Induction of direct endosome to endoplasmic reticulum transport in Chinese hamster ovary (CHO) cells (LdlF) with a temperature-sensitive defect in epsilon-coatomer protein (epsilon-COP).

In the present study we demonstrate that ricin, apparently without passing through the Golgi apparatus, reaches the endoplasmic reticulum (ER) and intoxicates cells in which the Golgi apparatus has been vesiculated by depletion of epsilon-COP, a subunit of COPI. LdlF cells contain a temperature-sensitive mutation in epsilon-COP. At the nonpermissive temperature epsilon-COP is degraded, and the Golgi apparatus undergoes a morphological change. To study ricin transport in these cells we used ricin sulf-2, a modified ricin molecule containing glycosylation and sulfation sites. Measurements of the incorporation of radioactive mannose into ricin sulf-2 showed that ricin reached the ER in cells depleted of epsilon-COP. Importantly, by investigating the glycosylation of ricin sulf-2 that was modified with radioactive sulfate in the trans-Golgi network, it was demonstrated that transport of ricin to the ER via the Golgi apparatus was severely inhibited. Moreover, we found that ricin was able to intoxicate ldlF cells depleted of epsilon-COP in the presence of brefeldin A. In contrast, control cells were completely protected against ricin by brefeldin A. In conclusion, our results suggest that in ldlF cells depleted of epsilon-COP ricin might be transported to the ER by an induced brefeldin A-resistant pathway that circumvents the Golgi apparatus.

In the present study we demonstrate that ricin, apparently without passing through the Golgi apparatus, reaches the endoplasmic reticulum (ER) and intoxicates cells in which the Golgi apparatus has been vesiculated by depletion of ⑀-COP, a subunit of COPI. LdlF cells contain a temperature-sensitive mutation in ⑀-COP. At the nonpermissive temperature ⑀-COP is degraded, and the Golgi apparatus undergoes a morphological change.
To study ricin transport in these cells we used ricin sulf-2, a modified ricin molecule containing glycosylation and sulfation sites. Measurements of the incorporation of radioactive mannose into ricin sulf-2 showed that ricin reached the ER in cells depleted of ⑀-COP. Importantly, by investigating the glycosylation of ricin sulf-2 that was modified with radioactive sulfate in the trans-Golgi network, it was demonstrated that transport of ricin to the ER via the Golgi apparatus was severely inhibited. Moreover, we found that ricin was able to intoxicate ldlF cells depleted of ⑀-COP in the presence of brefeldin A. In contrast, control cells were completely protected against ricin by brefeldin A. In conclusion, our results suggest that in ldlF cells depleted of ⑀-COP ricin might be transported to the ER by an induced brefeldin A-resistant pathway that circumvents the Golgi apparatus.
It is well established that a number of protein toxins reach the cytosol and exert their toxic effects after retrograde transport from the cell surface to the Golgi apparatus and to the endoplasmic reticulum (ER) 1 (for review see Ref. 1). The plant toxin ricin consists of two polypeptide chains (A and B) linked by a disulfide bond. Ricin binds to both glycolipids and glycoproteins with terminal galactose via its B-chain and can therefore serve as a membrane marker. From the endosomal compartment ricin is partly recycled to the cell surface, partly degraded in the lysosomes, and partly transported to the Golgi apparatus and the ER (1,2). From the ER ricin is translocated to the cytosol where the A-chain inhibits protein synthesis. Coat proteins regulate membrane traffic and may be involved in one or several steps in ricin trafficking (3,4). In this study we have investigated the role of an intact COPI coat for ricin trafficking. The COPI coat has been reported to be involved in transport between the Golgi cisternae, the Golgi apparatus and the ER, early and late endosomes, and the endosomal compartment and the Golgi apparatus (5)(6)(7)(8)(9)(10)(11). One way to interfere with coat assembly is to use the fungal metabolite brefeldin A (BFA), which inhibits the binding of COPI to membranes (12). As a consequence of this inhibition, the Golgi stacks disassemble and redistribute into the ER (13,14). Endocytosis of ricin and its transport to the trans-Golgi network (TGN) is unchanged in the presence of BFA (15). However, BFA protects against ricin in cells with a BFA-sensitive Golgi apparatus (15,16). The protective effect of BFA against ricin toxicity might be due to a requirement of COPI for the transport of ricin from the TGN to the ER, or it might be caused by the redistribution of the Golgi stacks into the ER. Another way to investigate the involvement of COPI in trafficking is to use the CHO-derived mutant cell line ldlF (17). This cell line contains a temperature-sensitive mutation in the COPI subunit ⑀-COP that causes its degradation at the nonpermissive temperature (18). Following incubation at the nonpermissive temperature, the ER to Golgi transport (17) and the early to late endosome transport (9, 10) are impaired. In addition, the Golgi apparatus is disrupted and the structure of the early endosomes is changed (9,19). It was recently reported that ricin can intoxicate ldlF cells depleted of ⑀-COP, and it was suggested that intoxication takes place after retrograde transport of ricin from the Golgi apparatus to the ER and to the cytosol (20). To investigate the various steps of ricin transport in cells depleted of ⑀-COP, we have measured the incorporation of radioactive sulfate and radioactive mannose into a modified ricin molecule containing sulfation and glycosylation sites in ldlF cells lacking ⑀-COP. Interestingly, our results suggest that ricin is transported to the ER from the endosomal compartment by a novel mechanism bypassing the Golgi apparatus.  (21) to a specific activity of 30,000 -40,000 cpm/ng.

