Quantification of prolactin messenger ribonucleic acid, pituitary content and plasma levels of prolactin, and detection of immunoreactive isoforms of prolactin in pituitaries from turkey embryos during ontogeny.

The content of prolactin mRNA as well as total prolactin content and type of isoforms of prolactin were measured in single pituitary glands from turkey embryos and poults. Levels of mRNA and pituitary content of prolactin remained low until 5 days before hatching, while plasma concentrations remained low until 2 days before hatching. Levels of prolactin mRNA then increased until the day of hatch, stayed stable during the 3 first days of age, and significantly increased until 2 wk of age. Similar changes were observed in pituitary content and plasma levels of prolactin. Two immunoreactive bands of apparent molecular masses of 24 and 27 kDa, corresponding to the nonglycosylated and glycosylated form of prolactin, respectively, were visualized on Western blots. In pituitary glands from embryos at 22 days of incubation, 31.5% of the protein was glycosylated, whereas in embryos at 27 days of incubation and poults at 1 and 7 days of age, 48.6%, 48.0%, and 56. 0% of prolactin was glycosylated, respectively. The results indicate that the increases in the synthesis and the release of prolactin occur mainly around and after the time of hatching in the turkey embryo. Higher percentages of glycosylated isoforms were associated with increasing levels of total prolactin in the pituitary gland. Thus, the synthesis of prolactin and its post-translational modifications may be important factors involved in the physiologic changes occurring around the time of hatching.     

Glycosylation abolishes an in vitro insulin-like action of prolactin

Ovine prolactin stimulated 14C-CO2 production from labeled glucose in adipose tissue of hypophysectomized rats in vitro, an insulin-like activity. Glycosylation of the hormone by attachment of a carbohydrate unit at asparagine31 abolished this in vitro insulin-like action. However, neither nonglycosylated nor glycosylated prolactin exhibited in vivo insulin-like action, as they did not lower serum glucose or non-esterified fatty acids in fasted hypophysectomized rats. Hindrance of receptor binding by the carbohydrate unit may account for the absence of in vitro insulin-like activity in glycosylated prolactin, but the dichotomy between in vivo and in vitro insulin-like actions for prolactin remains obscure.     

Prolactin isoforms secreted by human prolactinomas.

Prolactin (hPRL) secreted by human prolactinoma cells in culture was purified by gel filtration, lectin affinity chromatography and gel electrophoresis in order to identify the different isoforms of the hormone and to test their respective immunoreactivities and bioactivities. The nonglycosylated hPRL (NG-hPRL), unbound to lectins, was the major form and was a species (NG1-hPRL), of 23,000 (M(r)) apparent molecular weight. The lectin-bound glycosylated hPRL (G-hPRL) consisted of three forms, G1-, G2- and G3-hPRL, of identical molecular weights (25,000 M(r)). Endoglycosidase treatment indicated that these three forms differed by the heterogeneity of their carbohydrate chains. These G-PRLs proved to be 68% less immunoreactive and 50% less bioactive than NG-hPRL. It is concluded from these data that, in prolactinomas, the main variant of the hormone is the nonglycosylated form of PRL.     

Binding of N- and C-terminal anti-prion protein antibodies generates distinct phenotypes of cellular prion proteins (PrPC) obtained from human, sheep, cattle and mouse

Prion diseases are neurodegenerative disorders which cause Creutzfeldt-Jakob disease in humans, scrapie in sheep and bovine spongiform encephalopathy in cattle. The infectious agent is a protease resistant isoform (PrP(Sc)) of a host encoded prion protein (PrP(C)). PrP(Sc) proteins are characterized according to size and glycoform pattern. We analyzed the glycoform patterns of PrP(C) obtained from humans, sheep, cattle and mice to find interspecies variability for distinct differentiation among species. To obtain reliable results, the imaging technique was used for measurement of the staining band intensities and reproducible profiles were achieved by many repeated immunoblot analysis. With a set of antibodies, we discovered two distinct patterns which were not species-dependent. One pattern is characterized by high signal intensity for the di-glycosylated isoform using antibodies that bind to the N-terminal region, whereas the other exhibits high intensity for protein bands at the size of the nonglycosylated isoform using antibodies recognizing the C-terminal region. This pattern is the result of an overlap of the nonglycosylated full-length and the glycosylated N-terminal truncated PrP(C) isoforms. Our data demonstrate the importance of antibody selection in characterization of PrP(C).     

