Turnover of plasma membrane proteins in rat hepatoma cells and primary cultures of rat hepatocytes.

The half-lives of turnover of plasma membrane proteins in rat hepatoma tissue, culture cells, and in primary cultures of rat hepatocytes have been analyzed after resolution by two-dimensional gel electrophoresis. Cell membranes were externally labeled via iodination catalyzed by lactoperoxidase and glucose oxidase. A bimodal pattern of turnover was found for the externally oriented plasma membrane proteins of rat hepatoma cells. Three glycoproteins analyzed in these cells had an average t 1/2 of 22 h while eight proteins which did not bind to concanavalin A had an average t 1/2 of 80 h. In contrast, more heterogeneous rates of turnover were found for the externally oriented plasma membrane proteins of primary cultures of hepatocytes. Most, if not all, of the membrane proteins accessible to iodination in these cells were glycoproteins. Among the glycoproteins resolved by two-dimensional polyacrylamide electrophoresis, the receptors for asialoglycoproteins had the shortest half-lives (18 h). Other glycoproteins, mostly with higher molecular weights and different isoelectric points, showed a spectrum of half-lives ranging from 16 to 99 h. The turnover rates of membrane proteins of primary cultures of rat hepatocytes were also determined with [3H]- and [35S]methionine labeling of cells. Heterogeneous rates of turnover again were found among the labeled glycoproteins and nonglycoproteins. Among the 10 glycoproteins individually analyzed, the half-lives range from 17 to 67 h. Among the 21 proteins which do not bind to concanavalin A, the half-lives range from 18 h to more than 100 h. Three proteins analyzed showed an apparent biphasic pattern of turnover, having a fast phase with a half-life of 4-6 h and a slow phase with a half-life of 15-29 h. Several nonglycoproteins, including clathrin and actin associated with membrane vesicles had extremely long half-lives. The more than 5-fold difference in the half-life between clathrin and the receptors for asialoglycoproteins, which coexist in coated pits indicates that intrinsic proteins of the coated pits turn over at a different rate than peripheral components.

Turnover of Plasma Membrane Proteins in Rat Hepatoma Cells and Primary Cultures of Rat Hepatocytes* (Received for publication, October 11, 1983) Fong-Fong Chu and Darrell Doyle From the Department of Biological Sciences, State Uniuersity of New York a t Buffalo, Buffalo, New York 14260 The half-lives of turnover of plasma membrane proteins in rat hepatoma tissue, culture cells, and in primary cultures of rat hepatocytes have been analyzed after resolution by two-dimensional gel electrophoresis. Cell membranes were externally labeled via iodination catalyzed by lactoperoxidase and glucose oxidase. A bimodal pattern of turnover was found for the externally oriented plasma membrane proteins of rat hepatoma cells. Three glycoproteins analyzed in these cells had an average t1/2 of 22 h while eight proteins which did not bind to concanavalin A had an average In contrast, more heterogeneous rates of turnover were found for the externally oriented plasma membrane proteins of primary cultures of hepatocytes. Most, if not all, of the membrane proteins accessible to iodination in these cells were glycoproteins. Among the glycoproteins resolved by two-dimensional polyacrylamide electrophoresis, the receptors for asialoglycoproteins had the shortest half-lives (18 h). Other glycoproteins, mostly with higher molecular weights and different isoelectric points, showed a spectrum of halflives ranging from 16 to 99 h.
The turnover rates of membrane proteins of primary cultures of rat hepatocytes were also determined with [3H]and [35S]methionine labeling of cells. Heterogeneous rates of turnover again were found among the labeled glycoproteins and nonglycoproteins. Among the 10 glycoproteins individually analyzed, the halflives range from 17 to 67 h. Among the 21 proteins which do not bind to concanavalin A, the half-lives range from 18 h to more than 100 h. Three proteins analyzed showed an apparent biphasic pattern of turnover, having a fast phase with a half-life of 4-6 h and a slow phase with a half-life of 15-29 h. Several nonglycoproteins, including clathrin and actin associated with membrane vesicles had extremely long half-lives. The more than &fold difference in the half-life between clathrin and the receptors for asialoglycoproteins, which coexist in coated pits indicates that intrinsic proteins of the coated pits turn over at a different rate than peripheral components. The ways by which the mammalian cell regulates the concentration of specific proteins in its plasma membrane, although fundamentally important, are not well understood. One of the problems that needs to be solved in this context is * This work was supported by Grant GM3403201 from the National Institutes of General Medical Sciences and Grant CA3877301 from the National Cancer Institute. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore he hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
to unravel the basic mechanism by which plasma membrane proteins are assembled, inserted into the membrane, and eventually removed from the membrane or degraded. The turnover of plasma membrane proteins is easier to examine than the proteins of most other intracellular organelle membranes because they can be labeled externally. Hence, one can follow the fate of the labeled proteins without going through the laborious steps involved in subcellular fractionation. A major question concerning the turnover of the plasma membrane is whether the membrane or subdomains of the membrane behave as units with respect to turnover or whether a mosaic pattern or heterogeneity of turnover is exhibited by the constituents of this complex cell organelle. Heterogeneous rates of turnover have been found for some of the proteins from the total plasma membrane fraction of macrophages, hepatocytes, and Chinese hamster ovary cells (1-3). However, more homogeneous turnover has been found for the proteins of the endocytic membrane fractions of macrophages, fibroblasts, and Chinese hamster ovary cells (4-6). This suggests that unit turnover may occur in subdomains of the plasma membrane, particularly those domains involved in endocytosis.
