Isolation and characterisation of clones from the Caco-2 cell line displaying increased taurocholic acid transport

Human adenocarcinoma epithelial cell lines are being used to investigate a variety of intestinal functions. For example, cellular differentiation of intestinal epithelial cells has been extensively studied (Pinto et al. 1983; Rousset, 1986; Neutra and Louvard, 1989; Zweibaum et al. 1990). Recently, it was proposed that the Caco-2 cell line, derived from a human colon carcinoma (Fogh et al. 1977), represented a propitious model for studying intestinal transport processes and permeability (Hilgers and Burton, 1988; Cogburn et al. 1989; Hidalgo et al. 1989; Artursson, 1990; Wilson et al. 1990). When Caco-2 cells are grown on permeable supports, in chambers, they form a confluent monolayer with several properties characteristic of differentiated absorptive epithelial cells in the distal ileum. The monolayers are morphologically polar and have well developed brush borders at their apical surface. Physiologically, the Caco-2 model has been characterised with respect to carrier-mediated transport of taurocholic acid and amino acids (Hidalgo et al. 1989; Wilson et al. 1990), receptor-mediated transport of cobalamin (Dix et al. 1990) and shown to display low permeability to the passive transepithelial passage of macromolecules (Wilson et al. 1990). The long-term viability and reproducibility of this intestinal epithelial cell model, together with the ability to perform quantitative and qualitative transport studies, confer significant advantages over other in vitro models (see Osiecka et al. 1985) used to study intestinal function.

The Caco-2 cell line is a unique in vitro model with respect to the polar transport of taurocholic acid. However, the amounts transported are relatively modest and, since the Caco-2 cell line is not homogeneous with respect to certain morphological parameters (Woodcock et al. 1989), we have explored the possibility that it is functionally heterogeneous and attempted to select clones with increased transport rates. In this paper, a clone is described that transports significantly increased amounts of taurocholic acid. In addition, we have characterised it with respect to other parameters associated with the parental population; namely, growth characteristics, transmembranal electrical resistance (TER) and morphology.

The Caco-2 cell line (Fogh et al. 1977) was obtained from Professor Colin Hopkins, Imperial College, University of London. Cells of passage numbers 80–100 for the parental population, and 1–20 for the clones, were used. Caco-2 cells were routinely grown in medium (maintenance medium) composed of Dulbecco’s modified Eagle’s medium (DMEM), containing 10% (v/v) foetal calf serum (FCS), 1% (v/v) non-essential amino acids, and 0.03% (w/v) glutamine, in an atmosphere of 90% air and 10% CO₂. Routine passaging of cell stocks was carried out by removing cells with a solution containing 0.25% (w/v) trypsin (Sigma Chemicals, UK, T8128) and 0.2% (w/v) EDTA, and seeded at a density of 4×10⁴ cells cm⁻² in 75 cm² and 150 cm² flasks (Sterilin, UK). For filter-cultures, cells were grown in 30 mm diameter Millicell-HA chambers (Millipore, Bedford, MA, USA), nitrocellulose filters (uncoated; pore size 0.45 μm). These chambers were inserted into the wells of 6-well plates (Costar, UK). Cells were seeded at 2×10⁶ per Millicell-HA chamber. Cells in 2 ml of maintenance medium, containing penicillin (100i.u.ml⁻¹)/streptomycin (100 μg ml⁻¹) were placed on the filters: 2 ml of the same medium was placed under the filter in the well of the culture plate. The filter-grown cultures were incubated at 37 °C in a humidified atmosphere of 90% air, 10% CO₂. Medium was changed every 48 h. Cells were maintained on filters for up to 30 days.

Caco-2 cells, passage number 82, were treated with 2 ml of 0.25% trypsin/0.2% EDTA solution and seeded onto 75 cm² flasks at a density of 100 cells per flask. The cells were cultured in 50% DMEM supplemented with 20% FCS and 50% conditioned medium (medium exposed to a confluent monolayer of Caco-2 cells for 48 h, collected, centrifuged at 1600 g for 10 min and passed through a 0.22 /an filter) and incubated for 14 days. Small colonies were visible on the bottom of the flask. The medium was replenished and the colonies were left to grow to 1mm in diameter. Selected colonies were individually trypsinised using 5 mm stainless steel cloning rings sealed with high-vacuum grease (Dow Corning) and the resulting cell suspension placed in a well of a 24-well plate (Costar). The medium was replaced every 48 h and when confluent the cells were trypsinised and placed in the well of a 6-well plate (Costar). Further stocks of the clones were obtained from subsequent passage. One clone (clone 40) was recloned at passage number 11 using the same limiting dilution method as described above. Clones and reclones were cultured and cryo-preserved until required for characterisation.

