Reconstitution of Galactosyl Receptor Inactivation in Permeabilized Rat Hepatocytes Is ATP-dependent*

A subpopulation of galactosyl receptors (GalRs) on isolated rat hepatocytes undergo a reversible inacti- vation and reactivation process during constitutive recycling Biochemistry 27, 2061-2069). Here, we report the reconstitution of this GalR inactivation in digitonin-permeabilized rat hepatocytes. Permeabilization of freshly isolated cells at 4 “C with 0.002% (w/v) digi- tonin releases cytosol containing 35-40% of the total cellular protein, 10-15% of a lysosomal marker, and 5-10% of an early endosomal marker. Incubation of permeabilized cells with cytosol at 37 “C results in a time-dependent reduction of total ’251-asialooroso-mucoid binding activity, which proceeds with first order kinetics (txh = 11.3 min). Only half of the total cellular GalRs are affected; maximal GalR activity loss, obtained by 30 min, is 50.5 f 9.5% (n = 21) of the control (4 “C) value. Increasing the digitonin concentration up to 0.055% does not increase the extent of inactivation. Permeabilized cells with reduced GalR activity were assessed for GalR protein content by Western blot analysis and by binding of anti-GalR antibody. The results show that the reduced “‘I-as- ialoorosomucoid

A subpopulation of galactosyl receptors (GalRs) on isolated rat hepatocytes undergo a reversible inactivation and reactivation process during constitutive recycling (McAbee, D. D., and Weigel, P. H. (1988) Biochemistry 27,[2061][2062][2063][2064][2065][2066][2067][2068][2069]. Here, we report the reconstitution of this GalR inactivation in digitoninpermeabilized rat hepatocytes. Permeabilization of freshly isolated cells at 4 "C with 0.002% (w/v) digitonin releases cytosol containing 35-40% of the total cellular protein, 10-15% of a lysosomal marker, and 5-10% of an early endosomal marker. Incubation of permeabilized cells with cytosol at 37 "C results in a time-dependent reduction of total '251-asialoorosomucoid binding activity, which proceeds with first order kinetics (txh = 11.3 min). Only half of the total cellular GalRs are affected; maximal GalR activity loss, obtained by 30 min, is 50.5 f 9.5% (n = 21) of the control (4 "C) value. Increasing the digitonin concentration up to 0.055% does not increase the extent of inactivation. Permeabilized cells with reduced GalR activity were assessed for GalR protein content by Western blot analysis and by binding of anti-GalR antibody. The results show that the reduced "'I-asialoorosomucoid binding is due to GalR inactivation rather than receptor protein degradation. GalR inactivation does not occur in the absence of cytosol or in the presence of dialyzed cytosol. The cytosol also loses its GalR inactivating ability in the presence of an ATPdepleting system. GalR inactivation in the absence of cytosol is achieved by incubating permeabilized washed cells at 37 "C with ATP but not with ADP, AMP, or other NTPs. The rate and extent of inactivation are dependent on the ATP concentration. Halfmaximal and maximal GalR inactivation are obtained a t 0.3 and 3.0 mM ATP, respectively. In the presence of cytosol, permeabilized hepatocytes could replenish cytosolic ATP by oxidative phosphorylation. As a result, similar levels of GalR inactivation were obtained with 500-fold lower ATP concentrations. We conclude that ATP is the only cytosolic component necessary for GalR inactivation in permeabilized rat hepatocytes.
Most class I1 receptors (1) are believed to undergo constitutive recycling (2)(3)(4)(5)(6). The biochemical events that drive the receptor recycling process and the itinerary of these receptors inside the cell are not yet clearly understood. The hepatic * This work was supported by National Institutes of Health Grant GM 30218. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ To whom correspondence should be addressed. asialoglycoprotein receptor or GalR' has been extensively characterized and serves as a convenient model for dissecting the endocytic pathway. Treatment of hepatocytes with functionally diverse agents such as metabolic energy poisons (6)(7)(8)(9), lysosomatropic amines (10)(11)(12)(13), microtubule depolymerizing drugs (14, 15), or proton ionophores (11,16,17) causes an inactivation and/or redistribution of cell surface GalRs and disrupts the recycling process (18). We have previously shown that in intact rat hepatocytes, depletion of ATP in the absence of ligand causes 50% of the cell surface GalRs to be trapped intracellularly in an inactive form (6,19). Restoration of cellular ATP to normal levels allows reactivation and reexpression of these GalRs on the cell surface, even in the absence of protein synthesis. Detailed kinetic analysis of GalR internalization, inactivation, reactivation, and externalization was performed using anti-GalR '2sI-IgG and "'I-ASOR to follow GalR protein and GalR activity, respectively. This study revealed that both the GalR inactivation and the GalR reactivation processes occur intracellularly (19). Additional evidence to support an intracellular site of GalR inactivation was obtained when hepatocytes were treated with hyperosmolar medium to inhibit coated pit formation (20) and disrupt internalization (21,22). In this case, GalRs failed to undergo inactivation in cells treated with monensin or chloroquine (23). Such an intracellular inactivation step during normal GalR recycling would explain why the segregation of dissociated receptor and ligand into different intracellular compartments for subsequent differential processing is so efficient.