Materials-Ricin
Cells and Cell Culture-Wild type CHO and the temperature-sensitive mutant ldlF cells were kindly provided by Dr. Monty Krieger (Massachusetts Institute of Technology, Cambridge, MA) and cultured as previously described (19). The cells were maintained in Ham's F-12 medium (Bio-Whittaker, Verviers, Belgium) supplemented with 5% fetal bovine serum, 100 units/ml penicillin, 100 g/ml streptomycin, 2 mM L-glutamine at 34°C (permissive temperature), and shifted or not to 40°C (nonpermissive temperature) overnight before running the experiments.
Measurements of Endocytosis and Degradation-Endocytosed ricin was measured as the amount of 125 I-labeled toxin that could not be removed by incubating the cells with a 0.1 M lactose solution for 5 min at 37°C. The degradation of 125 I-labeled ricin was measured as the amount of radioactivity that could not be precipitated by trichloroacetic acid after 4 h (22).
Measurements of Protein Synthesis-The cells were incubated in medium without leucine in the presence of 2 Ci/ml [ 3 H]leucine for 20 min at 37°C. The medium was then removed and the cells were washed twice with 5% (w/v) trichloroacetic acid for 10 min. Finally, the cells were solubilized in KOH (0.1 M) and the acid-precipitable radioactivity was measured. The results are expressed in percent of [ 3 H]leucine incorporated in cells incubated without toxin. Deviations between duplicates did not vary by more than 10%.
Sulfation and Mannosylation of Ricin Sulf-2-Ricin A-chain sulf-2 containing a sulfation site and three partially overlapping N-glycosylation sites in the carboxyl terminus was produced, purified, and reconstituted with ricin B-chain to form ricin sulf-2 as previously described (23). The cells were incubated with 600 Ci/ml Na 35 SO 4 in DMEM without sulfate or with 70 Ci/ml [ 3 H]mannose in DMEM without glucose (Invitrogen). 30 min later ricin sulf-2 (ϳ500 ng/ml) was added, and the incubation was continued for 6 h at 37°C or for 4 h and then 4 h in the absence of ricin and radioactive sulfate. The cells were then washed with a 0.1 M lactose solution at 37°C and once with cold phosphate-buffered saline, lyzed (lysis buffer: 0.1 M NaCl, 10 mM Na 2 HPO 4 , 1 mM EDTA, 1% Triton X-100, pH 7.4) in the presence of a protease inhibitor mixture (Roche Diagnostics), and centrifuged to remove the nuclei for 10 min at 5000 rpm in an Eppendorf centrifuge (model 5415). The supernatant was immunoprecipitated with rabbit anti-ricin antibodies immobilized on protein A-Sepharose CL-4B (Amersham Biosciences). Finally, the beads were washed with cold phosphate-buffered saline containing 0.35% Triton X-100, and the adsorbed material was analyzed by SDS-PAGE (12%) under reducing conditions. The radioactivity in the cell lysates was measured to detect differences in the total amount of isotope incorporated in the different conditions. SDS-PAGE-SDS-PAGE was done as described by Laemmli (24). The gels were fixed in 4% acetic acid and 27% methanol for 30 min and then treated with 1 M sodium-salicylate (pH 5.8) in 2% glycerol for 15 min. Dried gels were exposed to Kodak XAR-5 films (Rochester, NY) at Ϫ80°C or to phosphor screens (Amersham Biosciences).

