Interaction of lectins with proteins of the endoplasmic reticulum and golgi system of rat liver

The membranes of the endoplasmic reticulum contain a number of different enzymes for glycosylation of 2 types of proteins, secretory and membrane glycoproteins. With the exception of albumin all secretory proteins which are transported through the channel system of the endoplasmic reticulum through the Golgi apparatus to the blood contain covalently bound sugar residues. In contrast to the secretory proteins the nature and the function of membrane glycoproteins have not yet been established. There are indications that some microsomal enzymes and ribophorin are glycoproteins (Bischoff, Tran-thi & Decker, 1975; Evans & Gurd, 1973; Dean, 1974; Haugen & Coon, 1976; Grebenau, Sabatini & Kreibich, 1977).

The sugar content of endoplasmic reticulum membranes includes the neutral sugars mannose and galactose, the amino sugar glucosamine and the charged sialic acid (Miyajima, Tomikawa, Kawasaki & Yamashina, 1969; Bergman & Dallner, 1976). The individual glycoproteins have not yet been isolated and their oligosaccharide sequences not yet characterized. However, it is possible to obtain information by using various lectins. The interaction of lectins with membranes has so far been studied mainly with plasma membranes (Cook & Stoddart, 1973). On the other hand, there are also studies which demonstrate that various intact intracellular membranes exhibit a relatively limited interaction with a few types of lectins (Nicolson, Lacourbière & Delmonte, 1972; Henning & Uhlenbruck, 1973; Monneron & Segretain, 1974; Keenan, Franke & Kartenbeck, 1974). If isolated intracellular membranes are solubilized, the ability of integral membrane glycoproteins to associate with lectins increases to a significant extent (Winqvist, Eriksson & Dallner, 1974).

In this paper we have investigated the interaction of a number of lectins with rough and smooth microsomal and Golgi membranes and also analysed lectin interaction with specific submicrosomal particles. It was found that some of the microsomal electron-transport enzymes are inhibited by lectins, which would suggest that they are glycoproteins.

Male Sprague-Dawley rats weighing 180–200 g were starved for 20 h. Total microsomes and rough and smooth subfractions were prepared as described earlier (Dallner, 1974). For preparation of the Golgi fraction, rats were given 1·2 g 50% ethanol/ioo g body weight by stomach tube 90 min before decapitation. That portion of the Golgi complex which has a density less than that of 0·86 M sucrose (Golgi I and II) was used in all experiments (Ehrenreich, Bergeron, Siekevitz & Palade, 1973). In order to remove adsorbed and luminal secretory proteins all fractionswere subjected to theTris-water-Tris washingprocedure(Dallner, 1974). Submicrosomal particles were prepared by using washed total microsomes (Winqvist & Dallner, 1976). Fractionation was performed on a continuous gradient ranging between 0·29 M and 1·75 M sucrose and containing 0·19% deoxycholate. After isopycnic equilibration the 5 bands were collected. In in vivo experiments 300 μCi D-[I-³ H]glucosamine (2000 mCi/mmol, Radiochemical Centre, Amersham, England) were injected into the portal vein of rats under pentobarbital anaesthesia 60 min before decapitation. The specific activities of the washed fractions were: 19000 cpm/mg protein for rough membranes, 5 3 000 cpm/mg protein for smooth membranes, and 12000 cpm/mg protein for Golgi membranes.

The incubation mixture contained 0·3 ml of lectin-Sepharose suspended in 0·15 M Tris-Cl buffer, pH 7·8 (50 % sedimented gel, vol/vol), 0·5 ml buffer or 0·5 ml 0·2 M lectin inhibitor in buffer, 0·05 ml 10 % deoxycholate (excluded when intact membranes were studied) and 02 ml membrane fraction (about 2 mg protein). The amount of lectin used was in excess to the amount of lectin-binding components. The following lectin inhibitors were used: for Con A, a-methyl mannose; for WGA, N-acetyl glucosamine; for Crotalaria lactose and for crude lipase fraction, N-acetylneuraminic acid (NANA). The mixture was incubated for 30 min at 25 °C with gentle shaking. After incubation the lectin-Sepharose complex was washed 3 times by centrifugation using 5 ml Tris-Cl buffer containing 0·1 % deoxycholate. The pellet was suspended in 0·2 ml water and transferred to a scintillation vial with 10 ml Bray’s solution (Bray, 1960).

