine jejunum and colon. However, vimentin was not detected by western-blot in Caco-2 cell proteins (Figure 5I, J). Although immunocytochemistry results for α-actin indicated a very low contamination of bovine epithelial cell cultures by mesenchymal cells, western blot results indicated that this phenomenon is to low to be detected in most culture stages.

RT-PCR analyses allowed us to investigate the expression of bovine enterocyte markers namely: villin (actin-caping protein in microvilli), zonula occludens (ZO1, associated with the cytoplasmic surface of tight junctions), fatty acid binding protein (FABP), small intestine peptidase (IP) and E-Cadherin. Expressions studies were also used to confirm the presence of the vimentin transcript in bovine cells. Similarly to western blot analyses, gene expressions studies were carried out on total RNA extracted from bovine samples of: (1) freshly scrapped epithelia, (2) organoid suspensions used to seed cultures and (3) intestinal cell cultures of both types. As can be seen in figure 6 (A–G), cDNA amplification products presented the expected size (see methods) using specific primers for coding regions of the epithelial markers, such as the villin gene (Figure 6C), the ZO1 gene (Figure 6D) and the E-cadherin (Figure 6G), that were uniformly expressed in cells from primary cultures of bovine enterocytes. Besides these structural components expected in any epithelium, we also investigated the expression of functional markers of digestive epithelia. Among them, the gene coding for the fatty acid binding protein (FABP, Figure 6B), presented a constant expression in both jejunocytes and colonocytes in vitro. The second tissue specific marker studied was the gene coding for the intestinal peptidase (IP, Figure 6E). In this case, the expression was restricted to jejunocyte samples, as logically expected on the basis of the functional specialization of the jejunum (nutrient processing and absorption). The IP expression could then be considered as suitable jejunocyte marker. In accordance with immunostaining results, samples from jejunocyte and colonocyte cultures expressed the vimentin gene (Figure 6F). Interestingly, while western blots analyses failed to reveal vimentin either in freshly removed epithelia or in organoids suspensions used to seed cultures, the transcript of vimentin gene was clearly detected in those samples, indicating that the vimentin is constitutively expressed in jejunum, as well as in colon from the bovine intestine. However, western blot results suggested that vimentin synthesis seemed to be restricted in cultured cells with a progressive increase over the culture passage number (Figure 5I, J).

The chosen enzymes were a disaccharidase (maltase) and the alkaline phosphatase. As for gene expression studies, enzyme activities were investigated in bovine samples of: (1) freshly scrapped epithelia, (2) organoid suspensions used to seed cultures and (3) intestinal cell cultures of both types. Assays performed on fresh epithelial tissue confirmed that jejunum and colon differed in respect to enzyme specific activities (SA), the jejunum presenting the highest level. Maltase SA (i.e. differentiation marker) clearly decreased in primary cultures of both cells types (Figure 7A).

Furthermore, the cell isolation procedure selected disaccharidase depleted material, since organoid suspensions presented a 50% reduction in regard to the fresh epithelium preparation. It is interesting to note that, compared to the organoids suspension, jejunocytes from the first culture passage did not present a drastic reduction of the maltase SA. However, with subsequent culture passages the decreasing of disaccharidase SA was progressing toward a stable low level. Similar results were obtained from measurements of the intestinal alkaline phosphatase SA (IAP, Figure 7B), a ubiquitous enzyme that is a marker of brush border in intestinal epithelial cultures. Indeed, as expected from functional differences between jejunum and colon in vivo, epithelia homogenates, organoids suspensions, and cultures at first passages from bovine jejunum presented higher SA for this enzyme. Moreover, similarly to what was noted for maltase SA, IAP presented a strong decrease from fresh tissue to subcultured cells, so that jejunocyte and colonocyte SA values joined together to the same low level as the culture passage number increased.

Regarding a given passage number, the maltase SA decreased over the culture duration (Figure 8A). These results reflected a loss of cell differentiation in vitro compared to the in vivo level. Adjustments of cell culture medium were made to improve the differentiation status of jejunocytes and colonocytes in vitro. To this end the glucose was substituted by inosin in the culture medium; this condition was described to stimulate the acquisition of the enterocyte differentiated phenotype in vitro 29. As seen in figure 8B, the use of a inosin-containing glucose-free culture medium during 7 days, led to an increase of maltase activity. However this effect diminished as the passage number increased, suggesting that glucose substitution by inosin should be complemented by other culture medium modifications. Addition of sodium butyrate (1–2 mM), a substance thought to promote enterocyte differentiation through a stimulation of the CDX2 homeobox gene expression 30, proved to be efficient to stimulate maltase SA (Figure 8C).

