INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

During mitosis in animal cells, there is a marked inhibition of membrane traffic (13, 20, 67, 68), and the Golgi apparatus fragments into vesiculotubular clusters that are dispersed throughout the metaphase cytoplasm (37-39, 62, 69). These clusters become the template for reassembly of 100 to 200 Golgi stacks which are then partitioned as the telophase daughter cells separate (37). Golgi clusters are also formed when cells are treated with phosphatase inhibitors such as okadaic acid (OA) (35), and their structure is morphologically indistinguishable from that of the Golgi clusters of mitosis. Since OA also induces arrest of the membrane traffic (12, 35), it provides an important tool for the study of the poorly understood process of Golgi cluster formation.

We have proposed a hypothesis to explain the generation of Golgi clusters (34). In this scheme the clusters arise because of an imbalance in membrane traffic through the Golgi organelle, which causes the Golgi cisternae to shrink and form essentially tubular remnants. The shrinkage, we suggest, would stem from continued export (from the trans-Golgi network [TGN] and via recycling to the endoplasmic reticulum [ER]) in the face of arrested import (via inhibition of ER-Golgi membrane traffic and recycling to the TGN from the plasma membrane) (36). This hypothesis is now supported by two principle lines of evidence. The first comes from quantitative electron microscopy (EM) of Golgi clusters in mitotic and OA-treated cells (35-37), showing that cluster formation is accompanied by a dramatic reduction in the amount of identifiable Golgi membrane. The Golgi clusters of mitotic HeLa cells contain only one-quarter of the Golgi membrane found in telophase cells, and those of OA-treated HeLa cells hold only one-half of the membrane found in untreated controls (36, 37). Importantly, major populations of Golgi resident proteins remain within the Golgi clusters so that membrane depletion leads to a two- to-threefold increase in their concentration (36, 60). The second line of evidence comes from studies of Golgi membrane traffic which have revealed that import predominates over export. Thus, import pathways from the rough ER (RER) and from the plasma membrane via endocytosis are significantly inhibited in both mitotic cells and OA-treated cells (12, 13, 24, 35, 67), while export pathways such as those carrying glycosaminoglycans and the TGN marker TGN 38 out of the TGN are much less affected and appear to be active during both mitosis (24) and OA treatment (22).

In the present study we have tested our hypothesis further by studying protein traffic between the RER and Golgi stack during OA treatment. Proteins in the RER reach the Golgi stack via the vesiculotubular intermediate compartment (IC) (16, 17, 26, 53), from which they either continue on the anterograde pathway to the TGN or, if they are RER proteins which have leaked into the IC, are retrieved and returned to the RER. The retrograde route (30, 31) requires the specific signal KDEL (for many soluble RER proteins) (41) or K(X)KXX (for type I membrane RER proteins) (23, 42) and is dependent on the function of COPI-coated vesicles (7, 11, 25, 29, 45, 46). Two closely related lectin-like type I membrane proteins, p58 (54) and its human homolog p53 (ERGIC 53) (56), contain double-lysine ER retrieval signals and appear to recycle from the IC to the ER via a COPI binding mechanism (29, 55, 58, 63, 66). Available evidence also indicates a requirement for COPI coat proteins in anterograde transport from the IC to the Golgi stack and between Golgi cisternae (46, 51).

The exit of proteins from the ER is controlled by a second coat complex, COPII (6, 8). COPII components appear to be involved in selective binding of exported cargo in the ER and its inclusion in COPII-coated buds (8). Recognizable COPII coats are recruited at the transitional element regions of the IC (44, 64) and take up proteins destined for anterograde transport (vesicular stomatitis virus [VSV] G protein) (5, 64) or recycling to the ER (KDEL receptor and p58) (5, 64) while excluding RER proteins such as ribophorin, calnexin, and BIP. Once formed, COPII structures (52) appear to lose their COPII coats and subsequently recruit COPI components which then function in retrograde and anterograde trafficking out from the IC (2, 5, 52).

