Hormonal Stimulation of Mg2' Uptake in Hepatocytes REGULATION BY PLASMA MEMBRANE AND INTRACELLULAR ORGANELLES*

Collagenase dispersed rat liver hepatocytes release Mg2+ when stimulated with norepinephrine or accumulate Mg2+ when stimulated with vasopressin, respectively. M 6 + fluxes in either direction account for a net loss or gain of approximately 10% of total cell magnesium and are rapidly reversible. Both stimulated Mg2+ efflux and Mgz+ influx require physiological concentration of extracellular NaCl and Caz+. In the absence of extracellular Na+, Mg2+ efflux, but not influx, can be observed in the presence of extracellular C1-. Under these conditions, the efflux is inhibited by the CI-/HCO; exchanger inhibitor 4,4'-dinitrostilbene-2,2'-disulfonic acid. In hepatocytes, Mg2+ influx, but not efflux, is completely inhibited by thapsigargin, a specific inhibitor of the endoplasmic reticulum Ca2+ ATPase. Several lines of evidence, such as measurements of cytosolic Ca2+ or of cytosolic Ca2+ buffering, indicate that the effect of thapsigargin in inhibiting Mg2+ influx could not be explained by an increase in cytosolic Ca2+. In- stead, the inhibition of hepatocyte Mg2+ influx was be the result of the depletion of the Ca2+ stored within the endoplasmic reticulum.

In hepatocytes, Mg2+ influx, but not efflux, is completely inhibited by thapsigargin, a specific inhibitor of the endoplasmic reticulum Ca2+ ATPase. Several lines of evidence, such as measurements of cytosolic Ca2+ or of cytosolic Ca2+ buffering, indicate that the effect of thapsigargin in inhibiting Mg2+ influx could not be explained by an increase in cytosolic Ca2+. Instead, the inhibition of hepatocyte Mg2+ influx was found to be the result of the depletion of the Ca2+ stored within the endoplasmic reticulum.
Magnesium is one of the most abundant cations within mammalian tissues and cells (1)(2)(3)(4). Recently, there has been a surge of interest in the role of MgZ+ in preventing acute and chronic metabolic derangements in the human body. Highly publicized findings of the role of extracellular magnesium in preventing ischemic damage to tissue have been reported. These observations have been paralleled by findings that an increased number of cellular enzymes, channels and receptors are regulated by changes in concentrations of Mg2' (1-4).
All these observations are hampered in terms of physiological relevance by the limited intellectual framework presently available on the regulation of intracellular and extracellular M e homeostasis. Both intra-and extracellular Mg2' homeostasis are far less understood than that of other cations such as H+, CaZ+, Na+, or K' . The single main reason for such limited knowledge is the fact that available techniques for measuring free or total magnesium within cells or in extracellular fluid are less developed than those available for measuring other cations.
The interest in the regulation of intracellular Mg2+ has *This research was supported by National Institutes of Health Grant HL 18708. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. dressed.
3 To whom correspondence and reprint requests should 'be ad-grown significantly in recent years. A large number of recent observations have shown that under physiological (5-8) and pathological (9-13) conditions, a major redistribution of Mg2' across cells and organelles occurs in liver (14)(15)(16)(17)(18)(19), heart (5,12,(18)(19)(20), and other tissues (6,21,22). For instance, this and other laboratories have shown that a net Mg2+ efflux can be induced in perfused organs or isolated cells upon noradrenergic stimulation and consequent increase in cell cAMP (18,19). It was further demonstrated that some of this efflux can be accounted for by a release of M e from mitochondria induced by cAMP interacting with the adenine nucleotide translocase (16). Another recent observation showed that either vasopressin or carbachol, via stimulation of protein kinase C, can induce a massive uptake of Mg2+ into hepatocytes through a pathway that remains to be elucidated (15).
Even less well understood is the mechanism through which various cell types handle large net fluxes of Mg2+ across their plasma membranes. Evidence has been provided in the literature that extracellular Na' is important to permit Mg2' efflux in mammalian (17,20,21) and non-mammalian cells (1,6,23,24).
In this study, we investigated the role of extracellular Na+, and of other cations, in regulating not only Mg2+ efflux but also Mg2' influx in liver cells. We also investigated cellular mechanisms responsible for the recently observed hormonal regulation of cellular M e influx. Evidence is provided that the endoplasmic reticulum, and its Ca2+ storage properties within hepatocytes, is necessary for the observed stimulated M$+ influx but not M e efflux.

