Dependence of Ricin Toxicity on Translocation of the Toxin A-chain from the Endoplasmic Reticulum to the Cytosol*

Ricin acts by translocating to the cytosol the enzymatically active toxin A-chain, which inactivates ribosomes. Retrograde intracellular transport and translocation of ricin was studied under conditions that alter the sensitivity of cells to the toxin. For this purpose tyrosine sulfation of mutant A-chain in the Golgi apparatus, glycosylation in the endoplasmic reticulum (ER) and appearance of A-chain in the cytosolic fraction was monitored. Introduction of an ER retrieval signal, a C-terminal KDEL sequence, into the A-chain increased the toxicity and resulted in more efficient glycosylation, indicating enhanced transport from Golgi to ER. Calcium depletion inhibited neither sulfation nor glycosylation but inhibited translocation and toxicity, suggesting that the toxin is translocated to the cytosol by the pathway used by misfolded proteins that are targeted to the proteasomes for degradation. Slightly acidified medium had a similar effect. The proteasome inhibitor, lactacystin, sensitized cells to ricin and increased the amount of ricin A-chain in the cytosol. Anti-Sec61α precipitated sulfated and glycosylated ricin A-chain, suggesting that retrograde toxin translocation involves Sec61p. The data indicate that retrograde translocation across the ER membrane is required for intoxication.

the A-chain appears to be translocated to the cytosol (12). On the other hand, evidence for translocation from endosomes has also been reported (13,14).
In attempts to elucidate if translocation from the ER is the main entry route, we have applied a number of conditions that increase or reduce the toxin sensitivity of cells and studied intracellular trafficking and membrane translocation of the toxin. It is an old observation that cells depleted for calcium or kept at moderately acidic conditions are partly protected against ricin (15,16). On the other hand, treatment with low concentrations of monensin increases the sensitivity of cells to ricin, whereas higher concentrations provide a partial protection (16). In neither case could the changed sensitivity be accounted for by altered rates of endocytosis or altered enzymatic action on ribosomes, and we have therefore here considered the possibility that it could be due to interference with the transport of the toxin from endosomes to the ER or to interference with translocation of the A-chain from ER to cytosol. A mutant ricin A-chain containing a C-terminal KDEL sequence was shown to be more toxic than wild-type ricin (17,18), and it was proposed that this was due to increased transport to the ER.
In the present work we have taken advantage of a recently developed system where we fused tyrosine sulfation and Nglycosylation signals onto the C-terminal end of ricin A-chain (12). Tyrosine sulfation takes place in the Golgi apparatus and in the trans-Golgi network (19,20). Therefore, when ricin containing the sulfation signal is added to cells growing in medium containing 35 SO 4 2Ϫ , that minority of toxin molecules that is transported to the Golgi apparatus becomes labeled selectively. Furthermore, the addition of glycosylation sites onto the toxin A-chain allows us to follow the labeled toxin to the ER (12). When the toxin reaches this organelle, the A-chain becomes glycosylated, which can easily be monitored by a decreased migration rate in SDS-PAGE.
We here provide evidence that monensin interferes with transport of ricin from the cell surface to the Golgi apparatus, whereas depletion of calcium and reduced pH do not but instead, inhibit translocation of the A-chain across the ER membrane. KDEL-tagged ricin is more efficiently transported from the Golgi apparatus to the ER. Finally, we present evidence that ricin utilizes the ER translocon for retrograde translocation to the cytosol. from Dr. Thomas Sommer, Max-Delbruck-Centrum, Berlin. Rabbit anti-human immunodeficiency virus-Tat was obtained from ImmunoDiagnostics. Lactacystin was provided by Dr. E. J. Corey, Harvard University, Cambridge, MA. Other chemicals were from Sigma.
Cell Culture-Cells were maintained and propagated under standard conditions (5% CO 2 in Eagle's minimal essential medium containing 5% fetal calf serum). Two days before the experiment, the cells were seeded into 12-or 24-well Costar plates at a density of 10 5 and 5 ϫ 10 4 cells/well, respectively.