The carbohydrates of the isoforms of three avian riboflavin-binding proteins.

The carbohydrate chains of nine isoforms of chicken egg-white riboflavin-binding protein (RfBP) and six isoforms each of quail egg-white and yolk RfBP have been structurally characterized. The two N-glycosylation sites, Asn36 and Asn147, of the most abundant isoform of each of the three proteins were analyzed in further detail leading to the identification of different glycosylation patterns. In both chicken and quail egg-white RfBP the carbohydrates attached to position 36 had a lower degree of branching and, in the case of the quail protein, this site was only partially glycosylated. A very heterogeneous mixture of complex structures was characteristic of the other glycosylation site. Analysis of the two sites in quail yolk RfBP confirmed this result which agrees with what has been established for hen yolk RfBP. The presence in the three proteins of a highly heterogeneous mixture of differently branched glycans suggests that the differences in isoelectric points, which is a peculiarity of the different isoforms, are probably indeed due to differences in carbohydrate structure.     

Identification and characterization of a novel,intracellular isoform of fibroblast growth factor receptor-1 (FGFR-1)

A novel, low molecular weight, intracellular isoform of FGF receptor-1 (FGFR-1) was identified in embryonic chicken tissues using several antibodies specific for different domains of FGF receptors. This low molecular weight isoform differs from the previously characterized isoforms of FGFR-1 in that it contains little or no carbohydrate. Furthermore, in contrast to the other isoforms of FGFR-1, this novel isoform is located exclusively intracellularly. However, it is capable of binding ¹²⁵I-FGF-2 and it possesses intrinsic kinase activity. Pulse-chase experiments indicate that this isoform of FGFR-1 is not simply a precursor to glycosylated FGFR-1 since it can be detected long after the appearance of glycosylated FGFR-1 in the cells. These results suggest that the novel FGFR-1 isoform plays a role in regulating FGF activity distinct from cell surface, glycosylated FGFR-1. The possible roles of this FGFR-1 variant in FGF signaling are discussed. © 1996 Wiley-Liss, Inc.     

Multiple forms of hypophyseal prolactin. A new glycosylated form of prolactin with increased biological activity

Different forms of prolactin obtained from porcine hypophysis and differing in terms of molecular weight, electrophoretic mobility and biological activity were studied by polyacrylamide gel electrophoresis. A glycosylated form of prolactin with Mr 25 000 and possessing an increased biological activity towards pigeon crop was revealed. It was found that the carbohydrate component of this prolactin form is attached to asparagine at position 31; no differences were revealed between the amino acid composition of the major and glycosylated forms of the hormone. In the hypophysis, the glycosylated form content makes up to 30-40% of the total prolactin monomer content. A disulfide dimeric form of prolactin with Mr of about 50 000 was isolated and characterized.     

Characterization of cell-surface glycopeptides from mouse cerebral cortex that inhibit cell growth and protein synthesis.

The occurrence of multiple forms of rat prolactin with different molecular weights (size heterogeneity) was studied with anterior pituitary extracts, purified rat prolactin and 125I-labelled rat prolactin. In each case, three main forms of the hormone were detected by gel filtration on Sephadex G-100: a major one (80--90%) corresponding to monomeric prolactin (mol.wt. 22000--25000), a peak (8--20%) that could be a dimer (mol.wt. 45000--50000) and a small quantity (1--5%) of a component of much greater molecular weight. On freezing and thawing of 125I-labelled rat prolactin, there was little interconversion of monomer and 'dimer' peaks, but both were converted substantially to very high-molecular-weight material. All three peaks of 125I-labelled rat prolactin could be precipitated by anti-(rat prolactin) serum and all three gave similar patterns of radioactive peptides after digestion with chymotrypsin followed by high-voltage paper electrophoresis. On sodium dodecyl sulphate/polyacrylamide-gel electrophoresis, the monomer peak of 125I-labelled prolactin migrated as a single component of mol.wt. 22000, the very high-molecular-weight peak largely dissociated to a component running in the same position as the monomer, and the 'dimer' peak migrated partly as a component of mol.wt. 45000 and partly as a component migrating with monomeric prolactin. No treatment was found that could dissociate the 'dimer' peak completely to monomeric prolactin.     

Impaired intracellular migration and altered solubility of nonglycosylated glycoproteins of vesicular stomatitis virus and Sindbis virus.

