INTRODUCTION

Extracellular ATP is a potent signaling factor that modulates a variety of cellular functions through the activation of P₂ purinergic receptors in the plasma membrane. These receptors are widely distributed among different liver cell types, including hepatocytes, cholangiocytes, macrophages, and endothelial cells, but the physiologic roles have not been fully defined. Cells release ATP in response to both osmotic and mechanical stimuli, and one mechanism may involve opening of a channel-like pathway (1, 2). In respiratory epithelia, ATP release stimulated by cytosolic cAMP activates outwardly rectified Cl channels coupled to P_2U receptors and enhances Cl secretion (3). Recent studies in a model liver cell line support an alternative pathway where increases in cell volume induce conductive ATP efflux. In these cells, removal of extracellular ATP or P₂ receptor blockade prevents both Cl channel activation and volume recovery (1). These findings suggest functional interactions between ATP release, P₂ receptor stimulation, and Cl channel opening in epithelial secretion and volume regulation.

Members of the ATP-binding cassette (ABC)¹ protein family are likely to be relevant to this volume regulatory pathway for two reasons. First, while the molecular basis for the transmembrane ATP conductance has not been established, heterologous expression or up-regulation of ABC family members in some cell models is associated with enhanced electrodiffusional ATP release. In cystic fibrosis respiratory epithelia, cAMP fails to stimulate channel-mediated ATP efflux, a response that is present in native epithelia; CFTR gene transfer restores the ATP conductance (3, 4). In other cell lines, ATP release is proportional to the expression of mammalian and Drosophila Mdr1 P-glycoproteins (5, 6). Second, in some but not all cell types, Mdr1 P-glycoproteins regulate swelling-activated Cl currents (I_Cl-swell). Effects include enhancement of I_Cl-swell and endowment of Cl channel sensitivity to protein kinase C (mdr1 gene transfer) and increase in I_Cl-swell for a given hypotonic stress (P-glycoprotein overexpression) (7, 8). The cellular mechanisms involved in these responses and the implications for other cell types have yet to be clarified.

In hepatocytes, P-glycoproteins transport both amphipathic compounds and phospholipids across canalicular membranes into bile (9, 10). However, the functions of multiple other ABC members present in liver cells are unknown. In light of the putative association of certain ABC proteins with channel-mediated ATP and Cl transport, we sought to investigate the role of hepatocellular ABC proteins in these processes. Findings in rat HTC hepatoma cells were compared with those in a selected population of HTC cells (HTC-R) that overexpress both endogenous and novel Mdr proteins (11). These studies demonstrate that inhibition of P-glycoprotein transport prevents recovery from swelling and that overexpression of Mdr proteins is associated with enhanced ATP release, volume recovery, and cell survival during hypotonic stress. Therefore, by modulating ATP transport, one or more hepatic ABC proteins may represent important regulatory elements that directly influence numerous volume- and ATP-dependent liver cell functions.

EXPERIMENTAL PROCEDURES

The purinergic agonists used were ATP, UTP, and ATPS, and the purinergic antagonists were suramin and reactive blue 2. The ATP scavengers used were apyrase and hexokinase. The P-glycoprotein blockers used were cyclosporin A and verapamil, and the Cl channel blocker was 5-nitro-2-(3-phenylpropylamino)benzoic acid. The Ca²⁺ chelator used was EGTA. The cell viability fluorochrome was propidium iodide. All experimental reagents were obtained from Calbiochem or Sigma and were exposed to cells or cell suspensions as described.

HTC cells possess ion channels and purinergic receptors similar to those found in primary hepatocytes and exhibit rapid recovery from swelling that is mediated by the opening of Cl channels (12, 13). HTC-R cells represent specially adapted HTC cells originally selected to resist normally toxic concentrations of glycocholic acid methyl ester. HTC-R cells have special advantages for these studies since they overexpress both endogenous and novel Mdr proteins (11). Both cell types were grown at 37 °C in a 5% CO₂ and 95% air atmosphere in minimal essential medium (Life Technologies, Inc.) containing 5% fetal calf serum, 2 mM L-glutamine, 100 IU/ml penicillin, and 100 µg/ml streptomycin. Glycocholic acid methyl ester was synthesized as described previously (11) and added to the HTC-R cell medium at concentrations of 300-650 µM. For all physiologic studies, the culture medium was replaced with standard isotonic extracellular buffer that contained 140 mM NaCl, 4 mM KCl, 1 mM KH₂PO₄, 2 mM MgCl₂, 1 mM CaCl₂, 10 mM glucose, and 10 mM HEPES/NaOH (pH 7.4). Hypotonic buffer was made by decreasing the NaCl concentration by 20-40% (84-112 mM) as indicated. As determined by a vapor pressure osmometer, the isotonic extracellular buffer osmolarity was ~295 mosmol/kg, and the hypotonic buffer (40% decrease in NaCl) osmolarity was ~205 mosmol/kg.

