Functionally different pools of Shiga toxin receptor, globotriaosyl ceramide, in HeLa cells Thomas Falguie ` res 1, *, Winfried Ro ¨ mer 1 , Mohamed Amessou 1 , Carlos Afonso 2 , Claude Wolf 3 , Jean-Claude Tabet 2 , Christophe Lamaze 1 and Ludger Johannes 1 1 Laboratoire Trafic et Signalisation, Unite ´ Mixte de Recherche 144, Institut Curie ⁄ CNRS, Paris, France 2 Laboratoire de Chimie Structurale Organique et Biologique, Unite ´ Mixte de Recherche 7613, Universite ´ Pierre et Marie Curie, Paris, France 3 Centre Hospitalier Universitaire Saint-Antoine, Unite ´ Mixte de Recherche 538, INSERM ⁄ UMPC, Universite ´ Pierre et Marie Curie, Paris, France Globotriaosyl ceramide (Gb 3 or CD77) is a glyco- sphingolipid that was initially described as the rare P K blood group antigen [1]. Gb 3 has also been identified as a germinal center B-cell marker [2] that is overex- pressed by Burkitt’s lymphomas [3] and other centro- follicular lymphomas. More recent studies have revealed that several hematopoietic malignancies and solid tumors express Gb 3 [4,5]. The physiologic func- tion of Gb 3 is still unknown. Some studies have suggested that Gb 3 could regulate the function of signaling molecules, such as type I interferon receptors and CD19 [6]. Indeed, Gb 3 ligation has been shown to lead to several signaling events such as apoptosis [7], cytokine release [8], and nitric oxide production [9]. On Burkitt’s lymphoma B-cells, Gb 3 binding by nat- ural ligands or antibodies has been shown to induce apoptosis [7]. Gb 3 has also been identified as a pathogen receptor. Although its exact role in HIV infection remains to be established [10], it is well recognized that Gb 3 is the Keywords globotriaosyl ceramide; HeLa cells; membrane microdomains; molecular species; Shiga toxin Correspondence L. Johannes, Unite ´ Mixte de Recherche 144, Institut Curie ⁄ CNRS, 26 rue d’Ulm, 75248 Paris cedex 05 Fax: +33 1 42 34 65 07 Tel: +33 1 42 34 63 51 E-mail: johannes@curie.fr *Present address University of Geneva, Science II, Depart- ment of Biochemistry, Geneva, Switzerland (Received 4 July 2006, revised 23 August 2006, accepted 27 September 2006) doi:10.1111/j.1742-4658.2006.05516.x Many studies have investigated the intracellular trafficking of Shiga toxin, but very little is known about the underlying dynamics of its cellular recep- tor, the glycosphingolipid globotriaosyl ceramide. In this study, we show that globotriaosyl ceramide is required not only for Shiga toxin binding to cells, but also for its intracellular trafficking. Shiga toxin induces globotria- osyl ceramide recruitment to detergent-resistant membranes, and subse- quent internalization of the lipid. The globotriaosyl ceramide pool at the plasma membrane is then replenished from internal stores. Whereas endo- cytosis is not affected in the recovery condition, retrograde transport of Shiga toxin to the Golgi apparatus and the endoplasmic reticulum is strongly inhibited. This effect is specific, as cholera toxin trafficking on GM 1 and protein biosynthesis are not impaired. The differential behavior of both toxins is also paralleled by the selective loss of Shiga toxin associ- ation with detergent-resistant membranes in the recovery condition, and comparison of the molecular species composition of plasma membrane globotriaosyl ceramide indicates subtle changes in favor of unsaturated fatty acids. In conclusion, this study demonstrates the dynamic behavior of globotriaosyl ceramide at the plasma membrane and suggests that globo- triaosyl ceramide-specific determinants, possibly its molecular species com- position, are selectively required for efficient retrograde sorting on endosomes, but not for endocytosis. Abbreviations CTxB, cholera toxin B-subunit; DRM, detergent-resistant membrane; ER, endoplasmic reticulum; Gb 3 , globotriaosyl ceramide; PPMP, 1-phenyl-2-hexadecanoyl-amino-3-morpholino-1-propanol; STxB, Shiga toxin B-subunit; Tf, transferrin; TfR, transferrin receptor; TGN, trans-Golgi network. FEBS Journal 273 (2006) 5205–5218 ª 2006 The Authors Journal compilation ª 2006 FEBS 5205 cellular receptor of Shiga toxin and the closely related verotoxins (or Shiga-like toxins). These are produced by Shigella dysenteriae and by enterohemorrhagic strains of Escherichia coli [11]. Notably, Shiga toxin- producing E. coli O157:H7 has developed into an emerging cause of foodborne illness, and has been identified among the principal causes of postdiarrheal hemolytic uremic syndrome leading to acute renal fail- ure in infancy and childhood. The homopentameric B-subunits of these toxins (STxB) bind to 10–15 mole- cules of Gb 3 at the plasma membrane [12] and allow the intracellular transport of the holotoxin and the delivery of the monomeric catalytic A-subunit into the cytosol, leading to the inhibition of protein biosynthe- sis [13,14]. In numerous cell lines [15], it has been shown that Shi- ga toxin follows the retrograde transport route from the plasma membrane to the endoplasmic reticulum (ER), via the early endosome and the Golgi apparatus, cir- cumventing the degrading environment of the late endo- cytic pathway [16–18]. The molecular mechanisms underlying the most critical step in the retrograde route, i.e. escape from the endocytic pathway, are beginning to be unraveled. Shiga toxin transport from early⁄ recycling endosomes to the trans-Golgi network (TGN) involves the small GTPase Rab6a¢, soluble N -ethyl maleimide- sensitive factor attachment protein receptor (SNARE) complexes around the heavy chain t-SNAREs syntaxin 16 [19,20] and syntaxin 5 [21], clathrin [22,23], the phos- phatidylinositol lipid-binding clathrin adaptor epsinR [22], golgin-97 [24], and the GPP130 protein [25]. Fur- thermore, evidence was provided for a role of membrane microcompartmentalization in Shiga toxin sorting to the retrograde route [26,27]. Although it is clear that Gb 3 is critical for Shiga toxin binding to cells, very few studies have aimed at investigating the lipid directly. A correlation has been described between the sensitization of cells to Shiga toxin following exposure to butyric acid and