The major subunit of the rat asialoglycoprotein receptor can function alone as a receptor.

Mammalian hepatic asialoglycoprotein receptors (ASGP-R) are composed of two unique, but closely related polypeptides, which in the rat are designated rat hepatic lectins 1 and 2/3 (RHL 1, RHL 2/3). Despite numerous studies, the composition of a functional ASGP-R has remained unclear. We examined this question in rat hepatoma tissue culture (HTC) cells (which lack endogenous ASGP-R) that were co-transfected with cDNAs for both RHL 1 and RHL 2/3. The original population was cloned, but derivatives were unstable. We therefore used fluorescence-activated cell sorting to separate a subpopulation of cells (positive) that specifically endocytosed fluoresceinated asialoorosomucoid (ASOR) from one that did not (negative). We then used indirect immunofluorescence with polypeptide-specific ASGP-R antibodies, immunoanalysis, and binding and uptake studies with two Gal ligands (ASOR and NAc-galactosylated poly-L-lysine (Gal-Lys] to further define the ASGP-R status in these two populations. As reported by others, we found that expression of both RHL 1 and RHL 2/3 in the positive cells resulted in binding, uptake and degradation of ASOR, the most commonly used ASGP-R ligand. The negative cells expressed only RHL 1 and neither bound nor processed ASOR. However, the presence of RHL 1 was sufficient for specific high affinity binding and processing of the synthetic ligand, Gal-Lys, by negative cells. These results show that RHL 1 can function as an ASGP-R, given a highly galactosylated ligand, and that RHL 2/3 must play an important role in the organization of native ASGP-R in the membrane.


The Major Subunit of the Rat Asialoglycoprotein Receptor Can Function Alone as a Receptor*
(Received for publication, August 10, 1988) Lelita T. Braiterman, Suzette C. Chance, William R. Porter, Yuan C. Lee Mammalian hepatic asialoglycoprotein receptors (ASGP-R) are composed of two unique, but closely related polypeptides, which in the rat are designated rat hepatic lectins 1 and 2/3 (RHL 1, RHL 2/3). Despite numerous studies, the composition of a functional ASGP-R has remained unclear.
We examined this question in rat hepatoma tissue culture (HTC) cells (which lack endogenous ASGP-R) that were co-transfected with cDNAs for both RHL 1 and RHL 2/3. The original population was cloned, but derivatives were unstable. We therefore used fluorescence-activated cell sorting to separate a subpopulation of cells (positive) that specifically endocytosed fluoresceinated asialoorosomucoid (ASOR) from one that did not (negative). We then used indirect immunofluorescence with polypeptide-specific ASGP-R antibodies, immunoanalysis, and binding and uptake studies with two Gal ligands (ASOR and NAc-galactosylated poly-L-lysine (Gal-Lys)) to further'define the ASGP-R status in these two populations. As reported by others, we found that expression of both RHL 1 and RHL 2/3 in the positive cells resulted in binding, uptake and degradation of ASOR, the most commonly used ASGP-R ligand. The negative cells expressed only RHL 1 and neither bound nor processed ASOR. However, the presence of RHL 1 was sufficient for specific high affinity binding and processing of the synthetic ligand, Gal-Lys, by negative cells. These results show that RHL 1 can function as an ASGP-R, given a highly galactosylated ligand, and that RHL 2/3 must play an important role in the organization of native ASGP-R in the membrane.
Mammalian hepatic asialoglycoprotein receptors (ASGP-R)' mediate the specific endocytosis and lysosomal degradation of a wide variety of glycoconjugates, all of which bear either a terminal galactose (Gal) or N-acetylgalactosamine (GalNAc) residue (for reviews, see Ashwell and Harford, 1982;GM29133 (to A. L. H.), DK31376 (to R. R. T.), and DK09970 (to Y. * This work was supported by National Institutes of Health Grants C. L.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ T o whom correspondence should be addressed Dept. of Cell Biology and Anatomy, 725 N. Wolfe St., Baltimore, MD 21205.
