Glycoprotein synthesis in the adult rat pancreas. IV. Subcellular distribution of membrane glycoproteins

Zymogen granule membranes from the rat exocrine pancreas displays distinctive, simple protein and glycoprotein compositions when compared to other intracellular membranes. The carbohydrate content of zymogen granule membrane protein was 5-10-fold greater than that of membrane fractions isolated from smooth and rough microsomes, mitochondria and a preparation containing plasma membranes, and 50-100-fold greater than the zymogen granule content and the postmicrosomal supernate. The granule membrane glycoprotein contained primarily sialic acid, fucose, mannose, galactose and N-acetylglucosamine. The levels of galactose, fucose and sialic acid increased in membranes in the following order: rough microsomes less than smooth microsomes less than zymogen granules. Membrane polypeptides were analyzed by polyacrylamide gel electrophoresis in sodium dodecyl sulfate. The profile of zymogen granule membrane polypeptides was characterized by GP-2, a species with an apparent molecular weight of 74 000. Radioactivity profiles of membranes labeled with [3H]glucosamine or [3H]leucine, as well as periodic acid-Schiff stain profiles, indicated that GP-2 accounted for approx. 40% of the firmly bound granule membrane protein. Low levels of a species similar to GP-2 were detected in membranes of smooth microsomes and the preparation enriched in plasma membranes but not in other subcellular fractions. These results suggest that GP-2 is a biochemical marker for zymogen granules. Membrane glycoproteins of intact zymogen granules were resistant to neuraminidase treatment, while those in isolated granule membranes were readily degraded by neuraminidase. GP-2 of intact granules was not labeled by exposure to galactose oxidase followed by reduction with NaB3H4. In contrast, GP-2 in purified granule membranes was readily labeled by this procedure. Therefore GP-2 appears to be located on the zymogen granule interior.     

Glycoprotein synthesis in the adult rat pancreas: II. Characterization of Golgi-rich fractions

Golgi-rich fractions were prepared from homogenates of adult rat pancreas by discontinuous gradient centrifugation. These fractions were characterized by stacks of cisternae associated with large, irregular vesicles and were relatively free of rough microsomes, mitochondria, and zymogen granules. The Golgi-rich fractions contained 50% of the UDP-galactose: glycoprotein galactosyltransferase activity; the specific activity was 12-fold greater than the homogenate. Such fractions represented < 19% of thiamine pyrophosphatase, uridine diphosphatase, adenosine diphosphatase, and Mg²⁺-adenosine triphosphatase. Zymogen granules and the Golgi-rich fractions were extracted with 0.2 m NaHCO3, pH 8.2, and the membranes were isolated by centrifugation. The glycoprotein galactosyltransferase could not be detected in granule membranes, while the specific activity in Golgi membranes was 25-fold greater than the homogenate.At least 35 polypeptide species were detected in Golgi membranes by polyacrylamide gel electrophoresis in 1% sodium dodecylsulfate. These ranged in molecular weight from 12,000 to <160,000. There were only minor differences between Golgi membranes and smooth microsomal membrane. In contrast, zymogen granule membranes contained fewer polypeptides. A major polypeptide, which represented 30–40% of the granule membrane profile, accounted for less than 3% of the polypeptides of Golgi membranes or smooth microsomal membranes.     

A possible role of Golgi membrane-associated galactosyltransferase in the formation of zymogen granule glycoproteins

A galactosyltransferase activity in smooth microsomes and Golgi membrane-rich fractions from rat pancreas glycosylated endogenous acceptors during incubation with UDP-[¹⁴C]galactose in the absence of exogenous glycoproteins. To evaluate the role of this activity in secretion, the endogenous products were partially characterized. Galactose-labeled fractions were sequentially extracted in 0.2 m NaHCO₃ and 0.25 m NaBr to prepare membranes and soluble acceptors. Bound radioactivity was equally distributed between these two fractions. Analysis by polyacrylamide gel electrophoresis in sodium dodecyl sulfate indicated that the particulate galactose-labeled polypeptides were distinct from the soluble galactose acceptors. Rabbit antisera against highly purified zymogen granule membranes precipitated approximately 40% of the radioactivity of the particulate fraction when solubilized in nonionic detergents. In polyacrylamide gels, the galactose-labeled species of the immunoprecipitate migrated with zymogen granule membrane glycoproteins. Rabbit antisera against secretory proteins cross-reacted with less than 5% of the galactose-labeled soluble acceptors. Mature zymogen granule membranes neither contained detectable galactosyltransferase activity nor served as galactosyltransferase acceptors. These results suggest that galactosyltransferase activity associated with membranes derived from the Golgi complex glycosylated zymogen granule membrane precursors. Analysis of [¹⁴C]galactolipids did not implicate lipid intermediates in this process.     

