Expression of the cystic fibrosis gene and the major pancreatic mucin gene, MUC1, in human ductal epithelial cells

The major feature of the inherited disorder cystic fibrosis (CF) is the build-up of mucous secretions in the respiratory, digestive and reproductive systems. Eighty-five per cent of patients with CF are pancreatic-insufficient. This is due to obstruction of pancreatic ducts preventing normal release of digestive enzymes into the duodenum. Gradual autolysis of the pancreas results in acinar tissue being replaced by cystic spaces. There has been controversy as to whether the substances causing the blockage of the ducts are inspissated mucous glycoproteins, or other proteins that have precipitated due to the abnormal ionic environment within the CF pancreatic duct.

Mucous secretions have long been an area of investigation in CF, but results have often been inconclusive. However, following the cloning of the genes for several mucins, including the complete cDNA for the major pancreatic and mammary gland mucin MUC1 (Lan et al., 1990; Gendler et al., 1990), partial cDNAs for the intestinal mucins MUC2 and MUC3 (Gum et al., 1989, 1990), and the tracheobronchial mucin MUC4 (Porchet et al., 1991), it is now possible to start investigating the involvement of mucins in CF at a molecular level.

The cystic fibrosis transmembrane conductance regulator (CFTR) protein encoded by the CF gene (Rommens et al., 1989; Riordan et al., 1989) is known to be a specific type of small chloride ion channel (Gray et al., 1988, 1989, 1990; Bear et al., 1992). However, due to the substantial structural homology observed between CFTR and members of the ABC transporters (Hyde et al., 1990), it remains possible that CFTR is a bifunctional protein that, in addition to acting as an ion channel, is an ATP-dependent transporter of an as yet unknown substrate.

How defects in a small chloride ion channel could cause the effects on mucus hydration and clearance remains unclear, though it is possible that lack of water movement accompanying Cl⁻ secretion might have a role. There has been some speculation that the CFTR protein might be directly involved in mucus secretion at the apical membrane of the ductal epithelial cells in which it is expressed, hence disease-associated mutations might directly affect mucus release. However, this would appear unlikely on the basis of what is now known about the mechanism of MUC1 secretion (Ligtenberg et al., 1992). Alternatively, the abnormalities in chloride and bicarbonate secretion from ductal epithelia might sufficiently alter the ionic environment in the duct to produce a ‘salt stress’ response in the epithelial cells, causing increased mucus secretion. This phenomenon has been observed in a bladder carcinoma epithelial cell line in response to lactate and chloride stress (Bader and Harris, 1987). Another possibility to account for the relative viscosity of mucous secretions in CF is abnormal glycosylation of the MUC1 glycoprotein.

We have been investigating the interaction of mucins and the CFTR in primary fetal pancreatic and male genital ducts, where both genes have been shown to be expressed at high levels in vivo (Harris and Coleman, 1988; Coleman and Harris, 1991; Harris et al., 1991). We have subjected primary epithelial cells to salt stress to examine the effect on MUC1 secretion. The investigation of potential interactions between mucins and CFTR is now being extended to adult pancreatic ducts that also express the CFTR and MUC1 genes (Crawford et al., 1991; Marino et al., 1991; Lan et al., 1990). The lack of normal adult pancreatic duct culture systems necessitated an examination of pancreatic adenocarcinoma cell lines. We have examined expression of both genes in eleven pancreatic adenocarcinoma cell lines and further investigated the expression of CFTR in relation to culture confluence.

Pancreatic adenocarcinoma cell lines BxPC3 (Tan et al., 1986) and Capan 1 (Fogh et al., 1977) were purchased from the American Type Culture Collection and grown in RPMI 1640 or Dulbecco’s modified Eagle’s medium (DMEM). The CF-PAC1 cell line (Schoumacher et al., 1990), a pancreatic adenocarcinoma cell line derived from a CF patient who is homozygous for the F508 deletion, was donated by Dr R. A. Frizzell (Birmingham, Alabama, USA). Lines HPAF (Kim et al., 1989), HS766T (Owens et al., 1976), ASPC1 (Tan et al., 1981), Capan 2 (Wright et al., 1981), Colo357 (G. Moore, University of Colorado, Denver, CO, USA), Panc1 (Lieber et al., 1975) Panc89 and MIAPaCa (Yunis et al., 1977) were grown by Dr M. A. Hollingsworth at Duke University, North Carolina, RNA was extracted post-confluency and RNA samples were shipped on dry ice. The colon carcinoma cell line HT29 (Huet et al., 1987) and the breast cancer cell line MCF7 (Soule et al., 1973) were grown in DMEM.