The ligand affinity-purified ASGP-R of rat liver is composed of three polypeptides, a major subunit called rat hepatic lectin 1 (RHL 1) at 42,000 daltons and two minor subunits, RHL 2 and RHL 3, at 49,000 and 54,000 daltons, respectively. Examination of primary sequence  has indicated that the three polypeptides are closely related, and molecular cloning studies have identified two unique, full length ASGP-R cDNAs (Holland et al., 1984;Halberg et al., 1987). RHL 2 and 3 are products of one gene and presumably differ only in the presence of polylactosamine linkages on RHL 3 (Halberg et al., 1987), giving rise to the RHL 2/3 designation. The major difference between RHL 1 and RHL 2/3, each of which has only one transmembrane segment, is the presence of a unique 18-amino acid sequence near the NH2-terminal cytoplasmic domain of the latter. Drickamer and colleagues have shown that each of the three RHL subunits has at least one ectoplasmic carbohydrate recognition domain (Halberg et al., 1987).
Despite extensive work, the precise composition of a functional ASGP-R has remained elusive. Binding studies using defined synthetic oligosaccharides, glycopeptides, and glycoproteins have resulted in a model for the ASGP-R in which occupancy of three Gal/GalNAc combining sites (a triad) is required for high affinity binding (Lee et al., 1984b;Townsend, 1987, for review). Results of cross-linking experiments to assess the oligomeric state of the rat ASGP-R have indicated that RHL 1 and RHL 2/3 are not physically linked and thus may be independent galactose-binding proteins (Halberg et al., 1987). However, more recent studies using antibody-induced surface receptor loss as well as chemical cross-linking suggest that the human analogues of RHL 1 and 2/3 are physically associated with each other in HepG2 cells (Bischoff et al., 1988). Finally, results of cDNA transfection experiments have indicated that both rat subunits are essential for receptor function (McPhaul and Berg, 1986).
In this study, we have examined two sub-populations of rat hepatoma tissue culture cells (HTC) derived from a population originally transfected with cDNA's for both RHL 1 and 2/3 (McPhaul and Berg, 1986). We show that one subset of cells, which expresses only the RHL 1 polypeptide, has lost its ability to bind ASOR but expresses a high affinity for, internalizes, and degrades a highly galactosylated synthetic ligand. Thus, RHL 1 can function as a receptor in the absence of RHL 2/3, but its organization in the membrane must be different than that of the native ASGP-R, implying that RHL 213 and RHL 1 do associate with one another when both are present.

Cell Culture
Rat HTC (Thompson, 1979) co-transfected cells (McPhaul and Berg, 1986) were cultured using G418 selection conditions as described (Southern and Berg, 1982), except that the DMEM was supplemented with 10% fetal calf serum and contained no penicillin or streptomycin.

Identification of ASGP-receptor Subunits in BC6 Cells
Immunofluorescence Analysis-Chamber slides were coated with poly-D-lysine (200 pg/slide according to manufacturer's instructions) and then plated with 5 X lo' cells. After 2 days in culture, the cell monolayers were fixed (20 min, 2% paraformaldehyde, 0.075 M lysine, 0.01 M NaI04, 0.037 M NaP04, pH 7.4). The slides were subsequently processed at 22 "C with 0.22 pm filtered solutions and 5 min incubations except as indicated. The cell monolayers were rinsed twice in PBS, quenched in 0.25% NH4Cl (w/v)-PBS, permeabilized with 0.06% digitonin-PBS, and then nonspecific sites blocked with 0.2% gel-PBS. The first antibody (either anti-1D (10 pglml), anti-2/3D (1:120) or diluent (0.2% gel-PBS)) was applied for 15 min followed by three rinses in PBS. The second antibody (FITC-goat anti-rabbit in 0.2% gel-PBS) was then applied for 15 min, followed again by three rinses in PBS. The slides were mounted with freshly prepared anti-fade solution (Johnson et al., 1982) made with 25% glycerol. Cells were observed by epifluorescence and phase contrast with an Axioplan Universal Microscope (Zeiss, West Germany). Photographs were taken with Kodak T MAX 400 film at an ASA setting of 800.