Biochemical and localization studies have documented arrest of RER-to-Golgi transport during mitosis (13) and OA treatment (12, 35), but the exact location has not been determined. In this study we used quantitative immuno-EM to show that in OA-treated CHO cells, the G protein of Chandipura (CHP) virus (40) is arrested exclusively in the RER during OA treatment. The block in transport is specific for RER-IC transport, since G protein that accumulated in the IC at 15°C could gain access to Golgi stack enzymes when transport was resumed in the presence of OA. We also examined recycling of proteins from the IC to the RER by using subcellular fractionation and quantitative immuno-EM and found that p58 and its human homolog p53 (1, 27, 28, 56, 57) accumulate in the RER during OA treatment. During this accumulation, p58 and p53 can no longer gain access to sec 13p structures, providing evidence that OA induces a very early transport block at or prior to the COPII machinery.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Electron microscopy. Monolayers of CHO cells, grown and infected with CHP virus as previously described for VSV (35), were fixed for 30 min in 0.5% (vol/vol) glutaraldehyde-200 mM PIPES [piperazine-N,N'-bis(2-ethanesulfonic acid)]-KOH, pH 7. Cells were scraped from dishes and centrifuged at 13,000 × g for 15 min. Pellets were cryoprotected by infusion with 2.1 M sucrose and frozen in liquid nitrogen, and ultrathin cryosections were prepared and immunolabelled by using polyclonal antibodies to CHP virus G protein (49) or polyclonal antibodies to p58 (a gift from Jaakko Saraste, University of Oslo, Bergen, Norway) followed by protein A-7-nm gold (36). Double labelling for CHP virus G protein (protein A-12-nm gold) and p58 (protein A-7-nm gold) was carried out as detailed by Prescott et al. (48).

Immunofluorescence. HeLa cells grown to subconfluency on sterile coverslips in 30-mm-diameter six-well plates (Costar, High Wycombe, United Kingdom) were fixed in precooled methanol (20°C) at room temperature for 5 min. After a 5-min blocking step in 0.2% fish skin gelatin-150 mM NaCl-10 mM NaP_i (pH 7.4) (FSG-PBS), the coverslips were incubated in antibodies diluted in 0.2% FSG-PBS. A mouse monoclonal antibody was used to detect human p53 (a gift from Hans-Peter Hauri, Basel, Switzerland) and a rabbit polyclonal antibody for sec 13p (a gift from Wanjin Hong, Singapore). Secondary antibodies conjugated to Texas red and fluorescein isothiocyanate were used to detect p53 and sec 13p, respectively. Observations were made on an MRC-600 confocal laser scanning microscope on single optical sections.

Separation of RER and Golgi membranes. CHO cells were treated with 1 µM OA for 2 h in modified Eagle's medium containing 20 mM HEPES-KOH (pH 7.4) and 10% fetal calf serum at 37°C. One hour into the treatment, the cells were transferred to fresh medium containing 1 µM OA, and during the final 30 min, both the control cells and the OA-treated cells were incubated with 20 µg of cytochalasin B (10) per ml and 10 µg of nocodazole per ml to disrupt the cytoskeleton. Following this treatment, the now loosely adherent cells were detached from the culture dishes and harvested by centrifugation at 500 × g for 2 min at 4°C. The cells (8 × 10⁷) were resuspended and swollen in 10 ml of ice-cold 150 mM KCl-10 mM triethanolamine (pH 7.4) for 10 min and then pelleted and washed twice in 15 ml of ice-cold 150 mM KCl-50 mM HEPES-KOH (pH 7.4)-10 mM EGTA-2 mM MgCl₂ (KHEM) containing 1 mM EDTA, 0.2 mM phenylmethylsulfonyl fluoride, 10 mM sodium fluoride and 1 mM sodium orthovanadate. The final pellet of cells was resuspended in 1 ml of ice-cold KHEM containing protease and phosphatase inhibitors and homogenized by 12 passes through a ball-bearing homogenizer with a 0.016-mm clearance (3, 4). Under these conditions, 95% of the cells were stained with trypan blue. Centrifugation at 1,000 × g for 5 min at 4°C produced a postnuclear supernatant for both OA-treated and interphase cells. Postnuclear supernatants containing equal amounts of protein were loaded onto the tops of step gradients of sucrose (3 ml of 0.8 M sucrose, 4 ml of 1.0 M sucrose, 4 ml of 1.2 M sucrose, and 1.0 ml of 1.6 M sucrose in KHEM containing protease inhibitors and phosphatase inhibitors as described above) and centrifuged for 1 h at 100,000 × g in an SW40Ti rotor (Beckman) at 4°C. Fractions of 2 ml were collected and diluted with KHEM, and the membranes were recovered by centrifugation for 1 h at 200,000 × g in an SW55Ti rotor (Beckman) at 4°C. The membrane pellets were solubilized in 1% (wt/vol) Triton X-100 containing 50 mM HEPES-KOH (pH 7.4), 0.25 M sucrose, 0.2 mM phenylmethylsulfonyl fluoride, and phosphatase inhibitors. Insoluble material was sedimented by centrifugation at 14,000 × g for 5 min at 4°C, and the supernatant was analyzed by immunoblotting and for marker enzyme activities. The distribution of Golgi membrane on the sucrose gradients was assayed by galactosyltransferase activity (9), and the distribution of ER membrane was assayed by KCN-resistant NADH-cytochrome c oxidoreductase activity (61).