Plasmids-Ricin A-chain-sulf1 and ricin A-chain-sulf2 were prepared as described (12). To form ricin A-sulf2-KDEL, a C-terminal extension was introduced by polymerase chain reaction using pMal-CN-RA-sulf2 as template and 5Ј-GCAATCACTCATCTTTTCACTGATGTT-C-3Ј as forward primer and 5Ј-GCGTAAGCTTTATCACATAATGACG-CACTGGGATGTGTTATTTTTGGTGCCGTT-3Ј as reverse primer. The product was cut with BglII and HindIII and cloned into pMAL-cN-RA-sulf2 that had been cut with the same enzymes. All plasmids were transfected into Escherichia coli DH5␣.
Purification of Recombinant Proteins and Reconstitution with Ricin B-chain-After induction with isopropyl ␤-D-thiogalactoside, bacteria were harvested, washed with PBS, resuspended in column buffer (20 mM Tris-HCl, pH 7.5, 200 mM NaCl, and 0.1 mM phenylmethylsulfonyl fluoride) and sonicated. After centrifugation at 10,000 rpm for 10 min, the clear supernatant was applied to a column with amylose resin. The fusion proteins were eluted with column buffer containing 10 mM maltose. Free ricin A-chain (with a modified C terminus) was cleaved off with factor Xa, mixed with ricin B-chain, and dialyzed extensively against PBS to remove reducing agents and allow a disulfide bridge to be formed between the two chains, as described in Rapak et al. (12). In an alternative way, to produce mutant ricin A-chain, the ricin-A-sulf2 construct was cloned into the expression vector pUTA (21) using standard procedures. The mutant A-chain was expressed and purified as described (21).
Measurement of Cytotoxicity-Vero cells were incubated with increasing amounts of toxin for 4 h in Hepes medium without leucine. The cells were then transferred to Hepes medium containing 1 Ci/ml [ 3 H]leucine and no unlabeled leucine and incubated for 20 min at 37°C. The cells were then extracted with 5% trichloroacetic acid for 10 min, followed by a brief wash in 5% trichloroacetic acid and subsequently dissolved in 0.1 M KOH. The cell-associated radioactivity was measured.
Permeabilization of Cells with Streptolysin O (SLO)-Vero cells were incubated with 35 SO 4 2Ϫ and reconstituted ricin for 4 h. Then 10 mM N-ethylmaleimide was added, and the cells were further incubated for 10 min and subsequently washed twice with PBS containing 0.1 M lactose. The cells were then placed on ice, and 2 g/ml SLO preactivated with 10 mM 2-mercaptomethanesulfonic acid in Hepes medium, pH 7.5, was added as described (12). After 10 min the cells were washed briefly to remove unbound toxin and incubated for 10 min at 37°C for permeabilization to occur. Subsequently, the cells were kept on ice for an additional 30 min to allow components of the cytosol to diffuse into the buffer, which was collected. Finally, the cells were scraped from the dish, centrifuged, and lysed with 0.5% Triton X-100 in PBS. The lysate and buffer were exposed to immobilized anti-ricin antibodies, and the adsorbed material was subjected to SDS-PAGE.
SDS-PAGE-SDS-PAGE was carried out as described (22) in the absence or presence of 2-mercaptoethanol, as indicated. The gels were fixed in 4% acetic acid, 27% methanol for 30 min, and then, in the case of proteins labeled with Na 2 35 SO 4 , treated with 1 M sodium salicylate, pH 5.8, in 2% glycerol for 30 min. Dried gels were exposed to Kodak XAR-5 films at Ϫ80°C for fluorography.
Immunoprecipitation-Cells were lysed in lysis buffer containing 1% Triton X-100, the nuclei were removed by centrifugation, and the supernatant was treated with antibodies immobilized on protein A-Sepharose (Amersham Pharmacia Biotech). The adsorbed material was analyzed by SDS-PAGE and fluorography.
