Protein Kinase Cδ Is Activated by Shiga Toxin and Regulates Its Transport*

Protein kinase C (PKC) isozymes regulate different vesicular trafficking steps in the recycling or degradative pathways. However, a possible role of these kinases in the retrograde pathway from endosomes to the Golgi complex has previously not been investigated. We report here the involvement of a specific PKC isozyme, PKCδ, in the intracellular transport of the glycolipid-binding Shiga toxin (Stx), which utilizes the retrograde pathway to intoxicate cells. Upon binding to cells, Stx was shown to specifically activate PKCδ and not PKCα. The involvement of PKCδ and PKCα in the retrograde transport of Stx was then monitored biochemically and by immunofluorescence after inhibition or depletion of the isozymes. PKCδ, but not PKCα, was shown to selectively regulate the endosome-to-Golgi transport of StxB. Upon inhibition or knockdown of PKCδ, StxB molecules colocalized less with giantin and more with EEA1, indicating that the molecules were accumulated in endosomes, unable to reach the Golgi complex. The inhibition of Golgi transport of Stx was reflected by a strong reduction in the toxic effect, demonstrating that transport of Stx to the cytosol is dependent on PKCδ activity. These results are in agreement with our previous data, which show that Stx is able to stimulate its own transport.

Shiga toxin (Stx) 2 is an AB 5 -toxin consisting of an enzymatically active A-subunit noncovalently linked to a stable pentamer of B-chains (StxB) that binds to the glycosphingolipid Gb3 and mediates transport of the toxin (1). To reach its intracellular target, the toxin has to be endocytosed and transported via early endosomes (EEs) to the Golgi complex. The toxin is then transported further to the endoplasmic reticulum from where the A-subunit is translocated to the cytosol (for reviews see Refs. 1 and 2). In the cytosol the protein synthesis is inhibited by enzymatic modification of the 28 S rRNA. Endocytosis of the toxin is mediated by a clathrin-dependent process in several cell types, although clathrin-independent mechanisms are also partly involved (2). The toxin is transported to the EE, which is considered a sorting station for many endocytosed molecules. Some receptors are recycled to the plasma membrane to be recharged with ligands, like the transferrin receptor, whereas others enter the late endocytic pathway for degradation, like the epidermal growth factor receptor. Stx has been shown to utilize the retrograde pathway, which deviates from the recycling and degradative pathways in the EE and allows the toxin to be directly transported to the trans-Golgi network (TGN) and then further to the Golgi and the endoplasmic reticulum (3). This pathway is also used for retrieval of endogenous proteins, like TGN46 and mannose 6-phosphate receptors (reviewed in Ref. 4). Several proteins have been shown to regulate the transport of StxB in this pathway (4). Recently, microtubuli and dynein were also shown to be involved in Stx trafficking, and interestingly, Stx seemed to be able to stimulate microtubule assembly (5). Importantly, the requirements for retrograde transport of various protein toxins such as Stx and ricin differ, suggesting that there is more than one route from EE to the TGN (6,7).
Stx has been shown to induce signaling cascades leading to apoptosis in several cell types (reviewed in Ref. 8). In some cells the induction of apoptosis is mediated by the ribotoxic stress induced by the A 1 -chain after entry into the cytosol, whereas in other cells both Stx and the Stx B-subunit have been shown to induce apoptosis independent of intracellular transport (8). Several kinases are rapidly activated by Stx, such as the Src kinases Yes (9,10) and Lyn (11), and the tyrosine kinase Syk (11,12). In human renal carcinoma-derived ACHN (human, Caucasian, kidney, adenocarcinoma) cells, Stx has been shown to induce phosphorylation of actinbinding proteins such as paxillin and ezrin and also to increase the levels of cortical actin and microtubuli (13). Importantly, we have shown that Stx is able to stimulate its own clathrin-mediated entry in a process dependent on toxin concentration (14,15) and by induction of signaling cascades that leads to tyrosine phosphorylation of clathrin and activation of Syk (12). Moreover, in the monocytic cell line THP-1, Shiga-like toxin 1 was reported to activate protein kinase C (PKC) (16), and the involvement of PKC in toxin traffic has previously been indicated by the demonstration of increased sensitivity to several toxins after stimulation of cells with 12-O-tetradecanoylphorbol-13-acetate (17), including Stx. 3 In this study we therefore wanted to investigate the role of PKC isozymes in Stx transport and whether Stx is able to activate these kinases to modulate its own transport.
PKC isozymes are serine-threonine kinases that are implicated in diverse cellular functions, such as growth, differentiation, apoptosis, motility, ruffling, and vesicular trafficking (for reviews see Refs. 18 and 19). The isozymes can be divided into subgroups based on structure and cofactor requirements. The classic PKC (cPKC) isozymes (␣, ␤ I , ␤ II , and ␥) are activated by Ca 2ϩ and 1,2-diacylglycerol, the novel PKC (nPKC) isozymes (␦, ⑀, , and ) lack the Ca 2ϩ -binding domain, but are activated by 1,2-diacylglycerol, whereas the atypical PKC (aPKC) isozymes ( and /) are both Ca 2ϩ -and 1,2-diacylglycerol-independent but might be activated by inositol 1,4,5-trisphosphate (18,19). Different PKC isozymes have been shown associated with several intracellular structures involved in vesicular trafficking such as caveolae, multiple endocytic compartments, the Golgi complex, and lysosomes. PKC activity has been shown to regulate endocytosis of an increasing number of ligands and to regulate different vesicular trafficking steps in the recycling or degradative pathways (reviewed in Ref. 20). However, a role of PKC isozymes in the retrograde pathway has previously not been reported. Recently, both PKC␣ and PKC␦ were shown to mediate correct trafficking of Fc␣R vesicles (21), and we therefore wanted to investigate whether PKC␣or -␦ activity were implicated in targeting Stx into the retrograde pathway. First, the ability of StxB to rapidly activate these isozymes was determined, and we found that PKC␦ was specifically activated by StxB binding but not PKC␣. Then, PKC␣ and -␦ were inhibited or depleted by the use of specific drugs or siRNA oligonucleotides, and the main Stx transport steps were monitored; the endocytosis, the endosome-to-Golgi transport, and the toxicity exerted in the cytosol. PKC␦ was found to regulate the endosome-to-Golgi transport of Stx, whereas PKC␣ activity did not seem to be involved in any of the intracellular transport steps. Upon PKC␦ inhibition or depletion the StxB en route to the Golgi accumulated in endocytic structures early in the transport pathway. The inhibition of endosome-to-Golgi transport of Stx was reflected by reduced toxicity of Stx. These results show that PKC␦ is an important regulator of Stx trafficking and are in agreement with the idea that Stx is able to stimulate its own intracellular transport.