Intestinal fragments of proximal region of jejunum and of spiralled colon from adult animals were obtained from a local slaughterhouse. Tissue removals were made in agreement with relevant local animal welfare laws, guidelines and policies. After several washes in warm (37°C) divalent ion free PBS (PBS: 13.7 mM NaCl, 0.27 mM KCl, 0.43 mM Na₂HPO₄, 0.14 mM KH₂PO₄, pH 7.4), fragments were washed in warm PBS supplemented with 1% Antibiotic-Antimycotic solution (Gibco/BRL), 2.7 mg/ml D-Glucose (Sigma), and 4 mM L-Glutamine (Gibco/BRL). Animal organs were then immersed in the same warm supplemented PBS and transported within 30 minutes to the laboratory. All following steps were performed under a laminar flux hood (Heraeus Instruments). After several washes, fragments of intestine were filled with PBS containing 1 mM 1.4-dithiothreitol (ICN Biomedicals Inc.) and, after closing their extremities, incubated for 5 minutes in a shaking bath at 37°C, to get the epithelium surface rid of mucous before performing the cell isolation procedure. Then this incubation medium was replaced by the digesting solution consisting in supplemented PBS added of collagenase (Sigma; 300 U/ml) and dispase (Gibco/BRL; 0.1 mg/ml in PBS), for 15 minutes (for jejunum) or 20 minutes (for colon) at 37°C in a shaking bath. Using again this medium, a second digestion step lasting 45 minutes (for jejunum) or 60 minutes (for colon) was carried out in the same conditions. The lumen content (bearing mostly dead cells, as demonstrated using a viability test described below) was discarded after enzymatic incubations. Then, each intestinal segment was longitudinally wide opened and the pre-digested epithelium was scraped from the digestive mucosa using a sterile scalpel blade. The resulting material was incubated in PBS containing 1 mg/ml dispase for 10 minutes whilst active pipetting movements were done to help the dissociation of epithelium fragments. Cells and organoids (i.e. cell aggregates) were pelleted by centrifugation at 140 × g for 3 minutes.

As the large amount of single cells present in the pellet is likely to include contaminant lymphocytes (initially located in basolateral spaces between epithelial cells) and fibroblasts (released from the conjunctive subepithelial layer), the following two methods were used to impoverish the preparation in those cells: the pellet was suspended in 30 ml of Dulbecco's modified Eagle's (D-MEM) medium containing 2% of sorbitol (Sigma) and centrifuged at 50 × g for 3 minutes (isopicnic centrifugation). Under these conditions, a large part of the single cells remained on the top of the sorbitol density cushion whereas organoids accumulated at the bottom of the tube. The supernatant was then discarded and the pellet was washed in sorbitol containing D-MEM and centrifuged at 50 × g for 3 minutes. The pellet wash was repeated for about 5 times, to obtain a clear supernatant. The final pellet content, not completely devoid of single cells, was then allowed to settle down on a cell culture surface, under conditions favouring the adhesion of possible residual fibroblasts (selective attachment): the medium used to suspend the pellet was D-MEM supplemented with a high amount of FBS (10%; Gibco/BRL) and the suspension was incubated for 1 hour at 37°C in a 175 cm² culture flask (Greiner Bio-One) devoid of any coating permitting the attachment of epithelial cells. Unattached material was then removed from the flask, centrifuged at 140 × g for 3 minutes and finally the pellet suspended in the medium used to culture epithelial cells in collagen coated flasks (see below).

The culture medium used was high glucose D-MEM (Gibco/BRL) supplemented with 100 nM hydrocortisone (Sigma), 20 nM triiodothyronine (Sigma), 1 ng/ml Epidermal Growth Factor (Sigma), 1 μg/ml insulin (Actrapid, Novo Nordisk A/S, Denmark), 10 μg/ml Acid linoleic/Albumin (Sigma), 1% Glutamax (Gibco/BRL) and 1% Non Essential Amino-Acids (NEAA, Gibco/BRL) 49, with 1% Antibiotic-Antimycotic solution and 2% FBS (Hyclone Perbio Sciences). The organoids were seeded at a high density, in collagen I (Roche Diagnostics; 17 μg/cm²) coated culture flasks (to permit epithelial cell adhesion), and the first medium change was done after 20 hours. The cultures were maintained at 37°C in a humidified incubator (Heraeus Instruments) in a 5% CO₂ atmosphere. Medium was changed every 2 or 3 days. Confluent cells were subcultured at a split ratio of 1:2. To this end, cells were detached by incubation for 3 to 5 minutes, at 37°C, with 40 μl/cm² trypsin/EDTA (Gibco/BRL). Cells were harvested in a 10 fold greater volume of supplemented culture medium and enzyme was washed away by centrifugation at 140 × g for 3 minutes. The pellet was suspended in the supplemented culture medium and the cell suspension redistributed in collagen coated culture flasks. Samples from both cell culture types (colonocytes or jejunocytes) were frozen and stored in liquid nitrogen, at 1.5 10⁶ cells per vial, in 1 ml of D-MEM containing 10 % DMSO (Merck) and 10 % FBS (Hyclone Perbio Sciences). We checked that frozen cells were suitable to reconstitute a culture.

