Introduction

Mammalian cells regulate magnesium ion (Mg2+) homeostasis via a complex system of transport, buffering, and signaling mechanisms. As a result of this regulation, free Mg2+ concentration within the cytoplasm and the mitochondrial matrix is approximately 0.8 mM at front of 18–20 mM total Mg2+ content present within nucleus, mitochondria, and endoplasmic reticulum [1, 2]. As cellular Mg2+ content is maintained below its electrochemical equilibrium [1, 2], specific and powerful extrusion mechanisms must operate in the cell membrane to maintain Mg2+ concentrations. A Na+-dependent mechanism putatively termed Na+/Mg2+ exchanger [3] appears to be the main pathway utilized by mammalian cells to extrude Mg2+ across the plasma membrane [1–3]. Because this exchanger has not been cloned as yet, all the available information is based on pharmacological and indirect evidence, and can be summarized as follows: (1) this exchanger is ubiquitous (see ref. [2] for a list of cells and tissues); (2) it strictly depends on extracellular Na+ to operate; and (3) it becomes active following the phosphorylation by cAMP [4] irrespective of the Gαs-coupled receptor activation. In this respect, activation of the exchanger has been observed following stimulation of β1-adrenergic receptors in cardiac myocytes [5, 6], β2-adrenergic, or glucagon receptors in hepatocytes [7–9], β2-adrenergic receptors in red blood cells [10], or PGE2 receptors in lymphocytes [11] just to name a few cell models. Consistent with a role of cAMP in Mg2+ extrusion, the Na+/Mg2+ exchanger can be activated by forskolin or cell-permeant cAMP analogs [5–9], and inhibited by Rp-cAMP stereoisomer [12]. No specific inhibitor for this exchanger is currently available, but its operation can be blocked by non-selective agents such as amiloride, imipramine, or quinidine [5, 13], or the removal of extracellular Na+ [3].

The operation of the Na+/Mg2+ exchanger, its strict dependence on extracellular Na+, and its inhibition by amiloride, imipramine, and quinidine have been confirmed in plasma membrane vesicles from liver [14, 15], heart [16], and intestine [17]. In both liver [15] and heart [16] plasma membrane vesicles, the activation of the exchanger via cAMP-dependent phosphorylation has been validated under in vitro conditions. In the case of the liver, the operation of the exchanger appears to be specifically confined to the basolateral domain of the hepatocyte [15].

While the operation and regulation of this exchanger have been corroborated in a variety of cell types and experimental models, uncertainty still remains about its stoichiometry of exchange. The original assessment that the exchanger is electroneutral [18] in chicken and turkey red blood cells has not been confirmed in human [19] or ferret erythrocytes [20], in which an electrogenic 1Na +in :1 Mg 2+out exchange ratio has been measured. A similar electrogenic ratio has been determined in liver plasma membrane vesicles [21].

The different stoichiometry ratios reported in the literature raise the question as to which extent experimental conditions and medium composition affect the operation of this transport mechanism. In particular, experimental evidence suggests a modulating role by intracellular and/or extracellular Cl− on the operation of the exchanger. It has been reported that the addition of increasing extracellular concentrations of Cl− (as choline chloride) stimulates Mg2+ extrusion from liver cells via a DIDS inhibited pathway [22]. In erythrocytes, intracellular Cl− appears to stimulate the Na+/Mg2+ exchanger through a cooperative effect [23]. A similar effect has been observed in the dialyzed squid axon [24], in which transport of Cl− via the Na+/Mg2+ exchanger has been observed. It remains unclear, however, whether Cl− ions can affect the stoichiometry ratio of the exchanger.

This present study was undertaken to assess the role played by extra- and intra-cellular Cl− on the stoichiometry exchange ratio of the Na+/Mg2+ antiport in liver cells. For this purpose, basolateral liver plasma membrane vesicles loaded with MgCl2 or Mg-gluconate were incubated in the presence of varying extravesicular concentrations of NaCl or Na-isethionate. The results reported here indicated that Cl− ions are not required for the Na+/Mg2+ exchanger to operate, but their absence converts the stoichiometry exchange ratio of the antiport from electrogenic to electroneutral. Furthermore, Cl− appears to be extruded from the vesicles in concomitance with Na+-induced Mg2+ extrusion. The change in ratio and the Cl− extrusion can have important implications for Na+, Mg2+, and Cl− cellular homeostasis, and have important repercussions on the membrane potential of the hepatocyte.