Immunoblotting. Immunoblotting with rabbit polyclonal antibodies to p58 and sec 13p was done as described by Pryde (49). Antibodies were detected by enhanced chemiluminescence, and the images were digitized and the pixel density was estimated by using a GDS 7600 system (752 × 582 pixel resolution) and 486 computer with an image acquisition card (UVP Products Ltd., Cambridge, United Kingdom). A Windows control software package (GRAB-IT) from Microsoft Corporation was used to estimate band densities. Estimates were made on a linear range of antigen-antibody binding.

Endoglycosidase H digestion and immunoprecipitation of [³⁵S]methionine-labelled G protein. CHP virus-infected CHO cells (5 × 10⁶ cells/ml) were incubated for 1 h in RPMI 1640 medium (ICN Biomedicals Ltd., High Wycombe, United Kingdom) lacking methionine and cysteine and supplemented with 2% (vol/vol) dialyzed fetal calf serum, 25 mM HEPES-KOH (pH 7.4) and were labelled with 100 µCi of [³⁵S]methionine (Tran³⁵S-label, 1,000 Ci/mmol; ICN Biomedicals Ltd., Thame, United Kingdom) per ml. G protein was extracted and treated with endoglycosidase H (35) and immunoprecipitated with a polyclonal antibody to CHP virus G protein as previously described (49).

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

G protein accumulates in the RER during OA treatment. To establish the site at which OA arrests protein transport, we used quantitative immuno-EM of CHP virus G protein as a marker for anterograde transport. Unlike VSV, which due to safety regulations cannot be used in our laboratories, CHP virus does not have a temperature-sensitive mutant in which the G protein is localized to the RER. Thus, to synchronize G protein transport from the RER, the secretory pathway was cleared of transported G protein by incubating virally infected cells with cycloheximide for 1 h to prevent translation of new viral G protein (14). In control cells both the RER and Golgi were heavily labelled for G protein (data not shown), but after incubation for 1 h in cycloheximide, both the RER (Fig. 1A) and the Golgi (Fig. 1B) were unlabelled, even though there was a high expression of virus at the cell surface. To test the effects of 1 µM OA, the phosphatase inhibitor was added during the second hour of a 2-h incubation with cycloheximide, and protein synthesis was then resumed in the presence of OA for 1 to 4 h by washing out the cycloheximide. Under these conditions, the RER was intensively labelled with gold particles (Fig. 1C).

