Increased hepatic Na,K-ATPase activity during hepatic regeneration is associated with induction of the beta1-subunit and expression on the bile canalicular domain.

Cellular and molecular mechanisms regulating the activity of the sodium pump or Na,K-ATPase during proliferation of hepatocytes following 70% liver resection have not been defined. Na,K-ATPase may be regulated by synthesis of its α- and β-subunits, by sorting to either the sinusoidal or apical plasma membrane domains, or by increasing membrane lipid fluidity. This study investigated the time course of changes during hepatic regeneration for Na,K-ATPase activity, lipid composition and fluidity, and protein content of liver plasma membrane subfractions. As early as 4 h after hepatic resection, Na,K-ATPase activity was increased selectively in the bile canalicular fraction. It reached a new steady state at 12 h and remained elevated for 2 days. Although hepatic regeneration was associated with a reduced cholesterol/phospholipid molar ratio and increased fluidity, measured with two different probes, these changes in lipid metabolism were in the sinusoidal membrane domain. The Na,K-ATPase β1-subunit, but not the α1-subunit, was increased selectively at the bile canalicular surface as shown by immunoblotting of liver plasma membrane subfractions and the morphological demonstration at both the light and electron microscopic levels. Furthermore, cycloheximide blocked the rise in β1-subunit mRNA levels. Since the time course for β1-subunit accumulation was similar to that for activation of Na,K-ATPase activity, this change implicated the β1-subunit in activating sodium pump activity.

Na,K-ATPase (EC 3.6.1.37) is a plasma membrane enzyme whose activity mediates cation fluxes (1). The extrusion of sodium in exchange for potassium is responsible in part for the transport of nutrients, electrical potential, and regulation of protein synthesis. Most models of the sodium pump suggest that the holoenzyme consists of a heterodimer (2). The catalytic ␣-subunit consists of 10 transmembrane-spanning domains and contains all the known binding sites of the functional unit, including ouabain binding, ATP hydrolysis, and sodium and potassium recognition sites (1). Three isoforms have been cloned, but in adult rat liver, only the ␣ 1 -subunit has been identified (3,4). The ␤-subunit, on the other hand, is of variable size due to tissue-specific glycosylation patterns at three potential N-linked asparagine sites (3,5). At least two isoforms have been identified (4,6). The precise function of ␤-subunits is unclear (5), but they have been proposed to be involved in the conformational folding of the ␣-subunit in the endoplasmic reticulum (7,8), to promote cation exchange (9,10), to be involved in cell adhesion (11), and to promote the mitogenic response (12).
In liver, several controversies have revolved around the composition and location of the sodium pump (13). Although all authors have detected ␣-subunit protein and mRNA, there has been variable agreement on the ␤-subunit. Some reports indicated that liver does not contain the ␤-subunit (14 -18), while more recent studies demonstrated ␤ 1 -but not ␤ 2 -subunits in liver (19 -21). Also, Simon et al. (21) demonstrated that the ␤ 1 -subunit was localized to the sinusoidal membrane surface, while the ␣ 1 -subunit was present at both domains, consistent with most, but not all, previous morphological and biochemical studies in liver. Na,K-ATPase activity has also been shown to be sensitive to lipid composition and structure (fluidity) (22)(23)(24)(25). Thus, in hepatocytes, Na,K-ATPase may be regulated by selection of isoform subunits, levels of gene expression, spatial distribution, membrane lipid structure, and response to various modulators.
During hepatic regeneration, Na,K-ATPase activity has been reported to be increased in association with both increased ␤ 1 -subunit mRNA and liver plasma membrane fluidity (26 -31), while ␣ 1 -subunit mRNA and protein were either not altered or increased (12,30,32). To examine the possible importance of ␤-subunit induction and membrane lipid fluidity in the regulation of hepatic Na,K-ATPase, we measured the time course of changes in Na,K-ATPase activity, membrane lipid composition and fluidity, and protein expression in liver plasma membrane subfractions. The results indicated that hepatic regeneration selectively increased Na,K-ATPase activity in the BCM 1 domain, in association with increased ␤ 1 -subunit levels, while the alterations in lipid composition and fluidity occurred in the SM domain and were not associated with changes in enzyme activity or subunit composition. Furthermore, the markedly increased ␤ 1 -subunit mRNA level peaked between 4 and 8 h and was dependent on de novo protein synthesis. olis, IN). Animals were maintained on a standard 12-h light-dark cycle and fed commercial laboratory chow and water ad libitum. Rats were fasted prior to surgery. Under light ether anesthesia, the median and left lateral lobes were excised according to the method of Higgins and Anderson (33). In sham-operated animals, the appropriate portions of the liver were exteriorized for the same length of time as those undergoing the 70% resection. At various times post 70% partial hepatectomy, the animals were killed, and the liver was removed and washed in PBS. Approximately 1 g was prepared for mRNA, and the remaining liver tissue was used to isolate plasma membrane subfractions as described below. Additional animals were exposed to light ether anesthesia only. In another set of rats, in addition to sham operation and 70% resection, each of these groups also received 5 mg of cycloheximide (Sigma)/100 g of body weight 15 min prior to surgery (34). There were only two operative deaths in all the experiments.
