THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1989 by The American Society for Biochemistry and Molecular Biology, Ine.

Vol. 264, NO. 19, Issue of July 5, pp. 1145a11467 1989 Printed in d.S.A.

The Cation-dependent Mannose 6-Phosphate Receptor STRUCTURAL REQUIREMENTS FOR MANNOSE 6-PHOSPHATE BINDING AND OLIGOMERIZATION*

(Received for publication, February 28, 1989)

Nancy M. DahmsS and Stuart Kornfeldg From the Departments of Medicine and Biological Chemistry, Washington university School of Medicine, St. Louis, Missouri 63110

The structural requirements for oligomerization and the generation of a functional mannose 6-phosphate (Man-6-P) binding site of the cation-dependent man- nose 6-phosphate receptor (CD-MPR) were analyzed. Chemical cross-linking studies on affinity-purified CD-MPR and on solubilized membranes containing the receptor indicate that the CD-MPR exists as a homo- dimer. To determine whether dimer formation is nec- essary for the generation of a Man-6-P binding site, a cDNA coding for a truncated receptor consisting of only the signal sequence and the extracytoplasmic do- main was constructed and expressed in Xenopus laevis oocytes. The expressed protein was completely soluble, monomeric in structure, and capable of binding phos- phomannosyl residues. Like the dimeric native recep- tor, the truncated receptor can release its ligand at low pH. Ligand blot analysis using bovine testes 6-galac- tosidase showed that the monomeric form of the CD- MPR from bovine liver and testes is capable of binding Man-6-P. These results indicate that the extracyto- plasmic domain of the receptor contains all the infor- mation necessary for ligand binding as well as for acid- dependent ligand dissociation and that oligomerization is not required for the formation of a functional Man- 6-P binding site.

Several different mutant CD-MPRs were generated and expressed in X. laevis oocytes to determine what region of the receptor is involved in oligomerization. Chemical cross-linking analyses of these mutant pro- teins indicate that the transmembrane domain is im- portant for establishing the quaternary structure of the CD-MPR.

The transport of newly synthesized lysosomal enzymes to lysosomes is known to be accomplished by the phosphoman- nosyl recognition system in numerous cell types (for reviews see von Figura and Hasilik, 1986; Kornfeld, 1987). Lysosomal enzymes bearing phosphomannosyl residues bind specifically

* This work was supported in part by United States Public Service Grant CA 08759 and a Monsanto Company/Washington University Biomedical Research Grant. 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.

$ Supported by American Cancer Society Fellowship PF-2873. Present address: Department of Biochemistry, Medical College of Wisconsin, 8701 Watertown Plank Rd., Milwaukee, WI 53226,

To whom correspondence should be addressed: Division of He- matology-Oncology, Washington University School of Medicine, 660 S. Euclid, St. Louis, MO 63110.

to mannose 6-phosphate receptors (MPR)‘ in the Golgi ap- paratus and the resulting receptor-ligand complex is trans- ported to an acidic prelysosomal compartment where the low pH mediates the dissociation of the complex. The lysosomal enzymes become incorporated into lysosomes and the recep- tors can recycle back to the Golgi to repeat the process.

Two distinct MPRs have been identified and recently their primary structures have been elucidated through sequence analyses of their cDNA clones (Lobel et al., 1987, 1988; Morgan et al., 1987; MacDonald et al., 1988; Oshima et al., 1988; Pohlmann et al., 1987; Dahms et al., 1987). The cation- independent (CI) MPR, which does not require divalent cat- ions to bind its ligand, was the first receptor to be character- ized (Rome et al., 1979; Sahagian et al., 1981). The deduced 2,499-amino acid precursor of the bovine CI-MPR has a calculated molecular weight of 275,000 and consists of a signal sequence, an extracytoplasmic domain that is composed of 15 homologous repeating units of -147 residues in length, a single putative transmembrane region, and a carboxyl-termi- nal cytoplasmic domain (Lobel et al., 1988). This receptor is identical to the insulin-like growth factor I1 receptor (Morgan et al., 1987). Ligand binding studies have revealed that the receptor binds 2 mol of Man-6-P and 1 mol of IGF-II/ polypeptide chain (Tong et al., 1988,1989). The second recep- tor, the cation-dependent (CD) MPR, has optimal binding to ligand in the presence of divalent cations (Hoflack and Korn- feld, 1985a, 1985b). The bovine CD-MPR cDNA encodes a 279-amino acid precursor with a calculated molecular weight of 31,000 that consists of a signal sequence, a 159-residue extracytoplasmic domain, a single putative transmembrane region, and a carboxyl-terminal cytoplasmic domain. Analysis of the sequence reveals that the extracytoplasmic domain of the CD-MPR has significant sequence similarities with each of the 15 repeating domains of the CI-MPR (Dahms et al., 1987; Lobel et al., 1988). Ligand binding studies on the CD- MPR indicate that this receptor binds only 1 mol of Man-6- P/polypeptide chain rather than 2 as is found with the CI- MPR (Tong and Kornfeld, 1989). Although the CD-MPR has a narrower pH range than the CI-MPR for optimal Man-6-P binding (pH 6.0-6.3 uersus 6.0-7.4, respectively), neither receptor can interact with ligand below pH 5 (Hoflack and Kornfeld, 1985a; Hoflack et al., 1987; Tong et al., 1989; Tong and Kornfeld, 1989). This latter property is shared with numerous receptors that function in the delivery of ligands to acidified compartments (Goldstein et al., 1985). Unlike the

The abbreviations used are: MPR, mannose 6-phosphate receptor; CD-MPR, cation-dependent M P R CI-MPR, cation-independent MPR Man-g-P, mannose 6-phosphate; SDS, sodium dodecyl sulfate; DSS, disuccinimidyl suberate; DTSSP, 3,3’-dithiobis(sulfosuccin- imidylpropionate); BSA, bovine serum albumin; Hepes, 4-(2-hydrox- yethy1)-1-pipierazineethanesulfonic acid; MES, 24N-rnorpholino) ethanesulfonic acid LDL, low density lipoprotein.

11458

Dimerization and Ligand Binding of the CD-MPR 11459

CI-MPR, the CD-MPR does not bind insulin-like growth factor II (Tong et al., 1988; Kiess et al., 1988).

Recent cross-linking experiments indicate that the CD- MPR exists as a dimer in U937 monocyte membranes (Stein et al., 1987b). However, studies performed on the receptor in solution yielded trimeric and tetrameric cross-linked species (Stein et al., 1987a; Li et al., 1988). We have utilized chemical cross-linking reagents on both affinity-purified and mem- brane-associated CD-MPR from various sources to try to verify the receptor's oligomeric state. The observation that the receptor exists as an oligomer leads to two models for the formation of the ligand binding site. In one model, two or more separate CD-MPR polypeptide chains would have to interact in a noncovalent fashion to form an active binding site. In the second model, each receptor monomer could fold into an independent ligand binding site. To distinguish be- tween these two alternatives and to determine what regions of the receptor are required for its oligomerization, a series of cDNAs encoding mutant proteins were constructed and ex- pressed in Xenopus laevis oocytes. The resultant proteins were analyzed with respect to their ligand binding capabilities as well as their subunit conformation. The results presented here indicate that the extracytoplasmic domain of the CD- MPR in a monomeric conformation is sufficient for Man-6- P binding and suggest that the transmembrane region is involved in the oligomerization of the receptor.

