The bioavailability of the nutritionally essential amino acid, L-lysine, from the diet depends upon its efficient transport across the intestinal mucosa. Different transport systems have been shown to accept cationic amino acids, including L-lysine, and they are believed to participate in L-lysine uptake into most cell types and in its efflux out of specialized cells such as the epithelium of the small intestine and of the renal proximal tubules (Maillard et al. 1995). In addition, some of these transporters were shown to interact with both cationic and zwitterionic amino acids, and a trans-stimulating effect was described for the uptake of cationic amino acids by the neutral amino acid L-leucine (Maillard et al. 1995, White 1985). Most of the functional information on these systems was obtained from experiments on animals in vivo or ex vivo. More recently, the use of alternative experimental strategies (i.e., expression cloning and methods based on partial sequence homology) has yielded several cDNA sequences coding for cationic amino acid membrane transport-related proteins, providing the first available structural information on these systems (Bertran et al. 1994).

Still very little is known of the transport of L-lysine and other cationic amino acids across the human small intestine. The sodium-independent uptake of the cationic amino acid L-arginine from the apical (AP)⁴ membrane of the human intestinal Caco-2 cell line has recently been ascribed (Pan et al. 1995, Thwaites et al. 1996) to the ubiquitous system y⁺ encoded by mCAT-1 and to system b^0,+ , encoded by rBAT and originally described in blastocysts but now known to be expressed in different cell types; in addition, a nonsaturable transport component and other transporters such as system y⁺L, recently described in human erythrocytes and placenta (Bertran et al. 1994, Eleno et al. 1994, Maillard et al. 1995), may contribute to cationic amino acid uptake in human intestinal cells.

Studying the transepithelial passage of L-lysine across Caco-2 cells, we have reported the presence of distinct transport mechanisms, exhibiting different requirements for sodium, operating on the AP membrane for the uptake into the cell and on the basolateral (BL) membrane for the efflux out of the cell (Ferruzza et al. 1995). Although the AP uptake of L-lysine did not require extracellular sodium, in accord with the presence of at least two sodium-independent transport systems as previously reported (Pan et al. 1995, Thwaites et al. 1996), we described a strong sodium dependency for the BL efflux of this amino acid (Ferruzza et al. 1995).

Caco-2 cells grown and differentiated on permeable filter supports have frequently been used to study the transepithelial passage of nutrients or drugs at the level of the intestinal mucosa (Alvarez-Hernandez et al. 1994, Chandler et al. 1993, Chen et al. 1994, Ferruzza et al. 1995, Meunier et al. 1995, Ranaldi et al. 1994). This system allows discrimination of the mechanisms responsible for uptake from those involved in the efflux of the nutrient under study, as recently shown for the transport of the amino acid L-methionine (Chen et al. 1994). We have employed this experimental model to characterize the efflux of L-lysine out of the BL membrane. The results show that more mechanisms appear to be responsible for the efflux of lysine out of the BL membrane of Caco-2 cells: a nonsaturable, sodium and energy-independent pathway, an active efflux mechanism that is stimulated by the presence of an inwardly directed sodium or lithium gradient and a sodium and energy-independent exchange mechanism that operates when other amino acids are present in the extracellular fluid and results in the counter-exchange of L-lysine for zwitterionic or cationic amino acids.

MATERIALS AND METHODS

For efflux experiments, the cells were seeded on polycarbonate filter cell culture chamber inserts (Transwell, 24-mm diameter, 4.7 cm² area, 0.45 µm pore diameter; Costar Europe, Badhoevedorp, The Netherlands) at a density of 2·10⁶ cells/filter; the high seeding density allows confluency to be reached within 48 h. Caco-2 cells were left to differentiate for 13-17 d after confluency; the medium was regularly changed three times a week. The intactness of the cell monolayer and the full development of the tight junctions were monitored before every experiment by determining the transepithelial electrical resistance (TEER) of filter-grown cell monolayers using a commercial apparatus (Millicell ERS; Millipore, Bedford, MA) as previously described (Ferruzza et al. 1995). Only cell monolayers with TEER > 1000 ·cm² were used for efflux experiments.