the change of the molecular species composition of the cel- lular Gb 3 [28,29]. In in vitro binding assays, the fatty acid chain of Gb 3 was found to influence the binding to Shiga toxin [30,31]. In this study, we investigated the Gb 3 distribution and dynamics underlying the internalization and retro- grade transport of Shiga toxin, a poorly described aspect of the cell biology of this pathogenic protein. Gb 3 was surprisingly dynamic, in that after its Shiga toxin-induced internalization, the plasma membrane pool of Gb 3 rapidly recovered. However, we observed that retrograde transport to the Golgi apparatus and the ER was significantly less efficient on recovered Gb 3 than under control conditions, whereas internalization was not affected. In parallel, Shiga toxin association with detergent-resistant membrane (DRM) was reduced in the recovery condition. Using appropriate controls, i.e. another glycosphingolipid-binding pro- tein, cholera toxin, we created an experimental situ- ation in which the Shiga toxin–Gb 3 system was selectively targeted, and our data strongly suggest the existence of plasma membrane Gb 3 pool-specific fac- tors, possibly the molecular species composition of Gb 3 itself, that are selectively required for efficient ret- rograde transport. Results Gb 3 is required for retrograde transport of Shiga toxin from endosomes to the TGN The glycosphingolipid Gb 3 is required for Shiga toxin binding to cells, but it is not known to what extent it is also involved in later steps of retrograde toxin trans- port. To address this question, we treated HeLa cells with the glucosylceramide synthase inhibitor 1-phenyl-2- hexadecanoyl-amino-3-morpholino-1-propanol (PPMP) to reduce cellular Gb 3 to levels below 5% of those in untreated control cells. Under these conditions, the 4 °C binding protocol used for control cells does not allow detectable amounts of STxB to associate with cells. Therefore, the cells were continuously incubated with high concentrations of STxB to permit endocytosis by fluid-phase uptake. Whereas in control cells, STxB efficiently colocalized with the Golgi marker CTR433 (Fig. 1A, upper panel), it failed to do so in PPMP-trea- ted cells (Fig. 1A, lower panel), in which the protein remained in the endocytic pathway, partly colocalized with the transferrin receptor (TfR) (Fig. 1B, lower panel). Using sensitive biochemical assays (sulfation and glycosylation assays [32]), it was confirmed that STxB did not enter the retrograde route in PPMP-trea- ted cells (data not shown). These studies thus demon- strate that Gb 3 is required for Shiga toxin transport from endosomes to the TGN, and that no other cellular component can substitute for this activity. Shiga toxin recruits Gb 3 to DRMs The above-described experiment shows that Gb 3 is critical not only for Shiga toxin binding to cells, but also for intracellular toxin trafficking. The question then arises as to whether Shiga toxin in return influen- ces the cellular properties of Gb 3 . In a first experiment, we analyzed whether Shiga toxin would recruit Gb 3 to DRMs. Conditions were established in which, at steady state, about 10% of cellular Gb 3 was in DRM Gb 3 dynamics in HeLa cells T. Falguie ` res et al. 5206 FEBS Journal 273 (2006) 5205–5218 ª 2006 The Authors Journal compilation ª 2006 FEBS fraction 2 (Fig. 2A,B). After incubation of cells with STxB at saturating concentrations, Gb 3 association with DRMs was increased 2.5-fold. Gb 3 thus behaved like protein receptors whose association with mem- brane microdomains of the raft type often increases upon ligand binding. Plasma membrane dynamics of Gb 3 In the next step, we investigated how Shiga toxin influences the plasma membrane dynamics of Gb 3 . Ultrastructural studies on lipids are difficult because of several limitations, such as lack of antibodies, and fix- ation procedures that keep lipids in place during immunostaining. We therefore chose a biochemical approach in which the plasma membrane of HeLa cells was enriched on density gradients following cell surface silica coating [33]. The plasma membrane fraction was characterized using several compartment-specific mark- ers (Fig. 3A). On average, about 90% of the plasma membrane marker alkaline phophodiesterase was recovered in this fraction. The DRM markers caveolin-1 and flotillin-1 were also highly enriched in the plasma membrane fraction (Fig. 3A). The preparation con- tained 10% of total protein, and low amounts of other compartment markers such as Golgi (mannosidase, 4%), lysosomes (b-hexosaminidase, 20%), ER (calnex- in, 9%), and early endosomes (EEA1, 5%) (Fig. 3A). The amounts of Gb 3 and cholesterol in the plasma membrane-enriched fractions were then quantified. It was found that about 50% of the Gb 3 and 56% of the cholesterol were present at the plasma membrane of HeLa cells at steady state (Fig. 3A). These values may be overestimates, considering the contamination of the plasma membrane fractions by other organelles (see above). The plasma membrane dynamics of Gb 3 was then studied using the protocol described in Fig. 3B. HeLa cells were incubated on ice with saturating concentra- tions of STxB, and after different periods of time at 37 °C (0–60 min), the proportion of Gb 3 in plasma membrane fractions was determined. At the 0 min time point, about 50% of Gb 3 was in plasma membrane fractions (Fig. 3C), as described above (Fig. 3A). Fol- lowing a short incubation at 37 °C, a transient decrease of Gb 3 in these fractions to 28% was observed. Sixty minutes after the shift to 37 °C, a time point at which STxB is quantitatively localized in the Golgi apparatus [17], Gb 3 levels in plasma membrane fractions returned to 45%, which is somewhat lower than the levels found on control cells (Fig. 3C). How- ever, with the current sample size, this difference was not statistically significant. This 60 min time point was termed the ‘recovery condition’ (Fig. 4). Three days after STxB internalization, Gb 3 levels in plasma mem- brane fractions were close to those found in the recov- ery condition (Fig. 3C). These experiments led to