View larger version (135K): [in this window] [in a new window]   FIG. 1.   Immunogold localization of G protein. CHO cells infected with CHP virus either were incubated in medium containing 10 µg of cycloheximide for 1 h and then fixed (A and B) or were incubated for a further 5 h either in 1 µM OA (C) or at 15°C (D, E, and F) with removal of cycloheximide after 1 h of treatment in order to reinitiate protein synthesis. In the presence of OA, G protein accumulated in the RER (arrows) and at 15°C in groups of vesiclulotubular profiles (arrows in panels D and E) as well as in the RER and nuclear envelope (arrowheads in panel F). Arrowheads in panels A, B, and C indicate the plasma membrane. g, Golgi complex; n, nucleus. Bars, 100 nm.

View larger version (19K): [in this window] [in a new window]   FIG. 2.   Quantitation of gold labelling for G protein over the RER, vesiculotubular structures, and Golgi stack after treatment at 15°C or with OA. CHO cells were infected for 4 h and treated with cycloheximide for 1 h. Some cells were fixed, while others were treated at 15°C or with 1 µM OA, initially for 1 h in the presence of cycloheximide and then for a further 4 h in its absence. Cryosections were immunogold labelled for G protein, and the number of gold particles associated with membrane-bound organelles was quantified. After 1 h of cycloheximide treatment, low levels of labelling were observed. However, when synthesis was reinitiated at 15°C, accumulation of gold label occurred both in the RER and in structures with vesicular and tubular form, mostly in close association with the Golgi stacks. In the presence of OA, the accumulation within the RER was more marked, but labelling in vesiculotubular structures could be detected only at levels below those seen after 1 h of cycloheximide treatment. Gold labelling is expressed as particles per 1,000 µm3, and standard errors were calculated by using gold counts per profile (cycloheximide, n = 38; 15°C, n = 28; OA, n = 25).

G protein accumulated at 15°C gains access to Golgi stack enzymes in the presence of OA. Our data were consistent with an OA-induced transport block between the RER and IC, and we next tested whether export of CHP virus G protein from the IC was also inhibited by OA. Our strategy was to accumulate G protein in the IC at 15°C and test its ability to gain access to Golgi stack enzymes when the cells were warmed to 37°C in the presence of OA. Access to the Golgi stack enzymes was measured by resistance of N-linked oligosaccharides to endoglycosidase H digestion, which is a consequence of the action of GlcNAc-transferase I located in the medial cisternae of the Golgi. In control experiments, G protein pulse-labelled for 10 min with [³⁵S]methionine (Fig. 3A, lanes 1 and 2) became resistant to endoglycosidase H after a 1-h chase at 37°C (Fig. 3A, lanes 3 and 4). Preincubation of the cells at 37°C with OA (Fig. 3A, lanes 5 and 6) or at 15°C with (Fig. 3A, lanes 7 and 8) or without (Fig. 3A, lanes 9 and 10) OA prevented the development of this endoglycosidase H resistance, reflecting the inhibition of transport already documented by immuno-EM (Fig. 2). Crucially, when pulse-labelling at 15°C was followed by OA treatment at 15°C and the cells were warmed to 37°C, in the continued presence of OA, the G protein acquired endoglycosidase H resistance (Fig. 3A, lanes 13 and 14), indicating that OA could not prevent G protein, located in the intermediate compartment, from accessing the medial-Golgi enzymes. This most likely indicates that forward transport from the IC continues in the presence of OA, but we were concerned to rule out other explanations such as reduced effectiveness of OA at 15°C or substantial recycling of medial-Golgi enzymes into the membranes of the RER-IC. As shown in Fig. 3B, when G protein was pulse-labelled following OA treatment at 15°C and then chased at 15 or 37°C, the G protein remained sensitive to endoglycosidase H digestion, showing that OA was fully effective at the lower temperature and that recycling of functional medial-Golgi enzymes to the RER was unlikely to have occurred.