Since the biochemical basis for GalR inactivation is not yet known, we have used digitonin-permeabilized rat hepatocytes to develop a cell-free in vitro system to study the requirements for GalR inactivation. At subsaturating concentrations, digitonin binds to and disrupts preferentially the cholesterolenriched plasma membrane (24). Fiskum et al. (25) have shown by electron microscopy that rat hepatocytes permeabilized with 0.005% digitonin maintain an extensive cytoskeleton and intact intracellular organelles. Such selective permeabilization allows one to bypass the permeability barrier across the plasma membrane with minimum disturbance of the normal intracellular structural organization. It is thus possible to monitor directly the responses of various intracellular organelles to exogenously added agents (26). In fact, such permeable cells are metabolically active and are capable of continuing complex processes such as lipid synthesis, oxidative phosphorylation, and secretion (27,28). Several inves-' T h e abbreviations used are: GalR, galactosyl receptor; ASOR, asialoorosomucoid; BSA, bovine serum albumin; HEPES, 4-(2-hy-droxyethy1)-1-piperazineethanesulfonic acid; EGTA, [ethylenebis(oxyethy1enenitrilo)ltetraacetic acid; SDS, sodium dodecyl sulfate; SDS-PAGE, sodium dodecyl sulfate-polyacrylamide gel electrophoresis. 8771 tigators have used permeabilized cells to study various cellular events including mobilization of calcium ion fluxes (29,30), modulation of enzyme activity (31,32), receptor phosphorylation (33,34), and the requirements for receptor-ligand dissociation (35,36). Here, we report the time, temperature-, and ATP-dependent partial inactivation of GalRs in permeabilized rat hepatocytes.
Preparation of Hepatocytes-Rat hepatocytes, isolated by a modification (40) of the collagenase perfusion procedure of Seglen (411, were suspended in medium 1/BSA (2 X lo6 cells/ml) and incubated at 37 "C for 1 h in a gyratory water bath to increase and stabilize the cell surface GalR number (42). Viable cells (>95%) selected by centrifugation at 350 X g through discontinuous Percoll gradients (7) were counted, washed, and resuspended in cytosolic buffer (2 X lo6 cells/ml).
Permeabilization of Hepatocytes and Preparation of Cytosol-To maintain an appropriate environment for intracellular organelles, the permeabilization of hepatocytes was done in cytosolic buffer to mimic the cytoplasm with respect to ion concentrations. Cells prepared as above were resuspended in cytosolic buffer (2 X lo6 cells/ml) with or without digitonin and incubated at 4 "C for 20 min. For most experiments (unless specified otherwise), 0.006% digitonin was used to obtain mildly permeabilized hepatocytes. The permeabilized cell SUSpension was either used directly, or the cells were first centrifuged at 350 X g for 5 min to remove the supernatant (cytosol), washed twice tion medium (2 X lo6 cells/ml). The cytosol, which has -800 pg of more with cytosolic buffer, and resuspended in the specified incubaprotein/ml, contains cytosolic components at roughly 1-2% of the concentration present in the cytoplasm of intact cells.