In a rat hepatoma tissue culture (HTC') cell line, the bulk of glycoproteins are turned over faster than the minimally glycosylated proteins (proteins which do not bind to concanavalin A) of the membrane (7,8). But, most of the proteins present in either the glycoprotein or the non or minimally glycosylated protein classes turn over at equivalent rates relative to other members in the class. Primary cultures of hepatocytes, in contrast to HTC cells, are well differentiated and still maintain the typical structural segregation of the cell membrane into the sinusoidal face, the bile canalicular face, and intercellular junctions. Receptors for hormones, growth factors, and asialoglycoproteins are concentrated on the sinusoidal surface of these cells. 5"Nucleotidase is concentrated on the bile canalicular surface (9), and junctional proteins are present on the intercellular surface. A relatively long half-life was found for bulk liver plasma membrane proteins when examined in situ in the animal and in cell cultures (10, 11). The turnover of proteins of the hepatocyte proceeds with heterogeneous rates. The gap junctional proteins have a reported half-life between 5 and 19 h (12,13). Insulin receptors have a half-life of 10 h (14), while receptors for asialoglycoproteins have a half-life of 20 h when measured in primary cell culture and 80 h when measured in the intact animal (10, 11). A 110,000-dalton glycoprotein isolated from rat liver plasma membrane was reported to have a half-life of The abbreviations used are: HTC, hepatoma tissue culture; SDS, sodium dodecyl sulfate; PMSF, phenylmethylsulfonyl fluoride; EGTA, ethylene glycol his@-aminoethyl ether)-N,N,N',N'-tetraacetic acid; Tricine, N-[2-hydroxy-l,l-bis(hydroxymethyl)ethyl]-glycine; HEPES, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid.

3097
70-78 h for the peptide backbone while terminal sugars such as fucose, galactose, and sialic acid residues had half-lives of 12.5, 20, and 33 h (15) with respect to turnover. There are at least two possibilities to explain these heterogeneous rates of turnover among plasma membrane proteins of the hepatocytes. One, since the plasma membrane of rat hepatocytes has many diverse physiological functions and many of these functions are segregated into distinct domains of the membrane, homogenous turnover could occur for those proteins localized in a specific area of the membrane and performing a single function, e.g. receptor proteins in the coated pit or gap junction proteins between cells or bile canalicular proteins at the appositions of cell membranes. Or, proteins may turnover according to characteristic features of their structure including extent of glycosylation. The purpose of this study is to begin to attempt to distinguish among these possibilities.

EXPERIMENTAL PROCEDURES
Cells-A cloned line of HTC cells originally derived from a hepatoma in Buffalo rats, was kindly provided by Dr. T. D. Gelehrter (University of Michigan). Cells were grown in Earle's minimal essential medium for suspension culture (GIBCO), supplemented with 10% fetal calf serum, 50 mM Tricine, 0.1% sodium bicarbonate, and 50 pg/ml of neomycin and kanamycin. Cells were maintained in spinner culture, and were used as such in all experiments.
Hepatocytes were prepared from male Buffalo rats, 200-300 g, according to a modified method of Seglen (16). Briefly, 200 ml of perfusion buffer containing 0.14 mM NaCI, 6.7 mM KC1, and 10 mM HEPES, pH 7.4, was used to perfuse the liver. Then hepatocytes were isolated by in situ perfusion of 200 ml of collagenase buffer, containing 0.1 g of type I collagenase (Sigma), 67 mM NaCI, 6.7 mM KC1, 5 mM CaC12, and 0.1 M HEPES, pH 7.6. After isolation, cells were washed in Dulbecco's minimal essential medium (GIBCO) supplemented with 10% heat-inactivated fetal calf serum, 2.5 mM HEPES, penicillin a t 100 pg/ml, and streptomycin a t 100 pg/ml. Cells were iodinated in suspension at this point. Collagenase treatment of cells or iodination had no significant effect on either the isoelectric point or the molecular weight of the plasma membrane proteins analyzed herein, since unlabeled membrane proteins and glycoproteins isolated directly from liver had the same mobility in two-dimensional gel electrophoresis as their counterparts isolated from collagenase treated and iodinated hepatocytes. The iodinated cells were plated on biomatrix-coated plates at a density of 2-3 X lo6 cells/lOO-mm Petri dish (Corning).
The biomatrix was fractionated from mouse liver according to the method of Rojkind et al. (17). Cells were maintained in washing medium without HEPES in a 5% CO, incubator. Cell viability as determined by trypan blue exclusion was greater than 95% for HTC cells and greater than 90% for hepatocytes throughout the experiments.
Iodination-Cells were iodinated according to the method of Tweto et al. (18) with the exception that carrier iodide was omitted. For HTC cells, 10' washed cells were suspended in 1.2 ml of phosphatebuffered saline without calcium and magnesium, supplemented with 50 mM Tricine, pH 7.6, 20 mM glucose, 80 pg of lactoperoxidase (Sigma), 40 pg of glucose oxidase (Calbiochem-Behring), and 2.5 mCi of Na'T or 5 mCi of Na13'I (Amersham Corp.). Iodination was performed for 30 min at 4 "C or room temperature as indicated in each experiment. For hepatocytes, 5 X lo7 washed cells were resuspended in 1 ml of Hanks' buffer (GIBCO) supplemented with 2.5 mM of HEPES, pH 7.4, 50 pg of gentamycin, 50 mM of glucose, 200 pg of lactoperoxidase, 100 ,ug of glucose oxidase, and 5 mCi of Nalz5I or 10 mCi of Na13'I. Iodination was performed at 4 "C for 30 min unless otherwise mentioned.