the conclusion that Gb 3 was cointernalized with Shiga toxin, and that the plasma membrane pool of Gb 3 was then rapidly replenished with Gb 3 from internal stores. Cell biological analysis of the recovery condition As the steady-state plasma membrane Gb 3 pool was mobilized by Shiga toxin internalization and then recovered, we tested whether this resulted in changes of STxB binding to cells. For this, a protocol like the one described in Fig. 3B was used. However, instead A B Fig. 1. Gb 3 -dependent retrograde transport of STxB. HeLa cells that were pretreated for 6 days PPMP (+ PPMP) or control cells were incubated for 45 min at 37 ° C continuously with 25 l M (0.25 mgÆmL )1 ) STxB for PPMP-treated cells, or after prebinding with 1 l M STxB for control cells. Cells were fixed and permeabi- lized. STxB and the Golgi marker CTR433 (A) or the endosomal marker TfR (B) were visualized by indirect immunofluorescence. Note that in PPMP-treated cells, STxB does not colocalize with the Golgi marker and partially overlaps with TfR labeling [arrows in (B)], whereas the protein is efficiently accumulated in the Golgi appar- atus in control cells. Bars: 10 lm. T. Falguie ` res et al. Gb 3 dynamics in HeLa cells FEBS Journal 273 (2006) 5205–5218 ª 2006 The Authors Journal compilation ª 2006 FEBS 5207 of applying the plasma membrane enrichment proce- dure at the end of each incubation period at 37 °C, radiolabeled [ 125 I]STxB was bound to the cells on ice. At the 0 min time point, [ 125 I]STxB binding was strongly reduced, as expected (Fig. 5A). Upon incuba- tion at 37 °C, binding then readily recovered, parallel- ing the recovery of plasma membrane Gb 3 described in Fig. 3C. The plateau level of [ 125 I]STxB rebinding to cells was reached after 60 min at 82% (Fig. 5A). These results thus confirm the Gb 3 quantification data of Fig. 3C. Seventy-eight percent of the binding sites found on control cells were still detected on recovery cells, as shown by Scatchard analysis, and the apparent affinity of STxB for cells was not significantly changed (Table 1). In control cells, K d values and numbers of binding sites per cell were in good agreement with our previous studies [26]. To create a control condition that simulates the slight loss of binding sites, as observed in the recovery condition, Gb 3 levels were reduced using a 5 h treatment with the glucosylceramide synthase inhib- itor PPMP (‘PPMP condition’, Fig. 4). This treatment A B C Fig. 3. Plasma membrane dynamics of Gb 3 . (A) HeLa cell plasma membrane was enriched using the silica-coating method. The total lysate and plasma membrane-enriched fractions were characterized for total protein, DRM markers caveolin-1 and flotillin-1, cholesterol, and several compartment-specific markers: alkaline phosphodiesterase (plasma membrane), mannosidase (Golgi apparatus), b-hexosaminidase (lyso- somes), calnexin (ER), and EEA1 (early endosomes). The percentage of Gb 3 in the plasma membrane fraction was determined by glycolipid extraction and TLC overlay (dashed bar). Results are presented as the plasma membrane fraction ⁄ total lysate signal ratio, and means (± SEM) of five independent experiments are shown. (B) Schematic representation of the recovery experiments. After STxB binding to HeLa cells for 30 min at 4 °C, the cells were shifted for the indicated times to 37 °C. The cells were then either processed for plasma membrane enrichment and Gb 3 quantification [see (C)], or incubated at 4 °C with [ 125 I]STxB in a rebinding assay (Fig. 5A). (C) Presence of Gb 3 in plasma membrane fractions at the indicated times after the shift to 37 °C, following STxB binding on ice. See (B) for the experimental protocol. The 60 min time point was termed the ‘recovery condition’. The chi-square test showed that the observed differences in Gb 3 levels in plasma membrane fractions are significant (P<0.001) for the 5 and 10 min time points (indicated by *), and not significant for the 60 min and 3 day time points (indicated by #). AB Fig. 2. STxB recruits Gb 3 to DRMs. (A) HeLa cells were incubated (+ STxB) or not incubated (– STxB) with 1 lM STxB for 30 min at 4 °C. After washes, cells were lysed in 1% Triton X-100, and DRMs were prepared. After extraction of neutral glycolipids, Gb 3 was quantified in each fraction using TLC and overlay assays. DRMs are enriched in fraction 2. The percentage of Gb 3 in the DRM fraction is indicated. (B) Means (± SEM) of three independent experiments as shown in (A). Gb 3 dynamics in HeLa cells T. Falguie ` res et al. 5208 FEBS Journal 273 (2006) 5205–5218 ª 2006 The Authors Journal compilation ª 2006 FEBS led to a reduction of binding sites to about 75% of con- trol levels without loss in affinity (Table 1). The num- ber of binding sites for control, recovery and PPMP conditions are reported in Fig. 4. These three conditions (Fig. 4) were then used to characterize a number of cell biological phenomena related to retrograde transport to the ER. We found that STxB enrichment in DRMs was significantly reduced in the recovery condition, when compared to the control and PPMP conditions (Fig. 5B). As we had previously observed that DRM association corre- lated with efficient retrograde transport [26], we tested Shiga toxin trafficking to the Golgi apparatus and the ER under all conditions. In the recovery condition, a strong inhibition of sulfation on sulfation-site-carrying STxB was observed (Fig. 5C), indicating that arrival in the TGN was inhibited. In PPMP-treated cells, sulfation was also reduced, reflecting at least in part the lower number of binding sites under these condi- tions. However, comparing the PPMP and recovery conditions, it can be stated that sulfation was more than three-fold more strongly inhibited in the recovery condition, due to a direct effect on retrograde trans- port. Glycosylation analysis was used to confirm these observations (Fig. 5D). Indeed, this assay allows measurement of the relative quantity of glycosylated, ER-associated STxB over total cell-associated STxB under given conditions, and is therefore insensitive to differences in binding