The mean cell diameter was measured in cell suspensions at 25 °C by electronic cell sizing utilizing the Coulter Multisizer, Coulter Sample Stand II, and Multisizer Accucomp Version 2.0 software (Coulter Electronics, Inc., Hialeah, FL) as described previously (14). Briefly, the mean cell volume of ~10,000 cells/time point was calculated from the mean cell diameter using the formula 4/3r³. Changes in size over time are expressed as relative volume by normalizing volume to the basal period. Any significant alterations in basal cell volume by experimental reagents are noted. Percent recovery for a given time point was calculated as follows: ((peak relative volume  relative volume at that time point)/(peak relative volume  1)) × 100. 

Measurement of Cl Currents

Whole-cell currents were measured ~24 h after plating using patch-clamp recording techniques (15). Cells on a coverslip were placed in a chamber (~400 µl) and perfused at 4-5 ml/min. Swelling was induced by switching to a buffer containing 20% less NaCl. The standard pipette (intracellular) solution contained 130 mM KCl, 10 mM NaCl, 2 mM MgCl₂, 10 mM HEPES/KOH (pH ~7.2), and free Ca²⁺ adjusted to ~100 nM (0.5 mM CaCl₂ and 1 mM EGTA) with a total Cl concentration of 145 mM (16). Cells were viewed through an Olympus IMT-2 inverted phase-contrast microscope using Hoffman optics at a magnification of ×600. Patch pipettes were pulled from Corning 7052 glass and had a resistance of 3-10 megaohms. Recordings were made with an Axopatch ID amplifier (Axon Instruments, Inc., Foster City, CA) and were digitized (1 kHz) for storage on a computer and analyzed using pCLAMP Version 5.5 programs (Axon Instruments, Inc.). Pipette voltages refer to the bath. Current-voltage relations were measured by 400-ms steps to potentials between 120 and +100 mV in 20-mV increments. In the whole-cell configuration, the pipette voltage corresponds to the membrane potential, and upward deflections of the current trace indicate outward membrane current. Whole-cell currents (pA) and current density (pA/pF) refer to measurements at a pipette voltage of 80 mV (E_K) to minimize any contribution of K⁺ currents.

Under physiologic conditions, ATP currents are below the limits of detection (3). To enhance measurements of ATP transport, whole-cell recordings were performed with high ATP concentrations in both the bath and pipette solutions including 100 mM MgATP, 5 mM MgCl₂, and 10 mM HEPES/Tris-OH (pH 7.4) as described previously (1, 4, 5). Swelling was induced by 20% dilution of the bath solution (1 ml) with water (200 µl/ml), and currents were measured before and 5 min after dilution.

Cellular toxicity was assessed by propidium iodide fluorescence as described previously (17). Propidium iodide (50 µg/ml) is thought to penetrate only damaged membranes to form intercalation complexes with double-stranded DNA with an amplification of nuclear fluorescence (excitation at 530 nm with a 620-nm emission filter). In wells containing subconfluent HTC and HTC-R cell monolayers, the number of positively staining nuclei was quantitated at timed intervals during incubation in isotonic and hypotonic buffers.

Pooled data are presented as means ± S.E. or S.D., where n represents the number of cell suspensions analyzed by the Coulter Multisizer, the number of cells for patch-clamp studies, and the number of wells for propidium iodide studies. Statistical comparisons were made using the paired or unpaired t test as indicated, and p < 0.05 was considered significant.

