Biosynthesis of the Human Transferrin Receptor in Cultured Cells*

The biosynthesis and degradation of the cell surface transferrin receptor has been investigated. The recep- tor is a glycoprotein, and evidence is presented that the mature receptor contains both complex and high man- nose N-asparagine-linked oligosaccharides that are synthesized via a common high mannose intermediate as previously described for other glycoproteins. It is shown that fatty acid is associated with only the mature form of the receptor and that addition of fatty acid to the receptor can occur as long as 48 h after synthesis. Glycosylation is not an absolute requirement for the receptor to act as acceptor for fatty acid, nor for transport to the cell surface, although the efficiency of both processes may be reduced in tunicamyein-treated cells. The protein moiety of the transferrin receptor is de- graded with a half-life of approximately 60 h.

Radiolabeling Procedures-Cells incubated with or without tunicmycin (4-5 pg/ml) were labeled with [3H]palmitic acid or [35S] methionine. Tunicamycin dilutions were obtained from a frozen stock solution (5 mg/ml in dimethyl sulfoxide DMSO) and incubation was for 90 min prior to labeling. For [3H]palmitic acid labeling, [3H] palmitic acid in benzene was dried and then redissolved in 80% (w/v) ethanol/H20. Labeled palmitate was added to cell suspensions such that the ethanol concentration did not exceed 0.3%. Cells (4 x IO6 ce&/ml) were labeled with 100 pCi of [3H]palmitic acid/ml in RPMI 1640 medium supplemented with 5% fetal calf serum dialyzed against PBS.
Metabolic labeling of cells with [35S]methionine was carried out in methionine-free RPMI 1640 medium containing 5% dialyzed fetal calf serum. For endo H experiments, cells (6 X lo6 cells/ml) were pulselabeled with [36S]methionine for 10 min using 150-200 pCi/ml (0.6 ml total volume), then diluted with 0.4 ml of RPMI 1640 medium supplemented with 10 mM L-methionine and 10% fetal calf serum. For the turnover rate of [3H]palmitate-and [35S]methionine-labeled transferring receptor, labeled cells were washed twice using RPMI 1640 medium supplemented with 10% horse serum, 10 p~ 2-mercaptoethanol, and 100 p~ palmitic acid, then incubated in the same medium for various chase periods.
Immunological Procedures-Monoclonal anti-transferrin receptor antibody (B3/25) has been described previously (15,16) and ascitic fluid was used for immunoprecipitation. To obtain immunoprecipitates, labeled cells were lysed with 1% Nonidet P-40 in PBS (1-2 X IO' cells/ml). After 20 min at 4 "C, nuclei were removed by centrifugation for 2 min in an Eppendorf model 3200 microcentrifuge. The lysate was then mixed with a solution containing 10% Nonidet P-40, 10% sodium deoxycholate, and 1% SDS in PBS (101, v/v, lysate: detergent solution), followed by the addition of 60 pl of 10% (v/v) formaldehyde-fixed Staphylococcus aureus (21) per ml of lysate. After 20 min at 4 "C, the bacteria were removed by centrifugation at 20,000 X g for 60 min. Antigen-antibody complexes were formed by incubating samples of the cell extract with the B3/25 monoclonal antibody for 15 min, then with goat anti-mouse IgG for an additional 15 min. Immune complexes were then collected using a 1% (v/v) suspension of S. aureus (150 pl of bacteria suspension/extracts from 1-2 X lo6 cells) and washed 3 times with 0.4 ml of 0.5% deoxycholate, 0.5% Nonidet P-40, 0.05% SDS in PBS. The antigens were released from the S. aureus bacteria by boiling for 2 min in electrophoresis sample buffer (22). Control immunoprecipitates for nonspecific binding were similarly prepared except that the monoclonal B3/25 antibody was not added.
Trypsin Digestion-Labeled cells (1-2 X lo7 cells/ml) were digested with TPCK-trypsin (256 units/mg, 50 pg/ml) at 23 "C in a total volume of 150 pl. After 15 min, proteolysis was terminated by adding 100 pl of 5 mg/ml of ovomucoid trypsin inhibitor. Supernatants were removed after centrifugation of the cells in a microcentrifuge for 1 min. Immunoprecipitates were then prepared from either the supernatants or from the remaining cells after washing once with 0.6 ml of PBS.
Treatment with Endo-P-N-acetylglucosaminidase H-Endo H was a kind gift from Dr. Phillips Robbins, Massachusetts Institute of Technology. S. aureus immunoprecipitates (three for each time point) were isolated and washed as described above. An additional washing step was carried out using 0.4 ml of 10 mM Tris-HC1 containing 0.1% Nonidet P-40 (pH 7.5). The procedure used for endo H digestion was similar to that described by Rothenberg and Boyse (23) except that the incubation was for 15-18 h at 37 "C. One of the three immunoprecipitates was stored at -20 "C and the remaining two were incu-bated at 37 "C in the presence or absence of endo H (5-10 pg/ml). No differences were observed upon SDS-polyacrylamide gel analysis between samples stored a t -20 "C and those incubated at 37 "C without endo H.
Miscellaneous Methods-SDS-polyacrylamide gel electrophoresis was carried out as described previously (24) using 7.5 or 10% polyacrylamide gels. Processing of gels for fluorography was as described by Bonner and Laskey (25). Protein synthesis was estimated from the amount of radioactivity incorporated into trichloroacetic acid-precipitable, ethanol-insoluble material obtained from 1% Nonidet P-40 extracts of labeled cells. Gels were scanned using a Zeineh scanning densitometer (Biomed Instruments Inc.).