For labelling of terminal galactose of isolated and purified cytochrome b₅ (Ozols, 1974), 0·5 mg protein in 0·1 M phosphate buffer, pH 7·4 was incubated with 0·05 mg (6 units)galactose oxidase (Worthington, Freehold, New Jersey) overnight at 37 °C. A small crystal containing about 2·5 mCi of sodium PH]borohydride (700 mCi/mmol, The Radiochemical Centre) was added to the mixture and the incubation continued for another 10 min at 37 °C. The suspension was applied to a Sephadex G-25 column equilibrated with water and the void volume was collected and freeze-dried. Part of the cytochrome b₅ was dissolved in 4 % sodium dodecyl sulphate sample buffer containing 5 % mercaptoethanol and used for polyacrylamide gel electrophoresis according to Weber & Osborn (1969). Another portion was dissolved in 01 M phosphate buffer, pH 7·4, and incubated with an excess of anti-cytochrome b₅ antiserum (kindly supplied by Dr A. Elhammer) for 2 h at 25 °C and overnight at 4 °C. The mixture was then applied to a protein A-Sepharose column. After washing with 10 column volumes (no more radioactivity was eluted), the cytochrome b₅-antibody complex was eluted with 0·2 M citrate buffer, pH 3.

Protein was determined according to Lowry, Rosebrough, Farr & Randall (1951) with bovine serum albumin as standard. N-acetylneuraminic acid was liberated by hydrolysis in 0·05 M H₂SO₄, 60 min 80 °C, purified by ion exchange chromatography (Svennerholm, 1963) and the free sialic acid was measured with the Warren procedure (Warren, 1963). N-acetyl-neuraminic acid type VI from Sigma Co. (St Louis, Miss.) was used as a standard. The various enzyme activities were measured according to methods described earlier (Eriksson, 1973; Beaufay et al. 1974).

Con A and WGA as well as Con A-Sepharose and WGA-Sepharose were purchased from Pharmacia Fine Chemicals AB, Uppsala, Sweden. Lectins from Crotalaria, Vicia cracca, Vicia ervilia, Vicia sativa, Vicia villosa (mannose-or N-acetyl galactosamine-specific) were all kindly supplied by Dr B. Ersson, Uppsala University, Sweden.

For preparation of crude lipase lectin, 10 g wheat germ lipase (Type I, Sigma) were dissolved in 400 ml 0·05 M phosphate buffer, pH 7·4. After centrifugation for 30 min at 10000 g the supernatant was decanted and applied to an affinity column, where glucosamine had been coupled to CH-Sepharose (Pharmacia). After washing with 0·2 M NaCl the lectin was eluted with 0·15M N-acetyl glucosamine in 0·05 M phosphate buffer, pH 7·4, dialysed against water and freeze-dried. Typically, 5 mg of protein were recovered. In sodium dodecyl sulphatepolyacrylamide electrophoresis, this preparation, designated as crude lipase lectin, gave 2 bands, a weak one at mol. wt 25 000 and a strong one at mol. wt 38000.

The various lectins were coupled to CNBr-activated Sepharose according to the method of Cuatrecasas (Cuatrecasas, 1970).

Rough and smooth membranes labelled in vivo with glucosamine were incubated with Con A, WGA and Crotalaria lectins covalently bound to Sepharose. The intact membrane demonstrated a small amount of interaction, about half of which could be inhibited by the specific sugar inhibitors (Table 1). When, on the other hand, the membranes were solubilized with detergent prior to incubation, the percentages of the total radioactivity bound to Con A were 37, 28 and 20% in rough, smooth and Golgi membranes, respectively. Only an insignificant portion of this binding could not be specifically inhibited. Nor did Sepharose without lectin bind any significant amount of radioactivity (Winqvist et al. 1974). Binding to WGA was less than that to Con A and was lowest in rough, moderate in smooth, and highest in Golgi membranes. Binding to Crotalaria lectin was low in all 3 fractions.

