Nonacylated Human Transferrin Receptors Are Rapidly Internalized and Mediate Iron Uptake*

The human transferrin receptor is post-translation-ally modified by the addition of a fatty acyl moiety. transient expression uptake,

Functional studies indicated that these nonacylated mutant receptors were internalized efficiently and mediated iron uptake from human transferrin at a similar rate to that of wild type receptors. We conclude, therefore, that acylation of the human transferrin receptor is not essential for endocytosis and recycling.
The transferrin receptor (TR)' binds the serum iron transport protein, transferrin (Tf), and mediates uptake of iron into the cell by the process of receptor-mediated endocytosis (reviewed in Ref. 1). Iron is a component of a variety of essential enzymes and electron transport carriers, and iron transport is closely coupled to cell proliferation. The human TR is a disulfide-bonded homodimer of two 95-kDa subunits. It has an external domain of 671 amino acids, a transmembrane region of 28 amino acids, and an amino-terminal cytoplasmic domain of 61 amino acids (2,3 of the class of cell membrane proteins that are post-translationally modified by the covalent attachment of palmitate (4,5). Treatment of the TR isolated from human cells with hydroxylamine at neutral pH cleaved the lipid moiety from the receptor suggesting it is covalently linked to the receptor by a thioester bond. Subsequent analysis of mutant human TRs transiently expressed in simian Cos cells indicated that, in these cells, the major fatty acid attachment site of the receptor was CyP, with CY#~ a possible minor site (6). Both Cys6* and CYSTS are located close to or within transmembrane region of the receptor on the cytoplasmic face of the plasma membrane. The lipid attachment sites of other palmitoylated membrane proteins have also been identified as cysteine residues that lie close to the cytoplasmic face of the cell membrane (reviewed in Refs. 7 and 8). Despite the fact that a large number of membrane proteins are known to be palmitoylated, little is known about the functional significance of this post-translational modification (7,8). We have previously shown that palmitoylation of the human TRs transiently expressed in Cos cells is not necessary for transport to the cell surface (6). However, the functional properties of mutant TRs lacking a lipid moiety could not be analyzed further, because the human receptors were only expressed on a small fraction of the transfected cells and their ability to bind human Tf and mediate iron transport could not be distinguished from that of endogenous simian TRs. We have recently described an experimental system that overcomes these problems in which human TRs are expressed in chick embryo fibroblasts (CEF) using a helper-independent retroviral vector (9). High expression of human TRs can be obtained and, as infectious virus is produced by cells transfected with the retroviral vector containing the human TR cDNA, cultures of CEF in which virtually all the cells express recombinant human TRs can be obtained within l-2 weeks following transfection.
Most importantly, as chick TRs do not bind human Tf, mutant human receptors can be analyzed without a direct contribution of endogenous chick receptors in functional assays.
In the studies reported here, we first expressed mutant human TRs in which CyP and CYSTS had been altered to serine residues because such mutant receptors were not detectably acylated in Cos cells (6). However, we found that in CEF, Ser62.67 mutant receptors were still acylated. Tf. cells were then removed from the wells with 1 M NaOH, and radioactivity in the medium, the acid wash, and the cell pellet was determined.
The distribution of human TRs on the cell surface and inside the cell at steady state was determined by incubating cells, plated out and preincubated as described for binding studies, with 4 rg/ml '*$Ilabeled Tf at 37 "C for 15,30, and 60 min. After incubation for these periods, the acid wash procedure described for the kinetic experiments was used to distinguish surface-bound and internalized '*?-labeled Tf.
59Fe Uptake Measurements-Human apoTf was labeled with K9Fe (69FeC13, Amersham Corp.) to a specific activity of 5-10 pCi/mg using nitrilotriacetate (14). Cells were plated at a density of 5-7 x  Receptors-wild type and Ser6', Ser67, and Ser62.67 mutant human TRs were expressed in CEF using the helperindependent retroviral vector RCAS as described under "Experimental Procedures." As preliminary experiments indicated that the Ser62,67 mutant receptors were acylated, mutant receptors were also constructed in which Cys6* and CYSTS were modified to alanine and expressed in CEF. The number and affinity of wild type and mutant human receptors expressed on the surface of infected CEF were determined by quantitative binding studies with lz51-labeled human Tf. The level of surface expression varied from 0.9 x lo5 receptors per cell for the Ser'j7 mutant to 4.9 X lo5 receptors per cell for the Ser62*67 mutant. Dissociation constants for the wild type and mutant receptors ranged from 3.2 to 6.6 nM. These values are similar to those reported previously for the expression of human TRs in this system (9).