To determine the intracellular concentration of L-lysine in the soluble pool, at the end of the loading period, duplicate filters were transferred on ice and rapidly washed with cold saline; the filters were carefully removed from the plastic ring with a scalpel and placed in 1 mL cold 50 g/L trichloroacetic acid (TCA) for 2 h at 4°C. To optimize the recovery, a plastic scraper was used to collect the cell extract from the filters. At the end of the incubation, the TCA-precipitated pellet, representing amino acid incorporation into newly synthesized proteins, was separated from the soluble pool by centrifugation at 15,000 × g for 5 min at 4°C, and the precipitate and the soluble fraction were assessed for radioactivity. The concentration of soluble L-lysine in the cellular pool was calculated using an intracellular volume of 3.49 µL/mg protein, as previously calculated for Caco-2 cells (Burnham and Fondacaro 1989).

After loading, the cells to be used for efflux measurement were washed three times with cold saline, the appropriate solutions, prewarmed to 37°C, were added to the AP and BL compartments and the Transwell plates were transferred to a water bath at 37°C. During efflux, unless otherwise stated, the AP medium was sodium-containing HBSS and the BL medium was HBSS with different additions or substitutions as described below. During efflux, to avoid amino acid backflow, the BL medium was changed after each time point with fresh prewarmed saline. Efflux was usually measured between 1 and 15-17 min, with time points taken every 1, 2 and 4 min. At the end of the efflux experiment, cells were transferred on ice and treated as described for cells after the loading period to determine the residual L-lysine concentration in the intracellular soluble pool.

The radioactive amino acid (³H-L-lysine) in the intracellular pool and in the efflux medium was analyzed by liquid scintillation spectrometry using a Beckmann LS1801 instrument (Beckmann Instruments, Fullerton, CA) after diluting with liquid scintillation cocktail (Ready Safe; Beckmann Instruments).

To determine the ionic requirements for efflux, NaCl in HBSS was replaced with equimolar amounts of either choline chloride or LiCl, and sodium salts were replaced with their potassium equivalents. Alternatively, ouabain (1 mmol/L) was added to the sodium- or lithium-containing HBSS in the BL compartment during the loading and the efflux periods, to inhibit the Na,K-ATPase responsible for maintaining the sodium gradient across the cell membrane. For energy depletion, the medium in both compartments was supplemented with 1 mmol/L NaN₃ and 50 mmol/L 2-deoxy-glucose during both the loading and the efflux periods. For trans-stimulation experiments, the BL medium was supplemented with 10 mmol/L (unless otherwise stated) of nonlabeled amino acid.

To study the effects of N-ethylmaleimide (NEM) on L-lysine efflux, cells were loaded under control conditions for 30 min; then 500 µmol/L NEM was added to the BL medium and loading was continued for an additional 10 min. NEM was also maintained in the BL media during the efflux period.

The binding of ³H-L-lysine to cell-free filters was measured and found to be <0.005% of the total applied radioactivity (data not shown).

The identity of the ³H-label (in the AP compartment or the intracellular soluble pool, or that collected from the BL compartment after efflux) was determined by TLC on cellulose plates as previously described (Ferruzza et al. 1995); in all cases, the ³H-radioactivity was found to migrate in the position corresponding to the ³H-L-lysine standard; no other radioactive products were detected (data not shown).

For total protein determination, filter-grown Caco-2 cells were dissolved in 1 mol/L NaOH and the protein assayed by a colorimetric method (Lowry et al. 1951). Differentiated Caco-2 cells on filter had a total protein content of 1.31 ± 0.12 mg for a corresponding surface area of 4.7 cm² .

Initial rates and kinetic parameters of L-lysine efflux were calculated by fitting the data to theoretical equations by nonlinear regression analysis using a computer program (Sigma Plot; Jandel Scientific, Corte Madera, CA) based on the Marquardt-Levenberg algorithm (Marquardt 1963). Data were analyzed by one-way ANOVA followed by the Scheffé F-test to determine significant differences among means (Kleinbaum and Kupper 1978). In addition, where appropriate, two-way ANOVA was used to determine the significance of different experimental conditions and the interactions among them. ANOVA and the Scheffé F-test were performed using the Statview 512 computer program (Brainspower, Calabasas, CA). Values in the text are means ± SD.