RESULTS
Glycosylation of the Transferrin Receptor-The biosynthesis of the transferrin receptor in the human leukemic cell line CCRF-CEM was studied using the monoclonal antibody designated B3/25 to isolate the receptor by immunoprecipitation. As shown in Fig. 1, newly synthesized receptor detected by labeling cells for 10 min with [""Slmethionine had an apparent molecular weight of 88,000 ( = 88,000 species removed all labeled mannose residues while only partial loss of mannose occurred during digestion of the mature M, = 95,000 receptor (data not shown). The interpretation of these results is that as early as 10 min after synthesis, the transferrin receptor is glycosylated and the oligosaccharides initially present are all of the high mannose type. Subsequently some, but not all, of these oligosaccharides are converted to complex oligosaccharides. Glycopeptide analysis by gel filtration of the receptor metabolically labeled with either mannose, galactose, or glucosamine was also consistent with the presence of both complex and high mannose oligosaccharides on the mature receptor.
Glycosylation of the transferrin receptor was also studied using the antibiotic tunicamycin, which inhibits asparaginelinked glycosylation by blocking the formation of the donor lipid-linked oligosaccharide (28,29). As shown in Fig. 2c, a major radiolabeled species with a molecular weight of approx- imately 79,000 could be detected after incubating CCRF-CEM cells with tunicamycin (5 pg/ml). Under these conditions, none of the normally glycosylated species (Fig. 26) was found and overall protein synthesis was not affected, whereas incorporation of [2-:'H]mannose into cellular glycoproteins was inhibited by more than 90%. A minor species (Mr = 74,000) also appeared to be specifically immunoprecipitated from tunicamycin-treated cells. Some of the unglycosylated transferrin receptors could be transported to the cell surface, since trypsin treatment of intact tunicamycin-treated cells resulted in the release of a tryptic fragment of the receptor into the supernatant (Fig. 2f). Under the same conditions, pulse-labeled (5 min) intracellular receptor was not cleaved by trypsin, indicating that under the conditions used, transferrin receptors at an intracellular location were not sensitive to proteolysis (data not shown). However, assuming that the efficiency of immunoprecipitation of the unglycosylated receptor and its tryptic fragment are identical with their glycosylated counterparts, then it is evident from the data presented in Fig. 2

Transferrin Receptor Biosynthesis
receptor could be distinguished by their migration on SDSpolyacrylamide gel electrophoresis and susceptibility to endoglycosidase-H digestion, it was possible to investigate when, during biosynthesis, fatty acid becomes associated with the receptor. As shown in Fig. 3, even when cells were briefly labeled with [:'H]palmitate (10 min), only the mature form of the transferrin receptor incorporated detectable amounts of radioactivity (Fig. 3, compare tracks a-f with g-j). This observation then raised the question of whether fatty acid becomes bound to the receptor at a defined late stage in its biosynthesis after oligosaccharide processing is complete or whether the mature form of the transferrin receptor can be labeled with ["Hlpalmitate at any time subsequent to its insertion into the plasma membrane. This point was investigated by incubating CCRF-CEM cells in the presence of tunicamycin for various periods of time, then labeling cells with ['"Hlpalmitate. Transferrin receptors synthesized in tunicamycin-treated cells could be distinguished from receptors that were made before the addition of tunicamycin on the basis of their mobility in SDS-polyacrylamide gels. As shown in Fig. 4, the transfemn receptor could still be labeled with ["Hlpalmitate in cells previously exposed to tunicamycin for as long at 48 h. Since the labeled receptor had a molecular weight of 95,000 corresponding to glycosylated molecules made prior to the addition of tunicamycin, it can be concluded that molecules that have been synthesized more than 48 h previously were labeled. The conclusion that transferrin receptors synthesized many hours earlier can still be labeled with palmitate was also supported by similar experiments in which protein synthesis was inhibited with emetine (10 pg/ m l ) up to 15 h prior to labeling cells with palmitate (data not  240 min (g-1). Incubation in the absence of (-) or presence of endo H (+) was carried out on transferrin receptor immunoprecipitates as described under "Materials and Methods." Exposure was for 19 days.
shown). Careful inspection of the autoradiographs shown in Fig. 4 also reveals that unglycosylated transferrin receptors were labeled with palmitate, suggesting that glycosylation is not obligatory for association of the receptors with fatty acid, although judging from the amount of radioactivity associated with the unglycosylated receptor, tunicamycin may have a quantitative effect on the efficiency of the process.