Antibody and Ligand Binding Assays-Cell suspensions prepared as above were incubated at either 4 or 37 "C with continuous gyration at 100 rpm. Aliquots (1 ml each) were removed at different times into 2 ml of ice-cold cytosolic buffer. The cell suspensions were then centrifuged, and the supernatant was removed by aspiration. To measure total cellular GalR activity, cell pellets were resuspended and incubated at 4 "C for 1 h in 500 p1 of medium l/BSA with 1.5 pg/ml "'I-ASOR (specific activity, -50-100 cpm/fmol) and 0.055% digitonin. The presence of 0.055% digitonin fully permeabilizes the cells and allows for the assessment of total (surface and internal) receptor number and activity (43). To measure total cellular antibody binding, cell pellets were resuspended and incubated for 1 h at 4 "C in 500 p1 of Hanks' with 100 pg/ml immune anti-GalR chicken IgY (for total binding) or preimmune chicken IgY (for nonspecific binding) with 7.5 mM EGTA and 0.055% digitonin. After removing unbound primary antibody with three washes each of 4 ml of Hanks' containing 7.5 mM EGTA, the cells were incubated for 45 min at 4 "C with 2 pg/ml of rabbit antichicken 12'I-IgG (specific activity -50-100 cpm/fmol). After incubation with either "'I-ASOR or '*'I-IgG, the cells were washed three times with 4 ml each of Hanks', and the final cell pellets were resuspended in either DNA assay buffer or 0.3 N NaOH. Aliquots were then assessed for radioactivity and assayed for DNA or protein content. Nonspecific 12'II-ASOR binding, determined by the radioactivity remaining after stripping specifically bound ligand with EGTA, was only 5-10% of the total in all cases. Nonspecific lZ51-IgG binding determined by using preimmune chicken IgY as the primary antibody was 25-35% of the total. All binding assays were done in duplicate, and the error bars indicate the sample standard deviation.

SDS-PAGE and Western Blot Analysis-
The procedure was the same as that described previously (38). Permeabilized hepatocytes incubated at either 4 or 37 "C were washed at 4 "C and extracted with 1% Triton X-100. The solubilized GalR was immunoprecipitated using anti-receptor goat IgG-Sepharose, washed three times with BIC 10 buffer, resuspended in 150 p1 of the same buffer, dissociated from the Sepharose by boiling for 5 min, and separated by SDS-PAGE after reduction with 5% 2-mercaptoethanol. The proteins were then transferred to nitrocellulose and processed for Western blotting using anti-receptor chicken IgY and alkaline phosphatase conjugated to rabbit anti-chicken IgG as the primary and secondary antibodies, respectively. The receptor bands were visualized by incubation with 0.225 mg/ml 5-bromo-4-chloro-3-indolylphosphate and 0.45 mg/ml nitro blue tetrazolium as described previously (38).
General-Cell viability was determined by the ability to exclude trypan blue. ATP concentrations were quantitated by a modification (44) of the lucerifin-luciferase procedure of Stanley and Williams (45). Protein content was measured by the method of Bradford (46) using BSA as a standard. N-Acetylglucosaminidase was measured by a colorimetric assay using p-nitrophenyl-N-acetylglucosaminide as the substrate and p-nitrophenol as the standard (47). Cellular DNA was determined spectrofluorometrically by the DNA binding dye Hoechst 33258 procedure of Labarca and Paigen (48) with calf thymus DNA as the standard. '251-Radioactivity was determined using a Packard Multiprias 2 y-spectrometer.

RESULTS
GalR Inactivation Occurs in Permeable Cells-When hepatocytes were permeabilized at 4 "C with 0.006% digitonin and then directly incubated at 37 "C, there was a time-dependent reduction in their ability to bind JZ5I-ASOR when subsequently assayed at 4 "C ( Fig. 1). This reduction in GalR activity was not seen after incubAion of permeabilized cells a t 4 "C or in intact cells incubated at either 4 or 37 "C. The loss of ligand binding activity proceeded with first order kinetics and a tlh of 11.3 min (Fig. 1). GalR activity loss was almost complete by about 30-45 min and, in a large number of experiments, resulted in 50.5 k 9.5% ( n = 21) reduction in ligand binding compared with controls incubated a t 4 "C. Extended incubation times of up to 2 h did not show any further reduction in GalR activity. Several alternative explanations for the decreased GalR activity were investigated. For example, degradation of GalR by proteases released due to permeabilization of lysosomes could result in decreased binding. However, addition of a mixture of protease inhibitors including phenylmethylsulfonyl fluoride, N-ethylmaleimide, q , , j EDTA, pepstatin, leupeptin, and bacitracin during permeabilization and the subsequent 37 "C incubation gave similar results (not shown). Degradation by proteases is therefore probably not responsible for the decreased binding. Another possibility is that partially processed glycoproteins with terminal galactosyl residues could be released from the Golgi as a result of permeabilization. Binding of these endogenous ligands to GalR would then interfere with the subsequent binding of '"'I-ASOR. However, an EGTA wash a t 4 "C to remove such endogenously bound ligands also did not prevent the GalR activity loss (not shown).