ASOR was fluoresceinated (F-ASOR) as follows. ASOR (1 mg) and FITC (5 mg) in 2.8 ml of 0.35 M NaHC03, pH 9, were incubated at 23 "C for 4 h, and then at 4 "C for 16 h in the dark. F-ASOR was recovered in the void volume of a PD-10 column equilibrated and run in 0.15 M NaC1, 0.02 M NaP04, pH 7.4.

Endocytosis Studies
F-ASOR (IS Ligand-Routinely, 5 X lo5 cells were plated in 35-mm wells and grown for 2 days. The cell monolayers were rinsed with DMEM, and then 1 pg/ml F-ASOR in DMEM was added. After incubation at 37 "C for 2.5-3 h, the cells were rinsed two times with GKN, removed from the wells with 0.05% trypsin, 0.02% EDTA, and resuspended in 10% fetal calf serum/DMEM. 100-fold excess unlabeled ASOR in the presence of F-ASOR was added to assess nonspecific uptake by the cells. The fluorescence intensity of the cells was analyzed in an EPICS 752 flow cytometer (Coulter Electronics Inc., Hialeah, FL). When positive and negative populations were separated, the labeled cells were analyzed in a FACS I1 flow cytometer (Becton-Dickinson FACS Systems, Mountain View, CA). To prevent ligand degradation, the cells were kept at 4 "C during analysis.
125Z-Ligands-Cells were plated as above, rinsed, and the appropriate concentration of radioactive ligand in DMEM was added. After incubation at 37 "C for designated times, medium was removed and precipitated in ice-cold 10% (w/v) trichloroacetic acid and held at 4 "C from 0.5 to 24 h. After centrifugation at 4 "C (1500 X g, 15 min, TJ-6R, Beckman), the supernates and pellets were separated and the radioactivity in them, plus the whole medium, was determined as above. The cell monolayers were rinsed two times in DMEM + 0.1% BSA, two times in DMEM, and then scraped into 0.1 N NaOH. The acid-soluble and -insoluble radioactivities in the cell lysates plus two rinses from each well were determined as above. A 100-fold excess unlabeled ASOR was added to assess non-specific uptake of lZ5I-ASOR. Nonspecific uptake of '251-Gal-Lys was determined by incubation of the cells for 10 min in 25 mM GalNAc in DMEM + 0.1% BSA, prior to addition of the '251-Gal-Lys.

Binding Studies
Surface and total ASGP-R were detected using a modification of the digitonin permeabilization assay as described previously (Weigel et al., 1983). Cells were plated as described for the endocytosis studies. The medium was removed and replaced with 2 ml of DMEM. The cell monolayers were incubated at 37 "C for 1 h and then at 4 "C for 30 min. After one rinse in DMEM-0.1% BSA, the radioactive ligand in DMEM-0.1% BSA without or with 0.075% (w/v) digitonin was added. After incubation at 4 "C for 2.5-4 h, the radioactive medium was removed, the cell monolayer rinsed twice with DMEM k 0.1% BSA, twice with DMEM, then scraped into 0.1 N NaOH. The radioactivity in the cell lysates plus two rinses from each plate was determined as above. Nonspecific binding of '"I-ASOR was defined as binding in the presence of 100-fold excess of unlabeled ASOR. Nonspecific binding of '251-Gal-Lys was determined by incubating the cells for 10 min in 25 mM GalNAc in DMEM + 0.1% BSA with or without 0.075% (w/v) digitonin before addition of the '251-Gal-Lys.
Since soluble protein was lost by digitonin permeabilization (Weigel et al., 1983) and cells were lost during the rinsing procedure after digitonin treatment, we determined the number of cells remaining in digitonin-treated monolayers as follows. The cell number from untreated wells was determined by measuring total cell protein (Lowry et al., 1951) (357 pg of protein/l X lo6 HTC cells). The resulting cell number was multiplied by the fraction of incorporated ['Hlthymidine remaining in digitonin-treated wells (40-60%). This latter value was obtained from cells in companion wells (to those used for 'Z61-ligand binding) that were labeled with 0.5 pCi of ['Hlthymidine over a 16-h period just prior to the binding studies. These wells were processed with or without digitonin as for the radioactive ligand, except that the '?-ligand was omitted. Aliquots of cell lysates were precipitated in 10% (w/v) trichloroacetic acid, the precipitates solubilized in Protosol according to the manufacturer's instructions and the radioactivity determined (LS 7000, Beckman). The amount of incorporated [3H]thymidine remaining in digitonin-treated cells was compared to that in untreated cells and expressed as a fraction, which was used to obtain accurate cell numbers for those treated with digitonin.

