Fatty acylation of the rat and human asialoglycoprotein receptors. A conserved cytoplasmic cysteine residue is acylated in all receptor subunits.

Functional rat or human asialoglycoprotein receptors (ASGP-Rs) are hetero-oligomeric integral membrane glycoproteins. Rat ASGP-R contains three subunits, designated rat hepatic lectins (RHL) 1, 2, and 3; human ASGP-R contains two subunits, HHL1 and HHL2. Both receptors are covalently modified by fatty acylation (Zeng, F.-Y., Kaphalia, B. S., Ansari, G. A. S., and Weigel, P. H. (1995) J. Biol. Chem.270, 21382-21387; Zeng, F.-Y., Oka, J. A., and Weigel, P. H. (1996) Biochem. Biophys. Res. Commun.218, 325-330). We report here that the single Cys residue in the cytoplasmic domain of each RHL or HHL subunit is fatty acylated. The degree of acylation is ≥90% per subunit. Deacylation of affinity-purified ASGP-Rs with hydroxylamine results in the spontaneous formation of dimers through reversible disulfide bonds, indicating that deacylation concomitantly generates free thiol groups. Reaction of hydroxylamine-treated ASGP-R with [14C]iodoacetamide resulted in the specific incorporation of radioactivity into all RHL and HHL subunits, verifying that fatty acids are attached via thioester linkages. To identify the Cys residue involved in the thioester linkages, 14C-carboxyamidomethylated RHL subunits were separated by SDS-polyacrylamide gel electrophoresis and digested in-gel with trypsin, and the resulting peptides were separated by reverse-phase high performance liquid chromatography. Amino acid sequence of radioactive peptides revealed that Cys35 in RHL1 and Cys54 in RHL2 and RHL3 were radiolabeled and, therefore, are fatty acylation sites. Fatty acylation of HHL subunits was analyzed by site-directed mutagenesis. Metabolic labeling of Cos7 cells transfected with wild type HHL1 cDNA resulted in substantial incorporation of [3H]palmitate into purified HHL1. Incorporation of [3H]palmitate into a C36S mutant of HHL1 was negligible (∼1%) compared with wild type. This result also shows that Cys57 within the transmembrane domain of HHL1 is not normally palmitoylated. We conclude that Cys35 in RHL1, Cys54 in RHL2 and RHL3, and Cys36 in HHL1 are fatty acylated. Cys57 in HHL1 and probably Cys56 in RHL1 are not palmitoylated.

The hepatic asialoglycoprotein receptor (ASGP-R) 1 has been a good model system for studying receptor-mediated endocyto-sis (1,2). ASGP-Rs mediate the endocytosis of plasma glycoconjugates containing terminal galactosyl or N-acetylgalactosaminyl residues through a coated pit pathway. The receptorligand complexes are rapidly endocytosed and dissociated within endosomes. Dissociated ligands are delivered to lysosomes for degradation, and receptors are recycled back to the cell surface. Previous studies by us (2)(3)(4) and others (5)(6)(7) have demonstrated that ligands are endocytosed and intracellularly processed via two distinct pathways by two functionally distinct receptor populations, designated State 1 ASGP-Rs and State 2 ASGP-Rs (4).
That State 2 ASGP-Rs undergo a transient inactivation/ reactivation cycle during receptor recycling indicates that regulation of receptor activity may be an important, and perhaps general, characteristic during endocytosis and receptor recycling (8 -10). In permeable rat hepatocytes, this State 2 receptor population is inactivated by ATP in the absence of cytosol (9), and the ATP-inactivated receptors can be quantitatively reactivated by the simple addition of palmitoyl-CoA (10,11). Recently, we showed in both rat hepatocytes (12,13) and human hepatoma cells (14) that State 2 ASGP-Rs are palmitoylated and that depalmitoylation with hydroxylamine causes inactivation of this same receptor population. We proposed that a dynamic fatty acylation/deacylation cycle may be the molecular basis for the State 2 ASGP-R inactivation/reactivation cycle in vivo (12,14).