Estimation of Band Intensity-To compare the intensity of bands in polyacrylamide gels, the gels were scanned with a Personal Densitometer SI with ImageQuant software (Molecular Dynamics, Sunnyvale, CA). To correct for the background, adjacent areas of the gels with the same size but containing no visible bands were scanned, and the values obtained were subtracted from those obtained with the bands of interest.

RESULTS
Toxicity of Modified Ricin-Ricin A-chain was modified to contain a C-terminal tyrosine sulfation signal without (ricin A-sulf1) and with (ricin A-sulf2) three partly overlapping Nglycosylation sites. In one case (ricin A-sulf2-KDEL) a C-terminal KDEL sequence was added as well (Fig. 1A). The constructs were expressed as fusion proteins with maltose-binding protein to allow their purification on an amylose column. Subsequently, the maltose-binding protein was cleaved off with factor Xa, a site that had been inserted between the two proteins.
The modified A-chains were reconstituted with ricin B-chain to form holotoxins. In the following the reconstituted toxins will be referred to as wild-type ricin, ricin-sulf1, ricin-sulf2, and ricin-sulf2-KDEL. When increasing concentrations of the reconstituted toxins were added to Vero and U2OS cells and the ability of the cells to incorporate [ 3 H]leucine into trichloroacetic acid-precipitable material was measured after 4 h, there was in all cases a dose-dependent reduction in incorporated radioactivity that did not differ much from that obtained with wildtype ricin (Fig. 1, A and B). Ricin-sulf2-KDEL was somewhat more toxic than the parent ricin-sulf2, particularly in U20S cells. Incubation of cells for shorter times (1 or 2 h) gave similar results (data not shown). This demonstrates that the alterations introduced into ricin A-chain did not strongly interfere with the toxic effect.
Sulfation and Glycosylation of Modified Ricin-When ricinsulf-2 was added to cells in the presence of Na 2 35 SO 4 , two labeled bands were observed ( Fig. 2A, lane 1) in accordance with earlier data (12). The lower band represents the unglycosylated form of ricin-A-sulf2, and the upper band represents the glycosylated form (gA). When a C-terminal KDEL sequence was added (ricin-sulf2-KDEL), the amount of unglycosylated protein was considerably reduced (lane 2), suggesting that the KDEL sequence increased transport of the protein to the ER.
The presence of brefeldin A, which disrupts the Golgi apparatus, completely prevented sulfation of ricin-A-sulf2-KDEL (Fig. 2B, lane 2) as we have earlier found with ricin-A-sulf-2 (12). Tunicamycin prevented the appearance of the heavier band (lane 3), confirming that this band represents the glycosylated form of ricin-A-sulf2-KDEL.

FIG. 1. Wild-type and mutant ricin A-chains and their toxicity to Vero and U2OS cells after reconstitution with ricin B-chain.
A, amino acid sequence of the C terminus of wild-type (wt) ricin A-chain, ricin A-chain tagged with a tyrosine sulfation site (indicated in bold face, Ricin A-sulf1), ricin A-chain tagged with both the sulfation site and N-glycosylation sites (Ricin A-sulf2), ricin A-chain containing in addition a KDEL motif (Ricin A-sulf2-KDEL). B and C, increasing amounts of wild-type and mutant ricin A-chains reconstituted with ricin B-chain were incubated with Vero (B) and U20S cells (C) for 4 h. Then the ability of the cells to incorporate of [ 3 H]leucine was measured as described under "Experimental Procedures." Effect of Monensin on Ricin Toxicity-When low concentrations of monensin (0.01-0.1 M) were added to the culture medium, the cells were approximately 3-fold more sensitive to ricin than in the absence of the compound (Fig. 3, A and B). In the presence of higher concentrations (1-10 M monensin), the sensitivity to the toxin was somewhat reduced. This is in accordance with earlier observations (16).