Autoradiography
Cells were plated on uncoated chamber slides and grown for 2 days. For the 4 "C binding studies, cells were incubated with 0.5 pg/ml ASOR or 2 pg/ml '251-Gal-Lys in the presence and absence of digitonin as described above. After rinsing, the cells were fixed for 30 min in 2% glutaraldehyde, 0.1 M sodium cacodylate, 2 mM CaC12, pH 7.4.
The slides were brought to room temperature during fixation, rinsed for 5 min in 0.1 M sodium cacodylate then dehydrated in a graded EtOH series and allowed to air-dry prior to emulsion coating.
Endocytosis experiments were carried out at 37 "C. Cells were rinsed with DMEM, incubated for 4 h with one of the '251-ligands, rinsed four times with DMEM at 4 "C, then incubated for 15 min at 4 "C with 2 ml of GKN & 2 mM EGTA, pH 7.4. Nonspecific binding was assessed using 25 mM GalNAc or excess ASOR as described above. The cells were fixed, dehydrated, and dried.
Processed slides were coated with K5 emulsion, exposed for 3-8 days and developed as described earlier (Zeitlin and Hubbard, 1982). They were mounted, examined on a Zeiss photomicroscope, and photographed as described above.

RESULTS
Desialylated plasma glycoproteins have been used routinely in studies of ASGP-R-mediated endocytosis. ASOR was reported by Ashwell and colleagues to exhibit one of the highest affinities for this receptor among the many naturally-occurring plasma glycoproteins tested (Pricer and Ashwell, 1971) and consequently has been the ligand used most frequently to study this receptor. McPhaul and Berg (1986) used the uptake of F-ASOR as a criterion for expression of a functional ASGP-R and found that transfection of both RHL 1 and RHL 2 cDNAs into HTC cells was required to obtain such expression. In their study, doubly transfected cells were incubated with F-ASOR at 37 "C for 12 h and then analyzed by FACS. Cells exhibiting the brightest fluorescence (top 8%) were collected, grown up for 2 weeks, resorted, and the top 8% again selected. Upon obtaining this sorted population from McPhaul, we examined them by light microscopic immunofluorescence using affinity-purified antibody to RHL 1 (not depleted of RHL 2/3 cross-reactive determinants) and found a variable level of staining (data not shown), indicating that the quantity of receptor from cell to cell was quite different. To obtain a homogeneous population, we cloned the doubly transfected cells.
Clones were selected by their ability to bind '261-ASOR at 4 "C, and one with high ligand-binding activity, BC6, was analyzed further. Upon passage in culture, the level of "'1-ASOR bound to BC6 cells decreased (data not shown). When the BC6 cells, passage 13, were analyzed in the FACS for their ability to accumulate F-ASOR at 37 "C, a biphasic distribution of fluorescence intensity was observed (Fig. lA). One population accumulated significant amounts of F-ASOR in a ligand-specific manner (Fig. 1B) and was designated ASORpositive. One population did not take up the ligand, since its fluorescence was coincident with cells that were incubated with unlabeled competitor and F-ASOR. We designated this the ASOR-negative population (negatiue, Fig. 1C). As reported previously (McPhaul and Berg, 1986), we found that the parent HTC cell line did not appear to take up the ligand (data not shown). Cells from the top 45% and the bottom 45% were separated (Fig. U), and the two populations were characterized further by immunological procedures, binding and uptake studies, and light microscopic autoradiography. Due to the instability of these transfected populations, all studies reported here were performed using cells passaged from two to six times after FACS separation.

