Carboxy-terminal proteolytic processing at a consensus furin cleavage site is a prerequisite event for quail ZPC secretion

In avian species, a glycoprotein homologous to mammalian ZPC is synthesized in the granulosa cells of developing follicles. We have previously reported that the newly synthesized ZPC (proZPC) in granulosa cells is cleaved at a consensus furin cleavage site to generate mature ZPC prior to secretion. In the present study, we examined the effect of the proteolytic cleavage of proZPC on ZPC secretion by using a specific inhibitor of furin endoprotease and site-directed mutagenesis of the furin cleavage site. Western blot analysis demonstrated that the furin inhibitor efficiently blocked both the proteolytic cleavage of proZPC and the subsequent ZPC secretion. A site-directed mutant that possessed a mutated sequence for furin cleavage was not secreted from the cells. The immunocytochemical observations indicated that proZPC produced in the presence of a furin inhibitor or those produced by the site-directed mutant of the furin cleavage site had accumulated in the endoplasmic reticulum. These results indicate that proZPC is proteolytically cleaved at the consensus furin cleavage site with furin-like protease, and the failure of this cleavage results in its accumulation in the endoplasmic reticulum. Therefore, the C-terminal proteolytic processing of proZPC at the consensus furin cleavage site is a prerequisite event for quail ZPC secretion.     

Secretion of egg envelope protein ZPC after C-terminal proteolytic processing in quail granulosa cells.

In avian species, an egg envelope homologous to the mammalian zona pellucida is called the perivitelline membrane. We have previously reported that one of its components, a glycoprotein homologous to mammalian ZPC, is synthesized in the granulosa cells of the quail ovary. In the present study, we investigated the proteolytic cleavage of the newly synthesized ZPC and the secretion of ZPC from the granulosa cells. Western blot analysis of the cell lysates demonstrated that the 43-kDa protein is the precursor of mature ZPC (proZPC), and is converted to the 35-kDa protein before secretion. The accumulation of proZPC in the presence of brefeldin A, and conversion of proZPC to ZPC in the presence of monensin, indicate the possibility that the proteolytic processing of ZPC occurs in the Golgi apparatus. An analysis of amino-acid sequence identified that the C terminus of mature ZPC protein is Phe360, and the N-terminal amino-acid sequence of the proZPC-derived fragment was determined as Asp363. These results suggest that newly synthesized ZPC is cleaved at the consensus furin cleavage site, and the resulting two basic residues at the C terminus are subsequently trimmed off to generate mature ZPC prior to secretion.     

Asparagine-linked oligosaccharide-independent secretion of egg envelope glycoprotein ZPC of the Japanese quail (Coturnix japonica)

In avian species, a glycoprotein homologous to mammalian ZPC is synthesized in the granulosa cells of developing follicles. We have previously reported that the newly synthesized ZPC (proZPC) in the granulosa cells is cleaved at the consensus furin cleavage site to generate mature ZPC prior to secretion. In the present study, we examined the role of asparagine (N)-linked oligosaccharides in the proteolytic processing of proZPC and the subsequent secretion of ZPC by using site-directed mutagenesis of the consensus sequence for N-glycosylation, and tunicamycin, an inhibitor for N-glycosylation of glycoprotein. Western blot analysis demonstrated that tunicamycin did not block either proteolytic cleavage of proZPC or the subsequent ZPC secretion. Moreover, a site-directed mutant that possesses a mutated sequence for N-glycosylation was efficiently secreted from the cells. These results indicate that proteolytic cleavage of proZPC, and the subsequent ZPC secretion occur in the absence of N-linked oligosaccharides. Therefore, the addition of N-glycans to ZPC polypeptide is not required for quail ZPC secretion.     

Novel blockade by brefeldin A of intracellular transport of secretory proteins in cultured rat hepatocytes