Physiologically, this receptor inactivation/reactivation cycle during endocytosis could explain why endocytosis mediated by a variety of recycling receptors is so efficient (2). Receptor inactivation shortly after internalization would eliminate the possibility of dissociated ligand rebinding to receptors during the segregation step in which the two are being separated and directed to different intracellular pathways. Transient receptor inactivation also ensures that the efficiency of the segregation process would not decrease as the extracellular ligand concentration (and presumably the need for efficient clearance) increased.
The rat ASGP-R is an integral transmembrane glycoprotein composed of three polypeptide subunits, designated rat hepatic lectin (RHL) 1, 2, and 3, with molecular masses of 41.5, 49, and 54 kDa, respectively. The three subunits are the products of two different genes (15). RHL2 and RHL3 contain the same polypeptide backbone but differ in the type and extent of posttranslational carbohydrate modification (16). The human ASGP-R contains two subunits designated HHL1 and HHL2, also encoded by two genes (1). Each RHL and HHL subunit contains a single conserved Cys in its cytoplasmic domain within four residues of the transmembrane domain (15). All hepatic lectin subunits also contain a conserved Cys residue * This research was supported by National Institutes of Health Grant GM 49695. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Cys residues at or near the junction between cytoplasmic and transmembrane domains are frequently potential palmitoylation sites. Such cysteines are palmitoylated in many membrane proteins such as the transferrin receptor (17), the G proteincoupled receptors (18,19), the HLA-D-associated invariant chain (20) and the influenza virus hemagglutinin (21).
We recently found that all subunits in both rat and human ASGP-Rs are fatty acylated (13,14). In both cases [ 3 H]palmitate could be readily incorporated into active ASGP-Rs, and mild treatment with hydroxylamine caused fatty acid deacylation and concomitant loss of ligand binding activity. Although the chemical sensitivity of thioester linkages to cleavage by hydroxylamine is an indication that this labile bond is present, it is not proof. Normal ester bonds, for example involving Ser or Thr, or even peptide bonds, can be cleaved by hydroxylamine if these bonds are under appropriate conformational strain. In order to understand the molecular mechanism by which the ligand-binding activity of the ASGP-R is regulated by fatty acylation, we have determined the fatty acylation sites of the three receptor subunits. Using a novel chemical method combined with site-directed mutagenesis, our results show that the single Cys in the cytoplasmic domain of each subunit is palmitoylated, whereas the conserved transmembrane domain Cys is not. Preparation of Active ASGP-Rs-Isolated hepatocytes from Sprague-Dawley rats (SasCo, Oklahoma City, OK) were prepared by a modification (23) of a collagenase perfusion procedure (24). The cells were first incubated in medium 1/bovine serum albumin at 37°C for 1 h to increase and stabilize the total cell ASGP-R activity. Active ASGP-Rs were then purified by detergent extraction and by affinity chromatography using ASOR-Sepharose as described (12).

Materials-Hydroxylamine
Treatment of ASGP-Rs with Hydroxylamine and Iodoacetamide-Freshly purified, active ASGP-Rs were incubated with 1 M hydroxylamine in buffer A (10 mM Hepes, pH 7.4, 137 mM NaCl, 6.7 mM KCl) containing 10 mM EGTA at 4°C or at room temperature for the indicated times prior to SDS-PAGE. To titrate the generation of thiol groups by hydroxylamine treatment, samples were incubated with 500 M [ 14 C]iodoacetamide (60 mCi/mmol) at room temperature. The samples were then immediately mixed with an equal volume of concentrated Laemmli (25) sample buffer (125 mM Tris-HCl, pH 6.8, 20% glycerol, and 2% SDS) and subjected to SDS-PAGE.
Enzymatic Digestion-The individual 14 C-carboxyamidomethylated subunits were digested with enzymes in-gel (26). Briefly, radioactive labeled rASGP-Rs were separated by nonreducing SDS-PAGE, the individual subunits (RHL1, 2 and 3) were visualized by Coomassie Blue R-250, excised, and cut into small slices. The gel pieces were washed twice with 0.25 ml of 50% acetonitrile in 200 mM ammonium carbonate, pH 8.9, for 20 min each at 30°C and left to semi-dry at room temperature. The gel pieces were then partially rehydrated with 10 l of 200 mM ammonium carbonate, pH 8.9, containing 0.02% Tween 20 and then 5 l of enzyme solution (0.5 mg/ml in 200 mM ammonium carbonate, pH 8.9) was added. After absorption of the enzyme solution, ammonium carbonate buffer was added until the gel slices were immersed. The digestions were carried out at 37°C for 20 h and terminated by the addition of 3 l of trifluoroacetic acid . The resulting peptides were recovered by two extractions of 30 min each with 200 l of 60% acetonitrile in 0.1% trifluoroacetic acid at 30°C with shaking. The extracts were combined and concentrated to dryness in a SpeedVac.