Glycosylation of rat liver cytochrome b5 on the ribosomal level

The distribution and site of synthesis of cytochrome $\text{b}{}_5$ was studied by antibody precipitation of the enzyme labeled $\text{in}\text{vivo}$. The enzyme is present in rough and smooth microsomes, Golgi and outer mitochondrial membranes. The cytochrome is synthesized only on bound ribosomes, where glucosamine and galactose moieties are also added. The enzyme seems to be devoid of mannose and sialic acid residues.     

Localization of protein-bound carbohydrate residues on the cytoplasmic surface of rough and smooth microsomes and golgi vesicles from rat liver

Rough and smooth microsomes and Golgi membranes isolated from rat liver were treated with proteolytic enzymes under conditions which removed 30–40% of the surface proteins without seriously disrupting the membrane structure. This treatment also removed 40–60% of protein-bound mannose, galactose and glucosamine. When protease treatment was combined with neuraminidase treatment, 80% of the sialic acid was removed from intact rough microsomal and Golgi vesicles and about half of the sialic acid of smooth microsomes was solubilized. It appears that half, or probably more, of the membrane glycoproteins are associated with the cytoplasmic surface of these membranes.     

COMPOSITION OF CELLULAR MEMBRANES IN THE PANCREAS OF THE GUINEA PIG : I. Isolation of Membrane Fractions

The subcellular components involved in the synthesis, transport, and discharge of secretory proteins in the guinea pig pancreatic exocrine cell have been isolated from gland homogenates by differential and gradient centrifugation. They include rough and smooth microsomes derived respectively from the rough endoplasmic reticulum and Golgi periphery, a zymogen granule fraction consisting mainly of mature zymogen granules and a smaller population of condensing vacuoles, and a plasmalemmal fraction. Membrane subfractions were obtained from the particulate components by treatment with mild (pH 7.8) alkaline buffers which extract the majority (>95%) of the content of secretory proteins, allowing the membranes to be recovered from the extracting fluid by centrifugation. The purity of the fractions was assessed by electron microscopy and by assaying marker enzymes for cross-contaminants. The rough and smooth microsomes were essentially free of mitochondrial contamination; the smooth microsomes contained <15% rough contaminants. The zymogen granule fraction and its derived membranes were free of rough microsomes and contained <3% contaminant mitochondria. The plasmalemmal fraction was heterogeneous as to origin (deriving from basal, lateral, and apical poles of the cell) and contained varying amounts of adherent fibrillar material arising from the basement membrane and terminal web. The lipid and enzymatic composition of the membrane fractions are described in the following reports.     

Alterations in labeling of cell-surface glycoproteins from normal and diabetic rat intestinal microvillous membranes

The effect of chronic streptozotocin-induced diabetes was studied on intestinal microvillous membrane surface carbohydrate groups. After 7 weeks of diabetes, purified microvillous membranes were prepared from rat small intestine and surface galactoproteins identified by labeling with galactose oxidase/sodium boro[³H]hydride. Membrane surface sialic acid residues were labeled using the sodium metaperiodate/sodium boro[³H]hydride technique. Membranes were solubilized in SDS and protein labeling analyzed by acrylamide electrophoresis. Membranes from diabetic rats showed an 81% increase in galactoprotein labeling ($\text{P< 0.02}$) while labeling of sialic acid residues was unchanged. The greatest increase in galactoprotein labeling occurred in protein monomers of $\text{M}{}_{\mathrm{<mtext>r</mtext>}}$ 116 000–200 000, where there was a 155% increase in labeling ($\text{P< 0.005}$). These results indicate that intestinal microvillous membrane protein glycosylation is altered in chronic diabetes. This increase in surface membrane carbohydrates could explain the decreased rates of proteolytic degradation previously described for at least one microvillous protein. An increase in membrane galactose groups has also been noted in hepatocyte and kidney glomerular basement membranes, which suggests the presence of a systematic change in membrane protein glycosylation occurring as a result of the diabetic state.     