EXPERIMENTAL PROCEDURES
Isolated Hepatocytes-Collagenase dispersed hepatocytes were prepared from male Sprague-Dawley rats (200-250 g), according to the procedure of Seglen (25). After isolation, cells were resuspended in a buffer containing (in mM): 120 NaC1,3 KCl, 1 KH2P04, 1.2 MgC12, 1 CaC12, 10 glucose, 12 NaHCOa, 10 HEPES (pH 7.2 at 37 "C) in the presence of 02:CO2 (95:5 v/v). Cell viability (88 * 2%, n = 81, assessed by the trypan blue exclusion test, did not change appreciably during the first 5 h after isolation. For measuring Mg2' movements, hepatocytes (300-350 pg/ml protein) were incubated in the medium described above, in the absence of added MgZ' (?+@+-free buffer). At selected times, aliquots of the incubation mixture were withdrawn and quickly sedimented in microcentrifuge tubes. The supernatant was removed and the Mg2+ content determined by atomic absorbance spectrophotometry using a Varian "575 instrument. The MgZ' present as contaminant in the M$+-free buffer ranged between 2 and 5 JLM as determined by atomic absorbance spectroscopy. In the experiments performed to evaluate the Na' dependence of M$+ movements, the Na+ concentration was decreased to 60 or to 0 mM. In both cases Na+ was isosmotically replaced with choline or N-methyl-o-glucamine. No changes in hepatocyte viability were observed during the first 15 min of incubation with either of the Na+ replacement buffers (87 -t 4%, n = 6).
In another set of experiments, hepatocytes were incubated in a Mg2+ Transport in Hepatocytes Nu+-free buffer having the same composition as the M$+-free buffer described above, but with NaCl isosmotically replaced with choline chloride. At the times indicated in the figures, 120 mM NaCl (final concentration) was added to the incubation system by diluting the reaction mixture with an identical volume of isosmotic Mg2c-free buffer containing 240 mM NaCI, or different concentrations of sodium isothionate, previously equilibrated at pH 7.2 and 37 "C. An identical procedure was used to investigate the role of extracellular C1-on M e movements. For these experiments, hepatocytes were incubated in 200 mM sucrose, 20 mM HEPES, Tris (pH 7.4 at 37 "C). At the times indicated in the figures, the final concentrations of C1-were obtained by diluting the reaction mixture with an identical volume of buffer containing twice the desired C1-concentration, previously equilibrated at the same pH and temperature.
Loading of the Hepatocytes with BAPTA-After isolation, hepatocytes were resuspended, at the final concentration of 10' cells/ml, in the medium, reported under Isolated Hepatocytes, containing 1.2 mM MgC1, and loaded for 30 min with 5 p~ BAPTA-AM' at room temperature. After loading, hepatocytes were washed 3 times with the same medium and incubated in the M$+-free buffer for measuring Mg2+ movements. Mg2c determinations were carried out as previously reported.
Loading of the Hepatocytes with Fura-2-After isolation, hepatocytes were resuspended, at the final concentration of 10' cells/ml, in the resuspension medium containing 1.2 mM MgC1, and loaded for 20 min with 2 p~ Fura-2-AM at room temperature. After loading, hepatocytes were washed 3 times with the same medium and incubated in the Mg2c-free buffer, in order to detect cytosolic CaZ+ movements. Calibration was performed at the end of the experiments as reported elsewhere (26).
Other Techniques-In all of the procedures, protein amount was determined with the Bradford technique (27). Alternatively, hepatocytes were counted in a hemacytometer. In all of the preparations tested (n = 5), the protein concentration was found to be 2.7 f 0.3 mg of protein/million cells.