View larger version (42K): [in this window] [in a new window]   FIG. 3.   Transport of G protein to the Golgi from the IC is not inhibited by OA. (A) CHO cells were pulsed for 10 min (lanes 1 and 2) with [35S]methionine at 37°C and then chased in the presence of cycloheximide (lanes 3 and 4) for 1 h. G protein was extracted into Triton X-114; half of each sample was treated with endoglycosidase H (Endo H) and then immunoprecipitated with antibody to G protein. The immunoprecipitated proteins were solubilized in sample buffer containing 1 mM dithiothreitol. After treatment with 10 mM iodoacetamide, the G protein was resolved on a 10% (wt/vol) polyacrylamide gel. Cells were also preincubated with 1 µM OA for 1 h at 37°C before being radiolabelled for a further 1 h (lanes 5 and 6). Cells were held at 15°C for 1 h, radiolabelled for 1 h, then incubated for a further 1 h in the presence (lanes 7 and 8) or absence (lanes 9 and 10) of 1 µM OA and 10 µg of cycloheximide per ml at 15°C. When moved to 37°C, the G protein of cells incubated in the presence (lanes 13 and 14) or absence (lanes 11 and 12) of OA became resistant to endoglycosidase H. (B) Cells were held at 15°C for 1 h and then for a further 1 h in the presence of 1 µM OA before being radiolabelled with [35S]methionine for 1 h (lanes 1 and 2). The radiolabelled G protein was chased at 15°C (lanes 3 and 4) or 37°C (lanes 5 and 6) for 1 h in the presence of 10 µg of cycloheximide per ml.

P58 accumulation in the RER during OA treatment. We predicted that membrane depletion of the early secretory pathway would be induced if continued recycling occurred during inhibition of export from the RER. To address this issue, we studied the location of the membrane protein p58, which previous work had assigned to both the IC and the RER (55). In cryosections of control cells (Fig. 4A), we found that 56% of immunolabelling for p58 was localized to the RER and 42% was localized to vesiculotubular profiles (Fig. 5). Double labelling showed that vesiculotubular structures positive for p58 were also labelled for G protein in CHP virus-infected cells incubated at 37 or 15°C (Fig. 4B), indicating that, like IC structures, they formed part of the secretory pathway (33, 47). It is important to point out that although the immunolabelling for p58 appears sparse, this is primarily due to dispersion of the label over an extensive RER compartment. Expressed on a cellular basis, the amount of labelling for p58 over the RER is actually more than 200 particles per cell, with large vesicular structures of greater than 100 nm in diameter, similar to those found to contain G protein (see above), containing less than 1% of the label. Small isolated vesicular profiles that might represent dispersed IC contained only 1.2% of the total labelling.

View larger version (98K): [in this window] [in a new window]   FIG. 4.   Immunolocalization of p58 on thawed cryosections. In control cells (A) labelling for p58 was located mainly over vesiculotubular clusters situated close to the Golgi stack. When synthesis of G protein was reinitiated during incubation at 15°C (B), labelling for G protein (large gold particles) accumulated in clusters of tubules that were also labelled for p58 (smaller gold particles marked with arrows). Incubation of cells in 1 µM OA induced a increase in p58 labelling over the RER (C); quantitative analysis showed this to be a large increase. Bars, 100 nm.

View larger version (21K): [in this window] [in a new window]   FIG. 5.   Quantitation of labelling for p58 over the membrane-bound structures. Cell profiles were systematically scanned at a magnification of ×20,000, and gold particles were assigned to different categories of structure. The numerical density of gold labelling in a cell volume was calculated by using estimates of the area of cell examined and the section thickness (see Materials and Methods). In control cells approximately half of the labelling is located over Golgi stack and tubules, and the other half is located over cisternae of the RER. On treatment with OA for 3 h in the presence of cycloheximide, the labelling over the RER increases approximately fourfold and the labelling over the Golgi decreases by half. Very little labelling was found over either small vesicles (<100 nm) or larger membrane-bound structures.

View larger version (27K): [in this window] [in a new window]   FIG. 6.   CHO cells were incubated for 1 h in 50 µM AlF4 (AlF) (applied as described by Orci et al. [43]), and p58 accumulated in the RER. AlF4 incubation was continued for a further 1 h in the presence of 1 µM OA, and there was no change in the distribution of p58 labelling. OA induced an accumulation of p58 labelling in the RER, but further incubation in AlF4 failed to modify this accumulation.