EXPERIMENTAL PROCEDURES
Materials-Hepes, bovine serum albumin, MESNa (mercaptoethanesulfonic acid), n-octylglucopyranoside, and tetracycline were purchased from Sigma. Na 2 35 SO 4 and [ 3 H]leucine were from Amersham Biosciences (Little Chalfont, Buckinghamshire, UK). Mouse anti-Stx antibodies (13C4 and 3C10) were from Toxin Technology (Sarasota, FL), rabbit anti-PKC␦ antibody was from Santa Cruz Biotechnology (Santa Cruz, CA) and mouse anti-␣-tubulin antibody and mouse anti-ricin anti-body were from Sigma. Rottlerin, BIM, Gö6976 (Calbiochem) were resuspended in Me 2 SO, and frozen aliquots were stored at Ϫ20°C. When used in experiments, the Me 2 SO concentration never exceeded 0.2%. Shiga toxin (Stx) was provided by Dr. J. V. Kozlov (Academy of Sciences of Russia, Moscow, Russia) and by Dr. J. E. Brown (U.S. Army Medical Research Institute of Infectious Diseases, Fort Detrick, MD). The plasmid encoding StxB-sulf 2 was a kind gift from Dr. B. Goud (Institute Curie, Paris, France). The protein concentrations in the lysates were determined by the BCA protein assay (Pierce) with the use of bovine serum albumin as the standard.
Cells and Transfection-HeLa, HEp-2, and Vero cells were grown under 5% CO 2 in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum, 100 units/ml penicillin, 100 g/ml streptomycin, and L-glutamine at 2 mM. For sulfation experiments, the cells were seeded out in 6-well plates at a density of 2 ϫ 10 5 cells/well and grown for 24 h before the experiments. For endocytosis and toxicity experiments the cells were seeded out in 24-well plates at a density of 4 ϫ 10 4 cells/ well and grown for 24 h before the experiments. Two different siRNAs targeting non-overlapping parts of the mRNA sequence were designed for PKC␦ and one for PKC␣. They were selected according to their physicochemical profiles, their specificity to the target mRNA by BLAST analysis profile (www.ncbi.nlm.nih.gov/BLAST/), and, finally, fitted as closely as possible to the criteria depicted previously (22). The sequences for the PKC␦ targeting constructs were GGCUA-CAAAUGCAGGCAAU and AACUCUUCGAGUCCAUC-CGUGU (siRNA 1 and 2, respectively), and for the PKC␣ targeting construct UGUGACACCUGCGAUAUGA (named siRNA 2). The constructs were ordered high-performance liquid chromatography-purified from MWG Biotech (Ebersberg, Germany). We also used the PKC␣-specific oligonucleotide AAAGGCUGAGGUUGCUGAU (23) named PKC␣ siRNA 1 in this report, the PKC␦-specific oligonucleotide CCAU-GAGUUUAUCGCCACCUU (Santa Cruz Biotechnology, Santa Cruz, CA), named PKC␦ siRNA 3 in this report or a negative control (Eurogentec, Seraing, Belgium). For siRNA transfection the cells were seeded out without antibiotics, grown for 24 h, and transfected using Oligofectamine (Invitrogen) according to the manufacturer's procedure. After 5 h of transfection, the medium was changed to complete growth medium containing serum and antibiotics, and the cells were grown for 2 or 3 days before the experiments, as indicated in the figure legends.
Determination of Shiga Toxin and Ricin Cytotoxicity-Cytotoxicity of Shiga toxin and ricin was determined by the reduction of protein synthesis in toxin-treated cells. HeLa cells seeded out in 24-well plates were washed twice in leucine-free Hepes medium, then the cells were preincubated with the different PKC inhibitors in the same medium for 30 min, before increasing concentrations of the toxins were added. The incubation was continued for 1.5 h (Stx) or 2 h (ricin), then the medium was replaced with leucine-free Hepes medium containing 2 Ci/ml [ 3 H]leucine, and the cells were incubated for a further 15 min. The proteins were extracted with 5% trichloroacetic acid, washed once in trichloroacetic acid, and then dis-solved in 0.1 M KOH. The incorporation of radioactivity was finally quantified.
Endocytosis Experiments-Measurements of internalization of 125 I-labeled ricin and transferrin were performed essentially as described previously (24). The endocytosis of Stx was quantified using a modified version of the procedure described previously (12,15), which takes advantage of a special ruthenium label quantitated in a highly specialized electro-chemiluminescent detection instrument provided by BioVeris Corp. (Gaithersburg, MD). Briefly, Stx was biotinylated with the reducible EZ-Link Sulfo-NHS-SS-Biotin (Pierce Biotechnology, Rockford, IL), and a monoclonal antibody against Stx (3C10, Toxin Technology, Sarasota, FL) was labeled with the special BV-TAG label containing a Tris(bipyridine)-chelated ruthenium(II) atom (BioVeris Corp.). To measure the uptake of biotinylated Stx in cells treated with different PKC inhibitors, the cells were washed once in Hepes-buffered minimal essential medium and then preincubated with the inhibitors before biotin-labeled Stx (0.33 nM) was added. The incubation was continued at 37°C for the indicated time periods, still in the presence of the inhibitors. Control cells were treated with Me 2 SO alone. To distinguish internalized toxin from total cell-associated toxin (bound plus internalized), the cells were washed (0.14 M NaCl, 2 mM CaCl 2 , 20 mM Hepes, pH 8.6), and one half of the plate was treated with 0.1 M MESNa in the same buffer for 30 min on ice to reduce the SS-linked biotin in cell surfacebound toxin. The other half was mock treated. The cells were washed in cold buffer (0.14 M NaCl, 2 mM CaCl 2 , 20 mM Hepes, pH 7.0) and lysed (1% Triton, 60 mM n-octylglucopyranoside, 100 mM NaCl, 5 mM MgCl 2 , 50 mM Hepes). The cell lysate was incubated with the BV-TAG-labeled anti-Stx antibody (0.5 g/ml), and simultaneously, to selectively quantify biotinylated Stx, the biotin-labeled Stx was captured by streptavidin-coated magnetic beads (0.1 mg/ml, Invitrogen) by gentle shaking for 1.5 h in assay diluent (0.2% bovine serum albumin, 0.5% Tween 20 in phosphate-buffered saline). The amount of streptavidincaptured Stx complexed to the BV-TAG-labeled antibody was quantified by an M1R Analyzer (BioVeris Corp.). Counts from cells treated with MESNa represent the amount of internalized toxin, whereas counts from untreated cells represent the total amount of toxin associated with the cells (bound plus internalized). Endocytosis of Stx was reported as internalized toxin in percentage of total cell-associated toxin.