We examined the effect of brefeldin A, an antiviral antibiotic, on protein synthesis, intracellular processing, and secretion in primary culture of rat hepatocytes. The secretion was strongly blocked by the drug at 1 microgram/ml and higher concentrations, while the protein synthesis was maintained fairly well. Pulse-chase experiments with [35S]methionine demonstrated that brefeldin A completely blocked the proteolytic conversion of proalbumin to serum albumin up to 60 min of chase, although its conversion was observed as early as 20 min in the control cells. The drug also inhibited the terminal glycosylation of oligosaccharide chains of alpha 1-protease inhibitor and haptoglobin. These two modifications have been shown to occur at the trans region of the Golgi complex. The drug, however, had no effect on the proteolytic processing of the haptoglobin proform which takes place within the endoplasmic reticulum. Such an effect by brefeldin A is very similar with that induced by the carboxylic ionophore monensin. However, in contrast to evidence that monensin causes a delayed secretion of the unprocessed forms of these proteins, brefeldin A allowed the completely processed forms to be secreted after a prolonged accumulation of the unprocessed forms. Morphological observations demonstrated that the endoplasmic reticulum was markedly dilated by treatment with the drug at 10 micrograms/ml which continuously blocked the secretion. On the other hand, brefeldin A caused no inhibitory effect on the endocytic pathway as judged by cellular uptake and degradation of 125I-asialofetuin. These results indicate that brefeldin A is a unique agent which primarily impedes protein transport from the endoplasmic reticulum to the Golgi complex by a mechanism different from those considered for other secretion-blocking agents so far reported.     

Secretion of collagen by insects: Autoradiographic study of l-proline 3H-5 incorporation by the firebrat Thermobia domestica

The biosynthesis of collagenous endoskeletal endosterna by fibroblasts was studied by electron microscope autoradiography after tritium-labelled proline administration in the firebrat Thermobia domestica at various time intervals. Autoradiographs were quantitatively analyzed and the relative concentration of label was determined for various cellular compartments. The labelling in the rough endoplasmic reticulum, ground cytoplasm, Golgi apparatus and endosterna was compared during the secretion of radioactive products. The label, initially located in the rough endoplasmic reticulum was found in the Golgi apparatus before it accumulated in the endosternum. The involvement of the rough endoplasmic reticulum and Golgi apparatus in the transport of secreted collagen is discussed, comparing our results with current knowledge of collagenous secretion.     

A single mutation prevents the normal intracellular transport of multiple lysosomal proteins from the rough endoplasmic reticulum

Dictyostelium discoideum strain HMW-426 has been previously shown to be defective in the proteolytic processing of the lysosomal enzyme precursor to alpha-mannosidase. We have now shown that the mutant is defective in the proteolytic processing of a second lysosomal enzyme, beta-glucosidase. Digestion of the HMW-426 alpha-mannosidase and beta-glucosidase precursors with endoglycosidase H revealed that the majority of oligosaccharide side chains on both precursors were sensitive to cleavage by this enzyme, indicating that both precursors fail to reach the Golgi apparatus. Subcellular fractionation experiments demonstrated that these two mutant precursors accumulated inside the lumen of the rough endoplasmic reticulum. The alpha-mannosidase precursor is conformationally altered, as evidenced by its abnormal protease susceptibility, suggesting that altered conformation is responsible for a generalized defect in transport of lysosomal protein precursors from the rough endoplasmic reticulum in the mutant.     

Pulmonary surfactant phosphatidylcholine transport bypasses the brefeldin A sensitive compartment of alveolar type II cells

Brefeldin A (BFA) causes disassembly of the Golgi apparatus and blocks protein transport to this organelle from the endoplasmic reticulum. However, there still remains considerable ambiguity regarding the involvement of the Golgi apparatus in glycerolipid transport pathways. We examined the effects of BFA upon the intracellular translocation of phosphatidylcholine in alveolar type II cells, that synthesize, transport, store and secrete large amounts of phospholipid for regulated exocytosis. BFA at concentrations as high as 10 microg/ml failed to alter the assembly of phosphatidylcholine into lamellar bodies, the specialized storage organelles for pulmonary surfactant. The same concentration of BFA was also ineffective at altering the secretion of newly synthesized phosphatidylcholine from alveolar type II cells. In contrast, concentrations of the drug of 2.5 microg/ml completely arrested newly synthesized lysozyme secretion from the same cells, indicating that BFA readily blocked protein transport processes in alveolar type II cells. The disassembly of the Golgi apparatus in alveolar type II cells following BFA treatment was also demonstrated by showing the redistribution of the resident Golgi protein MG-160 to the endoplasmic reticulum. These results indicate that intracellular transport of phosphatidylcholine along the secretory pathway in alveolar type II cells proceeds via a BFA insensitive route and does not require a functional Golgi apparatus.     

Proteolytic conversion of hepatitis B virus e antigen precursor to end product occurs in a postendoplasmic reticulum compartment.