Materials and methods

Materials

All chemicals were of the purest analytical grade (Sigma, St. Louis, MO). Nitrex nylon mesh was from Tetko (Briarcliff Manor, NY). 36Chloride was from American Radiolabeled Chemicals, Inc (St Louis, MO).

Animals ethics

Male Sprague–Dawley rats (220–250 g body weight) were used as organ donors. Animals were maintained and handled in accordance with the Guide for the Care and Use of Laboratory Animals (Institute of Laboratory Animal Resources, Commission on Life Science, National Research Council 1996), as approved by the Animal Resource Center (ARC) at Case Western Reserve University, Cleveland, Ohio.

Isolation of total plasma membrane vesicles

Animals were anesthetized by intra peritoneal injection of saturated sodium pentobarbital solution (65 mg/kg body weight) provided through the ARC. Total liver plasma membrane vesicles (tLPM) were isolated on Percoll gradient and stored as described in detail by Cefaratti et al. [25]. Plasma membrane purity was determined by assessing the activities of 5′-nucleotidase, cytochrome-c oxidase, and glucose 6-phosphatase as enzymatic markers for plasma membrane, mitochondria and endoplasmic reticulum, respectively (see [25] for detail). The orientation of loaded tLPM vesicles was determined by measuring Na+/K+-ATPase and 5′-nucleotidase activities [25]. The comparison of these activities to those of detergent-disrupted vesicles (considered as 100%) confirmed that ≥90% of tLPM vesicles were in "inside-in" configuration following the loading with Mg2+, as previously reported [25].

The plasma membrane endogenous carryover of cations (Ca2+, Na+, K+, and Mg2+) and adenine phosphonucleotides were measured by atomic absorbance spectrophotometry (AAS) and HPLC, respectively, and found to be negligible (not shown).

Isolation of basolateral plasma membrane vesicles (bLPM)

After isolation, tLPM were washed with five volumes of 250 mM sucrose, 25 mM Hepes/Tris (pH 7.4), and sedimented at 34,500g ×10 min in a SS-34 Sorvall rotor to remove residual Percoll from the purified fraction. The tLPM vesicles were further homogenized as described in detail elsewhere [14] to purify basolateral plasma membrane vesicles (bLPM). Na+/K+-ATPase, alkaline phosphatase, and 5′-nucleotidase were used to assess purity and orientation of bLPM vesicles (see ref. 14 for detail). The BLPM vesicles were stored in liquid nitrogen until used.

The data relative to total and basolateral plasma membrane vesicles orientation and purity were similar to those reported by us in previous publications [14, 25]. Hence, they are not reported here for simplicity.

Vesicles loading

Five mL aliquots of tLPM, or bLPM were resuspended in 25 ml of 250 mM sucrose, 25 mM Hepes (pH 7.4 with Tris) in the presence of 20 mM MgCl2, or 20 mM MgCl2 labeled with 1 μCi 36Cl− where specified, and loaded by four passes in a Thomas C Potter with a tight-fitting pestle at 4°C [25]. This Mg2+ concentration was utilized based on previous reports indicating it as optimal to study Mg2+ transport in our experimental model [14, 15, 25]. The loading mixture was sedimented at 34,500g ×10 min in a Sorvall SS-34 to remove excess extravesicular cation and isotope labeling. Mg2+-loaded vesicles were resuspended in 5 ml of 250 mM Sucrose, 25 mM Hepes/Tris (pH 7.4) devoid of Mg2+ (Mg 2+-free medium) and stored in ice until used. Efficiency of loading was determined by exposing plasma membrane vesicles to ionophore (A23187, 2 μg/ml), or 0.05% Triton X-100, and measuring the amount of Mg2+ extruded in the extravesicular space or retained within the vesicle pellet using atomic absorbance spectrophotometry (AAS) in a Perkin Elmer 3100, as reported previously [14, 25]. Spontaneous leakiness of Mg2+ and 36Cl− from loaded vesicles were assessed by measuring Mg2+ and 36Cl− content in the extravesicular milieu by AAS and β-scintillation counter (Beckman LS6000), respectively, and found to be negligible.