EXPERIMENTAL PROCEDURES

Materials-Restriction endonucleases, RNA polymerases, and T4 DNA ligase were purchased from Bethesda Research Laboratories or New England Biolabs. The deoxy- and dideoxynucleoside triphos- phates, the cap analogue G(5')ppp(5')G, and BSA (DNase- and RNase-free) were from Pharmacia LKB Biotechnology Inc. RNasin was from Promega Biotec (Madison, WI). The Bluescript KS and SK plasmid vectors were obtained from Stratagene. The chemical cross- linkers disuccinimidyl suberate (DSS), 3,3'-dithi0bis(sulfosuc- cinimidylpropionate) (DTSSP), and dimethyl pimelimidate were ob- tained from Pierce Chemical Co. Protein A-Sepharose as well as other biochemicals were from Sigma. Radionuclides were purchased from Amersham Corp. ([35S]methionine, 1000 Ci/mmol; [~x-~'Ss]dATp, 650 Ci/mmol; [5-3H]UTP, 13.7 Ci/mmol; [1251]sodi~m iodine, carrier- free). Adult female X . laeuis frogs were from NASCO (Fort Atkinson, WI). 'rota1 0-galactosidase was purified from bovine testes as de- scribed (Distler and Jourdian, 1973) and was kindly provided by Walter Gregory of this laboratory. High uptake P-galactosidase was prepared by applying total @-galactosidase to a column of immobilized bovine CI-MPR and eluting with buffer containing 10 mM Man-6-P. The eluted @-galactosidase was radiolabeled using Enzymobeads (Bio- Rad) and NaIZ5I, and re-purified on a bovine CI-MPR affinity column. Murine L cells transfected with bovine CD-MPR cDNA were a kind gift of K. Johnson?

Plasmid Constructions-The Bluescript KS or SK plasmid vector containing the wild-type bovine CD-MPR cDNA was constructed as described previously (Dahms et al., 1987). This plasmid contains the Mad-Ps t I fragment (nucleotides 32-1071) of the CD-MPR to which a poly(A) tail from a human cathepsin D clone was attached. To produce mutant receptors, cassette mutagenesis (Wells et al., 1985) was employed to insert a stop codon, and in some cases to change the amino acid sequence of the transmembrane region, in the CD-MPR cDNA. Synthetic oligonucleotides were synthesized on an Applied Biosystems Inc. model 380A oligonucleotide synthesizer and were purified in acrylamide/urea gels as described in Maniatis et al. (1982). The oligonucleotides were annealed by heating 15 min at 68 "C and were then allowed to cool to room temperature in a buffer containing 1.5 mM ATP, 0.1 M Tris-HC1 (pH 7.6), 1.5 mM spermidine, 15 mM MgCI,, 22.5 mM dithiothreitol, and 0.3 mg/ml BSA. For the STOP155 construct, the annealed synthetic oligonucleotides form a 101-bas.e pair MaeI-BglII fragment which encodes nucleotides 492-592 of the CD-MPR except that the three proline (residue 155) codons present in the wild-type sequence (CCA, nucleotides 584-586) have been changed to a stop codon (TGA). A 480-base pair XbaI-Mael fragment

K. Johnson and S. Kornfeld, submitted for publication.

containing nucleotides 32-491 of the wild-type receptor sequence plus 20 base pairs of Bluescript KS polylinker (XbaI-SmaI), the annealed mutagenic oligonucleotides forming the Mad-BglII fragment, and a 3650-base pair BglII-Xbal fragment that consists of the Bluescript KS plasmid vector, nucleotides 593-1071 of the wild-type CD-MPR sequence, and a poly(A) tail from a human cathepsin D clone were incubated with T4 DNA ligase at 14 'C. Constructs STOP'", TM'", TMCIMPR, and TMLDLR were generated as above except that the annealed synthetic oligonucleotides for each construct form a BglII- PstI fragment which was then ligated to a 3750-base pair fragment consisting of the Bluescript KS plasmid vector, nucleotides 32-588 of the wild-type CD-MPR sequence, and a poly(A) tail from a human cathepsin D clone. The BglII-PstI fragment for each of these con- structs encodes nucleotides 589-682 of the CD-MPR cDNA with the following changes. STOP'% nucleotides 599-601 (CTT) and nucleo- tides 602-604 (AGC) have been changed to stop codons (TGA); TMIa7: nucleotides 680-682 (CTG) have been changed to a stop codon (TGA); TMCIMPR: nucleotides 680-682 (CTG) have been changed to a stop codon (TGA) and nucleotides 599-673 have been replaced with nu- cleotides 6940-7008 of the bovine CI-MPR cDNA (Lobe1 et al., 1988); TMLDLR: nucleotides 680-682 (CTG) have been changed to a stop codon (TGA) and nucleotides 599-673 have been replaced with nu- cleotides 2365-2430 from the human LDL receptor (Yamamoto et al., 1984). Escherichia coli JM109 cells were transformed with the various cDNA constructs and the plasmids were purified. For each construct, the region in the plasmid encompassing the mutagenic oligonucleo- tides was subjected to double-stranded DNA sequence analysis (Chen and Seeburg, 1985) to confirm the predicted sequence.

In Vitro Transcription-RNA was synthesized from Bluescript SK or KS plasmids containing the wild-type CD-MPR sequence or the various mutant cDNAs as described previously (Dahms et al., 1987).

Microinjection and Translation of mRNA in X . hevis Oocytes- Stage VI oocytes (Dumont, 1972) were removed from anesthetized X. lueuis females, microinjected with either the wild-type or mutant CD- MPR RNA, and incubated in a modified Barth's solution (Gurdon, 1976) containing [35S]methionine as described previously (Dahms et al., 1987). X . iueuis oocytes contain a relatively small pool size of methionine of -30 pmol/oocyte (Colman, 1984). When high specific activity [35S]methionine is used, the incorporation of radioactivity into protein which is precipitable with trichloroacetic acid remains linear up to -20 h and thus incubations longer than -20 h constitute a chase (data not shown). After incubation, the media were recovered and the oocytes were homogenized by sonication in ice-cold solubi- lization buffer (50 J/oocyte) containing 100 mM Tris-HC1 (pH 8.0), 1% Triton X-100, 0.1% sodium deoxycholate, 10 mM Na2EDTA, 1% aprotinin, 1 mM phenylmethylsulfonyl fluoride, antipain (4 pg/ml), benzamidine (20 pg/ml), and 2 pg/ml each of leupeptin, chymostatin, and pepstatin. After 1 h of incubation on ice, the suspension was centrifuged for 10 min at 12,000 X g at 4 "C, and the detergent- solubilized oocyte lysate was stored at -70 "C. In some experiments, the oocytes were homogenized in the presence of either 100 mM iodoacetamide or 10 mM N-ethylmaleimide and incubated at room temperature for 30 min. After centrifugation, the lysates were dialyzed to remove the alkylating reagent. In other experiments, the media were recovered and the oocytes sonicated in buffer (100 pl/oocyte) containing 20 mM Tris-HCl (pH 7.6), 150 mM NaCl, 10 mM Mg(OAc)z, and 0.2% aprotinin in order to prepare membranes. The suspension was centrifuged at 100,000 X g for 30 min. The membranes were washed twice in the above buffer by sonication and recentrifu- gation. The supernatants were combined and the membranes were solubilized with detergent as described above.

Column Chromatography-To assay for ligand binding ability, detergent-solubilized oocyte lysates were passed over either a 0.5-1111 phosphomannan-Sepharose column or a 0.5-ml pentamannosyl phos- phate-agarose column at 4 "C in buffer containing 50 mM imidazole (pH 6.51, 150 mM NaCI, 10 mM MnClz, 0.05% Triton X-100, and 0.01 mg/ml BSA (Hoflack and Kornfeld, 1985b). The run-through and the material that was eluted from the column in the above buffer con- taining either 2 mM glucose 6-phosphate or 5 mM Man-6-P was immunoprecipitated as described below. To determine subunit stoi- chiometry, the Man-6-P eluates from the phosphomannan-Sepharose columns were chromatographed on a Sephadex G-75 fine column (1 X 56 cm). The column was run at 4 'C in a buffer containing 50 mM imidazole (pH 6.71, 150 mM NaCl, 5 mM /3-glycerophosphate, 5 mM Na2EDTA, and 0.05% Triton X-100. The flow rate was 3.3 ml/h, and 0.4-ml fractions were collected. The excluded volume, Vo, was deter- mined by blue dextran 2000 (Mr 2,000,000; Pharmacia LKB Biotech- nology Inc.) and the totally included volume, V,, was determined by

11460 Dimerization and Ligand Binding of the CD-MPR [6-'H]galactose (20 Ci/mmol, Amersham Corp.). Protein standards were obtained from Sigma.