Viability of freshly isolated material was tested with a mixture of 10 mg/ml ethidium bromide (Sigma) and 5 mg/ml acridin orange (Sigma) in PBS. The mixture was added 1/1000 to the cell suspension and immediately observed using a fluorescence microscope (Nikon Eclipse TE 200). Under these conditions dead cells nuclei were stained in red while nuclei of living cells corresponded to green spots.

Several antibodies were used to characterize the primary cell cultures, namely: monoclonal anti-pan-cytokeratin (mixture of clones C-11, PCK-26, CY-90, KS1A3, M20, A53 and B/A2, Sigma), monoclonal anti-E-cadherin (Becton Dickinson Biosciences), monoclonal anti-α smooth muscle actin (clone 1A4, Sigma), monoclonal anti-vimentin (clone V9, Sigma), monoclonal anti-epithelium specific antigen (ESA, Sigma) and monoclonal anti-cytokeratin peptide 18 (clone KS-BA2, Sigma). Prior to immunostaining, cells were fixed with 5% paraformaldehyde into PBS for 5 minutes at room temperature. Cells were then permeabilized by incubation for 15 minutes with 0.2% Triton X-100 (Roche) dissolved in PBS containing 5% goat serum (Gibco/BRL) to block non-specific binding. After 3 washes in PBS, for 5 minutes each, cells were incubated for 60 minutes at room temperature with the primary antibody in the blocking solution. Dilutions used were those recommended by the manufacturer. Cells were washed three times, for 5 minute each, in PBS and then incubated with the secondary antibody for 30 minutes at room temperature in darkness. The secondary antibody was anti-mouse IgG conjugated to FITC (Sigma) or TRITC to detect the antigen-antibody complex. After another three washes in PBS, cells were observed with a fluorescent microscope (Nikon Eclipse TE 200). In a few cases, direct immunofluorescence was performed using an anti-pan-cytokeratin (clone C11) FITC conjugate (Sigma) and anti-vimentin (clone V9) Cy3 conjugate (Sigma). For epithelia specific staining, Caco-2 cells (human carcinoma colonic cells, received from Prof. YJ Schneider, Catholic University of Louvain, Belgium) were used as positive controls. For all other immunodetections, 3T3 fibroblasts (a generous gift from Prof. E. Heinen, Histology Laboratory, University of Liege, Belgium) and bovine cell samples submitted to the staining procedure, but omitting the incubation with primary antibody, were used as negative controls.

Samples of fresh mucosa and cell suspension used for starting the cultures were prepared by washing the cells twice into PBS and suspending the pellet in ultrapure water (1 ml). Samples of confluent cultured cells were prepared by discarding the culture medium, rinsing the cell monolayer twice with PBS and then cell scraping in ultrapure water (300 μl ultrapure water for a 25 cm² culture flask). Samples were sonicated (30 sec on ice, Sonic Power Company Cell Disrupter) to accomplish cell lysis. Twenty μg of proteins from each sample were separated by electrophoresis on 10% SDS-PAGE 50 and transferred to Bio Trade PVDF Transfer Membrane (0.45 μm, Pall Corporation, Life Sciences). After 1 hour incubation at 37°C in the blocking solution (PBS containing 0.2% of Tween 20 and 5% of milk powder; Nestle), membranes were incubated for 1 hour at 37°C (or overnight at 4°C) with primary antibody (see immunocytochemistry section) diluted in the blocking solution according the manufacturer's instruction. Three washes in the blocking solution were followed by the secondary antibody incubation. The secondary antibody was a peroxydase conjugate (Antimouse IgG peroxydase conjugate; Sigma) diluted at 1:1000 in the blocking solution. The blots were developed using the peroxydase substrate 3,3-diaminobenzidine (DAB; Sigma) containing 0.03% H₂O₂ (Merck). Cultured Caco-2 cells and 3T3 fibroblasts, submitted to the same treatment, were respectively used as positive and negative controls for epithelial immunostaining.