Sulfation of StxB-sulf 2 or Ricin-sulf 1 -Ricin-sulf 1 , a modified ricin A-chain containing a tyrosine sulfation site reconstituted with ricin B-chain, was produced, purified, and reconstituted as previously described (25). The StxB containing a tandem of sulfation sites in the C terminus (StxB-sulf 2 ) was produced in Escherichia coli BL21(DE3) cells as previously described (26). For sulfation experiments, HeLa, HEp-2, or Vero cells (2 ϫ 10 5 /well in 6-well plate) were washed twice in sulfate-free Dulbecco's modified Eagles' medium supplemented with 2 mM L-glutamine, before they were preincubated with 0.2 mCi/ml Na 2 35 SO 4 for 3 h in sulfate-free Dulbecco's modified Eagles' medium. For the inhibitor studies, the drugs were added for the last 30 min of this preincubation. Then StxB-sulf 2 or ricin-sulf 1 was added (2 g/ml), and the incubation continued for 45 min (StxB) or 2 h (ricin), in the presence or absence of the inhibitors.
To accumulate StxB in endosomes, the cells were starved in sulfate-free medium for 2 h, and then StxB-sulf 2 was internalized at 19.5°C for 1 h, before the cells were transferred to 37°C and the incubation continued for 45 min. The cells were washed in cold phosphate-buffered saline, lysed (0.1 M NaCl, 10 mM Na 2 HPO 4 , 1 mM EDTA, 1% Triton X-100, and 60 mM n-octylglucopyranoside, supplemented with a mixture of protease inhibitors (Roche Applied Science), pH 7.4), and scraped. The lysate was cleared by centrifugation (10,000 ϫ g, 10 min), and StxB or ricin was immunoprecipitated from the supernatant with monoclonal antibodies prebound to protein-A/Sepharose beads (Amersham Biosciences) overnight at 4°C. The beads were washed twice in phosphate-buffered saline/0.35% Triton X-100, before the adsorbed material was analyzed by 4 -20% SDS-PAGE under reducing conditions and transferred onto a polyvinylidene difluoride membrane (Immobilon-P, Millipore, Billerica, MA). The bands were detected by exposing the membrane to a PhosphorImager screen, and the signal intensities were quantified by using ImageQuaNT 5.0 software (Amersham Biosciences). To determine the level of protein sulfation in general in cells treated with the different inhibitors or transfected with siRNA, the amount of 35 S-labeled proteins in the lysate was measured by trichloroacetic acid precipitation. Only minor changes in the sulfation of proteins in general were observed under the different conditions. 3 In addition, the amount of protein in each well of transfected cells was determined by the BCA protein assay (Pierce) with the use of bovine serum albumin as the standard. Only minor changes in the level of proteins were observed. 3 Immunoprecipitation and Immunoblotting-PKC␦ detection was performed as follows. HeLa cells were starved for 4 h and treated with StxB (250 ng/ml) for the indicated time, before lysates were prepared (1% Triton X-100, 0.5 M Tris, 20 mM EDTA, 10 mM NaF, and 30 mM sodium pyrophosphate decahydrate, supplemented with a mixture of protease inhibitors (Roche Applied Science) and a mixture of phosphatase inhibitors (Inhibitor Mixture 1, Sigma), pH 7.4). The lysates were sonicated for 10 s, and then the cleared supernatants were used for immunoprecipitation with mouse anti-PKC␦ (BD Biosciences) precoated on protein-G beads overnight at 4°C. The immunoprecipitation was then washed twice in the lysis buffer, and the adsorbed material was analyzed by 4 -20% SDS-PAGE under reducing conditions and transferred onto a polyvinylidene difluoride membrane. The phosphorylation state of PKC␦ was determined by rabbit anti-phospho-PKC␦ (Thr 505 , Cell Signaling Technology) and the enhanced signal chemiluminescence reagent (Pierce Biotechnology). The total level of PKC␦ was monitored by rabbit anti-PKC␦ (Santa Cruz Biotechnology, Santa Cruz, CA). The phosphorylation state of PKC␣ was analyzed in parallel. A fraction of the whole cell lysate from the PKC␦ experiment was analyzed by Western blot using antiphospho-PKC␣ (Ser-657, Millipore, Billerica, MA). Total PKC␣ was determined by mouse anti-PKC␣ (BD Biosciences).
To determine the level of PKC␦ or PKC␣ upon siRNA treatment, lysates were prepared and PKC content was analyzed by Western blotting. The level of PKC␦ was detected by rabbit anti-PKC␦ (Santa Cruz Biotechnology). PKC␣ was detected by mouse anti-PKC␣ (BD Biosciences). The level of ␣-tubulin was detected by mouse anti-␣-tubulin (Sigma) and was used as a loading control. Signal intensities of the bands were quantified using ImageQuaNT 5.0.