Liver Plasma Membrane Subfraction Isolation-Liver plasma membrane subfractions were isolated concomitantly from liver homogenate as described previously (35). Briefly, liver slices were homogenized in a chilled buffer (300 mM mannitol, 5 mM EGTA, 18 mM Tris-HCl, 1 mM phenylmethylsulfonyl fluoride, pH 7.4) using a Polytron (Brinkmann Instruments) in 15 ml of buffer for 45 s. The solution was centrifuged at 48,000 ϫ g for 30 min. The resulting pellet was resuspended in buffer, and 15 mM Mg 2ϩ precipitation was followed by centrifugation for 15 min at 2445 ϫ g. The pellet was saved for SM isolation, while the supernatant was centrifuged at 48,000 ϫ g for 30 min to obtain the BCM fraction. SMs were isolated from the initial Mg 2ϩ precipitation pellet on a discontinuous sucrose gradient using 41% (5 ml) and 37.5% (12 ml) as overlay layers in cellulose acetate tubes centrifuged at 88,000 ϫ g for 2.5 h in a Beckman Model L8 -70 ultracentrifuge and SW 28 rotor. The float layer was carefully harvested from the top of the gradient. Both fractions were washed and stored in 0.5 ml of 1 mM NaHCO 3 plus 1 mM phenylmethylsulfonyl fluoride at both Ϫ20 and Ϫ80°C.
Membrane Enzyme Activity Measurements-The purity of each SM and BCM fraction was determined from the measurement of membrane-bound specific enzyme activities, leucine aminopeptidase (BCM), cytochrome c reductase (endoplasmic reticulum), and succinate dehydrogenase (mitochondrial) in liver homogenate, SM, and BCM as described previously (35). Na,K-ATPase and Mg 2ϩ -ATPase were measured in the Ϫ20°C stored material after overnight freeze-thawing by means of an enzyme-coupled kinetic assay with pyruvate kinase and lactate dehydrogenase, assuming ouabain (2.5 mM) inhibition measured the Na,K-ATPase activity. Enzyme activities are expressed as mol/ h/mg of protein. Enrichment is defined as specific activity in the membrane subfraction/specific activity in the homogenate. Protein was determined by the method of Lowry et al. (36) using bovine serum albumin as a standard.
Lipid Composition and Fluidity Measurements-Total lipids were extracted from Ϫ80°C stored membrane fractions (1-2 mg of protein) by the method of Bligh and Dyer (37). Cholesterol was quantitated using coprostanol as the internal standard by chromatographic methods as described previously (38). Total phospholipids were determined by the method of Ames and Dubin (39).
Fluorescence polarization measurements were done on a Model 4800 polarization spectrofluorometer (SLM-AMINCO, Urbana, IL) with fixed emission and excitation filters as described (35). Fluorescence intensity was measured perpendicularly and parallel to the polarization phase of the excitation light (excitation wavelength of 360 nm; KV 389 emission filter (Schott, Duryea, PA)). Measurements were made at 37°C. Nonsignificance of membrane vesicle light scattering was verified by insertion of sample without probe. The probes 1,6-diphenyl-1,3,5-hexatriene (DPH) and DL-12-(9-anthroyloxy)stearic acid (12-AS) (Molecular Probes, Inc., Junction City, OR) were dissolved in tetrahydrofuran to a final concentration of 0.6 g/ml. Probes were added to intact membranes in a total volume of 1.2 ml (containing 72 g of protein) and frequently vortexed for 12-15 min at 37°C. Measurements were done in triplicate, and results were analyzed as described previously (35). Membrane lipid fluidity was increased in vitro by the addition of 2-(2-methoxyethoxy)ethoxylethyl 8-(cis-2-n-octylcyclopropyl)octanoate (A 2 C) (Sigma) at a 6 M final concentration (25).