RESULTS

ATP Regulates Swelling-activated Cl Currents in HTC-R Cells

In HTC cells, recovery from swelling depends on channel-mediated ATP release, stimulation of membrane P₂ receptors, and activation of Cl currents (I_Cl-swell) (Fig. 1) (1). Whole-cell patch-clamp and cell-sizing studies were initially performed to confirm that a similar purinergic signaling pathway contributes to volume regulation in HTC-R cells, a model derived from HTC cells that overexpresses ABC proteins (11).

In whole-cell recordings of HTC-R cells, basal Cl currents in isotonic buffer were small. However, incubation in hypotonic buffer (20% less NaCl) resulted in an increase in Cl currents from 219.7 ± 38.8 to 3938.8 ± 389.4 pA (n = 13) (Fig. 2, A and D). Removal of extracellular ATP by the addition of the ATPase/ADPase apyrase (1 unit/ml) to hypotonic buffer significantly inhibited activation of I_Cl-swell (201.8 ± 44.8 pA, n = 3; p < 0.001). This effect was reversible upon return to apyrase-free solutions (Fig. 2, B-D). The addition of the nonselective P₂ blockers suramin (100 µM) and reactive blue 2 (10 µM) to hypotonic buffer also inhibited current activation (suramin: 380.9 ± 118.6 pA, n = 4; reactive blue 2: 166.0 ± 48.9 pA, n = 3; p < 0.001), confirming that the effects of extracellular ATP involve stimulation of membrane P₂ receptors (Fig. 2, C and D). Exposure to ATP (2.5-10 µM) in the absence of swelling markedly increased Cl currents with similar characteristics to I_Cl-swell (data not shown). The P₂ receptor subtype involved is unknown since UTP (a P_2U agonist) at similar concentrations did not activate Cl currents, and in previous studies in HTC cells, neither P_2Y-preferring (2-MeSATP) nor P_2X-preferring (,-MeATP) agonists stimulated channels (18).

Under isotonic conditions, HTC-R mean cell diameter was 23.81 ± 0.164 µm, corresponding to a mean cell volume of 7068 ± 147 µm³ (n = 10). Notably, the average basal volume for HTC-R cells is 238 ± 6% that of HTC cells (2958 ± 112 µm³, n = 10; p < 0.001). As shown in Fig. 3A (Control), exposure to hypotonic buffer (40% less NaCl) caused a rapid increase in values by 2 min (mean cell diameter = 25.93 ± 0.21 µm; mean cell volume = 9132 ± 219 µm³), which corresponds to a relative cell volume of 1.29 ± 0.03 (n = 10; p < 0.001). Swelling was followed by a regulatory volume decrease (RVD) to a relative volume of 1.07 ± 0.02 by 30 min. The anion channel blocker 5-nitro-2-(3-phenylpropylamino)benzoic acid (50 µM; no preincubation) completely inhibited volume recovery (p < 0.001), confirming that RVD is dependent on channel-mediated Cl efflux (Table I).

Table I. Percent volume recovery for HTC and HTC-R cells The effects of anion channel blockade (5-nitro-2-(3-phenylpropylamino)benzoic acid), extracellular ATP removal (apyrase and hexokinase), P2 receptor blockade (suramin and reactive blue 2), and P-glycoprotein inhibition (verapamil) on volume recovery were evaluated. HTC and HTC-R cell volume changes during incubation in hypotonic buffer (40% less NaCl) for 30 min were determined with a Coulter Multisizer. As a measure of volume recovery, percent RVD was calculated as follows: ((peak relative volume at 2 min  relative volume at 30 min)/(peak relative volume 1)) × 100. Except for ATPS, which was added 5 min after exposure to hypotonic buffer, all experimental agents were added to hypotonic buffer at time 0. Values represent means ± S.D. and n  3 for all experiments. Statistical significance was determined between control and experimental values. Experimental Condition HTC cells HTC-R cells RVD p Value RVD p Value % % Control 64  ± 6a 78  ± 7b NPPBc (50 µM) 0  ± 6 <0.001  8  ± 8 <0.001 Apyrase (3 units/ml) 13  ± 6a <0.001 24  ± 8 <0.001 Apyrase (3 units/ml) + ATPS (25 µM) 57  ± 9a NS Hexokinase (1 units/ml) 29  ± 8a <0.001 Suramin (100 µM) 33  ± 5a <0.001 15  ± 6 <0.001 Reactive blue 2 (100 µM) 33  ± 7a <0.001 Verapamil (20 µM) 19  ± 11 <0.001 a Data were previously published (1). b Volume recovery at 30 min was significantly greater for HTC-R cells compared with HTC cells (p < 0.01). c NPPB, 5-nitro-2-(3-phenylpropylamino)benzoic acid; NS, not significant.