To determine the level of acylation of wild type and mutant receptors, replicate cultures of cells were metabolically labeled with either [35S]methionine or [3H]palmitate, and human TRs were isolated by immunoprecipitation with monoclonal antibody B3/25 (Fig. 1) -----------w--d----83125: Polyacrylamide gels were fluorographed and then sequentially treated with 1 M hydroxylamine at pH 6.6 and 11.0 and fluorography repeated after each treatment. The figure shows the results from two independent experiments of hydroxylamine treatment of mut.ant and wild type human TRs. In each case, fluorographs were exposed for 3 days.
(data not shown). The radioactivity from ["Hlpalmitate incorporated into the Alafi2s6' mutant receptors, which was lower in other experiments (see Fig. 2), is ascribed to metabolic conversion to amino acids. receptors were treated sequentially with 1 M hydroxylamine at pH 6.6 and 11.0. Treatment with hydroxylamine at neutral pH virtually quantitatively removed the lipid moieties of wild type and Ser"' mutant TRs. In contrast, treatment with hydroxylamine at neutral pH only partially cleaved the ["Hlpalmitate label from the Ser" mutant TR and removed essentially none of the acyl groups attached to the Ser62*67 mutant TR (Fig. 2). However, the ["Hlpalmitate label was cleaved from both these two mutant TRs by hydroxylamine at pH 11.0. These results suggest that the fatty acid linkage of wild type human TRs expressed in CEF is a thioester bond as reported previously for TRs isolated from the human T leukemic cell line, RPM1 8402 (4). The lipid moiety of the Ser6' mutant receptor is also linked exclusively via a thioester bond as judged by its cleavage with hydroxylamine at pH 6.6. In contrast, the lipid moiety of the Serfi2,67 mutant TRs is attached via an ester linkage. The low efficiency with which the Ser6' mutant TRs were labeled precludes a firm conclusion as to the fraction of palmitate attached to Cysa and Serfi7, respectively. These data, together with the observation that the radioactivity incorporated from ["Hlpalmitate into SerG2 and Ser'*16' mutant TRs is about half that of wild type and Serfi mutant TRs, indicate that both Cys"' and Cys6' serve as acylation sites; furthermore, serine substituted at position 67 but not position 62 is also acylated in mutant receptors expressed in CEF.

Pulse-Chase Experiments of Metabolically
Labeled TRs-Previous studies in the human leukemic cell line, CCRF-CEM, suggested that the acyl moiety of the human TR may turn over more rapidly than can be accounted for by degradation of the receptor (5). The acyl groups of other palmitoylated proteins have also been found to turn over rapidly (8,15,16), and it has been suggested that fatty acylation and deacylation may play a role in protein transport from the Golgi cisternae (17,18). To re-examine whether there is differential turnover of the lipid moiety of wild type TRs and to compare the rate of turnover of the acyl group linked to the Ser62,67 mutant TR via an ester linkage, pulse-chase experiments of receptors biosynthetically labeled with ["HIpalmitate or ["'Slmethionine were performed. In addition, to investigate whether turnover may be related to receptor recycling, A3-59 mutant TRs essentially lacking a cytoplasmic domain that are internalized with -10% of the efficiency of wild type receptors were studied (9). As shown in Fig. 3, the rate of turnover of the 13H]palmitate incorporated into the wild type human TRs was much too slow for acylation and deacylation to be an obligatory step for the internalization and recycling of the receptor in CEF and was similar to the rate of degradation of the [""Slmethionine-labeled receptor protein. The kinetics of turnover of the lipid moieties of the Ser62,67 and A3-59 mutant TRs were similar to the wild type receptors. Thus, neither the linkage of the acyl group to the receptor nor the frequency with which the receptors are internalized and recycled had a significant effect on the rate of fatty acid turnover.