The TER of the cell monolayer was determined using a modification of the von Bonsdorff et al. (1985) method. Briefly, Minimum Essential Medium (MEM)+0.1% (w/v) bovine serum albumin (BSA), 5 ml and 3 ml were added to the apical and basolateral chamber, respectively, and allowed to equilibrate at 25°C for 15 min. Salt bridges (2% agar, 2 M KC1) were connected to each chamber and a current of 100 μA was applied across the monolayer. The potential difference across the monolayer was measured and the electrical resistance (ohms cm^{− 2}) derived from Ohm’s Law.

Caco-2 cells and clones were seeded into 6-well plates (Costar) at a plating density of 2×10⁵ cells per well and were cultured under standard conditions. The medium was replenished every 48 h. Cells were trypsinised on a daily basis and the resulting single cell suspensions diluted and counted on a ZM Coulter Counter.

Experiments were performed using Caco-2 cells cultured in Millicell-HA chambers as outlined above. Cells were used at varying times after addition to the filters as indicated. [¹⁴C]tauro-cholic acid (Amersham, UK; 56 mCi mmol^- b, 0.25–0.5 μCi in 2 ml medium, was added to the apical fluid. Specific transport from the apical to basolateral side was determined from competition with a 400-fold excess of taurocholic acid (1 mg ml⁻¹) added to the apical side. After 4h at 37 °C (95% relative humidity and 10% CO₂) apical and basolateral samples were removed, 14 ml of scintillant (Lumagel, LKB) was added and they were counted using a Beckman LS1801 liquid scintillation spectrometer. Results were corrected to disintsmin^-1 by comparison with standard quench curves and are expressed as nmol [¹⁴C]taurocholic acid transported from the apical to basolateral side per 4 h across a filter of cells and calculated as the difference between [¹⁴C]taurocholic acid transported with and without excess unlabelled taurocholic acid.

Celle on filters were washed in fresh maintenance medium and fixed with a solution containing 2.5% (v/v) glutaraldehyde and 4% (w/v) formaldehyde (freshly prepared from paraformaldehyde) in 0.1M cacodylate buffer (pH 7.3) for 1 h at 25 °C. The cells were then rinsed in 0.1M cacodylate buffer and fixed with 1% osmium tetroxide in 0.1M cacodylate buffer for 1 h at 25°C. The cells were finally rinsed in 0.1M cacodylate buffer. Filters were removed from the plastic surround, quartered and placed in 0.1 M cacodylate buffer. Filter-grown cells were dehydrated through an ascending series of ethanol, infiltrated, embedded in Epon resin and cured at 60°C for 48 h. The filter quarters were embedded in flat embedding moulds. Cells were sectioned on a Reichert OM4 microtome, stained with uranyl acetate and lead citrate, visualised and photographed on a Philips CM10 electron microscope. Materials used throughout were of an appropriate electron microscopy grade.

Cells were fixed and dehydrated using the procedures for transmission electron microscopy (TEM). At the final 100% ethanol wash, the samples were dried in an Edwards (Cambridge, UK) critical point drier. Filters were glued onto an SEM stub, and a fine gold film was evaporated onto the surface in an Edwards sputter coater. The cells were viewed in a Cambridge Stereoscan 200 electron microscope.

Forty three stable clones were isolated using the method described above. All of these were cultured on filters and assessed for TER over a period of 30 days in culture. Fig. 1A shows a selection of clones with similar TER (clone 35) and higher TER (clones 12 and 40) to the parental population. Clone 12 had a TER of 2115 ohms cm⁻² after 9 days in culture and this decreased to 1405 ohms cm⁻² at day 29. It represented the clone displaying the highest TER over this time period. In the case of clone 40 the resistance profile was maintained for 20 passages. Clone 19 displayed very low resistance up to day 18, then increased dramatically to 2124 ohms cm⁻². On day 5 clone 1 had a relatively high resistance, 763 ohms cm⁻², but it decreased over a 25-day period to 429 ohms cm⁻². Reclones of clone 40, clone 40.1 and 40.2, had similar resistances to the parent clone (Fig. 1B). The reclones also displayed stability with respect to this feature for 10 passages.

A comparison of cell numbers between the parental population, clone 40 and the reclones of clone 40 was made after seeding and for subsequent days up to day 16. The cells were seeded at 2.0×10⁵ per well. Fig. 2 indicates a short lag phase (days 0–2), an exponential growth phase between days 2 and 6 (see inset) and a reduction in growth rate from day 6 onwards, with the cell number reaching 2.4×10⁶ per well at day 16. There was no significant difference (P>0.05) between the growth curves in the exponential phase for the parental population, clone 40 and the reclones of clone 40 (clones 40.1 and 40.2) with a generation time, calculated during the exponential growth phase, of 22.2±0.72 h (Fig. 2, inset). This is consistent with the generation time derived from results published by Pinto and coworkers (1983) but significantly lower than the value of 43.8 h reported by Fabricant and Broitman (1990).