Immunofluorescence-HeLa cells grown on coverslips were washed in Hepes medium and preincubated with or without rottlerin (2.5 M) for 30 min, then StxB-sulf 2 (2 g/ml) was added and the incubation continued for 10 min. The cells were washed three times in warm Hepes medium, before incubation for a further 20 min at 37°C in warm Hepes medium with or without rottlerin. The cells were fixed in 3% paraformaldehyde in phosphate-buffered saline, permeabilized in 0.1% Triton X-100, and blocked in 5% fetal calf serum. Stx was stained by mouse anti-Stx antibodies (13C4, Toxin Technology, Sarasota, FL) followed by Cy2-labeled donkey anti-mouse IgG (Jackson ImmunoResearch Laboratories, West Grove, PA), and the cells were labeled with either rabbit anti-giantin antibodies (BabCO, Berkeley, CA) and Cy3-labeled donkey anti-rabbit IgG (Jackson ImmunoResearch Laboratories), or human anti-EEA1 serum (a gift from Dr. Harald Stenmark, The Norwegian Radium Hospital, Norway) and rhodamine-labeled donkey anti-human IgG (Jackson ImmunoResearch Laboratories). The cells were mounted in Mowiol and analyzed in a Zeiss LSM 510 Meta confocal microscope (Zeiss, Jena, Germany) equipped with a Neo-Fluar 100 ϫ/1.45 oil immersion objective. Images were taken of thin single plane sections, and the signal intensities were quantified by using the histogram-function in the Zeiss LSM Image Browser software (Version 3).

PKC␦ Is Specifically Activated by StxB-We have recently
shown that the tyrosine kinase Syk is activated by Stx and regulates toxin uptake (12). We therefore wanted to investigate whether other key regulators of intracellular transport are activated by Stx and have a role in toxin transport. PKC isozymes are involved in sorting and intracellular transport of several receptors, and recently both PKC␣ and ␦ activity were implicated in key transport steps. First, we wanted to study whether Stx binding would activate these specific PKC isozymes. Several PKC isozymes have been detected in HeLa cells, including ␣, ␤ II , ␦, ⑀, , and (27,28). Phosphorylation of Thr 505 located at the activation loop of PKC␦ is known to increase the activity of this kinase (29). As shown in Fig. 1, upon addition of StxB, PKC␦ was rapidly phosphorylated at Thr 505 . Quantification of the amount of phosphorylated PKC␦ from three independent experiments revealed that there was an increase already after 10 min, and that the peak was observed at 20 min with an increase to ϳ300% of unstimulated control cells. The phosphorylation was then reduced, but did not return to control levels after 60 min. 3 Notably, there was a basal PKC␦ phosphorylation even in unstimulated control cells (Fig. 1). The phosphorylation status of PKC␣ was determined in parallel. Autophosphorylation of Ser 657 is a hallmark of PKC␣ activation (29). However, no increase in PKC␣ phosphorylation was detected in the lysate (Fig. 1). From these data we conclude that PKC␦ is specifically activated by StxB.
HeLa Cells Are Protected against Stx Toxicity by the PKC␦ Inhibitor Rottlerin-Knowing that Stx was able to activate PKC␦, we wanted to study whether this kinase had a role in intracellular transport of the toxin. First, the general toxicity of Stx, determined as inhibition of cellular protein synthesis, was measured in HeLa cells treated with different PKC inhibitors. Fig. 2A shows a typical toxicity experiment, and Fig. 2B illustrates the -fold protection against Stx obtained by pretreatment with the different inhibitors. Treatment with the general PKC inhibitor BIM, which inhibits both classic and novel PKC isozymes, gave a 2-fold protection against Stx, whereas treatment with rottlerin, which is specific for PKC␦, induced a 5-fold protection. In contrast, treatment with Gö6976, which selec-FIGURE 1. PKC␦ is activated by StxB. HeLa cells were starved (4 h) before treatment with or without StxB (250 ng/ml) for the indicated times. PKC␦ was immunoprecipitated from the lysate, and the amount of phosphorylated PKC␦ (Thr 505 ) was detected by Western blot. The blot was stripped and reprobed with PKC␦ to show equal loading. The presented blot is a typical result, and the histogram shows the percent increase in PKC␦ phosphorylation compared with untreated control (mean value Ϯ S.D. from three independent experiments). The amount of phosphorylated PKC␣ (Ser 657 ) was determined in parallel in the whole cell lysate (WCL) by Western blot. The presented blot is representative for two independent experiments. tively inhibits PKC␣ and -␤ I , gave no protection. Importantly, the protein synthesis was not significantly changed upon treatment with the inhibitors themselves, not even rottlerin, which has been reported to deplete the cellular ATP in certain cell types at higher concentrations (30). 3 Thus, the inhibitors were non-toxic to the cells during the short incubations here used, and the observed reduction in protein synthesis upon Stx treatment is caused by the toxin only. To verify the involvement of PKC␦ in Stx toxicity, HeLa cells were specifically depleted for PKC␦ by transfection with siRNA oligonucleotides complementary to PKC␦ mRNA. Control cells were transfected with a nonspecific control siRNA and are in the rest of the report referred to as "control-transfected cells." Both a typical toxicity experiment (Fig. 2C) and pooled data from three independent experiments (Fig. 2D) show a Ͼ3-fold protection against Stx upon depletion of PKC␦ by siRNA 1. The effect of the siRNA was routinely verified by immunoblotting, and as shown in supplemental Fig. S1C, the PKC␦ protein level was strongly reduced (50 nM siRNA 1, 48 h, with toxicity and sulfation experiments done in parallel). Importantly, as shown in Fig. 2D, the effect of the plant toxin ricin, which has the same intracellular target as Stx, was not reduced by depletion of PKC␦, implying that PKC␦ specifically affects Shiga toxicity. Taken together, PKC␦ activity seemed to be important for the intoxication by Stx in HeLa cells, whereas PKC␣ activity did not seem to be involved in this process.