Membrane Protein Electrophoresis and Immunoblotting-SDS-polyacrylamide gel electrophoresis and immunoblotting were carried out as described previously (21). For ␣-subunits, liver plasma membrane subfractions were diluted 1:1 in 2 ϫ loading buffer without boiling or reduction and were immediately loaded (15-30 g/lane) onto 7.5% gels using the electrophoresis system of Laemmli (40). ␤-Subunits were similarly detected except that 1% ␤-mercaptoethanol was added to samples after the plasma membrane fractions were boiled for 1 min. Proteins were transferred overnight onto 0.1-m nitrocellulose filters (Micron Separation Inc., Westboro, MA) by the procedure of Towbin et al. (41) at 150 mA overnight at 4°C. Efficiency of protein transfer was determined by protein staining of the gel and generally was Ͼ90%. Na,K-ATPase subunits were identified in SM and BCM fractions using rabbit polyclonal antisera (Upstate Biotechnology, Inc.) diluted 1:750. Polymeric IgA receptor was identified in plasma membrane subfraction samples that were boiled for 2 min in buffer containing 10 mM EDTA, 25 mM dithiothreitol, 2% SDS (Bio-Rad). After cooling, 2.3 mM iodoacetamide was added and then probed using a polyclonal anti-polymeric IgA receptor (gift from Dr. K. Howell) (42). Bands were identified after washing in Tris-buffered saline at 22°C with second-step alkaline phosphatase conjugate antibodies (Tago, Inc.) diluted 1:5000, incubated for 90 min at 22°C, and washed three times. Alkaline phosphatase (Kirkegaard & Perry Laboratories, Inc.) gel immunostaining density was quantitated using Bio-Rad laser densitometry.
Immunofluorescence and Electron Microscopy-Tissue from eight rats was obtained at either 24 or 32 h post 70% hepatectomy and compared with sham-operated animals. Livers from these animals were cleared of blood by portal perfusion (30 s) with PBS containing 2% sucrose and fixed with 4% paraformaldehyde-lysine-periodic acid fixative for 5 min (43). The livers were removed and sliced into 1-2-mmthick slices, and fixation was continued for 50 min at room temperature. The tissue was washed once and stored in PBS at 4°C until further processing. For the preparation of 5-m cryostat sections for immunofluorescence, tissue was infiltrated with sucrose in PBS in stepwise increments of 10, 15, 20, and 25% for 30 min each at 4°C. The tissue was then embedded in O.C.T. compound (Tissue Tek) and rapidly frozen in liquid nitrogen-cooled isopentane. Sections (5 m) were cut with a cryostat, mounted on a slide, dried, and stored at Ϫ70°C until immunolabeled. Ultrathin cryosections were prepared from tissue cryoprotected with 2.1 M sucrose in PBS according to Griffith (44) and Tokuyasu (45) and mounted on Formvar-coated grids.
The thawed ultrathin cryosections were immunolabeled as indicated above for fluorescence labeling except that at step 4, the sections were incubated with rabbit anti-mouse serum (Cappel) diluted 1:400 for 30 min, washed four times for 5 min each, incubated with goat anti-rabbit IgG conjugated to 10-nm colloidal gold (Biocell, Ted Pella, Inc.) diluted 1:40, and washed four times for 5 min each. The specimens were washed briefly with water and embedded in 2% methylcellulose containing 0.2% uranyl acetate. Specificity was confirmed by failure of nonrelevant monoclonal antibody to demonstrate similar labeling patterns in either sham-operated or regeneration animals. Samples were coded and interpreted with knowledge of the experiment model.
RNA Isolation and Northern Blot Analysis-Poly(A) ϩ mRNA was rapidly extracted from livers using PolyATtract system 1000 (Promega). mRNA was fractionated on 1.2% agarose-formaldehyde gels in borate buffer at a constant voltage (140 V) for 4 h at 22°C. RNA was transferred to Hybond N ϩ (Amersham Corp.) by capillary action and fixed by ultraviolet cross-linking. cDNA probes were labeled with [ 32 P]dCTP to a specific activity of ϳ2 ϫ 10 6 cpm/ng of DNA using the Megaprime random-primed labeling system (Amersham Corp.). Hybridization was performed as described previously (21) with the following cDNA probes: 1) the rat ␣ 1 -subunit (provided by J. Lingrel), 2) the rat ␤ 1 -subunit (provided by E. Benz), or 3) the rat asialoglycoprotein receptor (provided by M. McPhaul). Gels were hybridized using Rapid-Hyb buffer (Amersham Corp.) at 65°C for 16 h; washed twice for 20 min each with 2 ϫ SSC, 0.1% SDS and twice with 0.1 ϫ SSC, 0.1% SDS; and exposed to autoradiographic film for 1-16 h with one intensifying screen.
Data Analysis-The data are expressed as the means Ϯ S.E. The statistical significance of the results between samples obtained from sham-operated and experimental rats was determined by the unpaired Student's t test or one way analysis of variance with Student-Newman Keul analysis for multiple comparisons. Significance was accepted at the p Ͻ 0.05 level.