sites. Again, retrograde trans- port of STxB was inhibited about three-fold under recovery conditions, while 5 h of PPMP treatment had only a minor effect (Fig. 5D). Using the same technique, we also analyzed retrograde transport effi- ciency several days after a first-wave internalization (Fig. 5E). We found that even if the Gb 3 pool is lar- gely restored at the plasma membrane within an hour of first-wave STxB internalization (Fig. 3C), the arri- val of second-wave STxB in the ER is still partially impaired after up to 3 days (Fig. 5E). This surprising persistence of the recovery phenotype could be explained by the fact that the amount of cell-associ- ated STxB remains the same between 60 min and 3 days of first-wave STxB internalization (Fig. 5F, Cells), indicating that once STxB is present in the Golgi apparatus, it remains stably associated with the cells. This material might be capable of sequestering neo-synthesized Gb 3 or hypothetical licensing factors (see Discussion). As opposed to retrograde transport to the TGN and the ER, endocytosis of STxB was not inhibited in the recovery condition (Fig. 6A), and neither was that of transferrin (Tf) (Fig. 6B). These results document the specificity of the recovery effect, and show that whereas STxB can enter cells independently of its association with DRMs, the efficiency of intracellular sorting to the retrograde route strongly correlates with its presence in DRM fractions, consistent with our pre- vious work [26]. To test the specificity of the recovery phenotype, we then measured retrograde transport of cholera toxin to the TGN. Cholera toxin also binds to a glycosphingo- lipid, the ganglioside GM1, is associated with DRMs, and follows the retrograde route to the ER [34]. A sulfation site-carrying peptide was chemically coupled to cholera toxin B-subunit (CTxB). When sulfation analysis was performed under the same conditions as those of Fig. 5C, it became apparent that cholera toxin transport in the retrograde route was not affected in the recovery condition (Fig. 6C). Furthermore, CTxB association with DRMs was, if anything, increased (Fig. 6D), and cholesterol levels in plasma membrane fractions were similar in the control and recovery con- ditions (Fig. 6E). To rule out a possible toxic effect of a contaminant in our STxB preparation, protein bio- synthesis was measured after 1 or 72 h of internalizat- ion of first-wave STxB. No significant difference in protein biosynthesis could be detected in comparison with nontreated cells, whatever the duration of STxB internalization (Fig. 6F). No effect on cell division was detected (data not shown). These data show that the Fig. 4. Schematic representation of control, recovery and PPMP conditions. STxB bind- ing to Gb 3 leads to clustering of the lipid, as suggested from the DRM association data of Fig. 2. The number of STxB-binding sites is indicated as a percentage of control for each condition. See text for further details. T. Falguie ` res et al. Gb 3 dynamics in HeLa cells FEBS Journal 273 (2006) 5205–5218 ª 2006 The Authors Journal compilation ª 2006 FEBS 5209 recovery phenotype is restricted to the STxB–Gb 3 sys- tem, and presents a highly selective way of interfering with its dynamics while leaving many other membrane parameters intact. Analysis of the molecular species compositions of Gb 3 pools Several studies have suggested that specific molecular species of Gb 3 are correlated with efficient retrograde transport [28,29]. Therefore, we analyzed the molecular species composition of the plasma membrane and internal pools of Gb 3 , under both control and recovery conditions (Fig. 7). After plasma membrane or DRM enrichment, glycolipids were extracted, and Gb 3 was isolated from TLC plates and analyzed by nanospray tandem MS. The proportion of each molecular species in the analyzed fractions was determined. Owing to technical limitations, only the most abundant lipids could be detected. In adherent HeLa cells, the most abundant mole- cular species were C16:0, C22:0, C24:0, and C24:1 (Fig. 7). This composition was similar to the one previ- ously described for human astrocytoma cells [29], with A B D F E C Fig. 5. Shiga toxin trafficking in the recovery condition. (A) Rebinding assay following a protocol as described in Fig. 3B. In the recovery con- dition (60 min shift to 37 °C), the plateau of rebinding was reached. (B) DRM preparations under control (black bars), PPMP (white bars) and recovery (dashed bars) conditions. Results are represented as the percentage of STxB present in each fraction of the gradients, including DRM fraction 2. Note that in the recovery condition, STxB association with DRM was reduced. (C) Sulfation assay. After prebinding of STxB–Sulf 2 , cells were incubated for 20 min at 37 °C in the presence of radioactive sulfate. Sulfation of STxB–Sulf 2 was reduced in PPMP conditions (reduced Gb 3 expression in cells), and strongly reduced under recovery conditions, indicating that retrograde transport to the TGN was inhibited. (D) Glycosylation assay. After prebinding of [ 125 I]STxB–Glyc–KDEL, cells were incubated for 4 h at 37 °C. In the recovery con- dition, retrograde transport to ER was strongly inhibited, as indicated by reduced glycosylation of [ 125 I]STxB–Glyc–KDEL (arrow). (E) Progres- sive restoration of STxB glycosylation efficiency after several days of recovery. Experiments were performed as in (D), with the following modifications: [ 125 I]STxB–Glyc–KDEL was bound to cells after 0–3 days of recovery, as indicated, and this was followed by 16 h incubations at 37 °C. (F) First-wave internalized STxB remains stably associated with cells. Prebound iodinated STxB was incubated with HeLa cells at 37 °C for 0, 1, 24, 48 or 72 h. Using trichloroacetic acid precipitation (see Experimental procedures), cell-associated STxB (Cells), STxB in the culture medium (Culture Med.) and degraded STxB were determined for each time point. For each assay, means of three independent experiments (± SEM) are shown. Gb 3 dynamics in HeLa cells T. Falguie ` res et al. 5210 FEBS Journal 273 (2006) 5205–5218 