Chemicals-Collagenase (CLS-2) was from Worthington (Freehold, NJ). Thapsigargin was from LC Services Corp. (Woburn, MA). BAPTA-AM and Fura-2-AM were from Molecular Probes (Eugene, OR). All other chemicals and reagents were from Sigma. Hepatocytes-Fig. lA shows that the addition of 10 p~ NE t o a suspension of collagenase dispersed hepatocytes induces the release of 6 nmol of M e / million cells during an 8-min period, consistent with previously reported data (15,16,18). The addition of TPA, 4 min after the treatment with NE, results in a reversal of the release and in a large, time-dependent Mg2+ accumulation by the hepatocytes.  Fig. 1 ( A and B ) were also observed when forskolin or vasopressin were added instead of NE or TPA, respectively (not shown). These data indicate that either a large M%+ influx or efflux could be observed in hepatocytes upon the addition of two classes of agonists and that this stimulation was rapidly reversible.

Magnesium Transport in
Recently, Gunther and co-workers (17) provided evidence that a Na+/M%+ exchanger regulates M$+ efflux from liver cells. This pathway, blocked by amiloride (17), appears to be also located in the plasma membrane of cardiac cells (20), chicken erythrocytes (23), and thymocytes (21). The next set of experiments was designed to determine if this exchanger is involved not only in the CAMP-mediated M P efflux but also The abbreviations used are: BAPTA, 1,2-bis(2-aminophen-0xy)ethane-N,N,N',N'-tetraacetic acid; ER, endoplasmic reticulum; TPA, 12-0-tetradecanoylphorbol-13-acetate; NE, norepinephrine; DNDS, 4,4'-dinitrostilbene-2,2'-disulfonic acid. in the TPA (or vasopressin)-induced M$+ influx. Fig. 2 shows that both the NE-induced M%+ efflux and the TPA-stimulated Mg2+ uptake require the presence of a "physiological" concentration of NaCl(l20 mM) in the extracellular medium. In fact, both the efflux and the influx of M F , which were maximal in the presence of 120 mM extracellular NaC1, were consistently decreased, or totally abolished, when cells were incubated in the presence of 60 or 0 mM NaCl, respectively. Under these experimental conditions, Na+ was isosmotically replaced with choline and no changes in cell viability were observed during the first 10 min of Na+ replacement.
The same results were also observed when the cells were stimulated with concentrations of N E or TPA higher than those reported in Fig. 2 or by other stimulatory agents ( i e . forskolin or vasopressin; data not shown), suggesting that the decrease, or the absence, of stimulation is not attributable to a reduced effectiveness of these agents.
Additional evidence that M%+ movements require extracellular physiological NaCl concentrations is provided by the experiments shown in Fig. 3. Hepatocytes incubated in the absence of external NaCl and stimulated by NE or TPA neither released (Fig. 3A) nor accumulated M e (Fig. 3B).
However, the addition of 120 mM NaCl after 4 min of incubation fully restored both Mg2+ efflux (Fig. 3A) and M%+ influx (Fig. 3B). In the absence of NE or TPA stimulation, the addition of 120 mM NaClper se did not induce any change in M e movements (not shown). Fig. 4 shows the results of two experiments where the effect of changes in extracellular Na+ or C1-on M g + efflux or M e uptake was investigated. Hepatocytes were incubated in the M$+-free medium described under "Experimental Procedures," where NaCl content was isosmotically replaced with sucrose, and were stimulated with 10 FM NE or with 20 nM vasopressin. Fig. 4 ( A and B ) shows that in this medium the addition of either agonist had no effect on M g + efflux or influx in hepatocytes during the first 6 min. Fig. 4A shows that the addition of sodium isothionate caused an efflux of M$+ in cells previously stimulated with NE and an uptake in cells previously stimulated with TPA. In contrast, Fig. 4B shows that when C1-was added to stimulated cells in the form of choline chloride M e efflux, but not M e uptake, was restored.
These experiments indicate that the presence of extracellular Na+ is necessary for both stimulated Mg2+ efflux and uptake. On the other hand, extracellular C1-alone is sufficient for M e efflux, but not for M$+ uptake.
The effective concentrations of C1-necessary to induce M e efflux after addition of NE were further investigated by incubating the hepatocytes in the M$+-free buffer reported under "Experimental Procedures." Cells were stimulated with various concentrations of choline chloride, under conditions where osmolarity changes of the system were minimized (see "Experimental Procedures"). Fig. 5A shows that the addition of C1-induced a consistent Mg2+ efflux, which was already maximal in the presence of 10 mM C1-. When the same experiment was repeated in the presence of 50 p~ DNDS, an inhibitor of the plasma membrane CI-/HCO; exchanger (28, 29), M P efflux was almost completely abolished in the presence of 25 mM C1-or significantly decreased in the presence of 75 mM C1- (Fig. 5B). Fig. 6 shows that, in addition to extracellular Na+ and C1-, extracellular Ca2+ was also required for both stimulated efflux and influx of M$+ in hepatocytes. The decrease in extracellular ca2+ concentration from 1.2 to 0.3 mM and to 0 mM decreased or abolished both the vasopressin-mediated Mg2+ uptake and the NE-stimulated Mg2+ release.