T o obtain more quantitative data in support of bona fide GalR inactivation rather than protein degradation, cells were assayed for both '2sI-ASOR and anti-GalR IgG binding as a measure of GalR activity and GalR protein, respectively (Fig.  2). Almost identical amounts of '"'I-IgG bound to both intact and permeabilized cells incubated a t either 4 or 37 "C. However, permeabilized cells treated a t 37 "C bound only -40% of the "'I-ASOR bound by intact cells or by permeabilized cells incubated a t 4 "C. These results provide evidence for GalR inactivation since there is no loss of GalR protein but substantially reduced GalR activity in mildly permeabilized cells incubated a t 37 "C in the presence of cytosol. The results also show that the chicken anti-GalR antibody specifically binds to both active and inactive GalR protein.
Partial degradation of GalR protein could eliminate ligand binding without interfering with antibody binding. This possibility was examined by Western blot analysis (Fig. 3). GalR was immunoprecipitated with goat anti-receptor IgG from extracts of permeabilized cells incubated a t either 4 or 37 "C. Although cells treated a t 37 "C had only 42% of the "'I-ASOR binding activity of those a t 4 "C, the positions and the staining intensity of the three GalR subunits were identical in both cases, showing that there was no detectable GalR degradation. Based on these observations, we conclude that incubation of permeabilized unwashed cells at 37 "C causes inactivation of GalRs. This inactivation does not require the cytosolic buffer, since similar results were obtained when buffer 1 was used for permeabilization and the subsequent 37 "C incubation (not shown).
Only One of Two GalR Subpopulations Is Inactivated in Vitro-In the experiments described above, there was always only a 40-60% reduction in total cellular GalR activity. The average extent of GalR inactivation, calculated from 21 experiments, was 50.5 & 9.5% of the control. To address the possibility that GalR inactivation may depend on the extent of permeabilization, hepatocytes were treated with different concentrations of digitonin. The extent of permeabilization of various intracellular compartments was assessed using different markers (Fig. 4A), and the extent of GalR inactivation induced a t 37 "C was also measured at each digitonin concentration (Fig. 4B). At <0.001% digitonin, there was no significant permeabilization of hepatocytes. Between 0.001 and 0.002% digitonin, the extent of plasma membrane permeabilization increased very sharply as evidenced by the increase in the percentage of total protein released and the decrease in cell viability. Maximum loss of protein (40% of total cellular content) was seen a t 20.002% digitonin. Permeabilization of lysosomes started a t 0.002% digitonin with a 15% loss of the lysosomal marker enzyme, N-acetylglucosaminidase. The percentage release steadily increased with the concentration of digitonin. Significant amounts of an early endosomal marker (internalized "'I-ASOR) were released only a t concentrations of >0.003% digitonin. At 37 "C, -50% of the total cellular GalR activity was lost a t 0.002% digitonin. This extent of GalR inactivation did not  Samples were also chilled to 4 "C and centrifuged, and the amounts of protein (V) and N-acetylglucosaminidase ( 7 ) released into the supernatant and retained in the cell pellet were measured. To estimate the extent of permeabilization of early endosomes (O), intact cells were first allowed to internalize surface-bound '251-ASOR a t 37 "C for 5 min (as in Ref. 7 ) before chilling to 4 "C and washing with medium 1 containing 7.5 mM EGTA to remove residual surface-bound ligand. Cells were then washed and resuspended in cytosolic buffer and permeabilized with the indicated concentration of digitonin a t 4 "C for 20 min. After permeabilization, the supernatant and cells were separated by centrifugation. '2sII-ASOR in permeabilized vesicles but still bound to receptor was released by washing the pellet once with 7.5 mM EGTA. The first supernatant and the EGTA wash were combined, and radioactivity was determined. This value is expressed as the percentage of the total internalized radioactivity. B, aliquots of permeabilized cells were further incubated a t 4 (0) or 37 "C (0) for 40 min and were assessed for total cellular GalR activity as described under "Experimental Procedures." increase substantially with increasing concentrations of digitonin and did not correlate with the amount of lysosomal or endosomal markers released. Even a t 0.005% digitonin, when most intracellular vesicles were permeabilized and almost 100% of both lysosomal and endosomal markers were released, the extent of inactivation was still -50% (not shown). This latter result also suggests that GalR degradation by lysosomal proteases is not responsible for the decreased ligand binding activity of GalR.
