HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

J. Biol. Chem., Vol. 277, Issue 18, 16249-16256, May 3, 2002

This Article Abstract Full Text (PDF) All Versions of this Article: 277/18/16249    most recent M109834200v1 Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Wolf, A. A. Articles by Lencer, W. I. Search for Related Content PubMed PubMed Citation Articles by Wolf, A. A. Articles by Lencer, W. I. Social Bookmarking                      What's this?

Uncoupling of the Cholera Toxin-G_M1 Ganglioside Receptor Complex from Endocytosis, Retrograde Golgi Trafficking, and Downstream Signal Transduction by Depletion of Membrane Cholesterol^*

Anne A. Wolf, Yukako Fujinaga, and Wayne I. Lencer^§¶

From  Gastrointestinal Cell Biology, Department of Pediatrics, Children's Hospital and Harvard Medical School, Boston, Massachusetts 02115 and the ^§ Harvard Digestive Diseases Center, Boston, Massachusetts 02115

Received for publication, October 11, 2001, and in revised form, February 2, 2002

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

To induce toxicity, cholera toxin (CT) must first bind ganglioside G_M1 at the plasma membrane, enter the cell by endocytosis, and then traffic retrograde into the endoplasmic reticulum. We recently proposed that G_M1 provides the sorting motif necessary for retrograde trafficking into the biosynthetic/secretory pathway of host cells, and that such trafficking depends on association with lipid rafts and lipid raft function. To test this idea, we examined whether CT action in human intestinal T84 cells depends on membrane cholesterol. Chelation of cholesterol with 2-hydroxypropyl -cyclodextrin or methyl -cyclodextrin reversibly inhibited CT-induced chloride secretion and prolonged the time required for CT to enter the cell and induce toxicity. These effects were specific to CT, as identical conditions did not alter the potency or toxicity of anthrax edema toxin that enters the cell by another mechanism. We found that endocytosis and trafficking of CT into the Golgi apparatus depended on membrane cholesterol. Cholesterol depletion also changed the density and specific protein content of CT-associated lipid raft fractions but did not entirely displace the CT-G_M1 complex from these lipid raft microdomains. Taken together these data imply that cholesterol may function to couple the CT-G_M1 complex with raft domains and with other membrane components of the lipid raft required for CT entry into the cell.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Vibrio cholerae causes worldwide epidemics of life-threatening secretory diarrhea by colonizing the intestinal lumen and producing cholera toxin (CT),¹ a potent enterotoxin that invades the intestinal epithelial cell as a fully folded protein. Structurally, CT consists of two components. The pentameric B-subunit binds stoichiometrically to five G_M1 gangliosides on the apical (lumenal) surface of intestinal epithelial cells, and the enzymatic A-subunit activates adenylyl cyclase inside the cell by catalyzing the ADP-ribosylation of the heterotrimeric GTPase G_s (1). Activation of adenylyl cyclase in intestinal crypt epithelial cells leads to Cl secretion, the fundamental transport event in secretory diarrhea.

To induce disease, both A- and B-subunits must enter the host epithelial cell as a fully assembled holotoxin by moving retrograde through the biosynthetic pathway into the ER,² where the A-subunit unfolds, dissociates from the B-pentamer, and translocates to the cytosol, presumably by dislocation through the protein conducting channel sec61p (2). The time between CT binding to G_M1 at the cell surface and induction of toxicity has been termed the "lag phase." The lag phase corresponds to the time required for trafficking CT into the ER, unfolding of the A-subunit by interaction with protein-disulfide isomerase, and finally dislocation of the A-subunit into the cytosol (2, 3).

In the model polarized intestinal epithelial cell line T84, CT function depends on B-subunit binding to (and possibly clustering) ganglioside G_M1. Binding to G_M1 anchors CT to the host cell membrane and associates CT with cholesterol-rich detergent-insoluble membrane microdomains, termed DIGs (detergent-insoluble glycosphingolipid-rich membrane microdomains) or lipid rafts (4). G_M1 exhibits specificity for CT action in human intestinal epithelial cells, as the ganglioside G_D1a does not substitute for G_M1 as a receptor for toxin action or for association with lipid rafts in this cell type (4, 5). Chimeric toxins, for example, containing the A-subunit of CT assembled with the B-subunit of Escherichia coli toxin LTIIb, bound only to G_D1a, did not associate with lipid rafts, and did not lead to a functional response (4). Based on these studies, we have recently proposed that the B-subunit-G_M1 complex represents the sorting motif necessary for toxin trafficking retrograde into the Golgi apparatus and possibly ER of host eukaryotic cells. We have also proposed that such trafficking depends on association with lipid rafts and lipid raft function. To test this idea, we now examine whether CT action depends on membrane cholesterol.