RESULTS
We initially examined the time course of changes in Na,K-ATPase following 70% hepatic resection since previous studies on Na,K-ATPase activity during hepatic regeneration have resulted in discordant reports (26 -32, 46, 47). The results shown in Fig. 1 indicated that Na,K-ATPase activity measured in liver homogenates was not significantly increased until 8 h, reaching a plateau at 12 h, which was maintained for 48 h before returning to control values at 4 days. In contrast, Mg 2ϩ -ATPase specific activity did not significantly change (data not shown). No significant changes in Na,K-ATPase activity were measured in either the sham-operated or anesthesia controls (data not shown). Up-regulation of Na,K-ATPase followed a pattern in which the activity increased during the prereplicative stage and remained elevated for two rounds of cell division.
Functionally active Na,K-ATPase in rat liver is predominantly localized to the SM domain (21,25). Using cell fractionation techniques that permitted the separation of SM and BCM fractions from the same homogenate, we determined whether increased Na,K-ATPase activity during hepatic regeneration was localized to the sinusoidal membrane domain. SM and BCM fractions were isolated at specific times after hepatic resection, and the purification of each fraction was determined. Marker enzyme enrichments shown in Fig. 2A indicated that the relative enrichment of the plasma membrane subfractions were not altered during regeneration. Leucine aminopeptidase, a marker of the BCM domain (48), was enriched 48 Ϯ 2-fold in control BCM fractions and was not significantly altered during regeneration. Furthermore, neither cytochrome c reductase (marker of the endoplasmic reticulum) nor succinate dehydrogenase (mitochondrial enzyme) indicated significant changes in enrichments from control values during regeneration. In particular, the BCM fraction enrichment values for these intracellular markers were low in both sham-operated and resected animals.
The possibility that regeneration increased distribution of SM components into the BCM fraction was also examined. Fig.  2B shows the distribution of the polymeric IgA receptor measured by immunoblotting of liver plasma membrane subfractions. In controls, the polymeric IgA receptor was located predominantly on the SM domain with lesser density in the BCM (Fig. 2B) as previously shown (49). The distribution pattern was not altered at either 12 or 24 h following hepatic resection. In SM fractions (but not BCM fractions) from controls and regeneration, additional lower molecular mass bands appeared, suggesting possible contamination with the endoplas-mic reticulum. This is consistent with the presence of cytochrome c reductase activity in the SM fraction, but not the BCM fraction. Taken together, these studies indicated that isolated BCM fractions were reasonably pure, and more important, their distribution was not altered during hepatic regeneration. Fig. 3 shows the effects of hepatic regeneration on Na,K-ATPase activity in the SM (panel A) and BCM (panel B) fractions during 2 weeks following resection. Na,K-ATPase activity in the SM fraction remained unchanged compared with shamoperated animals. In contrast, in the BCM fractions, Na,K-ATPase activity was significantly elevated as early as 4 h after resection and continued to increase, reaching a new plateau at 12 h and remaining elevated at 48 h. Na,K-ATPase activity returned to undetectable levels in the BCM at 4 days and remained unchanged at 2 weeks.
The observation that Na,K-ATPase was selectively increased in the BCM fraction was consistent with previous in vitro activation of latent activity with the fluidizing agent A 2 C (25). We therefore examined the hypothesis that increased Na,K-ATPase activity was secondary to increased BCM fluidity. To test this hypothesis, we reasoned that in contrast to control BCM fractions, the in vitro addition of A 2 C would not further activate Na,K-ATPase activity during hepatic regeneration. On the other hand, if enzyme activity was derived from the sinusoidal membrane surface by either contamination or reorganization from endogenous sites, A 2 C would further increase activity. Fig. 4A demonstrates that the in vitro addition of 6 M A 2 C to sham BCM fractions activated latent Na,K-ATPase activity. However, up to 4 days after hepatic resection, A 2 C had no further activating effect on the BCM fraction (Fig. 4B). The ability of A 2 C to activate was restored 2 weeks posthepatic resection, when activation of latent Na,K-ATPase activity returned to control levels. The addition of A 2 C to SM fractions from either sham-operated or regeneration animals had no effect on Na,K-ATPase activity (data not shown).
Membrane lipid composition and bulk lipid fluidity are known to regulate Na,K-ATPase activity (22-24, 50 -52). Although it has been shown that hepatic regeneration alters liver plasma membrane lipid composition and fluidity, it was not clear which domain was affected. The changes in lipid composition and membrane lipid fluidity were therefore determined in liver plasma membrane subfractions. The effect of hepatic regeneration on the cholesterol/phospholipid molar ratio, a major determinant of membrane fluidity, is shown in Fig. 5. As previously shown, the ratio in the BCM (0.80 Ϯ 0.09) was significantly greater than that in the SM (0.48 Ϯ 0.02) in control animals (35). During hepatic regeneration, the cholesterol/phospholipid molar ratio rapidly and markedly decreased in the SM fractions as early as 4 h after resection and remained reduced for at least 2 days. In contrast, the cholesterol/phospholipid molar ratio in the BCM fraction was unchanged until 12 h after resection. These changes in the molar ratio were due to decreased cholesterol content in both the SM and BCM fractions since total phospholipid was not significantly different from control (data not shown). These studies, however, surprisingly indicated that the major early changes in lipid composition during hepatic regeneration were at the SM and not the BCM surface.