In Table 2 are shown results using an additional lectin, i.e. crude lipase, which represents a partially purified preparation from commercial wheat germ lipase. Here substantial binding of all 3 fractions was observed. When the fractions were pretreated with neuraminidase, there were no changes in Con A-binding, whereas the increase in binding to Crotalaria indicates that the dominant sugar moiety next to terminal NANA in these membrane glycoproteins is galactose. Neuraminidase pretreatment decreased the WGA binding of all 3 fractions to some extent, while the binding of rough and Golgi membranes to crude lipase was greatly diminished. Pretreatment with low concentrations of deoxycholate, known to abolish the membrane permeability barrier (Kreibich, Debey & Sabatini, 1973), did not in itself alter the binding, but neuraminidase treatment in the presence of deoxycholate removed almost all the binding sites for crude lipase in smooth membranes.

Affinity chromatography experiments were performed to investigate the binding to crude lipase lectin. Fetuin and asialofetuin (where the sialic acid had been removed by hydrolysis) were applied to a crude lipase-Sepharose column (Fig. 1.). When fetuin was used, almost none appeared in the eluate, but after application of NANA, all fetuin in the column was displaced and could be recovered in subsequent fractions. Asialofetuin, on the other hand, passed through the column and could be collected at the front. NANA could not displace any further amounts from the column. This indicates interaction of NANA with the crude lipase. To exclude the possibility that the neuraminidase employed is contaminated with glucosaminidase, thin-layer chromatography was run on the products of the hydrolysis (Fig. 2). The only sugar released by neuraminidase was NANA.

Chemical determination of NANA in rough and smooth microsomes after neuraminidase treatment also supports the view that crude lipase interacts with terminal NANA residues. This treatment removes a large part of the NANA from rough but not from smooth microsomes (Table 3). This Table also demonstrates that after pretreatment with protease, neuraminidase has access to additional NANA residues in rough microsomes and a large part of the NANA in smooth membranes can also be removed now. This combined treatment only affects the outer surface of the membrane, since permeability studies showed that the vesicles were still impermeable to large macromolecules such as Dextran 70 ∞o. Liver microsomes are known to retain their original inside-outside orientation even after various enzyme treatments (Nilsson et al. 1978).

A sizeable fraction of the protein-bound neutral and amino sugar residues are contained in the trypsin-sensitive portion of intact cytoplasmic membranes (Bergman & Dallner, 1976). Therefore, trypsin-treated solubilized membranes were tested for lectin binding (Table 4). Trypsin treatment did not influence the amount of labelled glycoproteins interacting with Con A. On the other hand, proteolysis of both rough and smooth microsomes decrease their binding to WGA and Crotalaria. In fact, Crotalaria binding to smooth membranes from which NANA had been removed was almost abolished. The lack of effect of trypsin on binding to Con A does not necessarily exclude the presence of terminal mannose at the cytoplasmic surface of the membrane, since there may still be enough sugar left for binding.

In Table 5 the 3 types of membrane were solubilized and incubated with 5 different lectins all known to be inhibited by α-methyl mannose. Since both the amount and the pattern of binding vary with the lectin used, it would seem that the binding is influenced by several factors and not exclusively dependent on the presence of a specific sugar moiety (Allen, Neuburger & Sharon, 1973; Pereira, Kisailus, Gruezo & Kabat, 1978).

Submicrosomal particles may be prepared by isopycnic equilibration of microsomal membranes through a deoxycholate-containing continuous sucrose gradient (Winqvist & Dallner, 1976). Five separate fractions are recovered after centrifugation, with a phospholipid/protein ratio which decreases from 2·5i in the top fraction to 0·n in the bottom one. The 5 fractions also exhibit specific patterns of enzyme activity (Table 6). AMPase and other microsomal hydrolases are mainly localized in fraction 1. The enzymes participating in NADH oxidation are enriched in fraction 2, and those involved in NADPH oxidation in fraction 3. Glucose-6-phosphatase activity is present in fractions 1 and 4, but the highest activity by far is in fraction 5.

After labelling in vivo all of these subfractions displayed protein-bound glucosamine label in decreasing amounts in the order 2, 1, 3, 4 and 5 (Table 7). Microsomes were also treated with neuraminidase in the presence of low concentrations of deoxycholate before preparing submicrosomal particles. The greatest losses of radioactivity were observed in fractions 1, 2 and 3.