Immunoprecipitation and Gel Electrophoresis-Samples in solubi- lization buffer or in solubilization buffer lacking 0.1% sodium deoxy- cholate were incubated with rabbit polyclonal antisera specific for the bovine CD-MPR and subjected to gel electrophoresis on slab gels using the buffer systems of Laemmli (1970) as described previously (Dahms et al., 1987). Radiolabeled materials were detected on the slab gels by fluorography of gels impregnated with EN'HANCE (Du Pont-New England Nuclear). Quantitation of the labeled material was obtained by removal of the radioactive slices, rehydration in water, and then solubilization in NCS tissue solubilizer (Amersham Corp.) as recommended by the manufacturer. Scinti Verse I (Fisher) was used as the scintillation mixture.

Cross-linking Reactions-The Man-6-P eluates of samples purified on pentamannosyl phosphate-agarose columns were dialyzed against 0.1 M Hepes (pH 7.0-8.0). 0.1 M NaCl (buffer A) and then incubated with the various cross-linkers in buffer A containing Triton X-I00 (0.05-1%) as indicated in the figure legends. The cross-linking reac- tion was stopped by the addition of glycine to 100 mM followed by a 10-min incubation. Soybean trypsin inhibitor (4 pg) was added as carrier to each sample prior to precipitation with 10% trichloroacetic acid. The precipitated material was resuspended in sample buffer and analyzed on SDS-polyacrylamide gels. Oocyte membranes were sol- ubilized in buffer A containing 1% Triton X-100 and incubated with cross-linking reagents as described in the figure legends. After ter- mination of the reaction with 100 mM glycine, the samples were immunoprecipitated with CD-MPR-specific antisera and analyzed by SDS-polyacrylamide gel electrophoresis. Transfected L cell mem- branes were prepared in buffer A containing 10 mM N-ethylmaleim- ide, washed twice in buffer A, and solubilized in buffer A containing 0.5% Triton X-100. After centrifugation at 15,000 X g for 10 min, the resulting supernatants were incubated with cross-linking agents as indicated in the figure legends and precipitated with trichloroacetic acid prior to analysis on SDS-polyacrylamide gels. The above cross- linking reactions were performed a t 25 "C. For cross-linking of sam- ples immobilized on pentamannosyl phosphate-agarose, the following procedure was performed 1) Samples were passed over pentaman- nosy1 phosphate-agarose columns (75 pl bed volume) as described above and extensively washed with buffer A (pH 7.0) containing 0.05% Triton X-100.2) The resin was transferred to 500-p1 tubes and incubated with 1 mM DTSSP in the above buffer for 30 min a t 4 "C. After 30 min, additional DTSSP was added to a final concentration of 2 mM and the samples were incubated for 30 min a t 25 "C. Glycine was added to 100 mM to stop the reaction. 3) The samples were eluted from the resin with 10 mM Man-6-P, immunoprecipitated with CD- MPR-specific antisera, and analyzed by SDS-polyacrylamide gel elec- trophoresis.

Immunoblot and Ligand Blot Analysis-Proteins were transferred from SDS-polyacrylamide gels to nitrocellulose (Schleicher & Schuell) according to the method of Burnette (1981). For immunoblot analysis, the nitrocellulose membrane was treated as described by Messner et al. (1989). For ligand blot analyses, affinity-purified bovine liver and bovine testes CD-MPR (1 pg) were loaded onto an SDS- polyacrylamide gel (consisting of an 8% resolving gel and a 5% stacking gel) in a buffer containing 65 mM Tris (pH 6.8), 10% glycerol, 0.5% SDS, and 0.1% bromphenol blue without prior heating. After electrophoresis, the proteins were transferred to nitrocellulose and the membrane was treated as described below following a procedure of Westlund and K~rnfe ld .~ The membrane was rinsed twice (2 min each) in 0.05% Tween 20. The membrane was rinsed for 5 min in buffer B (50 mM imidazole (pH 6.5), 150 mM NaCI, 0.01% Triton X- 100, 10 mM MnC12, 1 mg/ml BSA) and then incubated in buffer B for 2 h. The membrane was incubated in buffer B containing 150,000 cpm/ml '2sII-labeled bovine testes @-galactosidase for 6 h in the presence or absence of 5 mM mannose 6-phosphate. The membrane was washed six times (5 min each) in buffer B. All membrane incubations and washes were performed at room temperature. Auto- radiograms were obtained by exposing the nitrocellulose of Kodak XAR-5 film for 6-12 h at -70 "C with Cronex Lightning Plus inten- sifying screens (Du Pont).

RESULTS

Oligomeric Structure of the CD-MPR-It has been observed that a significant amount of the CD-MPR synthesized in the

B. Westlund and S. Kornfeld, unpublished data.

murine BW5147 lymphoma cell line migrates as an apparent dimer in nonreducing SDS-polyacrylamide gels (Gasa and Kornfeld, 1987). This phenomenon was also seen with the bovine CD-MPR expressed in X. laeuis oocytes. When [35S] methionine-labeled oocyte membranes were solubilized with detergent, immunoprecipitated with CD-MPR-specific anti- sera, and analyzed by SDS-polyacrylamide gel electrophoresis in the absence of reductant, a monomeric species of 38 kDa was detected and -50% of the receptor migrated as an appar- ent dimer of 74 kDa (Fig. lA, lane 1). However, in the presence of 2-mercaptoethanol nearly all of the receptors migrate as a monomer in SDS-polyacrylamide gels (data not shown). When solubilization was performed in the presence of iodo- acetamide, the amount of the CD-MPR dimer was drastically reduced (compare lanes 1 and 2, Fig. 1A). Alkylation with N- ethylmaleimide gave similar results (Fig. lB, lane 1). These results indicate that intermolecular disulfide bonds form be- tween two CD-MPR polypeptide chains after the oocyte mem- branes are solubilized and these artifactual covalent dimers can be prevented by alkylating cysteine residues during the solubilization procedure. The observation that a significant portion of the receptor can be isolated in the form of these disulfide-linked dimers suggests that the CD-MPR exists largely as noncovalently associated dimers.

To confirm that the bovine CD-MPR exists predominantly as a dimer in oocyte membranes, chemical cross-linking ex- periments were performed. [35S]Methionine-labeled oocyte membranes were solubilized in the presence of N-ethylmal- eimide and then incubated with the cleavable cross-linking agent DTSSP for varying times. The receptor was immuno- precipitated and analyzed on an SDS-polyacrylamide gel un- der nonreducing conditions (Fig. lB, lanes 1-4). The presence of 1 mM DTSSP for 5 min resulted in the formation of a species that migrated with an apparent molecular weight of 74,000 which is consistent with this species being a homodi- mer of the receptor (Fig. lB, lane 2). With increasing incu- bation times, the amount of the dimer form increases but no trimers or tetramers are seen. In addition, the electrophoretic mobility of both the monomeric and dimeric forms of the CD- MPR progressively increases with increasing incubation times. This increased mobility probably occurs due to the formation of intramolecular cross-links which could create

A R

66- - 92 - 103-

0 ?' 67-

.g 42-

22 - 28-

0 45- I" 0

5 31- "-. ""

I 2 1 2 3 4 5 6 7 8

FIG. 1. Cross-linking of CD-MPR from X. Zaeuis oocytes in the presence of alkylating reagents. A, membranes from [%SI methionine-labeled X . laeuis oocytes that had been microinjected with wild-type CD-MPR RNA were homogenized with 1% Triton X- 100 and 0.1% deoxycholate in the presence or absence of 100 mM iodoacetamide (see "Experimental Procedures"), immunoprecipitated with CD-MPR-specific antisera, and analyzed on a 12% SDS-poly- acrylamide gel under nonreducing conditions. B, membranes from [35S]methionine-labeled X. laeuis oocytes that had been microinjected with wild-type CD-MPR RNA were solubilized with 1% Triton X- 100 and 0.1% deoxycholate in the presence of 10 mM N-ethylmaleim- ide and incubated with 1 mM DTSSP for the indicated times (see "Experimental Procedures"). The samples were immunoprecipitated with CD-MPR-specific antisera and analyzed on 10% SDS-polyacryl- amide gels under nonreducing (lanes 1-4) or reducing (lanes 5-8) conditions. The migration of molecular weight standards is indicated.