In HTC cells, volume recovery is abolished by (i) removal of extracellular ATP by the ATP scavengers apyrase and hexokinase, an effect that is reversed by exposure of the nonhydrolyzable P₂ agonist ATPS (25 µM); and (ii) concentration-dependent P₂ receptor blockade with suramin and reactive blue 2 (Table I) (1). Similarly, in HTC-R cells, apyrase (3 units/ml; no preincubation) reduced percent RVD at 30 min from 78 ± 7% (controls) to 24 ± 8% (p < 0.001) (Table I and Fig. 3A). In the presence of suramin (100 µM), HTC-R cell swelling was more pronounced (28 ± 7% increase compared with controls at 2 min; p < 0.001), and percent volume recovery was significantly inhibited (15 ± 6% at 30 min; p < 0.001) (Table I and Fig. 3A). These findings suggest that P₂ receptor stimulation by extracellular ATP mediates Cl current activation and volume recovery during HTC-R cell swelling.

Same-day Coulter studies were performed on both cell populations to assess whether overexpression of ABC proteins is associated with a difference in volume recovery rates. As shown in Fig. 3B, volume recovery was significantly faster for HTC-R cells compared with HTC cells. Enhanced recovery was most notably early in the volume recovery phase; percent RVD at 5 min was 32 ± 8% for HTC-R cells versus 5 ± 9% for HTC cells (p < 0.001). By 30 min, HTC-R cell volume was closer to basal values (relative volumes of 1.06 ± 0.024 and 1.12 ± 0.023 for HTC-R and HTC cells, respectively; p < 0.001). These findings were unexpected since the lower surface/volume ratio should favor a slower volume recovery phase in the larger cells (HTC-R) and suggested that factors responsible for RVD, such as ATP and/or Cl efflux, may be up-regulated in these cells.

To establish whether volume regulation is Mdr protein-dependent, additional Coulter studies were performed in which verapamil, an inhibitor of P-glycoprotein-mediated substrate transport, was added to hypotonic buffer. As shown in Fig. 3C, verapamil (20 µM) markedly attenuated HTC cell volume recovery. The inhibitory effects was apparent at all time points beyond 2 min, and percent volume recovery at 30 min was 19 ± 11% in the presence and 64 ± 6% in the absence of verapamil (p < 0.001) (Table I). These findings suggest that Mdr proteins may contribute directly to the cellular mechanisms that regulate volume recovery.

Inhibition of P-glycoprotein Transport Prevents Activation of Cl Currents

Since Mdr1 P-glycoproteins have been implicated in the regulation of I_Cl-swell in other cell types, an increase in Cl permeability could explain the enhanced rate of volume recovery in HTC-R cells. Therefore, the effects of verapamil and cyclosporin A, another P-glycoprotein inhibitor, on HTC cell I_Cl-swell were assessed. In control studies, exposure of HTC cells to hypotonic buffer (20% less NaCl) led to a marked increase in whole-cell Cl currents from 81.1 ± 24.4 to 1697.0 ± 121.2 pA (n = 13) (Fig. 4, A and B) as described previously (1, 13). The addition of verapamil (20 µM) to extracellular buffer 1-2 min prior to hypotonic challenge abolished activation of I_Cl-swell (86.9 ± 15.2 pA, n = 5; p < 0.001) (Fig. 4, A and B). This inhibitory effect was reversible upon removal of verapamil from the bath solution. In similar studies, cyclosporin A (15 µM) also prevented current activation (151.1 ± 62.5 pA, n = 5; p < 0.001) (Fig. 4, A and B).