Next, we examined whether the changes in lipid composition were reflected in similar alterations in membrane lipid fluidity. Fluidity was directly determined using two different fluorescent probes that measured the two major components of fluidity. DPH fluorescence polarization measures largely the static component, while the 12-AS probe reflects the dynamic parameters of lipid fluidity. Fig. 6 (A and B) compares the changes in DPH fluorescence polarization in the sinusoidal and bile canalicular membranes from sham-operated and regeneration animals. As previously reported, SM polarization values are lower than the BCM fraction values, indicating greater fluidity (35). Following 70% resection, the polarization values in the SM fraction were rapidly and dramatically decreased (increased fluidity) as early as 4 h and remained low for at least 2 days. Decreased polarization values returned to control values at 4 days. In contrast, during the first 12 h following resection, no changes in polarization values were detected in the BCM compared with sham-operated animals. After 24 h, regeneration significantly decreased BCM polarization, which returned to control values at 4 days.
To further pursue this unexpected result, we used another probe (12-AS). Similar changes in polarization were measured with 12-AS during regeneration (Fig. 6, C and D). Polarization values were selectively decreased in the SM as early as 4 h after resection and only returned to control values at 12 h; and again, no changes in BCM fluidity were detected with 12-AS during 24 h posthepatic resection.
The dichotomy between increased Na,K-ATPase activity in the BCM and the measurements of increased fluidity and decreased cholesterol content at the sinusoidal membrane pole strongly indicated that changes in lipids could not account for the dramatic increase in BCM Na,K-ATPase activity. Additional possibilities that could explain increased BCM Na,K-ATPase activity during hepatic regeneration included altered distribution and/or increased content of Na,K-ATPase subunits. Fig. 7A shows that regeneration did not alter the content of the ␣-subunit in either the SM or BCM fractions at 24 h, a time when Na,K-ATPase activity was maximally active. On the other hand, the ␤-subunit density was selectively increased in the BCM fraction (Fig. 7B), but was unchanged in the SM fraction until 24 h, when a modest decrease in its content was measured. The increase in ␤-subunits in the BCM fraction was time-dependent, reaching a new steady state at 12 h after resection and remaining elevated for at least 4 days. This time course was similar to that for the progressively increased Na,K-ATPase activity measured in the BCM fraction, strongly supporting the hypothesis that sorting of ␤-subunits to the BCM was responsible for activating the latent sodium pump. Furthermore, we were unable to identify other ␣or ␤-isoforms either in sham-operated or regenerating liver samples (data not shown).
The ␤-subunit has been localized by immunofluorescence to the sinusoidal membrane domain (21). However, its ultrastructural location and the effect of hepatic regeneration on the plasma membrane spatial distribution have not been reported. Fig. 8 shows the immunofluorescent location of the ␤-subunit using the monoclonal antibody IEC 1/48. As shown in Fig. 8A, the ␤-subunit in sham-operated animals was equally distributed along the acinar lobule and localized to the sinusoidal membrane domain (Fig. 8B). The distribution of ␤-subunit immunofluorescence remained evenly distributed along the hepatic lobule following hepatic resection, examined at 24 and 32 h post-resection (Fig. 8C). However, as shown in Fig. 8D, hepatocytes were enlarged, and although ␤-subunit immunofluorescence is still observed at the sinusoidal membrane surface, fluorescence is more evident along the lateral surface and  4. Effect of A 2 C on Na,K-ATPase activity in bile canalicular membrane fractions. Na,K-ATPase activity was determined at the indicated time points. A shows the effect of A 2 C (Ç) on Na,K-ATPase activity compared with untreated (å) bile canalicular membrane fractions from sham-operated rats. B compares the effect of A 2 C (E) and diluent (q) on Na,K-ATPase activity in bile canalicular membrane fractions from regenerating liver. 6 M A 2 C or diluent was added to isolated fractions in vitro, and enzyme activity was measured by a coupled enzyme assay as described under "Experimental Procedures." Results are expressed as the means Ϯ S.E. of separate determinations in four plasma membrane preparations at each time point. may be present as well in BCM structures (Fig. 8D).