Tunicamycin, an antibiotic which prevents the glycosylation of newly synthesized proteins, inhibits the replication of both vesicular stomatitis virus and Sindbis virus. In tunicamycin-treated infected cells, all of the viral proteins are synthesized but the glycoproteins are devoid of carbohydrate. The nonglycosylated glycoproteins could not be detected on the outside of the plasma membrane by lactoperoxidase labeling, indirect immunofluorescence staining, or chymotrypsin treatment of intact cells, whereas the glycosylated glycoproteins were readily detected by all three methods. These results indicate that the bulk of the nonglycosylated glycoproteins are unable to undergo the normal migration to the cell surface. In contrast to the normal glycosylated viral glycoproteins, the nonglycosylated glycoproteins were insoluble in nonionic detergents such as Triton X-100. The nonglycosylated glycoprotein of vesicular stomatitis virus could be solubilized using a combination of 6 M guanidine hydrochloride and 0.2% Triton X-100, but precipitated when the 6 M guanidine was removed by dialysis. These results suggest that the lack of carbohydrate alters the properties of the glycoproteins, which may explain their impaired mobility through the intracellular membranous system.     

A single-amino-acid substitution eliminates the stringent carbohydrate requirement for intracellular transport of a viral glycoprotein.

In this report, we have investigated the contribution of primary sequence to the carbohydrate requirement for intracellular transport of two closely related glycoproteins, the G proteins of the San Juan and Orsay strains of vesicular stomatitis virus. We used site-directed mutagenesis of the coding sequence to eliminate the two consensus sites for glycosylation in the Orsay G protein. Whereas the nonglycosylated San Juan G protein required at least one of its two asparagine-linked oligosaccharides for transport to the plasma membrane at 37 degrees C, a fraction of the Orsay G protein was transported without carbohydrate. Of the 10 amino acid differences between these two proteins, residue 172 (tyrosine in San Juan, aspartic acid in Orsay) played the major role in determining the stringency for the carbohydrate requirement. The rates at which the glycosylated and nonglycosylated Orsay G proteins were transported to the cell surface were the same, although a smaller fraction of the nonglycosylated protein was transported. These results suggest that the carbohydrate does not promote intracellular transport directly but influences a polypeptide folding or oligomerization step which is critical for transport.     

Expression, purification, and kinetic characterization of full-length human fibroblast activation protein

Human fibroblast activation protein (FAP), an integral membrane serine protease, was produced in insect cells as a hexa-His-tagged protein using a recombinant baculovirus expression system. Two isoforms of FAP, glycosylated and nonglycosylated, were identified by Western blotting using an anti-His-tag antibody and separated by lectin chromatography. The glycosylated FAP was purified to near homogeneity using immobilized metal affinity chromatography and was shown to have both postprolyl dipeptidyl peptidase and postgelatinase activities. In contrast, the nonglycosylated isoform demonstrated no detectable gelatinase activity by either zymography or a fluorescence-based gelatinase activity assay. The kinetic parameters of the dipeptidyl peptidase activity for glycosylated FAP were determined using dipeptide Ala-Pro-7-amino-trifluoromethyl-coumarin as the substrate. The k(cat) is 2.0 s(-1) and k(cat)/K(m) is 1.0 x 10(4) M(-1) s(-1) at pH 8.5. The pH dependence of k(cat) reveals two ionization groups with pK(a1) of 7.0 and pK(a2) of 11.0. The pH profile of k(cat)/K(m) yields similar results with pK(a1) 6.2 and pK(a2) 11.0. The neutral pK(a1) is associated with His at the active site. The basic pK(a2) might be contributed from an ionization group that is not involved directly in catalysis, instead associated with the stability of the active site structure.     

Glycoprotein biosynthesis in myeloma cells. Characterization on nonglycosylated immunoglobulin light chain secreted in presence of 2-deoxy-D-glucose.