Identification of ASGP-receptor Subunits in Positive and
Negative Cells Immunofluorescence-Indirect immunofluorescence was performed on positive and negative cells using subunit-specific immunological reagents. As shown in Fig. 2, both positive (panel A) and negative (panel B ) cells reacted with RHL 1specific antibody. Upon examination of cells with RHL 2/3specific antibody, the positive cells showed reactivity (panel C); however, the negative cells gave a signal that was close to background (panel D). A small percentage, usually 2-5%, but sometimes up to lo%, of the negative cells (passages 2-6) reacted with anti-2/3D, giving a signal similar to that found on the positive cells (data not shown). These cells were judged to be a small spillover of positive cells within the negative population. Therefore, the ASOR-positive cells expressed both subunits, while the ASOR-negative cells expressed only RHL 1. When present, labeling was detected at both the cell surface and in a perinuclear location.
Immunoblot Analysis-To determine the levels of RHL 1 and RHL 2/3 in the two populations, we applied anti-.ASGP-R 6 and anti-RHL 2/3 antibodies to nitrocellulose transfers of purified rat liver ASGP-R and HTC cell extracts (Fig. 3). The anti-ASGP-R 6 antibody bound to each subunit from a purified ASGP-R preparation, but in a slightly different ratio than did Coomassie Blue R (Fig. 3A). Less RHL 2/3 was detected on immunoblots than was present on protein-stained gels, giving RHL 1 to 2/3 ratios of 4:l by immunoblot but 3:l by Coomassie blue. Thus, the level of RHL 2/3 that we detected by immunoblot may be an underestimate of the

FIG. 3. Identification of ASGP-R subunits by immunoanalysis.
A, ASGP-R purified from rat liver; Coomassie Blue staining (CB, -46 pg of protein) with rabbit anti-ASGP-R 6 serum followed by 1251-protein A. In lane CB, the stoichiometry of RHL 1 to RHL 2/ 3is 3:l. In lane ZB, the stoichiometry of RHL 1 to RHL 2/3 is 4:l. B, detergent extracts (4 X lo5 cells/lane) of alkaline-extracted BC6positive (Pos) and -negative (Neg) cell homogenates were precipitated in trichloroacetic acid and then solubilized in SDS as described (Hubbard et al., 1985). Nitrocellulose transfers were labeled with affinity-purified rabbit lz5I-anti ASGP-R 6. In the BC6-positive cell lane, the stoichiometry of RHL 1 to RHL 2/3 is 41. In all lanes using anti-ASGP-R 6, a -90-kDa (D) band was observed that is apparently a dimer of RHL 1 generated during sample preparation (22). C, detergent extracts from liver (4.2 X lo4 cells/lane), BC6-positive (Pos, 2 x IO5 cells/lane) and negative (Neg, 2 X IO5 cells/lane) cells were prepared as for B. Nitrocellulose transfers were labeled with anti-RHL 2/3 antibody and 1251-protein A. In addition to the RHL 2/3 subunits identified in the positive cells, two other bands of faster mobility than RHL 2 appeared. These bands (0) were routinely present in BC6 cell extracts and absent in liver extracts. They may represent unglycosylated RHL 2 or RHL 2 breakdown occurring either within the cell or during sample preparation. One band comigrates with the band directly above RHL 1 from the positive cell extract (see B ) , the other band co-migrates with RHL 1. The band labeled 0 is nonspecific as judged by its appearance at higher inputs of all BC6 cell homogenates. The slower mobility observed for RHL 3 in positive cells may be due to different glycosylation enzymes in these cells.  Table I. quantity actually present. When the positive cell homogenate was examined (Fig. 3B), RHL 1 and RHL 2 had comparable mobilities to those from a rat liver homogenate used as a standard (lane not shown). Low levels of RHL 3 were detected in the positive cells (see Fig. 3C for clarification). By immunoblot, the stoichiometry of RHL 1 and RHL 2/3 in the positive population was 4:1, which is comparable to that in both the purified receptor and homogenate from rat liver.
When the negative cell homogenate was examined, a sig-  nificant amount of the RHL 1 subunit was detected (Fig. 3B). By densitometry, the level of RHL 1 in negative cells was 33% of that in the positive cells. There was no detectable RHL 2 or 3, even when four times more cell protein was analyzed. The high level of RHL 2 and RHL 3 detected by immunolocalization in 2-5% of the negative cells was undetectable at the level of cell homogenate used for immunoblotting.