Dimerization and Ligand Binding of the CD-MPR 11461

tightly packed globular structures. A similar increase in elec- trophoretic mobility is seen in chemical cross-linking studies of both the chicken hepatic lectin (Loeb and Drickamer, 1987) and the LDL receptor (van Driel et al., 1987). When the cross- linked samples were analyzed on SDS-polyacrylamide gels in the presence of 2-mercaptoethanol, all of the dimeric form of the CD-MPR collapsed to the monomeric species (Fig. 123, lanes 5-8). These results indicate that DTSSP is forming covalent intermolecular cross-links between two CD-MPR polypeptide chains and thus suggests that the CD-MPR exists as a homodimer in X. laevk oocytes. Similar results were obtained with the noncleavable cross-linking agents, DSS and dimethyl pimelimidate (see below, and data not shown).

To determine the oligomeric form of the receptor in mam- malian cells, bovine CD-MPR cDNA-transfected murine L cells were utilized. These cells express high levels of the bovine CD-MPR and thus facilitate detection of oligomeric species of the receptor. Membranes from these transfected cells were prepared in the presence of N-ethylmaleimide, solubilized with detergent, and treated with varying concentrations of either the cleavable cross-linking agent DTSSP or with the noncleavable cross-linking agent DSS. Both agents caused a high proportion of the receptor to be cross-linked to a dimeric species of 74 kDa (Fig. 2, lanes 1-5). The presence of 2- mercaptoethanol in the SDS sample buffer caused the major- ity of the DTSSP-cross-linked species to be converted to their monomeric forms while no effect was seen on the noncleavable DSS-cross-linked species (Fig. 2, lanes 6-10). Even though this transfected cell line expresses a large amount (contains 0.44 ng of CD-MPR/pg protein which is -100-fold higher than endogenous levels)' of the bovine CD-MPR, no trimer, tetramer, or higher oligomeric species were detected. When intact, surface-iodinated transfected mouse L cells were in- cubated with DTSSP, again only dimers of the receptor were detected (data not shown). These results indicate that the CD-MPR exists as a dimer in this transfected mouse L cell line.

Expression of a Truncated CD-MPR-Since the CD-MPR appears to exist as a dimer, we wanted to determine if the quaternary structure of the receptor was necessary for the formation of a ligand binding site. To do this, we constructed a truncated form of the receptor by introducing a stop codon just prior to the putative transmembrane region (the STOP'5s mutant, Fig. 3). This mutant receptor was then expressed and its oligomeric and ligand binding properties analyzed. Fig. 4

Nonreduced

20s -

0 103- b 67-

4 2 - 1

20 - 1 2 3 4 s 6 7 8 910

FIG. 2. Cross-linking of CD-MPR from transfected mouse L cells. Membranes from bovine CD-MPR-transfected mouse L cells were prepared in the presence of 10 mM N-ethylmaleimide, solubilized with 0.5% Triton X-100, and incubated for 30 min at 25 'C with either DSS (pH 8.0) or DTSSP (pH 7.0) at the concentrations of cross-linking agent indicated. The samples were precipitated with trichloroacetic acid, separated on an 8% SDS-polyacrylamide gel in the absence (lanes 1-5) or presence (lanes 6-10) of 2-mercaptoethanol, and prepared for immunoblot analysis (see "Experimental Proce- dures").

Tronsmembrone domain \ vr DSSLACSPBISE LSVGSILLVTLASLVAVIIIGPLY ORLWGAIo(B0P~LAP

STOP155 DSSLACS

STOP160 DSSLACSPRISB

l'HlE7 DSSLACSPRISE LSVCSILLVRASLVAVIIICPLT (ur

l'HlW DSSLACSPRISB L S V C S I L L v r U s L V A V I I ~ ~ S DDCS

T@"PR DSSLACSPBISB AVCAVLSLLLVALTACLLTLLLT OR

TIImM DSSLACSPRISB ALSIVLPIVLLWLCLCVPLLY OR

FIG. 3. Amino acid sequence of the wild-type and mutant CD-MPRs. The schematic diagram of the wild-type CD-MPR lists the number of the amino acid at the start of each domain. This numbering differs slightly from that described in Dahms et al. (1987) in that the signal sequence here consists of 28 residues instead of 22. This assignment is based on SIGSEQ2 (Folz and Gordon, 1987), a computer program based on the method of Von Heijne (1986), and on amino-terminal amino acid sequence obtained from the mature human CD-MPR (Stein et al., 1987a). The amino acid sequence encompassing the mutations in each CD-MPR construct is listed. The STOP constructs contain a stop codon at the amino acid position indicated by the superscript and encode proteins consisting of only the signal sequence and extracytoplasmic domains. The TM con- structs contain a putative transmembrane domain in addition to the signal sequence and extracytoplasmic domains, and the superscripts indicate either the amino acid position where a stop codon has been incorporated or the name of the protein whose transmembrane region has been substituted for the CD-MPRs transmembrane region. The underlined amino acids indicate those residues in the mutant con- structs which differ from the wild-type ( W T ) CD-MPR sequence.

shows that X. lueuis oocytes injected with mRNA for this truncated receptor synthesize three major species of approx- imately 34, 31, and 28 kDa that are specifically immunopre- cipitated with CD-MPR-specific antisera. Partial endoglyco- sidase H digestions of the STOPlSS mutant demonstrated that the multiple species of this truncated receptor are due to varying levels of N-linked glycosylation (data not shown). Quantitation of these results shows that -20% of the STOP'55 mutant molecules contain two (28-kDa species), -50% con- tain three (31-kDa species), and -30% contain four (34-kDa species) N-linked carbohydrate chains. This differs from the wild-type CD-MPR which, when expressed in oocytes, is efficiently glycosylated with -90% of the molecules contain- ing four carbohydrate chains (data not shown). Fig. 4 also shows that the STOP'= mutant is gradually secreted into the media (lanes 8 and 10). To demonstrate directly that this truncated receptor, which lacks the cytoplasmic and trans- membrane regions, is soluble, oocyte extracts prepared in the absence of detergent were separated into soluble and mem- brane fractions. When the distribution of the STOPlSS mutant was analyzed, virtually all of it was found in the soluble fraction (data not shown).

Ligand Binding Capabilities of the STOP'5s Mutant CD- MPR-To determine if this truncated CD-MPR is capable of binding ligand, detergent-solubilized oocytes that had been

11462 Dimerization and Ligand Binding of the C D - M P R

0

0 47 c X

18

1 2 3 4 5 6 7 8 9 1 0 FIG. 4. Expression of the truncated receptor STOP"" in X.

laevis oocytes. X. laeuis oocytes were injected with -50 nl of RNA (200 ng/pl) that was transcribed from a plasmid containing the truncated receptor DNA. After injection, the oocytes were incubated in buffer containing 1 mCi/ml [%]methionine. At the various times indicated, the media were recovered and the oocytes were solubilized in buffer containing 1% Triton X-100 and 0.1% deoxycholate. The cells (C) and media (M) were immunoprecipitated with CD-MPR- specific antisera and analyzed on a 12% SDS-polyacrylamide gel under reducing conditions. The noninjected lanes represent oocytes that were radiolabeled for 48 h but were not injected with truncated receptor STOP'5s mRNA.