The above results indicate that only a portion of the total cellular GalR content can be inactivated in the in vitro permeable cell system. This partial inactivation could reflect the behavior of two different GalR subpopulations. We have previously characterized two functionally distinct GalR populations in isolated rat hepatocytes, which we have termed State 1 and State 2 GalR (49-51). These two GalRpopulations process internalized ligand in two different intracellular pathways, and their activity is differentially modulated by a variety of drugs, including azide (52). The State 2 GalRs are inactivated and accumulate inside the cell in azide-treated cells (19,52). State 1 GalRs remain active under these conditions. It is possible, therefore, that only one of these two GalR subpopulations (the State 2 GalRs) becomes inactivated in the permeable cells. To test this, intact cells were treated at 37 "C with sodium azide to down-modulate the State 2 GalRs. When these cells were subsequently washed, permeabilized at 4 "C, and then incubated at 37 "C in the presence of cytosol, no GalR inactivation occurred (Fig. 5). Furthermore, the control (non-azide-treated) permeable cells lost an amount of GalR activity in vitro equal to what the azide-treated cells had lost in vivo. We conclude that GalR inactivation in the permeable cells represents the loss of State 2 GalR activity; State 1 GalR activity appears to be unaffected. ATP Is the Only Cytosolic Factor Required for GalR Inactivation-To identify the components responsible for inactivation, permeabilized cells and the released cytosol were separated from each other. GalRs in permeabilized cells that had been washed twice to remove cytosolic components and then resuspended in cytosolic buffer were not inactivated after incubation at 37 "C ( Table I). Incubation of these washed cells with cytosol resulted in GalR inactivation, indicating that at least one inactivating factor(s) is cytosolic. The permeabilized washed hepatocytes thus provide a convenient system to reconstitute the GalR inactivation process. When cytosol was extensively dialyzed at 4 "C, it failed to cause GalR inactivation in permeabilized washed hepatocytes ( Table I). Storage of cytosol at 4 "C for the same length of time only marginally decreased its GalR inactivating ability, suggesting that the inability of dialyzed cytosol to inactivate GalR was due to depletion rather than inactivation of a factor(s). This result indicates that a relatively small (molecular mass < 10 kDa) cytosolic component is required for GalR inactivation. Since any proteases, if present, would still be in the dialyzed cytosol, this result also corroborates the conclusion that GalR degradation by proteases is not responsible for the decreased GalR activity.
The finding that a cytosolic factor is responsible for GalR inactivation suggests that the resistance of a fraction of the total cellular GalRs to inactivation, as discussed above, may be due to the depletion or denaturation of a cytosolic inactivating factor during the incubation at 37 "C. To address this possibility, permeabilized hepatocytes were first incubated at 37 "C with cytosol. After this first incubation, only 50-60% of the total cellular GalRs were active. These cells were washed and then again incubated at 37 "C with fresh cytosol.  In two different experiments, hepatocytes (2 X lo6 cells/ml) were incubated for 20 min at 4 "C in cytosolic buffer with or without 0.006% digitonin. After an aliquot (unwashed cells) was removed, the permeabilized cells were centrifuged and the cytosol was saved. The cell pellet was washed twice with cytosolic buffer and resuspended at immediately, stored at 4 "C for 20 h, or dialyzed against 0 . 0 1~ cytosolic buffer at 4 "C for 16 h followed by dialysis against IX cytosolic buffer for 4 h at 4 "C. Aliquots of intact cells, permeabilized unwashed cells, or permeabilized washed cells were resuspended at 2 X lo6 cells/ml in either cytosolic buffer, cytosol, dialyzed cytosol, or stored cytosol and incubated at 37 or 4 "C for 30 min. Total cellular GalR activity was then determined as described under "Experimental Procedures." Results are expressed as a percentage of the control activity remaining after incubation at 4 "C. FIG. 6. Effect of cytosolic ATP depletion on GalR inactivation in permeable cells. Hepatocytes were permeabilized as in Fig.  1 and centrifuged, and the cytosol was removed. The cells were washed twice with and resuspended at 4 X lo' cells/ml in cytosolic buffer. The cells were then diluted to 2 X lo6 cells/ml in cytosolic buffer (O), cytosol (A), cytosol preincubated at 37 "C with 25 units/ml hexokinase, and 5 mM glucose for 10 min (+) or in cytosolic buffer containing 7.5 mM ATP (W) and incubated at 37 "C. At the indicated times, samples were removed and total cellular GalR activity was determined.