Metabolic Labeling of Hepatocytes-Two hours after plating of rat liver hepatocytes, unattached cells were removed with Hanks' balanced salt solution without phenol red (GIBCO). Cells were labeled with either 0.5 mCi of [35S]methionine (1053 Ci/mmol; New England Nuclear) or 0.2 mCi of [3H]methionine (5.7 Ci/mmol; Amersham Corp.) in 3 ml of Hanks' balanced salt solution and 0.5 ml of heatinactivated fetal calf serum in each 100-mm Petri dish. Two hours after the addition of isotope, 6 ml of Dulbecco's modified essential medium without methionine and containing 10% heat-inactivated fetal calf serum was added. Cells were incubated for another 3 h before the radioisotope label was removed. At the end of the 5-h labeling, all plates labeled with 3H and one plate labeled with 35S were harvested and frozen a t -20 "C. methionine-labeled cells based on DNA and protein assays. The mixed cells were sonicated with a Sonifier cell disruptor (Branson Sonic Power, Danbury, CT) a t scale 4 for 30 s in Hanks' buffer in the presence of 1 mM phenylmethylsulfonyl fluoride (PMSF). Membranes were pelleted a t 150,000 X g for 60 min. The pellet was resuspended with sonication in a small volume of 10 mM HEPES containing 1 mM PMSF at pH 7.4. Aliquots of this membrane preparation were used for concanavalin A chromatography.
Concanaoalin A Chromatography-Membrane pellets were resuspended with sonication in 0.5 ml of 10 mM Tris-HC1 for HTC cells or 10 mM HEPES-HCI for hepatocytes at pH 7.4 containing 1 mM PMSF. Aliquots of resuspended membrane were mixed with an equal or larger volume of 1% sodium deoxycholate in 10 mM Tris-HC1 a t pH 8.0 and 1 mM PMSF. The mixture was sonicated for 10 s. Insoluble material was pelleted at 15,000 X g for 5 min.
The supernatant fractions were applied to pre-equilibrated concanavalin A-Sepharose columns. Two procedures were used for the affinity binding. Both methods were equivalent in binding and elution efficiency. For HTC cells, concanavalin A chromatography mostly was performed in a small column containing 0.3 ml of beads. The excluded fraction was reapplied two additional times. Then the column was washed thoroughly with the same buffer containing deoxycholate. Bound proteins were eluted with 0.5 ml of 0.2 M a-methylmannoside and 0.2 M amethylglucoside in deoxycholate buffer by incubation a t 37 "C for 30 min. This elution was repeated twice. Proteins in the combined eluates were concentrated by precipitation with 10% trichloroacetic acid followed by washing three times with 100% ethanol. The washed pellet was then taken up in lysis buffer for two-dimensional gel electrophoresis. With hepatocytes, concanavalin A binding was mostly performed in Eppendorf centrifuge tubes.
The sample was allowed to equilibrate with the concanavalin A beads with end to end rotation (Fisher, Roto-Rack) for 1 h at 4 "C. Unbound proteins were removed in buffer containing deoxycholate by centrifugation. Bound proteins were eluted directly off the beads with lysis buffer without a-methylmannoside for analysis by two-dimensional gel electrophoresis.
Isolation of Receptors for Asialoglycoproteins-Asialofetuin was a kind gift from Dr. R. Bernacki (Roswell Park Memorial Institute, Buffalo, NY). It was covalently linked to CNBr-activated Sepharose 4B (Pharmacia) as recommended by the manufacturer. Receptor isolation is based on the modified method of Warren and Doyle (10). Affinity binding was performed in Eppendorf centrifuge tubes with 0.2 ml of coupled beads for each sample. Before binding, the asialofetuin beads were pre-equilibrated with binding buffer containing 10 mM HEPES, 50 mM CaCl,, 0.15 M NaCl, and 0.5% Triton X-100 a t pH 7.2. T o aliquots of membrane pellet prepared the same way as for concanavalin A binding was added Triton buffer containing 1% Triton X-100, 0.4 M NaC1, 10 mM HEPES, pH 7.2, and 2 mM PMSF. Proteins were solubilized by sonication in this buffer and then equilibrated a t 4 "C for 30 min. The insoluble proteins were pelleted by centrifugation a t 15,000 X g for 5 min. The supernatant solution was added to 50 mM CaC1, for binding to asialofetuin-Sepharose. The binding was performed for 1 h at 4 "C with end to end rotation of the tube. The unbound proteins were removed with the binding buffer or the column equilibration buffer. Then, the heads were washed with 10 mM HEPES, pH 7.2, to remove high salt. The bound receptor was eluted off the asialofetuin-Sepharose with lysis buffer prior to gel electrophoresis.