In principle, extracellular Ca2+ could exchange for intracellular M$+ and such an exchange could account for the release of M$+ induced by NE. On the other hand, lack of extracellular Ca2+ should facilitate, rather than inhibit, the Mg2+ uptake induced by vasopressin.
It is possible that changes in extracellular CaZ+ result in changes of intracellular Ca2+ or total cell Ca2+ bound within the cytosol or internalized within organelles. Fig. 7 shows the results of an experiment where cytosolic free Ca2+ in hepatocytes was highly buffered by the previous induced M e efflux (Fig. 5 B ) . Data  loading with the Ca2+ buffer BAPTA. In the presence of a physiological extracellular Ca2+ concentration, the buffering of cytosolic free Ca2+ abolishes both the NE-dependent Mg2+ release (Fig. 7 A ) and the vasopressin-dependent M$+ uptake (Fig. 7B).
The data of Fig. 7A, where BAPTA inhibits stimulated MgZ+ efflux, cannot be unequivocally explained in term of cytosolic Ca2+ buffering, since BAPTA might also buffer cytosolic M$+. This explanation, however, is not consistent with the data in Fig. 7B, which shows that even the stimulated M P uptake is inhibited by BAPTA. Under these conditions cytosolic M$+ buffering should have enhanced, or maintained unaffected, M$+ uptake. Fig. 8 shows the results of an experiment where intact hepatocytes were incubated in the presence of 5 nM thapsigargin, a specific blocker of the ER Ca2+-ATPase (30, 31). Thapsigargin, which is lipophilic, can cross the plasma membrane of liver cells, reach its intracellular target, and effectively deplete the ER Caz+ store(s) (30, 31). When thapsigargin was added to a suspension of intact hepatocytes, no detectable loss or gain in MgZ+, with respect to control samples, was observed. However, under these conditions a small, but long lasting, increase in cytosolic free Ca2+ was observed in cells loaded with the calcium-selective fluorescent dye Fura-2 (see below), consistent with a depletion of the ER Ca2+ pool(s) following the inhibition of the ER Ca2+-ATPase. When thapsigargin-treated hepatocytes were stimulated by adding 10 p~ NE or 50 p~ forskolin (Fig. M), M e effluxes, similar to those reported in Fig. lA and Fig. 2, were observed. In contrast, no M e influx was observed when 20 nM TPA (Fig. 8 B ) or 20 nM vasopressin (not shown) were used under similar experimental conditions. Higher doses of these agents were equally ineffective.
The increase in cytosolic Ca2+ following the inhibition of the ER Ca2+-ATPase by thapsigargin is a process that requires several minutes for a maximal effect to be observed (31). Therefore, intact hepatocytes were treated with 5 nM thapsigargin for varying periods of time (20 s, 1 min, 3 min, or 5 min) before being stimulated by 20 nM vasopressin. As shown in Fig. 9 (bottom panel), hepatocytes were still able to accumulate Mg2+ from the extracellular medium when vasopressin was added 20 s or 1 min after the addition of thapsigargin. In contrast, in hepatocytes treated for 3 min or for 5 min with thapsigargin, the vasopressin-induced M$+ influx was cornpletely abolished. Parallel experiments, where intracellular free Ca2+ was measured with the Ca2+ fluorescent dye Fura-2 ( Fig. 9, top panel), indicated that after 20 s or 1 min of thapsigargin treatment, cytosolic free Ca2+ increased and some ER Ca2+ could still be mobilized by vasopressin. Hepatocytes treated for 2 min with thapsigargin, instead, presented a level of cytosolic free Ca2+ consistently higher than the untreated cells and no Ca2+ was further mobilizable by vasopressin from the endoplasmic reticulum Ca2+ pools. The inhibition of M$+ uptake by thapsigargin could be due to the increase in cytosolic Caz+ following the inhibition of the ER Ca2+ pump or to the depletion of Ca2+ from the ER itself. The data of Fig. 9 provide evidence supporting the latter alternative.