ª 2006 The Authors Journal compilation ª 2006 FEBS the exception of C24:1, which was more abundant in our HeLa cell clone. We observed, however, that another clone, HeLa S3, had lower levels of C24:1 (data not shown). The Gb 3 molecular species composi- tion was similar in plasma membrane (Fig. 7A) and internal pools (Fig. 7B), indicating that at steady state, Gb 3 localization is not dictated by parameters such as membrane thickness. As a further test, we compared the molecular species composition of Gb 3 in DRMs before and after recruitment by STxB (Fig. 7C). Again, the results were similar under both conditions. When comparing the molecular species compositions of control and recovery conditions in each preparation, it became apparent that they were also very similar. The only notable exceptions were the C22:1 and C23:1 species in plasma membrane fractions, which were enriched two-fold in the recovery condition. However, it must be noted that C22:1 and C23:1 are minor spe- cies, and it remains to be determined directly to what extent such subtle differences in the overall species profile can account for the major effects that were observed in the recovery condition on DRM associ- ation and retrograde transport. Discussion Owing to technical limitations, very little is known about the dynamics and intracellular transport of sphingolipids. In this study, we used a plasma mem- brane enrichment method to analyze the dynamics of the Shiga toxin receptor Gb 3 . We found that Gb 3 was mobilized during Shiga toxin internalization, and the plasma membrane Gb 3 pool was then rapidly replen- ished from internal stores. Strikingly, retrograde trans- port in the recovery condition was significantly less efficient than in controls. We hypothesize that the recovery and control conditions are explained by plasma membrane steady-state Gb 3 pool-specific deter- minants that modify the efficacy of retrograde trans- port. AB C F E D Fig. 6. In-depth characterization of the recovery phenotype. (A) STxB endocytosis assay. No effect on STxB endocytosis was observed in the recovery condition. (B) Tf endocytosis assay. No effect on Tf endocytosis was observed in the recovery condition. (C) Retrograde trans- port assay with CTxB. Retrograde transport of CTxB to the TGN was not affected in the recovery condition, as determined by sulfation ana- lysis. This is in striking contrast to retrograde transport of STxB (Fig. 5C). (D) In the recovery condition, the association of CTxB DRMs was slightly increased. This is in striking contrast to the reduced DRM association of STxB under these conditions (Fig. 5B). (E) Cholesterol meas- urement in plasma membrane fractions. The cholesterol content was measured at the plasma membrane in control and recovery HeLa cells. No change was observed in the recovery condition. (F) Measurement of protein biosynthesis. HeLa cells were incubated or not with 1 l M STxB on ice, and this was followed by shift to 37 °C for 1 or 72 h. Protein biosynthesis was then measured by incorporation of [ 35 S]methion- ine. Results are expressed as a percentage of protein synthesis measured on control cells. For all experiments in this figure, means of at least three independent experiments (± SEM) are shown. Table 1. Scatchard analysis of control, PPMP and STxB-treated HeLa cells. HeLa cells were mock-treated (Control) or treated with 5 l M PPMP for 5 h (PPMP), or with 1 lM STxB for 30 min at 4°C, followed by a 1 h internalization at 37°C (Recovery). Then, 30 n M to 1 l M [ 125 I]STxB–Glyc–KDEL was bound to the cells for 2 h at 4°C. After washes and lysis of the cells, the results were expressed as a Scatchard representation, and K d and number of sites per cell were deduced for each condition. Means (± SEM) of three different experiments are shown. Treatment K d (nM) Number of sites (· 10 6 per cell) Control 30.5 ± 9.3 54.7 ± 3.7 (100%) PPMP 23.3 ± 8.1 41.4 ± 4.6 (75%) Recovery 26.2 ± 8.1 42.8 ± 4.1 (78%) T. Falguie ` res et al. Gb 3 dynamics in HeLa cells FEBS Journal 273 (2006) 5205–5218 ª 2006 The Authors Journal compilation ª 2006 FEBS 5211 Many of our attempts to identify these pool-specific determinants were not successful, in that no differences could be detected between control and recovery condi- tions for the following parameters: plasma membrane cholesterol levels (Fig. 6E), protein biosynthesis (Fig. 6F), band patterns of plasma membrane proteins crosslinked to STxB, and STxB-induced cytoskeletal rearrangements (data not shown). In our search for these pool-specific determinants, we also analyzed the molecular species composition of Gb 3 in plasma mem- brane fractions under control and recovery conditions. Indeed, a role for specific molecular species in Shiga toxin trafficking and intoxication of cells had previ- ously been hypothesized, based on the observation that butyric acid treatment of cells leads to a change of the molecular species composition of Gb 3 and to a con- comitant sensitization to Shiga toxin [28,29]. Using tandem MS, a two-fold increase in the recovery condi- tion was selectively observed for two minor molecular species, C22:1 and C23:1. Building on this finding, future work will have to address two critical questions: does Shiga toxin indeed induce the clustering of Gb 3 in lipid patches, and does spiking these patches with low doses of specific molecular species lead to a loss of microdomain organization? Response elements in favor of the first point are the apparent capacity of Shiga toxin to bind up to 15 Gb 3 molecules at a time [12] (but see also [35]), and the recruitment of Gb 3 to DRMs after ligation by STxB, as shown in this study. As for the second point, it remains to be explained how C22:1 and C23:1 species could have a strong effect on DRM association despite the presence of high quantities of another unsaturated species, C24:1, in plasma membrane preparations from both control and recovery conditions. Another interpretation suggests that specific factors are associated with the plasma membrane Gb 3 pool under steady-state conditions. Upon first-wave Gb 3 binding by STxB, the activity