Measurement of Mg2+ fluxes

Mg2+ fluxes were measured by AAS. An aliquot of Mg2+-loaded vesicles (tLPM, or bLPM) was incubated in the Mg 2+-free medium mentioned above, at 37°C, under continuous stirring, at the final concentration of 300 μg protein/ml. Magnesium (Mg2+) present as contaminant in the medium was assessed by AAS and found to be ≤2 μM (1.87 ± 0.35 μM, n = 9). After 2 min of equilibration, aliquots of the incubation mixture were withdrawn in duplicate at 1 min intervals. The vesicles were sedimented in microfuge tubes at 7,000g ×45 s, and total Mg2+ content of the supernatants was measured by AAS. The vesicle pellets were digested overnight in 500 μl 10% HNO3. Following sedimentation of denatured protein in microfuge tubes, Mg2+ content of the acid extracts was measured by AAS. The first two time points after the equilibration period (i.e., t = 0 and t = 1 min) were used to establish a baseline. Following the withdrawal of t = 1 min sample, the concentrations of NaCl and Na-isethionate reported in the figures were added to the incubation mixture, and the incubation continued for four additional minutes. Because Mg2+ content in the supernatant could vary considerably among preparations as a result of loading carry-over, the data are reported as the net variation in extravesicular Mg2+ content, normalized per miligram of protein, for simplicity. To calculate net Mg2+ extrusion, Mg2+ content in the supernatant or in the pellet at t = 0 min and t = 1 min were calculated, averaged, and subtracted from the values of the subsequent incubation time points. Sodium content accumulated within the vesicles in exchange for entrapped Mg2+ was also measured by AAS in acid extracts of tLPM and bLPM vesicles.

For the experiments in which amiloride, imipramine, DIDS, niflumenic acid, bumetanide, or furosemide were used, the drug of choice was added to the incubation system together with the bLPM vesicles.

Determination of Cl− fluxes

Following loading of the vesicles with 20 mM MgCl2 labeled with 36Cl−, aliquots of the incubation mixture were withdrawn in duplicate and rapidly filtered onto glass fiber filters (Whatman filters, 0.25 μm) under vacuum. Filters were washed with 5 ml 250 mM "ice-cold" sucrose under vacuum, and dried overnight. The radioactivity retained within the vesicles trapped onto the filter was detected in a β-scintillation counter (Beckman LS6100). In some experiments, 50 μg/ml digitonin were added at selected time points to ensure the released of residual 36Cl− entrapped within the vesicles and assess loading homogeneity.

Assessment of stoichiometry ratio

To determine the stoichiometry ratio of the Na+/Mg2+ exchanger, the amount of Na+ accumulated within and the amount of Mg2+ mobilized from the vesicles was assessed by acid digestion of the vesicles at the end of the incubation protocol. To this end, MgCl2-loaded bLPM vesicles were incubated as described previously in a nominal Mg2+-free medium, and stimulated by the addition of varying doses of extravesicular NaCl (or Na-isethionate). At 1-min interval, aliquots of the incubation mixture were withdrawn in duplicate and sedimented at 7,000 rpm × 45 s through an oil layer (dioctyl-phthalate:dibutyl-phthalate 2:1, v/v). The supernatant and the oil layer were removed by vacuum suction, and the pellet vesicles digested overnight in 500 μl 10% HNO3. The following day, the acid mixture was sedimented at 14,000 rpm × 3 min. The amounts of Na+ and Mg2+ present in the supernatant (acid extract) were measured by AAS after proper dilution. The denatured protein pellet was digested in 1 N NaOH for protein determination. The amount of Na+ accumulated within the vesicles and the amount of Mg2+ loss from the vesicles were normalized for protein content and used to calculate the ratio of exchange.