At least two proteolytic events are involved in the biogenesis of hepatitis B virus e antigen. The first proteolytic event removes the signal peptide and results in the translocation of the precursor protein, P22, into the lumen of the endoplasmic reticulum (ER). The second proteolytic event removes the carboxy-terminal arginine-rich sequence of P22 and converts it to the 16-kDa hepatitis B virus e antigen end product. In contrast to the first proteolytic event, the second proteolytic event is suppressed by brefeldin A, a chemical that inhibits the transport of protein from the ER to the Golgi apparatus. In subcellular fractionation experiments, P22 was detected in both the ER and the Golgi fractions, but P16 was detected only in the Golgi fraction. On the basis of these results, we conclude that the conversion of P22 to P16 occurs ina post-ER compartment, mostly likely the Golgi apparatus.     

Alkaline phosphatase biosynthesis in the endoplasmic reticulum and its transport through the Golgi apparatus to the plasma membrane: cytochemical evidence

Enzyme induction of HeLa cell placental alkaline phosphatase with various agents such as prednisolone, sodium butyrate, hyperosmolality (NaCl), or combination of these inducers resulted in the appearance of enzyme activity in the rough endoplasmic reticulum, nuclear envelope, Golgi apparatus, and plasma membrane. In the Golgi apparatus, intense reaction product deposits tended to be concentrated on its trans side, with small vesicles and granules also being positively stained. Inhibition of protein synthesis with cycloheximide was followed by the disappearance of enzyme activity from these cytoplasmic organelles but not from the plasma membrane. Treatment with monensin, a secretory protein transport inhibitor, uniformly increased activity in the rough endoplasmic reticulum while causing marked dilatation of the intensely positive Golgi cisternae. These results suggest that intracellular alkaline phosphatase is newly synthesized in the endoplasmic reticulum and then passes en route through the Golgi apparatus to the plasma membrane. Accordingly, the present system could represent the biosynthesis, transport, and incorporation of the model cell surface enzyme protein to add to the vesicular stomatitus virus glyco-1 (VSV-G) protein and acetylcholine receptor model systems for studying the dynamics of cell surface protein genesis, transport, and membrane integration.     

Mutations in the cytoplasmic domain of the influenza virus hemagglutinin affect different stages of intracellular transport

Mutations have been introduced into the cloned DNA sequences coding for influenza virus hemagglutinin (HA), and the resulting mutant genes have been expressed in simian cells by the use of SV40-HA recombinant viral vectors. In this study we analyzed the effect of specific alterations in the cytoplasmic domain of the HA molecule on its rate of biosynthesis and transport, cellular localization, and biological activity. Several of the mutants displayed abnormalities in the pathway of transport from the endoplasmic reticulum to the cell surface. One mutant HA remained within the endoplasmic reticulum; others were delayed in reaching the Golgi apparatus after core glycosylation had been completed in the endoplasmic reticulum, but then progressed at a normal rate from the Golgi apparatus to the cell surface; another was delayed in transport from the Golgi apparatus to the plasma membrane. However, two mutants were indistinguishable from wild-type HA in their rate of movement from the endoplasmic reticulum through the Golgi apparatus to the cell surface. We conclude that changes in the cytoplasmic domain can powerfully influence the rate of intracellular transport and the efficiency with which HA reaches the cell surface. Nevertheless, absolute conservation of this region of the molecule is not required for maturation and efficient expression of a biologically active HA on the surface of infected cells.     

Effects of weakly basic amines on proteolytic processing and terminal glycosylation of secretory proteins in cultured rat hepatocytes.

We examined the effects of weakly basic amines on the secretion and post-translational modifications of secretory proteins in cultured rat hepatocytes. Weakly basic amines such as methylamine, chloroquine and NH4Cl strongly inhibited not only protein secretion, but also the proteolytic conversion of a proform of complement C3, allowing the precursor to be released into the medium. The amines, however, had no effect on the proteolytic conversion of prohaptoglobin into its subunits. Since available evidence indicates that the conversion of pro-C3 occurs at the Golgi complex while that of prohaptoglobin takes place in the endoplasmic reticulum, it is most likely that the weak bases specifically affect the proteolytic event occurring at the Golgi complex. Electron microscopic observations confirmed that the amines caused morphological changes of the Golgi complex, consisting of dilated cisternae and swollen vacuoles. When the glycosylation of alpha 1-protease inhibitor and haptoglobin was examined, it was found that the amines caused a marked accumulation in the cells of both glycoproteins corresponding to the mature secreted forms. Neuraminidase digestion demonstrated that the glycoproteins accumulating in response to the amines had acquired terminal sialic acid. The results indicate that the amines do not significantly affect terminal glycosylation, in contrast with their definite effect on proteolytic processing, despite the fact that both modifications take place in the Golgi complex.     