of such factors would be altered, in such a way as to reduce the efficiency of retrograde transport in the recovery condition. The existence of these factors remains hypothetical, and as mentioned above, we have been unable to identify recovery condition-specific STxB crosslinking products. It must, of course, be considered that the licensing factors might be cytosolic. For example, several protein kinases are activated after Shiga toxin binding to Gb 3 [36–41], and further work will be required to address their potential functions in retro- grade Shiga toxin transport in control and recovery conditions. A surprising finding of our study is that the recovery phenotype can be perpetuated over several generations of cell divisions. Indeed, 3 days after first-wave STxB internalization, Gb 3 levels at the plasma membrane are almost fully restored (Fig. 3C), but STxB targeting to the retrograde route is still partially impaired (Fig. 5E). One possible explanation of these unex- pected results is the existence of licensing factors whose activity would be required for Gb 3 association with DRMs and ⁄ or correct sorting to the plasma mem- brane. Even if neo-synthesized, these hypothetical fac- tors would remain trapped in ER ⁄ Golgi structures that contain first-wave internalized STxB–Gb 3 complexes for at least 3 days. Similarly, neo-synthesized Gb 3 could be sequestered by free binding sites on ER ⁄ Golgi-localized first-wave-internalized STxB–Gb 3 com- plexes. In the recovery condition, the association of Shiga toxin with DRMs was selectively reduced. In parallel, retrograde transport to the TGN and the ER was spe- cifically inhibited, without affecting toxin endocytosis. These observations are consistent with the possibility A B C Fig. 7. Analysis of Gb 3 molecular species under control and recov- ery conditions at the plasma membrane, on internal membranes, and in DRMs. Plasma membrane (A), internal membranes (B) and DRMs (C) of HeLa cells in control (white bars) and recovery (gray bars) conditions were purified, Gb 3 was extracted, and molecular species were analyzed by nanospray tandem MS-MS. Results rep- resent the percentage of each detected molecular species of Gb 3 . Means (± SEM) of three independent experiments are shown. In some cases, error bars are too small to be seen. Gb 3 dynamics in HeLa cells T. Falguie ` res et al. 5212 FEBS Journal 273 (2006) 5205–5218 ª 2006 The Authors Journal compilation ª 2006 FEBS that Shiga toxin can enter cells via several endocytic routes. Indeed, it has been reported that, on the one hand, Shiga toxin can be detected in clathrin-coated vesicles [42], and on the other hand, interfering indi- rectly [26,43] or directly [22,23] with clathrin function has minimal effects on Shiga toxin endocytosis, show- ing that Shiga toxin can enter cells efficiently via clath- rin-independent endocytic mechanisms. As opposed to its endocytosis, retrograde sorting of Shiga toxin on early ⁄ recycling endosomes appears to be very selective. Our previous studies have implicated membrane micro- compartmentalization in the early⁄ recycling endo- somes-to-TGN transport step [26]. These studies relied in part on harsh cholesterol extraction conditions. Therefore, it is of importance that the selective recov- ery protocol, as presented in the current article, pro- vides an independent confirmation. Two recent studies have come to the conclusion that early ⁄ recycling endo- somes-to-TGN transport is also dependent on clathrin coats [22,23]. Although unexpected, the possibility of clathrin-dependent trafficking implicating membrane microdomains of the raft type is not entirely unpre- cedented. Activation of the B-cell receptor induces clathrin heavy chain phosphorylation in raft-type microdomains [44], the endocytosis of DRM-associated anthrax toxin is clathrin-dependent [45], and the epi- dermal growth factor receptor could be localized in nascent coated pits that almost invariably contained raft membranes [46]. How raft-type microdomains could favor clathrin-coated pit formation on the early endosome remains to be established. Different scenar- ios can be proposed, such as local overconcentration of lipid-modifying enzymes whose activity would be required for membrane recruitment of clathrin adaptor proteins such as epsinR, a critical factor for efficient retrograde transport at the early ⁄ recycling endosomes– TGN interface [22]. In conclusion, our study provides evidence for the existence of functionally different Gb 3 pools in cells. These pools are in dynamic exchange and are likely to be associated with factors that determine the efficiency of retrograde transport to the ER. In agreement with our earlier studies [22,26], the current work further establishes that the critical step for Shiga toxin trafficking into cells is its retrograde sorting on early ⁄ recycling endosomes, via a mechanism that depends on clathrin coats and involves membrane mic- rocompartmentalization. However, further studies will be necessary to precisely identify the licensing factors necessary for Gb 3 association with DRM and ⁄ or sort- ing at the plasma membrane and, more generally, to unravel the molecular mechanisms involved in the intracellular dynamics of the Gb 3 glycosphingolipid. Experimental procedures Cells and reagents HeLa cells were cultured as previously described [16]. STxB, STxB–Glyc–KDEL, STxB–Sulf 2 , and STxB–K 3 were purified as previously described [16,17,26]. Anti-CTR433 and anti-TfR H68.4 IgG, and cationic colloidal silica, were kind gifts from M. Bornens (UMR 144-Institut Cur- ie ⁄ CNRS, Paris, France), I. Trowbridge (The Salk Institute, San Diego, CA), and D. Stolz (Department of Pathology, Pittsburg, PA), respectively. The monoclonal (13C4) and polyclonal antibodies against STxB were obtained as previ- ously described [16,17]. PPMP (Calbiochem, La Jolla, CA), Texas-red coupled anti-rabbit serum, fluorescein isothiocya- nate-coupled anti-mouse serum and alkaline phosphatase- coupled secondary antibodies (Jackson Immunoresearch, West Grove, PA), HPTLC plates (Merck, Darmstadt, Germany), enhanced