Reverse-phase HPLC and Peptide Sequencing-The peptides were redissolved in 0.1% trifluoroacetic acid and separated by reverse-phase HPLC on a Vydac RP-300 C18 column (250 ϫ 4.6 mm, 5 m) using a linear gradient of acetonitrile in 0.1% trifluoroacetic acid at a flow rate of 1 ml/min and monitored at 220 nm. Five hundred-l fractions were collected, and radioactivity (0.1 ml) was determined using a liquid scintillation counter (Packard Tri-Carb 2300TR). The radioactive fractions were concentrated in a SpeedVac. Peptide sequencing performed by the Molecular Biology Resource Facility (OUHSC) was carried out by automatic Edman degradation using a protein sequencer (Applied Biosystems Model 470A) equipped with an HPLC chromatograph (model 120A) to identify the phenylthiohydantoin derivatives of amino acids.
Molecular Biology Reagents-Oligonucleotides were obtained from Ransom Hill Bioscience (Ramona, CA). Enzymes used for molecular cloning were from Promega. Plasmid Miniprep and EndoFree plasmid maxi kits were from Qiagen. The expression vector pcDNA3.1(ϩ) and calcium phosphate transfection kit were from Invitrogen. Sequenase Version 2.0 DNA sequencing kit was from USB. Ligations were performed using the Ligation Expression kit from Clontech. General molecular biology procedures were performed as described in Sambrook et al. (27). DNA sequencing was performed by the dideoxynucleotide procedure of Sanger et al. (28).
Site-directed Mutagenesis-Mutants were generated using Promega's Altered Site ® II in vitro mutagenesis system according to the manufacturer's instructions. Plasmids pGA1, containing full-length HHL1, and pKA2b, containing full-length HHL2b (29), were generously provided by Dr. Martin Spiess (University of Basel, Switzerland). HHL1 cDNA was subcloned into pALTER Ex-1 Vector (Promega) using HindIII and EcoRI. HHL2 cDNA was subcloned into pALTER Ex-1 Vector using BamHI and EcoRI. The Cys residues in the cytoplasmic and transmembrane domains of HHL1 were replaced by Ser residues, yielding the single mutants HHL1(C36S) and HHL1(C57S). The mutagenic (sense) oligonucleotides used were the following (the altered nucleotides are in bold): 5Ј-C CTG CAG CGT CTC AGC TCC GGA CCT CG-3Ј (C36S); and 5Ј-T GTG GTT GTC AGC GTG ATC GGA TCC C-3Ј (C57S). All mutations were verified by DNA sequencing.
Transfection and Metabolic Labeling-Wild type and mutant HHL1 cDNAs were subcloned into the pcDNA3.1(ϩ) expression vector using HindIII and EcoRI restriction sites. For transfection, plasmid DNA was purified from Escherichia coli Top10FЈ using the EndoFree plasmid maxi kit. COS-7 cells (obtained from ATCC) were maintained in Dulbecco's modified Eagle's medium supplemented with 10% (v/v) fetal calf serum, 2 mM L-glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin in an atmosphere of 5% CO 2 at 37°C. Cells were transfected at ϳ50% confluency with pcDNA3.1(ϩ) plasmids using a calcium phosphate (30) transfection kit according to the manufacturer's instructions. Cells were labeled 24 h posttransfection with 400 Ci/ml [ 3 H]palmitic acid at 37°C for 3 h, and expressed HHL1 protein was purified by affinity chromatography using ASOR-Sepharose 4B as described previously (14).