Monensin affects the structure of the Golgi apparatus, which swells up at higher concentrations (23). To study if the altered sensitivity was due to altered transport to or altered function of the Golgi apparatus, we measured the ability of ricin-sulf2 to be sulfated and glycosylated in the presence of monensin. The results in Fig. 3, C and D, demonstrate that at the lower concentrations of monensin there was a moderate increase in labeling of the toxin with sulfate, whereas at the higher concentrations, the labeling was considerably reduced. This indicates that low concentrations of monensin increase the transport of toxin to the Golgi apparatus, whereas higher concentrations either inhibit the transport or reduce the labeling due to swelling of the Golgi apparatus. On the other hand, the relative fraction of labeled toxin that became glycosylated was not affected, as the relation between the two labeled bands was approximately the same at the different concentrations of monensin (Fig. 3, E and F). The data therefore indicate that monensin interferes either with the transport of toxin to the Golgi apparatus or with the sulfate labeling but not with the further retrograde transport to the ER.
The electrophoresis was carried out under nonreducing conditions, and in Fig. 3C the gel was run to also visualize the separated chains to check to what extent cell-induced reduction of the toxin A-and B-chains had taken place. Such cell-mediated reduction could be a requirement for translocation to the cytosol. The data demonstrate that monensin did not strongly influence the relative amount of toxin that was reduced by the cells.
Altogether, there is good correlation between the toxic effect on cells and labeling of the toxin and subsequent glycosylation. The findings are therefore consistent with the possibility that transport through the Golgi apparatus to the ER is necessary for intoxication.
Ability of Low pH of the Medium to Protect Cells against Ricin-When the cell culture medium is slightly acidified under conditions that do not inhibit endocytosis from coated pits (24), the cells are partly protected against ricin (16). The data in Fig. 4, A and B, demonstrate that this is also the case with ricin-A-sulf2 and ricin-sulf2-KDEL.
Experiments using radioactive sulfate showed that there was no reduction in the labeling and glycosylation of the protein at low pH (Fig. 4C). In fact, in some cases there was rather an increased labeling of the toxin. In Vero cells the KDEL construct was less labeled than the parent ricin-sulf2, but there was no difference between the two pH values with respect to sulfate labeling and glycosylation. It therefore appears that incubation at slightly acidic pH provides protection against ricin by a mechanism occurring after the toxin has reached the ER.
To test if the altered sensitivity to ricin in cells incubated at pH 6.5 was due to interference with the translocation of the A-chain to the cytosol, we treated the cells with SLO to selectively permeabilize the surface membrane (12, 25, 26) and release soluble proteins from the cytosol into the buffer. Due to technical problems in carrying out SLO experiments with  U2OS cells (B, D, and F). A and B, cells were preincubated with monensin for 30 min. Then increasing amounts of ricin-sulf2 were added, and the cells were incubated for 4 h more. Finally, the ability of the cells to incorporate [ 3 H]leucine during 20 min was measured. C and D, cells were labeled with 35 SO 4 2Ϫ and preincubated with monensin for 30 min. Then the cells were incubated with ricin-sulf2 for 4 h more. Finally, the cells were washed, lysed, immunoprecipitated with anti-ricin, and analyzed by SDS-PAGE under nonreducing conditions followed by fluorography. A, A-chain; gA, glycosylated A-chain; R, ricin; gR, glycosylated ricin. E and F, the bands corresponding to unreduced ricin (glycosylated and nonglycosylated) in the fluorograms in C and D were scanned, and the densities of the bands were measured. E, unglycosylated A-chain; q, glycosylated A-chain. The concentration of ricin that reduced protein synthesis to the half (IC 50 ) at the different concentrations of monensin was estimated from panels A and B and plotted into the diagrams in panels E and F (OE).
U2OS cells, the experiments were performed only with Vero cells. After centrifugation, we then analyzed the buffer and the cellular pellet for labeled material. In a previous paper it was shown that mainly the glycosylated form of ricin-A-sulf2 was found in the cytosol after SLO treatment (12). In accordance with this, the data in Fig. 4D demonstrate that mainly glycosylated ricin-A-sulf2 was recovered from the cytosol of cells kept at pH 7.5. In cells incubated at pH 6.5 very little material was found in the cytosol, indicating that the translocation from the ER lumen to the cytosol was inhibited. A brief exposure to medium with neutral pH was, however, sufficient to allow the A-chain to appear in the cytosolic fraction (data not shown).