COMPOSITION OF CELLULAR MEMBRANES IN THE PANCREAS OF THE GUINEA PIG : II. Lipids

The lipid composition of rough and smooth microsomal membranes, zymogen granule membranes, and a plasmalemmal fraction from the guinea pig pancreatic exocrine cell has been determined. As a group, membranes of the smooth variety (i.e., smooth microsomes, zymogen granule membranes, and the plasmalemma) were similar in their content of phospholipids, cholesterol and neutral lipids, and in the ratio of total lipids to membrane proteins. In contrast, rough microsomal membranes contained much less sphingomyelin and cholesterol and possessed a smaller lipid/protein ratio. All membrane fractions were unusually high in their content of lysolecithin (up to ~20% of the total phospholipids) and of neutral lipids, especially fatty acids. The lysolecithin content was shown to be due to the hydrolysis of membrane lecithin by pancreatic lipase; the fatty acids, liberated by the action of lipase on endogenous triglyceride stores, are apparently scavenged by the membranes from the suspending media. Similar artifactually high levels of lysolecithin and fatty acids were noted in hepatic microsomes incubated with pancreatic postmicrosomal supernatant. E 600, an inhibitor of lipase, largely prevented the appearance of lysolecithin and fatty acids in pancreatic microsomes and in liver microsomes treated with pancreatic supernatant.     

INTRACELLULAR TRANSPORT OF SECRETORY PROTEINS IN THE PANCREATIC EXOCRINE CELL : II. Transport to Condensing Vacuoles and Zymogen Granules

In the previous paper we described an in vitro system of guinea pig pancreatic slices whose secretory proteins can be pulse-labeled with radioactive amino acids. From kinetic experiments performed on smooth and rough microsomes isolated by gradient centrifugation from such slices, we obtained direct evidence that secretory proteins are transported from the cisternae of the rough endoplasmic reticulum to condensing vacuoles of the Golgi complex via small vesicles located in the periphery of the complex. Since condensing vacuoles ultimately become zymogen granules, it was of interest to study this phase of the secretory cycle in pulse-labeled slices. To this intent, a zymogen granule fraction was isolated by differential centrifugation from slices at the end of a 3-min pulse with leucine-¹⁴C and after varying times of incubation in chase medium. At the end of the pulse, few radioactive proteins were found in this fraction; after +17 min in chaser, its proteins were half maximally labeled; they became maximally labeled between +37 and +57 min. Parallel electron microscopic radioautography of intact cells in slices pulse labeled with leucine-³H showed, however, that zymogen granules become labeled, at the earliest, +57 min post-pulse. We assumed that the discrepancy between the two sets of results was due to the presence of rapidly labeled condensing vacuoles in the zymogen granule fraction. To test this assumption, electron microscopic radioautography was performed on sections of zymogen granule pellets isolated from slices pulse labeled with leucine-³H and subsequently incubated in chaser. The results showed that the early labeling of the zymogen granule fractions was, indeed, due to the presence of highly labeled condensing vacuoles among the components of these fractions.     

The pancreatic membrane protein GP-2 localizes specifically to secretory granules and is shed into the pancreatic juice as a protein aggregate

GP-2 is the major secretory granule membrane glycoprotein of the exocrine pancreas and appears in the pancreatic juice in a modified sedimentable form. We have localized GP-2 in the rat pancreas at the electron microscopic level using affinity-purified antibodies and found it to be concentrated in the zymogen granules and in the acinar lumen. Label was also present on the apical and basolateral plasma membranes but prior treatment of the sections with periodate to eliminate the contribution of highly antigenic oligosaccharide moieties reduced substantially the staining of the basolateral surface. Approximately 45% of the GP-2 in the granules was not membrane-associated but appeared instead in the granule lumen. Parallel biochemical characterization of GP-2 in isolated secretory granules demonstrated that 60% fractionated with the membranes after granule lysis while 40% remained in the content fraction. Unlike the membrane-associated form of the protein, which is linked to the membrane via glycosyl-phosphatidylinositol (GPI), GP-2 in the content did not enter the detergent phase upon Triton X-114 extraction; nor was it sedimentable at 200,000g, as is characteristic of the form collected in the pancreatic juice. In addition, GP-2 in the pancreatic juice was recovered in the aqueous phase during Triton X-114 extraction and yet remained sedimentable after detergent extraction, demonstrating that its ability to remain in large aggregates was independent of lipid. These results are consistent with a life cycle for the protein that begins with synthesis of a membrane-associated precursor that can be converted by lipolytic or proteolytic cleavage to a soluble form within the zymogen granule. Further modification to a sedimentable form may then occur in the pancreatic juice.     