6t 9: -

g 4:

i 3'

22

92 - 66-

FIG. 5. Ligand affinity chromatography of the STOPI5" construct. A, the solubilized oocyte lysate from a 72-h [35S]methio- nine labeling of oocytes injected with the truncated receptor mRNA was passed over a 0.5-ml pentamannosyl phosphate-agarose column. The column was eluted first with 2 mM glucose 6-phosphate and then with 5 mM Man-6-P. The run-through ( R T ) , glucose 6-phosphate ( G ) , and Man-6-P (M6P) eluates were immunoprecitated with CD- MRP-specific antisera and analyzed on a 12% SDS-polyacrylamide gel under reducing conditions. B, phosphomannan-Sepharose col- umns (0.5 ml) were run on 24-h [35S]methionine-labeled samples as described above except that the columns were eluted (6 column volumes each) first with 2 mM glucose 6-phosphate ( G ) ; a buffer containing 50 mM MES (pH 4.6), 150 mM NaCl, 0.05% Triton X- 100,lO mM MnC12, and 0.01 mg/ml BSA; and 5 mM Man-6-P (M6P). The eluates were immunoprecipitated and then analyzed on a 12% SDS-polyacrylamide gel under reducing conditions. Only 61% of the STOP'" mutant protein binds to phosphomannan-Sepharose while virtually all of this molecule binds to a pentamannosyl phosphate affinity column. This differs from the wild-type receptor in which greater than 90% of the molecules bind to both types of ligand (data not shown).

injected with the STOP155 mutant mRNA were passed over a pentamannosyl phosphate-agarose column. After extensive washing, the column was first eluted with glucose 6-phosphate and then with Man-6-P. Virtually all (-99%) of the STOP'" mutant protein was recovered in the Man-6-P eluate (Fig. 5A).

Since the CD-MPR does not bind lysosomal enzymes a t low pH (Hoflack and Kornfeld, 1985a; Hoflack et al., 1987),

we wanted to determine if this truncated CD-MPR would also exhibit ligand dissociation at an acidic pH. To do this, deter- gent-solubilized oocytes were prepared as before and passed over a phosphomannan-Sepharose column which was eluted sequentially with glucose 6-phosphate, a buffer at pH 4.6, and finally with Man-6-P. Fig. 5B illustrates that the truncated CD-MPR is eluted completely from the phosphomannan- Sepharose column with low pH. Taken together, these results demonstrate that the extracytoplasmic domain of the CD- MPR contains all the information necessary for ligand bind- ing as well as for acid-dependent ligand dissociation.

Oligomeric Structure of the STOP'ss Mutant CD-MPR- Chemical cross-linking studies were undertaken to analyze the quaternary structure of the STOP'Ss mutant. Wild-type and STOP1S5 CD-MPRs were expressed in X . laeuk oocytes, labeled with [35S]methionine, alkylated, and purified on affin- ity columns of pentamannosyl phosphate. The receptors were then incubated with two concentrations of the cleavable (DTSSP) and noncleavable (DSS) cross-linking agents for varying times, and the reaction products were analyzed on SDS-polyacrylamide gels (Fig. 6). At all incubation times and concentrations of cross-linking agent, a significant amount of the wild-type CD-MPR was cross-linked to a dimeric species of 74 kDa (Fig. 6A, lanes 3-16). In contrast, no oligomeric species of the STOP1s5 construct was seen after incubation with either reagent (Fig. 6B, lanes 3-16). Both cross-linking agents were capable of forming covalent bonds with this truncated receptor since increasing times of incubation caused an increased electrophoretic mobility of the monomeric form of the STOP'55 construct (Fig. 6B, lanes 3-16). In addition, when DTSSP-cross-linked species of either the wild-type or mutant receptors are compared to their noncross-linked coun- terparts on SDS-polyacrylamide gels in the presence of re- ductant, the DTSSP-cross-linked species migrate with a slower mobility (Fig. 6C). This decreased electrophoretic mo- bility suggests that the DTSSP reagent is covalently attached to the polypeptide chain of both the wild-type and mutant receptors.

Fig. 6 also shows that the STOP155 mutant, in contrast to the wild-type CD-MPR, does not form an oligomeric species when isolated from the oocyte in the absence of alkylating agents (compare Fig. 6A, lanes 1 and 2 with Fig. 6B, lanes 1 and 2). This suggests that the transmembrane and/or cyto- plasmic regions of the wild-type receptor are responsible for the formation of intermolecular disulfide bonds during the isolation procedure. The bovine CD-MPR has two cysteine residues at positions 214 and 218 in its cytoplasmic domain, and thus either one or both of these residues are implicated in this process.

One possibility for the failure of the STOP'55 mutant to be cross-linked with either DTSSP or DSS is that these agents may not be of the proper length or in the proper orientation to span the two polypeptide chains in a dimer. Therefore, a different cross-linking agent, dimethyl pimelimidate, was tried, but it also failed to cause the STOP1s5 mutant to oligomerize (data not shown). In addition, cross-linking ex- periments were performed at different pH values and temper- atures, on nonalkylated receptors, and in the presence of Man-6-P. Regardless of the experimental conditions oligo- meric species of the STOPlSs mutant were not detected, whereas the wild-type CD-MPR was cross-linked under all conditions tested (data not shown). Although the presence of Man-6-P during the cross-linking reaction had no effect, another experiment was performed to determine if ligand induces the STOP'55 mutant to oligomerize. Following alkyl- ation, the wild-type and truncated CD-MPRs were allowed to

Dimerization and Ligand Binding of the CD-MPR 11463

l+l Wild-type STOP'55

z 43- ,

29 -

4 3 -

'I -

22 - 11 345678910111213141516

C

200

97 68

0

z - 43

29

1 2 3 4 FIG. 6. Cross-linking of affinity-purified wild-type and

STOP'66 CD-MPRs. A , ["S]methionine-labeled X . laeuis oocytes that had been microinjected with wild-type CD-MPR RNA were homogenized with 1% Triton X-100 and 0.1% deoxycholate in the absence (lane 1) or presence (lanes 2-16) of 10 mM N-ethylmaleimide, purified on a pentamannosyl phosphate-agarose column, and incu- bated with either DSS (pH 8.0) or DTSSP (pH 7.4) for the times and cross-linking agent concentrations indicated. The samples were pre- cipitated with trichloroacetic acid and analyzed on a 10% SDS- polyacrylamide gel under nonreducing conditions (see "Experimental Procedures"). E , same as in A; only [3SS]methionine-labeled X . laeuis oocytes that had been microinjected with the STOP'" mutant CD- MPR RNA were used. C, duplicate samples from A and B were prepared but were analyzed on an SDS-polyacrylamide gel in the presence of 2-mercaptoethanol: lane I, identical sample preparation as A , lane 2; lane 2, identical sample preparation as A , lane 14; lane 3, identical sample preparation as B, lane 2; lane 4, identical sample preparation as B, lane 14.

bind to separate pentamannosyl phosphate-agarose columns. The immobilized receptors were incubated with the cleavable cross-linking agent DTSSP, eluted with Man-6-P, and then analyzed on an SDS-polyacrylamide gel under nonreducing conditions (Fig. 7). The noncross-linked run-through frac- tions, in which no oligomeric species are seen (Fig. 7, lanes 1 and 4), are shown for comparison. Nearly 55% of the wild- type CD-MPR was cross-linked to a dimeric species of 74 kDa when immobilized on the affinity resin, whereas none of the truncated receptor was cross-linked to dimers of the predicted molecular weight (Fig. 7, lanes 3 and 6). However, in contrast to the monomeric form of the STOP''s mutant which runs predominantly as a triplet of 25-, 28-, and 31-kDa species, 8% of the truncated receptor was detected as a triplet of 37-, 40-,

< 31-

22 - f t 1 2 3 4 5 6

-

FIG. 7. Cross-linking of ligand immobilized wild-type and STOP'66 CD-MPRs. [%]Methionine-labeled X . faeuis oocytes that had been microinjected with either wild-type or the STOP'" mutant CD-MPR RNA were homogenized with 1% Triton X-100 and 0.1% deoxycholate in the presence of 10 mM N-ethylmaleimide and chro- matographed on pentamannosyl phosphate-agarose columns. The resin with the bound wild-type or bound STOP'" mutant CD-MPR was then incubated with DTSSP as described under "Experimental Procedures." The cross-linked material was then eluted with 10 mM mannose 6-phosphate and the run-through (RT), glucose 6-phosphate ( G ) , and cross-linked Man-6-P (M6P) eluates were immunoprecipi- tated with CD-MPR-specific antisera and analyzed on a 10% SDS- polyacrylamide gel under nonreducing conditions. Man-6-P eluted greater than 95% of each receptor from the affinity resin as deter- mined by boiling the eluted resin with SDS sample buffer containing 2-mercaptoethanol (data not shown).

and 44-kDa species (Fig. 7, lane 6) . The addition of 2-mercap- toethanol caused these higher molecular weight forms to collapse to monomers on SDS-polyacrylamide gels (data not shown). Therefore, the higher molecular weight forms of the truncated receptor may be dimers that run anomalously on SDS-polyacrylamide gels or may be receptor that is cross- linked to a different protein. The poor level of apparent dimerization obtained with DTSSP suggests that ligand does not cause a significant amount of the ST0Pls5 mutant to oligomerize. The low amount of cross-linking that is seen may be due to the fact that the STOP'" mutant had been concen- trated on the resin (-4 fmol of radiolabeled receptors were applied to the affinity columns), resulting in a small percent- age of the molecules being in close physical proximity for DTSSP to covalently cross-link two polypeptide chains to- gether.

