Retrograde Transport of Pseudomonas Exotoxin a by the KDEL Receptor 469 Prepared Locally

Journal of Cell Science 112, 467-475 (1999) 467 Printed in Great Britain © The Company of Biologists Limited 1999 JCS4636

The KDEL retrieval system is exploited by Pseudomonas exotoxin A, but not by Shiga-like toxin-1, during retrograde transport from the Golgi complex to the endoplasmic reticulum

Michelle E. Jackson1,*, Jeremy C. Simpson1,*, Andreas Girod2,*, Rainer Pepperkok2, Lynne M. Roberts1 and J. Michael Lord1,‡ 1Department of Biological Sciences, University of Warwick, Coventry, CV4 7AL, UK 2Light Microscopy Laboratory, Imperial Cancer Research Fund, 44 Lincoln’s Inn Fields, London WC2A 3PX, UK *All three authors contributed equally to this study ‡Author for correspondence (e-mail: [email protected])

Accepted 27 November 1998; published on WWW 25 January 1999

SUMMARY

To investigate the role of the KDEL receptor in the retrieval when they express additional KDEL receptors. These data of protein toxins to the mammalian cell endoplasmic suggest that, in contrast to SLT-1, PE can exploit the KDEL reticulum (ER), lysozyme variants containing AARL or receptor in order to reach the ER lumen where it is believed KDEL C-terminal tags, or the human KDEL receptor, have that membrane transfer to the cytosol occurs. This been expressed in toxin-treated COS 7 and HeLa cells. contention was confirmed by microinjecting into Vero cells Expression of the lysozyme variants and the KDEL antibodies raised against the cytoplasmically exposed tail receptor was confirmed by immunofluorescence. When of the KDEL receptor. Immunofluorescence confirmed that such cells were challenged with diphtheria toxin (DT) or these antibodies prevented the retrograde transport of the Escherichia coli Shiga-like toxin 1 (SLT-1), there was no KDEL receptor from the Golgi complex to the ER, and this observable difference in their sensitivities as compared to in turn reduced the cytotoxicity of PE, but not that of SLT- cells which did not express these exogenous proteins. By 1, to these cells. contrast, the cytotoxicity of Pseudomonas exotoxin A (PE) is reduced by expressing lysozyme-KDEL, which causes a redistribution of the KDEL receptor from the Golgi Key words: Retrograde transport, KDEL receptor, Pseudomonas complex to the ER, and cells are sensitised to this toxin exotoxin A, Endoplasmic reticulum

INTRODUCTION (see reviews by Sandvig and van Deurs, 1994; Lord et al., 1994; Lord and Roberts, 1998). The ER is an attractive Toxins such as ricin, Pseudomonas exotoxin A (PE), diphtheria contender for a translocation compartment as it contains factors toxin (DT), Shiga toxin (ST) and Escherichia coli Shiga-like which could facilitate entry to the cytosol by reducing and toxin 1 (SLT-1), kill cells by inhibiting protein synthesis. The unfolding the toxins prior to their membrane partitioning. It intoxication process can be divided into four main stages: cell also contains protein transport machinery, usually thought to binding, endocytosis, membrane translocation to the cytosol be involved in the translocation of newly synthesised proteins and, finally, enzymatic modification of a cytosolic target. DT into the ER lumen, but which, as in the case of the Sec61p and PE inhibit protein synthesis by catalysing the ADP- complex, has also been implicated in reverse translocation of ribosylation of elongation factor 2 (EF-2) (Pappenheimer, misfolded proteins from the ER lumen into the cytosol (Wiertz 1977; Iglewski and Kabat, 1975), whereas ricin, ST and SLT- et al., 1996; Pilon et al., 1997; Plemper et al., 1997). 1 hydrolytically cleave a specific N-glycosidic bond in the A number of studies have provided evidence to support the large rRNA of the 60S subunit of 80S ribosomes (Endo et al., physiological routing of certain toxins to the ER of mammalian 1987). The consequent release of a single adenine residue from cells. Firstly, brefeldin A (BFA) treatment, which disrupts the the highly conserved target rRNA sequence renders the Golgi stack causing it to merge with the ER, protects cells ribosome incapable of protein synthesis. In order to reach their against PE, ricin and SLT-1, but not against DT (Yoshida et al., cytosolic targets, the toxins, once internalised by the cell, must 1991; Sandvig et al., 1991a). BFA does not block endocytosis cross an intracellular membrane. Unlike DT, which enters the nor does it prevent the toxins from reaching the trans-Golgi cytosol from acidified endosomes, the intracellular site of network (TGN) in the cell lines tested (Yoshida et al., 1991; translocation of PE, ricin and ST/SLT-1 is thought to be the ER Sandvig et al., 1991a). Epidermal cell differentiation factor, an 468 M. E. Jackson and others