The Endocytic Uptake of Stx Is Slightly Reduced by Rottlerin-To intoxicate cells, Stx has to be transported all the way from the plasma membrane, via EE to the Golgi, and further to the endoplasmic reticulum before translocation to the cytosol. We wanted to investigate which of the one or more steps in this intracellular transport might be regulated by PKC␦ activity and lead to the protection observed in Fig. 2. First, we measured the very first step in the process, the endocytic uptake of Stx. HeLa cells were pretreated with the different PKC inhibitors used above, and the internalization of labeled Stx was measured after 15 min of incubation. As shown in Fig. 3A, although Gö6976 did not affect Stx endocytosis, both BIM and rottlerin gave a small, but significant reduction in toxin uptake after 15 min, indicating that PKC␦ activity is partially involved in the initial uptake of Stx. The endocytic rates of Stx in control and rottlerintreated cells are compared in Fig. 3B. Rottlerin did not seem to increase the recycling of Stx, because an increased recycling would lead to a reduction in the amount of internalized Stx over time. In contrast, the reduction in endocytosis upon PKC␦ inhibition was observed already from early time points, and it was constant over time (Fig. 3B). However, it is unlikely that the small reduction in Stx uptake would lead to the 5-fold protection against Stx observed in rottlerin-treated cells, and we therefore went on to investigate whether the next step in the internalization process of Stx, the endosome-to-Golgi transport, would also be reduced by rottlerin.
The Sulfation of StxB Is Strongly Inhibited by Rottlerin-To quantitate the transport of toxin molecules to the TGN biochemically, we employed StxB molecules with tandem sulfation sites recombinantly added to the C terminus. These sites are sulfated by a sulfotransferase upon arrival to the TGN (31, 32). The experiment is performed in the presence of radioactive  sulfate, and the amount of sulfated StxB can be visualized by autoradiography. As shown in Fig. 4, pretreatment of HeLa cells with rottlerin resulted in a marked, concentration-dependent inhibition of StxB sulfation, whereas treatment with Gö6976 had no effect. BIM also gave a significant reduction in StxB sulfation, but not as pronounced as rottlerin. These data suggest that PKC␦ activity is strongly involved in the endosometo-Golgi transport of StxB, and the inhibition of this step is large enough to account for the protection against Stx in rottlerin-treated cells (Fig. 2). Again, PKC␣ activity did not seem to be involved. The possibility existed that pretreatment with the PKC inhibitors would reduce the general level of protein sulfation in the cells, and as described under "Experimental Procedures," we routinely measured the total amount of sulfated proteins and found it unchanged under the different conditions presented herein. 3 To investigate whether the StxB transport was sensitive to rottlerin also in other cell types, StxB sulfation was measured both in Vero and HEp-2 cells. In the two cell types the StxB sulfation was reduced to 30.0 Ϯ 4.4% and 16.0 Ϯ 7.9% of untreated control, respectively (mean Ϯ S.D., n ϭ 3 independent experiments) upon rottlerin treatment (5 M) without affecting the total sulfation of proteins. Clearly, the involvement of PKC␦ in StxB sulfation is not restricted to HeLa cells.
Sulfation of StxB Is Inhibited by siRNA-induced Knockdown of PKC␦-To verify that PKC␦ is involved in endosome-to-Golgi transport of Stx, HeLa cells were depleted for PKC␦ by transfection with three different siRNA oligonucleotides complementary to non-overlapping regions of PKC␦ mRNA. The effect of the siRNAs were tested by immunoblotting, and as shown in Fig. 5A, 48 h after transfection of HeLa cells with siRNA 1, the PKC␦ protein level was strongly reduced (down to 37% of control-transfected cells) already at 25 nM siRNA oligonucleotide. The protein level was further reduced in a concentration-dependent manner until it was virtually knocked out at 100 nM siRNA 1 (down to 16% of control-transfected cells). The blot was stripped and reprobed with anti-␣-tubulin to show equal loading. Then StxB sulfation was determined in the siRNA-transfected cells. As shown in Fig. 5B for one typical experiment and in Fig. 5C from several pooled experiments, upon transfection with siRNA 1, the sulfation of StxB was strongly reduced already at 25 nM and was further reduced in a concentration-dependent manner at 50 and 100 nM. Thus, there seemed to be a strong correlation between the PKC␦ protein level and the extent of StxB sulfation. Transfection with two other siRNAs against PKC␦, siRNA 2 and 3, reduced the protein level of PKC␦ to ϳ45% of that in control-transfected cells (Fig. 5D), and in these cells the StxB sulfation was reduced to ϳ65% of the control (Fig. 5E).
To investigate whether PKC␦ activity is a general regulator of retrograde transport routes or whether such activity might be specific for the pathway used by StxB, we compared the endosome-to-Golgi transport of StxB with that of ricin in cells treated with rottlerin or PKC␦ siRNA. The sulfation of the two toxins was measured in parallel. As shown in supplemental Fig.  S1A, although the StxB sulfation was inhibited in cells treated with rottlerin, the ricin sulfation was not reduced. In most experiments the ricin sulfation was actually slightly increased. The same result was obtained when the ricin sulfation was determined after 45 min of incubation, although the band intensity was lower. 3 Moreover, as shown in supplemental Fig. S1B, although transfection with PKC␦ siRNA 1 strongly reduced the StxB sulfation, again the ricin sulfation was slightly increased. In addition, the small inhibitory effect that rottlerin seemed to exert on the initial uptake of Stx seems to be specific to this toxin. Rottlerin treatment reduced neither the general clathrin-mediated endocytosis, as measured by uptake of 125 Ilabeled transferrin (113 Ϯ 13% of control, n ϭ 4 independent experiments) nor the uptake of 125 I-ricin (101 Ϯ 10% of control, n ϭ 5 independent experiments). Taken together, the retrograde transport of Stx and ricin seem to be differentially regulated with respect to PKC␦.
To demonstrate that rottlerin is specific for PKC␦ in our system, we performed sulfation experiments with the combined treatments of rottlerin and PKC␦ siRNA. We reasoned that, if the inhibition of StxB sulfation was not increased in the presence of both treatments compared with the inhibition observed with one treatment alone, rottlerin acts only on PKC␦. In other words, the removal of the drug's target should abolish the effect of the drug. As shown in Fig. 5 (F and G), no significant additive effect was observed in cells treated with both rottlerin and PKC␦ siRNA compared with both treatments alone. This confirms that rottlerin is specific for PKC␦ in our assay.