Quantitation of p58 movement between Golgi-enriched and RER-enriched membrane fractions. To monitor the movement of p58 on sucrose gradients, a clear separation of Golgi and RER membrane was required, and this was achieved only when CHO cells were treated with cytochalasin B and nocodazole to disrupt microfilaments and microtubules, respectively (10, 31). On these gradients the distribution of RER and Golgi membranes was assessed by using assays for NADH-cytochrome c oxidoreductase (61) and galactosyltransferase, respectively. Galactosyltransferase is a trans-Golgi marker (50) and was used because it does not recycle during mitosis or during OA treatment (35, 36). The use of earlier cis/medial-Golgi markers such as mannosidase II (65) and also of IC markers other than p58 was precluded because of the possibility that they may recycle to the RER under these conditions.

View larger version (23K): [in this window] [in a new window]   FIG. 7.   Separation of Golgi and RER membranes on a sucrose gradient. Postnuclear supernatants from control cells and OA-treated cells (treated with cytochalasin B and nocodazole before homogenization) were fractionated on sucrose gradients containing steps of 0.8, 1.0, 1.2, and 1.6 M sucrose. Membranes were recovered from the fractions and assayed for galactosyltransferase (Gal T) activity (Golgi marker) and NADH-cytochrome c oxidoreductase (Cyt red) activity (RER marker). The activities are expressed as a percentage of the activity of postnuclear supernatant loaded onto the gradients. Data are expressed as the means from three experiments, and bars indicate standard errors.

View larger version (21K): [in this window] [in a new window]   FIG. 8.   Immunoblotting of sucrose gradient fractions for p58. Gradient fractions 1 to 6 are from cells treated with OA (+) or untreated (). After treatment with OA, there is a shift in the distribution of p58 from the galactosyltransferase fractions (1 to 3) into the fractions (4 to 6) enriched in NADH-cytochrome c oxidoreductase.

View larger version (18K): [in this window] [in a new window]   FIG. 9.   Quantitation of immunoblot staining shown in Fig. 8. Values represent the means from three experiments, and bars indicate standard errors.

Segregation of p53 and p58 from sec 13p during OA treatment. In order to determine whether the p58 arrest in the RER was related to COPII-containing structures, immunofluorescence microscopy and cell fractionation were used to compare the distribution of p58 with that of the COPII coat component sec 13p (59, 64). For immunofluorescence experiments we used a monoclonal antibody raised against p53, the human homolog of p58 in HeLa cells. This allowed double labelling with rabbit anti-sec 13p antibodies. In untreated HeLa cells, confocal fluorescence images showed very similar distributions for p53 and sec 13p, with strong fluorescence in the perinuclear region as well as in numerous peripheral punctate structures (Fig. 10A and B). The presence of diffuse cytoplasmic fluorescence and nuclear envelope staining for p53 was in accordance with the ER localization demonstrated for p58 at the EM level and by subcellular fractionation (see above).

View larger version (118K): [in this window] [in a new window]   FIG. 10.   Immunofluorescence microscopy of HeLa cells. In untreated interphase cells, sec 13p (A) shows a juxtanuclear Golgi-like distribution with additional punctate structures in the cell periphery. In the same cells, p53 (B) colocalizes with juxtanuclear sec 13p and is also present in punctate structures of the cell periphery. There is clear evidence for a reticular RER-like pattern with weak staining of the nuclear envelope (arrowhead in panel B). In OA-treated cells, the juxtanuclear location of sec 13p (C) is lost, and sec 13p is present mainly in punctate structures situated throughout the cell (arrows). In contrast, p53 (D) is mainly present in the nuclear envelope (arrowhead in panel D) and in a diffuse RER-like distribution. Little p53 staining colocalizes to sec 13p-positive punctate structures. Bars, 25 µm.