To analyze the quaternary structure of the STOP"' mutant receptor in a different manner, affinity-purified truncated receptor was chromatographed on a Sephadex G-75 gel filtra- tion column (Fig. 8). The column profile shows a minor peak of radioactivity in the void volume of the column followed by a major peak of radioactivity in the position of a protein with a molecular size of 31 kDa. Immunoprecipitation of the radio- active peaks revealed that greater than 95% of the STOP'" mutant was in peak B (Fig. 8, inset). Analysis of the total radioactive proteins in the two peaks showed that peak B contains only truncated receptor while peak A contains a small amount of the truncated receptor plus the endogenous CD- and CI-MPRs of the X. laeuis oocyte (data not shown). Gel filtration of the STOP''s mutant on a Bio-Gel P-100 column gave an apparent size of 28 kDa (data not shown). These values are probably overestimates of the true molecular weight since glycoproteins often appear larger than they ac- tually are on gel filtration. The protein portion of the trun- cated receptor is predicted to have a molecular weight of -17,000. Since the truncated CD-MPR synthesized in oocytes has predominantly two, three, or four N-linked oligosaccha- rides, the predicted molecular weight of the mutant receptor ranges from 21,000 to 25,000. Thus, these data are consistent with the STOP''' mutant being a monomer.

11464 Dimerization and Ligand Binding of the CD-MPR 200

1 : 1 150 -

V 2 1 0 0 -

50-

O’ -6.2 6 i2 014 0.6 018 ‘ AV

FIG. 8. Gel filtration of the STOP’66 construct on Sephadex G-75. The mannose 6-phosphate eluates from the pentamannosyl phosphate-agarose columns (see Fig. 5) were chromatographed on a Sephadex (2-75 fine column (1 X 56 cm) (see “Experimental Proce- dures”). The peaks of radioactivity (peaks A and B) were pooled, immunoprecipitated using CD-MPR-specific antisera, and subjected to SDS-polyacrylamide gel electrophoresis on a 12% slab gel under reducing conditions (inset). The arrows indicate the elution positions of the standards: I, ovalbumin (M, 43,000); 2, carbonic anhydrase (M, 29,000); and 3, cytochrome c (M, 12,400). The distribution coefficient, Kav, represents (V. - Vo/( V, - VO) where V. is the elution volume, Vo is the excluded volume as determined by blue dextran 2000 (M, 2,000,000), and V, is the totally included volume as determined by [6- JH)galactose.

202

103 b 67-) ; X L

A2

2a

1 2

FIG. 9. Ligand blot analysis of CD-MPR from bovine liver and testes. A mixture of affinity-purified bovine liver and testes CD- MPR (1 pg) was subjected to electrophoresis in 8% slab gels contain- ing 0.1% SDS under nonreducing conditions. The proteins were transferred to nitrocellulose and incubated with 1251-labeled 8-galac- tosidase (&Gal) in the absence (lane I) or presence (lane 2) of 5 mM Man-6-P (M6P) as described under “Experimental Procedures.” The nitrocellulose was then subjected to autoradiography.

In order to obtain further evidence that the monomeric form of the CD-MPR is capable of binding phosphomannosyl residues, affinity-purified CD-MPR from bovine liver and testes was electrophoresed on an SDS-polyacrylamide gel, transferred to nitrocellulose, and incubated with 1251-labeled 8-galactosidase in the absence (Fig. 9, lane 1) or in the presence (Fig. 9, lane 2) of 5 mM Man-6-P. The autoradi- ograph of the nitrocellulose membrane reveals that the mon- omeric form of the receptor (-35 kDa) is capable of binding 8-galactosidase (Fig. 9, lane I ) , and this binding is specifically inhibited by mannose 6-phosphate (Fig. 9, lane 2). It is pos- sible, although unlikely, that the receptor could form dimers when immobilized to the nitrocellulose membrane. This ex- periment also shows that the apparent dimeric (-70 kDa), trimeric (-110 kDa), and tetrameric (-170 kDa) forms of the

receptor are capable of binding phosphomannosyl residues. The apparent trimeric and tetrameric forms of the receptor were detected only upon long exposures of the nitrocellulose membrane to film and thus represent only a small fraction of the total oligomeric species.

Structural Requirements for CD-MPR Dimerization-We next analyzed the portions of the CD-MPR required for dimer formation. A series of mutant receptors was generated by cassette mutagenesis, expressed in X . laevis oocytes, and their ligand binding capabilities and oligomeric structures analyzed. Fig. 3 lists the various constructs that were made and their amino acid differences with the CD-MPR. All of the con- structs bound efficiently to a pentamannosyl phosphate-aga- rose column (data not shown). The oligomeric structure of each of these constructs was assayed by their ability to be chemically cross-linked. In each experiment, the wild-type CD-MPR was included as a positive control. Construct TM187 was designed to test whether or not the cytoplasmic region of the receptor is necessary for the formation of oligomers. This construct contains the transmembrane region of the CD-MPR plus 2 residues of the cytoplasmic domain (Fig. 3). Fig. 10A shows that this mutant is efficiently cross-linked to an ap- parent dimeric species of 50 kDa by DSS (lanes 5-8). These results indicate that the cytoplasmic domain of the CD-MPR is not necessary for dimer formation. Construct TM’” was obtained as a by-product of the construction of TMlE7 and was detected during the sequencing of various isolates. It lacks the entire cytoplasmic region of the CD-MPR and, in addition, contains amino acid changes in the carboxyl-termi- nal portion of the transmembrane region (Fig. 3). Construct TMIW is also efficiently cross-linked to an apparent dimeric species by DSS (Fig. 10A, lanes 9-12). We next determined whether a specific amino acid sequence in the transmembrane region was required for self-association of the CD-MPR to occur. The TMCIMPR and TMLDLR constructs were made in which the transmembrane region of the CD-MPR was substi- tuted with the transmembrane region of either the bovine CI- MPR or the human LDL receptor, respectively (Fig. 3). Fig. 10B shows that both the TMCIMPR (lanes 5-8) and TMLDLR (lanes 9-12) constructs are efficiently cross-linked to an ap- parent dimeric species by DSS. All four TM constructs were found to associate exclusively with the membrane fraction during isolation and all were efficiently glycosylated with greater than 80% of the molecules containing four carbohy- drate chains (data not shown). To rule out the possibility that the five amino acids prior to the putative transmembrane region are responsible for the covalent cross-linking to a dimer by DSS, the STOPIGO construct was made (Fig. 3). Fig. 1OC shows that this mutant is not cross-linked by DSS (lanes 1- 4). Further studies indicated that the STOPIGO mutant has properties very similar to that of the STOP’” mutant: 1) it is secreted by X. laeuk oocytes; 2) it is completely soluble in the absence of detergent; 3) it contains varying levels of N-linked glycosylation with the predominant species containing two, three, and four carbohydrate chains; and 4) it migrates as a monomer of 31 kDa on a Sephadex G-75 gel filtration column (data not shown).