ADP-ribosyltransferase from Staphylococcus aureus E-1, for the cytotoxicity of PE (Chaudhary et al., 1990), and its causes a similar breakdown of the Golgi to that observed with replacement with KDEL (Seetharam et al., 1991) even BFA, and this metabolite also protects cells from the effects of enhances its potency. These data suggest that PE may directly ricin (Sugai et al., 1992a,b). Secondly, HRP conjugates of ST interact with the KDEL receptor, a view supported by the in have been directly visualised in the ER lumen of cells that have vitro association with immobilised KDEL receptors (Kreitman been sensitised to the toxin (Sandvig et al., 1992; 1994). and Pastan, 1995), although this has never been convincingly Indeed, the ST B fragment alone can be transported to the ER demonstrated in vivo. Endocytosed CT has also been visualised (Kline and Lingwood, 1994; Sandvig et al., 1994). Thirdly, in the Golgi and the ER (Sandvig et al., 1996), and an overexpression of trans-dominant mutant GTPases involved in immunofluorescence study demonstrated the retrograde regulating vesicle traffic between the ER and the Golgi transport of endocytosed CT from the Golgi to the ER (Majoul complex significantly reduces the cytotoxic potency of PE, et al., 1996). While it seems likely that CT, with a KDEL ricin and SLT-1, suggesting the importance of transport to or tetrapeptide, is also transported from the Golgi to the ER by through the Golgi stack (Simpson et al., 1995). Finally, interacting with the KDEL receptor, once again this has not modified forms of ST and ricin have been shown, by been directly demonstrated. acquisition of core oligosaccharides, to enter the ER after Involvement of the KDEL retrieval system with the transport endocytosis from the cell surface (Johannes et al., 1997; Rapak of other toxins is, however, less clear. For example, cholera B et al., 1997). chain (which lacks a KDEL sequence; the KDEL of CT is on Should PE, ricin and ST/SLT-1 reach the cytosol from the the A chain) has been visualised in all Golgi cisternae (Sandvig ER, the question remains as to the mechanism of retrograde et al., 1994) and ST, which also lacks a KDEL-like motif, has transport. Following uptake by clathrin-dependent or clathrin- been seen in the Golgi and the ER of sensitised A431 cells independent endocytosis (Sandvig et al., 1991b; Simpson et al., (Sandvig et al., 1992, 1994). Furthermore, removal of the 1998), the toxins are delivered to endosomes, and a large RDEL sequence from LT was reported to have no effect on its proportion is subsequently delivered to lysosomes for potency (Ciepak et al., 1995), although a different study degradation (reviewed in Sandvig and van Deurs, 1996). suggested that disruption of the LT-RDEL did indeed slow However, a small fraction of internalised ricin and ST has been down the toxin-induced secretion of chloride ions (Lencer et visualised in the TGN (van Deurs et al., 1988). Such toxin is al., 1995). Like ST and SLT-1, ricin does not contain a KDEL most likely in association with the membrane having exploited or KDEL-like motif. However, the fact that addition of a KDEL a route utilised by, for example, unoccupied mannose-6- tag to ricin A-chain increases its cytotoxic effect (Wales et al., phosphate receptors, TGN38 or furin (Goda and Pfeffer, 1988; 1993) at least suggests that a fraction of internalised ricin must Bos et al., 1993; Schafer et al., 1995). That the TGN represents encounter and, in this case, interact with recycling KDEL- a port for subsequent transport steps is indicated by the receptors, and that such interaction facilitates eventual delivery correlation between an intact Golgi and sensitivity to ricin and to the cytosol. ST (Sandvig et al., 1991a; Yoshida et al., 1991), and the In the present study we have investigated in vivo whether protective effects of low temperatures known to block the the KDEL retrieval system plays a part in the intracellular delivery of toxins to this compartment (van Deurs et al., 1987; transport of SLT-1 and PE. Cells were treated in ways that Sandvig et al., 1989). affected the cellular level, intracellular distribution, or In addressing retrograde transport from the TGN to the ER, functioning of the KDEL receptor. The predicted modifications it has been postulated that the toxins may exploit an ER were shown to have occurred by immunofluorescence. The retrieval system (Pelham et al., 1992). There are a number of effects of these modifications on the susceptibility of the cells signals which function as retrieval markers for escaped resident to intoxication demonstrate that PE exploits the KDEL retrieval ER proteins in the Golgi compartments. Type I and type II system to reach the ER, whereas SLT-1 utilises a different transmembrane resident ER proteins contain cytosolically carrier to reach the same destination. located sequences at their C and N termini, respectively, which act as ER retrieval signals (Jackson et al., 1993; Schutze et al., 1994). The only retrieval signal that has been identified for MATERIALS AND METHODS resident lumenal proteins is the K/RDEL (HDEL in yeast) motif found at the C terminus of soluble ER proteins (Munro Materials and Pelham, 1987). Escaped lumenal ER proteins containing Unless otherwise stated all materials were purchased from Sigma this signal are able to bind to KDEL receptors normally Chemical Co. (St Louis, MO, USA), Boehringer Mannheim GmbH distributed throughout the Golgi (Griffiths et al., 1994; (Mannheim, Germany) or BDH Lab Supplies (Poole, UK). Tissue Miesenböck and Rothman, 1995). Binding results in retrieval culture media, foetal calf serum (FCS) and restriction enzymes were of the receptor-ligand complex to the ER, where a shift in pH purchased from GIBCO BRL (Paisley, UK). The lipofection reagent triggers release of the ligand (Wilson et al., 1993). It has been DOSPERTM and protease inhibitors (CompleteTM tablets) were demonstrated that KDEL-tagged molecules can be retrieved obtained from Boehringer Mannheim GmbH. Rabbit polyclonal anti- from the TGN (Peter et al., 1992; Miesenböck and Rothman, chick lysozyme antibodies were obtained from Chemicon 1995), the compartment where ricin and ST have been directly International Inc. (Temecula, CA, USA) and 9E10 mouse monoclonal anti-c-myc antibodies were a gift from John M. Jarvis (MRC, visualised. Exploitation of this system may occur for certain Cambridge). Anti-mouse IgG antibodies (Texas Red-conjugated), toxins, such as PE, which contains a KDEL-like motif anti-rabbit IgG antibodies (FITC-conjugated) and [35S]PromixTM (REDLK) at its C terminus, cholera toxin (CT) with a C- were purchased from Amersham International plc (Amersham, UK). terminal KDEL and E. coli heat-labile toxin (LT) with C- DT, Moviol and PansorbinTM cells were obtained from Calbiochem terminal RDEL. Certainly, the REDLK sequence is important (La Jolla, CA, USA). PE was obtained from Sigma, and SLT-1 was Retrograde transport of Pseudomonas exotoxin A by the KDEL receptor 469 prepared locally. 1,4-diazabicyclo(2,2,2)-octane (DABCO) was 50 µl of prewashed PansorbinTM cells were added and incubated at 4°C obtained from Aldrich Chemical Co. (Milwaukee, WI, USA). The for 1 hour. The PansorbinTM cells were then washed, resuspended in anti-KDEL receptor antibody was generated against the 21 C-terminal sample buffer, and immunoprecipitations analysed by SDS- amino acids of the mammalian KDEL-receptor (Tang et al., 1993) by polyacrylamide gel electrophoresis and autoradiography. The levels of methods previously described (Pepperkok et al., 1993). The antibodies lysozyme synthesis are reported as a percentage of lysozyme found in were affinity-purified and Fab fragments were generated as described transfected cells that were not treated with toxin. (Pepperkok et al., 1993). Coumarin-conjugated bovine serum albumin (BSA) was purchased from Molecular Probes (Eugene, OR, USA). Microinjection experiments The plasmids pHYA, pHYK and pHYAE-4 were a gift from Hugh R. Vero cells were plated as drops onto glass coverslips in dishes in a 6 B. Pelham (MRC, Cambridge) (Lewis and Pelham, 1992). µl volume 2 days prior to microinjection. The following day the dishes Autoradiograms were quantified using a Molecular Dynamics were flooded with MEM (5% FCS) and the cells allowed to grow for Phosphorimager (Sunnyvale, CA, USA). a further 24 hours. The number of cells on each coverslip was counted to allow a value of protein synthesis per cell to be calculated. On the Plasmids and cell culture coverslips used for microinjection, all the cells (typically 400) were The plasmids pHYA, pHYK and pHYAE4 are COS 7 cell expression injected, as described by Pepperkok et al. (1993). The coverslips were vectors, which include the SV40 origin of replication. pHYA and then transferred to MEM containing dilutions of toxins designed to pHYK contain hen lysozyme cDNA modified to encode additional C- reduce protein synthesis to 10-20% of that in the toxin-free control terminal AARL and KDEL sequences, respectively (Lewis and cells. Cells were incubated for 4 hours, then starved for 30 minutes Pelham, 1992). pHYAE4 contains the cDNA for hen lysozyme in methionine-deficient MEM, before being pulsed with 35S- modified to encode an additional C-terminal AARL sequence and also methionine (NEN Life Science Products, Hounslow, UK) for 30 an erd2 cDNA, modified to include a c-myc epitope to the C terminus minutes. Protein synthesis was terminated with MEM containing 100 of the protein (Townsley et al., 1993). Each gene was expressed from µg/ml cycloheximide. Cells were washed twice with a copy of the adenovirus major late promoter. For use with the MEM/cycloheximide before lysis and protein precipitation using vaccinia expression system, erd2 and lysozyme (AARL or KDEL- TCA, according to established methods. TCA precipitates were tagged) were cloned into pET3a vectors (Novagen Inc., Milwaukee, washed five times with acetone before protein synthesis measurement WI, USA). The genes were amplified by PCR from the pHYAE4 and by liquid scintillation counting. Final protein synthesis levels were pHYK vectors using primers which introduced NdeI and BamHI sites determined as counts per cell. For immunofluorescence Vero cells at the 5′ and 3′ ends respectively. These sites were then used to clone were fixed with methanol at −20°C and subsequently rehydrated with into a BamHI and NdeI restricted pET3a vector, placing the genes several washes in PBS. Thereafter they were immunostained with under the control of a T7 promoter within the plasmid. COS 7, HeLa respective antibodies as described (Pepperkok et al., 1993, 1998). and Vero cells were routinely cultured in Dulbecco’s modified Eagle’s Photographs of the respective immunostainings were taken with a medium (DMEM) containing 5% FCS and supplemented with microscope system as previously described (Scales et al., 1997). penicillin, streptomycin and glutamine. Cells were grown in a humidified incubator at 37°C. RESULTS Transfections COS 7 cells were seeded into 35 mm dishes at 2.5×105 cells/dish and Immunofluorescence localisation of the KDEL- allowed to adhere overnight. After washing with OptiMEMTM cells receptor and KDEL/AARL-tagged lysozyme µ µ were incubated with 1 g DNA and 15 l of the lipofection reagent Firstly we demonstrated that we were able to repeat the ligand- DOSPERTM in 1 ml OptiMEMTM per dish, for 6 hours. The transfection mix was then removed and replaced with DMEM (5% induced redistribution of the human KDEL receptor (hERD2 FCS); 24 hours after initial transfection the medium was again replaced protein) from the Golgi complex to the ER (as described by with fresh DMEM (5% FCS). Immunofluorescence studies and toxin Lewis and Pelham, 1992) in COS 7 cells. We expressed a version treatments were carried out 48 hours post-transfection. HeLa cells were of hERD2 that had been modified to include a c-myc epitope at transfected using a vaccinia expression system described previously the C terminus (Lewis and Pelham, 1992) to allow detection with (Fuerst et al., 1986). Cells were seeded into 35 mm dishes at 2.5×105 the anti-myc monoclonal antibody 9E10. Cells expressing both cells/dish, and allowed to adhere overnight. After 30 minutes infection hERD2 and either KDEL- or AARL-tagged lysozyme were by a recombinant vaccinia virus carrying a T7 RNA polymerase gene labelled for immunofluorescence. In keeping with the earlier (vTF7-3), cells were transfected with 1 µg of DNA and 15 µl of TM TM findings, when hERD2 and lysozyme-AARL were co-expressed, DOSPER per dish in 1 ml of OptiMEM . 5 hours post- the receptor was localised to perinuclear regions typical of the transfection, the cells were treated with toxin. Golgi, whilst lysozyme-AARL appeared to be distributed Toxin treatments and immunoprecipitations throughout the secretory pathway (data not shown). Conversely, Following transfection (as described above) cells were washed with when both hERD2 and lysozyme-KDEL were expressed in the phosphate-buffered saline (PBS) and quadruplicate portions of DT, PE same cells, both the receptor and lysozyme appeared to largely or SLT-1 added in DMEM (2.5% FCS) to individual dishes at the localise to the ER (data not shown). This indicates that the concentrations indicated. Cells were then incubated in the presence of receptor was able to bind the KDEL-tagged lysozyme and that toxin for 3 hours (DT and SLT-1) or 5 hours (PE), at 37°C. Following this binding caused the anticipated redistribution of hERD2 (and intoxication the level of lysozyme synthesis was quantified. COS 7 presumably the endogenous KDEL receptors) and lysozyme to cells were incubated for 30 minutes and HeLa cells for 10 minutes in the ER, in agreement with the earlier observations (Lewis and modified Eagle’s medium (MEM) deficient in methionine and cysteine, Pelham, 1992). and then radiolabelled for 1 hour or 20 minutes, respectively, with 25 µ 35 TM Ci [ S]Promix per dish. Cells were lysed in ice-cold lysis buffer Effects of additional KDEL ligands and KDEL (1% Triton X-100, 10 mM Tris, pH 7.4, 150 mM NaCl) containing protease inhibitors (CompleteTM tablets). The lysates were spun at receptors on toxin potency 15,000 g for 5 minutes, and 0.5 µg polyclonal anti-lysozyme antibodies PE, ST and DT exert their cytotoxic effects by inhibiting were added to the supernatants. Following a 2 hour incubation at 4°C, protein synthesis in target cells. Within certain concentration 470 M. E. Jackson and others ranges, increasing amounts of toxin will inhibit protein 120 synthesis in a dose-dependent fashion. Following the rate of protein synthesis in toxin-treated cells allows the construction 100