Ala6*," Mutant TRs Are Internalized Efficiently and Mediate Fe Uptake-To investigate whether acylation of the human TR is required for its role in iron uptake, the ability of the  Ala"'."' mutant receptors to internalize Tf and mediate iron uptake was determined. Kinetic experiments in which cells were incubated at 4 "C with '*'I-Tf, washed, and then incubated at 37 "C for various times detected no difference in the internalization and recycling of wild type and Ala"'.67 mutant receptors (data not shown). The efficient internalization of the mutant receptors was confirmed by the fact that, at steady state, 74.4 + 3.6% of the mutant receptors were found inside the cell compared with 69.2 f 3.5% of wild type receptors (Fig. 4). The rate of cellular "'Fe accumulation from "'Fe-Tf, after normalization for differences in expression of human TRs on the CEF cultures assayed, provides an independent estimate of their steady-state rate of internalization and recycling. As shown in Fig. 5 and chick TRs do not bind human Tf, the functional status of mutant human receptors can also be assessed by growth assays under conditions in which mammalian Tf is the only source of iron (9). Consistent with the results of the "Fe uptake studies, CEF expressing Alafi'*'j7 mutant TRs were able to accumulate sufficient iron from human Tf to proliferate normally (Fig. 6). DISCUSSION In the studies reported in this paper, we set out to investigate the biological role of acylation of the human TR using an experimental system in which wild type and mutant receptors were expressed at high levels in CEF using a helperindependent retroviral vector (9). Initially, we expressed mutant receptors in which cysteine residues at positions 62 and 67 were altered to serine. Previous studies in Cos cells had indicated that Cys'" was the major acylation site in the human TR with CY@~ a possible minor site and that substitution of serine for cysteine abolished acylation. We found that this was not the case when, the Ser62*67 mutant human TRs were expressed in CEF. The mutant receptors were still acylated and hydroxylamine treatment of the [3H]palmitate-labeled receptors strongly suggested that the fatty acid moiety was linked via an ester bond. Analysis of single mutants in which cysteine residues at positions 62 and 67 were independently replaced by serines is consistent with Serfi being the major ester-linked lipid attachment site. The ability of a serine residue substituted for a cysteine to serve as an alternative acylation site in membrane-associated proteins has not been documented previously. Three well-characterized  examples of  acylated proteins, the vesicular stomatitis  virus G protein,  p21H-"", the transforming  protein  of the Kirsten  sarcoma  virus, and the human &-adrenergic receptor, are normally palmitoylated on cysteine residues. Alteration of these residues, CyP' in the viral G protein (19), CyP and CyP in p21H-'"" (20), and CyP in the human &-adrenergic receptor (21), to serine abolishes palmitoylation when the mutant proteins are expressed in Chinese hamster ovary cells. However, this probably does not reflect an intrinsic structural difference between these proteins and the human TR because the receptor itself is not acylated on serine in Cos cells (6). A more likely explanation is that there is a palmitoyl acyltransferase activity in CEF capable of transferring an acyl group to serine that is not present in Cos cells or Chinese hamster ovary cells. Whether this represents a protein O-acyltransferase distinct from the acyltransferase that mediates the palmitoylation of cysteine remains to be determined. The primary structural requirements for palmitoylation of proteins are not known. However, during studies of the role of the human TR cytoplasmic domain we expressed a mutant receptor lacking residues 3-59 of the 61-amino acid cytoplasmic tail in CEF. This mutant receptor is acylated as efficiently as the wild type molecule, indicating that residues amino-terminal to LysGo are not required for addition of palmitate to the receptor (Fig. 3).
with the inner surface of the plasma membrane and for transforming activity. It is now known, however, that this cysteine residue is modified by the addition of an polyisoprenoid, farnesol, and that some ras proteins are not palmitoylated but still bind, albeit less tightly, to the plasma membrane and induce transformation (20,22). More recently, it has been reported that mutation of CyP in the carboxyl tail of the human &-adrenergic receptor exhibits a drastically reduced ability to mediate isoproterenol stimulation of adenyl cyclase although the molecular basis of the functional impairment is unknown (21). However, the functional significance of the lipid moieties of most acylated proteins remains unknown.
Because Ser'j7 of mutant human TRs can serve as an acylation site in CEF, we constructed the A1a62*67 mutant receptor in order to evaluate the role of acylation in receptor function. In all major respects, the mutant receptors were indistinguishable from wild type receptors. The A1a6'*'j7 mutant receptors bind human Tf and were internalized efficiently, and their ability to mediate iron uptake was not significantly impaired. Thus, acylation does not appear to be required for functional human TRs. Nevertheless, this does not preclude the possibility that acylation of the receptor may have a more subtle influence on receptor internalization and recycling than could be detected by the assays we employed or that it influences other properties of the receptor. In the latter, however, we did not detect any difference in the rate of degradation of wild type, A1a62,67, or Ser6*+j7 mutant receptors biosynthetically labeled with [%S]methionine* (Fig. 3). The effect of palmitoylation on the biological properties of other membrane-associated proteins has not been extensively investigated.
It was originally thought that palmitoylation of p21'"" on CyP6 was required for association of the protein ' S. Jing and I. S. Trowbridge, unpublished results.