The parental population and selected clones were cultured for 14 days in Millicell chambers and their ability to transport taurocholic acid was determined. Like the TER results, certain clones showed marked differences in the amount of taurocholic acid transported from the apical to basolateral side. Fig. 3 shows that clones 1 and 16 did not specifically transport taurocholic acid in the apical to basolateral direction. Interestingly, these clones did display a TER consistent with an integral monolayer, suggesting that the clones lack an inherent part of the taurocholic acid transport machinery. The integrity of the monolayer is further supported by the low level of passive transport of taurocholic acid demonstrated by these clones. Clone 19 did not transport taurocholic acid to the basolateral side specifically at day 14 although large amounts of passive transport resulted. This is consistent with the TER results where clone 19 displayed low resistance for the first 14 days in culture, <100 ohms cm⁻² (Fig. 1A), suggesting a freely permeable monolayer. Clone 43 (18.2±3.7 ng filter⁻¹ 4h⁻¹), specifically transported similar amounts of taurocholic acid to the parental population (24.1±1.0), whilst others, clones 12 (63.6±18.4), 15 (46.1±3.9), 33 (42.2±2.7) and 40 (43.8±4.1), transported significantly more than the parental cells from the apical to basolateral side. Clone 40 was stable with respect to specific apical to basolateral taurocholic acid transport over 10 passages (Fig. 4).

Clone 40 and the parental population were cultured on Millicell chambers for 21 days and the ability to transport taurocholic acid from the apical to basolateral side was tested at days 7, 14 and 21 (Fig. 5). Results indicate that the parental population was unable to transport taurocholic acid specifically at day 7 and the amount of transport varied considerably. On the other hand, clone 40 did transport taurocholic acid specifically at day 7 (21.0±3.8ng filter⁻¹ 4h⁻¹) with markedly better reproducibility. Thereafter, the amount of specific taurocholic acid transported by clone 40 was consistently higher at each time point compared to the parental population. This was most evident in the amount transported at 14 days (43.5±4.0 for clone 40 compared to 24.0±1.0 for the parental population) and at 21 days where the amount transported by the parental cells (51.3±4.8) was less than one third of the activity expressed by clone 40 (162.7±21.8).

Fig. 6 shows the specific, apical to basolateral, taurocholic acid transport results of the parental population, clone 40 and six reclones of clone 40 at day 15. The amount of specific taurocholic acid transported varies between the reclones, with reclone 40.7 (136.4±6.0ng filter^-1 4h^-1) transporting significantly more than clone 40 (110.3±10.3), and clone 40.1 (115.0±4.3) transporting similar amounts to clone 40 and significantly more than the parental population (69.3 ±8.4). Whereas, reclones 40.3 (47.3±1.0), 40.6 (44.9±1.5) and 40.4 (31.5±1.3) transported less than the parental population, with 40.5 (5.8±1.6) showing a marked reduction in taurocholic acid transported in comparison to the parental population.

A micrograph of the parental population at day 15 (Fig-7) shows that the microvilli profile was predominantly straight. The subcellular organelles were generally well preserved. Heterogeneity of microvilli, investigated by SEM of the apical surface of the parental monolayer (Fig-8), showed the presence of dense microvilli on individual cells, clusters of microvilli or cells having low numbers of microvilli. Clones 1, 15, 16, 33 and 43 showed similar morphology to the parental population at days 5 and 15 (data not shown).

Clone 12 did not display a distinct monolayer but rather a succession of overlapping cells (Fig. 9A). The apical cells had very prominent convoluted lateral interdigitations. This clone displayed extensive microvilli and numerous transverse profiles, showing that the structure was preserved in a position parallel to the apical surface. At high magnification the lateral interdigitations of the cells possessed obvious desmosomes and the cytoplasm in this area appeared devoid of organelles (Fig. 9B).

Interestingly, clone 40 was homogeneous on days 5 and 15 (Fig. 10), in that only electron-dense cells with straight microvilli profiles and non-convoluted lateral interdigitations were seen. By SEM clone 40 showed a homogeneous layer of apical microvilli that occurred in discrete groups (Fig. 11).

The Caco-2 cell line when grown on permeable supports represents a functional model of the GI epithelial barrier. We have previously shown the parental population to be morphologically heterogeneous (Woodcock et al. 1989) and the work described in this paper confirms this observation using electrophysiological and biochemical approaches.