Assessment of TPP+ distribution

A tetraphenylphosphonium (TPP+) electrode was used to monitor changes in membrane potential under our experimental conditions. The TPP+-sensitive electrode was made as detailed by Cefaratti [21]. In brief, a solution of 100 mg sebacic acid bis-2-ethylhexyl ester, 97.5 mg polyvinylchloride (PVC) and 20 mg Na-tetraphenylborate (TPB) was prepared in 1 ml tetra-hydofuran (THF). All the components were added and solubilized in the PVC–THF solution, which was poured into a 2.5-cm-diameter ring plate and left to evaporate for 48 h. A piece of membrane was then glued onto 2.5-mm-diameter PVC tubing. The TPP+ electrode was filled with 1–100-μM TPP+ in 10 mM KCl. The reference was an Ag–AgCl electrode filled with 3 M KCl. The response of the electrode in the nano or micromolar range was logarithmic with a slope of 63 mV [21].

Protein determination

Protein was measured according to the procedure of Lowry et al. [26], using bovine serum albumin as a standard.

Statistical analysis

Data are presented as means ± S.E. Data were first analyzed by one-way ANOVA. Multiple means were then compared by Tukey’s multiple comparison test or by the Student–Newman–Keuls method performed with a level of statistical significance designated as P ≤ 0.05.

Results

The putative Na+/Mg2+ exchanger is specifically located in the basolateral domain of the hepatocyte [20]. Hence, for the experiments reported here, only this vesicle subpopulation was used.

Basolateral (bLPM) liver plasma membrane vesicles incubated in an ion free-medium did not release entrapped Mg2+ over several minutes of incubation (Fig. 1a). The addition of extravesicular NaCl, however, elicited a rapid and large Mg2+ extrusion (Fig. 1a). This extrusion occurred in a dose-dependent fashion based upon the concentration of external Na+ added, and it was already maximal within 1 min from the addition of NaCl irrespective of the concentration added, not changing significantly at later time points (Fig. 1a). Hence, for the remainder of the experiments showed here, the net amount of Mg2+ extrude at t = 5 min from bLPM following NaCl addition is reported for simplicity. The net amounts of Mg2+ extruded from bLPM vesicles at 5 min following the addition of varying concentrations of extravesicular NaCl are reported in Fig. 1b. To determine whether extracellular Cl− plaid any role in Mg2+ extrusion, Mg-loaded bLPM vesicles were stimulated by the extravesicular addition of varying concentrations of Na-isethionate. The net amount of Mg2+ extruded under these conditions accounted for by 33.43 ± 1.34, 85.17 ± 5.11, and 170.43 ± 11.59 nmol Mg2+/mg prot/5 min for vesicles stimulated by 1, 5, and 10 mM Na-isethionate (n = 6 for each experimental condition). These values were essentially similar to those that elicited by comparable doses of NaCl (Fig. 1b). Lastly, for the majority of the experiments reported here, an extravesicular concentration of 10 mM NaCl (or Na-isethionate) was used as this concentration is close to the calculated Km of ~14 mM for the Na+/Mg2+ exchanger [14, 25].

Fig. 1
figure 1

Mg2+ extrusion from tLPM and bLPM vesicles. Total (tLPM, a) and basolateral (bLPM, b) vesicles were loaded with MgCl2 and incubated for Mg2+ extrusion as indicated in the “Materials and Methods” section. After a few minutes of equilibration, varying doses of extravesicular NaCl were added to induce Mg2+ extrusion from the vesicles. Data are means ± S.E. of four different preparations, each tested in duplicate. *Statistically significant versus control value

Full size image

Under the experimental conditions described in Figs. 1 and 2, a release of 36Cl− from bLPM vesicles was observed (Fig. 2a). Also this extrusion was maximal within 3 min from the addition of extravesicular Na+, and occurred in a dose-dependent manner relative to the concentration of Na+ added extravesicularly. In vesicles stimulated with 10 mM Na-isethionate, however, the extrusion of 36Cl− was reduced by ~20–25% as compared to the extrusion occurring in vesicles stimulated with 10 mM NaCl (Fig. 2a). When normalized to the total 36Cl− content entrapped within the vesicles, the amount of Cl− extruded from the vesicles varied from ~12 to 40% of the total Cl− loading depending on the extravesicular concentration of NaCl added (Fig. 2b).