The ultrastructural location of the ␤-subunit was then determined using immunogold labeling. In Fig. 9 (A and B), the ␤-subunit was localized to the sinusoidal membrane and lateral surfaces in sham-operated animals without gold particles in the BCM domain (Figs. 9B and 10A), strongly supporting pre-vious biochemical and morphological localization. Following hepatic resection at 24 and 32 h, the bile canaliculus was observed to be enlarged with marked reduction of microvilli (Fig. 10B) compared with sham-operated animals (Figs. 9B and 10A). Immunogold labeling was also clearly seen at the BCM surface 24 h after resection (Fig. 10B). This appearance was not seen in every canaliculus, but rather involved a selective number of hepatocytes without an apparent relationship to lobular location or mitosis. Fig. 11 (A and B) shows the time course of changes in the ␣ 1 -   10. Immunolocalization of Na,K-ATPase ␤-subunit in bile canaliculi of normal and regenerating liver. Colloidal gold labeling of the ␤-subunit was performed as described for Fig. 9. A is a micrograph of a bile canaliculus (BC), illustrating the distribution of the ␤-subunit in normal liver. As in Fig. 9, the canalicular plasma membrane is not labeled, whereas the lateral plasma membrane is labeled with colloidal gold (arrowheads) to the level of the tight junctional complexes (arrows). In B, the arrowheads demonstrate the distribution of the ␤-subunit on the plasma membrane of the bile canaliculus in regeneration. The tight junctions are indicated by arrows. The bars are 0.2 m. and ␤ 1 -subunit mRNA densities relative to sham-operated animals determined over 48 h following hepatic resection. As previously shown by some (12), but not others (32), ␣ 1 -subunit mRNA was not significantly changed over 48 h of regeneration, while ␤ 1 -subunit mRNA was increased dramatically as early as 4 h after resection, rising to 21-fold above sham expression. The elevated levels of ␤ 1 -subunit mRNA returned to control levels by 24 h. Neither in sham-operated animals nor during regeneration could we demonstrate ␣ 2 -, ␣ 3 -, or ␤ 2 -subunit mRNA isozymes in rat liver (data not shown). Thus, the increase in ␤ 1 -subunit mRNA occurred prior to the onset of DNA synthesis and the appearance of the ␤ 1 -subunit peptide in the BCM.
Following hepatic resection, a number of genes, the immediate-early genes, are activated within minutes of resection and are independent of protein synthesis (34). Subsequently, many of these genes induce the delayed-early response genes. These latter genes are induced within a few hours of hepatectomy, and their transcription requires protein synthesis. We sought to determine if ␤ 1 -subunit induction was part of the early or delayed-early response genes. Fig. 12 demonstrates the changes in ␤ 1 -subunit mRNA relative to sham-operated animals during the first 4 h after hepatic resection. ␤ 1 -Subunit mRNA levels did not rise significantly until 4 h after partial resection. Cycloheximide administration 15 min prior to resection blocked the mRNA rise at 4 h. Sham surgery alone did not significantly alter ␤ 1 -subunit mRNA levels. The results are consistent with the ␤ 1 -subunit mRNA response is consistent with a delayed-early gene.

DISCUSSION
The liver is one of the few adult organs that demonstrates a physiological growth response (53-55); thus, it provides an animal model of events associated with cellular proliferation. The sequence of events during hepatic regeneration is synchronized and well characterized (55). Within minutes following 70% partial hepatectomy, the majority of cells in the remnant liver that are normally quiescent rapidly re-enter the cell cycle associated with the induced expression of a number of growth response genes (34). The onset of DNA synthesis occurs at 12-16 h posthepatectomy in hepatocytes with a peak level at 24 h, followed by a peak of mitosis at 32-34 h, and the mass of the liver is restored by 7-10 days. During this acute period, the liver must maintain both vital metabolic and synthetic functions as well as initiate the induction of genes involved in the cell cycle.
The sodium pump or its enzymatic equivalent, Na,K-ATPase, is activated ubiquitously with cellular proliferation (56 -60), including hepatic regeneration. Na,K-ATPase is a heterodimeric transmembrane complex whose expression is dependent on numerous factors, including tissue specificity, developmental stages, subunit isoforms, and post-translational mechanisms, e.g. cellular trafficking, membrane sorting, and interactions with membrane lipids (61). The expression of many genes has been shown to increase during regeneration, including Na,K-ATPase ␤ 1 -subunit mRNA, suggesting that it may be involved in the mitogenic response (12). Also, hepatic regeneration is well recognized to alter hepatic lipid metabolism and to increase membrane lipid fluidity (62,63). However, it is unclear whether these changes in mRNA and lipid metabolism are related to growth or rather to metabolic adaptation of the sodium pump during hepatic regeneration. This study was therefore undertaken to answer the following questions. 1) Is increased Na,K-ATPase activity during hepatic regeneration due to changes at the sinusoidal surface, the BCM surface, or FIG. 11. Time course for Na,K-ATPase ␣ 1 -and ␤ 1 -subunit mRNAs during hepatic regeneration. Rats were subjected to either 70% hepatic resection or sham operation and killed at the same time of day. A shows the means Ϯ S.E. of four separate experiments in regeneration for changes in the ␣ 1 -subunit (E) and ␤ 1 -subunit (q) relative to sham-operated animals. B shows Northern blot gels for the ␤ 1 -subunit, ␣ 1 -subunit, and asialoglycoprotein (ASGP) receptor, which was used as a loading control. RNA was isolated, and 5 g of mRNA was loaded per lane as described under "Experimental Procedures." Exposure time for the ␤ 1 -subunit was 12-16 h. To show low levels of basal ␤ 1 -subunit mRNA, subsequent samples had to be overexposed.