General-SDS-PAGE was carried out by the method of Laemmli using 12% polyacrylamide gels (25). The protein bands were visualized by staining with silver (31) or Coomassie Blue R-250. The protein content was determined by the method of Bradford (32) using bovine serum albumin as a standard. Western blotting was performed using receptor subunit-specific antibodies as described earlier (33). Fluorography was performed as described previously (12).

RESULTS
Fatty acids are usually attached to proteins through amide or ester linkages. An ester linkage is labile to mild alkaline treatment, and a thioester bond can be easily cleaved by hydroxylamine, whereas an amide linkage is essentially resistant to such treatments (34). At steady state, ASGP-Rs contain both palmitic acid and stearic acid, and mild treatment with hydroxylamine releases Ͼ70% of the total fatty acids from each of the three subunits (13). The same treatment with hydroxylamine releases almost all (Ͼ90%) of the metabolically incorporated [ 3 H]palmitic acid from active rat or human ASGP-Rs, suggesting that this palmitate is attached to cysteine residues through thioester bonds (12,14). To investigate further the linkage of fatty acids to individual subunits, hydroxylamine-treated rat ASGP-Rs were analyzed by SDS-PAGE under nonreducing conditions.
As shown in Fig. 1A, hydroxylamine treatment of rASGP-Rs at pH 7.4 results in the formation of several new protein bands, primarily in the range of 90 -110 kDa, in a time-dependent manner. The control treatment with Tris had no effect. A small amount of larger oligomers was also observed. Reduction with ␤-mercaptoethanol completely abolished these larger bands (Fig. 1B) and regenerated the normal RHL subunits, indicating that these new, larger proteins were formed via disulfide bonds after hydroxylamine treatment. Taking the molecular masses into account, most of the newly formed proteins are likely to be dimeric RHLs. Disulfide bonds could form between any two RHL subunits if free thiol groups are generated after removal of fatty acids. The same behavior is found with purified hASGP-Rs, which also form dimers upon hydroxylamine treatment (Fig. 2).
To determine the composition of the dimeric RHLs, we performed Western analyses using two receptor subunit-specific antibodies (33), raised against synthetic peptides corresponding to the C-terminal 17 amino acids of RHL1 and RHL2/3 (Fig.  3). The 90 kDa-band was recognized by the RHL1-specific antibody but not by the RHL2/3-specific antibody, whereas the broad band at 95-110 kDa reacted with the RHL2/3-specific antibody but not the RHL1-specific antibody (Fig. 3). These results clearly show that deacylation of active rASGP-Rs with hydroxylamine under mild conditions results in the formation of predominantly homodimers (RHL1-RHL1 and RHL2/3-RHL2/3). A similar analysis of the disulfide-bonded dimer composition from hASGP-R (Fig. 2) was not feasible, since the HHL1 and HHL2 subunits are not usually resolved by SDS-PAGE.
We interpret the above results to mean that spontaneous disulfide bond formation occurs after fatty deacylation of ASGP-R subunits. The assumption that the dimers formed during hydroxylamine treatment were deacylated was confirmed by examining ASGP-Rs that had been metabolically labeled with [ 3 H]palmitic acid ( Fig. 2 and Fig. 4). After hydroxylamine treatment, essentially all of the palmitate was removed from the monomeric HHL or RHL subunits, and no radioactivity was found with the dimeric subunits ( Fig. 2A,  lane 2 and Fig. 4B, lane 2).
To verify the presence of a thioester linkage in each subunit, the OSH groups generated by deacylation with hydroxylamine were reacted with the alkylating reagent iodoacetamide. Hydroxylamine-induced dimer formation was effectively prevented by alkylating agents such as iodoacetamide and Nethylmaleimide (not shown). Since alkylation occurs more readily at room temperature than at 4°C, we carried out the reaction at room temperature. Hydroxylamine treatment under these conditions caused dimerization of about 50% of rASGP-Rs (Fig. 1A, lane 6). When purified rASGP-Rs were incubated with [ 14 C]iodoacetamide in the absence of hydroxylamine, none of the subunits were detectably labeled, suggesting that free thiol groups, or other reactive groups, are absent in active ASGP-R preparations (Fig. 5B, lane 1). Treatment with 1 M hydroxylamine resulted in extensive incorporation of radioactivity into all three subunits (Fig. 4B, lane 3). The control incubation in the presence of 1 M Tris, pH 7.4, gave a much smaller increase in the amount of labeled RHL subunits (Fig. 5B, lane 2). Densitometric quantitation showed that the increase with Tris was about 19% of the hydroxylamine effect. The effect of Tris treatment reflects the lability of thioesters and our previous findings that freshly purified State 2 ASGP-Rs lose activity with storage (12). Essentially identical results were obtained with purified hASGP-R (not shown). These results show that [ 14 C]iodoacetamide specifically labels hydroxylamine-generated free thiol groups and no other functional groups in the ASGP-R. This finding became the basis of our biochemical approach to determine the sites of thioesterlinked fatty acylation.