Turnover of [35S]Methionine-and [3H]Palmitate-labeled
Transferrin Receptor-It is widely believed that the transferrin receptor is internalized during iron transport (e.g. Refs. 7 and 30), although there is no direct evidence for this view. This raises the question of whether the transferrin receptor is degraded during internalization or is recycled back to the cell surface. If the latter possibility is true, then the additional question is raised whether the fatty acid modifkation of the receptor may play a role, for example, as a signal for part of the process given that association of palmitate with the receptor can occur long after synthesis of the glycoprotein. With regard to both of these points, the rate of turnover of the protein and lipid moieties of the transferrin receptor are of some relevance. As shown in Fig. 5, pulse-chase experiments with [35S]methionine-labeled receptor demonstrate that the receptor protein is relatively stable and is degraded with a half-life of approximately 60 h. The palmitate moiety appears to turn over more rapidly than the receptor protein, and 60% of the labeled palmitate is lost within 12 h. There is some evidence that the rate of turnover of the palmitate is heterogeneous, however. There are several explanations for this, including the possibility that the receptor contains more than one class of bound lipid moieties.

DISCUSSION
The results reported in this paper provide further information about the transferrin receptor of human cells. The mature cell surface glycoprotein contains both complex and high mannose I?-asparagine-linked oligosaccharides, and all the evidence is consistent with these oligosaccharides being derived from a lipid-linked oligosaccharide precursor via a high mannose intermediate present on the immature intracellular form of the glycoprotein, as has been well established for other secreted and membrane-bound glycoproteins (23, 27, 31-34). The time required for oligosaccharide processing appears to be about 4 h, which is somewhat longer than that for the processing of the oligosaccharides attached to the vesicular stomatitis virus G glycoprotein (35). Inhibition of asparagine-linked glycosylation by tunicamycin does not completely block insertion of the transferrin receptor into the plasma membrane. This is analogous to other membrane glycoproteins, including HLA (36) and IgM (37), but different from the Thy-1 glycoprotein from a mouse lymphoma mutant cell line where abnormal glycosylation completely prevented cell surface expression (38). As noted earlier, however, it is probable that tunicamycin treatment may well reduce the efficiency with which transferrin receptors reach the cell surface.
We have previously shown that the transferrin receptor can be labeled with [3H]palmitate and provided evidence that the fatty acid is probably covalently attached to the region of the molecule containing the membrane-associated portion of the molecule (19). Here we have shown that only the mature form of the receptor is associated with fatty acid and that transferrin receptors may be labeled with [3H]palmitate 48 h after their synthesis. These results differ in several respects from those reported for the attachment of fatty acids to the membrane glycoproteins of vesicular stomatitis and Sindbis viruses. The results reported by Schmidt and Schlesinger (39) suggest that fatty acid is bound to the viral glycoprotein shortly before oligosaccharide processing was complete, and they suggest the Golgi complex as a likely site for fatty acid incorporation into the glycoprotein. Further, nonglycosylated forms of the viral glycoprotein failed to act as fatty acid acceptors (40, 41), whereas we have detected [3H]palmitate labeling of the unglycosylated transferrin receptor found in tunicamycin-treated cells. The fist difference may arise because, unlike the viral glycoproteins which are assembled into virions, transferrin receptors may, during their lifetime on the cell surface, be repeatedly internalized and recycled back to the plasma membrane, affording multiple opportunities for association with fatty acid, if, in fact, the Golgi complex, for example, is the site for incorporation. The fact that unglycosylated transferrin receptors act as fatty acid acceptors is because, unlike vesicular stomatitis virus G or Sindbis virus E l and E2 glycoproteins, at least a fraction of the unglycosylated molecules reach the cell surface, and it is these molecules that are labeled with [3H]palmitate. Thus, as previously suggested (39-42), the failure of unglycosylated viral glycoproteins to reach the cellular site at which fatty acid addition occurs, rather than their structure per se, accounts for their inability to bind fatty acid.
The availability of monoclonal antibodies enabled us to investigate the turnover rate of transferrin receptors and the results clearly show that the protein moiety is degraded only slowly (t1/2 -60 h). This means that, whatever the role of the receptor in iron transport, the transferrin receptor is not degraded in this process. Thus, if, as is widely believed, the receptor is internalized together with transferrin, it must be recycled back to the cell surface. The situation with the palmitate moiety of the receptor is less clear. The evidence suggests that fatty acid can be added to the protein moiety at any time after synthesis is completed and that it may be removed at a rate faster than can be accounted for by total degradation of the receptors. It has been calculated that the entire population of transferrin receptors may be internalized in about 6 min if this is obligatory for iron transport (7). If, then, addition and removal of fatty acid were a necessary event in this cycle, one would expect a turnover rate of [3H] palmitate molecules of the order of a few minutes. This does not appear to be the case.