Fig. 2
figure 2

Cl− extrusion from bLPM vesicles. Basolateral (bLPM) vesicles loaded with MgCl2 plus 1 μCi 36Cl− were incubated for Cl− extrusion as indicated in the “Materials and Methods” section. The amplitude of Cl− extrusion following the extravesicular addition of varying NaCl concentrations is reported in a. The amplitude of Cl− mobilization as a percent of the amount entrapped within the vesicles is reported in b. Digitonin (50 μg/ml) was added after t = 4 min to induce the release of residual Cl− from the vesicles and assess homogeneity in vesicle loading. Data are means ± S.E. of five different preparations, each tested in duplicate. All the values of Cl− efflux (a) are statistically significant versus corresponding control values. Statistical labeling is omitted for simplicity. b All the values of Cl− efflux are statistically significant versus corresponding control values. Statistically significant labeling (*) is limited for clarity

Full size image

To determine whether intracellular Cl− was required for the Na+-dependent Mg2+ extrusion to occur, bLPM vesicles were loaded with 20 mM Mg-gluconate and the efflux was induced by addition of either NaCl or Na-isethionate. The results reported in Fig. 3 indicate that the amplitude of Mg2+ extrusion induced by Na-isethionate appears to be ~20% larger than that induced by an equivalent concentration of NaCl, but the trend did not attain statistical significance. The Mg2+ loading of bLPM was essentially similar irrespective of the moiety used (397.5 + 7.3 and 401.5 ± 9.1 nmol Mg2+/mg protein following loading with MgCl2 and Mg-gluconate, respectively, n = 9 for both experimental conditions). When normalized for the total Mg2+ loading, the amount of Mg2+ extruded by the vesicles varied from ~6 to 44% of the intravesicular Mg2+ content based on the concentration of external Na+ added irrespective of the nature of the Na+ moiety added (Table 1). Table 1 also reports the net amount of Na+ accumulated within bLPM vesicles loaded with either MgCl2 or Mg-gluconate as a function of the Na+/Mg2+ exchanger activity. When used to assess the stoichiometry exchange ratio of the Na+/Mg2+ antiporter, these values supported the operation of the exchanger on electrogenic bases in the presence of intravesicular Cl− (Mgout/Nain 1.10 + 0.003, Table 1) and on electroneutral bases in their absence (0.53 ± 0.003 in the presence of external Cl- and 0.46 ± 0.001 in their absence, Table 1). This change in stoichiometry ratio was confirmed by the absence of TPP+ accumulation within Mg-gluconate loaded bLPM vesicles following the addition of NaCl or Na-isethionate (Fig. 4).

Fig. 3
figure 3

Net Mg2+ extrusion from bLPM vesicles loaded with Mg-gluconate. Basolateral (bLPM) vesicles loaded with Mg-gluconate were stimulated for Mg2+ extrusion by addition of varying concentrations of NaCl or Na-isethionate as reported under “Materials and methods”. Net Mg2+ extrusion from the vesicles is reported. Data are means ± S.E. of four different preparations, each tested in duplicate. *Statistically significant versus control value

Full size image
Table 1 Changes in intra-bLPM Mg2+ and Na+ content upon loading with MgCl2 or Mg-gluconate and addition of extravesicular NaCl or Na-Isethionate
Full size table
Fig. 4
figure 4