of a ‘kill curve’ (Fig. 1). In order to construct such a curve, and

s i

to subsequently use this approach to assess the effects of s

e 80 h

expressing either a receptor ligand or the KDEL receptor itself t

y s on protein synthesis in toxin-treated cells, it was necessary to

e 60 m

look at protein synthesis solely in the small population of y

o s

transfected cells (typically approx. 30%). This was y l 40 accomplished by transfecting cells with plasmids encoding % lysozyme-KDEL or lysozyme-AARL, or by co-transfecting cells with plasmids encoding lysozyme-AARL and the hERD2 20 gene for comparison with cells transfected with lysozyme- AARL alone. In all the experiments, the lysozyme variant was 0 1 10 100 used as a marker protein to monitor protein synthesis during a PE (ng/ml) short radioactivity pulse. In this way, we could be certain that only the transfected cell population was being analysed. Fig. 2. PE toxicity is inhibited in cells expressing lysozyme-KDEL Following toxin treatment of cells, both 35S-labelled lysozyme and increased in cells expressing additional KDEL receptors. variants could be immunoprecipitated from cell extracts using Quantitation of expressed lysozyme from PE-treated COS 7 cells polyclonal rabbit anti-chick lysozyme antibodies. The transiently expressing either lysozyme-AARL (open circles), immunoprecipitates were resolved by SDS-PAGE (Fig. 1A) lysozyme-KDEL (ﬁlled circles), or lysozyme-AARL together with additional KDEL receptors (open squares). Data are the means ± s.d. and the amount of lysozyme produced in toxin-treated cells, from four replicate samples. relative to that made in cells where no toxin was added, was quantiﬁed using a Phosphorimager. As a negative control we utilised DT, which is known to be into the cytosol by a mechanism that depends crucially on the endocytosed to early endosomes, from where it translocates low pH environment within this compartment (Draper and Simon, 1980; Sandvig and Olsnes, 1980). DT should not, therefore, utilise the KDEL retrieval system to facilitate its A entry into cells. Using the approach described above, we could NTC 0.1 1 5 20 50 100 ng/ml show that expression of lysozyme-KDEL (Fig. 1B) and expression of additional KDEL receptors (Fig. 1B) had no effect on the cytotoxicity of DT towards COS 7 cells. PE contains a KDEL-like motif at its C terminus and there is evidence from mutagenesis studies that this toxin is B recognised by the KDEL receptor (Chaudhary et al., 1990; Seetharam et al., 1991). Fig. 2 demonstrates that expression of 10 0 lysozyme-KDEL protects COS 7 cells from the cytotoxic effects of PE. The cells were less sensitive since greater