The stoichiometry of fatty acylation was estimated from these radiolabeling experiments based on the specific radioactivity of [ 14 C]iodoacetamide and the amount of ASGP-R analyzed. RHL1, RHL2, and RHL3 were estimated to contain an average of 0.94, 1.0, and 1.3 fatty acyl groups/subunit, respectively. This stoichiometry of approximately one fatty acid per subunit is consistent with our previous gas-liquid chromatography-mass spectrometry results (13).
The single Cys residue in the cytoplasmic domain of RHL1, RHL2/3, and HHL1 is a potential fatty acylation site. The fact that the N-terminal amino acids of the three RHL subunits are blocked precluded direct determination of the 14 C-carboxyamidomethylated Cys by N-terminal sequence analysis of intact subunits. However, we have observed that partial digestion of ASOR⅐rASGP-R complexes with subtilisin in the presence of Ca 2ϩ first produces a 40-kDa fragment and then a 22-kDa fragment, both containing the carbohydrate recognition domain (CRD) still bound to ligand (33). These CRD fragments are specifically released from the ASOR-Sepharose by elution with EGTA (Fig. 6, lane 2). Both fragments were identified as the products of RHL1 by using subunit-specific antibodies (Fig. 6, lanes 3 and 4). Under our digestion conditions, RHL2/3 appears to be completely degraded; no CRD domains are recovered from these subunits. Treatment of the two RHL1derived proteins with hydroxylamine and [ 14 C]iodoacetamide resulted in specific 14 C incorporation into the 40-kDa protein but not the 22-kDa CRD (Fig. 6, lane 5). This result corroborates the conclusion that the extracellular CRD contains no free thiol groups or other reactive groups that are present in the untreated ASGP-Rs or generated by hydroxylamine treatment.
To identify the radiolabeled amino acid, the 14 C-carboxy-amidomethylated 40-kDa protein was transferred onto a polyvinylidene difluoride membrane followed by N-terminal sequence analysis, in which the cycle fractions were also assessed for radioactivity (not shown). A major peptide was present whose N-terminal sequence of 20 amino acids corresponded to Gln 21 -Arg-Gly-Pro-Pro-in RHL1 (RHL1 cleaved at Leu 20 -Gln 21 ). A minor peptide that was two amino acids shorter than the major peptide was also detected. The 15th amino acid of the major peptide and the 13th amino acid of the minor peptide could not be detected. The peak of radioactivity (88% of total) was detected in cycles 14 -16, corresponding to Cys 35 (at cycle 15) of the major peptide. A small amount of radioactivity (6% of total) was also detected in the 13th cycle, consistent with Cys 35 in the minor peptide. These results are consistent with the conclusion that iodoacetamide specifically and uniquely modified hydroxylamine-generated free cysteines. The same approach failed to identify palmitoylation sites of the minor RHL2 and RHL3 subunits, since subtilisin readily digested both proteins into small peptides (Fig. 6). To determine their palmitoylation sites, 14 C-carboxyamidomethylated RHL2/3 subunits (prepared as in Fig. 5) were subjected to in-gel digestion with trypsin. The recovered peptides were separated by reverse-phase HPLC, and radioactive fractions were identified (Fig. 7). RHL2 (Fig. 7B) and RHL3 (Fig. 7C) showed similar peptide maps, and radioactivity was detected in fractions 4 (peak 1) and 15/16 (peak 2). Repurification of the combined fractions 15 and 16 using a shallower gradient gave only a low recovery