View larger version (15K): [in this window] [in a new window]   FIG. 11.   Immunoblotting of sec 13p on sucrose gradient fractions from untreated cells (A) and OA-treated cells (B). The majority of labelling remains in the top fraction after OA treatment, in contrast to the shift of p58 labelling into denser fractions (Fig. 8). Equal amounts of protein were loaded on each fraction.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Our study provides evidence that OA induces selective arrest of protein transport at the membrane of the RER. First, our quantitative immuno-EM showed that newly synthesised CHP virus G protein accumulates exclusively in the ER, and an extensive qualitative search for other membrane-bound structures containing accumulations of G protein was unsuccessful. In fact, we could find only a very small proportion of immunolabelling in structures composed of tubular or vesicular profiles which might have represented the IC structures in which we observed G protein accumulation on incubation at 15°C. Thus, although previous low-resolution immunofluorescence studies had suggested an RER-like localization of VSV G protein in OA-treated cells, our EM studies now effectively rule out significant accumulations of G protein in post-RER structures.

OA also induced accumulation of p58 and its human homolog p53 in the RER, which we documented by using quantitative immuno-EM and immunofluorescence. The accumulation was accompanied by a reduction in the amount of detectable p58 and p53 over identifiable IC structures, indicating a net translocation via a recycling pathway. To document this further, we successfully separated Golgi-related and RER membranes and showed that p58 disappeared from fractions rich in Golgi resident proteins and appeared in those fractions containing RER markers. To examine whether the accumulation was due to accelerated recycling, rather than a block in forward transport, we used aluminum fluoride to inhibit retrograde traffic. Aluminium fluoride strongly inhibited p58 accumulation in the RER, and we therefore conclude that arrested entry of p58 into the IC was the principal factor in producing p58 accumulation.

To further localize the site of the transport arrest, we compared the localization of p58 with that of the COPII coat component sec 13p. In other studies sec 13p and other COPII components have consistently been localized to non-RER structures of the IC region which also contain exported proteins that either are en route for the Golgi stack (VSV G) or are destined for retrograde transport to the RER (p58 and the KDEL receptor [5, 64]). Our data are consistent with these results and show that in untreated interphase HeLa cells, sec 13p colocalizes with p58 and p53 in juxtanuclear and punctate structures. At the EM level these structures correspond to the tubulovesicular IC regions which label for p58 and accumulate G protein at 15°C. Thus, our data indicate that recruitment and budding of COPII occur in the IC region to form RER export complexes which contain exported p58. Significantly, when OA was applied to the cells, sec 13p failed to redistribute to the RER and was no longer associated with p58, which itself had accumulated in the RER. These data therefore suggested that p58 had been prevented from entering COPII-coated membrane structures of the IC, but it was important to rule out the possibility that COPII components had, during OA treatment, dissociated from IC membranes and accumulated at some other site. Analyses of the quantity and distribution of sec 13p by using gradient centrifugation showed that after OA treatment, COPII components remained associated with membranes that were distinct from the RER and were of similar density to the lightest Golgi membranes. These results are therefore consistent with the notion that during OA action COPII remains associated with cognate membrane structures which can no longer import p58.

The final line of evidence for OA action at the RER membrane came from a morphological analysis of RER exit sites. At these sites, transitional element tubules emerge from a smooth ribosome-free area of the RER membrane, and it is these groups of tubules which appear to be coated with COPII components (47). Our data demonstrate that the number of these structures is reduced to less than 10% of control levels, indicating a defect in the transport machinery at the level of the earliest identifiable structures involved in the RER-IC transport.

What are the possible mechanisms of the RER-IC transport block induced by OA? One idea is that exit from the RER is regulated by a quality control mechanism which is sensitive to the oligomerization state of newly synthesized proteins. For technical reasons we were unable to assess the oligomerization of CHP virus G protein, but we successfully assayed the oligomerization of p58, which is known to dimerize in the RER (19). Using native polyacrylamide gel separation followed by immunoblotting, we found that nearly all detectable p58 is present as a dimer in both untreated and OA-treated CHO cells (unpublished observations), indicating that a defect in oligomerization cannot explain the arrested transport of p58. Such a conclusion was supported by the observation of Hammond and Helenius (18) that even improperly folded VSV G protein can pass the junction between RER and IC.