The biosynthesis and secretion of a glycosylated, K-type immunoglobulin light chain (K-46) was studied in a mouse myeloma tumor, mineral oil plasmacytoma-46B. Viable single cell suspensions were prepared from excised tumors and optimal conditions were established for incorporation of amino acid and carbohydrate precursors into the protein synthesized and secreted by the cells. The glucose analog, 2-deoxy-D-glucose, was utilized as an inhibitor of glycosylation to determine the role of glycosylation in the biosynthesis, intracellular transport, and export of the protein from the cell. It was determined that 6 mM 2-deoxyglucose prevents the incorporation of glucosamine, mannose, and galactose into secreted protein, but permits the incorporation of leucine at approximately 40% of control values. The nonglycosylated protein, secreted in the presence of 2-deoxyglucose, was characterized as a nonglycosylated form of K-46 light chain by the following criteria: (a) electrophoresis in polyacrylamide gels containing sodium dodecyl sulfate, (b) reactivity of the nonglycosylated protein with antisera prepared against native, fully glycosylated, K-46 light chain, (c) analysis of the protein by gel filtration techniques, (d) behavior of the protein on lectin-derivatized Sepharose, and (e) analysis of tryptic peptides derived from the protein. We have concluded that 2-deoxyglucose-inhibited cells synthesize and secrete the normal polypeptide chain of K-46 devoid of its carbohydrate side chain indicating that glycosylation is not an essential step in the biosynthesis, intracellular transport, or export of this protein that is normally synthesized and secreted in a glycosylated form. Under conditions of 2-deoxyglucose inhibition, the nonglycosylated form of K-46 light chain constitutes a significantly greater proportion of accumulated intracellular protein, suggesting that the biosynthesis of the polypeptide chain of K-46 light chain proceeds at a nearly normal rate, but that the absence of the carbohydrate side chain of the protein retards, but does not prevent, the intracellular transport of the protein and its export from the tumor cell.     

Biosynthesis of glycosylated human lysozyme mutants.

Complementary DNA encoding human lysozyme was subjected to oligonucleotide-directed mutagenesis. At one of three selected positions, amino acid residues 22, 68, or 118, the signal for N-linked glycosylation was created. The mutant DNAs were inserted into a eucaryotic vector and transfected into cultured hamster cells. The three mutant cDNAs directed synthesis of lysozyme mutants, which were named LI, LII, and LIII. The mutant lysozymes LI and LII comprised mixtures of glycosylated and nonglycosylated forms. The glycosylated and nonglycosylated forms of mutant LI were found to have an enzymatic activity similar to normal human milk lysozyme. The usage of the glycosylation sites in the mutants was similar in Chinese hamster ovary (CHO) and baby hamster kidney cells. Approximately two of every three molecules in mutant LI, approximately one of every eight molecules in mutant LII, and practically no molecules in mutant LIII became glycosylated. In CHO cells, the processing of the oligosaccharide side chains yielded several larger products than in baby hamster kidney cells. This size variability of glycosylated lysozyme from CHO cells may be explained by the presence of biantennary and triantennary endo-beta-N-acetylglucosaminidase H-resistant oligosaccharides with N-acetyllactosamine repeats of variable length and by the presence of hybrid oligosaccharides, as suggested by affinity to several lectins and sensitivity to endo-beta-galactosidase. In both cell types, the majority of the glycosylated forms were secreted and thus behaved similarly to nonglycosylated lysozyme. A small proportion of mutant LI lysozyme remained associated with the cells. The retained lysozyme was recruited predominantly from the molecules bearing high mannose oligosaccharides. These molecules were targeted to lysosomes, and their carbohydrate was trimmed to an endo-beta-N-acetylglucosaminidase H-resistant form. Owing to the small size of mutant LI lysozyme, minor changes in the size of its carbohydrate moiety result in detectable changes in the electrophoretic mobility of the whole glycoprotein. We suggest that this novel glycoprotein could be used as a reporter in studies on processing and segregation of glycoproteins.     

Attributes of glycosylation in the establishment of the unfolding pathway of soybean agglutinin

Soybean agglutinin (gSBA) is a tetrameric legume lectin, each of whose subunits are glycosylated. Earlier studies have shown that this protein shows exceptionally high stability in terms of free energy of unfolding when compared to other proteins from the same family. This article deals with the unfolding reactions of the nonglycosylated recombinant form of the protein rSBA and its comparison with the glycosylated counterpart gSBA. The nonglycosylated form features a lower stability when compared to the glycosylated form. Further, the unfolding pathways in the two are widely different. Although the glycosylated form undergoes a simple two-state unfolding, the nonglycosylated species unfolds via a compact monomeric intermediate that is not a molten globule. Representative isothermal and thermal denaturation profiles show that glycosylation accounts for a stabilization of approximately 9 kcal/mol of the tetramer, whereas the difference in T(m) between the two forms is 26 degrees C. Computational studies on the glycan-protein interactions at the noncanonical interface of the protein show that quite a number of hydrogen bond and hydrophobic interactions stabilize the glycoprotein tetramer.     