and no significant improvement in purity (not shown). RHL1 gave a different peptide map (Fig. 7A) with radioactivity in fractions 4 (peak 1), 27 (peak 2), and 29 (peak 3). The individual radioactive fractions were directly analyzed by N-terminal sequencing (Table I). For RHL2 and RHL3, fractions 15 and 16 gave the same sequences, both consisting of two peptides in a molar ratio of 3:1 to 2:1. The major peptide was matched with Thr 35 -Glu-Asn-Pro-Arg 39 (RHL2/3 cleaved at Arg 34 -Thr), and the minor peptide was Leu 53 -(Cys)-Ser-Lys 56 which represents RHL2/3 cleaved at Arg 52 -Leu (Table I). Radioactivity was detected only in the second cycle, indicating that this cycle contained 14 C-carboxyamidomethylated cysteine. These results demonstrate that Cys 54 of RHL2 and RHL3 is modified by attachment to fatty acid. Fraction 4 (P1) in all cases (Fig. 7, A-C) probably contains free [ 14 C]iodoacetamide, since no peptide sequence could be detected. The sequence analysis of peak 2 from RHL1 also gave two peptides in a molar ratio of about 3:1. The major peptide corresponds to Leu 103 -Val-Glu-Ser-Gln-Leu-Glu-Lys 110 of RHL1 (cleaved at Lys 102 -Leu). The minor peptide was identified as Leu 34 -(Cys)-Ser-Gly-Phe 38 derived from RHL1 by cleavage at Arg 33 -Leu (Table I). The second residue of this peptide should be Cys, and as expected, radioactivity was detected only in the second cycle. Peak 3 gave a major sequence corresponding to Gln 75 through Lys 89 of RHL1 and several minor peptides at similar molar ratios that have not been further analyzed, but radioactivity was detected only in the second cycle. These results are consistent with the above sequence analysis of the 40-kDa fragment (seen in Fig. 6) and verify that Cys 35 in RHL1 is fatty acylated. Thus, the single cysteine residue in the cytoplasmic domains of all three rat subunits is identified as an essentially stoichiometric fatty acylation site. Similar attempts to isolate and sequence 14 C-labeled peptides from hASGP-R, prepared as in Fig. 5, failed. The above studies did not determine whether the transmembrane Cys, which is conserved in all RHL and HHL subunits, is fatty acylated. No radiolabeled peptides of this region could be identified after hydroxylamine treatment and reaction with iodoacetamide. On the other hand, freshly purified ASGP-R from human or rat showed no significant reactivity with this reagent, suggesting that either no free OSH groups are present in active receptor or any free Cys residues are inaccessible to the alkylating agent. No reaction with iodoacetamide could be demonstrated even in the presence of SDS.
To resolve this uncertainty, we used site-directed mutagenesis to change Cys to Ser in the cytoplasmic and transmembrane domains of HHL1. This subunit is expressed well alone when COS-7 cells are transfected with a cDNA encoding wild type protein (Fig. 8A, lane 1). The expressed protein is readily detected after metabolic labeling with [ 3 H]palmitate (Fig. 8B,  lane 1). In contrast, the HHL1(C36S) mutant incorporates virtually no palmitate, although the protein is expressed as well as the wild type (Fig. 8, lanes A2 and B2). This result shows that Cys 36 in HHL1 is fatty acylated, like the homologous Cys 35 in RHL1. Additionally, dimer formation was much greater for wild type HHL1 compared with HHL1(C36S), probably reflecting greater spontaneous deacylation and disulfide bond formation during the purification of HHL1 containing Cys 36 .