Effect of Ca 2ϩ -Depletion of Cells on Translocation of Mutant Ricin A-chain-When cells were briefly treated with EGTA and then incubated in Ca 2ϩ -free medium, they were found to be less sensitive to wild-type ricin than cells kept in normal medium (15). As demonstrated in Fig. 5, A and B, this is also the case with ricin-sulf2 and ricin-sulf2-KDEL. When cells were incu-bated with mutated ricin in the presence of 35 SO 4 2Ϫ , two labeled bands were obtained corresponding to glycosylated and nonglycosylated A-chain both with and without Ca 2ϩ depletion (Fig.  5C). This indicates that depletion of Ca 2ϩ does not inhibit transport of toxin to the ER.
Translocation of misfolded proteins to the cytosol for degradation requires normal Ca 2ϩ concentration in the ER (27). In cells depleted for Ca 2ϩ , the concentration of Ca 2ϩ in the ER is reduced (28). To investigate if the depletion of Ca 2ϩ inhibits the translocation to the cytosol of ricin, we permeabilized the cells with SLO and analyzed the cytosol. In the Ca 2ϩ -depleted cells, the translocation from the ER to the cytosol was strongly inhibited (Fig. 5D). Also in this case a short exposure of the cells to normal medium containing Ca 2ϩ was sufficient to let the A-chain appear in the cytosolic fraction (data not shown).
Sensitization of Cells to Ricin by the Proteasome Inhibitor Lactacystin-It has been reported that misfolded proteins that FIG. 4. Ability of acidic medium to protect against ricin intoxication. A and B, Hepes medium adjusted to pH 6.5 or 7.5 was added to Vero (A) and U20S cells (B). Subsequently, increasing amounts of ricin mutants were added, and the cells were incubated for 4 h. Protein synthesis was then measured during the next 20 min as incorporation of [ 3 H]leucine. C, U20S and Vero cells were preincubated with 35 SO 4 2Ϫ , and then the pH of the medium was adjusted to 6.5 or 7.5. Ricin mutants as indicated were added, and the cells were incubated for 4 h more. Finally, the cells were lysed, and the immunoprecipitated ricin was analyzed by SDS-PAGE and fluorography. A, A-chain; gA, glycosylated A-chain. D, translocation of ricin A-chain to the cytosol. Vero cells were incubated with Na 2 35 SO 4 and ricin containing A-sulf-2 for 4 h at pH 7.5 or pH 6.5. Then N-ethylmaleimide (10 mM) was added, and the cells were further incubated for 10 min and then washed twice with PBS containing 0.1 M lactose. The cells were permeabilized with SLO. The buffer was collected and treated with immobilized anti-ricin, and the adsorbed material was submitted to SDS-PAGE and fluorography.

FIG. 5. Effect of calcium depletion on the toxicity and sulfation of ricin mutants. Vero (A) and U2OS cells (B)
were washed briefly with 1 mM EGTA in PBS or with PBS alone. Then DMEM with or without 1 mM CaCl 2 and increasing amounts of ricin-sulf2 or ricin-sulf2-KDEL was added, and the cells were incubated for 4 h more. Finally, the rate of protein synthesis was measured. C, Vero and U20S cells were preincubated with 35 SO 4 2Ϫ and washed with or without EGTA as above, and then DMEM containing Na 35 SO 4 with or without CaCl 2 was added. Then the cells were incubated with the ricin mutants for 4 h. Finally, the cells were lysed, and the immunoprecipitated ricin was analyzed by SDS-PAGE and fluorography. A, A-chain; gA, glycosylated A-chain. D, translocation of ricin A-chain to the cytosol. Vero cells were incubated with Na 2 35 SO 4 and ricin containing A-sulf-2 for 4 h in the absence and presence of CaCl 2, as above. Then N-ethylmaleimide (10 mM) was added, and the cells were further incubated for 10 min and then washed twice with PBS containing 0.1 M lactose. The cells were permeabilized with SLO. The buffer was treated with immobilized anti-ricin, and the adsorbed material was submitted to SDS-PAGE and fluorography. are translocated retrograde from the ER to the cytosol become ubiquitinated by a membrane-bound enzyme and thereby targeted for degradation by proteasomes in the cytosol (29 -31).