Examination of Fig. 1 suggests several potential sites for action of verapamil and cyclosporin A on the Cl conductance, including direct effects on Cl channels themselves and/or indirect effects via inhibition of ATP release. The former possibility is unlikely since swelling-induced Cl current density was not different between HTC and HTC-R cells (approximately 50 pA/pF; not significant) (Fig. 5B). Moreover, in the presence of verapamil (20 µM), the addition of 10 µM ATP to the bath solution led to a marked increase in HTC whole-cell Cl currents, which were indistinguishable from those activated by ATP in the absence of verapamil (Fig. 4C). These findings strongly suggest that verapamil is not functioning as a Cl channel blocker and that Mdr proteins may therefore regulate a more proximal component in the signaling cascade, such as ATP release (Fig. 1).

4

4

4

4

Enhanced HTC-R cell volume recovery may be related to an up-regulation of purinergic signaling processes secondary to increased availability of ATP. To characterize the ATP conductance, whole-cell studies were performed with ATP⁴ as the major charge carrier (1). Compared with HTC cells, the basal ATP conductance was greater in HTC-R cells (2.7 ± 0.5 versus 0.6 ± 0.21 pA/pF; p < 0.01). Exposure to hypotonic buffer (20% less NaCl) markedly increased HTC-R cell ATP currents to 46.2 ± 8.3 pA/pF. Notably, swelling-induced ATP current density was ~3-fold greater in HTC-R cells compared with HTC cells (16.1 ± 10.4 pA/pF; p < 0.001) (Fig. 5A).

These findings indicate that HTC-R cells exhibit enhanced ATP release and more efficient volume homeostasis. To address the physiologic significance of this response, additional studies examined HTC and HTC-R cell viability with propidium iodide (50 µg/ml) at timed intervals during a 24-h exposure to either isotonic (control) or hypotonic buffer (Fig. 6). HTC and HTC-R cell death during incubation in isotonic buffer did not differ (<1% at 0-12 h and <3% at 24 h; not significant), and viability during hypotonic exposures 6 h was similar to control values. In contrast, longer hypotonic exposures increased cell death compared with controls (87 and 79% increases for HTC and HTC-R cells at 24 h, respectively). In addition, HTC cell death during incubation in hypotonic buffer for 12 and 24 h was significantly greater compared with HTC-R cells (cell death at 12 h: 3.2 ± 0.55% for HTC cells and 1.14 ± 0.48% for HTC-R cells; at 24 h: 19 ± 2.20% for HTC cells and 11.5 ± 2.32% for HTC-R cells; p < 0.001). These findings suggest that enhanced ATP release confers a selective survival advantage during prolonged periods of osmotic stress.

DISCUSSION

Hepatocytes release ATP in response to multiple stimuli, including increases in cell volume, extracellular adenosine, and mechanical stress (1, 2, 19). Transmembrane ATP efflux occurs through a channel-like pathway in accordance with an electrochemical gradient that favors movement into the extracellular space. In rat hepatoma cells, ATP functions as a local autocrine agonist that regulates cell volume by opening Cl channels coupled to membrane P₂ receptors (1). The present studies provide evidence that ABC proteins play a important role in volume-sensitive ATP transport. First, overexpression of Mdr proteins is associated with an increase in ATP conductance. Second, inhibition of P-glycoprotein transport eliminates volume recovery and activation of I_Cl-swell. Finally, the effects of P-glycoprotein inhibition are bypassed by exogenous ATP. Enhanced ATP efflux is associated with more efficient volume regulation and enhanced cell survival during hypotonic stress.

Liver cells are localized in a critical anatomic site where they receive dual perfusion from the systemic and portal circulations. Consequently, they are subject to unusually large osmotic demands related to high rates of protein synthesis, gluconeogenesis, and organic solute transport (20). Moreover, varying exposure to nutrients, hormones, and oxidative stress between the fed and fasted states leads to dramatic changes in ion transport at rates up to 10¹⁰ ions/cell/s. Despite this, conductive efflux of solutes such as Cl, K⁺, and taurine maintains volume within a narrow physiologic range (13, 21, 22). Extracellular ATP, which is present in liver circulation, stimulates membrane Cl channels similar to those activated by swelling. Furthermore, both ATP and decreases in cell volume have catabolic effects, presumably related to modulation of gene and protein expression (19, 20, 23, 24). Based on recent studies implicating volume-sensitive ATP release in recovery from cell swelling, identification of the cellular mechanisms involved in ATP efflux represents an important site for regulation of liver cell and organ function.