RESULTS

To investigate the previously observed sodium requirement for L-lysine efflux, after the loading period, the BL medium was substituted with fresh HBSS containing either sodium or choline. In the presence of sodium in the BL medium, efflux led to an 80% depletion of intracellular L-lysine in 40 min; conversely, when sodium in the BL medium was substituted with choline, efflux was slower, leading to a 34% depletion in 40 min. (Fig. 2A). The rate of BL efflux of L-lysine was reduced by 72% with choline-containing HBSS in the BL medium compared with the sodium-containing control (Fig. 2B). However, if sodium was substituted with an equimolar concentration of lithium ions in the BL medium, efflux was faster (~130%) than in the presence of sodium. The presence of ouabain in the sodium-containing BL medium did not affect loading (data not shown), but significantly decreased L-lysine efflux by ~27% in the presence of sodium and by 37% in the presence of lithium (Fig. 2A, B).

The energy requirements of the BL efflux of L-lysine were investigated by performing the experiment in the presence of 1 mmol/L NaN₃ and 50 mmol/L 2-deoxy-glucose to inhibit energy metabolism, as previously described (Ferruzza et al. 1995). Although loading the cells with L-lysine in the presence or absence of energy inhibitors did not significantly affect the intracellular concentration of soluble L-lysine that was achieved in 30 min (P > 0.05; Fig. 3A), the rate of efflux was strongly reduced under energy-deprived conditions by 63 and 56% in Na-containing or Li-containing HBSS, respectively (Fig. 3B, C). Lowering the temperature during the efflux period to 4°C also dramatically reduced the rate of L-lysine efflux by more than 93% of control (Fig. 3B, C).

The efflux of L-lysine was investigated at different intracellular concentrations of soluble amino acid by loading the cells from the AP side with increasing extracellular concentrations (10 µmol/L-1 mmol/L) of L-lysine. This resulted in intracellular concentrations of soluble L-lysine ranging from 0.20 ± 0.014 mmol/L to 10.8 ± 0.42 mmol/L, corresponding to a 20- to 11-fold accumulation, respectively, of L-lysine in the intracellular pool with respect to the extracellular concentration. Figure 4 shows the initial rates of efflux at all L-lysine concentrations tested. Efflux was measured in the presence of sodium or choline in the BL medium. The rates of efflux in the absence of sodium (choline) were measured at increasing intracellular concentrations and the results fitted to the equation v = K_d [AA] [Equation (2)], representative of a diffusional component, where v is the velocity of transport, [AA] is the amino acid concentration and K_d is the diffusion constant for a nonsaturable component. The apparent K_d for efflux in the presence of choline was 23.5 ± 0.5 pmol/(min·mg protein·mmol·L¹). The rates of efflux obtained in the presence of sodium were corrected by subtracting this diffusional component, and the resulting data were fitted to the following equation: where v is the velocity of efflux, [AA] is the amino acid concentration, V_max is the maximal velocity of efflux, K_m is the amino acid concentration at which the velocity is half-maximal. For the corrected efflux, the calculated values of the apparent kinetic parameters were K_m 3.57 ± 0.93 mmol/L, V_max 466 ± 56 pmol/(min·mg protein).

The effects of amino acids on the trans-side of the membrane on L-lysine efflux were investigated by measuring efflux in the presence of different amino acids at 10 mmol/L in the BL medium. Figure 5A shows the trans-stimulating effect of several amino acids on L-lysine efflux in a typical experiment. Figure 5B shows that the initial velocity of efflux was strongly trans-stimulated by four amino acids and more weakly by three amino acids, with the following ranking L-leucine > L-methionine = L-ornithine = L-arginine  L-serine > L-phenylalanine = glycine. As shown in Figure 5A, the most powerful trans-stimulating amino acid L-leucine led to >95% depletion of intracellular L-lysine concentration in 5 min.

We further investigated the requirements of efflux in the presence of extracellular trans-stimulating amino acids by studying the effects of sodium or choline (Fig. 6A) and of the energy inhibitors NaN₃ and 2-deoxyglucose (Fig. 6B). The trans-stimulation of L-lysine efflux by the cationic amino acids L-ornithine and L-arginine was not affected by sodium substitution with choline, or by energy deprivation. Conversely, for the neutral amino acids L-leucine, L-methionine and L-serine, maximal trans-stimulation could be achieved only in the presence of extracellular sodium ions and of normal levels of cellular energy. The trans-stimulating effect of L-leucine and L-arginine was also investigated at different amino acid concentrations in the BL medium. Figure 6C shows that, although L-leucine exhibited a stronger trans-stimulation effect than L-arginine, which increased for both amino acids with concentration from 0.1 to 10 mmol/L, the minimal extracellular concentration capable of eliciting a significant trans-stimulating effect was 0.1 mmol/L for L-leucine and 1 mmol/L for L-arginine.