The presence and absence of RHL 2/3 in the two HTC cell populations were confirmed by immunoblot analysis using anti-RHL 2/3 antibody (Fig. 3C). When RHL 2/3 subunit specific antibody was applied to nitrocellulose transfers of rat liver and HTC cell extracts, again there was no specific reactivity to proteins in extracts of the negative cells. However, RHL 2 and RHL 3 subunits were clearly present in both the liver and positive cell homogenates. The mobilities of RHL 2 and 3 in the positive cells corresponded to bands identified using anti-ASGP-R antibody 6 (Fig. 3B). With either antibody, RHL 3 from the positive cells appeared as a broad diffuse band with slower mobility than RHL 3 identified in liver.

Binding and Uptake Studies
Biochemical Analysis-We compared the 4 "C binding properties of the positive (RHL 1 + RHL 2/3) and negative (RHL 1 only) cells using ASOR and Gal-Lys, a high affinity synthetic ligand composed of a 40,000-dalton poly-L-lysine polymer to which an average of 100 galactose moieties were attached . This ligand binds to the ASGP-R with a 10-fold lower & than does ASOR on isolated hepatocytes* and presumably has more degrees of freedom for binding due to the high density of galactosyl residues and multiple conformations attributable to flexible aglycon and lysine side chains.
When the total ligand binding activities in positive and negative populations were examined with lZ5I-ligands at 4 "C in the presence of digitonin, we found that the positive cells specifically bound '251-Gal-Lys as well as lZ51-ASOR. A representative Scatchard plot is shown in Fig. 4, and Table I summarizes information from multiple experiments. In the positive population, we measured -390,000 specific binding sites/cell with lZ5I-ASOR (& = 24 f 8.4 X lo-' M) and 340,000 specific binding sites/cell with lZ5I-Gal-Lys (Kd = 3.6 f 1.8 X R. T. Townsend and Y. C. Lee, unpublished results. lo-' M). The binding sites for the two ligands appeared to be largely overlapping, since ASOR added before lZ5I-Gal-Lys effectively blocked (7040%) the latter's association with positive cells (data not shown). The negative population also bound lZ5I-Gal-Lys with a K d of 2.3 f 0.9 X lo-' M, which is in good agreement with that of the positive cells. However, we detected only 51,000 specific lZ5I-Gal-Lys-binding sites/ cell or -15% of the binding sites found in the positive cells. When the negative cells were examined using lZ5I-ASOR, <2% of the binding sites found in the positive cells were detected, even at lZ5I-ASOR concentrations as high as 400 nM.
The distribution of binding sites between the cell surface and intracellular sites was next determined for '251-ASOR in the positive cells and for '251-Gal-Lys in both cell populations. In all cases, -20% of the total ligand-binding sites were found at the surface of both positive and negative cells (i.e. sites detected in the absence of digitonin), with the remaining 80% found inside (data not shown). These distributions are comparable to those found for isolated and in situ hepatocytes (Weigel et al., 1983;Geuze et al., 1983).
We next examined the uptake and degradation at 37 "C of the '251-ligands in both cell populations. 41,000 lZ5I-ASOR molecules/cell were processed by the negative cells after 21 h of exposure to the ligand, which represented only 0.8% of the amount of 1251-ASOR processed by the positive population (data not shown). In contrast, we were able to detect significant levels of lZ5I-Gal-Lys uptake by the negative cells (Fig.  5). After only 90 min of exposure to the ligand, the negative cells had processed 35,000 molecules/cell, or -10 times more than the number of lZ5I-ASOR molecules endocytosed by the same cells per unit of time. When the uptake of '251-Gal-Lys by positive and negative cells was compared, the latter processed 16% that of the former, which is in agreement with the difference in 4 "C binding of '251-Gal-Lys between the two cell populations. Both the time course of endocytosis and number of molecules endocytosed by the positive cells were comparable for both ligands (Fig. 5). Although we detected little degradation of lZ5I-Gal-Lys in either population after only 90 min of continuous endocytosis, degradation had occurred by later time points (data not shown).