The possibility existed that the strong reduction in StxB sulfation we observed upon transfection with the PKC␦-specific siRNAs could be due to reduced binding and/or uptake of the toxin. To exclude this, PKC␦ siRNA 1-3 were tested in endocytosis-experiments. The transfection process itself did not alter the Stx uptake, because toxin uptake was unchanged in cells transfected with nonspecific siRNA compared with untransfected cells. 3 Furthermore, in cells transfected with the PKC␦-specific siRNAs only a small reduction in Stx uptake was detected, 3 comparable to the reduction in toxin uptake induced by rottlerin treatment of the cells (Fig. 3).
To further strengthen our hypothesis that the observed reduction in sulfation was caused by reduced endosome-to-Golgi transport of StxB and not by reduced uptake, the cells were incubated at 19.5°C for 1 h to accumulate StxB in endosomes, before the incubation was continued for 45 min at 37°C to allow further transport to the Golgi. As shown in Fig. 5H, StxB sulfation upon PKC␦ depletion was strongly reduced also after incubation at 19.5°C. The amount of sulfated StxB under these conditions was only slightly higher than under the normal conditions without the accumulation in endosomes (Fig.  5C, 50 nM). Comparable results were obtained by rottlerin treatment (2 M) during the last 30 min of the 19.5°C step. 3 This strengthens the notion that the reduction in StxB sulfation observed upon inhibition or knockdown of PKC␦ was not caused by reduced endocytic uptake of the toxin but rather a specific inhibition of the endosome-to-Golgi transport step. The amount of cell-associated StxB in PKC␦ siRNA-transfected cells was further investigated by measuring the cell-associated 125 I-StxB after 45 min at 37°C. The amount of 125 I-StxB was only slightly reduced upon PKC␦ depletion, 3 indicating that the reduction in sulfation was not associated with an increased degradation of StxB.
StxB Sulfation Is Unchanged by Knockdown of PKC␣ by siRNA Oligonucleotides-Data from StxB sulfation experiments upon treatment with Gö6976 (Fig. 4) suggested that PKC␣ was not involved in endosome-to-Golgi transport of StxB. We wanted to confirm this by depletion of PKC␣ by two different The cells were preincubated for 3 h in sulfate-free medium containing 0.2 mCi/ml Na 2 35 SO 4 , then StxB-sulf 2 (5 g/ml) was added, and the incubation continued for 45 min before lysis. StxB was immunoprecipitated from the lysate, and the sulfated StxB was visualized by SDS-PAGE and autoradiography. C, the amount of sulfated StxB in cells transfected with nonspecific siRNA was normalized to 100%, and the amount of sulfated StxB after knockdown of PKC␦ with siRNA 1 was calculated as percentage of control-transfected cells for each of the three concentrations used. The data shown are mean Ϯ S.D. from seven independent experiments with 50 nM, 4 with 100 nM, or mean Ϯ S.D. between duplicates from one experiment with 25 nM. D, immunoblot analysis of the amount of PKC␦ present in the cell lysate after knockdown of PKC␦ with siRNA 2 or 3 (100 nM, 3 days). ␣-Tubulin was used as a loading control. E, sulfation of StxB after transfection with PKC␦ siRNA 2 or 3. HeLa cells were transfected with nonspecific or either of the two PKC␦-specific siRNAs (100 nM, 3 days). Sulfation of StxB was measured as described in B. The amount of sulfated StxB in cells transfected with nonspecific siRNA was normalized to 100%, and the amount of sulfated StxB after knockdown of PKC␦ with siRNA 2 or 3 was calculated as percentage of control-transfected cells. The data shown are mean values Ϯ S.D. of three independent experiments. F, HeLa cells were transfected with either nonspecific or PKC␦-specific siRNA 1 (50 nM, 48 h). After 2.5 h of starvation in sulfate-free medium containing 0.2 mCi/ml Na 2 35 SO 4 , the cells were preincubated with rottlerin (2 M), or control-treated, in sulfatefree medium for 30 min, then StxB-sulf 2 (5 g/ml) was added and the incubation continued for 45 min, still in the presence of the inhibitor. The amount of sulfated StxB was detected as described above. A representative autoradiogram is shown out of three independently performed experiments. G, quantification of the autoradiogram shown in F. The amount of sulfated StxB in control-treated cells transfected with nonspecific siRNA was normalized to 100%, and the amount of sulfated StxB in the other conditions were calculated as percentage of this control value. The data shown are mean Ϯ S.D. between duplicates. H, HeLa cells were controltransfected cells or transfected with PKC␦ siRNA 1 (50 nM, 48 h) and preincubated for 3 h in sulfate-free medium containing 0.2 mCi/ml Na 2 35 SO 4 . The cells were moved to 19.5°C, and then StxB-sulf 2 (5 g/ml) was added. The incubation was continued for 1 h at 19.5°C before the cells were moved back to 37°C and incubated for 45 min. The cells were lysed, and StxB was immunoprecipitated from the lysate as described under "Experimental Procedures." The amount of sulfated StxB in control-transfected cells was normalized to 100%, and the amount of sulfated StxB after knockdown of PKC␦ with siRNA 1 was calculated as percentage of control-transfected cells. The data shown are mean values Ϯ S.D. of three independent experiments. siRNA oligonucleotides complementary to non-overlapping regions of PKC␣ mRNA. As shown by immunoblotting in Fig. 6A, the two siRNAs virtually knocked out the PKC␣ protein at 50 nM 48 h after transfection. However, as shown for one typical experiment in Fig. 6B and from several pooled experiments in Fig. 6C, the StxB sulfation proceeded as normal despite the knockdown of the PKC␣ protein, which is in accordance with the inhibitor data.