Total cellular RNA was prepared by the method of Chirgwin et al. (1979). Reverse transcriptase-PCR (RT-PCR) was carried out in a modified version of that described previously (Harris et al., 1991). Briefly, for each reaction, cDNA was synthesized using the 3′ primer of the CFTR or MUC1 gene fragment, together with 1/10th the concentration (for co-amplification of CFTR cDNA) or an equal concentration (for co-amplification of MUC1 cDNA) of a 3′ primer homologous to the housekeeping gene glucocere-brosidase in the same tube. The 3′ primers were annealed to 1 μg of total RNA from the cell line of interest at 65°C for 10 min, and used to prime cDNA synthesis from each gene at 42°C for 60 min with reverse transcriptase (BRL). The PCR reaction (Saiki et al., 1988) for CFTR was 94°C for 5 min, then 30 cycles of 93°C for 1 min, 60°C for 1 min and 72°C for 5 min, with a final elongation step of 72°C for 5 min. For MUC1 the PCR conditions were the same except for annealing at 62°C. The following gene-specific primers were used: (1) for CFTR, primers E1R 3055 5′AGATTCTCCAAAGATATAGC3′ 3074 and E1L 3824 5′GAAATGTTGTCTAATATGGC3′ 3805, generating a 769 bp fragment (Riordan et al., 1989; Chalkley and Harris, 1991); (2) for MUC1, primers MUC1A 690 5′AGGCTCAGCTTC-TACTCTGG3′ 710 and MUC1B 1346 5′GACAGACAGC-CAAGGCAATG3′ 1326, producing a 656 bp fragment (Lan et al., 1990); (3) for glucocerebrosidase, GD67A 837 5′CAGAT-ACTTTGTGAAGTTCC3′ 856 and GDMID9B 1409 5′GACT-GTCGACAAAGTTACGC3′ 1390 producing a 572 bp band (Sorge et al., 1985). The cDNA fragments were resolved on 1.5% to 2% agarose gels. The specific fragments generated by each PCR reaction, i.e. for CFTR, MUC1 and B glucocerebrosidase, were verified by restriction enzyme digestion. All primer sets were shown not to amplify a product from genomic DNA.

Primary cultures of vas deferens epithelial cells (Harris and Coleman, 1989) were analysed at passage 3 on 6-well Nunc cell culture dishes in DMEM, with equal cell numbers seeded in each well. When cultures had just reached confluence, the culture medium was aspirated and replaced with 1 ml fresh medium per dish. After 24 h, this 1 ml was collected and replaced by 1 ml of fresh medium supplemented with 0 M, 0.05 M or 0.1 M NaCl. These concentrations were calculated on the basis of survival curves for primary vas deferens epithelial cells cultured in supplemented salt concentrations of 0.01 M to 0.5 M (not shown). This media replacement was repeated at 24 h intervals up to 72 h. After the final collection of medium, the cells were harvested and the total protein per well assayed using the Bio-Rad DC protein assay kit.

Amounts of MUC1 secreted into the medium were assayed by standard ELISA techniques, using the anti-MUC1 antibody Ca2 (Bramwell et al., 1985; Harris, 1987), donated by Dr M. Bramwell. This antibody was raised against the native Ca antigen (MUC1 protein) purified from human urine and is known to detect MUC1 glycoprotein secreted into the tissue culture medium from a variety of cells. Total protein in the culture media at each collection point was assayed by the Bio-Rad DC protein assay kit, measuring absorbance at 710 nm.

The pancreatic adenocarcinoma cell line Capan 1 and the intestinal epithelial cell line HT29 were plated out at densities of 1×10⁶ and 2.6 ×10⁶ cells, respectively, per 90 mm cell culture dish.

Cells were harvested at 48 h intervals commencing on day 6 after passaging for Capan1, and day 3 after passaging for HT29. RNA was prepared by the method of Chirgwin et al. (1979) and RT-PCR for CFTR and MUC1 was carried out as described above.

Detection of the CFTR and MUC1 transcripts in nine pancreatic adenocarcinoma cell lines is shown in Fig. 1. Each lane shows the RT-PCR product from 1 μg of total RNA using primers specific for CFTR (A) or for MUC1 (B). For the CFTR and MUC1 reactions the primers amplify a cDNA that spans multiple exons and has been shown not to amplify a product from genomic DNA. In both (A) and (B) the larger RT-PCR product (769 bp for CFTR and 656 bp for MUC1) is specific for the gene being assayed, and the smaller product (572 bp) represents the glucocerebrosidase control housekeeping transcript. The amount of glucocerebrosidase transcript in one cell type is relatively constant. In addition to the pancreatic adenocarcinoma cell lines, lanes 9 and 11 contain RNA from two control cell lines. The former, MCF7, is a mammary gland epithelial cell line (Soule et al., 1973), which we have shown to express MUC1 but not CFTR (A. Harris, unpublished result). The latter is the colon carcinoma cell line HT29 (Huet et al., 1987), which is known to express both CFTR, in relative abundance, and MUC1. (B) shows that all the pancreatic adenocarcinoma cell lines asssayed express at least some MUC1, though the amount varies greatly as has been observed previously by northern analysis (Lan et al., 1990). However, only one line, Capan 1 (A, lane 6), expresses CFTR at a significant level. In (A) no CFTR expression was detected in Panc89, Capan 2, HS766T, MIAPaCa, Colo357, HPAF, Aspc3 or Panc1. Further, no CFTR expression was detected in the BxPC3 or CF-PAC1 cell lines (not shown).