To study if ricin A-chain is partially degraded by proteasomes, we tested if an inhibitor of proteasomes, lactacystin, is able to sensitize cells to ricin. The data in Fig. 6, A and B, demonstrate that in the presence of lactacystin, the cells were approximately 3-fold more sensitive to ricin than in the absence of the inhibitor.
As shown in Fig. 6C, the amount of sulfate-labeled and glycosylated A-chain in the whole cells was essentially the same with and without lactacystin. On the other hand, the amount of A-chain present in the cytosolic fraction after SLO treatment of the cells was approximately 3-fold higher when the activity of the proteasomes was inhibited by lactacystin (Fig. 6D). The data therefore indicate that although a fraction of ricin A-chain translocated to the cytosol is able to escape degradation, a considerable part is degraded by the proteasomes.
Additive Sensitization of Cells by Lactacystin and Ca 2ϩ -To study if lactacystin and calcium sensitize the cells by acting at the same or different steps in the intoxication process, we tested the sensitivity of cells in the presence of each component as well as their combination. The data in Fig. 7 demonstrate that compared with Ca 2ϩ -depleted cells, calcium and lactacystin sensitized the cells to wild-type ricin to approximately the same extent. The combination of the two conditions sensitized the cells to an extent corresponding to an additive effect of the two conditions. The data therefore indicate that lactacystin and calcium sensitize the cells to ricin by two independent mechanisms.
Immunoprecipitation of Ricin-A-sulf2 with Anti-Sec61␣ Antibodies-Several recent studies have provided evidence that misfolded ER proteins destined for degradation are first transported to the cytosol in a Sec61p-dependent manner before being degraded by proteasomes (32)(33)(34)(35). To test if ricin also interacts with the Sec61p translocon, we first incubated cells with 35 SO 4 2Ϫ and ricin-sulf1 (Fig. 8, panel A) or ricin-sulf2 (Fig.  8, panel B). After a 4-h incubation, we lysed the cells and subjected the lysate to sequential immunoprecipitation. In a first round we used an irrelevant antibody (anti-HSP90 (Fig.  8A) or anti-human immunodeficiency virus-Tat (Fig. 8B)) coupled to protein A-Sepharose beads. Subsequently, we incubated the same lysate with antibodies against Sec61␣ coupled to protein A-Sepharose. Finally, the lysate was incubated with immobilized anti-ricin antibodies. As shown in Fig. 8, A and B, no ricin was precipitated with the control antibodies, whereas a small amount of sulfated and glycosylated ricin was precipitated with anti-Sec61␣. Control experiments demonstrated that the antibodies used did not precipitate 125 I-labeled ricin as such (data not shown).
Interestingly, in the material immunoprecipitated with anti-Sec61␣ in Fig. 8, panel B, the relation between the glycosylated and the nonglycosylated form of ricin A-chain was 1.6 and 1.9, whereas in the immunoprecipitate with anti-ricin antibodies, the relation was only 0.9. Clearly, the glycosylated form of the A-chain is preferentially associated with Sec61p. The data indicate that ricin interacts with the Sec61p complex in the ER. When compared with the total amount of labeled ricin in the cells, the amount immunoprecipitated with anti-Sec61␣ was found to be 0.93% in the case of Fig. 8A and 0.58 and 0.63% in the case of Fig. 8B.