8 0 amounts of toxin were necessary to reduce the synthesis of the

i s

e lysozyme marker. This desensitisation was not seen when cells

t n

y express lysozyme-AARL (Fig. 2). The protection observed was s

6 0 e

m partial, since the approach taken cannot ensure total occupancy

z o

s of KDEL receptors by the lysozyme-KDEL molecules, and

y 4 0 l

presumably also because very little toxin is required in the % cytosol to promote cell death. The IC50 values (the dose 2 0 required to inhibit lysozyme synthesis by 50%) indicate an approximately fourfold decrease in sensitivity to PE for cells 0 expressing lysozyme-KDEL. Fig. 2 also demonstrates the 0.1 1 10 100 opposite effect, where expression of additional KDEL DT (ng/ml) receptors partially sensitised the cells to PE, the presence of Fig. 1. DT toxicity is unaffected by the overexpression of lysozyme- additional receptors reducing the IC50 value approximately KDEL, lysozyme-AARL or additional KDEL receptors. (A) SDS- ninefold. Taken together, these data strongly suggest an polyacrylamide gel showing the effect of diphtheria toxin on involvement of the KDEL retrieval system in the cytotoxicity lysozyme-AARL synthesis in COS 7 cells. Transient expression of of PE. lysozyme-AARL, exposure to DT and immunoprecipitations of SLT-1 does not contain a KDEL or KDEL-like motif. It has lysozyme was carried out as described in Materials and methods. been speculated that toxins lacking a KDEL signal may bind NTC denotes non-transfected cells. (B) Graphical representation of lysozyme immunoprecipitations from DT-treated COS 7 cells indirectly to KDEL receptors via intracellular KDEL- expressing either lysozyme-AARL (open circles), lysozyme-KDEL containing carriers, which might allow toxin to piggy-back in (ﬁlled circles), or lysozyme-AARL together with additional KDEL order to reach the ER (Pelham et al., 1992). However, receptors (open squares). Data are the means ± s.d. from four expression of KDEL-tagged lysozyme (Fig. 3) or of additional replicate samples. KDEL receptors (Fig. 3) in HeLa cells (HeLa cells were used Retrograde transport of Pseudomonas exotoxin A by the KDEL receptor 471

120 from non-injected cells by immunoﬂuorescence (also conﬁrmed by the presence of the injected coumarin-BSA) (Fig. 100 4A,B). We exploited this feature of microinjected Sar1(H79G)

to characterise an antibody raised against a polypeptide

s i

s representing the 21 amino acids of the C terminus of the KDEL

e 80

h t

n receptor, which is thought to be exposed to the cytoplasm (Tang

e 60 et al., 1993). Microinjected monovalent Fab fragments of this

m y

z antibody accumulated in a juxtanuclear region representing the

s y

l Golgi complex, as veriﬁed by co-staining with antibodies 40

% against the Golgi resident protein galactosyl transferase GalT

20 (not shown). Co-injection of the Fab fragments with Sar1(H79G) significantly inhibited the redistribution of the KDEL receptor to the ER in contrast to the presence of 0 0.001 0.01 0.1 1 10 Sar1(H79G) alone (Fig. 4). At 1.5 hours post-injection more SLT-1 (ng/ml) than 70% of the injected cells showed a juxtanuclear distribution of the KDEL receptor (Fig. 4C,D), which appeared Fig. 3. SLT-1 toxicity is unaffected by the overexpression of to only slowly recycle to the ER at later time points (Fig. 4E,F). lysozyme-KDEL or additional KDEL receptors. Quantitation of This suggests that binding of the anti-KDEL receptor expressed lysozyme from SLT-1-treated HeLa cells expressing either antibodies to the C terminus of the KDEL receptor interferes lysozyme-AARL (open circles), lysozyme-KDEL (filled circles), or lysozyme-AARL together with additional KDEL receptors (open with its retrograde transport, whereas in the presence of the squares). Data are the means ± s.d. from four replicate samples. Sar1(H79G) mutant alone the redistribution occurs rapidly (Fig. 4A,B). In contrast, anterograde transport of the secretory markers vesicular stomatitis virus glycoprotein (see Pepperkok for experiments involving SLT-1 since they have more surface et al., 1993, 1998) or CD8 (Shima et al., 1998; Pepperkok et receptors for this toxin than COS 7 cells) did not reduce the al., 1998) was not affected by microinjection of the KDEL cytotoxicity of SLT-1 compared to cells expressing only receptor antibodies (not shown). Therefore, the anti-KDEL lysozyme-AARL. These data, like those for DT, suggest that receptor antibodies should be an ideal reagent to directly SLT-1 is not exploiting the KDEL retrieval system to facilitate interfere in vivo with the cycling and thus function of the intracellular transport during cell entry. KDEL receptor in Golgi complex to ER transport. Antibodies against a cytoplasmic domain of the Antibodies against the KDEL receptor protect cells KDEL receptor prevent its retrograde transport to against PE but not against DT or SLT-1 the ER The results described so far have suggested that PE but not SLT COPII-coated vesicles are involved in anterograde transport exploit the KDEL receptor retrieval system to enter the from the ER to the Golgi complex in yeast (Barlowe et al., cytoplasm via the ER. In order to obtain more direct evidence 1994; Bednarek et al., 1995). The assembly of the COPII coat for this hypothesis we microinjected the Fab fragments of the is thought to be regulated by the small GTPase Sar1p (Kuge et anti-KDEL receptor antibodies into Vero cells and tested their al., 1994). Here we microinjected Vero cells with Sar1(H79G), effect on cell intoxication by various protein toxins. a trans-dominant negative mutant of the mammalian Preliminary experiments were performed to establish toxin homologue of the GTPase, which blocks COPII-mediated concentrations that reduced cellular protein synthesis in Vero export of secretory cargo from the ER in vivo (Pepperkok et cells to around 10-20% of control cells (data not shown). al., 1998) and in vitro (Kuge et al., 1994). Sar1 promotes Microinjection of the KDEL receptor antibodies gave no vesicle budding from the endoplasmic reticulum but not Golgi protection against DT and SLT-1, but significantly reduced the compartments (Kuge et al., 1994), but does not interfere with sensitivity of the cells to PE by approximately 2.5-fold (Fig. retrograde transport from the Golgi complex to the ER of 5). Unlike DT, both PE and SLT-1 are thought to proceed from cycling proteins like ERGIC53 (Shima et al., 1998). In non- the Golgi to the ER before translocating into the cytosol (see injected cells the KDEL receptor accumulated in a juxtanuclear reviews by Sandvig and van Deurs, 1994; Lord et al., 1994; region, representing the Golgi complex (Fig. 4A, indicated by Lord and Roberts, 1998). In keeping with this, pretreating Vero arrowheads). In contrast, similar to ERGIC53, a gradual cells with brefeldin A resulted in a 5.2- and 2.3-fold protection redistribution from the Golgi complex to the ER was observed against PE and SLT-1, respectively, although this treatment for the KDEL receptor in the presence of Sar1(H79G) (Fig. was, as expected, without effect on the sensitivity of the cells 4A,B). The appearance of the Golgi in the presence of to DT. In the case of PE and SLT-1, brefeldin treatment restored Sar1(H79G), visualised by immunofluorescence with protein synthesis in the presence of either toxin to the control antibodies against Golgi residents, appeared to be unaltered level seen in the absence of toxin: the apparent discrepancy compared to non-injected cells within at least 2 hours of between a 5.2- and a 2.3-fold protection simply indicates that microinjection (not shown; R. Pepperkok and T. E. Kreis, at the toxin concentrations used, PE reduced cellular protein unpublished) demonstrating that the redistribution of the synthesis to less than 20% of the control value while SLT-1 KDEL receptor was due to its cycling activity and not to a non- was less effective, reducing protein synthesis to approx. 40% specific redistribution of Golgi membranes. Therefore, in the of the control. Protein synthesis in the non-toxin-treated presence of Sar1(H79G) recycling proteins accumulate in the control cells was unaffected by the presence of brefeldin A, as ER, and thus microinjected cells can be easily discriminated indicated by the relative protein synthesis value of 1 (Fig. 6). 472 M. E. Jackson and others

