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
Trang 1globotriaosyl ceramide, in HeLa cells
Thomas Falguie`res1,*, Winfried Ro¨mer1, Mohamed Amessou1, Carlos Afonso2, Claude Wolf3, Jean-Claude Tabet2, Christophe Lamaze1and Ludger Johannes1
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 (Gb3 or CD77) is a
glyco-sphingolipid that was initially described as the rare PK
blood group antigen [1] Gb3 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 Gb3 [4,5] The physiologic
func-tion of Gb3 is still unknown Some studies have
suggested that Gb3 could regulate the function of
signaling molecules, such as type I interferon receptors and CD19 [6] Indeed, Gb3 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, Gb3 binding by nat-ural ligands or antibodies has been shown to induce apoptosis [7]
Gb3 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 Gb3 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
GM1 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; Gb3, 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.
Trang 2cellular 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 Gb3 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 Gb3 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 Gb3 [28,29] In in vitro binding assays, the fatty
acid chain of Gb3 was found to influence the binding
to Shiga toxin [30,31]
In this study, we investigated the Gb3 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
Gb3 was surprisingly dynamic, in that after its Shiga
toxin-induced internalization, the plasma membrane
pool of Gb3 rapidly recovered However, we observed
that retrograde transport to the Golgi apparatus and
the ER was significantly less efficient on recovered Gb3
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–Gb3 system was selectively targeted, and our data strongly suggest the existence of plasma membrane Gb3 pool-specific fac-tors, possibly the molecular species composition of
Gb3 itself, that are selectively required for efficient ret-rograde transport
Results
Gb3is required for retrograde transport of Shiga toxin from endosomes to the TGN
The glycosphingolipid Gb3 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 Gb3 to levels below 5% of those in untreated control cells Under these conditions, the
4C 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 Gb3 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 Gb3to DRMs The above-described experiment shows that Gb3 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 Gb3 In a first experiment,
we analyzed whether Shiga toxin would recruit Gb3 to DRMs Conditions were established in which, at steady state, about 10% of cellular Gb3 was in DRM
Trang 3fraction 2 (Fig 2A,B) After incubation of cells with
STxB at saturating concentrations, Gb3 association
with DRMs was increased 2.5-fold Gb3 thus behaved
like protein receptors whose association with
mem-brane microdomains of the raft type often increases
upon ligand binding
Plasma membrane dynamics of Gb3
In the next step, we investigated how Shiga toxin
influences the plasma membrane dynamics of Gb3
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 Gb3 and cholesterol in the plasma membrane-enriched fractions were then quantified It was found that about 50% of the Gb3 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 Gb3 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
37C (0–60 min), the proportion of Gb3 in plasma membrane fractions was determined At the 0 min time point, about 50% of Gb3 was in plasma membrane fractions (Fig 3C), as described above (Fig 3A) Fol-lowing a short incubation at 37C, a transient decrease of Gb3 in these fractions to 28% was observed Sixty minutes after the shift to 37C, a time point at which STxB is quantitatively localized in the Golgi apparatus [17], Gb3 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, Gb3 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 Gb3 was cointernalized with Shiga toxin, and that the plasma membrane pool of Gb3 was then rapidly replenished with Gb3from internal stores
Cell biological analysis of the recovery condition
As the steady-state plasma membrane Gb3 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.
Trang 4of applying the plasma membrane enrichment
proce-dure at the end of each incubation period at 37C,
radiolabeled [125I]STxB was bound to the cells on ice
At the 0 min time point, [125I]STxB binding was
strongly reduced, as expected (Fig 5A) Upon
incuba-tion at 37C, binding then readily recovered,
parallel-ing the recovery of plasma membrane Gb3described in
Fig 3C The plateau level of [125I]STxB rebinding to
cells was reached after 60 min at 82% (Fig 5A) These
results thus confirm the Gb3 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, Kd 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, Gb3levels 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 #).
Fig 2 STxB recruits Gb 3 to DRMs (A) HeLa cells were incubated (+ STxB) or not incubated (– STxB) with 1 l M 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, Gb3was quantified
in each fraction using TLC and overlay assays DRMs are enriched in fraction 2 The percentage of Gb3in the DRM fraction is indicated (B) Means (± SEM) of three independent experiments as shown in (A).
Trang 5led 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 Gb3 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 Gb3 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.
Trang 6recovery phenotype is restricted to the STxB–Gb3
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 Gb3pools
Several studies have suggested that specific molecular
species of Gb3 are correlated with efficient retrograde
transport [28,29] Therefore, we analyzed the molecular
species composition of the plasma membrane and
internal pools of Gb3, under both control and recovery conditions (Fig 7) After plasma membrane or DRM enrichment, glycolipids were extracted, and Gb3 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 Gb3expression 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 [125I]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.
Trang 7the 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 Gb3molecular species
composi-tion was similar in plasma membrane (Fig 7A) and
internal pools (Fig 7B), indicating that at steady state,
Gb3 localization is not dictated by parameters such as
membrane thickness As a further test, we compared
the molecular species composition of Gb3 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 Gb3 We found that Gb3 was mobilized during Shiga toxin internalization, and the plasma membrane Gb3 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 Gb3 pool-specific deter-minants that modify the efficacy of retrograde trans-port
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 transtrans-port 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 l M STxB for 30 min at 4C,
followed by a 1 h internalization at 37C (Recovery) Then, 30 n M to
1 l M [125I]STxB–Glyc–KDEL was bound to the cells for 2 h at 4C.
After washes and lysis of the cells, the results were expressed as
a Scatchard representation, and Kdand number of sites per cell
were deduced for each condition Means (± SEM) of three different
experiments are shown.
Number of sites (· 10 6 per cell)
Trang 8Many 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 Gb3 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 Gb3 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 Gb3
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 Gb3 molecules at a time [12] (but see also [35]), and the recruitment of Gb3 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 Gb3 pool under steady-state conditions Upon first-wave Gb3 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 Gb3 [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, Gb3levels 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 Gb3 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–Gb3 complexes for at least 3 days Similarly, neo-synthesized Gb3 could be sequestered by free binding sites on ER⁄ Golgi-localized first-wave-internalized STxB–Gb3 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, Gb3was 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.
Trang 9that 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 diindi-rectly [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 Gb3 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 Gb3association 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 Gb3glycosphingolipid
Experimental procedures
Cells and reagents HeLa cells were cultured as previously described [16] STxB, STxB–Glyc–KDEL, STxB–Sulf2, and STxB–K3were 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 37C to allow its fluid-phase endocyto-sis in PPMP-treated cells; or (b) 1 lm STxB bound at 4C and then chased for 45 min at 37C 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, monopoly-clonal anti-CTR433, or monopoly-clonal 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 Gb3 expression from cells treated with PPMP was verified using the glycolipid extraction procedure (see below)
Trang 10Glycolipid 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 56C 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; Gb3 expression was then
quantified
Plasma membrane enrichment and
characterization
We used a published procedure [33] with some
modifica-tions For each enrichment experiment, 108HeLa cells were
used The cells were trypsinized, incubated or not with
1 lm of STxB on ice, and then shifted for 1 h to 37C
After this point of the procedure, all plastic and glass
materials were coated with SigmaCote After washes in
ice-cold NaCl⁄ Pi 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 4C 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 CaCl2 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⁄ Pi 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 Na2HPO4
(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 37C 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 (105cells per well) under the indicated control, PPMP (5 lm for 5 h at 37C), 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 GM1and
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
37C 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 4C After centrifugation at 13 000 g for 30 min at 4C 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