Another possibility is that OA modulates the function of the COPII structures which sort and concentrate cargo during export from the RER (52). COPII coats undergo a cycle of recruitment and dissociation on membranes of the IC region, and it is therefore conceivable that OA modulates this process. Our gradient subfractionation revealed that OA had no effect on the steady-state amount of sec 13p present in the Golgi fraction, which makes selective inhibition of assembly or activation of disassembly by OA unlikely. However, the data are consistent with a defect in uncoating reactions, although this alone does not explain the observed paucity of p58 in these structures. In the future it will be important to examine the dynamics of COPII assembly and disassembly in OA treated cells by using appropriate assays.

Our hypothesis of Golgi cluster formation predicts an imbalance in protein traffic which at the level of the IC would stem from reduced import coupled to continued export. The data presented in this report now provide additional support for this model (Fig. 12). An early block in RER-IC trafficking at the RER membrane would restrict import and helps to explain the dramatic decrease in the amount of Golgi membrane during OA treatment that we reported previously (36). If the RER-Golgi block was situated at a more distal site, then membrane accumulation would result, as occurs when transport out of the IC is arrested by incubation at 15°C.

View larger version (13K): [in this window] [in a new window]   FIG. 12.   Model of OA effects on the RER-IC region. OA induces transport arrest at the exit sites of the ER, preventing entry into COPII-coated structures. Export from the IC to the Golgi stack and recycling from the IC continue (arrows). Recycled proteins such as p58 and p53 then accumulate in the RER membrane. The dotted connection between the COPII- and COPI-coated structures reflects current evidence indicating COPII is recruited prior to COPI during transport or maturation of the post-RER-IC membranes (52).

One pathway for continued export from the IC was the recycling of p58 and p53 from the IC to the RER, which was observed by using quantitative immuno-EM (p58), immunofluorescence (p53), and cellular fractionation experiments (p58). p58 and p53 contain RER retrieval signals (28) and bind to COPI components (66) known to be required for retrieval of ER proteins (29). It is therefore likely that p58 recycles via a natural occurring mechanism rather than by one artifactually induced by OA, and this is supported by the lack of recycling of other Golgi-related proteins such as sec 13p (this work) and galactosyltransferase (36). If COPI-dependent recycling is unaffected by OA action, then this is entirely consistent with our previous data showing that the dynamics of COPI vesicle assembly on Golgi membranes are unaffected during OA-induced Golgi cluster formation in HeLa cells (36).

Another pathway for continued export was transport from the IC to the Golgi stack. We observed that when pulse-labelled G protein was accumulated in the IC at 15°C and transport was allowed to resume by warming in the presence of OA, the G protein acquired resistance to digestion by endoglycosidase H, indicating that the G protein had been transported into the Golgi stack. An alternative explanation for these results was that exit from the IC was actually blocked and that medial-Golgi enzymes had recycled from the Golgi stack to the IC. A recent study has demonstrated that medial-Golgi enzymes can recycle to the RER-IC. Crucially, however, the same study showed that recycling is actually blocked in OA-treated cells (21). This is also supported by our own unpublished study of CHP virus G protein N-linked oligosaccharides, which has revealed no evidence for further processing of oligosaccharides by stack enzymes, such as mannosidase I, which trims Man9-6GlcNAc2Asn to Man5GlcNAc2Asn (data not shown). It will now be important to document the extent of the anterograde movement of G protein within the Golgi stack, especially in view of the reported inhibition of protein transport between the cis- and medial-Golgi cisternae by the protein phosphatase inhibitor microcystin (12).

In summary, we have established that OA-induced arrest of RER-IC protein traffic occurs at the RER membrane and prevents access to COPII structures, an effect which when combined with continued export from the IC provides us with additional insight into how the Golgi clusters of mitotic and OA-treated cells form.