Two-dimensional Gel Electrophoresis-Two-dimensional gel electrophoresis was done according to a modification of O'Farrell's method (19). In the first dimension, proteins were separated by charge via isoelectric focusing. Samples were dissolved in lysis buffer, containing 9.6 M urea, 2% Nonidet P40 (Particle Data Laboratory, Elmhurst, IL), 0.4% of Ampholine, pH 3-10, 0.8% of Ampholine, pH 4-6, 0.8% of Ampholine, pH 6-8 (Pharmacia), and either 5% 0mercaptoethanol or 160 mM dithiothreitol (Sigma). A similar composition of Ampholine was used in the isoelectric focusing gel. In the second dimension, proteins were separated by molecular weight by Turnover of Plasma Membrane Proteins 3099 discontinuous gel electrophoresis in the presence of sodium dodecyl sulfate (SDS). Two methods were used in this dimension. For HTC cells, Laemmli's gel system was used (20). The Neville and Glossmann system (21) has a higher capacity for proteins and gives better resolution for glycoproteins and was therefore used to separate hepatocyte membrane proteins. In both systems, the acrylamide concentration of the separating gel was 7.5% while that of the stacking gel was 3%. Gels were stained either with 0.25% Coomassie Brilliant Blue R (Sigma) in 45% methanol and 10% glacial acetic acid for 1 h and then destained in 45% methanol and 10% acetic acid, or with silver nitrate according to the method of Wray et al. (22). Gels were dried on a Bio-Rad gel dryer for 3 h before exposure to Kodak XAR-5 x-ray film for autoradiography. Gels with or 35S label were impregnated with ENHANCE (New England Nuclear) before drying. In general, gels were exposed from 1 to 4 days for iodide in a well shielded container at room temperature and 1 to 2 weeks for / 3 emitters a t -80 "C. "C-methylated molecular weight standards (Amersham) used were myosin, 200,000; phosphorylase b, 92,500; bovine albumin, 69,000; ovalbumin, 46,000; carbonic anhydrase, 30,000; and lysozyme, 14,300.
Counting Procedures-Individual proteins separated by two-dimensional gel electrophoresis were cut out of the gel with the help of a transparent paper copy of the autoradiograph superimposed on the stained and marked gel. Background counts were obtained from counting a comparable area of gel without visible radioactivity. ' "I and 1311 labels were counted in a Beckman Biogamma counter. The two channels were set up in a way that the spillover of "' I into the 1311 channel was less than 1% while there was 25% spill of the ','I into the lZ5I channel. The counts were corrected with the aid of a Cromemco computer program. The half-life determinations of individual proteins by a least square method were also calculated using a Cromemco computer program.
To solubilize 35S or 3H radioactivity from the acrylamide gel, the following procedure was used. The gel spot was soaked in 50 p1 of water for 1 h, then 1 ml of 90% NCS tissue solubilizer (Amersham) was added to digest the proteins in an overnight incubation a t room temperature with shaking. Ten milliliters of scintillation fluid was then added. The fluid was a mixture of 97 ml of ScintiPrep 2 (Fisher), 300 ml of 100% ethanol, and 2 liters of toluene. This procedure gives 90-100% recovery of the radioactivity. A Beckman LS 7500 scintillation counter was used for counting the isotopes. Program 7 was routinely used for the double isotope counting. has a 1% spillover into the 35S channel, while "S spillover into the ,H is 25%. Counting error was in the range of 2% or less. The ratio, 35S/3H, was obtained from the corrected counts with an algorithim which accounted for the spillover of the 3'S into the 3H channel.
Preparation of Coated Vesicles-Coated vesicles were prepared from frozen rat brains and from liver. Typically 20 Sprague-Dawley rat brains or livers were used for one isolation. The procedure followed was that of Pearse (23) with modifications. Frozen tissues were thawed in 3 volumes of ice-cold buffer containing 10 mM HEPES, 0.15 M NaCI, 0.5 mM MgC12, 1 mM EGTA, 0.02% NaN,, and 1 mM PMSF at pH 7.2. They were homogenized in a Polytron (Brinkman Instruments) a t top speed for 1 min. After centrifugation at 2,000 X g for 30 min, RNase (10 units/ml) was added to the supernatant and allowed to react at room temperature for 30 min. The vesicles were pelleted by centrifugation for 1 h a t 55,000 X g. After resuspending the pellet in a small aliquot of homogenizing buffer, the suspension was layered on the top of a 5-60% continuous sucrose gradient, made in the same buffer. After a 2-h centrifugation a t 50,000 X g, a broad cloudy zone between solid white bands at about 50% sucrose and a reddish clear zone near the top of the gradient were harvested. These fractions were pooled and diluted 4-fold with the homogenizing buffer and then were pelleted a t 100,000 X g for 1 h. The pellet was resuspended in the same buffer containing 1% Triton X-100 and layered onto a 5-60% continuous sucrose gradient prepared with homogenizing buffer and Triton X-100. After 2 h of centrifugation a t 50,000 X g, the gradient was divided into 20 fractions. The clathrinenriched fractions were identified by both running the fraction samples on sodium dodecyl sulfate gels and by examining uranyl acetatestained samples from the sucrose gradient with the electron microscope for basket structures. Rabbit antibody to rat clathrin and preimmune serum were a kind gift of Dr. Richard G. W. Anderson (University of Texas Health Science Center, Southwestern Medical School, Dallas, TX). Hepatocytes labeled with [35S]methionine (1.5 mCi, 5 h, 37 "C/1 X 10' cells) were frozen and thawed twice. A crude total membrane fraction of these cells was sonicated in the presence x, total iodinated proteins obtained from 1251-labeled cells co-cultured with '311-labeled cells. 0, total iodinated proteins from '"I-labeled cells mixed with 13'I-labeled cells frozen at 0 h. 0, total concanavalin A-bound protein from co-cultured '"Iand I3'I-labeled cells. A, total concanavalin A-bound proteins from Iz5I-labeled cell cultures mixed with I3lI-labeled cells frozen at 0 h. of 1% Triton X-100,0.01% sodium dodecyl sulfate, 0.01% deoxycholate, 150 mM NaC1, 50 mM Tris-C1, pH 7.5, and 0.1 mM €"SF. Insoluble material was removed by centrifugation at 100,000 x g for 60 min in the 70.1 T i rotor. The supernatant fraction from this centrifugation contained 5.4 X 10' cpm and 1.8 mg of protein/ml. To 1 ml of this material was added 25 pl(500 pg of IgG) of either antibody to clathrin or preimmune serum. After 22 h a t 4 "C, 250 pl of prewashed protein A-Sepharose was added. Following a further incubation a t room temperature for 4 h, the immune complexes were washed extensively with 1% Triton X-100, 0.01% dodecyl sulfate, 0.01% deoxycholate, 500 mM NaC1, 50 mM Tris, pH 7.5. Finally the immune complexes were washed three times with water and were analyzed by two-dimensional gel electrophoresis. The actin standard obtained from an acetone powder of rabbit skeletal muscle was a kind gift of Dr. Roger Daley (Department of Anatomy, State University of New York a t Buffalo).