Lipid rafts are highly enriched in cholesterol, GPI-linked proteins, and glycosphingolipids, including ganglioside G_M1, the receptor for CT. Abundant evidence indicates that raft structure and function in cellular metabolism, including certain forms of ligand-induced signal transduction, protein and lipid sorting, endocytosis, and transcytosis, depend critically on cholesterol (4, 6-15). Demonstration of such sensitivity to membrane cholesterol has been taken widely as evidence for dependence on raft function in these cellular processes. With respect to CT and aerolysin toxin, another bacterial enterotoxin that binds membrane receptors located in lipid rafts, disruption of membrane cholesterol by the sterol-binding molecule filipin or by chelation with -cyclodextrin has been shown to effect endocytosis and toxin action (16, 17). On the other hand, whether lipid rafts function directly in CT endocytosis has recently been questioned (18). In some cellular systems, clathrin-mediated endocytosis displays sensitivity to disruption of membrane cholesterol (19, 20), and although CT binds G_M1 located both in lipid rafts and clathrin-coated pits, the toxin enters hippocampal neurons and A431 epithelial cells by clathrin-mediated mechanisms (18, 19).

In the current study, we utilize 2-hydroxy and methyl -cyclodextrin to examine CT trafficking and function in model epithelial cells acutely depleted in cholesterol (22-24). To control for general effects of cholesterol depletion on membrane dynamics, we utilize anthrax edema toxin (EdTx). Like CT, anthrax EdTx is a two component toxin. Both CT and EdTx enter polarized monolayers of intestinal T84 cells by receptor-mediated endocytosis, and both toxins induce an identical cAMP-dependent Cl secretory response, but by different mechanisms. The enzymatic component of EdTx, termed edema factor, is itself a potent calmodulin-dependent adenylyl cyclase that, unlike CT, translocates to the cytosol directly across the endosome membrane. Membrane translocation of edema factor and the induction of toxicity depend critically on entry of EdTx into an acidic endosomal compartment. Retrograde transport into Golgi cisternae or ER is not required (25). In addition, unlike CT, the membrane-binding component of EdTx, termed protective antigen, recognizes a receptor on T84 cell membranes that displays strict basolateral polarity, and the EdTx-receptor complex does not fractionate with lipid rafts (25).

Our data show that endocytosis and trafficking of CT into Golgi cisternae, as well as CT-induced Cl secretion, depend on membrane cholesterol. Cholesterol depletion altered lipid raft structure and presumably function, and only partially displaced the CT-G_M1 complex from lipid raft microdomains. Cholesterol depletion had no detectable affect on entry of anthrax EdTx into acidic endosomes, as assessed as an EdTx-induced Cl secretory response. These data indicate that CT trafficking depends on toxin association with lipid rafts and imply that cholesterol may function in lipid raft structure to couple the CT-G_M1 complex with raft domains and with other membrane components involved in membrane sorting or downstream signal transduction required for CT entry into the cell.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Materials and Antibodies-- Cholera toxin and cholera toxin B-subunit coupled to horseradish peroxidase (HRP) were obtained from Calbiochem (San Diego, CA) or were recombinantly expressed. Anthrax edema toxin was a kind gift from Dr. R. John Collier (Department of Microbiology, Harvard Medical School, Boston, MA). 2-Hydroxy -cyclodextrin, methyl -cyclodextrin, cholesterol, and cholesterol analogues were obtained from Sigma. Mouse monoclonal antibody to pp60^src was obtained from Upstate Biotechnology, Inc. (Lake Placid, NY) rabbit polyclonal antibody to caveolin-1 from Santa Cruz Biotechnology (Santa Cruz, CA) and mouse monoclonal antibody to ecto-5'-nucleotidase from Linda Thompson (Oklahoma Medical Research Foundation, Oklahoma City, OK). Rabbit polyclonal antibodies to CT B-subunits were previously described (26). All other chemicals were from Sigma unless otherwise stated.

Cell Culture-- T84 cells obtained from American Type Culture Collection (Rockville, MD) were cultured and passaged as previously described (27). Passages 80-92 were used for these studies. Monolayers used for electrophysiology and for toxin binding studies were grown on 0.33-cm² polycarbonate filters (Costar, Corning, NY) and used 12-14 days after plating. Monolayers used for isolation of lipid rafts were grown on 45-cm² filters and used 21 days after plating. Hanks' balanced salt solution (HBSS) to which 10 mM HEPES was added and pH adjusted to 7.4 was used for all assays unless otherwise noted.

Cholesterol Depletion and Electrophysiology-- Initial studies showed that cyclodextrins induced a progressive concentration- and time-dependent decrease in transepithelial resistance. We thus defined conditions for acute cholesterol depletion that preserved monolayer resistance using either 16.5 mM 2-hydroxypropyl-cyclodextrin (2OH-CD) or 4 mM methyl -cyclodextrin (m-CD) applied to apical or basolateral cell surfaces of T84 cell monolayers for 1 h.