This second incubation caused inactivation of only a small fraction of the residual active GalRs. The final GalR activity was still about 43% of the initial amount before the first incubation. This result indicates that depletion of an inactivating factor is not responsible for the ability of only roughly half of the total GalRs to maintain their activity, and it is consistent with the above conclusion that the sensitive and insensitive GalRs are different. To address the possible involvement of ATP, GalR inactivation was assessed in the presence of either ATP alone or ATP-depleted cytosol (Fig. 6). As expected, incubation of permeabilized hepatocytes at 37 "C with cytosol resulted in a rapid loss of '"I-ASOR binding activity with -50% inactivation by 30 min. However, when the cytosol was first depleted of ATP by using hexokinase and glucose and then incubated with the permeabilized washed cells, still in the presence of the ATP depleting system, the cytosol was unable to support GalR inactivation. Addition of glucose alone did not significantly alter the extent of GalR inactivation. However, hexokinase alone was able to partially block GalR inactivation (not shown). This can be explained because the K,,, of hexokinase for glucose is low (-8-30 p~) , and the endogenously available glucose is probably sufficient to support ATP depletion by hexokinase. ATP levels in cytosol were quantitated in the presence and absence of hexokinase and confirmed the ability of hexokinase to deplete ATP. Since ATP-depleted cytosol is no longer capable of supporting GalR inactivation, the absolute requirement for ATP by this process was suggested. When permeabilized washed hepatocytes were incubated a t 37 "C in the absence of cytosol but in the presence of 7.5 mM ATP, GalR activity loss occurred with kinetics similar to that obtained with cytosol (Fig. 6). Thus, ATP alone was sufficient to cause GalR inactivation. These results, taken together, indicate that ATP may be the only cytosolic component required for GalR inactivation.
The specificity of the requirement for ATP was tested by substituting other nucleoside triphosphates for ATP. Incubation of permeabilized washed hepatocytes at 37 "C with 3 mM ATP resulted in inactivation of 40-50% of total cellular GalRs by 20-30 min. However, incubation of the permeabilized washed cells with any of the other NTPs, ADP, or AMP did not result in any significant change in GalR activity (Fig. 7). Thus, other nucleotides cannot support GalR inactivation, and the ATP requirement for GalR inactivation in permeabilized cells is specific and absolute.
The extent and rate of ATP-induced GalR inactivation were dependent on the concentration of ATP (Fig. 8). The ATP dose-response curves for the rate and extent of GalR inactivation are almost identical. In the absence of ATP, permeabilized washed hepatocytes bound -800 fmol of ASOR/106 cells. At each ATP concentration tested, GalR inactivation continued at a linear rate for up to 45 min. The maximal rate of inactivation, obtained at -3.0 mM ATP, was 8 fmol of ASOR binding sites (1%) lost/min/1O6 cells. A halfmaximal rate of GalR inactivation was obtained at -0.3 mM ATP. The extent of GalR inactivation, measured after 45 min of incubation a t 37 "C, also increased with increasing concentrations of ATP. A half-maximal extent of GalR inactivation was obtained at -0.3 mM ATP. Maximal extent of inactivation, obtained at -3.0 mM ATP was 55% of the total cellular activity. Further increasing the ATP concentration to 7.5 mM did not significantly increase the extent or rate of GalR  Correlation coefficients ranged from -0.91 to -0.99.
inactivation. Thus, analogous to the results with cytosol, -50% of the total GalR population is inactivated and -50% is resistant to ATP-induced inactivation.
Cytosol Decreases the ATP Concentration Needed for GalR Inactiuation-From the above results, it is clear that cytosolic ATP is essential for GalR inactivation. However, the concentration of ATP in cytosol, determined by a luciferin-luciferase assay, is only -5-6 p~, and ATP by itself, at such low concentrations, does not cause any significant loss of GalR activity in washed permeabilized cells (Fig. 8). Nonetheless, the extent of GalR inactivation induced by cytosol was comparable to that caused by 3.0 mM ATP. Thus, in the absence of other cytosolic components, a -500-fold greater ATP concentration is required to attain similar levels of GalR inactivation. When washed, permeable hepatocytes are incubated with millimolar amounts of ATP at 37 "C, the ATP is hydrolyzed, and, consequently, its concentration steadily decreases with time until a much lower steady state value is reached (not shown). These observations suggest that the ATP-dependent GalR inactivation in the presence of cytosol may be enhanced by other cytosolic components.