Ricin Intoxicates Cells Lacking ⑀-COP by a BFA-resistant
Mechanism-Intoxication with ricin normally requires ricin endocytosis and transport from the endosomes to the Golgi apparatus and then to the ER, from where the toxin is translocated to the cytosol and then inhibits protein synthesis (1,2). In ldlF cells incubated at the nonpermissive temperature (40°C) ⑀-COP is degraded and COPI function is impaired. It has previously been shown (9, 10) that in these cells endocytosis continues but that the transport from late endosomes to lysosomes is impaired. In agreement with these results, the endocytosis of ricin continues in ldlF cells at the nonpermissive temperature (data not shown), but ricin degradation is severely reduced (Fig. 1).
As a consequence of impaired COPI function, the Golgi apparatus is partially redistributed into an ER-like structure and partially into patch-like structures (19,25). Importantly, as in many other cell lines (13,14), BFA causes the redistribution of the Golgi into the ER in ldlF cells depleted of ⑀-COP (25). To investigate whether ricin was able to reach the cytosol and intoxicate ldlF cells at the nonpermissive temperature both in the absence and in the presence of BFA, we measured the ability of ricin to inhibit protein synthesis both in the absence and in the presence of this drug. Interestingly, as shown in Fig.  2A, ricin was able to intoxicate ldlF cells grown at the nonpermissive temperature (40°C) in the presence of BFA, although at a lower extent compared with untreated cells. However, BFA severely inhibited the intoxication of ricin in CHO cells grown at 40°C (Fig. 2B). Because the delivery of ricin to lysosomes is strongly inhibited in ldlF cells grown at the nonpermissive temperature (Fig. 1), a toxicity experiment was also performed in the presence of bafilomycin A 1 , which will inhibit transport to lysosomes and degradation. Importantly, also in the presence of bafilomycin A 1 , ldlF cells and CHO cells grown at the nonpermissive temperature respond differently to BFA (Fig. 2, C and D), thus suggesting that the fact that ricin can intoxicate ldlF cells grown at the nonpermissive temperature in the presence of BFA is not due to the inhibition of ricin degradation and the higher intracellular amount of ricin that these cells may have. It should also be noticed that a similar concentration of ricin was required to intoxicate ldlF cells and CHO cells in the presence of bafilomycin A 1 (Fig. 2, C versus D, without BFA), thus suggesting that the differences in ricin toxicity between ldlF cells and CHO cells (Fig. 2, A versus B, without BFA) may be due to the larger amount of ricin that is degraded in CHO cells. Finally, in some experiments the effect of BFA on ricin toxicity in ldlF cells grown overnight at the permissive (34°C) or at the nonpermissive temperature (40°C) was compared. Because many transport steps are temperature-sensitive these experiments were performed at 37°C after testing that COPI remains unfunctional at the end of the experiment in ldlF cells incubated overnight at 40°C (data not shown). As shown in Fig. 2, E and F, in the presence of BFA ricin intoxicates ldlF cells lacking ⑀-COP (preincubated at 40°C), but not control ldlF cells (preincubated at 34°C). In conclusion, these experiments suggest that in ldlF cells lacking ⑀-COP ricin may reach the cytosol bypassing the Golgi apparatus, and that neither incubation of cells (CHO) at 40°C nor inhibition of lysosomal degradation are sufficient to induce a BFA-resistant intoxication pathway.
Ricin Is Transported to the ER in LdlF Cells Lacking ⑀-COP Bypassing the Golgi Apparatus-The previous toxicity experiments suggest that ricin enters the cytosol of ldlF cells lacking ⑀-COP by a different mechanism than in other cells.