To assay for cholesterol depletion, total cell lipids from T84 cell monolayers were extracted in chloroform, dried, and then resuspended in minimal volume (3 µl) of chloroform. Cholesterol content was analyzed per milligram of total cell protein by colorimetric assay using the Infinity cholesterol reagent according to the manufacture's directions (Sigma Diagnostics). Absorbance was read at 500 nM on a Spectronic21D spectrophotometer (Milton Roy Products).

For assessment of Cl secretion (measured as a short circuit current, Isc), cells were treated as above with the appropriate cyclodextrin analogue for 1 h at 37 °C before adding CT (20 or 1 nM) to apical or basolateral reservoirs, respectively, in the continued presence or absence of cyclodextrin. The time course of toxin-induced Cl secretion (Isc) and transepithelial resistance were measured as previously described (28). To provide two point calibration for all studies, the cAMP agonist vasoactive intestinal peptide (VIP, 10 nM) was applied to basolateral reservoirs at the end of each experiment (either 60 or 85 min after application of toxin). To replete monolayers with cholesterol, cholesterol and related sterols were complexed with m-CD as previously described and used at a final concentration of 0.2 mM sterol (29).

Toxin Binding-- Specific toxin binding to apical membranes of T84 cell monolayers treated or not treated with 16.5 mM 2OH-CD at 37 °C was measured. All incubations were done in buffer made of HBSS containing 0.1% bovine serum albumin. After pretreatment, monolayers were cooled to 4 °C, and incubated with buffer alone or 100 nM CT B-subunit (to measure nonspecific binding) in buffer for 20 min. Monolayers were then washed and incubated apically with 20 nM CT B-subunit-HRP in buffer for 30 min on ice. Unbound toxin was removed by three washes with ice-cold HBSS. Cell surface-bound CT B-HRP was assessed by colorimetric assay using 30% hydrogen peroxide in 1 mM 2,2'-azino-di-3-ethylbenzthiazoline sulfonic acid in 100 mM citrate-phosphate buffer, pH 4.2, as previously described (4). Specific binding of toxin to the cell surface was determined by subtracting nonspecific binding from total surface-bound toxin.

Endocytosis-- T84 cells grown for 2 weeks on 0.33-cm² filter supports were treated with 4 mM m-CD or buffer alone for 1 h at 37 °C, brought to 4 °C in HBSS, and incubated with 20 nM CT in the presence of 4 mM m-CD or buffer alone for 45 min at 4 °C. Monolayers were then warmed to 37 °C in the presence of apical 20 nM CT 16.5 mM m-CD or buffer for the indicated times, or kept at 4 °C for the duration of the experiment. All monolayers were then washed in ice-cold buffer to remove unbound toxin. Where indicated, cell surface-bound CT was removed by immersing each monolayer in HBSS, pH 7.4, at 37 °C for 10 s (to further release unbound toxin at the cell surface) and then immediately transferred to HBSS, pH 2.5, at 4 °C for 5 min. Each monolayer was then immersed for 5 min in HBSS, pH 7.4, at 4 °C, and incubated again in HBSS, pH 2.5, at 4 °C for an additional 5 min. 97.5% ± 16.5% (n = 4) of CT was stripped from the cell surface using this procedure. After removal of cell surface-bound toxin, monolayers were cut from their filter supports, immersed in 5% SDS, and boiled for analysis of total cell-associated CT B-subunit.

Sucrose Equilibrium Density Centrifugation-- Sucrose equilibrium density centrifugation was performed as described previously (4). One or two confluent 45-cm² monolayers of T84 cells were used for isolation of detergent-insoluble membranes. All steps were performed at 4 °C. Cells were scraped into 2 ml of ice-cold 50 mM Tris-buffered saline containing 16.5 mM Tween 20 and a protease inhibitor tablet (containing EDTA, pancreas extract, Pronase, thermolysin, chymotrypsin, trypsin, papain) from Roche Molecular Biochemicals and homogenized by five strokes in a tight-fitting Dounce homogenizer on ice. The homogenate was adjusted to 40% sucrose by adding 2 ml of 80% sucrose in Tris-buffered saline containing Tween 20, layered under a continuous 5-30% sucrose gradient, and centrifuged at 39,000 rpm for 16-20 h in a swinging bucket rotor (model SW41; Beckman Instruments, Palo Alto, CA). The presence of a floating membrane fraction was noted visually, and 0.5-ml fractions were collected from the top. For step gradients, the cell homogenates in 4 ml of 40% sucrose were layered first under 5 ml of 30% sucrose, and then 5 ml of 5% sucrose was added to the top of the gradient. The step gradients were centrifuged as described above. Raft fractions were collected from the 30-5% interface and combined. Soluble material remained in the 40% sucrose fraction at the bottom of the gradient. For all gradients, fractions were normalized for protein content (Pierce BCA protein assay), analyzed by SDS-PAGE, Western-blotted for the indicated proteins as previously described (30), and quantified by densitometry using a Kodak Digital Science Image Station 440 CF (PerkinElmer Life Sciences). All signals were in the linear range. Raft and soluble fractions from some experiments were analyzed for 5'-nucleotidase (CD73) activity as previously described (31).