Morphological Analysis-To determine the proportion of cells in the negative population that expressed an RHL capable of binding and processing ligand, we quantitated lz5I-Gal-Lys and lZ5I-ASOR 4 "C binding and 37 "C uptake by light microscopic autoradiography. The results are presented in Fig. 6. A majority (96%) of the negative BC6 cells not only exhibited substantial levels of specific Gal-Lys-binding activity a t 4 "C ( Fig. 6B') but the majority (80%) was also capable of internalizing this ligand a t 37 "C ( Fig. 6C'). The negative cells neither bound nor internalized significant amounts of '251-ASOR under the conditions used.

DISCUSSION
In the present study, we have demonstrated that RHL 1, the major subunit of the rat ASGP-R, can function as a carbohydrate receptor in the absence of RHL 2/3. The key to this finding was use of Gal-Lys, a ligand that contains 5-10 times more and differently organized galactosyl residues than does ASOR. We showed that cells expressing only the RHL 1 polypeptide were capable of binding, internalizing, and degrading this ligand. In agreement with the results of Mc-Phaul and Berg (1986), we found that these cells did not recognize ASOR. Our results imply that: 1) the structure of the Gal-ligand dictates whether both subunits are required for binding; 2) the RHL 213 subunit plays an important role in the organization of native ASGP-R in the membrane; 3) the native rat ASGP-R is a functional hetero-oligomer; and 4) the biosynthesis, assembly, and participation in endocytosis of RHL 1 polypeptides do not depend on expression of RHL 213.
Ligand Requirements for High Affinity Binding-Drickamer and colleagues have demonstrated that both rat ASGP-R subunits have Gal-binding sites, as defined by the ability of each isolated polypeptide to bind specifically to Gal-Sepharose (Halberg et al., 1987). However, the binding of glycoproteins to the intact ASGP-R in a biological membrane is most likely different and clearly more complicated than the interaction of a solubilized receptor or individual subunits with immobilized Gal residues for the following reasons. First, glycoproteins bind the intact ASGP-R with nanomolar affinity while monosaccharides bind it with affinities in the millimolar range (Lee et al., 1983). Second, glycoproteins bear multiple oligosaccharide chains of the complex type, each located at a specific glycosylation site and each having two to four Gals (Schmid et al., 1979). ASOR, which is the most commonly used ASGP-R ligand, has five N-linked oligosaccharides, exhibits considerable microheterogeneity at each glycosylation site and therefore represents a mixture of glycoforms bearing an average of 10-25 galactosyl residues per polypeptide (Graham, 1972). Finally, the two non-identical ASGP-R polypeptides, RHL 1 and RHL 2/3, whose expressions and functions have been the focus of this study, most likely have different affinities for various Gal-structures, but these are currently unknown, as are their associations with one another in the membrane (see the Introduction).
Lee and colleagues have clarified understanding of this complicated receptor-ligand interaction by using ligands of known structure to define the requirements for high affinity binding (Kd = 0.5 > 5 nM) to the ASGP-R on the surface of intact hepatocytes (reviewed by Townsend, 1987). They have proposed that the minimal "functional" unit of the ASGP-R is a triad of Gal-binding sites (Hardy et al., 1985;Lee et al., 1984). Two of the three binding sites in the triad are spaced -15 A apart in the membrane while the third site may be mobile to a limited (30 A ) extent. They also have proposed that the triads are themselves clustered in the membrane, since they found additional binding sites for a small triantennary glycopeptide on hepatocytes saturated with ASOR (Hardy et al., 1985). Both this higher level of triad organization in the membrane, as well as the triad itself, are most likely disorganized when the ASGP-R is solubilized, because there is more of a reduction (-100-fold) in affinity for triantennary oligosaccharides than for glycoproteins such as ASOR (-10-fold), which have more Gals in more orientations and thus can still make the essential three point interactions (Lee et al., 1984a).