FIG. 12. Effect of cycloheximide on Na,K-ATPase ␤ 1 -subunit steady-state mRNA levels during hepatic regeneration. Rats were subjected to 70% hepatic resection (HR) (q), cycloheximide (CX) alone (f), or hepatic resection ϩ cycloheximide (E). Cycloheximide (5 mg/100 g of body weight) was administered 15 min prior to surgery. RNA was isolated and quantitated as described under "Experimental Procedures." The effect of cycloheximide on the expression of the Na,K-ATPase ␤ 1 -subunit as -fold increase over sham-operated rats is shown. Asialoglycoprotein receptor (ASGP-R) mRNA levels were unaffected by experimental protocol (data not shown). Only 4-h posthepatic resection mRNA levels were significantly different from controls (p Ͻ 0.01). both? 2) Does increased membrane lipid fluidity activate latent enzyme activity? 3) Is increased ␤ 1 -subunit mRNA associated with increased protein expression, and which domain of the hepatocyte has increased ␤ 1 -subunit levels? The findings demonstrated that increased Na,K-ATPase activity was localized to the BCM domain and supported the hypothesis that the ␤ 1subunit peptide targeted to the apical domain, rather than alterations in membrane lipid composition and structure, was instrumental in regulating the increased enzyme activity.
After 70% hepatic resection, Na,K-ATPase activity increased during prereplication (0 -12 h) and remained elevated through mitosis, suggesting that increased activity was integrally involved in cell proliferation. Although during basal physiological states, Na,K-ATPase activity is localized to the SM surface following hepatic resection, activity was selectively increased at the BCM pole (Fig. 3). If one assumes that the canalicular surface area represents 0.5% of the hepatocyte membrane (64) and the BCM preparation recovered 12% of the fraction (35), the observed increase in activity in the BCM can account for the entire rise measured in liver homogenates.
Increased canalicular Na,K-ATPase activity during regeneration could result from contamination of BCM fractions with sinusoidal vesicles, activation of latent BCM enzyme, redistribution of sinusoidal enzyme into the BCM domain, or increase in the Na,K-ATPase ␣or ␤-subunit or both subunits in the BCM domain. Although it is difficult to exclude the possibility that the increased apical activity is not due to cross-contamination with SM fractions, several lines of evidence strongly argue against this possibility. First, the enrichment of the BCM marker enzyme leucine aminopeptidase activity was not altered significantly during regeneration. In addition, enzyme markers of intracellular organelles were either unchanged or showed de-enrichment in the BCM fraction. Second, the polymeric IgA receptor density, measured by immunoblotting, was unchanged in the BCM fraction during regeneration, indicating that the sinusoidal membrane fraction was not coisolated with the BCM fraction.
Na,K-ATPase activity is known to be sensitive to lipid composition and membrane fluidity (22)(23)(24)(25), and previous studies have shown increased hepatic membrane fluidity with regeneration (29,30). We therefore undertook additional studies to examine the possibility that regeneration activated latent activity by increasing fluidity. A 2 C increased Na,K-ATPase activity in the BCM fraction in sham-operated animals, but it had no effect on the increased Na,K-ATPase activity associated with regeneration (Fig. 4), which suggested that BCM Na,K-ATPase was activated during regeneration, rather than resulting from contamination or redistribution.
Activation may result either from increased BCM fluidity or from the association of newly synthesized ␤-subunits with endogenous ␣-subunit. This study demonstrated that hepatic regeneration was associated with dramatically increased lipid fluidity, measured with two fluorescent probes, DPH and 12-AS. However, the changes in fluidity were predominantly associated with the SM surface. In contrast, no significant changes in DPH and 12-AS polarization values were measured in the BCM fraction until 24 h, significantly after Na,K-ATPase activity had increased.