Organization of transport from endoplasmic reticulum to Golgi in higher plants

In plant cells, the organization of the Golgi apparatus and its interrelationships with the endoplasmic reticulum differ from those in mammalian and yeast cells. Endoplasmic reticulum and Golgi apparatus can now be visualized in plant cells in vivo with green fluorescent protein (GFP) specifically directed to these compartments. This makes it possible to study the dynamics of the membrane transport between these two organelles in the living cells. The GFP approach, in conjunction with a considerable volume of data about proteins participating in the transport between endoplasmic reticulum and Golgi in yeast and mammalian cells and the identification of their putative plant homologues, should allow the establishment of an experimental model in which to test the involvement of the candidate proteins in plants. As a first step towards the development of such a system, we are using Sar1, a small G-protein necessary for vesicle budding from the endoplasmic reticulum. This work has demonstrated that the introduction of Sar1 mutants blocks the transport from endoplasmic reticulum to Golgi in vivo in tobacco leaf epidermal cells and has therefore confirmed the feasibility of this approach to test the function of other proteins that are presumably involved in this step of endomembrane trafficking in plant cells.     

Subcellular fractionation studies on the post-translational processing of pro-adrenocorticotropic hormone/endorphin in rat intermediate pituitary

The subcellular localization of the post-translational processing steps which occur in the conversion of pro-adrenocorticotropic hormone (ACTH)/endorphin into beta-endorphin-sized molecules in rat intermediate pituitary has been studied. Primary cell cultures were incubated in radioactively labeled amino acids, and a subcellular fraction containing secretory granules was separated from a subcellular fraction containing rough endoplasmic reticulum and Golgi apparatus by centrifugation of homogenates on gradients on Percoll (Pharmacia Fine Chemicals). The radiolabeled beta-endorphin-related material in the granule and rough endoplasmic reticulum/Golgi apparatus fractions was quantitated by immunoprecipitation and sodium dodecyl sulfate polyacrylamide gel electrophoresis. A pulse-chase labeling experiment demonstrated that newly synthesized beta-endorphin-related material first appeared in the rough endoplasmic reticulum/Golgi apparatus fraction and after longer incubations (chase) appeared in the secretory granule fraction. After 2 h of chase incubation, about 85% of the beta-endorphin-related material synthesized during the 30-min pulse incubation had been transferred from the rough endoplasmic reticulum/Golgi apparatus to the secretory granule fraction. The conversion of most of the newly synthesized pro-ACTH/endorphin into beta-lipotropin occurred in the rough endoplasmic reticulum/Golgi apparatus fraction, whereas the conversion of most of the beta-lipotropin into beta-endorphin-sized molecules occurred in the secretory granule fraction.     

Biologically active APRIL is secreted following intracellular processing in the Golgi apparatus by furin convertase

Tumor necrosis factor (TNF) ligand family members are synthesized as transmembrane proteins, and cleavage of the membrane-anchored proteins from the cell surface is frequently observed. The TNF-related ligands APRIL and BLyS and their cognate receptors BCMA/TACI form a two ligand/two receptor system that has been shown to participate in B- and T-cell stimulation. In contrast to BLyS, which is known to be cleaved from the cell surface, we found that APRIL is processed intracellularly by furin convertase. Blockage of protein transport from the endoplasmic reticulum to the Golgi apparatus by Brefeldin A treatment abrogated APRIL processing, whereas monensin, an inhibitor of post-Golgi transport, did not interfere with cleavage of APRIL, but blocked secretion of processed APRIL. Thus, APRIL shows a unique maturation pathway among the TNF ligand family members, as it not detectable as a membrane-anchored protein at the cell surface, but is processed in the Golgi apparatus prior to its secretion.     