StxB Accumulates in Endosomes and Does Not Colocalize with Giantin upon Treatment with Rottlerin or PKC␦-specific siRNA-From the biochemical data, PKC␦, and not PKC␣, seemed to be involved in the endosome-to-Golgi transport of StxB, and we wanted to confirm these data by another approach. To this end HeLa cells were either pre-treated with rottlerin to inactivate PKC␦, or the cells were depleted of PKC␦ by siRNA, and the transport of StxB was monitored by confocal microscopy. As shown in Fig. 7, both rottlerin treatment (B) and transfection with the PKC␦-specific siRNA 1 (D) resulted in an almost complete loss of colocalization between StxB and the Golgi-marker giantin, compared with the extensive colocalization of these markers in both untreated control cells (A) and cells transfected with control siRNA (C). In the rottlerin-treated or siRNA 1-transfected cells StxB showed a punctate cytosolic staining, which was presumably labeling of endosomal structures. In the PKC␦depleted cells the StxB labeling was weaker, and many of the StxB-positive structures were actually smaller than in the rottlerin-treated cells. The colocalization between StxB and giantin was quantified in each of the four conditions, and calculated as a percentage of total giantin staining. Giantin was chosen as the reference, because the staining pattern was equal under all conditions. As shown in Fig. 7 (right panels) both rottlerin and siRNA1 transfection reduced the colocalization with giantin strongly, indicating that StxB was unable to reach the Golgi under these conditions. Similar phenotypes were obtained by the PKC␦ siRNA 2 and 3 (Fig. 7, F and  G), although under these conditions the phenotype was less obvious, in accordance with the partial knockdown of the PKC␦ protein using these siRNAs (see Fig. 5D). Inhibition or depletion of PKC␦ did not seem to increase the recycling of StxB, as there was no detectable increase in StxB staining on the cell surface of treated cells compared with controls (Fig.  7, B and D).
To investigate whether the Golgi transport of StxB and the Shiga holotoxin was affected in the same way by depletion of PKC␦, HeLa cells transfected with PKC␦ siRNA 1 or 2 were treated with Stx and processed for immunofluorescence as for StxB. In control-transfected cells Stx showed a dense, perinuclear staining identical to that observed for StxB (Fig. 7, compare C and H), and in the PKC␦ siRNA-transfected cells the staining pattern of Stx was equal to that of StxB, a punctate cytosolic distribution (Fig. 7, compare D-F with I-J).
To verify the lack of effect on Golgi transport by PKC␣ depletion (the sulfation data), cells transfected with PKC␣ siRNA was also monitored by immunofluorescence. As shown in Fig. 8  (A and B), transfection with PKC␣ siRNA 1 did not lead to an altered localization of StxB compared with control-transfected cells. Similar results were obtained with the other PKC␣ siRNA. 3 This is in accordance with the sulfation data and supports the conclusion that PKC␣ is not involved in the endosome-to-Golgi transport of StxB.
The structure of the Golgi itself, as judged by the giantin staining, did not seem to be affected by rottlerin or the lack of PKC␦. However, the TGN46 staining was slightly redistributed under these conditions. 3 A partial redistribution of TGN38/46 to endosomes has previously been shown in cells where endosome-to-Golgi transport is strongly inhibited, a situation that would also inhibit the retrieval of Golgi proteins (33,34). Note that transfection with PKC␣ siRNA, which had no effect on StxB sulfation (Fig. 6), did not induce a redistribution of TGN46 (Fig. 8B).
In an attempt to identify the cytosolic structures in which the StxB accumulated upon PKC␦ inhibition, control-and rottlerin-treated cells were labeled for the EE-marker EEA1 after internalization of StxB. As shown in Fig. 8C, the control cells revealed as expected, a dense, perinuclear staining of StxB and hardly any colocalization with the punctate EEA1 staining. Upon rottlerin treatment the StxB staining was punctate and no longer perinuclear, and in addition, there was a higher degree of colocalization between StxB and EEA1 (Fig. 8D). The colocalization of StxB and EEA1 was quantified and presented as the percentage of total EEA1 staining (Fig. 8, right panel). The marked increase in colocalization between StxB and EEA1 suggests that StxB is trapped in EE upon inhibition of PKC␦ activity.

DISCUSSION
In this study, we have shown that PKC␦ is activated upon StxB binding and regulates the endosome-to-Golgi transport of the toxin. In contrast, PKC␣ was neither activated nor involved in the retrograde transport of StxB. Upon inhibition or knockdown of PKC␦, StxB molecules colocalized less with giantin and more with EEA1, indicating that the molecules were accumulated in endosomes, unable to reach the Golgi. Importantly, not only was Golgi transport inhibited, but also the toxic effect of Stx was strongly reduced, indicating that transport of Stx to the cytosol is dependent on PKC␦ activity.
The endosome-to-Golgi transport of StxB monitored biochemically (the sulfation data) correlated well with the immunofluorescence data and the toxicity data. Both inhibition by the PKC␦-specific drug rottlerin or depletion of PKC␦ by siRNA resulted in a strong reduction of the endosome-to-Golgi transport of StxB, which was reflected by the reduced toxicity. The StxB staining showed a punctate pattern with little or no colocalization with giantin, while the colocalization with EEA1 increased. This indicates that StxB was trapped early in the transport pathway, unable to reach the Golgi. Such an accumulation of cargo in endosomes upon PKC inhibition has previously been shown for PKC-regulated recycling or degradative pathways, suggesting that PKC activity is important for exit from early endosomal compartments (27,(35)(36)(37)(38). However, this is the first report showing that PKC regulates the exit of cargo from endosomes to the retrograde pathway.
In the PKC␦-inhibited or -depleted cells, the general level of protein sulfation was unchanged, and the sulfation of ricin was not reduced (supplemental Fig. S1), dem- onstrating that the activity of the Golgi-resident enzyme sulfotransferase was not regulated by PKC␦. Furthermore, inhibition or depletion of PKC␦ did not seem to affect degradation of Stx or recycling to the plasma membrane. Under control conditions both the degradation and the recycling of Stx have been shown to be marginal in HeLa cells (39), and neither of the processes seemed to be changed by the inhibition or depletion of PKC␦. 3 Stx has been shown to activate several kinases leading to apoptosis in different cell types (reviewed in Ref. 8). We have recently demonstrated that Stx-induced signaling also stimulates its entry (12). Here we report that Stx specifically activated PKC␦, and the same kinase was found to regulate the endosome-to-Golgi transport of the toxin. In contrast, PKC␣ was neither activated nor involved in toxin transport. This indicates that Stx is able to increase its own retrograde transport by specific activation of PKC␦. However, it cannot be excluded that the basal activity of PKC␦ in HeLa cells (see Fig. 1) is sufficient to mediate the retrograde transport of Stx. It should also be noted that a potential role of the atypical PKC isozymes in Stx transport has not been determined by the treatments employed in this study.