Peptide Mapping-Peptide mapping by proteolysis in SDS-polyacrylamide gels was performed using either Laemmli's (20) or Neville and Glossman's procedure (21). Following one-dimensional gel electrophoresis in 7.5% polyacrylamide, the gel was stained in 0.25% Coomassie Brilliant Blue G-250 in 30% methanol and 7% glacial acetic acid for 5 min. After a destain in 30% methanol and 7% acetic acid for 5-10 min, the protein bands of interest were cut from the gel and equilibrated with one-fourth strength upper gel buffer containing 0.001% bromphenol blue for 1 h a t room temperature with one change of buffer. The gel was digested by placing the bands atop a second SDS gel which contained 10% acrylamide in the separating gel. (American Hoechst Co.) as described by Labarca and Paigen (24). Calf thymus DNA (Sigma) was used as standard. Briefly, small aliquots of sonicated cells were mixed with 1 ml of high salt buffer containing 2 M NaCl and 0.05 M of NaH2P04, pH 7.4. Then, 1 ml of Hoechst reagent was added at a concentration of 0.2 mg/ml. The mixture was vortexed and read in a fluorometer (Aminco) at an emission wavelength of 458 nm after excitation at 356 nm. Protein assay was done with a Bio-Rad protein assay kit, following the procedure for the macroassay. Bovine serum albumin was used as the protein standard.

RESULTS
Rates of Turnover of Plasma Membrane Proteins and Glycoproteins in HTC Cells-The turnover of externally oriented plasma membrane proteins of HTC cells was studied utilizing an experimental protocol involving labeling of the membrane proteins with either "' 1 or 1311. The loss of lZ51 label from proteins was used to assess their rates of turnover while I3lI label incorporated into the same membrane proteins served as an internal standard to monitor the recovery of isotope through protein separation. HTC cells in this experiment had a doubling time of 48 h, hence cellular proteins will be diluted 2-fold every 48 h. A sampling and mixing protocol was devised to correct for effects of increase in cell number on turnover. However in HTC cells membrane proteins turnover at their characteristic rates independently of whether the cells are dividing or in stationary growth (for documentation, see Refs. 7 and 18). Freshly iodinated cells were kept in culture medium for 2 h before the zero time point was taken. This allows any secretory proteins which were iodinated while en route to secretion to be secreted. During harvesting, the same amount of originally labeled cells are obtained at each time point. Thus, 4-fold more cells were collected at the 96-h time sample relative to the 0-h sample. However, the l3lI-1abeled proteins in co-cultured cells would show the same degree of dilution as lZ5I-labeled cells. Therefore, when one follows protein degradation by following the ratio changes of 1251/1311, the dilution factor cancels out. The co-cultured samples are used also to assay any isotope effects that the cells might detect. For example, 1311 has a much stronger emission energy than lZ5I. l3lI-1abeled proteins could be subjected to more radiation damage which might result in a faster degradation rate.
As shown in Fig. 1, the turnover of lZ51 and l3lI-1abeled membrane proteins in co-cultured HTC cells shows the same ratio with time, indicating that 1311 did not cause any additional radiation damage relative to IZ5I that affected the turnover behavior of the membrane proteins. When the rate of degradation of total labeled proteins was determined from the change in the ratio of the two isotopes in cultured lZ51-labeled cells mixed with frozen l3'I-labeled cells, a half-life of 87 h was obtained. The half-life of the glycoproteins in this fraction isolated based on their ability to bind to concanavalin A was 42 h.

Turnover of
Two-dimensional Acrylamide Gel Analysis of Plasma Membrane Turnover in HTC Cells-Membrane proteins and glycoproteins were resolved by electrophoresis on two-dimensional gels. As shown in Fig. 2, the pattern of iodinated HTC membrane proteins differed depending on the conditions used for lactoperoxidase-catalyzed iodination. When iodination was performed at room temperature, the labeled proteins (panel B ) showed a pattern similar but not identical to that of the Coomassie Blue-stained proteins (panel A ) in the membrane sample. However, when iodination was performed at 4 "C, the labeled proteins (panel C) had a pattern suggestive of the pattern displayed by membrane glycoproteins. The concanavalin A-bound iodinated proteins which were labeled in situ in cells at room temperature is shown in panel D.
The reason for the difference in accessibility for iodination of the different sets of membrane proteins as a function of temperature is not yet known but in no case are intracellular proteins labeled by the external labeling reagents. Actin for example is not labeled at either 4 "C or room temperature. The recovery of label after gel separation was monitored and is shown in Table I. In two different samples taken at different time points, 62% or 82% of the total membrane proteins were TABLE I Recovery of iodinated membrane proteins after isolation and gel electrophoresis Iodinated HTC cells were analyzed for the recovery of radioactivity after isolation of the membrane and gel electrophoresis of the iodinated membrane polypeptides. 1251-labeled cell samples from the first time point and the last time point of a turnover experiment were used. Frozen 13'I-labeled cells were mixed with IZ5I-labeled cells that were cultured for the indicated amount of time.