TPP+ Distribution in bLPM vesicles. Basolateral (bLPM) vesicles loaded with MgCl2 or Mg-gluconate were stimulated for Mg2+ extrusion by addition of 25 mM NaCl or Na-isethionate as reported under “Materials and methods”. Panel A shows sequential addition of 125nM TPP+ aliquots and 25 mM NaCl in a system devoid of membrane (the addition of 25 mM Na-isethionate was equally ineffective). Panel B shows sequential addition of TPP+ and 25 mM NaCl or Na-isethionate to 20 mM MgCl2 (or 20 mM Mg-gluconate) loaded bLPM. Trace a Na-isethionate addition to Mg-gluconate-loaded vesicles, Trace b NaCl addition to Mg-gluconate-loaded vesicles. Trace c Na-isethionate addition to MgCl2-loaded vesicles, Trace d NaCl addition to MgCl2-loaded vesicles. Accumulation of TPP+ in traces c and d support the notion that the net amount of charges carried out by Mg2+ efflux in the presence of intracellular Cl− is greater than the net amount of charges carried into the vesicles by Na+ entry. A typical experiment out of six for each experimental conditions is reported

Full size image

We have reported that imipramine selectively inhibits the operation of the Na+/Mg2+ exchanger in bLPM [14]. Amiloride is also able to inhibit Na+-dependent Mg2+ extrusion, at least in cells [5, 18]. To confirm pharmacologically that the observed extrusion of 36Cl− occurred through the Na+/Mg2+ exchanger in a manner similar to that described by Rasgado-Flores [24] in squid axon, bLPM vesicles loaded with MgCl2 labeled with 36Cl− were stimulated by addition of extra-vesicular NaCl in the presence of amiloride or imipramine as inhibitors of the Na+-dependent Mg2+ extrusion mechanism, and niflumic acid or DIDS as non-specific inhibitors of Cl− transport. The results reported in Fig. 5a indicate that only imipramine was fully effective at inhibiting Na+-induced Mg2+ (Fig. 5a) and Cl− (Fig. 5b) efflux from bLPM vesicles. All the other agents tested were essentially ineffective at inhibiting Mg2+ and Cl− efflux (Figs. 5a, b) or actually slightly increased Cl− extrusion (niflumic acid, Fig. 5b). Furosemide and bumetanide were similarly ineffective at inhibiting Mg2+ (Fig. 5a) and Cl− (Fig. 5b) extrusion. Doses of inhibitors larger than those reported in the figure were not tested as the doses utilized were already maximal [14, 25]. Addition of digitonin at t = 4 min (i.e., 3 min after extravesicular Na+ addition) resulted in a marked loss of 36Cl− associated with the vesicles, which reached the same end-point irrespective of the experimental conditions (Fig. 5c). This observation would exclude that the reduced Cl− extrusion observed in the presence of imipramine could be attributed to a limited vesicle loading.

Fig. 5
figure 5

Effect of various inhibitory agents on Mg2+ and Cl− extrusion from bLPM vesicles. Baso-lateral (bLPM) vesicles loaded with MgCl2 plus 1 μCi 36Cl− were incubated for Mg2+ and Cl− extrusion in the presence of 1 mM amiloride, 250 μM imipramine, 250 μM niflumic acid, 50 μM DIDS, 250 μM furosemide, and 20 μM bumetanide as indicated in the “Materials and methods” section. The amplitude of Mg2+ (a) and Cl− extrusion (b) following the extravesicular addition of 10 mM NaCl are reported. The amplitude of Cl− mobilization as a percent of the amount entrapped within the vesicles is reported in c. Digitonin (50 μg/ml) was added after t = 4 min to induce the release of residual Cl− from the vesicles and assess similarity in vesicle loading. Data are means ± S.E. of five different preparations, each tested in duplicate. a and b *Statistically significant versus corresponding time points of samples incubated with other inhibitors. c All the values of Cl− efflux are statistically significant versus corresponding control values. Statistically significant labeling (*) is limited for clarity

Full size image

Discussion

The operation of a Na+/Mg2+ exchange mechanism has been observed in all mammalian and non-mammalian cells tested so far [2]. Because the exchanger has not been cloned as yet, all the available information about its presence and operation is mostly indirect and based on non-specific inhibitors or experimental approaches. Owing to these limitations, the stoichiometry of the exchanger remains poorly defined, with both electroneutral [18] and electrogenic [19, 20] ratios being reported.