In summary, our results demonstrate that inhibition of KDEL affect productive toxin uptake. If it remains uncertain as to receptor cycling from the Golgi complex to the ER by whether all of the toxins possessing putative retrieval signals microinjection of antibodies against its C terminus, inhibit actually utilise these in vivo, the situation with the toxins intoxication by PE (containing a KDEL-like sequence) but not lacking such signals is even less clear. It is possible that such the intoxication by SLT-1 (lacking any KDEL-like motif). Thus toxins are able to reach the ER as opportunistic fluid phase the data suggest that PE but not SLT-1 uses the KDEL retrieval cargo within vesicles following the retrograde pathway. This system to reach the cytosol via the ER. could account for the finding that cells containing trans- dominant negative mutants of Sar1 and Rab1 are partially resistant to free ricin A-chain, presumably taken into the cells DISCUSSION by fluid phase endocytosis (Simpson et al., 1996). While such a route to the ER would clearly be extremely inefficient, the It has been proposed that the compartment for the translocation exquisite potency of these protein toxins might ensure that of a number of protein toxins into the mammalian cell cytosol intoxication still ensues. However, since the cytotoxic potency is the ER (Pelham et al., 1992; Pastan et al., 1992; Sandvig et al., 1992), and establishing this has been the focus of a number of studies (for a review, see Lord and Roberts, 1998). In the productive routing pathway that leads to cellular intoxication, a small proportion of the endocytosed toxin reaches the TGN. Dissection of the mechanisms by which such toxins may be transported from the TGN to the ER is now of particular interest. It has been known for some time that within the early secretory pathway, vesicle trafficking between the ER and the Golgi is bi-directional (for example, see Pelham, 1989). This serves to recycle proteins involved in inter-compartmental vesicle targeting and to retrieve escaped resident ER proteins. It has been suggested that this retrograde vesicular trafficking might be exploited by protein toxins in order to reach the ER, possibly to allow them to subvert the ER- associated protein degradation pathway (Hiller et al., 1996; McCracken and Brodsky, 1996; Werner et al., 1996; Kopito, 1997) to be exported from the ER lumen to the cytosol (Lord, 1996; Hazes and Read, 1997). How are protein toxins transported from the TGN to the ER lumen? A number of toxins, such as Escherichia coli LT, CT and PE, contain C-terminal KDEL or KDEL-like motifs, which are retrieval signals normally found on resident proteins of the ER lumen. The KDEL retrieval system has been implicated in the routing of both PE and CT to the ER (Pastan et al., 1992; Majoul et al., 1996; Sofer and Futerman, 1996), although it Fig. 4. Effects of anti-KDEL receptor antibodies on Golgi-to ER-transport. Vero cells were should be noted that the B chain of CT microinjected with purified Sar1(H79G) (A,B; final concentration 1 mg/ml) and coumarin- (which does not possess the retrieval conjugated BSA (B; final concentration 2 mg/ml) as a co-injection marker to identify signal) has also been visualised injected cells; or they were injected with Sar1(H79G) and monovalent Fab fragments of anti- KDEL receptor antibodies (C-F). Immediately after injection, cells were incubated at 37°C throughout the Golgi stack (Sandvig et al., for 1.5 (A-D) or 2.5 hours (E,F) before they were fixed and immunostained for endogenous 1994). The data for LT are contradictory, KDEL receptor (A,C,E) and microinjected Fab fragments (D,F). Asterisks indicate with reports that disruption of the RDEL microinjected cells. Arrowheads in A, C and D indicate the juxtanuclear accumulation of the tetrapeptide both does (Lencer et al., KDEL receptor (A,C) or injected antibodies (D). The arrow in E points to some KDEL 1995) and does not (Ciepak et al., 1995) receptor staining remaining in the juxtanuclear region 2.5 hours after injection. Bar, 15 µm. Retrograde transport of Pseudomonas exotoxin A by the KDEL receptor 473

2.5 6

) )