Membrane fluidity combines, at the very least, two components: the range of motion (or order parameter) and the dynamic component. DPH predominately measures the order parameter of membrane lipids located in the membrane bilayer core. Similar to the results with DPH, 12-AS demonstrated a marked and rapid decrease in fluorescence polarization selectively in the SM fraction. Thus, both major components of bulk membrane lipid fluidity were altered in the sinusoidal mem-brane without significant changes in the apical domain. Alterations in membrane fluidity are associated with time-dependent decreases in the cholesterol/phospholipid molar ratio, again selectively in the sinusoidal domain. Thus, it appears that the activity of Na,K-ATPase in the sinusoidal membrane is limited by subunits rather than by the membrane lipid environment. Similarly, A 2 C failed to further activate BCM Na,K-ATPase when the ␤-subunit was associated with ␣-subunits.
Most previous reports have shown, in mixed liver plasma membrane fractions, that Na,K-ATPase activity is increased during regeneration. However, one group was unable to demonstrate increased binding of monoclonal antibody to the ␣ 1subunit (30) or increased ␣ 1 -subunit mRNA levels (12), while recently, both ␣-subunit and mRNA levels were shown to be increased (32). On the other hand, both studies (12, 32) observed that ␤ 1 -subunit mRNA levels were increased from undetectable levels to marked elevations at 4 h. Since membrane fluidity was not associated with increased enzyme activity, we examined the possibility that hepatic regeneration altered Na,K-ATPase ␤-subunit synthesis and spatial location to the apical surface, which might result in activation of enzyme activity. If this hypothesis were correct, increased ␤-subunit levels in the apical domain should follow a time course similar to that for increased enzyme activity. The immunoblot density of the ␣-subunit in both the sinusoidal and BCM fractions did not significantly change, consistent with the measurement of ␣ 1 -subunit mRNA (Figs. 7A and 11A). Although the ␤-subunit density was not increased at the sinusoidal surface, its content progressively increased in the BCM fraction, following a time course consistent with changes in Na,K-ATPase activity. It is unlikely that the increase is due to contamination of BCM fractions with the SM since the increase was selective for the ␤-subunit and did not involve other markers of the SM, including the ␣-subunit and polymeric IgA receptor. In addition, these biochemical observations were confirmed by the morphological localization of the ␤-subunit to the BCM during regeneration using both immunofluorescence and immunogold labeling (Figs. 8 and 9). The ␤-subunit was localized to the SM surface equally throughout the liver lobule in control and regenerating liver. During regeneration, immunogold labeling of the ␤-subunit in the BCM was random and not specifically associated with cells undergoing mitosis. Loss of Na,K-ATPase activity and ␤ 1 -subunit polarity with regeneration is unique, for studies by Bartles et al. (65,66) have not detected loss of polarity using several sinusoidal and BCM antigens.
We have confirmed that hepatic regeneration is associated with increased ␤ 1 -but not ␣ 1 -subunit mRNA and extended the observation to demonstrate that the increase in ␤ 1 -subunit mRNA precedes the rise in ␤ 1 -protein levels, indicating a precursor-product relationship. Since it has been suggested that the ␤ 1 -subunit may be part of the mitogenic program initiating hepatic regeneration, we sought to determine the earliest time for its increase and if the ␤ 1 -subunit mRNA increase required de novo protein synthesis. ␤ 1 -Subunit mRNA levels did not significantly rise until 4 h post-partial hepatectomy, with a 20-fold increase (Figs. 11 and 12). Cycloheximide, administered to inhibit protein synthesis, completely blocked the rise in ␤ 1 -subunit mRNA levels. The time of maximum increase in ␤ 1 -subunit mRNA rise and its inhibition of increase with cycloheximide are consistent with ␤ 1 -subunit mRNA being a delayed-early gene rather than a mitogenic response. Increased ␤ 1 -subunit mRNA and Na,K-ATPase activity may be an adaptation to metabolic requirements. For example, sodiumdependent amino acid transport is increased in liver at approximately the same time as Na,K-ATPase activity peaks (67).
In summary, hepatic regeneration is associated with in-creased sodium pump activity from the prereplicative through the mitotic stages due to selectively increased ␤ 1 -subunit mRNA and protein in the BCM domain. Increased Na,K-ATPase activity was independent of alterations in plasma membrane fluidity, strongly suggesting that the ␤-subunit has a direct role in controlling pump function as suggested by others (68). That the Na,K-ATPase complex may be active in the BCM lipid environment should not be surprising since the ␣⅐␤-subunit complex is known to function at the apical pole of the choroid plexus and pigmented retinal cell (69). In liver, Na,K-ATPase activity can be regulated in the BCM domain by increasing only one subunit of the sodium pump, suggesting that the ␤-subunit may be involved in allosteric changes in the ␣-subunit, permitting enzyme activity independent of the membrane lipid environment.