CONDENSING VACUOLE CONVERSION AND ZYMOGEN GRANULE DISCHARGE IN PANCREATIC EXOCRINE CELLS: METABOLIC STUDIES

We have examined, in the pancreatic exocrine cell, the metabolic requirements for the conversion of condensing vacuoles into zymogen granules and for the discharge of the contents of zymogen granules. To study condensing vacuole conversion, we pulse labeled guinea pig pancreatic slices for 4 min with leucine-³H and incubated them in chase medium for 20 min to allow labeled proteins to reach condensing vacuoles. Glycolytic and respiratory inhibitors were then added and incubation continued for 60 min to enable labeled proteins to reach granules in control slices. Electron microscope radioautography of cells or of zymogen granule pellets from treated slices showed that a large proportion of prelabeled condensing vacuoles underwent conversion in the presence of the combined inhibitors. Osmotic fragility studies on zymogen granule suspensions suggest that condensation may result from the aggregation of secretory proteins in an osmotically inactive form. Discharge was studied using an in vitro radioassay based on the finding that prelabeled zymogen granules can be induced to release their labeled contents to the incubation medium by carbamylcholine or pancreozymin. Induced discharge is not affected if protein synthesis is blocked by cycloheximide for up to 2 hr, but is strictly dependent on respiration. The data indicate that transport and discharge do not require the pari passu synthesis of secretory or nonsecretory proteins (e.g. membrane proteins), suggesting that the cell may reutilize its membranes during the secretory process. The energy requirements for zymogen discharge may be related to the fusion-fission of the granule membrane with the apical plasmalemma.     

Use of colloidal gold particles in double-labeling immunoelectron microscopy of ultrathin frozen tissue sections

Complexes of protein-A with 5 and 16 nm colloidal gold particles (PA/Au5 and PA/Au16) are presented as sensitive and clean immunoprobes for ultrathin frozen sections of slightly fixed tissue. The probes are suitable for indirect labeling and offer the opportunity to mark multiple sites. The best procedure for double labeling was to use the smaller probe first, i.e., antibody 1 - PA/Au5 - antibody 2 - PA/Au16. When this was done, no significant interference between PA/Au5 and PA/Au16 occurred. Using this double-labeling procedure we made an accurate comparison between the subcellular distributions of amylase as a typical secretory protein and of GP-2 a glycoprotein, characteristic for zymogen granule membrane (ZGM) preparations. We prepared two rabbit antibodies against GP-2. One antibody (R x ZGM) was obtained by immunizing with native membrane material. The specificity of R x ZGM was achieved by adsorption with the zymogen granule content subfraction. The other, R x GP-2, was raised against the GP-2 band of the SDS polyacrylamide profile of ZGM. We found that the carbohydrate moiety of GP-2 was involved in the antigenic determinant for R x ZGM, while R x GP-2 was most likely directed against GP-2 polypeptide backbone. THe immunocytochemical observations showed that GP-2, on the one hand, exhibited the characteristics of a membrane protein by its occurrence in the cell membrane, the Golgi membranes, and its association with the membranes of the zymogen granules. On the other hand, GP-2 was present in the contents of the zymogen granules and in the acinar and ductal lumina. Also, a GP-2-like glycoprotein was found in the cannulated pancreatic secretion (Scheffer et al., 1980, Eur. J. Cell Biol. 23:122-128). Hence, GP-2 should be considered as a membrane-associated secretory protein of the rat pancreas.     