Streptomyces lividans glycosylates the linker region of a beta-1,4-glycanase from Cellulomonas fimi.

The beta-1,4-glycanase Cex of the gram-positive bacterium Cellulomonas fimi is a glycoprotein comprising a C-terminal cellulose-binding domain connected to an N-terminal catalytic domain by a linker containing only prolyl and threonyl (PT) residues. Cex is also glycosylated by Streptomyces lividans. The glycosylation of Cex produced in both C. fimi and S. lividans protects the enzyme from proteolysis. When the gene fragments encoding the cellulose-binding domain of Cex (CBDCex), the PT linker plus CBDCex (PT-CBDCex), and the catalytic domain plus CBDCex of Cex were expressed in S. lividans, only PT-CBDCex was glycosylated. Therefore, all the glycans must be O linked because only the PT linker was glycosylated. A glycosylated form and a nonglycosylated form of PT-CBDCex were produced by S. lividans. The glycosylated form of PT-CBDCex was heterogeneous; its average carbohydrate content was approximately 10 mol of D-mannose equivalents per mol of protein, but the glycans contained from 4 to 12 alpha-D-mannosyl and alpha-D-galactosyl residues. Glycosylated Cex from S. lividans was also heterogeneous. The presence of glycans on PT-CBDCex increased its affinity for bacterial microcrystalline cellulose. The location of glycosylation only on the linker region of Cex correlates with the properties conferred on the enzyme by the glycans.     

Temporal expression of HIV-1 envelope proteins in baculovirus-infected insect cells: implications for glycosylation and CD4 binding

Three different human immunodeficiency virus type I (HIV-1) envelope derived recombinant proteins and the full length human CD4 polypeptide were expressed in Spodoptera frugiperda (Sf9) cells. DNA constructs encoding CD4, gp120, gp160, and gp160 delta (full length gp160 minus the transmembrane and cytoplasmic region of gp41) were cloned into the baculovirus expression vector pVL941 or a derivative and used to generate recombinant viruses in a cotransfection with DNA from Autographa californica nuclear polyhedrosis virus (AcMNPV). Western blotting of cell extracts of the recombinant HIV-1 proteins showed that for each construct two major bands specifically reacted with anti-HIV-1 envelope antiserum. These bands corresponded to glycosylated and nonglycosylated versions of the HIV proteins as determined by 3H-mannose labeling and tunicamycin treatment of infected cells. A time course of HIV envelope expression revealed that at early times post-infection (24 hours) the proteins were fully glycosylated and soluble in nonionic detergents. However, at later times postinfection (48 hours), expression levels of recombinant protein reached a maximum but most of the increase was due to a rise in the level of the nonglycosylated species, which was largely insoluble in nonionic detergents. Thus, it appears that Sf9 cells cannot process large amounts of glycosylated recombinant proteins efficiently. As a measure of biological activity, the CD4 binding ability of both glycosylated and nonglycosylated recombinant HIV envelope proteins was tested in a coimmunoprecipitation assay. The results showed that CD4 and the glycosylated versions of recombinant gp120 or gp160 delta specifically associated with one another in this analysis. Nonglycosylated gp120 or gp160 delta proteins from tunicamycin-treated cultures did immunoprecipitate with anti-HIV-1 antiserum but did not interact with CD4. We conclude that production of native HIV envelope proteins, as measured by addition of carbohydrate side chains and ability to bind CD4, peaks early after infection in baculovirus-infected insect cells.     

Cysteine 17 of recombinant human granulocyte-colony stimulating factor is partially solvent-exposed