Several observations indicate that volume-sensitive ATP efflux is mediated by ABC proteins. First, both verapamil and cyclosporin A abolish swelling-induced Cl current activation, and verapamil inhibits volume recovery during hypotonic stress. It is notable that the effects of verapamil on Cl currents are reversed by exogenous ATP. Second, overexpression of ABC proteins in HTC-R cells is associated with an exaggerated rate of volume recovery. Finally, volume-activated ATP conductance is markedly enhanced in HTC-R cells.

Both verapamil and cyclosporin A have been shown to inhibit P-glycoprotein-mediated transport across hepatocyte membranes in multiple model systems (25-33). The effects of verapamil are not likely secondary to changes in intracellular Ca²⁺ since verapamil at the doses used does not change intracellular Ca²⁺ in isolated hepatocytes or alter passive Ca²⁺ transport across membrane vesicles (34, 35). Cyclosporin A also inhibits P-glycoprotein transport, but through a different mechanism by binding to canalicular membrane P-glycoproteins; induction of intrahepatic cholestasis in patients taking this drug may be related to attenuation of canalicular P-glycoprotein-mediated substrate transport (31, 36).

The specific ABC proteins involved in hepatic ATP trafficking are not known. The functions of several hepatocellular P-glycoproteins have recently been elucidated, including transport of amphipathic compounds (rodent mdr1a/1b and human MDR1) and phospholipids (murine mdr2 and human MDR3) across canalicular membranes into bile (9, 10). However, hepatocytes and other cells appear to possess multiple additional ABC proteins. In the liver, these novel proteins have been implicated indirectly in the transport of bile acids and hormones (11, 30). Consequently, it will be important to further evaluate the role of ABC proteins whose endogenous functions have not yet been determined.

Characterization of other important aspects of this signaling pathway will require further study. Most important, the mechanism(s) whereby ABC proteins such as P-glycoproteins and CFTR modulate ATP release, such as directly conducting ATP or regulating other as yet unidentified ATP efflux pathways, is presently unclear (37, 38). In addition, studies in isolated cells do not allow direct conclusions regarding polarity of membrane receptors and channels involved. Finally, the intracellular signals that regulate P-glycoprotein function and ATP transport have not yet been determined.

Since physiologic changes in liver cell volume are commonplace, it is interesting that these cells literally invest energy, in the form of ATP, on such an elaborate regulatory mechanism. In addition to activating conductive efflux pathways that maintain cellular integrity, the hydration state in itself directly alters cellular metabolism, solute transport, and gene expression (20). While these effects target the metabolic balance of hepatocytes, local release of ATP may also act as a paracrine signaling factor by modulating membrane transport in other cell types within the liver. For example, while increased bile formation during hepatocellular swelling has been attributed as least in part to enhanced canalicular bile acid transport, it is attractive to speculate that volume-sensitive ATP release into the bile duct lumen also contributes by stimulating secretion at the ductular level. Support for this model includes evidence that (i) ATP is present in physiologic concentrations in human bile; (ii) hypotonic perfusion of rat livers increases bile ATP levels; and (iii) stimulation of apical P₂ receptors in cholangiocytes (IC₅₀ ~ 2 µM) induces transepithelial Cl and fluid secretion across biliary cell monolayers (39, 40).

ATP transport also represents a potential pathway that couples the cellular metabolic state to membrane ion permeability. Swelling during anoxic depletion of cytosolic ATP is associated with loss of hepatocyte viability (41, 42). Volume recovery mediated by conductive ion efflux, which depends on ATP release, reduces cell death during these conditions (43, 44). The survival advantage of HTC-R cells during prolonged hypotonic stress may be related to enhanced ATP-dependent volume regulation. Admittedly, the role of other factors, such as larger cell size, cannot be discounted.

In conclusion, extracellular ATP has emerged as an important signaling factor capable of influencing volume-related and other cellular functions. ATP transport represents an exciting potential addition to the expanding list of substrate transport pathways regulated by ABC proteins in liver. Moreover, these findings may be relevant to other epithelia that possess ABC proteins and ATP-sensitive Cl channels as well. Further characterization of the cellular and molecular mechanisms involved in ATP transport will hopefully aid in the development of pharmacological approaches that favorably impact on bile formation and hepatocellular function.