Gene expression analyses of cell samples were realized using the combined mRNA reverse transcription-polymerase chain reaction (RT-PCR). Total RNA was extracted using the guadinium thiocyanate protocol 51. Prior to reverse transcription RNA was treated with DNAse (DNAse RQ1; Promega) in the appropriate DNAse buffer (RQ1 Buffer; Promega) for 30 minutes at 37°C to break up the possible contaminant DNA. Reverse transcription of mRNA was performed from 2 μg of total RNA, in presence of RNAse inhibitor (RNAguard 40 U/μl; Promega) using oligo-dT primers (Oligo dT-15 Primer, Promega), deoxynucleotides (dNTP 10 mM, Promega), Mo-MuLV reverse transcriptase (200 U/μl, Promega) and the reverse transcriptase buffer (Promega) in a 150 μl final volume. Reverse transcription was performed at 42°C for 1 h. Primers used for the specific amplification of cDNA fragments corresponding to coding segments of the following proteins were: villin forward 5'-ACC-TTC-ACA-GGC-TGG-TTC-CT-3' and reverse 5'-GGT-TTT-GTT-GCT-TCC-AT-3' (amplification product size: 384 bp); intestinal peptidase (IP): forward 5'-GCG-ATT-ATG-CCC-CTT-TCA-TT-3' and reverse 5'-CAG-CCT-GCA-GGA-AGC-T-3' (amplification product size: 276 bp); Fatty Acids Binding Protein (FABP) forward: 5'-TTC-AGC-AGC-TGG-TAG-GAA-3' and reverse 5'-TAA-CCA-AAG-AGA-TGA-CCC-TA-3' (amplification product size: 276 bp); Zonula Occludens 1 (ZO1) forward: 5'-GCG CTG AAA GAA GCA ATT CA-3' and reverse: 5'-AAA CAT GGT TCT GCC TCA TC-3' (amplification product size: 272 bp); vimentin: forward 5'-CCG-GAG-CTA-CGT-GAC-CAC-AT-3' and reverse 5'-CTC-GGC-TTC-CTC-TCT-CTG-AA-3' (amplification product size: 540 bp); E-cadherin: forward 5'-CGC-ACA-ACA-AAA-TGT-TCA-CC-3' and reverse 5'-CAT-TGG-TGA-CTG-GGT-CTG-TG-3' (amplification product size: 308 bp). RNA extraction and RT-PCR success were checked through the co-detection of the constantly expressed gene β-actin, using the primer pair forward: 5'-AGA-AAA-TCT-GGC-ACC-ACA-CC-3' and reverse: 5'-GTC-AGG-CAG-CTC-GTA-GCT-CT-3' (amplification product size 485 bp). PCR comprised 35 cycles, each cycle consisting in the succession of a denaturing step at 95°C for 1 minute, a primer annealing step (at 50 to 55°C, depending on the primer pair) for 1 minute and a primer elongation step 72°C for 1 minute. They were preceded by a first denaturing step for 3 minutes at 95°C and followed by a final elongation step at 72°C for 5 minutes. The PCR reaction contained 1 to 2 μl of cDNA, 1 μmol/L of each primer (Eurogentec, Belgium), 1.5 mmol/L of MgCl₂ (Promega), 200 μmol/L of deoxynucleotide triphosphates (dNTP 10 mM, Promega) and 0.5 units of Taq polymerase (Promega) in the appropriate buffer (Promega). Control reactions omitting the cDNA were included and yielded negative response.

Amplification products were visualized after electrophoresis on 1% agarose (Sigma) gels in Tris Borate EDTA buffer pH 8 (TBE; 89 mM Tris base, Sigma; 89 mM boric acid, Merck; 2 mM Na EDTA, Merck) stained with ethidium bromide (1‰ w/v, Sigma). Samples were dissolved in electrophoresis loading buffer (Promega). Simultaneous migration of DNA molecular weight standards (1 Kb ladder, Promega) allowed the size determination (in bp) of amplification products.

Samples of fresh mucosa, cell suspension used to start the cultures and epithelial cell cultures were prepared as described for western blot analysis. Maltase activity (brush border disaccharidase) was measured according to the spectrophotometric method developed by Dahlqvist 52. Alkaline phosphatase was estimated using 4-nitrophenyl phosphate as substrate (1.5 mM; Acros Organics) in ethyldiethanolamine buffer (150 mM; Sigma) and dinitrophenolphosphate (Sigma) as standard. After a 20 minute incubation time, the reaction was stopped by a 1 M NaOH solution. The absorbance of the enzyme reaction product was measured at 415 nm. Enzymatic activities were reported to the protein content measured by the method of Bradford 53 and expressed as specific enzymatic activities (SA, μmol/min.g).

For transmission electron microscopy, cells were grown in collagen I coated culture wells. Cultured cells were washed twice with PBS and then fixed in 2.5% glutaraldehyde solution at 4°C for 1 hour. Two other PBS washes were followed by a post-fixation in 2% osmium tetroxyde for 1.5 h. Then samples were dehydrated in serial ethanol dilutions, scrapped from their plastic support and embedded in epon-epoxy resin. Ultrathin sections were counterstained with uranyl acetate were fixed on grids and observed under a transmission electron microscope (JEOL JEM-100CX) at 80 kV.

Results are reported as mean ± standard deviation (SD). With regard to heteroscedasticity, statistical analysis was performed using Newman-Student-Keuls or Kruskal-Wallis ANOVA. A p-value below 0.05 was considered as statistically significant.