This present study investigated the possible contribution of intra- and extra-vesicular Cl− to the Na+-induced Mg2+ extrusion across the hepatocyte cell membrane and its stoichiometry of exchange. For our study, we utilized purified bLPM vesicles as we have previously characterized the presence and operation of the putative Na+/Mg2+ exchanger in this subpopulation [20], and excluded the occurrence of an osmotic mismatch as the possible cause of Mg2+ extrusion [25].

The obtained results indicate that Na+-induced Mg2+ extrusion occurs to a comparable extent irrespective of the absence or the presence of intravesicular Cl−. This is at variance of what reported in red blood cells [23], in which Mg2+ extrusion was markedly stimulated only in the presence of intracellular Cl−. This difference may depend on the biophysical model and selective intra- and extra-vesicular medium used here as compared to erythrocytes artificially loaded with MgCl2. Furthermore, we observed that the addition of external NaCl elicits a marked extrusion of intravesicular entrapped Cl− into the extravesicular milieu. This extrusion persists even in the absence of extravesicular Cl− (Na-isethionate) although the rate of transport appears to be smaller than in the presence of external Cl− (NaCl). Discounting the OH− groups present in the system, no other anion moieties are present under our experimental conditions, and yet Cl− is extruded at a rate of ~30 nmol/mg protein/1 min (the earliest time point that we could reliably detect with our experimental approach).

The addition of niflumic acid and DIDS as inhibitors of Cl− or anion transport does not prevent Cl− extrusion from bLPM vesicles, suggesting that Cl− is not extruded via one of the classical Cl− pathways. The absence of K+ ions in our experimental system, required for the operation of the sodium/potassium/chloride co-transporter (NKCC1, SLC12A2), and the lack of an inhibitory effect by furosemide or bumetanide, two fairly specific inhibitors of the NKCC1 transporter, exclude the involvement of this transport mechanism in Cl− extrusion. On the other hand, the inhibition of Cl− extrusion by imipramine, which quite specifically inhibits the putative Na+/Mg2+ exchanger in bLPM vesicles [14, and data reported here] would suggest the direct involvement of this exchanger in Cl−mobilization. As hepatocytes possess outwardly rectifying Cl− channels [27], it is possible, at least in principle, that these channels become active following the operation of the Na+/Mg2+ exchanger. However, these Cl− channels as well as bestrophin channels [28], are usually inhibited by DIDS [28, 29], which does not inhibit Cl− extrusion under our experimental conditions. Moreover, the Cl− channels require ATP hydrolysis to be active [28, 29], and no ATP is present in our preparation. Lastly, the involvement of volume-activated Cl− channels [29, 30] can be excluded based on the absence of Ca2+ in our system, a cation strictly required for the activation of these channels [29, 30]. The absence of ATP and the inability of DIDS to inhibit Cl− efflux also exclude the operation of CFTR as this channels is regulated by both ATP and DIDS [29]. Having excluded all these alternative mechanisms, our results in the presence of DIDS or niflumic acid while not fully probative suggest that Cl− is extruded via the Na+/Mg2+ exchanger itself.

Historically, the Na+/Mg2+ exchanger has been reported to be inhibited by amiloride [5, 18]. Yet, this agent does not appear to inhibit the transporter in bLPM [14, and data reported here]. The lack of inhibition on both Mg2+ and Cl− extrusions under our experimental conditions could perhaps be ascribed to the presence of a specific amiloride-insensitive isoform of the Na+/Mg2+ exchanger in the hepatocyte basolateral domain. In fact, amiloride retains the ability to inhibit Mg2+ extrusion in purified apical liver plasma membranes [14]. As neither the apical or the basolateral Mg2+ extrusion mechanisms have been cloned, we are presently unable to explain at the structural level the difference(s) between these two transport mechanisms.