s 5

i l l

2 s

t d

n 4

1.5 e

- e

t 3

/ n

p 1

d v

e 2

( n

r 0.5 1

i (

0 0 Control DT PE SLT-1 Control DT PE SLT-1 Fig. 5. Microinjection of Vero cells with an antibody to a Fig. 6. BFA dramatically reduces the toxicity of PE and SLT-1 to cytoplasmic domain of the KDEL receptor reduces the toxicity of Vero cells, but is without effect on DT. Cells were pre-treated with 5 PE, but not that of DT or SLT-1. Cells were injected with antibody or µg/ml BFA or left untreated for 30 minutes prior to incubation with not injected prior to a 4 hour treatment with the toxins as shown. The toxins, as shown, for 4 hours. The amount of toxins used were 0.2 amounts of toxins used were 0.2 ng/ml (DT), 500 ng/ml (PE) and 0.2 ng/ml (DT), 500 ng/ml (PE) and 0.2 ng/ml (SLT-1). Control cells ng/ml (SLT-1). These values had previously been determined to were not treated with toxin. A 30 minute pulse with 35S-methionine, reduce cellular protein synthesis to 20% of that seen in non-toxin- followed by TCA precipitation of proteins and scintillation counting treated control cells. The control cells used here were not treated was used to determine the level of protein synthesis in the cells after with any toxin. Protein synthesis (counts per cell) was determined this time. The data graphs depict the relative levels of protein following a 30 minute pulse with 35S-methionine, TCA precipitation synthesis (BFA-treated / non-BFA-treated) in the cells treatments, of the proteins, then scintillation counting. The data graphs show the and are the average of two independent experiments. relative protein synthesis values (injected / non-injected) of the cells following the various treatments, and are the average of two independent microinjection experiments. endosomes. Although PE has been shown, by toxin mutagenesis studies, to possess a critical KDEL-like sequence (REDLK) at its C terminus, the effects on cytotoxic potency of of toxins lacking a KDEL-like motif is the same as those that modulating available KDEL receptors in vivo has never before do have this signal, it seems likely that all the toxins probably been demonstrated. The terminal lysine residue of the REDLK interact with some natural carrier to facilitate retrograde sequence would predictably prevent interaction with the KDEL transport to the ER. This could be achieved through a direct receptor, and this has been demonstrated in vitro by assessing interaction with, for example, the KDEL receptor, or by the binding of synthetic peptides to membranes containing the interacting with another component that itself has an ER receptor (Kreitman and Pastan, 1995). It has recently been retrieval signal. shown that the terminal lysine residue of native PE is removed Here we have investigated a possible role for the KDEL during incubation with target cells (Hessler and Kreitman, retrieval system in the entry mechanisms of PE (a toxin which 1997). In the present study we show that perturbation of the does contain a KDEL-like motif at its C terminus), and E. coli KDEL retrieval system can disrupt or enhance the potency of SLT-1 (which does not possess such a sequence). This was native PE in a way that correlates with a dependence on the done by transiently expressing a population of KDEL-tagged availability of Golgi-located KDEL receptors. We have lysozyme proteins or additional KDEL receptors in COS 7 and confirmed that retrograde transport from the Golgi to the ER HeLa cells. Thus the population of lysozyme-KDEL proteins is required for toxicity, since blocking this transport by should compete with other KDEL-terminating proteins for microinjecting antibodies that interact with the cytoplasmically endogenous receptors. Should a significant population of exposed region of the receptor protects cells against PE (Fig. receptors be occupied in this way during the time frame of the 5). Although SLT-1 also must pass through the Golgi on route experiments, it was rationalised that receptors would not be to the ER (Fig. 6), it has no apparent requirement for the KDEL available to those toxins that require interaction for productive receptor itself (Fig. 5). entry. Likewise, the expression of additional KDEL receptors Exploiting the KDEL retrieval system is not the only way would be anticipated to increase the potency of such toxins. that protein toxins can reach the ER from the TGN. Our data Immunofluorescence confirmed expression of both show that SLT-1 and presumably ST (although produced by lysozyme-KDEL and myc-tagged KDEL receptors (data not different organisms, these related toxins can be regarded as shown). Co-expression resulted in the redistribution of both essentially identical since they differ in only 1 amino acid receptor and lysozyme to the ER, exactly as described earlier residue), do not require the KDEL retrieval system to reach the (Lewis and Pelham, 1992). The cytotoxicity data (Figs 1-3) ER, raising the question of how they accomplish this retrograde clearly demonstrate that the KDEL retrieval system is transport. Transport of these lipid-binding toxins might be irrelevant for DT and SLT-1, but that it is required for optimal somehow dependent on the lipid tails of the cell surface toxin entry of PE. That neither DT or SLT-1 cytotoxicities were receptors. SLT-1 and ST bind to the glycosphingolipids known affected by expressing KDEL ligands or additional KDEL as globotriose-(α-D-galactose-(1→4)-β-D-galactose-(1→4)- receptors would appear to rule out any indirect effects of these β-D-glucose-(1→4))ceramides, which are heterogeneous with proteins caused by, for example, increased ER-Golgi vesicle regards to the length and saturation of their lipid tails. In cells traffic. The results with DT are predictable, since this toxin is sensitised to ST where the toxin can be visualised in the ER, known to translocate into the cytosol from acidified it has been shown that the lipids have longer fatty acid tails 474 M. E. Jackson and others

(Sandvig et al., 1994). Since it is known that lipids can undergo Cosson, P. and Letourneur, F. (1994). Coatomer interaction with di-lysine retrograde transport from the Golgi to the ER (Hoffmann and endoplasmic reticulum retention motifs. Science 263, 1629-1631. Pagano, 1993), transport of SLT-1 might therefore occur from Draper, R. K. and Simon, M. I. (1980). The entry of diphtheria toxin into the mammalian cell cytoplasm: evidence for lysosomal involvement. J. Cell the cell surface to the ER via the original interaction with Biol. 87, 849-854. surface lipid receptors. It has also been reported that the Endo, Y., Mitsui, K., Motizuki, M. and Tsurugi, K. (1987). The mechanism addition of KDEL to the C terminus of SLT-1 B chain does not of action of ricin and related toxins on eukaryotic ribosomes. J. Biol. Chem. enhance the rate of transport to the ER (Johannes et al., 1997). 