PKC␦ has many potential targets at different cellular locations. The kinase might act directly on a key regulator of Stx transport, or alternatively, components important for endosome-to-Golgi transport of Stx might be relocated due to PKC␦ inhibition. PKC isozymes are known to translocate to the plasma membrane upon activation (18). In this study, although only a fraction of the initial Stx uptake seemed to be dependent on PKC␦, active translocated PKC␦ might be important for recruitment of specific proteins or lipids to the area of Stx internalization. This microenvironment could determine the destination of Stx-containing vesicles. Correct retrograde targeting of both Stx and cholera toxin (CT) have been shown to require localization of these toxins to lipid rafts (39 -41), and the fraction of Gb3 that is raft-associated was recently shown to increase upon Stx binding (40). Interestingly, activated PKC␣ and -␦ have been shown to translocate to plasma membrane rafts upon initiation of an immunoreceptor signaling cascade activated by cross-linking of Fc␣R (21). The signaling was shown to be required for correct trafficking of Fc␣R vesicles. Moreover, retrograde transport of CT has been shown to require an association between CT-GM1 and actin mediated via lipid rafts (42). The link between CT in the outer leaflet of the plasma membrane with actin on the cytoplasmic side was hypothesized to be mediated by the ezrin/radixin/moesin (ERM) family of proteins, which binds actin and phosphatidylinositol 4,5-bisphosphate (42). The actin cytoskeletal elements are important in vesicular trafficking and well known substrates for PKC (reviewed in Ref. 43). A direct interaction between PKC␦ and actin has been shown in airway epithelial cells (44), and PKC activity has been implicated in phosphorylation of ERM proteins in endothelial cells (45). Also, Stx has been shown to induce phosphorylation of ezrin in ACHN cells (13). Thus, regulation of actin polymerization, possibly via ERM proteins, might be a target for PKC␦ in HeLa cells, and in this way the kinase might modulate Stx transport.
Moreover, the major PKC-substrate myristoylated alaninerich C kinase substrate (MARCKS), which is mainly found on the plasma membrane, directly binds to filamentous actin, calmodulin, raft-localized phosphatidylinositol 4,5-bisphosphate and PKC in a competitive manner (reviewed in Ref. 46). Therefore, MARCKS is proposed to integrate both PKC-mediated and calcium/calmodulin-mediated signals into modulation of the actin cytoskeleton (43). Both the classic and the novel PKC isozymes have been shown to phosphorylate MARCKS in vivo (47)(48)(49) and in vitro (50), and MARCKS has been implicated in several transport pathways. Interestingly, in HeLa cells treated with the actin-stabilizing agent jasplakinolide the endosometo-Golgi transport of StxB was nearly doubled, whereas the initial uptake of the toxin remained unchanged. 3 This suggests that actin might be involved in endosome-to-Golgi transport of StxB, despite the fact that depolymerization of actin by cytochalasins was reported to have little effect on this step (51). Calmodulin does not seem to be involved in the endosome-to-Golgi transport of StxB, 3 however, any potential cross-talk between PKC and MARCKS, actin, or calmodulin at the plasma membrane that could regulate the subsequent sorting of StxB in endosomes cannot be excluded.
Stx stimulation has been shown to induce microtubuli assembly both in ACHN cells (13) and in Vero cells (5). In Vero cells, Stx was suggested to stimulate its own trafficking, based on the observed Stx-induced assembly of microtubules and the requirements of both microtubules and dynein in the Golgi transport of Stx (5). Notably, it was shown that the Stx-induced signaling activating microtubuli assembly was not mediated via Syk, suggesting that multiple signaling pathways are induced by Stx. Whether PKC␦ activity is involved in the increased assembly of microtubuli is unknown.
PKC␦ might regulate the exit of Stx from EEs into the retrograde pathway by affecting target proteins on the EE. The retromer complex has been detected on EE and is implicated in exit of mannose 6-phosphate receptors from EEs into the retrograde pathway (52). The yeast homolog of the retromer components sorting nexin 1 and 2, Vps5p, is a phosphoprotein (53). Therefore it might be speculated that sorting nexin 1 and/or 2 are potential candidates for PKC␦-mediated regulation of Stx transport. Moreover, a proposed complex of the sorting nexins 4, 41, and 42 is shown to mediate retrieval of the v-SNARE Snc1p from endosomes to the Golgi complex in yeast (54). A role for this complex in Stx transport awaits determination.
PKC␦ might also affect Stx transport at the Golgi level. The SNARE proteins are regulators of vesicle biology and known PKC substrates (55,56). The t-SNAREs involved in StxB transport are mostly localized to the TGN or Golgi, and PKC␦ has been shown to translocate to the Golgi under certain conditions (57,58). The endosome-to-Golgi transport of StxB and TGN46 has been shown to implicate Syntaxin 6 and Syntaxin 16 (59). Syntaxin 16 is a phosphoprotein, thus, PKC␦ could regulate the endosome-to-Golgi transport of StxB by acting at the SNARE complex. Sed5, the yeast homolog of Syntaxin 5, is also shown to be a phosphoprotein. However, only a conserved PKA phosphorylation site was described (60). Interestingly, Syntaxin 5 was reported to regulate the retrograde transport of StxB in a pathway parallel to the Syntaxin 16-regulated route (61).
In conclusion, PKC␦ is specifically activated by Stx and regulates the endosome-to-Golgi transport of the toxin. Clearly, future studies are required to fully understand the mechanisms of the different intracellular transport routes followed by Stx and to further elucidate the toxin's own role in cellular entry and transport.