Because the sulfhydryl-binding reagent NEM is known to inhibit the activity of certain amino acid transporters (Chillaron et al. 1996, Pan et al. 1995), the sensitivity of L-lysine efflux to NEM was determined under different experimental conditions. Figure 7A shows that adding NEM during the last 10 min of the loading period did not significantly affect the concentration of L-lysine in the intracellular soluble pool compared with the untreated control. Conversely, NEM treatment significantly decreased efflux in the presence of sodium or lithium (35% of control), whereas it did not significantly affect efflux in the presence of choline (Fig. 7B). When efflux was measured in the presence of trans-stimulating concentration (10 mmol/L) of L-leucine or L-arginine in the extracellular BL medium, NEM did not appear to have any effect on the initial rate of L-lysine efflux (Fig. 7C).

DISCUSSION

A recent report has shown that the AP uptake of the cationic amino acid, L-arginine, by 9-d-old Caco-2 cells occurs via a sodium-independent transport mechanism attributed to the presence of systems y⁺ and y⁺L (or b^0,+), defined as the L-leucine-insensitive and NEM-sensitive or L-leucine-sensitive and NEM-insensitive transport activity, respectively (Pan et al. 1995).

Taking advantage of our previous observations on the sodium dependency of the BL efflux of L-lysine (Ferruzza et al. 1995), we show that in the absence of sodium (replaced by choline) in the BL compartment, Caco-2 cells accumulated L-lysine in the intracellular soluble amino acid pool to levels between 10 and 20 times the extracellular concentration. This has allowed the study of the kinetics and requirements for L-lysine efflux from the BL membrane.

The cellular volume of differentiated Caco-2 cells has been reported to range between 3.49 and 3.66 µL H₂O/mg protein (Blais et al. 1987, Burnham and Fondacaro 1989), calculated by the (¹⁴C)3-O-methyl-D-glucose equilibration method. We have used the value of 3.49 µL H₂O/mg protein to estimate the intracellular amino acid concentrations reached after the loading period and the distribution ratio between intracellular and extracellular concentrations during the efflux period.

Because L-lysine is positively charged at the intracellular pH, its efflux out of the cell must occur against an unfavourable electrical gradient because of the more negative potential difference inside the cell. In the presence of a large concentration gradient (0.2-10.8 mmol/L intracellular vs. no L-lysine in BL medium) and no sodium (substituted by choline) in the BL medium, L-lysine efflux increased linearly with concentration and was nonsaturable (Fig. 4). In the presence of sodium in the BL medium, efflux was strongly stimulated, indicating that the presence of an inwardly directed sodium gradient across the BL membrane facilitated L-lysine efflux. In addition, substitution of sodium with lithium in the BL medium produced a significant increase in the rate of lysine efflux compared with the sodium-containing control. Thus, L-lysine efflux appears to be favored by the presence of an inwardly directed gradient of cations such as sodium or lithium. Addition of ouabain to the BL medium to block the Na,K-ATPase significantly reduced the efflux of L-lysine in sodium- or lithium-containing HBSS. However, the inhibitory effect of ouabain on L-lysine efflux (27% inhibition in sodium and 37% in lithium) was less than that observed with choline-containing HBSS in the BL medium (72% inhibition compared with sodium control). The effect of ouabain suggests that the Na,K-ATPase may be involved in the maintenance of the sodium gradient that facilitates L-lysine efflux. The ouabain effect in the presence of lithium in the BL medium is more difficult to explain; lithium gradients cannot be directly maintained by the operation of the Na,K-ATPase because the cytoplasmic activation site of the pump is highly specific for sodium (Skou 1988). However, it has been shown that certain sodium-dependent transport systems such as system N and ASC accept sodium substitution by lithium (Jacob et al. 1986, Kilberg et al. 1980).