A Proposed Role for RHL 2/3 in ASGP-R Organization-In the present study we confirmed the results of McPhaul and Berg (1986) that HTC cells containing only RHL 1 do not bind ASOR. This suggests that RHL 2/3 is needed, but for what function? Since immunofluorescence showed that RHL 1 was at the surface of these cells, RHL 2/3 must not be serving a delivery function. However, since RHL 1 alone is able to bind Gal-Sepharose (Halberg et al., 1987), we reasoned that an active RHL 1 in these negative cells might recognize molecules such as Gal-Lys that have a higher density of Gal residues (100 Gal residues per 250 lysine residues) than does ASOR. Indeed, Gal-Lys exhibited nanomolar affinity when bound to cells bearing only the RHL 1 polypeptide. Such binding is understandable if RHL 1 subunits were clustered in such a way as to allow the essential three interactions by the multiple Gals of Gal-Lys. Thus, the key findings of loss of ASOR-binding activity yet persistence of Gal-Lys activity in cells bearing only RHL 1 strongly suggest that the absence of RHL 2/3 profoundly affects the organization of the proposed triads required for the binding of asialoglycoproteins but not the cell surface expression of clustered RHL 1 Galbinding sites.
Several additional characteristics of RHL 2/3 suggest that it may be playing an important organizing role in the structure and therefore function of the native ASGP-R. RHL 2/3 is present in lesser amounts than RHL 1 (the mass ratio of rat liver RHL 1 to RHL 2/3 is 3 or 4/1), yet it is preferentially cross-linked to a photoaffinity-labeling reagent based on a triantennary glycopeptide found in asialo-fetuin . RHL 2/3 loses Gal-binding activity more readily than does RHL 1 (Halberg et al., 1987). Since, from the present study, RHL 2/3 is apparently not necessary for transport of RHL 1 through the biosynthetic or endocytic pathways, it must be serving another function. We postulate that the minor subunit is either itself forming a homodimer in the membrane with its Gal-binding sites -15 A apart, or it organizes or participates with RHL 1 subunits to form such a dimer. In either case, there must be additional RHL 1 subunits associated to accommodate the 3 or 4/1 mass ratio of RHL 1 to RHL 2/3 that we have observed. Therefore, we suggest that the native ASGP-R is a hetero-oligomer.
RHL 1 Can Act as a Receptor-The two HTC populations we have studied were originally derived from co-transfection of cells with cDNAs for both ASGP-R subunits. Although the negative cells do not express detectable RHL 2/3 protein, the status of the RHL 2/3 cDNA is not presently known. With this caveat, our results show that RHL 1 is targeted through the biosynthetic pathway and reaches the cell surface in a form that can bind a specific type of ligand. Since RHL 1 was also capable of internalizing significant quantities of Gal-Lys, it alone seems to contain all the information necessary for participation in the endocytic pathway as well. Preliminary experiments suggest that recycling may occur, because more ligand can be processed than total functional receptors present. 3 We have not performed the kinetic experiments that would indicate whether the efficiency of RHL 1 targeting through the various intracellular pathways is comparable in the presence and absence of expression of the minor subunit.

L. T. Braiterman and A. L. Hubbard, unpublished observations.
When the amounts of RHL 1 protein in the two HTC cell populations were compared on immunoblots, the negative cells contained -30% that of the positives. However, comparison of Gal-Lys binding activities showed that the negatives bound only 15% of the lZ5I-Gal-Lys that the positives did. There are two possible explanations for this difference: either 50% of the RHL 1 polypeptides are non-functional, in which case turnover measurements may reveal two degradation rates, one of which reflects an unstable pool of RHL 1; or 2 RHL 1 subunits are required for '251-Gal-Lys binding in the negative but not the positive cells.
Finally, we observed that the negative and positive HTC cells exhibited poorer affinities for Gal-ligands than do isolated primary hepatocytes. Nonetheless, the relative difference in K d values between ASOR and Gal-Lys (-6-10-fold) in the positive (RHL 1 + 2/3) cells was similar to that found in isolated hepatocytes. These results imply that there is yet another level of receptor organization that may involve additional cellular components (cytoskeletal elements? other membrane proteins?) not provided by the HTC cells. Clearly, much more work is needed to fully understand this complex receptor.