262, 5908-5912. Fuerst, T. R., Niles, E. G., Studier, W. and Moss, B. (1986). Eukaryotic Assuming that the SLT-1 B chain KDEL tag was accessible to transient-expression system based on recombinant vaccinia virus that recycling KDEL receptors, this observation suggests that SLT- synthesizes bacteriophage T7 RNA polymerase. Proc. Natl. Acad. Sci. USA 1 B chain is indeed transported to the ER regardless of any 83, 8122-8126. interaction with the KDEL receptors, a conclusion consistent Goda, Y. and Pfeffer, S. R. (1988). Selective recycling of the mannose-6- phosphate/IGF-II receptor to the trans-Golgi network in vitro. Cell 55, 309- with the data presented here. 320. At present little is known about the vesicular carriers that Griffiths, G., Ericsson, M., Krijnse-Locker, J., Nilsson, T., Goud, B., transport toxins to the ER. Because of their recognised role in Söling, H.-D., Tang, B. L., Wong, S. H. and Hong, W. (1994). Localization retrograde transport (Cosson and Letourneur, 1994; Letourneur of the Lys, Asp, Glu, Leu tetrapeptide receptor to the Golgi complex and the intermediate compartment in mammalian cells. J. Cell Biol. 127, 1557- et al., 1994), COPI-coated vesicles are potential carriers. The 1574. KDEL receptor has been found in retrograde COPI-coated Hazes, B. and Read, R. J. (1997). Accumulating evidence suggests that vesicles that bud from all cisternae in the Golgi stack (Orci et several AB-toxins subvert the endoplasmic reticulum-associated protein al., 1997). Experiments to analyse the nature of the retrograde degradation pathway to enter target cells. Biochem. 36, 11051-11054. Hessler, J. L. and Kreitman, R. J. (1997). An early step in Pseudomonas carriers of these toxins are currently underway. exotoxin action is removal of the terminal lysine residue, which allows In conclusion, and in agreement with circumstantial binding to the KDEL receptor. Biochem. 36, 14577-14582. evidence from other studies, the results presented here show Hiller, M. M., Finger, A., Schweiger, M. and Wolf, D. H. (1996). ER that PE, in contrast to SLT-1, exploits the KDEL retrieval degradation of a misfolded luminal protein by the cytosolic ubiquitin/proteasome pathway. Science 273, 1725-1728. system to facilitate its delivery to the ER prior to membrane Hoffmann, P. M. and Pagano, R. E. (1993). Retrograde movement of translocation. Use of an entry pathway and of a retrieval system membrane lipids from the Golgi apparatus to the endoplasmic reticulum of operating from the distal Golgi to the ER supports the notion perforated cells: evidence for lipid recycling. Eur. J. Cell Biol. 60, 371-375. that this reversal of the secretory pathway does not exist solely Iglewski, B. H. and Kabat, D. (1975). NAD-dependent inhibition of protein synthesis by Pseudomonas aeruginosa toxin. Proc. Natl. Acad. Sci. USA 72, for the opportunistic entry of protein toxins, but that it is a route 2284-2288. that may be utilised to some unknown extent by Jackson, M. R., Nilsson, T. and Peterson, P. A. (1993). Retrieval of physiologically relevant proteins. transmembrane proteins to the endoplasmic reticulum. J. Cell Biol. 121, 317-333. We thank Mike J. Lewis and Hugh R. B. Pelham for supplying Johannes, L., Tenza, D., Antony, C. and Goud, B. (1997). Retrograde plasmids, and John M. Jarvis for 9E10 antibodies (MRC Laboratory transport of KDEL-bearing B-fragment of Shiga toxin. J. Biol. Chem. 272, 19554-19561. of Molecular Biology, Cambridge). We are also grateful to Drs B. L. Kline, A. A. and Lingwood, C. A. (1994). Capping and receptor-mediated Tang and S. H. Wong (Institute of Molecular and Cell Biology, endocytosis of cell-bound Verotoxin (Shiga-like toxin)-1: chemical Singapore) for providing the monoclonal anti-KDEL receptor identiﬁcation of an amino acid in the B subunit necessary for efficient antibody used to stain endogenous KDEL receptor. In addition, we receptor glycolipid binding and cellular internalization. J. Cell. Physiol. 161, thank Dr D. Shima (Cell Biology Laboratory, Imperial Cancer 319-332. Research Fund, London) for anti-galactosyl transferase antibodies Kopito, R. R. (1997). ER quality control: the cytoplasmic connection. Cell 88, (GalT). Finally we also acknowledge the help of Debbie Lyon, Alex 427-430. Stokes and Peter Jordan (Light Microscopy Laboratory, ICRF, Kreitman, R. J. and Pastan, I. (1995). Importance of the glutamate residue of the KDEL in increasing the cytotoxicity of Pseudomonas exotoxin London). This work was supported by grant 88/T02035 from the UK derivatives and for increased binding to the KDEL receptor. Biochem. J. 307, Biotechnology and Biological Sciences Research Council. 29-37. Kuge, O., Dascher, C., Orci, L., Rowe, T., Amherdt, M., Plutner, H., Ravazzola, M., Tanigawa, T., Rothman, J. E. and Balch, W. E. (1994). REFERENCES Sar1 promotes vesicle budding from the endoplasmic reticulum but not Golgi compartments. J. Cell Biol. 125, 51-65. Lencer, W. I., Constable, C., Jobling, M. G., Webb, H. M., Ruson, S. R., Barlowe, C., Orci, L., Yeung, T., Hosobuchi, M., Hamamoto, S., Salama, Madara, J. L., Hirst, T. R. and Holmes, R. K. (1995). Targeting of cholera N., Rexach, M. F., Ravazzola, M., Amherdt, M. and Schekman, R. toxin and Escherichia coli heat-labile toxin in polarised epithelia: role of (1994). COPII: a membrane coat formed by Sec proteins that drive vesicle COOH-terminal KDEL. J. Cell Biol. 131, 951-962. budding from the endoplasmic reticulum. Cell 77, 895-907. Letourneur, F., Gaynor, E. C., Henneke, S., Demolliere, C., Duden, R., Bednarek, S. Y., Ravazzola, M., Hosobuchi, M., Amherdt, M., Perrelet, A., Emr, S. D., Riezman, H. and Cosson, P. (1994). Coatomer is essential for Schekman, R and Orci, L. (1995). COPI- and COPII-coated vesicles bud retrieval of dilysine-tagged proteins to the endoplasmic reticulum. Cell 79, directly from the endoplasmic reticulum in yeast. Cell 83, 1183-1196. 1199-1207. Bos, K., Wraight, C. and Stanley, K. K. (1993). TGN38 is maintained in the Lewis, M. J. and Pelham, H. R. B. (1992). Ligand induced redistribution of trans-Golgi network by a tyrosine-containing motif in the cytoplasmic a human KDEL-receptor from the Golgi complex to the endoplasmic domain. EMBO J. 12, 2219-2228. reticulum. Cell 68, 353-364. Chaudhary, V. K., Jinno, Y., Fitzgerald, D. and Pastan, I. (1990). Lord, J. M. (1996). Protein degradation: go outside and see the proteasome. Pseudomonas exotoxin contains a speciﬁc sequence at the carboxyl terminus Curr. Biol. 6, 1067-1069. that is required for cytotoxicity. Proc. Natl. Acad. Sci. USA 87, 308-312. Lord, J. M. and Roberts, L. M. (1998). Toxin entry: retrograde transport Ciepak, W., Messer, R. J., Konkel, M. E. and Grant, C. C. R. (1995). Role through the secretory pathway. J. Cell Biol. 140, 1-4. of potential endoplasmic reticulum retention sequence (RDEL) and Golgi Lord, J. M., Roberts, L. M. and Robertus, J. D. (1994). Ricin: structure, complex in the cytotoxic activity of Escherichia coli heat-labile enterotoxin. mode of action, and some current applications. FASEB J. 8, 201-208. Mol. Microbiol. 16, 789-800. Majoul, I. V., Bastiaens, P. I. H. and Söling, H.-D. (1996). Transport of an Retrograde transport of Pseudomonas exotoxin A by the KDEL receptor 475