When the cells were kept in the absence of Ca 2ϩ , where translocation of the A-chain to the cytosol is inhibited, we were not able to immunoprecipitate more ricin A-chain with anti-Sec61␣ than from cells kept under normal conditions (data not shown). This indicates that in calcium-depleted cells the inhibition of translocation is at a step before the A-chain interacts with Sec61p. DISCUSSION Previously we provided direct evidence that ricin is transported retrograde through the Golgi apparatus to the ER and subsequently translocated to the cytosol (12), but we did not exclude the possibility that there could be other translocation routes that could be more efficient. It has been reported that ricin can be translocated from endosomes (13,14) and that transport through the Golgi apparatus may not be required (36). In the present paper we have investigated in detail the effect on transport and translocation of different conditions that decrease or enhance the sensitivity of cells to the toxin. Under these conditions the toxic effect correlated with transport to the Golgi apparatus and translocation from the ER lumen to the cytosol.
Certain toxins have a C-terminal KDEL or KDEL-like sequence (37,38), which is known to act as an ER retrieval signal. It has been proposed that the KDEL system could facilitate transport of toxins to the ER (39). Wild-type ricin does not contain a KDEL sequence, but addition of this sequence to the C terminus of the isolated A-chain of ricin increased its toxicity to a varying extent in different cells (18,40). To check the effect of the KDEL sequence on the transport of ricin to the ER, we added this sequence to the C terminus of ricinA-sulf2. Ricin-sulf2-KDEL was more efficiently transported to the ER in U20S cells than the parent ricin-sulf2, and it was more toxic. Interestingly, ricin-sulf2-KDEL was labeled very weakly in Vero cells. Possibly, in these cells the mutant leaves the Golgi apparatus quickly such that sulfation is incomplete. Ricin-sulf2-KDEL was more extensively glycosylated in U2OS cells at lower pH, indicating more efficient transport to the ER, possibly due to a higher affinity to the KDEL receptor at lower pH (41).
Monensin, which is a carboxylic ionophore, prevents at higher concentrations acidification of intracellular compartments (23), thereby inhibiting the degradation of ricin by lysosomes. Also, endosomes and the Golgi apparatus swell in the presence of the drug, and the transport to and through the Golgi stacks is inhibited. The effect of submicromolar concentrations of monensin is less well understood. At lower concentrations of monensin, there was increased labeling of the toxin and increased toxicity. This supports earlier findings that transport of the toxin to and through the Golgi apparatus is required for intoxication (42,43).
Earlier it was demonstrated that the toxicity of ricin is reduced in cells depleted for calcium, incubated at slightly reduced pH or in the presence of high concentrations of monensin, although the binding of ricin to cells and endocytosis of the toxin was not altered under these conditions (15,16). Also, the rate of inactivation of ribosomes in a cell-free system was not inhibited.
When cells are kept in Ca 2ϩ -free medium, the concentration of Ca 2ϩ in the ER rapidly decreases from 500 -600 M to about 150 M (28). It is known that calcium plays an important role in the translocation of proteins across the ER membrane (44). The majority of chaperones, like BiP, protein disulfide isomerase, GRP94, calnexin, and calreticulin, which could assist ricin during translocation, are calcium-dependent (45,46). BiP is supposed to close the lumenal end of the Sec61p translocation channel (47) and would therefore have to be removed before ricin A-chain could enter the pore. Reduced calcium concentration could interfere with this process. Importantly, a decrease in the Ca 2ϩ concentration in the ER of yeast due to a mutated Ca 2ϩ pump resulted in an inability of the cells to degrade misfolded carboxypeptidase Y (27), a process occurring by the proteasomes after retrograde transport of the protein to the cytosol (48).
A moderate reduction in pH of the medium (from 7.5 to 6.5) was also found to strongly reduce the translocation of ricin A-chain from the ER to the cytosol. The reason for this is not clear. A reduction in extracellular pH will only moderately reduce the pH of the cytosol due to the correcting effect of the Na ϩ /H ϩ exchanger (49), which is operative in the bicarbonatefree Hepes medium used. The pH in the ER is close to that in the cytosol (50), and it would be expected to be slightly lowered by the transfer of the cells from the regular medium to the pH 6.5 medium. The question of whether the slightly reduced pH in the ER or that in the cytosol is the reason for the reduced translocation efficiency remains to be elucidated.