TABLE I1
Half-lives of individual HTC cell plasma membrane proteins resolved by two-dimensional gel electrophoresis The protein number corresponds to the spots on the two-dimensional gel in Fig. 2. The half-lives and the standard error of the halflife estimation were determined with data from three separate experiments.

Autoradiograph of rat hepatocyte proteins from cell surface-iodinated cells resolved by two-dimensional gel electrophoresis. Upper panel shows proteins from cells labeled at 4 "C,
while the lower panel shows proteins from cells labeled at room temperature (RT). Immediately after iodination, cells were washed and the membranes processed for two-dimensional gel electrophoresis. The two-dimensional gel system employed electrofocusing in the first dimension and the discontinuous SDS-gel system of Neville and Glossmann (21) for the second dimension is described under "Experimental Procedures." trichloroacetic acid precipitable. Sixty-one per cent and 28% of the total radioactivity was retained in the focusing gel in the 2-and 93-h samples, respectively. Before running the second dimension SDS gel, the focusing gel has to be equilibrated with the SDS sample buffer, containing 1% SDS and reducing agent. Label can be lost during this step. Assuming the SDS gel will retain all of the counts from the equilibrated focusing gel, between 14% and 30% of the starting counts are still present at this step of the procedure. The reasons for the differential loss of label was not further examined because the ratio of the two isotopes remains relatively constant. Hence to study protein turnover utilizing the technique of twodimensional gel electrophoresis, it is necessary to use a double isotope method to correct for variable loss of labels during sample preparation and analysis.
The half-lives of individual externally accessible plasma membrane proteins were determined by cutting protein spots out of the gel and counting them directly for the ratio of 1251/ 1311. The values obtained for the half-lives are shown in Table  11. Two different ranges of half-lives were found for the proteins analyzed. Proteins which did not bind to concanavalin A, with molecular weight ranging between 51,000 and 77,000 and isoelectric focusing point between pH 4 and 6.2, had half-lives of about 100 h. Three glycoproteins analyzed after enrichment by binding to concanavalin A had molecular weights between 85,000 and 115;OOO and pH between 4.0 and 5.5 and had half-lives of 22 h. This suggests that in HTC cells two classes of plasma membrane proteins, one consisting primarily of glycoproteins and.the other non or minimally glycosylated proteins, exist with respect to their rates of turnover but that individual members in each class have approximately equivalent half-lives for turnover.
Rates of Turnover of Plasma Membrane Proteins and Gly-

Properties of cell surface iodinated proteins from hepatocytes
Glycoprotein nature of cell surface iodinated proteins. Total iodinated proteins were derived from the membrane pellet of hepatocytes that had been surface iodinated then plated. The per cent column represents the phosphotungstic acid-precipitable radioactivity versus total radioactivity incorporated into the cells via lactoperoxidasecatalyzed iodination (0.4% phosphotungstic acid in 0.5 M HCI). The deoxycholate-soluble proteins represent the membrane proteins soluble in 1% deoxycholate buffer which was used for the concanavalin A binding; the Triton X-100-soluble proteins are those membrane proteins solubilized in 1% Triton X-100 buffer which were subsequently applied to a column of asialofetuin.   Fig. 5. R1, R2, and R3 represent the three major components of the asialoglycoprotein receptor (see Fig. 4). The half-lives were determined from the combined data of four experiments utilizing 4-5 sample points for each experiment. little effect on membrane protein turnover in the isolated hepatocyte system. In contrast to the situation in HTC cells, coproteins in Primary Cultures of Hepatocytes-Freshly isolated hepatocytes were iodinated in suspension and plated for 2 h. The 2-h incubation serves two purposes. One is to allow cell attachment. Since only viable cells will attach to the substratum, cell viability was above 95% after the 2-h incubation. The other purpose is to allow cells to secrete the serum proteins which were iodinated while en route to secretion. Cells cultured on plastic plates or biomatrix-coated plates were compared for plating efficiency and viability. Since adult hepatocytes undergo little DNA synthesis and no cell division, concanavalin A-bound proteins do not turn over differently from the total iodinated membrane proteins in hepatocytes (data not shown). Since the biomatrix-coated plates provide a slightly better substratum for cell culture, it was used throughout the following experiments. The iodinated cells exhibited a typical hepatocyte cord structure after plating. The cells flattened out with time. Cell viability was better than 90% on the matrix-coated plates throughout the culture period.
Two-dimensional Polyacrylamide Gel Analysis of Externally Labeled Membrane Proteins from Hepatocytes-In contrast to  5. A , two-dimensional gel analysis of hepatocyte membrane proteins metabolically labeled with methionine, mannose, or fucose. B is a diagram of total [35S]methionine-labeled membrane proteins resolved by two-dimensional gel electrophoresis. The numbered spots indicate glycoproteins. The alphabetically labeled spots indicated proteins that do not bind to concanavalin A or which are not labeled with [3H]fucose or [3H]mannose. These proteins were examined for their half-lives as shown in Table V. the pattern found with membrane proteins of HTC cells, the labeled proteins (Fig. 4). The concanavalin A-bound proteins iodinated proteins of hepatocytes had a very similar pattern had their isoelectric focusing point shifted slightly to the basic regardless of whether the cells were labeled at 4 "C or room end or the left side of the gel because the presence of residual temperature (Fig. 3). The proteins bound to concanavalin A deoxycholate interferes with protein focusing. As shown in also showed a pattern similar but not identical to the total Table 111, up to 40% of total labeled proteins could bind to The protein labeling system refers to Fig. 5B. The half-life determination is based on regression analysis performed on two experiments. Each experiment had five time points, ranging from zero to 70 h after labeling. Standard deviation is determined for the least square line. For proteins 10, f, and p, two half-lives are listed because their degradation follows a biphasic pattern. The first phase occurred between the first and second harvest, or between zero to 12 h after removal of labeled methionine from the medium. concanavalin A compared to the less than 10% binding of total iodinated membrane proteins from HTC cells. Cells labeled at 4 "C were used for the determination of the degradation half-lives of individual proteins.