The manganese cation disrupts membrane dynamics along the secretory pathway

The endoplasmic reticulum and Golgi apparatus play key roles in regulating the folding, assembly, and transport of newly synthesized proteins along the secretory pathway. We find that the divalent cation manganese disrupts the Golgi apparatus and endoplasmic reticulum (ER). The Golgi apparatus is fragmented into smaller dispersed structures upon manganese treatment. Golgi residents, such as TGN46, beta1,4-galactosyltransferase, giantin, and GM130, are still segregated and partitioned correctly into smaller stacked fragments in manganese-treated cells. The mesh-like ER network is substantially affected and peripheral ER elements are collapsed. These effects are consistent with manganese-mediated inhibition of motor proteins that link membrane organelles along the secretory pathway to the cytoskeleton. This divalent cation thus represents a new tool for studying protein secretion and membrane dynamics along the secretory pathway.     

Albumin secreted by rat liver bypasses Golgi apparatus cisternae

Albumin was isolated immunologically from various subcellular fractions from livers of adult male rats receiving an intraperitoneal injection of [³H]leucine to investigate the kinetics and pathway of subcellular transfer of newly synthesized albumin during secretion. At appropriate time intervals, livers were excised and fractionated into endoplasmic reticulum and Golgi apparatus. Golgi apparatus were further subfractionated into cisternae and secretory vesicles. In endoplasmic reticulum fractions, labeled albumin appeared within 7.5 min of injection of isotope, followed by a rapid decline in specific activity. Albumin in Golgi apparatus was labeled and concentrated in secretory vesicles over 25 min. The radioactivity in albumin per mg total protein was highest in secretory vesicles and insignificant in the cisternal fraction. Labeled albumin was present in serum by 30 min and radioactivity in serum albumin reached a plateau within 60–90 min after injection of isotope. Results provide evidence for the migration of albumin from its site of synthesis on endoplasmic reticulum membrane-bound polyribosomes to its site of secretion into the circulation via the Golgi apparatus. The pathway of albumin transport to secretory vesicles is suggested to involve peripheral elemenst of the Golgi apparatus. Secretory vesicle formation and maturation required 20 to 30 min for completion, via a mechanism whereby the inner spaces of the central saccules may be bypassed.     

Variability in transport rates of secretory glycoproteins through the endoplasmic reticulum and Golgi in human hepatoma cells

We have previously shown that newly synthesized liver secretory proteins are exported at three distinct characteristic rates, with intracellular retention half-times of 110-120 min (e.g. transferrin), 75-80 min (e.g. ceruloplasmin), and 30-40 min (e.g. alpha 1-protease inhibitor) (J. B. Parent, H. Bauer, and K. Olden (1985) Biochim. Biophys. Acta, in press). In the present study we have determined the average time required for specific glycoproteins to move through the various compartments of the intracellular transport pathway, consisting of endoplasmic reticulum and Golgi complex. Localization in particular compartments was monitored by the use of the following complementary approaches: (i) Percoll density gradient fractionation of the subcellular organelles, (ii) sensitivity of the glycan moiety of N-linked glycosylation to endo-beta-N-acetylglucosaminidase H, and (iii) by the lectin-binding characteristics. The cell fractionation studies revealed that alpha 1-protease inhibitor, ceruloplasmin, and transferrin were transported from the rough endoplasmic reticulum with a retention half-time of 10, 30, or 45 min, respectively. Measurements of the rate at which newly synthesized glycoprotein became endo H-resistant (an event localized near the medial region of Golgi) demonstrated that it took 60-70, 30, and 18 min for 50% of transferrin, ceruloplasmin, and alpha 1-protease inhibitor, respectively, to reach the medial Golgi. Consistent with this finding, maximal binding of transferrin to wheat germ agglutinin (also a medial Golgi event) and Ricinus communis agglutinin I (a trans Golgi event) required 75 and 90 min, respectively, and maximal binding of ceruloplasmin to both lectins occurred in approximately 30 min. Maximal binding of alpha 1-protease inhibitor to wheat germ agglutinin and Ricinus communis agglutinin I required 15 and 30 min, respectively. The results presented here clearly indicate that (i) the time required for protein secretion cannot be entirely accounted for by lag in transport from the rough endoplasmic reticulum to the Golgi since the glycoproteins examined are retained in the former organelle for no more than two-fifths of the total intracellular retention half-time, and (ii) the variability in rates of protein secretion is not due solely to differences in rates of transport from the rough endoplasmic reticulum to the Golgi as variability in retention within the Golgi is also demonstrated. The results are discussed in terms of their compatibility with receptor-mediated transport of glycoproteins in both the endoplasmic reticulum and Golgi.     