Preincubation of cytosol a t 37 "C for 60 min resulted in the complete hydrolysis of cytosolic ATP as determined by the subsequent quantitation of ATP. However, such ATP-depleted cytosol still supports GalR inactivation (Table 11), and ATP resynthesis is seen within minutes at 37 "C. This indicates that in the presence of cytosol, the permeable cells are capable of replenishing cytosolic ATP by either glycolysis or oxidative phosphorylation. This was found to be the case, since GalR inactivation by ATP-depleted cytosol was almost completely blocked when oxidative phosphorylation was inhibited by the addition of 1 mM atractyloside (Table 11) or 2 P M carbonyl cyanide p-trifluoromethoxyphenylhydrazone. Similarly, GalR inactivation was achieved in the absence of either ATP or cytosol by supplying the substrates for oxidative phosphorylation; uiz. inorganic phosphate, ADP, glutamate, and malate (Table 11). None of these substrates by itself was able to support GalR inactivation. It is also worth noting that the extent of GalR inactivation obtained with 0.5 mM ADP in the presence of these substrates was comparable to that obtained with 5 mM ATP. Thus, the continued resynthesis of ATP in digitonin-permeabilized hepatocytes is efficient and explains why lower concentrations of ATP are

Oxidative phosphorylation facilitates GalR inactivation i n permeabilized hepatocytes
Isolated hepatocytes were permeabilized as in Table I, cytosol was saved, and the cells were washed twice and resuspended a t 4 X lo7 cells/ml in cytosolic buffer. The cells were then diluted to a density of 2 X lo6 cells/ml in cytosol or cytosolic buffer containing the indicated additions. After incubation at 4 (control) or 37 "C for 40 min (Experiment 1) or 30 min (Experiment 2), aliquots of the cell suspension were removed, washed, and assessed for total cellular GalR activity as described under "Experimental Procedures." The results obtained for cells incubated a t 37 "C are expressed as a percentage of the corresponding cells incubated at 4 "C (control required in the presence of cytosol to achieve the same level of GalR inactivation compared to ATP alone.

DISCUSSION
The activities of several class I (1) receptors have been shown to be reversibly regulated. Vasopressin and angiotensin I1 receptors on rat hepatocytes are inactivated by a Ca2+dependent cytosolic protein. This inactivation is reversed by lowering the pH to 5.5 but not by EGTA (53). The aromatic hydrocarbon receptor in Hepa-1 cells is reversibly inactivated in a time-and temperature-dependent manner in the presence of cytochalasin B. The extent of the loss of ligand binding activity correlates with the depletion of cellular ATP. Reactivation is readily achieved by incubation of cells at 37 "C after removing cytochalasin B (54). The activity of the pregnenolone receptor in guinea pig adrenocortical cytosol requires phosphorylation by a cytosolic kinase. Alkaline phosphatase treatment at pH 9.0 causes its reversible inactivation (55). The calf uterus estradiol-17 @-receptor is also active in the phosphorylated form and inactive when it is dephosphorylated. A nuclear phosphatase and a cytosolic kinase are responsible for the reversible inactivation and reactivation process, respectively (56).
In addition to our reports on inactivation of the hepatic GalR (6,19,52), evidence has also been presented by others for the presence or generation of inactive transferrin (57,58) and mannose (59) receptors. In the case of these class I1 recycling receptors, which mediate the continuous endocytosis of large amounts of ligand, receptor inactivation may be an important cellular strategy to ensure the efficient segregation of receptor and ligand after their internalization and dissociation. Internalized ligand can be concentrated -104-fold in endosomes relative to the extracellular concentration. This high ligand concentration could drive the reassociation of ligand and receptor, even though the lower endosomal pH decreases the affinity constant for receptor-ligand complex formation. In other words, the decreased affinity of receptor for ligand at lower pH could be offset by the dramatically increased ligand concentration in the endosome. This would result in the return of occupied receptors to the cell surface; such receptors would, essentially, be nonfunctional. Transient inactivation of receptors during the segregation of free receptor and ligand would ensure that this separation is efficient. Additionally, receptor inactivation may be an alternative mechanism to mediate receptor-ligand dissociation. Receptor inactivation during constitutive recycling could also provide a mechanism for cells to regulate their endocytic and recycling processes.
The molecular mechanism responsible for the reversible inactivation of GalRs during constitutive recycling (6, 19) is not clear. GalR internalization is a prerequisite for its inactivation, since hyperosmolarity, which stops internalization, also protects against GalR inactivation by chloroquine and monensin (23). The intracellular accumulation of inactive GalRs in ATP-depleted intact cells indicates that both the GalR reactivation and GalR externalization processes require ATP. When cellular ATP levels are restored, the intracellular GalR activity returns to normal levels prior to the cell surface GalR activity, suggesting that GalR reactivation also occurs intracellularly (19). In order to understand the physiological significance of these observations, it is important to identify the biochemical events contributing to the regulation of GalR activity.