Incorporation of galactose from UDP-galactose into microsomal and golgi membranes of rat liver

Rough and smooth microsomes and Golgi membranes were incubated with UDP[¹⁴C]galactose and the incorporation of radioactivity into the lipid extract and into endogenous protein acceptors were measured. Antagonistic pyrophosphatases were inhibited with ATP and interference from β-galactosidase activity was greatly decreased by carrying out the incubation at pH 7.8. After incubation the particles were centrifuged to remove free oligosaccharide residues. Radioactivity was found in the lipid extract from Golgi membranes but not from rough and smooth microsomes. This radioactivity, however, was not associated with dolichol or retinyl phosphates. The incorporation of radioactivity into proteins of the Golgi fraction was more than double than that of the microsomal fractions. In addition, the transferases in these two types of particles exhibited different properties. Trypsin treatment of intact rough microsomal vesicles, smooth vesicles and Golgi membranes removed about 5, 15 and 50%, respectively, of newly incorporated protein-bound galactose, indicating that the proportion of the newly galactosylated proteins, which are localized at the cytoplasmic surface of the membrane, is lowest in rough microsomes, intermediate in smooth, and highest in Golgi membranes.     

Association of sialic acid with microsomal membrane structures in rat liver

The amount of sialic acid on phospholipid basis increases from rough, through smooth II and smooth I microsomes, to Golgi membranes, all of them free from most of the adsorbed and luminal protein. The incorporation rate of glucosamine-³H into sialic acid also follows a similar order. Deoxycholate removes phospholipid and sialic acid to an identical extent, and a significant part of the latter remains after trypsin and neuraminidase treatment. The sialic acid/phospholipid ratio decreases in phenobarbital-induced smooth but not in rough membranes, while the incorporation rate of glycosamine-³H into sialic acid decreases in both subfractions.     

Identification of the proteins exposed on the cytoplasmic surface of the pancreatic zymogen granule

Lactoperoxidase-catalyzed ¹²⁵I-iodination was used to label pancreatic zymogen granules. Membrane proteins facing the cytoplasmic surface were specifically labeled. Two low molecular weight proteins of 17000 and 15000 were intensely labeled at 0°C. Another small 13 kDa protein was strongly iodinated at 25°C along with some others, including the 29 kDa subunit of the ATP diphosphohydrolase. The major glycoprotein of the granule membrane was not iodinated but the presence of an iodinated 80 kDa protein suggests that proteolytic fragments of the 92 kDa glycoprotein were accessible to iodination on the intact granule. These proteins localized on the cytoplasmic surface of the granule are believed to play a major role in the exocytotic phenomenon of the exocrine pancreas.     

The major zymogen granule membrane protein GP-2 in the rat pancreas is not involved in granule formation.

The major membrane protein of zymogen granules in the rat pancreas is a glycoprotein of 78 kDa (GP-2), which is inserted into the membrane via a glycosyl-phosphatidylinositol (GPI) anchor. GP-2 occurs in both, a membrane-attached and a soluble form. Due to its specific luminal orientation and its quantitative contribution to the zymogen granule membrane, GP-2 has been postulated to play an important role in sorting of digestive enzymes into the granule and in the formation of the granule as a storage organelle. We have tested this hypothesis in the rat pancreas under three different functional conditions, where both the rates of enzyme/isoenzyme synthesis change drastically, and new zymogen granules form at a high rate: a) during prolonged hormonal stimulation of the adult rat pancreas, b) during the differentiation of AR4-2J cells induced by dexamethasone in vitro, and c) during embryonic development and early postnatal life, when gene expression is modulated due to the differentiation program. Both, GP-2 mRNA levels and the rate of GP-2 biosynthesis were quantitated and compared to the immunohistochemical localization of this protein in tissue sections. Under all three functional conditions, significant changes could be demonstrated at the level of digestive enzyme gene expression, but no concomitant modulation of GP-2 expression was observed. GP-2 mRNA is absent from the embryonic pancreas and for the first time is expressed after birth with a significant increase during the period of weaning. Furthermore, GP-2 mRNA and protein levels are not modulated by hormonal stimulation, either in the adult pancreas or in AR4-2J cells in culture. Therefore, we conclude that GP-2, in spite of its quantitative contribution to the zymogen granule membrane, is not involved in enzyme protein sorting or granule formation. Alternative functions for GP-2 are discussed.     