Experiments with lectins show that the submicrosomal particles described above contain various types of glycoproteins (Table 8). Interaction with Con A occurs with the upper 3 fractions and is most pronounced in fraction 2. The picture with WGA is somewhat different. Here the interaction is more general, being highest in fractions 2, 3, and 4. Binding to Crotalaria and ErviHa lectins is low in the various fractions, while binding to 7V-acetyl galactosamine-specific Villosa lectin is absent. Submicrosomal particles prepared from neuraminidase-treated microsomes exhibit a decreased WGA-binding by the upper 3 fractions, indicating that some of the interaction is due to NANA residues. The same 3 fractions display a highly significant (11- to 26fold) increase in interaction with Crotalaria lectin after neuraminidase treatment. Thus, the submicrosomal fractions prepared show both specific enzyme patterns and characteristic lectin interactions.

A number of experimental approaches are available which demonstrate that the interaction of lectins with the cell membrane has both structural and functional consequences. The behaviour of microsomal vesicles in the presence of various lectins was investigated (Table 9). In the presence of Con A, Ervilia and WGA all microsomal vesicles existed as separate entities. On the other hand, with increasing concentrations of Crotalaria microsomes demonstrated a parallel increase in aggregation.

Several microsomal enzyme functions were investigated after incubation of intact vesicles with various lectins. Prolonged incubation with Crotalaria lectin does not interfere with NADH-ferricyanide reductase which, however, is almost completely inhibited in the presence of Ervilia lectin (Fig. 3). NADH-cytochrome c reductase activity, which involves the interaction of flavoprotein with cytochrome b₅, is inhibited not only by Ervilia lectin but also by Crotalaria. One cannot expect a complete inhibition with Crotalaria lectin even if it interacts with cytochrome b₅, since the cytochrome is present in 1o-fold excess over the reductase. The inhibitions of NADH-ferricyanide reductase with Ervilia lectin and of NADH-cytochrome c reductase with both Ervilia and Crotalaria lectins indicate that both the flavoprotein and the cytochrome are glycoproteins. Lectin inhibition, however, does not necessarily imply that the enzyme itself is a glycoprotein. The lectin could bind to an adjacent protein in the membrane and thus prevent the penetration of the substrate to the membrane enzyme, particularly in the case of large substrates such as the cytochromes. Further evidence for the glycoprotein nature of cytochrome b₅ is given by treatment with galactose oxidase followed by labelling with tritiated borohydride. Detergent-isolated cytochrome b₅ was labelled with the galactose oxidase procedure. Sodium dodecyl sulphate-gel electrophoresis gave a single labelled peak at the same position as the non-labefled cytochrome (Fig. 4).

The galactose oxidase-labelled cytochrome b₅ was also incubated with specific antibodies and the resulting complex was adsorbed onto a protein A-Sepharose column (Fig. 5). The cytochrome-antibody complex was eluted and the labelled protein peak was identified as cytochrome b₅ by gel electrophoresis.

In this study the interaction of glycoproteins of the endoplasmic reticulum with various types of lectins was investigated. Sepharose-bound lectins do not interact appreciably with intact microsomal or Golgi vesicles, probably for 2 reasons. First, the oligosaccharide residues of the membrane glycoproteins at the cytoplasmic surface appear to be short and thus may not reach the lectin-binding site. And secondly, interaction may be prevented because the microsomal vesicle is larger than the channels in the Sepharose bead and thus may not reach lectin molecules bound inside the bead.

Detergent-solubilized membranes contain glycoproteins which react with various types of lectins. Concanavalin A, Crotalaria, WGA and a crude fraction from wheat germ lipase were used. The specificity of Con A for terminal mannose is well established, both by X-ray analysis (Becker et al. 1975) and inhibition studies (Portez & Goldstein, 1970). Crotalaria lectin interacts mainly with galactose and lactose (Ersson, Aspberg & Porath, 1973), while WGA binds mainly 7V-acetyl glucosamine (Allen, Neuberger & Sharon, 1973). A number of factors influence the binding of glycoproteins to lectins (Lis & Sharon, 1973), e.g., the nature of the terminal sugar, the length and the composition of the oligosaccharide chain, the type of sugar interaction and the structure ot the peptide. Still, the use of lectins is one of the most valuable methods for studying the nature of the terminal sugar residues of protein-bound oligosaccharide chains. Our experiments demonstrate that a crude preparation of wheat germ lipase contains a lectin interacting mainly with NANA. The preparation was not further purified and we cannot tell whether the binding is of the ‘lock-and-key’ type or involves a more unspecific electrostatic interaction. Nevertheless, the binding and inhibitor studies indicated that the preparation may be used as a lectin for probing terminal NANA residues. It should be added that pure WGA also interacts to some extent with terminal NANA, since neuraminidase treatment decreased binding of this lectin both to microsomes and submicrosomal particles. This is in agreement with previous studies (Allen et al. 1973; Greenaway & LeVine, 1973; Redwood & Polefka, 1976; Jordan, Basset & Redwood, 1977).