Rates of Degradation of Individual Plasma Membrane Proteins of Hepatocytes-The double isotope technique whereby ','"I-labeled frozen cells are mixed with '251-labeled cells harvested from culture every 12-24 h was used to assess degradation of the labeled hepatocyte membrane proteins. Hepatocytes obtained from normal adult livers do not divide, hence there is no need to consider a dilution factor. Cell number was determined by either DNA or protein assay: lo6 hepatocytes contain approximately 1 mg of protein or 25 pg of DNA. The half-lives of individual proteins are shown in Table IV, and the values are based on four separate experiments. Since relatively low counts were associated with each protein spot, there was some variation of the half-lives among experiments. The protein spots used for analysis are those which contained more than 100 cpm of "' 1 label above background. A statistical analysis of the data using a least square method to determine half-life was followed. The application of the least square analysis was based on the assumption of a linear relationship between the log of the ratio of 1251/1311 and time. It is formulated as Y = aX + b, where the X axis represents time and the Y axis is the log value of 1251/1311. The log value of the isotope ratio is used because the rate of protein degradation follows first order kinetics. The half-life is the value of X when Y = b/2. Heterogeneous rates of turnover were found for most of the hepatocyte membrane proteins so analyzed. The lectin for asialoglycoproteins was isolated using asialofetuin-coupled Sepharose 4B; we recovered three major subunit classes (Fig. 4) as described previously (10). The halflives of individual subunit classes were analyzed after resolving on the two-dimensional gel. Homogeneous turnover was found among these subunits of the receptor, they had a halflife of 18 h similar to that reported previously (10) ( Table   IV) .
The half-life of total [35S]methionine labeled membrane proteins in hepatocytes was also assessed, either by the loss of specific activity or by the decrease in the ratio of 35S/3H. The half-life was 55 h while the total concanavalin A-bound protein in the methionine-labeled membrane fraction had a 30-h half-life.
Half-lives of individual membrane proteins metabolically labeled with methionine were also determined after resolution by two-dimensional gel electrophoresis. A fluorograph of the resolved membrane proteins is shown in Fig. 5A (top left). Less than 8% of the methionine-labeled membrane proteins will bind to concanavalin A-Sepharose. Many of the concanavalin A-bound glycoproteins labeled with methionine contain both mannose and fucose residues as shown by comparing the concanavalin A-bound fraction to the glycoproteins labeled with [3H]mannose or [3H]fucose. The half-lives of individual proteins illustrated in Fig. 5B were determined by the decrease in the ratio 35S/3H and are shown in Table V. Protein 1, a glycoprotein, larger than 200,000 daltons was iodinated by surface labeling and was highly enriched by concanavalin A chromatography. The half-life of this protein was 45 k 6 h with "S/'H labeling and 42 f 10 h with ' "1 and I3'I labeling. The half-lives of the glycoproteins which bound to concanavalin A as a group were not faster than those that did not bind and heterogeneous half-lives were found in both populations of resolved proteins. Among the total labeled proteins in the membrane fraction from hepatocytes are two relatively abundant proteins which we have characterized as clathrin (b) and actin (0) as described below. Both of these proteins have half-lives greater than 100 h.
Identification of Clathrin and Actin-Coated vesicles prepared from rat brains or livers exhibited characteristic basket structures as shown in Fig. 6. The M, = 180,000 protein identified as clathrin was purified from these vesicles. To identify clathrin and actin in the rat hepatocyte membrane preparations, aliquots of three samples from hepatocyte membranes, clathrin purified from the coated vesicles, and actin purified from rabbit muscle extract were co-electrophoresed in two dimensions. Neither clathrin nor actin focus as discrete bands but in hepatocyte membranes, the protein that comigrates with clathrin is readily visible with Coomassie Blue stain (Fig. 7) as is the actin-like protein, although the recovery of actin from membrane preparations is quite variable.
T o further confirm the identity of hepatocyte clathrin, Staphylococcus protease or chymotrypsin digestion fragments of clathrin from the purified brain baskets and from hepatocyte membrane pellets resolved after polyacrylamide gel electrophoresis were compared. Similar digestion fragments assayed by silver staining were obtained by either protease treatment, but the clathrin from hepatocytes had a tendency to be degraded slightly faster than brain clathrin (data not shown).
Finally, antibodies to rat clathrin when reacted with ["SI methionine labeled extracts of hepatocytes recognize a M , = 170,000 protein of PI 6.0 which migrates in a two-dimensional gel at the same region as the protein identified in Table V and Fig. 7 as clathrin (Fig. 8).

DISCUSSION
In this study we compare the turnover behavior of plasma membrane proteins in hepatoma cells of rat liver origin to plasma membrane proteins in primary cultures of hepatocytes from rat liver. The morphology of the plasma membrane in the hepatoma cell line is not as complex as that of the hepatocyte when in the liver or in culture on a biomatrix support. The hepatoma cell lacks the polarity and the regional specialization, such as the sinusoidal domain, the bile canalicular domain, gap and tight junctions, etc. that characterize the differentiated hepatocyte in the liver. Hepatocytes when isolated from the liver by collagenase perfusion also "lose" their membrane domains and the polarity that they have in situ. However, with time in culture the hepatocytes regain the characteristic cord appearance that they have in the animal and the sinusoidal, gap and tight junctions and the bile canalicular domains reappear.