The vas deferens epithelium appears to have many physiological and biochemical similarities to the pancreatic duct. Most importantly, in this context the vas deferens is one of the sites primarily affected by the disease process in CF. Also, the vas deferens epithelial cells express CFTR and MUC1 at a high level (Harris and Coleman, 1989; Coleman and Harris, 1991). As salt stress has been shown to upregulate MUC1 secretion in a bladder carcinoma epithelial cell line (Bader and Harris, 1987), we investigated whether salt stress had the same effect on vas deferens epithelial cells in culture. Levels of MUC1 glycoprotein released into the tissue culture medium after exposure of cells to 0.05 M or 0.1 M added NaCl were recorded and expressed as an absolute figure relative to the total amount of protein in the medium (Fig. 2). No significant effect of salt stress on vas deferens epithelial cell MUC1 secretion was observed. However, it was noted that after reaching a peak at confluence, the total protein in the cell culture wells, as assayed on harvesting at the end of the time course, was substantially lower in the salt-treated cells than in the controls, suggesting that the salt exposure was slightly toxic to the cells even though they were not killed by the salt concentrations employed. (MUC1 measurements were also normalized to account for the total cellular protein in each culture well, again revealing no effect of salt stress on MUC1 secretion, data not shown).

The level of expression of CFTR in the Capan 1 cell line with increasing degree of confluence is shown in Fig. 3. In (A) variable expression of CFTR is shown by the amounts of the upper band (769 bp) relative to constant amounts of the glucocerebrosidase transcript (572 bp) in 1 μg of total RNA. A total of 1×10⁶ cells were seeded on 90 mm dishes and harvested at 6, 8, 10, 12 and 14 days after passaging. The amounts of CFTR transcript increases greatly with time in culture, reaching a plateau at confluence (10 days after passage; lane 3). In fact, after this time the levels of CFTR transcript in the cultures appear to fall slightly.

To investigate whether this phenomenon is specific to the Capan 1 cell line we carried out an identical experiment on the HT29 colon carcinoma cell line, which is known to express CFTR at a relatively high level. This time, 2.6×10⁶ cells were seeded onto 90 mm tissue culture plates and harvested after 3, 5, 7, 9 and 11 days. Again levels of CFTR expression relative to expression of the housekeeping gene glucocerebrosidase are shown (Fig. 3B). The abundance of the CFTR transcript is seen to rise with increasing cell density and to plateau at confluence (day 7). The level of MUC1 mRNA remains essentially constant relative to the amounts of glucocerebrosidase over the same time period in both Capan 1 and HT29 cell lines (Fig. 3C).

The vast majority of pancreatic adenocarcinomas are ductal in origin. The cystic fibrosis gene has been shown to be expressed at a high level in the pancreatic ducts, both at the levels of RNA and protein (Harris et al., 1991; Foulkes and Harris, 1993; Marino et al., 1991; Crawford et al., 1991). Hence, it might be expected that at least the more highly differentiated pancreatic adenocarcinoma cell lines would maintain expression of CFTR in culture. However, we have found that the majority of pancreatic adenocarcinoma cell lines examined do not express CFTR in vitro, irrespective of their state of differentiation. These include Panc89, Capan 2, HS766T, MIAPaCa, Colo357, HPAF, Aspc3 and Panc1. HPAF is a well differentiated adenocarcinoma while Panc1 and Hs766T are poorly differentiated. Only one adenocarcinoma cell line, Capan 1, from a well differentiated adenocarcinoma, was shown to express CFTR mRNA at a significant level. The level of CFTR mRNA in this line (Fig. 1, lane 6) is still substantially lower than that seen in the intestinal epithelial cell line HT29 (Fig. 1, lane 11).