Since mutation of the single cytoplasmic Cys completely abolishes palmitoylation of the protein, we also conclude, therefore, that transmembrane Cys 57 is not normally palmitoylated. Interestingly, neither the single HHL1(C57S) mutation nor the double HHL1(C36S,C57S) mutant was stably expressed in COS-7 cells. Little or no HHL1 protein could be detected. Although transmembrane Cys 57 is not palmitoylated and not involved in a disulfide bond, it may be necessary for assembly of HL subunits into a stable oligomeric ASGP-R. DISCUSSION The human or rat ASGP-Rs consist, respectively, of two or three polypeptide subunits that are the products of two different genes (15). RHL2 and RHL 3 have the same core polypeptide, differing only in the type and extent of carbohydrate modification. The primary structures of RHL1 and RHL2/3 show a high degree of identity. Each subunit has 10 cysteine residues at conserved positions (15), of which one is located in the N-terminal cytoplasmic domain, one in the transmembrane domain, and eight in the extracellular domain; the latter form four intra-chain disulfide bonds (35). In addition, HHL2 has a Cys at position 58 (1). Our results (Fig. 5) confirm the conclusion that all extracellular cysteines in RHL1 participate in the formation of disulfide bonds. Previous studies demonstrated that the rat ASGP-R is covalently modified by palmitic acid and   FIG. 7. Purification of 14 C-carboxyamidomethylated peptides by reverse-phase HPLC. Hydroxylamine-generated thiol groups were reacted with [ 14 C]iodoacetamide as in Fig. 4. The individual subunits RHL1 (A), RHL2 (B), and RHL3 (C) were separated by SDS-PAGE on a 12% (w/v) gel and digested with trypsin as described under "Experimental Procedures." The resulting peptides were separated by reversephase HPLC on a C18 column with a linear gradient of acetonitrile in 0.1% trifluoroacetic acid at a flow rate of 1 ml/min. Five hundred l-fractions were collected, and 100 l/fraction was removed to determine radioactivity using a liquid scintillation counter. The bars indicating A 220 (OD(220nm)) represent an absorbance of 0.04. stearic acid (13). Metabolic labeling studies verified that both rat and human ASGP-Rs contain palmitate (12)(13)(14). The lability of this palmitate linkage to hydroxylamine treatment suggested the involvement of thioester linkages of fatty acids to these proteins. Treatment with reducing agents such as dithiothreitol and ␤-mercaptoethanol also partially removed [ 3 H]palmitate from purified ASGP-Rs (not shown).
Hydroxylamine treatment causes a time-and temperaturedependent formation of dimeric RHL or HHL subunits through disulfide bonds. This observation suggests that hydroxylamine releases fatty acids and concomitantly generates free thiol groups, consistent with fatty acid attachment to cysteines through thioester bonds (34). The formation of new disulfide bonds upon hydroxylamine treatment has also been found in some palmitoylated proteins, such as the human tissue factor (36) and the vesicular stomatitis G glycoprotein (37). Since the same RHL dimer pattern is observed when intact hepatocytes are treated with hydroxylamine, this behavior is not an artifact of detergent solubilization or receptor purification (not shown). Although our approach of chemically deacylating and derivatizing the resulting free OSH group was successful with the ASGP-R, this method may not be generally applicable. We were fortunate that only one free Cys is present in the cytoplasmic domain of each subunit, which made analysis relatively simple.
The physiological significance of ASGP-R subunit dimer formation upon release of fatty acids was not addressed in this study and remains unknown. It raises the interesting possibility, however, that changes in the number of fatty acylated Cys residues (or the pattern of acylated subunits) within the oligomeric ASGP-R could lead to the formation or cleavage of disulfide bonds between subunits. Such changes could clearly alter the conformation of both internal and external domains of the receptor and lead to new or altered interactions with other regulatory proteins (e.g. protein kinases or thioesterases) or with ligand. Although several studies have demonstrated that the functional rat and human ASGP-Rs are hetero-oligomers (2,38,39), the structure of the native ASGP-R remains debated. Halberg et al. (40) showed only homo-oligomeric products in cross-linking experiments using hepatocyte microsomes, suggesting that no physical association of RHL1 and RHL2/3 exists. That we observe essentially exclusive formation of homodimeric RHL1-RHL1 or RHL2/3-RHL2/3 upon hydroxylamine treatment of affinity-purified ASGP-Rs suggests that rASGP-R exists in solution either in a homo-oligomeric form or in a particular hetero-oligomeric form in which like subunits are closer to, or more likely to react with, each other. Due to the lack of resolution between HHL1 and HHL2, similar data and conclusions for the human ASGP-R cannot be obtained.