DISCUSSION
Overall M$+ Fluxes in Hepatocytes-Previous reports from our laboratory have provided evidence that isolated cardiac (18,19) and liver cells (15,16,18) can regulate Mg2+ movements across the plasma membrane in response to hormonal stimulation. More specifically, the increase in cytosolic cAMP level, resulting from the stimulation of @-adrenergic receptors or from the use of forskolin or permeant cAMP analogs (i.e. dibutyryl CAMP, 8-chloro-cAMP, 8-bromo-cAMP), prompts a marked Mg2+ efflux from both cell types. A CAMP-dependent M$+ efflux has been reported to occur in thymocytes as well (21), supporting the idea that the CAMP-mediated M$+ efflux is a more general mechanism. We have also shown that this M$+ efflux mainly depends on the mobilization of a sizable Mg2+ pool located within the mitochondrial matrix, via the activation of the adenine nucleotide translocase (16).
On the other hand, cardiac and liver cells (15,16,18,19), as well as lymphocytes (22), can accumulate M P from the extracellular space, when the protein kinase C pathway is activated (32-35).
In the present study we investigated the possibility that the cAMP and protein kinase C pathways could be sequentially and alternatively activated in the same cell population. The data reported in Fig. 1 (A and B ) indicate that stimulated MgZ+ release could be reversed to Mg2+ uptake within seconds or vice versa, by alternatively activating signaling pathways.
Two striking conclusions can be drawn from these data. First, both Mg2+ efflux or influx are rapidly reversible. Second, the amount of net M$+ translocated in either direction within a few minutes is very large and corresponds to a net decrease or increase in total cell M P by 1 mM (-10%) (16,19). Based upon reported values of hepatocyte surface area (36, 37), either flux will correspond to a net flux of 0.2 nmol of M e / min/cm2, which is one of the highest values reported for exchangers operating in cell for net ion translocation.
Dependence of Cell M$+ Uptake and Release on Extracellular Ions-An increasing number of reports suggest that a Na+/Mg2+ exchanger, located in the plasma membrane of mammalian cells (17,20,21), chicken erythrocytes (6, 23), and giant squid axon (24), is the main mechanism responsible for Mg2f efflux under stimulatory conditions. Heretofore, no evidence has been provided about the dependence of M$+ influx on extracellular Na+ and whether or not this sodiumdependent pathway could be involved in M$+ uptake as well. The importance of a physiological concentration of Na+ in the extracellular medium for MgZ+ efflux is confirmed by the data reported in Figs. 2 and 3. On the other hand, these same data indicate that stimulated M$+ influx is equally dependent on extracellular Na+.
Hence, the proposed obligatov exchange of extracellular Na+ for intracellular M$+ during Mg2+ efflux may not be operative in hepatocytes for at least two reasons. First, extracellular Na+ is required for both uptake and release of M e . Second, the exchange ratio between Na' and Mg2+ measured in these hepatocytes is extremely variable, ranging between 1 and 4 Na+ for 1 Mg2+ (not shown).
Both types of evidence suggest that, although Na+ could exchange with M$+ under certain conditions, other cations or anions could be involved in exchanging for M P .
For instance, extracellular C1-can support a stimulated M$+ efflux, but not influx (Fig. 4B). This observation, which clearly demonstrates a dissociation between the two processes, requires additional investigation, since changes in extracellular C1-can result in changes in intracellular C1-concentration, pH, and membrane potential.
The extracellular Ca2+ requirement for both stimulated Mg2+ uptake and release can also be explained by different mechanisms. For instance, under certain conditions, extracellular Ca2+ could exchange for intracellular M$+, and this putative pathway could account for an exchange of Ca2+o for M$+i during the M$+ efflux mediated by NE. However, lack of extracellular Ca2+ resulted in a similar inhibition of cellular M$+ uptake by liver cells. If a Ca2+/Mg2+ exchange were involved in both Mg2+ release and uptake, a decrease in extracellular Ca2+ should affect only M e release and enhance or be without effect on M$+ uptake.