Mutations of the Rous sarcoma virus env gene that affect the transport and subcellular location of the glycoprotein products

The envelope glycoproteins of Rous sarcoma virus (RSV), gp85 and gp37, are anchored in the membrane by a 27-amino acid, hydrophobic domain that lies adjacent to a 22-amino acid, cytoplasmic domain at the carboxy terminus of gp37. We have altered these cytoplasmic and transmembrane domains by introducing deletion mutations into the molecularly cloned sequences of a proviral env gene. The effects of the mutations on the transport and subcellular localization of the Rous sarcoma virus glycoproteins were examined in monkey (CV-1) cells using an SV40 expression vector. We found, on the one hand, that replacement of the nonconserved region of the cytoplasmic domain with a longer, unrelated sequence of amino acids (mutant C1) did not alter the rate of transport to the Golgi apparatus nor the appearance of the glycoprotein on the cell surface. Larger deletions, extending into the conserved region of the cytoplasmic domain (mutant C2), resulted in a slower rate of transport to the Golgi apparatus, but did not prevent transport to the cell surface. On the other hand, removal of the entire cytoplasmic and transmembrane domains (mutant C3) did block transport and therefore did not result in secretion of the truncated protein. Our results demonstrate that the C3 polypeptide was not transported to the Golgi apparatus, although it apparently remained in a soluble, nonanchored form in the lumen of the rough endoplasmic reticulum; therefore, it appears that this mutant protein lacks a functional sorting signal. Surprisingly, subcellular localization by internal immunofluorescence revealed that the C3 protein (unlike the wild type) did not accumulate on the nuclear membrane but rather in vesicles distributed throughout the cytoplasm. This observation suggests that the wild-type glycoproteins (and perhaps other membrane-bound or secreted proteins) are specifically transported to the nuclear membrane after their biosynthesis elsewhere in the rough endoplasmic reticulum.     

Arginine/lysine residues in the cytoplasmic tail promote ER export of plant glycosylation enzymes

Plant N -glycan processing enzymes are arranged along the early secretory pathway, forming an assembly line to facilitate the step-by-step modification of oligosaccharides on glycoproteins. Thus, these enzymes provide excellent tools to study signals and mechanisms, promoting their localization and retention in the endoplasmic reticulum (ER) and Golgi apparatus. Herein, we focused on a detailed investigation of amino acid sequence motifs present in their short cytoplasmic tails in respect to ER export. Using site-directed mutagenesis, we determined that single arginine/lysine residues within the cytoplasmic tail are sufficient to promote rapid Golgi targeting of Golgi-resident N -acetylglucosaminyltransferase I (GnTI) and α-mannosidase II (GMII). Furthermore, we reveal that an intact ER export motif is essential for proper in vivo function of GnTI. Coexpression studies with Sar1p provided evidence for COPII-dependent transport of GnTI to the Golgi. Our data provide evidence that efficient ER export of Golgi-resident plant N -glycan processing enzymes occurs through a selective mechanism based on recognition of single basic amino acids present in their cytoplasmic tails.     

Secretion of killer toxin encoded on the linear DNA plasmid pGKL1 from Saccharomyces cerevisiae

By the kar1-mediated cytoduction, linear double-stranded DNA plasmids pGKL1 and pGKL2, encoding killer toxin complex, have been successfully transferred to the recipient strains with about 30% frequency. The killer toxin was found to be secreted through the normal yeast secretory pathway by introducing pGKL plasmids into the several Saccharomyces cerevisiae sec mutants and examining the secretion of killer toxin. S. cerevisiae cells, harboring newly isolated deletion plasmid pGKL1D, expressed only the 28K protein among three killer subunits, and secreted the 28K subunit at a level of zero to 20% efficiency of the cells containing intact pGKL1 plasmid. These data indicated that subunit interaction (cosecretion) of killer proteins is required for the efficient secretion of 28K subunit. The 28K precursor protein was found to translocate across the canine pancreatic endoplasmic reticulum membrane under the direction of its own signal peptide in vitro without any other subunits. From kex2 mutant cells harboring pGKL1 plasmid, the 97K subunit, and its precursor 128K protein were not secreted, however, the 28K subunit was secreted in the same amount as that secreted from KEX2 cells. These lines of evidence suggest that the final assembly of killer toxin complex after KEX2 site of Golgi apparatus is not essential for the secretion of 28K subunit, and therefore, that putative interaction between 128K protein and 28K subunit for the transport between endoplasmic reticulum and Golgi apparatus may be required for the efficient secretion of 28K subunit.