It is known that certain proteins may be transported retrograde from the ER to the cytosol. Wiertz et al. (30) provided evidence that this retrograde transport is mediated by the Sec61p channel. The HCMV gene US2 product bound to newly synthesized major histocompatibility complex class I heavy chains and the complex formed was rapidly transported from the ER to the cytosol, deglycosylated by N-glycanase, and subsequently degraded by the proteasomes (51). Also, it was demonstrated in yeast that misfolded secretory proteins in the ER are exported via the Sec61p channel to the cytosol for degradation (48,52). Using two cold-sensitive mutants of Sec61 at a temperature permissive for protein import, membranes were deficient for export of misfolded secretory proteins, which remained associated with Sec61p. In the presence of ATP and cytosol, the misfolded secretory protein was released from wildtype Sec61p and degraded. Mutations in two other translocon components, Sec62 and Sec63, only slightly affected export (34). In in vivo experiments using a mutated lumenal yeast carboxypeptidase Y, Sec63p and Kar2p (BiP) were found to be involved in export of carboxypeptidase Y to the cytosol for degradation (35).
We have found here that the toxin interacts with Sec61p. Compared with the total amount of labeled ricin in the cells, only a small fraction (0.6 -0.9%) is associated with Sec61p. The reason for this could be that only a small fraction of the toxin FIG. 8. Immunoprecipitation of ricin-A-sulf1 (A) and ricin-A-sulf2 (B) with anti-Sec61␣. Vero cells were incubated with Na 35 SO 4 and ricin-sulf1 or ricin-sulf2 for 4 h. The cells were lysed and subjected to immunoprecipitation with immobilized anti-HSP90 (A) or anti-Tat antibodies (B). The supernatant was subsequently treated with immobilized anti-Sec61␣ antibodies. In B the ricin left in the supernatant was finally immunoprecipitated with immobilized anti-ricin. The immunoprecipitated proteins were analyzed by SDS-PAGE under reducing conditions followed by fluorography. In the panel with anti-ricin the film was exposed for only half the time as compared with those with the other antibodies. The paired lanes in panel B represent duplicate samples. A, A-chain; gA, glycosylated A-chain.
A-chain is in the process of translocating at any time or that the association is weak and that only a minor part does not dissociate during the immunoprecipitation. In the absence of Ca 2ϩ the translocation is inhibited. However, this inhibition appears to be at some step before the translocation, since we were not able to immunoprecipitate more ricin A-chain with anti-Sec61␣ under these conditions.
A protein bound to the ER membrane, Cue1p, is a component of a ubiquitin system that interacts with the soluble Ubc7p enzyme at the ER surface. In the absence of Cue1p, the degradation substrates remained inside the ER, indicating that ubiqutination is an important process during retrograde transport from the ER to the cytosol (53). To test the possibility that ricin may use the ER-associated protein degradation (ERAD) pathway for entry into the cytosol (54), we used a specific proteasome inhibitor, lactacystin. Lactacystin increased the toxicity of both wild-type ricin and ricin mutants approximately 3-fold as well the amount of free A-chain in the cytosol. This indicates that a fraction of the A-chain translocated to the cytosol is degraded by the proteasomes. Possibly, therefore, the toxin may use the same translocation mechanism, as defect or misfolded proteins employ during transport through the Sec61p channel for destruction by the proteasomes. However, for toxins to kill the cell, they must at least to some extent avoid proteasome degradation. It was proposed that toxins that are translocated from the ER could be inefficiently ubiquitinylated due to a low lysine content of the A chain (55). In the course of evolution, ricin A-chain may have learned how to behave like a misfolded protein in the ER and as a correctly folded one after it has been translocated to the cytosol.