To investigate the routing of ricin in ldlF cells lacking ⑀-COP we used ricin sulf-2, a modified ricin molecule containing a tyrosine sulfation site and three partially overlapping Nglycosylation sites at the carboxyl terminus of the A-chain (23). Ricin sulf-2 has previously been used to monitor the transport of the toxin to the TGN (site of sulfation (26)) and to the ER (site of glycosylation). To investigate whether ricin ingly, ricin sulf-2 labeled with [ 3 H]mannose was detected in control ldlF cells (preincubated at 34°C) and in ldlF cells lacking ⑀-COP (preincubated at 40°C) (Fig. 3A), although the amount of labeled ricin in ldlF cells lacking ⑀-COP was reduced by ϳ30% (Fig. 3B). This experiment shows that ricin can reach the ER in ldlF cells lacking ⑀-COP. Furthermore, to test the hypothesis that ricin is transported to the ER and to the cytosol bypassing the Golgi apparatus, ldlF cells were incubated with ricin sulf-2 in the presence of radioactive sulfate. When ricin sulf-2 was added to control ldlF cells (preincubated at 34°C) in the presence of radioactive sulfate and immunoprecipitated from cell lysates, two main labeled bands were visible (Fig. 3C, left lane, the two upper bands). It has previously been shown (23) in other cells lines that the band with the lowest molecular weight of these two bands represents ricin that has been sulfated in the TGN, and the band with the highest molecular weight of these two represents ricin that has been both sulfated in the TGN and glycosylated in the ER. This seems also to be the case in control ldlF cells because only the band with the highest molecular weight was observed when ricin-sulf 2 and radioactive sulfate was removed from the medium after 4 h, and the incubation was continued for 4 h more to allow sulfated ricin to be transported to the ER and be glycosylated (data not shown). In some experiments a third labeled band with a lower molecular weight was observed (Fig. 3C, lowest band), this band probably being a degradation product. Interestingly, we found that ricin sulf-2 also was sulfated in ldlF cells preincubated at the nonpermissive temperature (40°C) (Fig.  3C) although to a lower extent (ϳ5-fold reduction) (in Fig. 3C, lane 34°C and lane 40°C show different film exposures). The molecular weight of the band corresponding to sulfated ricin was slightly different in cells preincubated at 34°C and 40°C (compare middle band in Fig. 3C). Interestingly, whereas in control cells (ldlF cells preincubated at the permissive temperature) ϳ45% of the total amount of sulfated ricin was transported to the ER (Fig. 3D), only minimal amounts of ricin sulf-2, which was both sulfated and glycosylated, were observed in ldlF cells preincubated at the nonpermissive temperature (Fig. 3C, 40°C), thus suggesting that the transport of ricin from the Golgi apparatus to the ER is severely inhibited in ldlF cells lacking ⑀-COP and supporting the idea that ricin is able to reach the ER and intoxicate these cells bypassing the Golgi apparatus.
Characterization of Ricin Trafficking in ⑀-COP-deficient Cells-It has previously been shown that ricin is internalized but does not intoxicate cells at 17°C. To investigate whether this was the case also in ldlF cells lacking ⑀-COP, a toxicity experiment was performed at 17°C with ldlF cells preincubated at the permissive or at the nonpermissive temperature. As shown in Fig. 4A, neither control ldlF cells grown at the permissive temperature (34°C) nor ldlF cells grown at the nonpermissive temperature (40°C) were intoxicated with ricin at 17°C, thus indicating that the transport of ricin to the ER in cells lacking ⑀-COP also requires a temperature-sensitive step. In addition, COPI or another BFA-sensitive coat may be involved in the induction of the BFA-insensitive pathway in ldlF cells depleted of ⑀-COP because when BFA was present during the overnight incubation of ldlF cells at 40°C as well as during the toxicity experiment, ricin was no longer able to intoxicate ldlF cells lacking ⑀-COP (Fig. 4B).
Direct trafficking from a specialized endosomal compartment (caveosome) to the ER was recently demonstrated for simian virus 40 (27). This transport step was reported to be dependent on microtubules, and we therefore tested whether ricin intoxication in ldlF cells lacking ⑀-COP could be inhibited by nocodazole. As shown in Fig. 5, neither disruption of microtubules with nocodazole (16 M), nor of the actin cytoskeleton with cytochalasin D (1 M), affected the toxicity of ricin in ldlF cells depleted of ⑀-COP. Because it has previously been found that cholesterol is important for ricin trafficking (2, 28), we also investigated whether depletion of cholesterol would affect ricin toxicity in ldlF cells. As shown in Fig. 5C, reduction of membrane cholesterol by addition of methyl-␤-cyclodextrin (2.5 mM) slightly increased the toxicity of ricin in control ldlF cells (34°C), but decreased the toxicity of the toxin in ldlF cells lacking ⑀-COP (40°C).