The ability of the Na+/Mg2+ exchanger to transport Cl− across the cell membrane has been suggested by early studies in erythrocytes [23] and in squid axon [24]. In red blood cells, the evidence provided by Gunther and collaborators is that intracellular Cl− plays a stimulatory role on the Na+-induced Mg2+ extrusion [23], but it remained undetermined whether Cl− was actually transported out of the cells together with Mg2+. In the case of the squid axon, the evidence provided by Rasgado-Flores and coworkers [24] suggests that Cl− ions are indeed transported by the Na+/Mg2+ exchanger in a manner similar to what occurs in NKCC1 transporter [28]. For experimental reasons, however, the transport in the dialyzed squid axon was performed forcing the exchanger to operate in the reverse configuration (Mg2+ entering the axon and Na+ exiting it). Hence, it could be argued that those results depended on the transporter operating in reverse-mode. A role of Cl− on Mg2+ efflux from intact hepatocyte has been reported by our laboratory [22]. The addition of increasing concentrations of extracellular Cl− (as choline chloride) to hepatocytes in suspension resulted in a dose-dependent Mg2+ extrusion from the cells, which was inhibited in the presence of DIDS [22]. At the time, the said study interpreted the Mg2+ extrusion as the consequence of extracellular Cl− exchanging for intracellular HCO3 − via the band 3 [22]. The loss of intracellular HCO3 − would then result in a cellular acidification that would alter the binding affinity of ATP and other cellular buffering components for Mg2+, and increase the free (unbound) Mg2+ concentration, which will be extruded in exchange for external Na+. Alternative possibilities supported by more recent observation could involve the activation of outwardly rectifying Cl- channels (ORCs) operating in combination with the Na+/Mg2+ exchanger to maintain cellular ion homoestasis [29, 31]. The involvement of the band-3 or alternative Cl- extrusion mechanisms, however, cannot be used to explain the results reported here, as the LPM experimental model does not include bicarbonate, and DIDS does not inhibit Mg2+ and Cl− extrusions. While increasing proportionally to the concentration of extravesicular NaCl added, even at its maximal amplitude (~30 nmol Cl−/mg protein/1 min) Cl− extrusion does not appear to be of a magnitude sufficient to complement the number of charges moved across the plasma membrane by the Na+/Mg2+ antiport. This would suggest that Cl− extrusion acts as a partial compensatory charge transport to maintain membrane potential and provide sufficient driving force for Mg2+ extrusion under physiological conditions. Whether protons are extruded from the vesicles for charge compensation via the Na/H antiport cannot be completely excluded despite the lack of an inhibitory effect by amiloride (a weak Na/H antiport inhibitor) on the amplitude of Mg2+ extrusion.

Lastly, based on the similar amplitudes of Mg2+ extrusion following the addition of NaCl or Na-isethionate, it appears that extravesicular (extracellular) Cl− ions do not play a detectable role on the operation of the Na+/Mg2+ exchanger, at least in our experimental model.

The Na+/Mg2+ exchanger is not unique in its modulation by extracellular and/or intracellular Cl−. A similar modulation has been reported for the Na+/H+ exchanger [32] and the NKCC transporter [33]. In the case of the Na+/H+ exchanger, removal of extracellular [34] and intracellular [35] Cl− is equally effective at inhibiting almost completely the transporter activity. Removal of intracellular Cl− also inhibits the operation of the NKCC transporter [33, 36]. All together, these data support a more general role of intracellular and extracellular Cl− in modulating various Na+-dependent transporters and non-selective cation channels in mammalian cells, with important repercussions on membrane potential, volume, and ion composition of the cell.

Conclusions

Taken together, our results and the inhibitory effect of imipramine are consistent with the notion that Cl− transport occurs via the Na+/Mg2+ exchanger, as proposed to occur in the squid axon by Rasgado-Flores and his collaborators [24]. Our observation suggests that significant changes in extra- and intra-cellular Cl− content may affect the rate of exchange of the putative Na+/Mg2+ exchanger, and impact both Na+ and Mg2+ homeostasis as well as the cell membrane potential.