The dominating interaction in the case of membrane glycoproteins was that with Con A, indicating a large amount of terminal or 1→2-linked mannose (Goldstein, Reichert & Misaki, 1974), in decreasing amount in the order rough and smooth microsomes and Golgi membranes. The total amount of glycosamine and NANA residues increases in the order rough and smooth microsomes and Golgi membranes, as could be demonstrated with WGA and crude lipase. This type of sugar distribution is in agreement with the subcellular distribution of the glycosylation processes, since in contrast to GDP-mannosyl transferase both UDP-glycosaminyl and CMP-sialyl transferases are enriched in the Golgi system (Nilsson et al. 1978; Schachter et al. 1970). There is a small amount of binding to Crotalaria lectin with all 3 fractions, and this can be increased by pretreating the membranes with neuraminidase. Obviously, the sugar next to the terminal NANA in these membranes is galactose. Both chemical measurements and lectin interaction studies show that a large part of the protein-bound oligosaccharide chains terminating with galactose-NANA are on the cytoplasmic side of rough and Golgi membranes, while in smooth microsomes about half of these chains are located on this side, but they appear to be buried by neighbouring protein molecules. Compartmentalization of the membrane glycoproteins was not studied in more detail, but is is clear from these treatments with various hydrolytic enzymes that some protein-bound sugar residues are located at the cytoplasmic side of the membrane.

High Con A-binding was observed with lipid- and AMPase-rich (fraction i), NADH oxidase-rich (fraction 2) and NADPH oxidase-rich (fraction 3) submicrosomal particles. The same fractions also exhibited WGA binding which was mainly neuraminidase-sensitive. WGA-binding was also observed with the glucose-6-phosphatase-rich particles (fractions 4 and 5), but this binding was not influenced by neuraminidase treatment. Thus, glucosamine appears to be the dominating protein-bound sugar in these fractions. The high WGA-binding by submicrosomal particles and also the large amount of binding to Crotalaria lectin after neuraminidase treatment can be explained by the fact that the membranes were subjected to a long deoxycholate treatment (48 h) which should result in uncovering and unfolding of polypeptide chains.

In this study glycoproteins were labelled with glucosamine 1 h before decapitation. Consequently, we cannot expect labelling of all protein-bound oligosaccharide chains and furthermore, the label is not randomly distributed. Clearly, those glycoproteins which are newly synthesized contain all the label. In spite of the intensive movement of the glycoproteins, partly within the endoplasmic reticulum and partly between the membrane and cytoplasmic compartment (Omura & Kuriyama, 1971; Autuori, Svensson & Dallner, 1975), it is not possible to obtain a complete labelling or an even distribution of the label and, therefore, the approach employed here should be interpreted in qualitative terms.

The lectin-inhibition studies suggest that the NADH-oxidizing flavoprotein is a glycoprotein with a terminal mannose residue and that cytochrome b₅ also has an oligosaccharide chain with terminal galactose. Chromatographic and electrophoretic analyses of the galactose oxidase-labelled cytochrome b₅ also supported the idea that cytochrome b₅ has a covalently bound oligosaccharide chain. Other investigations have demonstrated that cytochrome b₅ reductase and cytochrome b exhibit similar structures composed of a hydrophobic part in a lipid environment and a hydrophilic protruding into the water phase (Spatz & Strittmatter, 1971, 1973). It appears probable that the hydrophilic portion of these enzymes contains an oligosaccharide chain, which may represent some of the sugar residues of the outer surface of microsomal vesicles.

The authors wish to thank Dr. Bo Ersson, University of Uppsala, for a number of lectins and Dr. Ake Elhammer for purified cytochrome b₅ and rabbit antiserum against cytochrome b₅. This work has been supported by grants from the Swedish Medical Research Council and the Magnus Bergwall Foundation.