Changes in extracellular Ca2+ also resulted in changes in intracellular Ca2+ and in a decrease of the Ca2+ bound within cytosolic buffers and/or stored within intracellular organelles.
Cytosolic Ca2+, ER Calcium Store, and Cell M$+ Uptake- The possibility exists that cytosolic free Ca2+ may be important in mediating M$+ uptake from the extracellular medium. Another, not mutually exclusive, hypothesis is that intracellular organelles play a role on intracellular Mg2+ homeostasis. Accordingly, most of the observed M P uptake and release by hepatocytes could be the observable consequence of a previous M$+ uptake by or Mg2+ release from these organelles. Two lines of evidence support this hypothesis. First, the amount of net M e uptake or release observed 3-5 min after stimulation is equivalent to 10% of the total cellular M e content and is greater than the values of free cytosolic M e reported in the literature (1,2,4). Second, in the case of cellular Mg2+ release due to NE, most of the M e translocated outside the cell is M$+ mobilized from the mitochondria (16).
Under vasopressin stimulation, a transient rise in cytosolic Ca2+ occurs in liver cells (7, 38-40) (Fig. 9) following the mobilization of ER Ca2+ pool by inositol 1,4,5-trisphosphate (7, 41, 42). Additionally, protein kinase C (33, 42) appears to induce Ca2+ mobilization, probably via the same or by an alternative route. Less clear is the link between these stimulatory agents and M P . Bond et al. (7) reported a major redistribution of cellular M$+ in rat liver "in uiuo" stimulated by vasopressin and glucagon, whereas Baumann et al. (8) and Somlyo et al. (13) suggested that M$+ can be counter-transported during Ca2+ release from the ER or the sarcoplasmic reticulum, both in physiological and pathological conditions. An alternative or additional mechanism involved in Mg? influx and intracellular redistribution could be the exchanger localized in the inner mitochondrial membrane, responsible for the transport of Mg.ATP2-complex into the mitochondria, in exchange for the PO:-present in the matrix (43)(44)(45). This pathway appears to be activated by the rise in cytosolic Ca2+ that follows the vasopressin stimulation of liver cells (44,451. Taken together, the data presented here do not unequivocally support the hypothesis that changes in cytosolic free Ca2+ concentration alone can modulate Mg2+ uptake (or release) from hepatocytes. However, the results do indicate a role for the Ca2+ cycling by ER in the stimulated M e uptake, but not the Mg2+ release, in hepatocytes. There is only one set of experimental data to indicate that the rise in cytosolic Caz+ could explain Mg2+ uptake, either by exchanging with Mg2+ across the plasma membrane or by an alternative mechanism. This is the finding that agents which promote cellular MgZ+ uptake, such as vasopressin or TPA, increase cytosolic Ca2+, at least transiently (32, 38-42).
To the extent that the effect of thapsigargin is selectively confined to the Ca2+-ATPase of the ER, and no evidence to the contrary exists in the literature, the data presented in this report indicate that functional ER Ca2+ cycling is necessary for M e uptake into hepatocytes. The lag period observed between the inhibition of the CaZ+-ATPase after thapsigargin addition ( Fig. 9 and Ref. 31) and the inhibition of the stimulated Mg2+ uptake suggest that the size of the ER Ca2+ pool, rather than the Ca2+-ATPase itself, is responsible for cellular Mg2+ uptake.
Hence, all these data are consistent with the ER being the destination site for the observed M$+ uptake under conditions where Ca2+ uptake by the ER is present and Ca2+ is stored within the ER.
Previous results from this laboratory (16) and others (17) clearly indicate that the CAMP-stimulated M e release observed in hepatocytes depends, to a great extent, on the mobilization of Mg2+ from mitochondria, followed by the extrusion of cytosolic M P . We have shown that this release is not affected by thapsigargin. In contrast, ER functioning in term of ca2+ cycling is necessary for the observed M e uptake stimulated by vasopressin, carbachol, or TPA.
The process of stimulated Mg2+ uptake described in this work is novel although not mechanistically defined. However, the identification of two different cellular organelles as the origin and destination sites for hepatocytes Mg2+ release and uptake, and the disassociation of conditions affecting the two processes, provide a basic foundation and experimental tools for a better elucidation of the integrated Mg2+ redistribution and homeostasis in hepatocytes.