chemifluorescence substrate (Amer- sham Biosciences, Little Chalfont, UK), streptavidin cou- pled to horseradish peroxidase (streptavadin–horseradish peroxidase) (Roche, Basel, Switzerland), polyacrylic acid (Aldrich, St Louis, MO), anti-calnexin, anti-[early endo- somal antigen-1 (EEA1)] and anti-(caveolin-1) IgG (BD Biosciences, San Diego, CA), anti-(flotillin-1) IgG (Santa Cruz Biotechnology, Santa Cruz, CA) and immobilized streptavidin (NHS–SS–biotin) (Pierce, Rockford, IL) were obtained from the indicated commercial sources. Optiprep, Nycodenz, SigmaCote, thymidine-5¢-monophosphate-p- nitrophenyl ester, 4-methylumbelliferyl-d-mannopyranoside, 4-methylumbelliferyl-N-acetyl-b-d-glucosaminide, CTxB and o-phenylenediamine dihydrochloride peroxidase substrate were obtained from Sigma (St Louis, MO). Immunofluorescence analysis on PPMP-treated cells HeLa cells were treated or not treated with 5 lm PPMP for 6 days. Immunofluorescence was determined as previously described [17]. Briefly, cells were incubated with: (a) 25 lm STxB for 45 min at 37 °C to allow its fluid-phase endocyto- sis in PPMP-treated cells; or (b) 1 l m STxB bound at 4 °C and then chased for 45 min at 37 °C after washes in control cells. Cells were then fixed in 3% paraformaldehyde for 15 min at room temperature, quenched with ammonium chloride, and permeabilized with 0.05% saponin. STxB, the Golgi marker CTR433 and TfR were labeled with poly- clonal anti-STxB, monoclonal anti-CTR433, or monoclonal anti-TfR, and visualized with the use of adapted fluoro- chrome-coupled secondary antibodies. Then, coverslips were mounted and analyzed by confocal microscopy (Leica Microsystems, Mannheim, Germany). At the same time, the loss of Gb 3 expression from cells treated with PPMP was verified using the glycolipid extraction procedure (see below). T. Falguie ` res et al. Gb 3 dynamics in HeLa cells FEBS Journal 273 (2006) 5205–5218 ª 2006 The Authors Journal compilation ª 2006 FEBS 5213 Glycolipid extraction and analysis by TLC Glycolipid extraction was performed as previously des- cribed [26]. Briefly, HeLa cells were lysed in water and subjected to partition against chloroform to separate the neutral lipids from the other cellular components. After saponification for 1 h at 56 °C in methanol ⁄ KOH, the products were re-extracted with chloroform, dried under nitrogen, and spotted onto HPTLC plates. After migra- tion in chloroform ⁄ methanol ⁄ water (65 : 25 : 4), the plates were overlaid with STxB, polyclonal anti-STxB and alkaline phosphatase-coupled serum, and visualized by enhanced chemifluorescence; Gb 3 expression was then quantified. Plasma membrane enrichment and characterization We used a published procedure [33] with some modifica- tions. For each enrichment experiment, 10 8 HeLa cells were used. The cells were trypsinized, incubated or not with 1 lm of STxB on ice, and then shifted for 1 h to 37 °C. After this point of the procedure, all plastic and glass materials were coated with SigmaCote. After washes in ice- cold NaCl ⁄ P i and plasma membrane-coating buffer (PMCB) [20 mm 2-(N-morpholino)ethanesulfonic acid, 150 mm NaCl, 280 mm sorbitol], cells were incubated in a glass tube with 2% cationic colloidal silica in PMCB, and then neutralized with 1 mgÆmL )1 polyacrylic acid in PMCB. After washes in PMCB, cells were mechanically lysed in 1.3 mL of lysis buffer (2.5 mm imidazole, pH 7.0) through needles: 24 times with G22, and 12 times with G27. Lysates were mixed with 1 mL of 100% Nycodenz (50% final) and overlaid on 0.5 mL of 70% Nycodenz in an SW55 centrifuge tube. The rest of the lysate (300 lL) was used for the characterization of the procedure. Tubes were filled to 5 mL with lysis buffer and spun for 25 min at 20 000 g at 4 °C in a swinging bucket rotor (SW55, Beckman Coulter, Fullerton, CA). The supernatant was collected, and the silica content in the pellet and the 50– 70% interface were washed in lysis buffer, mixed in 50% Nycodenz, and submitted to another ultracentrifugation under the same conditions. The supernatant was collected and mixed with the first one. The pellet was washed three times with lysis buffer and resuspended in 1 mL of the same buffer for further analysis. Lysates and plasma membrane fractions were character- ized for their content of total proteins (Bradford Protein Assay; BioRad, Hercules, CA) and several organelle mark- ers. The plasma membrane marker alkaline phosphodiest- erase was colorimetrically assessed in 100 mm Tris ⁄ HCl (pH 9.0) ⁄ 40 mm CaCl 2 using 2 mgÆmL )1 thymidine-5¢- monophosphate-p-nitrophenyl ester as substrate; after 30 min, absorbance at 400 nm was detected. The Golgi marker mannosidase II was assessed fluorometrically in NaCl ⁄ P i containing 0.1% Triton X-100 using 5 mm 4- methylumbelliferyl-d-mannopyranoside as substrate. The lysosomal marker b-hexosaminidase was also assessed fluorometrically in 10 mm citric acid ⁄ 30 mm Na 2 HPO 4 (pH 4.5) with 0.1% Triton X-100 and 2.3 mgÆmL )1 4-meth- ylumbelliferyl-N-acetyl-b-d-glucosaminide as substrate. For the last two fluorometric assays, fluorescence was read after 30 min at 37 °C with excitation at 355 nm and emission at 460 nm. Free cellular cholesterol content was measured as described [26]. The ER marker calnexin, the early endosom- al marker EEA1 and the DRM markers caveolin-1 and flo- tillin-1 were assessed by western blot after migration on 10% SDS ⁄ PAGE, semidry transfer (BioRad) on nitrocellu- lose membrane, and successive incubation with primary antibodies and alkaline phosphatase-coupled secondary antibodies. After visualization with enhanced chemifluores- cence and scanning of membranes with phosphorimager (Amersham Biosciences) in the blue chemiluminescence mode, signals were quantified with imagequant (Amer- sham Biosciences). Results were expressed as the percentage of marker in the plasma membrane fraction compared to the total lysate. Biochemical analysis of STxB retrograde transport, association with DRMs, degradation and recycling These experiments were done on HeLa cells in 24-well plates (10 5 cells per well) under the indicated