Here, we have used permeabilized cells as an in vitro system to reconstitute the GalR inactivation observed in intact cells. Incubation of permeabilized hepatocytes at 37 "C in the presence of cytosol results in a reduction in GalR activity. This activity loss represents a real rather than trivial GalR inactivation for several reasons. The activity loss occurs without affecting the ability of the cells to bind anti-GalR antibody. A battery of protease inhibitors did not prevent GalR activity loss. Western blot analysis also shows that equal amounts of GalR with identical subunit sizes are immunoprecipitated from permeabilized control cells or cells in which 60% of all GalRs are inactivated. The extent of reduction in GalR activity does not correlate with the extent of release of lysosomal enzymes. Since GalR activity loss does not occur with dialyzed cytosol, which still contains most of the soluble cell proteins, degradation by proteases is not likely responsible for the decreased GalR activity. These results provide convincing evidence that this loss of ligand binding activity is due to GalR inactivation rather than GalR protein degradation. Further evidence that the observed GalR inactivation is physiologically relevant is the finding that ATP is required for inactivation. GalRs on permeabilized washed hepatocytes are inactivated without cytosol by incubation in the presence of ATP at 37 "C; other cytosolic components are not essential to reconstitute this GalR inactivation. Other membranebound enzymes and/or proteins must interact with ATP and GalR to cause its inactivation, since ATP by itself does not affect the activity of purified GalR.' Even though ATP could cause GalR inactivation independently of the presence of cytosol, a -500-fold higher concentration of ATP than present in the cytosol was required to cause maximal GalR inactivation. This apparent inconsistency is explained by the demonstration that permeabilized washed hepatocytes are able to resynthesize ATP continuously in the presence of cytosol. Since mitochondria in mildly permeabilized rat hepatocytes are still intact (25), they are functional and can support ATP synthesis by oxidative phosphorylation. The cytosol provides enzymes, substrates, and cofactors to support glycolysis and oxidative phosphorylation. This was shown by reconstituting GalR inactivation in permeabilized washed heptocytes in the absence of cytosol or added ATP by providing substrates for oxidative phosphorylation. In the absence of cytosol or in the presence of dialyzed cytosol, there is no oxidative phosphorylation because of the unavailability of substrates. As a result, much higher concentrations of ATP are required for a comparable extent of GalR inactivation when ATP regeneration cannot occur. Our results are consistent with those of Katz and Wals (27), who reported that mitochondria in hepatocytes permeabilized with 0.1% digitonin for 1-2 min at room temperature are capable of oxidative phosphorylation. All of these results suggest that cytosolic components facilitate GalR inactivation only by maintaining an appropriate cytosolic ATP concentration and that ATP alone is necessary and sufficient for GalR inactivation in permeabilized rat hepatocytes. In intact hepatocytes, GalR reactivation appears to require ATP (6,19). However, it is possible that both processes, GalR inactivation and its reactivation, require ATP, but with different K,,, values for ATP. Thus, in intact cells, if GalR inactivation has a very low K,, the process will still occur even in azide-treated, ATP-depleted cells. A residual cellular ATP content of 5% still represents -150 PM ATP. The reactivation process may have a substantially higher K , for ATP and would then be blocked in ATP-depleted intact cells.
Efforts to reconstitute GalR reactivation in the permeable cells have so far been unsuccessful. This could be explained if GalR reactivation requires cytosolic components that are either inactive or present at lower concentrations than required.
In the in vitro permeable system to reconstitute GalR inactivation, there is always only a 40-60% loss of GalR activity. This is not due to limiting amounts of ATP. This resistance of -50% of the total cellular GalRs to inactivation is consistent with earlier studies demonstrating the presence of two subpopulations of GalR on rat hepatocytes (49)(50)(51). In intact cells exposed to various drugs and metabolic poisons, only -50% of the total cellular GalRs, corresponding to the State 2 GalR subpopulation, are inactivated (52). The partial inactivation of GalRs in the permeable cells is also due to these State 2 GalRs. Only if active State 2 GalRs are present does GalR inactivation in the in vitro system occur. When azide-treated intact hepatocytes, which express only active State 1 GalRs, were permeabilization and incubated at 37 "C in the presence of cytosol no further GalR inactivation occurred. We propose that the in vitro inactivation of GalRs in permeabilized cells reconstitutes part of a normally occurring GalR inactivation and reactivation process. The biochemical changes induced by ATP that result in GalR inactivation still remain to be explained. Possible mechanisms may include phosphorylation of either the GalR or other regulatory membrane components. The ATP requirement may also be to support energy-dependent mechanisms, such as endosomeendosome fusion (60) or the maintenance of normal proton gradients across endosomal membranes (61). The in vitro system described here should be extremely useful in answering these and other questions regarding the regulation of GalR activity and function.