DISCUSSION
The most important finding of this study is that the toxin ricin seems to be transported to the ER and to the cytosol bypassing the Golgi apparatus in ldlF cells lacking ⑀-COP (Fig.  6). In ldlF cells lacking ⑀-COP after incubation at the nonpermissive temperature (40°C), the function of COPI is impaired and the Golgi apparatus is disrupted (19). Nevertheless, ricin is transported to the ER and intoxicates these cells. Under nor- mal conditions ricin intoxicates cells after being transported from the endosomal compartment to the Golgi apparatus, and then to the ER. However, our experiments indicate that ricin can be transported to the ER by an alternative mechanism in ldlF cells lacking ⑀-COP. As shown, the transport of ricin from the Golgi apparatus to the ER (revealed by studies of sulfatelabeled ricin) is severely inhibited in ldlF cells lacking ⑀-COP. Secondly, BFA only inhibits the toxicity of ricin to some extent in ldlF cells lacking ⑀-COP (incubated at the nonpermissive temperature, 40°C), but completely in control ldlF cells (incubated at the permissive temperature, 34°C). These results cannot be explained by a lack of effect of BFA on the Golgi structure because it has previously been shown that BFA causes the redistribution of the disturbed Golgi apparatus into the ER in ldlF cell depleted of ⑀-COP (25). Importantly, the difference in toxicity in the presence of BFA in ldlF cells and CHO cells at the nonpermissive temperature is not caused by the higher intracellular amount of ricin in the ldlF cells because a similar difference was observed when the experiment was performed in the presence of bafilomycin A 1 .
It is not clear at the moment why the toxicity of ricin is inhibited to some extent by BFA in ldlF cells depleted of ⑀-COP. The possibility exists that there is more than one transport route from the endosomal compartment to the ER under these conditions and that only one is sensitive to BFA, or it could be that there is one pathway that is only partially sensitive to FIG. 6. Transport of ricin to the ER in ldlF cells incubated overnight at 34°C or at 40°C. A, transport of ricin to the ER in control ldlF cells. Under normal conditions ricin intoxicates cells after being transported from the endosomal compartment to the Golgi apparatus, and then to the ER. Sulfation experiments show that ricin sulf-2 that has been sulfated in the TGN is glycosylated in the ER. The transport to the cytosol is BFA-sensitive. B, transport of ricin to the ER in ldlF cells lacking ⑀-COP. Ricin seems to be transported to the ER by an alternative mechanism in ldlF cells lacking ⑀-COP. Sulfation experiments show that ricin sulf-2 that has been sulfated in the TGN is not glycosylated in the ER. However, mannosylation experiments show that ricin incorporates radioactive mannose in the ER. The transport to the cytosol is BFA-resistant.

FIG. 4. Ricin does not intoxicate ldlF cells grown overnight at 40°C
with BFA or ldlF cells lacking ⑀-COP at 17°C. A, ldlF cells were incubated overnight at 34°C (permissive temperature) (E) or at 40°C (nonpermissive temperature) (•) in culture medium. The cells were then washed in Ham's F-12 medium without serum and incubated at 17°C with increasing concentrations of toxin. 4 h later protein synthesis was measured as described under "Experimental Procedures." B, ldlF cells were incubated overnight at 40°C (nonpermissive temperature) with (E) or without (•) BFA (2 g/ml), and then increasing concentrations of toxin were added at 40°C. 4 h later protein synthesis was measured as described under "Experimental Procedures." The figures show the average between duplicates of representative experiments. inhibition by BFA. However, it is unlikely that the inhibition of toxicity by BFA reflects an effect of the drug on the transport of ricin passing through the vesiculated Golgi, because there is only a very small fraction of sulfated ricin that is transported from the TGN (sulfation site) to the ER and glycosylated.
It was recently reported that ricin-intoxication of ldlF cells transfected with cDNA encoding wild type ⑀-COP at 40°C is only partially sensitive to BFA (20). It is not clear why BFA did not completely inhibit the toxicity of ricin in ldlF cells transfected with wild-type ⑀-COP at 40°C, but it is possible that there is still an effect of the temperature-sensitive form of ⑀-COP in these cells. A changed conformation or the degradation of the temperature-sensitive protein might still affect the cells.
It has been shown that the simian virus 40 is directly transported to the ER from non-classical endosomes termed caveosomes, and that this process requires intact microtubuli (27). Unlike the simian virus 40, transport of ricin to the cytosol does not require microtubuli, suggesting that caveosomes are not involved in ricin transport to the ER in ldlF cells lacking ⑀-COP.
Recent data indicate that ricin is transported to the Golgi apparatus from the early/recycling compartment (29). In agreement with this idea we show here that ricin can intoxicate ldlF cells depleted of ⑀-COP, where the transport from early to late endosomes is inhibited (10). Altogether our experiments suggest that ricin intoxicates ldlF cells lacking ⑀-COP by using an alternative pathway that bypasses the Golgi apparatus.