control, PPMP (5 lm for 5 h at 37 °C), or recovery conditions. STxB–Glyc–KDEL iodination, glycosylation and Scatchard analysis were performed as previously described [16]. Sulfa- tion analysis was performed as previously described [17], with similar results being obtained for 30 min or 4 h incu- bations. Iodinated STxB–Glyc–KDEL was used to measure the association of STxB with DRM. DRMs were isolated as previously described, and fraction 2 of each gradient was characterized as the DRM fraction that contains GM 1 and no TfR [26]. Degradation and recycling of first wave-internalized STxB were measured as follows. Prebound iodinated STxB–Glyc–KDEL was internalized into HeLa cells at 37 °C for 0, 1, 24, 48 or 72 h. Culture supernatants and cell lysates in 0.1 m KOH were submitted to 10% trichloroacetic acid precipitation for 30 min at 4 °C. After centrifugation at 13 000 g for 30 min at 4 °C in a bench- top centrifuge (Eppendorf, Hamburg, Germany), trichloro- acetic acid-precipitated and soluble materials were analyzed using a gamma-counter. Culture supernatant STxB was expressed as trichloroacetic acid-precipitated counts in the culture supernatant, and degraded STxB as trichloroacetic acid-soluble counts in culture supernatant and cell lysates. Gb 3 dynamics in HeLa cells T. Falguie ` res et al. 5214 FEBS Journal 273 (2006) 5205–5218 ª 2006 The Authors Journal compilation ª 2006 FEBS [...]... & Tesh VL (2000) Shiga toxin- induced tumor necrosis factor alpha expression: requirement for toxin enzymatic activity and monocyte protein kinase C and protein tyrosine kinases Infect Immun 68, 5183–5189 37 Foster GH & Tesh VL (2002) Shiga toxin 1-induced activation of c-Jun NH(2)-terminal kinase and p38 in the human monocytic cell line THP-1: possible involvement in the production of TNF-alpha J Leukoc... length on verotoxin ⁄ globotriaosyl ceramide binding in model membranes: comparison of a supported bilayer film and liposomes Glycocon J J13, 159–166 31 Binnington B, Lingwood D, Nutikka A & Lingwood CA (2002) Effect of globotriaosyl ceramide fatty acid alpha-hydroxylation on the binding by verotoxin 1 and verotoxin 2 Neurochem Res 27, 807–813 32 Mallard F & Johannes L (2002) Shiga toxin B-subunit as... ceramide induces intracellular signals that mediate cytoskeleton remodeling in human renal carcinoma-derived cells J Cell Sci 117, 3911–3922 40 Katagiri YU, Mori T, Nakajima H, Katagiri C, Taguchi T, Takeda T, Kiyokawa N & Fujimoto J (1999) Activation of src family kinase yes induced by shiga toxin binding to globotriaosyl ceramide (Gb3 ⁄ CD77) in low density, detergent-insoluble microdomains J Biol... distinct Gb3 ⁄ CD77 signaling pathways leading to apoptosis are triggered by antiGb3 ⁄ CD77 mAb and verotoxin-1 J Biol Chem 278, 45200–45208 8 van Setten PA, Monnens LA, Verstraten RG, van den Heuvel LP & van Hinsbergh VW (1996) Effects of verocytotoxin-1 on nonadherent human monocytes: binding characteristics, protein synthesis, and induction of cytokine release Blood 88, 174–183 5216 9 Yuhas Y, Kaminsky... cells for 30 min at 4 °C After lysis, DRMs were prepared as described [26], and the percentage of CTxB associated with DRMs was determined by quantitative western blotting Gb3 dynamics in HeLa cells SS–biotin was bound to cells under control, recovery and PPMP conditions The cells were then incubated from 0 to 40 min at 37 °C After washes, cells were split in to two equal fractions that were incubated... was determined as the ratio of radioactivity after acid wash (internal Tf) and signal without acid wash (total cell-associated Tf) Analysis of Gb3 molecular species Measurement of protein biosynthesis HeLa cells were cultured in 96-well plates, incubated or not with 1 lm STxB on ice, and shifted for 1 h or 3 days to 37 °C After washes, protein biosynthesis was determined using [35S]methionine incorporation,... and the level of protein synthesis under recovery conditions was calculated as a percentage of protein synthesis detected on mock-treated cells STxB and Tf internalization assays HeLa cells were used under the same conditions as described above STxB–K3 was coupled to NHS–SS–biotin following the manufacturer’s instructions, and the resulting protein (STxB–SS–biotin) was used to measure STxB internalization,... T, Sekino T, Sato N, Ohmi K, Nakajima H, Takeda T et al (2000) Globotriaosyl ceramide (CD77 ⁄ Gb3) in the glycolipid-enriched membrane domain participates in B-cell receptor-mediated apoptosis by regulating lyn kinase activity in human B cells Exp Hematol 28, 1260– 1268 39 Takenouchi H, Kiyokawa N, Taguchi T, Matsui J, Katagiri YU, Okita H, Okuda K & Fujimoto J (2004) Shiga toxin binding to globotriaosyl. .. cancer therapy Crit Rev Oncol Hematol 39, 99–106 5 Johannes L & Decaudin D (2005) Protein toxins: intracellular trafficking for targeted therapy Gene Ther 12, 1360–1368 6 Maloney MD, Binnington-Boyd B & Lingwood CA (1999) Globotriaosyl ceramide modulates interferonalpha-induced growth inhibition and CD19 expression in Burkitt’s lymphoma cells Glycocon J J16, 821– 828 ´ ` 7 Tetaud C, Falguieres T, Carlier... K, Lindberg AA & Keusch GT (1992) Shiga toxin: biochemistry, genetics, mode of action, and role in pathogenesis Curr Top Microbiol Immunol 180, 65–94 12 Ling H, Boodhoo A, Hazes B, Cummings MD, Armstrong GD, Brunton JL & Read RJ (1998) Structure of Shiga- like toxin I B-pentamer complexed with an analogue of its receptor Gb3 Biochemistry 37, 1777–1788 13 Falnes PO & Sandvig K (2000) Penetration of protein . species composition of the cel- lular Gb 3 [28,29]. In in vitro binding assays, the fatty acid chain of Gb 3 was found to in uence the binding to Shiga toxin [30,31]. In this study, we investigated. interfering indi- rectly [26,43] or directly [22,23] with clathrin function has minimal effects on Shiga toxin endocytosis, show- ing that Shiga toxin can enter cells efficiently via clath- rin-independent. microdomains of the raft type often increases upon ligand binding. Plasma membrane dynamics of Gb 3 In the next step, we investigated how Shiga toxin in uences the plasma membrane dynamics of Gb 3 . Ultrastructural