We were not able to detect CFTR mRNA in the CF-PAC1 cell line (Schoumaker et al., 1990), a pancreatic adenocarcinoma cell line from a CF patient who was homozygous for the F508 deletion (not shown). This cell line has been used extensively in CF ‘phenotype correction’ experiments where CFTR cDNA is transfected into the CFPAC cell line (Drumm et al., 1990). It would appear that rather than correcting the F508 mutation, these experiments may simply have been restoring a function lost due to switching off of CFTR transcription. It is of course possible that at early passage numbers in culture many of the pancreatic adenocarcinoma cell lines did express CFTR, but that with time CFTR transcription has gradually been down-regulated. Many attempts at in vitro immortalization of various ductal epithelial cell lines have been accompanied by more or less rapid down-regulation of CFTR transcription, irrespective of the immortalizing agent. Since CFTR is a specific type of small conductance chloride ion channel located in the apical membrane of specialized ductal epithelia (Gray et al., 1988, 1989, 1990; Bear et al., 1992), the fact that many ductal epithelial cell lines immortalized in vitro, exhibit a disorganized pattern of growth, with loss of the characteristic polarized monolayer, may be relevant.

The observation that salt stress (150 mM sodium lactate) could cause an increase in the amount of MUC1 secretion from the human bladder carcinoma cell line RT112 (Bader and Harris, 1987) suggested a possible basis for mucin overproduction in CF ductal epithelial cells. Lack of fluid secretion from ductal epithelia, resulting from defective chloride efflux, and in the pancreatic ducts, bicarbonate efflux, could effectively raise the molarity of ductal secretions. This could in turn reproduce the salt stress phenomenon observed in vitro in the RT112 cell line. Hence, it was appropriate to investigate the salt stress phenomenon in vas deferens epithelial cells, which secrete MUC1 and contain high levels of CFTR. However, with 0.05 M and 0.1 M sodium chloride (the maximimum concentration that cells could tolerate) added to the cell culture medium, no effect on MUC1 secretion by cultures of primary vas deferens epithelium cells was observed. This would suggest that rather than being a general phenomenon in ductal epithelial cells, hypersecretion of MUC1 in response to salt stress may be restricted to the bladder epithelium, which has to cope with widely fluctuating and extreme extracellular salt concentrations in vivo. It remains possible that CF epithelial cells may respond aberrantly to salt stress.

Initial experiments that had shown some fluctuations in the levels of expression of CFTR in the Capan 1 cell line led us to examine the effect of confluence on expression of this gene. The levels of CFTR in the Capan 1 cell line were seen to rise relative to the amounts of mRNA for the housekeeping gene glucocerebrosidase over a period, until the cells reached confluence at 8 days post-seeding (Fig. 3A, lane 3), at which time they reached a plateau. In fact, post-confluence the levels of the CFTR transcript in the cultures appears to fall slightly, and this correlates with the increasing tendency of the Capan 1 cell line to slough off cells from the monolayer into the culture medium post-confluence.

Parallel experiments were also carried out on the intestinal epithelial cell line HT29 to see if this was a general phenomenon in differentiated ductal epithelial cell lines expressing CFTR. Fig. 3B shows increasing levels of CFTR relative to the constant glucocerebrosidase transcript until the culture reaches confluence in HT29 cells. This pattern of regulation of CFTR in intestinal epithelial cell lines has recently been reported elsewhere (Sood et al., 1992). The correlation of CFTR expression with cell confluence and differentiation in culture thus appears to be a phenomenon seen in intestinal as well as pancreatic duct epithelial cell lines. The result is of importance for the analysis of normal CFTR function in epithelial cells and for the investigation of the effects of mutations in the CF gene on the function of CFTR in physiological systems.

Large amounts of MUC1 glycoprotein are known to accumulate after confluence in long-term cell lines, a phenomenon that we have also observed in the primary vas deferens epithelial cell lines. However, the levels of expression of MUC1 mRNA in Capan 1 and HT29, as assayed by RT-PCR do not appear to increase with confluence over the time period during which CFTR expression increases markedly, as is shown in Fig. 3.

In conclusion, we have identified a pancreatic adenocarcinoma cell line, Capan1, that expresses CFTR at a significant level and also produces MUC1. This identifies a suitable model, at confluence, for investigating the role of mutations in the CF gene in causing pancreatic duct obstuction by mucins in CF.

The failure to induce the phenomenon of mucus hyper-secretion, by salt stress, in primary vas deferens epithelial cells suggests that a more complex mechanism is responsible for the build-up of mucus secretions in CF. This may be a direct effect of defective chloride secretion from epithelial cells. However, it is also possible that another as yet unidentified function of CFTR may be of importance.

The authors would like to thank Drs M. Bramwell, M. Hollingsworth and D. Swallow for the CA2 antibody, RNA samples and helpful discussions, also Dr R. Frizzell for the CF-PAC cell line. This work was supported by the Cystic Fibrosis Research Trust.