These results indicate that the transmembrane region, but not the cytoplasmic domain, of the CD-MPR is necessary for its oligomerization.

DISCUSSION

In this report, we have analyzed the quaternary structure of the CD-MPR and have determined by chemical cross- linking studies the portion of the molecule that is involved in its oligomerization. We have also begun to define the regions

Dimerization and Ligand Binding of the CD-MPR 11465

A

205 - - 103- '9 67-

5 4 2 -

28 -

X

67- h i 4 x 42-

28 -

I 2 3 4 5 6 7 8 9 1011 12 13 14 15

9 7 - 66 -

0 43- 0

x

'- 31-

22-

1 2 3 4 5 6 7 0

FIG. 10. Cross-linking of affinity-purified wild-type, TM'", TM'", TMCIMPR, TMLDLR, and STOP'" mutant CD- MPRs. A, ["Slmethionine-labeled X. laeuis oocytes that had been microinjected with either wild-type (lanes 1 4 ) , TM'" ( l a n e s 5 4 , or TMIW ( l anes 9-12) mutant CD-MPR RNA were homogenized with 1% Triton-X-100 and 0.1% deoxycholate in the presence of 10 mM N-ethylmaleimide, purified on pentamannosyl phosphate-agarose columns, and incubated with 2 mM DSS (pH 8.0) for the times indicated. The samples were precipitated with trichloroacetic acid and analyzed on a 10% SDS-polyacrylamide gel under nonreducing conditions. B, same as in A; only [RsSS]methionine-labeled X. laeuis oocytes that had been microinjected with either wild-type (lanes 1- 5), TMCIMPR ( l anes 6-10), or TMLDLR (lanes 11-15) mutant CD-MPR RNA were used. C, [35S]methionine-labeled X. l a e u i oocytes that had been microinjected with either (lanes 1-4) or wild-type (lanes 5-8) CD-MPR RNA were solubilized with 1% Triton X-100 and 0.1% deoxycholate in the absence of alkylating agents, purified on penta- mannosyl phosphate-agarose columns, and incubated for 30 min a t 25 "C with the indicated concentrations of DSS a t pH 8.0. The samples were precipitated with trichloroacetic acid and analyzed on a 10% SDS-polyacrylamide gel under nonreducing conditions.

of the CD-MPR that are required for the formation of a Man- 6-P binding site.

Our studies with homobifunctional cross-linking agents indicate that the CD-MPR exists as a noncovalent homodi- mer. These results are consistent with the chemical cross- linking studies of Stein et al. (1987b) who reported that the receptor is a dimer in U937 monocyte membranes. However, with the various cross-linking agents we never obtained 100% conversion of the monomeric form of the receptor to the dimeric form. Therefore it is possible that the CD-MPR could exist in both forms and the particular quaternary structure

that the CD-MPR assumes could, perhaps, depend on its location in the cell. Covalent dimers of the receptor were also obtained in the absence of chemical cross-linking when the receptor was isolated from oocyte membranes in the absence of alkylating reagents. However, these dimeric species were lost when alkylating agents are present during homogeniza- tion. This observation indicates that this covalent cross- linking occurs via intermolecular disulfide bonds following solubilization of the oocyte. A similar finding has been re- ported for the LDL receptor (van Driel et al., 1987). The various CD-MPR mutants have localized these intermolecular cross-links to the cytoplasmic domain. Since there are 2 cysteine residues in the cytoplasmic portion of the CD-MPR, it is possible that disulfide bonds could form between many polypeptide chains during the isolation procedure and result in higher order cross-linked species. However, the highest order oligomer that is detected is a dimer, thus providing additional evidence that the CD-MPR exists as a noncovalent dimer in oocyte membranes. Oligomeric species of a higher order than a dimer were seen only when affinity-purified CD- MPR from bovine liver and testes that had been isolated in the absence of alkylating agents were analyzed on SDS- polyacrylamide gels in the absence of reductant (see Fig. 9). It is possible that the low levels of trimers and tetramers that were detected may be due to nonspecific aggregation of high concentrations (0.2 mg/ml) of receptor. This possibility may explain the results seen by Stein et al. (1987a) in which significant amounts of trimers and tetramers were detected upon chemically cross-linking the CD-MPR in solution at a high concentration (1 mg/ml).

The evidence that the CD-MPR's polypeptide chains self- associate to form homodimers and not heterodimers is as follows: 1) The molecular weight of either the disulfide- bonded or chemical cross-linked dimers is that expected for a homodimer; and 2) reduction of either the disulfide-bonded or the DTSSP-cross-linked species with 2-mercaptoethanol results in the appearance of only one species which migrates as the CD-MPR monomer on SDS-polyacrylamide gels. No other proteins were detected upon dissolution of this complex. Although it is possible that lipids may interfere with cross- linking of the CD-MPR in membranes to limit our detection to only dimers (Loeb and Drickamer, 1987), taken together our results suggest that the CD-MPR self-associates to form homodimers.

The production of the truncated STOP'55 receptor, which consists of only the signal sequence and extracytoplasmic domain, allowed us to address the question of whether oligo- merization is required for the generation of a functional ligand binding site. Analysis of this truncated receptor by SDS- polyacrylamide gel electrophoresis, chemical cross-linking, and gel filtration demonstrated that it exists entirely in the monomeric form. Affinity chromatography on columns bear- ing phosphomannosyl residues revealed that the STOP'5S mutant binds Man-6-P and releases its ligand at an acidic pH. Further support for this interpretation comes from the ligand blot analysis of affinity-purified CD-MPR from bovine liver and testes which shows that the monomeric form of the wild-type receptor can bind ligand. These results demonstrate that the extracytoplasmic domain of the CD-MPR is capable of folding into the proper conformation to allow binding of phosphomannosyl residues without the need for oligomeriza- tion.

These findings on the nature of the Man-6-P binding site of the CD-MPR may provide insight on how the 275-kDa CI- MPR binds its ligand. The extracytoplasmic domain of the CI-MPR is composed of 15 homologous repeating domains,

11466 Dimerization and Ligand Binding of the CD-MPR

each of which has significant homology to the extracyto- plasmic domain of the CD-MPR (Lobel et al., 1988). Since the extracytoplasmic domain of the CD-MPR can fold into a ligand binding monomer, it is possible that each of the 15 repeating domains of the CI-MPR folds independently to form functional ligand binding units. However, ligand binding ex- periments indicate that the CI-MPR contains only two high affinity sites for Man-6-P and one high affinity site for insulin-like growth factor I1 per polypeptide chain (Tong et ai., 1988,1989). Thus the CI-MPR, with its numerous domains that may each be capable of folding independently to form a binding site, may bind yet other ligands in addition to Man- 6-P and insulin-like growth factor 11.

The observation that the truncated STOP'55 mutant does not form oligomers indicated that the transmembrane region and/or the cytoplasmic domain is responsible for the associ- ation of the receptor into oligomers. To determine which region(s) of the CD-MPR is required for dimerization, a series of mutant receptors was analyzed with respect to their ligand binding capabilities and subunit conformation. The TMls7 mutant, which contains the extracytoplasmic and transmem- brane domains of the CD-MPR, was shown to exist as a dimer. This result demonstrates that the cytoplasmic domain of the receptor is not required for oligomerization. This differs from the LDL receptor in which chemical cross-linking of intact cells bearing truncated receptors indicate that the cytoplasmic domain is responsible for its self-association into oligomers (van Driel et al., 1987). Preoteolytic cleavage of the chicken hepatic lectin followed by incubation with chemical cross-linkers indicates that the self-association of this lectin is mediated by a region in or near the transmembrane region (Loeb and Drickamer, 1987). Since this region encompasses the transmembrane domain and portions of the extracyto- plasmic and cytoplasmic domains, it is not clear what domain of the chicken hepatic lectin is actually mediating its self- association. Interestingly, a specific transmembrane sequence is not required for self-association of the CD-MPR since truncated mutants that contain the transmembrane region of the CI-MPR (TMClMPR) or LDL receptor (TMLDLR) in place of the CD-MPR transmembrane domain form dimers when incubated with chemical cross-linkers. This observation sug- gests that the specific self-association of the CD-MPR to homodimers is mediated by the extracytoplasmic domain. One possible explanation for the role of a transmembrane region in oligomerization is that association of the receptor with the membrane alters the conformation of the extracytoplasmic domain such that residues in the extracytoplasmic domain are now in the correct orientation for specific noncovalent interactions to occur. Alternatively, hydrophobic interactions between the transmembrane regions of adjacent subunits may serve to stabilize the oligomer. This stabilizing role of the transmembrane domain in oligomeric structures has been proposed for the hemagglutinin trimers of influenza virus (Doms et al., 1986; Doyle et al., 1986).