Oh-edaet al. have shown instability of granulocyte-colony stimulating factor (G-CSF) upon storage abovepH 7.0 [J. Biol. Chem. (1990)265, 11,432–11,435]. To clarify the mechanism of this instability, the accessibility of a free cysteinyl residue at position 17 for disulfide exchange reaction was examined using a sulfhydryl reagent. The results show that the cysteine is partially solvent-exposed in both glycosylated and nonglycosylated forms, suggesting that the exposure of the cysteine plays a critical role in the instability of the protein. This is supported by the facts that at lowpH where the cysteine is protonated, both proteins have much greater stability and that a Cys17 Ser analog is extremely stable at neutralpH and 37°C. It was observed that the rate of sulfhydryl titration is slower for the glycosylated form than for the nonglycosylated form, suggesting that the cysteine residue is less solvent-exposed for the former protein or that the pK _a is somewhat more basic. In either case, the carbohydrate appears to affect the reactivity of the sulfhydryl group through steric hindrance or alteration in local conformation. Both the glycosylated and nonglycosylated proteins showed essentially identical conformation as determined by circular dichroism, fluorescence, and infrared spectroscopy. Unfolding of these two proteins, induced either by guanidine hydrochloride or bypH, showed an identical course, indicating comparable conformational stability. Contribution of conformational changes to the observed instability at higherpH is unlikely, since little difference in fluorescence spectrum occurs betweenpH 6.0 and 8.0. Based on these observations, G-CSF, whether glycosylated or not, should not be stored above pH 7.0 in solution. On the other hand, G-CSF is extremely stable in acidic solution as expected from the proposed mechanism.     

Glycosylated rat prolactin: isolation and structural characterization.

Isolation of glycosylated 26 kDa rat prolactin and subsequent proper carbohydrate characterization has so far not been reported. In the present work the hormone isoform was isolated to 95% homogeneity by preparative electrophoretic separation on Mini Prep Cell of rat pituitary homogenate. The isoform was then investigated by 2-mercaptoethanol gradient electrophoresis, Cleveland's sequential SDS-PAGE, digestion with endoproteinase Asp-N and N-glycanase. The glycosidic part of the isoform was examined in O-profiling and its monosaccharide composition obtained by FACE and HPAE-PAD analysis. The outcome of the experimental data is: 1) in contrast to unglycosylated 23 kDa rat prolactin, intra-chain S-S bridging is not affected in 26kDa rat prolactin, neither by transiting through a thiol gradient nor in sequential nonreducing/reducing SDS-PAGE; 2) the conformational availability of Asp residues involved in the endoproteinase Asp-N attack is the same in 23- and 26 kDa rat prolactin; the glycan moiety apparently does not cause steric hindrance at this level; 3) no glycosidic N-linkage could be detected, only O-linkage(s); 4) 26 kDa rat prolactin is no glycosyl-phosphaditylinositol-anchored protein; 5) in O-profiling an oligosaccharide chain of Mr +/- 1.4 kDa was recorded; 6) the monosaccharide composition obtained in FACE is peculiar in the sense that next to Fuc, Man, GalNac, GlcNac and NeuAc also Rib was determined; 7) HPAE-PAD analysis identified NeuAc subtypes; 8) in vitro, glycosylation of rat prolactin modulates immune recognition through steric hindrance of the access to the epitope sites.     

Posttranslational modifications of the lutropin receptor: mass spectrometric analysis

Our present knowledge of the lutropin (LH/hCG) receptor structure derives from deductions made from its amino acid sequence as established by studying the cDNA. To obtain direct experimental information, luteinizing hormone (LH) receptor expressed in L cells was immunopurified in sufficient amounts to warrant analysis by mass spectrometry and microsequencing. The mature receptor, complexed to human chorionic gonadotropin (hCG), was purified by using monoclonal antibodies recognizing the hormone, whereas the mannose-rich non-hormone-binding precursor was purified by use of antireceptor antibodies. Determination of the N-terminus showed that (2)/(3) of protein molecules started at Thr24 whereas (1)/(3) started at Ala28. All these molecules bound hCG, suggesting that the most N-terminal region of the receptor does not participate in hormone binding. Six N-glycosylation sites have been predicted from the amino acid sequence. One of them (Asn299) was found to be nonglycosylated in both the precursor and the mature protein. The most heavily glycosylated residue was Asn291, followed by Asn195 and Asn99. These three sites accounted for 82% and 97% of carbohydrate moieties in the mature receptor and in the mannose-rich precursor, respectively. The presence of some receptor molecules nonglycosylated at sites 99, 174, and 195 in hormone-receptor complexes dismisses a direct role of these glycosylation sites in hormone binding or in the correct folding of the protein. The mature carbohydrate chains were homogeneous at position 174, 195, and 313 (absence of Golgi mannosidase II activity at positions 174 and 313, absence of GlcNAc tranferases III and IV activity at position 195). Heterologous carbohydrates were present at sites 99 and 291. The latter, which is highly variable in carbohydrate chains, is unlikely to participate in a direct interaction with hormone. Site 313 thus remains as the main candidate for a role in hormone binding.