Intracellular transport of the major glycoprotein of zymogen granule membranes in the rat pancreas. Demonstration of high turnover at the plasma membrane

The intracellular transport and destination of the major glycoprotein associated with zymogen granule membranes in the pancreas (GP-2) was established. In suspensions of isolated acinar cells from rat pancreas, pulse-chase experiments were performed. The incorporation of the first newly synthesized GP-2 molecules into zymogen granule membranes occurred at about 60 min after beginning of the pulse. We demonstrated by using two different methods that newly made GP-2 reaches the cell surface within the same time span. After 6-8 h chase considerable more newly synthesized GP-2 has reached the cell surface than would be expected on account of secreted newly synthesized zymogens. These observations strongly suggest that at least part of the GP-2 molecules bypass the mature zymogen granule compartment on their way to the plasma membrane. GP-2 is the only protein that appears in discernable quantity in the plasma membrane during 1-4 h after a pulse label. Nevertheless GP-2 comprises only a small percentage of externally 125I-iodinated plasma membrane proteins. We conclude that GP-2 has a high turnover rate at the plasma membrane level. Treatment of the acinar cells with the N-glycosylation inhibitor tunicamycin does not block the intracellular transport of GP-2.     

Membrane detachment and release of the major membrane glycoprotein of secretory granules in rat pancreatic exocrine cells

The major glycoprotein of pancreatic zymogen granule membranes (GP-2) was detected in the medium of acinar cell suspensions from rat pancreas. Its release from the cells was studied in pulse-chase metabolic labeling experiments with radioactive methionine. GP-2 (apparent Mr = 80 000) was found to be processed to a form of slightly lower apparent Mr (75 000) after about 4 h chase. At about the same time this smaller form of GP-2 appeared in the medium. These results are in accordance with earlier findings in vivo. At different chase times acinar cells were extracted with Triton X-114 to separate water-soluble proteins from membrane-associated (hydrophobic) proteins. This experiment showed that GP-2 is slowly converted from a membrane-bound glycoprotein to a soluble glycoprotein after its reduction in apparent molecular mass, causing its detachment from the membrane. Further analysis indicated that the detachment process may occur at the zymogen granule membrane as well as the plasma membrane. Immunocytochemistry on ultrathin cryosections of pancreatic tissue showed that GP-2 is localized on zymogen granule membranes, plasma membranes and in the acinar lumen. Although in much smaller quantities, GP-2 is also present in the granule content. Thus, in summary, GP-2 is synthesized as a true membrane glycoprotein which is gradually processed to a soluble species and is found in the secretion.     

Synexin protein is non-selective in its ability to increase Ca2+-dependent aggregation of biological and artificial membranes

Synexin, a soluble protein which increases the specificity of Ca²⁺ to aggregate isolated bovine chromaffin granules was prepared from bovine adrenal medullary tissue by the method of Creutz, Pazoles and Pollard (J. Biol. Chem. $\text{253}$, 2858–2866, 1978). We also find that synexin increases both the initial rate and final amplitude of Ca²⁺-promoted aggregation of granule membranes. This effect is Ca²⁺-specific. However in contrast to Creutz $\text{et}\text{al}$, we find that synexin also potentiates aggregation of adrenal medulla and liver mitochondria and microsomes as well as phosphatidylserine vesicles. This lack of membrane specificity argues against the suggestion of Creutz $\text{et}\text{al}$ that synexin specifically binds the granule to the plasma membrane prior to exocytosis $\text{in}\text{vivo}$.     

Organisation of the proteins of the chromaffin granule membrane

The organisation of the protein components of bovine chromaffin granules has been investigated by labelling or digesting intact granules or broken membranes with the following reagents: lactoperoxidase/Na¹²⁵I as a reagent for tyrosine residues, $\text{N-}\text{(iodoacetylaminoethyl)-5-naphthylamine-1-sulphonic}$ acid as a reagent for cysteine residues, pronase, and galactose oxidase/KB³H₄. Following treatment, membranes were purified and washed and proteins were examined by polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulphate. Rather more than 60 bands were resolved, of which about 40 were relatively intense and reproducible. The bands were classified according to their molecular weights and sensitivity to reagents. Penetration of the membranes by the reagents was assessed by examination of intragranular proteins.The majority of chromaffin granule membrane polypeptides became labelled when intact granules were treated with impermeant reagents. Eleven were probably protected in the intact granules, reactive sites becoming exposed only on membrane lysis. By contrast, carbohydrate moieties of glycoproteins appear to be exposed only on the matrix side of the membrane. Two proteins were shown to span the membrane, although this is probably an underestimate.