An interesting finding was that the STOP'55 and STOP'" mutant receptors synthesized in X . laevis oocytes were less completely glycosylated compared to either the wild-type or mutant membrane-associated receptors. The wild-type and TM mutant receptors have four out of the five N-linked glycosylation sites efficiently glycosylated, whereas the solu- ble CD-MPRs have two, three, or four N-linked carbohydrate chains, with three chains being slightly more prevalent. Thus, the extracytoplasmic domain is glycosylated less efficiently as a soluble protein than when it is anchored to the membrane. However, the different levels of glycosylation on the soluble mutant receptors have no effect on its ability to bind phos-

phomannosyl residues since the receptors containing two, three, or four N-linked oligosaccharides all appear to bind equally well to the affinity column (see Fig. 5 and data not shown). Conversion of the human insulin receptor from a membrane-associated protein to a secretory protein also re- sults in a change in the glycosylation pattern of its a-subunit (Whittaker and Okamoto, 1988).

The production of a truncated CD-MPR containing only the extracytoplasmic domain has allowed us to being to define the structural requirements for a functional Man-6-P binding site. Additional deletion and/or site-specific mutants of the CD-MPR should aid in further defining the elements involved in Man-6-P binding.

Acknowledgments-We thank Walter Gregory for generously pro- viding CD-MPR-specific antisera and phosphomannosyl affinity res- ins and Greg Grant and Mark Frazier of the Protein Chemistry Facility for the production of synthetic oligonucleotides.

REFERENCES

Burnette, W. N. (1981) Anal. Biochem. 112,195-203 Chen, E. Y., and Seeburg, P. H. (1985) DNA (N. Y. ) 4,165-170 Colman, A. (1984) in Transcription and Translation: A Practical

Approach (Hames, B. D., and Higgins, S. J., eds) pp. 271-302, IRL Press, Oxford

Dahms, N. M., Lobel, P., Breitmeyer, J., Chirgwin, J. M., and Korn- feld, S. (1987) Cell 50, 181-192

Distler, J . J., and Jourdian, G. W. (1973) J. Biol. Chem. 2 4 8 , 6772- 6780

Doms, R. W., Gething, M.-J., Henneberry, J., White, J., and Helenius, A. (1986) J. Virol. 6 7 , 603-613

Doyle, C., Sambrook, J., and Gething, M.-J. (1986) J. Cell Biol. 103 ,

Dumont, J. N. (1972) J. Morphol. 136,153-180 Folz, R. J., and Gordon, J. I. (1987) Biochem. Biophys. Res. Commun.

146,870-877 Gasa, S., and Kornfeld, S. (1987) Arch. Biochem. Biophys. 2 5 7 , 170-

176 Goldstein, J. L., Brown, M. S., Anderson, R. G. W., Russell, D. W.,

and Schneider, W. J. (1985) Annu. Reu. Cell Biol. 1, 1-39 Gurdon, J. B. (1976) J. Embryol. Exp. Morphol. 3 6 , 523-554 Hoflack, B., and Kornfeld, S. (1985a) Proc. Natl. Acad. Sci. U. S. A.

Hoflack, B., and Kornfeld, S. (1985b) J. Biol. Chern. 260, 12008- 12014

Hoflack, B., Fujimoto, K., and Kornfeld, S. (1987) J. Biol. Chem. 262,123-129

Kiess, W., Blickenstaff, G. D., Sklar, M. M., Thomas, C. L., Nissley, S. P., and Sahagian, G. G. (1988) J. Biol. Chem. 263,9339-9344

Kornfeld, S. (1987) FASEB J. 1 , 462-468 Laemmli, U. K. (1970) Nature 227, 680-685 Li, M., Distler, J., and Jourdian, G. W. (1988) FASEB J . 2 , 1551

Lobel, P., Dahms, N. M., Breitmeyer, J., Chirgwin, J. M., and Korn-

Lobel, P., Dahms, N. M., and Kornfeld, S. (1988) J. Biol. Chem. 2 6 3 ,

Loeb, J. A,, and Drickamer, K. (1987) J. Biol. Chem. 262,3022-3029 MacDonald, R. G., Pfeffer, S. R., Coussen, L., Tepper, M. A., Brock-

lebank, C. M., Mole, 3. E., Anderson, J. K., Chen, E., Czech, M. P., and Ullrich, A. (1988) Science 2 3 9 , 1134-1137

Maniatis, T., Fritsch, E. F., and Sambrook, J. (1982) Molecular Cloning: A Laboratory Manual, Cold Spring Harbor Laboratory, Cold Spring Harbor, NY

Messner, D., Griffiths, G., and Kornfeld, S. (1989) J . Cell Biol., in

Morgan, D. O., Edman, J. C., Standring, D. N., Fried, V. A,, Smith, press

M. C., Roth, R. A., and Rutter, W. J. (1987) Nature 329, 301-307 Oshima, A., Nolan, C. M., Kyle, J. W., Grubb, J. H., and Sly, W. S.

(1988) J. Biol. Chem. 2 6 3 , 2553-2562 Pohlmann, R., Nagel, G., Schmidt, B., Stein, M., Lorkowski, G.,

Krentler, C., Cully, J., Meyer, H. E., Grzeschik, K.-H., Mersmann, G., Hasilik, A,, and von Figura, K. (1987) Proc. Natl. Acad. Sci.

Rome, L. H., Weissmann, B., and Neufeld, E. F. (1979) Proc. Natl.

1193-1203

82,4428-4432

(abstr.)

feld, S. (1987) Proc. Natl. Acad. Sci. U. S. A. 8 4 , 2233-2237

2563-2570

U. S. A. 84,5575-5579

Dimerization and Ligand Binding of the CD-MPR 11467

Acad. Sci. U. S. A. 76,2331-2334 van Driel, I. R., Davis, C. G., Goldstein, J. L., and Brown, M. S. Sahagian, G. G., Distler, J., and Jourdian, G. W. (1981) Proc. Natl. (1987) J , fjiol, them, 262, 16127-16234

Acad. Sci. U. S. A. 78, 4289-4293 Stein, M., Meyer, H. E., Hasilik, A., and von Figura, K. (1987a) Biol. 193

von Figura, K., and Hasilik, A. (1986) Annu. Reo. Biochem. 55, 167-

Chem. Hoppe-Seyler 368, 927-936 Stein, M., Braulke, T., Krentler, C., Hasilik, A,, and von Figura, K. von Heijne, G. (1986) Nucleic Acid Res. 14, 4683-4690

(1987b) Biol. Chem. Hoppe-Seyler 3 6 8 , 937-947 Wells, J. A., Vasser, M., and Powers, D. B. (1985) Gene (Amst.) 34, Tong, P. Y., and Kornfeld, S. (1989) J. Biol. Chem. 264, 7970-7975 315-323 Tong, P. Y., Tollefsen, S. E., and Kornfeld, S. (1988) J. Biol. Chem. Whittaker, J., and Okamoto, A. (1988) J . Bid. Chem. 263, 3063-

Tong, P. Y., Gregory, W., and Kornfeld, S. (1989) J. Bwl. Chem. Yamamoto, T., Damis, C. G., Brown, M. S., Schneider, W. J., Casey, 263,2585-2588 3066

264,7962-7969 M. L., Goldstein, J. L., and Russell, D. W. (1984) Cell 39, 27-38