As noted in Fig. 3, some oligomeric RHLs that are probably RHL1 trimers also formed upon hydroxylamine treatment of ASGP-Rs. These oligomers appear to be disulfide-bonded, since reducing agents convert them to monomeric RHLs. Freshly purified active ASGP-Rs do not contain interchain disulfide bonds, even though the transmembrane Cys residues are free. Formation of trimers after deacylation with hydroxylamine could be explained if the removal of fatty acid chains attached to the cytoplasmic domains enhanced the ability of two adjacent transmembrane cysteines to form a disulfide bond. One subunit could be disulfide-bonded to a second subunit through Cys 35 linkages and to a third subunit through a Cys 56 linkage.
We have proposed that a reorganization or altered packing of transmembrane domains occurs when oligomeric receptors are fatty acylated or deacylated in their cytoplasmic domains because the acyl chains will intercolate into the membrane among the transmembrane domains (14). Changes in the relative spacing of these membrane domains, as acyl chains are added or removed, would translate into changes in the spacing of the external CRDs and regulate ligand binding. The cytoplasmic Cys residues are palmitoylated in many proteins (34), and the transmembrane cysteines in some proteins (usually near the cytoplasmic domain junction), such as the transferrin receptor (41) and the cell surface glycoprotein CD4 (42), are also modified by fatty acids. Our mutagenesis results, however, show that the Cys residue in the transmembrane domain of HHL1 subunits is not palmitoylated.
In the present study we find that native ASGP-Rs contain close to one thioacyl group per subunit. Even though the transmembrane Cys is not palmitoylated, freshly purified ASGP-R is not carboxyamidomethylated despite partial denaturation of ASGP-Rs with SDS or by heating prior to alkylation. Either this Cys is not accessible to the polar reagent or the residual secondary and tertiary structure of the protein renders this thiol group nonreactive with iodoacetamide. Our biochemical and mutagenesis results show that Cys 35 in RHL1, Cys 54 in RHL2/3, and Cys 36 in HHL1 are modified by fatty acids. Saxena and Fallon (43) have reported preliminary results that HHL2 is likely palmitoylated at Cys 54 as well. It appears then  Fig. 7 were subjected to N-terminal amino acid sequence analysis, and the data were compared with the published sequences of RHL 1 and RHL 2/3 (15). A small portion (ϳ10%) of each fraction was subjected to scintillation counting. Radioactivity was detected only in the second cycle in all cases.  A2 and B2) and metabolically labeled with [ 3 H]palmitate as described under "Experimental Procedures." Purified HHL1 was analyzed by nonreducing SDS-PAGE followed by fluorography (B) or Western blotting (A) with rabbit antiserum to hASGP-R. The arrow indicates the broad HHL band at [35][36][37][38][39][40] that the same fatty acylated Cys residue is conserved in all subunits in both rat and human ASGP-Rs.
When the location of the acylation sites in the three RHL and two HHL subunits is compared with that in other integral membrane proteins in which Cys is near the transmembrane and cytoplasmic domain junction, it is clear that there is no unique or common requirement for the spacing of the Cys relative to the transmembrane domain. In the case of RHL1 and RHL2/3, the Cys is located four amino acids from the transmembrane domain; this distance is six amino acids in the vesicular stomatitis G protein (44), the influenza virus hemagglutinin (21), and the 63-kDa integral membrane protein (45). In contrast, there are one and two amino acid distances between the cytoplasmic Cys and the transmembrane domains identified in the cell surface glycoprotein CD4 (42) and the HLA-D-associated invariant chain (20), and 12-13 amino acid distances in the human ␤ 2 -adrenergic receptor (18) and bovine rhodopsin (46). There are also no known consensus sequences for the palmitoylation of membrane proteins although the palmitoylated Cys residues are frequently preceded and/or followed by basic amino acids such as are found in each hepatic lectin sequence (15). The positive charges in the vicinity of the acylation site might serve as part of the recognition site for the appropriate fatty acyltransferase (47). We are presently undertaking further site-directed mutagenesis studies in order to understand the structural requirements for palmitoylation and how ASGP-R structure and function is regulated by fatty acylation/deacylation.