The turnover behavior of the externally oriented plasma membrane proteins of the hepatoma cell is not as complex as the turnover behavior of the externally oriented plasma membrane proteins of the rat hepatocyte. Many of these proteins in HTC cells are not glycosylated or are minimally glycosylated. This is in contrast to the situation in rat hepatocytes where most of the plasma membrane proteins are glycosylated in that they will bind to concanavalin A or they can be labeled with either mannose or fucose. All of the non or minimally glycosylated proteins of the hepatoma cell that are accessible to external iodination and are resolved by two-dimensional polyacrylamide gel electrophoresis have half-lives for turnover in the range of approximately 100 h. Three major proteins that are accessible in these cells to lactoperoxidase-catalyzed iodination are glycoproteins. Similar to previous studies (7, 8) the glycoproteins of the hepatoma cell membrane have a faster rate of turnover corresponding to a half-life of 20-24 h than the non or minimally glycosylated proteins. With one exception (25), all of the externally oriented proteins of the hepatoma cell that we have thus far examined fall into either the 100-h half-life class, typical of the nonglycosylated proteins, or the 24-h half-life class, typical of the glycosylated proteins of the hepatoma cell.
The plasma membrane proteins accessible to lactoperoxidase-catalyzed iodination in both the hepatoma cells and the primary hepatocytes would include both transmembrane proteins and proteins in the exoskeletal of the cell. In fact, the long half-life class of iodinatable proteins in the hepatoma cell may be enriched in exoskeletal proteins. It is most probable that in the hepatoma cells most of the plasma membrane which is not well differentiated is continually being internalized and recycled back to the surface (26)(27)(28). It is also possible that fusion of internalized plasma membrane with lysosomes occurs as one step in the recycling phenomenon. Exoskeletal and perhaps also cytoskeletal proteins in fact may be more resistant to lysosomal hydrolysis during membrane recycling than are glycoproteins. Hence, relative resistance of the two classes of proteins to lysosomal hydrolysis during recycling of a single type domain membrane in hepatoma cells may be the basis for the two classes of membrane protein turnover in this cell type.

Rat Hepatocyte Membrane Pellet
The most likely basic mechanism for the turnover of the plasma membrane of HTC cells then is internalization of a membrane domain and fusion of this domain with lysosomes, followed by either recycling or some degradation of the glycoproteins and less frequently some degradation of the minimally glycosylated membrane or exoskeletal proteins. Superimposed upon this basic mechanism is some (rare) degradation of a plasma membrane protein via a nonlysosomal protease (25).
In primary cultures of hepatocytes the situation with respect to the regulation of plasma membrane protein turnover is more complex. In this cell type, the plasma membrane and exoskeletal glycoproteins that are accessible for lactoperoxidase-catalyzed iodination do not appear to fall into a small number of classes with respect to their turnover behavior. Rather, there is considerable heterogeneity in the turnover of these membrane proteins. However, the plasma membrane of the hepatocyte is complex, having many morphological and functional domains including sinusoidal, bile canalicular domains, gap and tight junctions, etc. Our working hypothesis is that if the different transmembrane glycoproteins of the hepatocyte plasma membrane could be assigned to their respective domains, then the turnover behavior of the proteins in the domain may be more homogeneous. As a beginning toward testing this hypothesis, we have analyzed the turnover behavior of the three polypeptide classes comprising the receptor for asialogycoproteins (10). Elsewhere, we will show that the three polypeptide classes are derived from at least two precursor polypeptides different in primary amino acid structure.2 The two precursor polypeptides are presumably derived from two distinct RNAs. This receptor is present primarily in the coated pit region of the hepatocyte sinusoidal surface. All of these glycopolypeptides had the same half-life for turnover (about 18-20 h). One of the major structural proteins of the coated pits is clathrin. However, clathrin is not accessible for iodination. Hence, we have examined the turnover behavior of clathrin relative to the other total membrane glycoproteins by metabolic labeling of hepatocytes with methionine. Since the membrane fraction from hepatocytes is not a homogeneous preparation of plasma membrane, not all of the methionine-labeled glycoproteins resolved by twodimensional polyacrylamide gel electrophoresis in this fraction will be plasma membrane proteins. Of the metabolically labeled membrane proteins, clathrin has a relatively long halflife of about 100 h. Hence, if the turnover behavior of clathrin reflects that of clathrin in the coated pit with the receptor for asialoglycoproteins, then the clathrin, a large molecular weight polypeptide is probably turned over independently of the other proteins in the pit, possibly by dissociating from the membrane proteins of the pit prior to pit turnover.
At present we are attempting to assign other proteins to their respective domains of the hepatocyte plasma membrane. Our goal is to compare the turnover behavior of these proteins of the sinusoidal region, and to other regions of the plasma membrane (12)(13)(14). In this way, we hope to eventually understand how the hepatocyte establishes and maintains the differentiated membrane domains that characterize this complex cell. We also wish to determine which of the most likely mechanisms are used by the hepatocyte to regulate the turnover of these domains. These mechanisms are: interiorization of membrane followed by fusion with lysosomes, neutral pro-' A. Le and D. Doyle, manuscript in preparation.

Turnover of Plasma
Membrane Proteins 3107 teases, and shedding. It is possible that more than one mechanism may in fact be used depending on physiological status of the cell. It is possible for example that those proteins that show biphasic kinetics for turnover are turned over partly by shedding and partly by one or more of the other most likely mechanisms for membrane protein degradation.