The significant energy dependency of efflux shown by the effect of metabolic energy inhibitors and by the strong effect of temperature reduction indicates that an active transport mechanism is responsible for a large proportion of L-lysine efflux. The rate of L-lysine efflux in the absence of sodium (or lithium) was similar to that obtained in conditions of reduced cellular energy (36 vs. 28% of control, respectively). This indicates that most of the efflux occurring in the presence of choline is likely to represent passive diffusion, as shown also by its nonsaturability at higher concentrations (Fig. 4). By subtracting this estimated passive efflux component from the efflux in the presence of sodium, we calculated apparent kinetic parameters for the active, sodium-stimulated efflux: K_m = 3.57 ± 0.93 mmol/L, V_max = 466 ± 0.56 pmol/(min·mg protein).

The idea that an active efflux mechanism may be present on the BL membrane of absorptive enterocytes to guarantee efficient transport of the nutritionally essential cationic amino acids to the blood was originally proposed by Cheesman on the basis of compartmental analysis of lysine movement across the vascularly perfused small intestine (Cheeseman 1992). Unfortunately, to the best of our knowledge, no study has ever addressed the ionic and energy requirements of the BL efflux of cationic amino acids from either intestinal or renal epithelial cells, and the vast majority of reports deal with their uptake, mostly from the AP and sometimes from the BL membrane. However, in analogy with the trans-stimulating effect of some neutral amino acids on the uptake of cationic amino acids (White 1985), the efflux of L-lysine from rat small intestine was shown to be greatly increased when L-leucine was present on the trans-side of the membrane (Lawless et al. 1987). The trans-stimulating effects of neutral (i.e., leucine, alanine, methionine) or cationic (lysine, arginine, histidine) amino acids on sodium-independent L-lysine efflux from the BL membrane of Caco-2 cells has recently been reported (Thwaites et al. 1996).

We therefore tested the effects of different amino acids in the extracellular medium on the rate of efflux of L-lysine from Caco-2 cells under different ionic and energy conditions. Zwitterionic amino acids such as L-leucine, L-methionine and the cationic amino acids L-ornithine and L-arginine strongly stimulated the efflux of L-lysine by 40 to 25 times. Interestingly, under conditions of trans-stimulation, the efflux exhibited energy-independence and was no longer accelerated by the presence of an inwardly directed gradient of sodium or lithium (Fig. 6). A notable difference was that only the zwitterionic but not the cationic amino acids required extracellular sodium (or lithium, data not shown) and energy to achieve maximal trans-stimulation of L-lysine efflux. This, together with the lack of effect of NEM on the trans-stimulated efflux of L-lysine, which markedly contrasts with the significant NEM inhibition of efflux under basal conditions (Fig. 7), strongly suggests that different efflux mechanisms may be involved depending on the extracellular availability of amino acids. Overall, these results indicate that multiple pathways exist for L-lysine efflux out of the BL membrane of Caco-2 human intestinal cells.

Among the different genes recently identified to code for proteins involved in the transport of cationic amino acids (Palacin 1994), Caco-2 cells have been shown to express mRNA highly homologous to mCAT1 (inducing y⁺-like activity) (Pan et al. 1995) and to rBAT (inducing b^0,+-like activity) (Thwaites et al. 1996), whereas data on the expression of 4F2hc mRNA (inducing y⁺L-like activity) in these cells are not available.

A recent report has shown that, in Caco-2 cells, system b^0,+-like activity occurs principally at the BL membrane and is likely to contribute to exchange efflux of lysine (Thwaites et al. 1996). Our data confirm the presence of a similar transport activity in the BL membrane of Caco-2 cells acting under sodium-free conditions to counter-exchange L-lysine with cationic or zwitterionic amino acids. In addition, however, we cannot exclude that other transporters (such as y⁺L or others with similar exchange characteristics) may contribute to L-lysine efflux in exchange for extracellular amino acids. In addition, we report an active, sodium (or lithium)-stimulated pathway as well as a nonsaturable sodium and energy-independent pathway that operate in the absence of extracellular amino acids. Whether the different efflux characteristics exhibited in the presence or absence of extracellular amino acids represent functional variations of the same transporter or distinct transport activities remains to be resolved.

In conclusion, we suggest that the functional differences of L-lysine efflux in Caco-2 cells may indicate that in the human intestine multiple mechanisms guarantee optimal bioavailability of this essential amino acid under different physiological conditions: the active and sodium (or lithium)-stimulated mechanism may ensure efficient exit of L-lysine against the electrochemical gradient even when amino acid flux is scarce (i.e., between meals or at the beginning of the absorptive phase); subsequently, during digestion, when amino acid flux is increased by the operation of several carriers, efflux by amino acid counter-exchange may prevail.