Caco-2 cells have been assigned morphological descriptions using parameters such as height, width, microvilli height and density, and how closely they resemble a typical ileal absorptive epithelial cell. These values differ between laboratories, probably due to the influence of a variety of culture conditions. Most workers have demonstrated particular electrophysiological and biochemical parameters, namely, TER (Grasset et al. 1984; Hidalgo et al. 1989; Wilson et al. 1990), taurocholic acid transport (Hidalgo and Borchardt, 1990; Wilson et al. 1990), amino acid transport (Hidalgo and Borchardt, 1988), cobalamin uptake (Dix et al. 1987; Muthiah and Seetharam, 1987), cobalamin transport (Dix et al. 1990), cobalamin pathway (Hassan et al. 1991), sugar transport (Blais et al. 1987), folic acid uptake (Vincent et al. 1985) and brush border hydrolase activity (Zweibaum et al. 1984). In addition, we have previously shown that Caco-2 monolayers displaying a TER above 400 ohms cm² were relatively impermeable to the transport of [³H]inulin and horseradish peroxidase, which are macromolecules commonly used as permeability markers (Wilson et al. 1990).

We set out to isolate clones from the parental population that displayed higher taurocholic acid transport and were homogeneous with respect to certain morphological parameters. The limiting dilution cloning regime proved effective, with 43 stable clones being isolated. All the clones were able to grow on permeable supports, have been passaged at least ten times and were tested for stability with respect to TER.

The clones showed marked differences in the amount of taurocholic acid actively transported. Whilst some clones did not actively transport taurocholic acid, others transported much higher amounts than the parental population. It is now clear that the amount of taurocholic acid transported by the parental population represents the sum from sub-populations. Moreover, it is not surprising that different groups publish different amounts of taurocholic acid transported at different time points. For example, Hidalgo and Borchardt (1990) report that taurocholic acid transport in Caco-2 cells reaches a plateau after 28 days in culture, whereas Wilson and coworkers (1990) have shown a plateau after 15 days. Clone 40 displayed a greater amount of active transport that started at earlier time points. Why the reclones of clone 40 should differ so markedly in their ability to transport taurocholic acid is not, in these circumstances, clear. It would appear that most of the features required for enhanced transport have been selected for, but the population still contains subsets of cells with additional characteristics. Nevertheless, clone 40 should prove a useful clone with which to study the bile acid transport pathway in epithelial cells.

The clones displayed a wide range of TER profiles. Most were similar to the parental population or somewhat higher. Other epithelial cells have been isolated that display a range of TER measurements. For example, two strains of Madin-Darby Canine Kidney epithelial cells (MDCK) (Madin and Darby, 1972) have been isolated which demonstrate markedly different electrophysiological properties. High resistance monolayers have less cell volume, are less columnar, have fewer, shorter microvilli (Barker and Simmons, 1981; Simmons, 1982) and have striking differences in glycosphingolipid content (Hansson et al. 1986). However, whether these characteristics represent a basis for the difference between the high and low resistance monolayers of MDCK cells remains unclear.

In the case of the Caco-2 clones, one clone, (clone 19) was markedly different, showing very low TER for 18 days then a dramatic increase, whilst another, clone 12, showed extremely high TER from early time points. The TER results of clone 19 are relatively straightforward to explain, in that the clone was not confluent until day 18. On reaching confluence, tight junction formation will lead to an increase in TER. The explanation for some clones displaying extremely high resistance is not known. However, Claude and Goodenough (1973) have shown that epithelia displaying higher TER possess more strands in their tight junctions than do epithelia displaying low TER. Similar studies would have to be performed on the Caco-2 parental population and clones to assess the importance of strand assembly.

Attempts to correlate morphological features with TER did not provide any clear indication of which features were important for high resistance, although some features are worth noting. For example, although clone 12 displayed a poorly developed brush border at day 5 it became extensive by day 15. Given that the TER decreased over this period it is unlikely that this feature contributed to the elevated results obtained. However, over the same period, several cells within this clone displayed extensive convoluted lateral interdigitations. This feature occurred in the parental population and other clones but not to the same degree and therefore may contribute to the high TER. However, it would be more convincing if other clones were isolated that had similarly high TER measurements and also possessed prominent lateral interdigitations.

Clone 40 presented a higher resistance than the parental population and this trait was found to be stable over 20 passages. In addition, reclones of clone 40 showed similar TER measurements and one of the reclones, clone 40.1, was stable for 10 passages. Cell growth curves were also found to be uniform for the parental population, clone 40 and the reclones. These results are consistent with the isolation of a stable clone from the parental population and the results obtained subsequent to recloning support this assertion.

We have successfully cloned the Caco-2 cell parental population and isolated a stable population of cells that significantly increased taurocholic acid transport over an extended period. These cells will be satisfactory for detailed analysis of the taurocholic acid transport pathway in epithelial cells. We believe that using clones of this kind also offers a significant advantage over the parental population in terms of stability and homogeneity of cell types.

We give warm thanks to Professor Colin Hopkins for discussion, and Isobel Brooks for advice and technical assistance during cell cloning. We thank John Hastewell for assistance with statistical analysis. We also thank Christine Malin for expert preparation of the manuscript.