external Lys-Asp-Glu-Leu (KDEL) protein from the plasma membrane to Golgi transport in living cells reveals a sequential mode of action for COPII the endoplasmic reticulum: studies with Cholera toxin in Vero cells. J. Cell and COPII. Cell 90, 1137-1148. Biol. 133, 777-789. Schafer, W., Stroh, H. F., Berghofer, S., Seiler, J., Vey, M., Kruse, M.-L., McCracken, A. A. and Brodsky, J. L. (1996). Assembly of ER-associated Kern, H. F., Klenk, H.-D. and Garten, W. (1995). Two independent protein degradation in vitro: dependence on cytosol, calnexin and ATP. J. targeting signals in the cytoplasmic domain determine trans-Golgi network Cell Biol. 132, 291-298. localization and endosomal trafficking of the proprotein convertase furin. Miesenböck, G. and Rothman, J. E. (1995). The capacity to retrieve escaped EMBO J. 14, 2424-2435. ER proteins extends to the trans-most cisterna of the Golgi stack. J. Cell Schutze, M. P., Peterson, P. A. and Jackson, M. R. (1994). An N-terminal Biol. 129, 309-319. double-arginine motif maintains type II membrane proteins in the Munro, S. and Pelham, H. R. B. (1987). A C-terminal signal prevents endoplasmic reticulum. EMBO J. 13, 1696-1705. secretion of luminal ER proteins. Cell 48, 899-907. Seetharam, S., Chaudhary, V. K., Fitzgerald, D. and Pastan, I. (1991). Orci, L., Stammes, M., Ravazzola, M., Amherdt, M., Perrelet, A., Sollner, Increased cytotoxic activity of Pseudomonas exotoxin and two chimeric T. H. and Rothman, J. E. (1997). Bidirectional transport by distinct toxins ending in KDEL. J. Biol. Chem. 266, 17367-17381. populations of COPI-coated vesicles. Cell 90, 335-349. Shima, D. T., Cabrera-Poch, N., Pepperkok, R. and Warren, G. (1998). An Pappenheimer, A. M. (1977). Diphtheria toxin. Annu. Rev. Biochem. 46, 69- ordered inheritance strategy for the Golgi apparatus: visualization of mitotic 94. disassembly reveals a role for the mitotic spindle. J. Cell Biol. 141, 955- Pastan, I., Chaudhary, V. and FitzGerald, D. J. (1992). Recombinant toxins 966. as novel therapeutic agents. Annu. Rev. Biochem. 61, 331-354. Simpson, J. C., Dascher, C., Roberts, L. M., Lord, J. M. and Balch, W. E. Pelham, H. R. B. (1989). Control of protein exit from the endoplasmic (1995). Ricin cytotoxicity is sensitive to recycling between the endoplasmic reticulum. Annu. Rev. Cell Biol. 5, 1-23. reticulum and the Golgi complex. J. Biol. Chem. 270, 20078-20083. Pelham, H. R. B., Roberts, L. M. and Lord, J. M. (1992). Toxin entry: how Simpson, J. C., Roberts, L. M. and Lord, J. M. (1996). Free ricin A chain reversible is the secretory pathway? Trends Cell Biol. 2, 183-185. reaches an early compartment of the secretory pathway before it enters the Pepperkok, R., Scheel, J., Horstmann, H., Hauri, H. P., Griffiths, G. and cytosol. Expl. Cell Res. 229, 447-451. Kreis, T. E. (1993). Beta-COP is essential for biosynthetic membrane Simpson, J. C., Smith, D. C., Roberts, L. M. and Lord, J. M. (1998). transport from the endoplasmic reticulum to the Golgi complex in vivo. Cell Expression of mutant dynamin protects cells against diphtheria toxin but not 74, 71-82. against ricin. Expl. Cell Res. 239, 293-300. Pepperkok, R., Lowe, M., Burke, B. and Kreis, T. E. (1998). Three distinct Sofer, A. and Futerman, A. H. (1996). Rate of retrograde transport of cholera steps in transport of vesicular stomatitis virus glycoprotein from the ER to toxin from the plasma membrane to the Golgi apparatus and endoplasmic the cell surface in vivo with differential sensitivities to GTPγS. J. Cell Sci. reticulum decreases during neuronal development. J. Neurochem. 67, 2134- 111, 1877-1888. 2140. Peter, F., Nguyen, V. P. and Söling, H.-D. (1992). Different sorting of Lys- Sugai, M., Chen, C-h. and Wu, H. C. (1992a). Bacterial ADP- Asp-Glu-Leu proteins in rat liver. J. Biol. Chem. 267, 10631-10637. ribosyltransferase with a substrate-speciﬁcity of the Rho-protein Pilon, M., Schekman, R. and Römisch, K. (1997). Sec61p mediates export disassembles the Golgi-apparatus in Vero cells and mimics the action of of a misfolded secretory protein from the endoplasmic reticulum to the BFA. Proc. Natl. Acad. Sci. USA 89, 8903-8907. cytosol for degradation. EMBO J. 16, 4540-4548. Sugai, M., Chen, C-h. and Wu, H. C. (1992b). Staphylococcal ADP- Plemper, R. H., Bohmler, S., Bordallo, J., Sommer, T. and Wolf, D. H. ribosyltransferase-sensitive small G-protein is involved in brefeldin A (1997). Mutant analysis links the translocon and BiP to retrograde protein action. J. Biol. Chem. 267, 21297-21299. transport for ER degradation. Nature 388, 891-895. Tang, B. L., Wong, S. H., Qi, X. L., Low, S. H. and Wong, H. J. (1993). Rapak, A., Falnes, P. O. and Olsnes, S. (1997). Retrograde transport of ricin Molecular cloning, characterization, subcellular localization and dynamics to the endoplasmic reticulum with subsequent translocation to cytosol. Proc. of p23, the mammalian KDEL receptor. J. Cell Biol. 120, 325-338. Natl. Acad. Sci. USA 94, 3783-3788. Townsley, F. M., Wilson, D. W. and Pelham, H. R. B. (1993). Mutational Sandvig, K., Garred, O., Prydz, K., Kolzov, J. V., Hansen, S. H. and van analysis of the human KDEL receptor: distinct structural requirements for Deurs, B. (1992). Retrograde transport of endocytosed Shiga toxin to the Golgi retention, ligand binding and retrograde transport. EMBO J. 12, 2821- endoplasmic reticulum. Nature 358, 510-512. 2829. Sandvig, K., Garred, O. and van Deurs, B. (1996). Thapsigargin-induced van Deurs, B., Petersen, O. W., Olsnes, S. and Sandvig, K. (1987). Delivery transport of cholera toxin to the endoplasmic reticulum. Proc. Natl. Acad. of internalized ricin from endosomes to cisternal Golgi elements is a Sci. USA 93, 12339-12343. discontinuous, temperature-sensitive process. Expl. Cell Res. 171, 137-152. Sandvig, K. and Olsnes, S. (1980). Diphtheria toxin entry into cells is van Deurs, B., Sandvig, K., Peterson, O. W., Olsnes, S., Simons, K. and facilitated by low pH. J. Cell Biol. 87, 828-832. Griffiths, G. (1988). Estimation of the amount of internalised ricin that Sandvig, K., Olsnes, S., Brown, J. E., Petersen, O. W. and van Deurs, B. reaches the trans-Golgi network. J. Cell Biol. 106, 253-267. (1989). Endocytosis from coated pits of Shiga toxin: a glycolipid-binding Wales, R., Roberts, L. M. and Lord, J. M. (1993). Addition of an protein from Shigella dysenteriae. J. Cell Biol. 108, 1331-1343. endoplasmic reticulum retrieval sequence to ricin A chain signiﬁcantly Sandvig, K., Prydz, K., Hansen, S. H. and van Deurs, B. (1991a). Ricin increases its cytotoxicity to mammalian cells. J. Biol. Chem. 268, 23986- transport in brefeldin A-treated cells: correlation between Golgi structure 23990. and toxic effects. J Cell Biol. 115, 971-981. Werner, E. D., Brodsky, J. L. and McCracken, A. A. (1996). Proteasome- Sandvig, K., Prydz, K., Ryd, M. and van Deurs, B. (1991b). Endocytosis dependent endoplasmic reticulum-associated protein degradation: An and intracellular transport of the glycolipid-binding ligand Shiga toxin in unconventional route to a familiar fate. Proc. Natl. Acad. Sci. USA 93, polarized MDCK cells. J. Cell Biol. 113, 553-562 13797-13801. Sandvig, K., Ryd, M., Garred, O., Schweda, E., Holm, P. K. and van Deurs, Wiertz, E. J. H. J., Torotella, D., Bogyo, M., Mothes, W., Jones, T. R., B. (1994). Retrograde transport from the Golgi complex to the ER of both Rapoport, T. A. and Ploegh, H. L. (1996). Sec61-mediated transfer of a Shiga toxin and the non-toxic Shiga B fragment is regulated by butyric acid. membrane protein from the endoplasmic reticulum to the proteasome for J. Cell Biol. 126, 53-64. destruction. Nature 384, 432-438. Sandvig, K. and van Deurs, B. (1994). Endocytosis and intracellular sorting Wilson, D. W., Lewis, M. J. and Pelham, H. R. B. (1993). pH-dependent of ricin and Shiga toxin. FEBS Lett. 346, 99-102. binding of KDEL to its receptor in vitro. J. Biol. Chem. 268, 7465-7468. Sandvig, K. and van Deurs, B. (1996). Endocytosis, intracellular transport, Yoshida, T., Chen, C., Zhang, M. and Wu, H. C. (1991). Disruption of the and cytotoxic action of Shiga toxin and ricin. Physiol. Rev. 76, 949-966. Golgi apparatus by brefeldin A inhibits the cytotoxicity of ricin, modeccin Scales, S. J., Pepperkok, R. and Kreis, T. (1997). Visualization of ER-to- and Pseudomonas toxin. Expl. Cell Res. 192, 389-395.