Retrograde Golgi Transport Assay-- CT holotoxin was engineered to contain the sulfation consensus motif SAEDYEYPS at the C terminus of its B-subunits.² Recombinant toxins were produced and purified as previously described (32). For in vivo sulfation experiments, T84 cells were first washed, incubated in sulfate-free HBSS for 30 min at 37 °C, and then incubated again for an additional 1 h in fresh sulfate-free HBSS. The monolayers were then incubated with 0.5 mCi/ml Na₂³⁵SO₄ in the same buffer for 30 min. The CT-variant toxin containing the sulfation motif was added to T84 cells both apically and basolaterally to a final concentration of 20 nM, incubated for 50 min at 37 °C, and then washed twice with ice-cold HBSS. Following total cell lysis, an immunoprecipitation using rabbit anti-CT-B antibodies as described previously (33) was performed. Samples were run on 10-20% denaturing Tris·HCl polyacrylamide gels (Bio-Rad), and analyzed with a PhosphorImager (Molecular Dynamics Inc, Sunnyvale, CA).

Statistics-- Data were analyzed using Statview 512+ software (Brainpower, Inc., Calabasa, CA)

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

We utilized the cyclodextrin analogues 2OH-CD and m-CD to determine whether the mechanism of CT action depends on membrane cholesterol. Initial studies showed that treatment of T84 cell monolayers with 2OH-CD caused a dose- and time-dependent loss of transepithelial resistance (TER), presumably because of direct effects of cholesterol depletion on tight junction structure (34, 35). Conditions were thus defined that depleted ~50% of total cell cholesterol but maintained monolayer resistance intact throughout the time course of each experiment. Under optimal conditions, however, transepithelial resistance still fell progressively to 15% of initial values by the end of each time course (from 1035 ± 54 to 151 ± 22 , n = 10), and the cAMP agonist VIP was used in all studies to calibrate measurement of Isc. Like CT, VIP increases intracellular cAMP in T84 cells by activating the heterotrimeric GTPase G_s. All subsequent events in the Cl secretory response are identical. The alternative sterol-binding agent filipin could not be used in these experiments to deplete membrane cholesterol, as short 15-min incubations with filipin (7.6 µM) caused complete loss of monolayer resistance that prevented assessment of cell function by electrophysiology.

T84 cells treated with 2OH-CD exhibited an attenuated response to apically applied CT. Fig. 1A shows a representative time course of CT-induced Cl secretion in monolayers depleted or not depleted in cholesterol. VIP was applied at 90 min to control for the effect of 2OH-CD on monolayer resistance as described above. 2OH-CD prolonged the "lag phase" by nearly 2-fold (Fig. 1, A and B) and diminished the apparent rate of toxin-induced Cl secretion by >60% (Isc = 0.98 ± 0.10 versus 0.39 ± 0.08 µA/cm²/min, n = 8, control versus 2OH-CD-treated monolayers, p < 0.05). Peak Isc responses to apically applied CT were inhibited by 50% when calibrated against that induced by VIP (Fig. 1, A and C). 2OH-CD, however, had no detectable effect on the time course of Isc induced by basolaterally applied CT (data not shown). Two lines of evidence indicated that treatment with 2OH-CD did not inhibit CT action by depleting the toxin's receptor ganglioside G_M1 from cell membranes. First, T84 monolayers treated or not with 2OH-CD exhibited nearly identical densities of apical membrane receptors for CT (Fig. 2A). Second, the effects of 2OH-CD were reversed completely by treatment with cholesterol alone (Fig. 2B). For these studies, 2OH-CD-treated monolayers were washed free of 2OH-CD and incubated for an additional 90 min in the presence or absence of excess cholesterol or the cholesterol analogues 25-hydroxycholesterol or cholestane-3,5,6-triol. Monolayers depleted in cholesterol and subsequently allowed to recover in buffer alone exhibited strongly attenuated CT-induced peak Isc values equal to that of T84 monolayers treated continuously with 2OH-CD (Fig. 2B, column 3 versus column 2). In contrast, monolayers pretreated with 2OH-CD and then allowed to recover in the presence of cholesterol exhibited peak secretory responses equal to that of control monolayers not treated with 2OH-CD (Fig. 2B, column 6 versus column 1). Cholesterol-treated monolayers also exhibited recovery of the lag phase, consistent with the idea that depletion of membrane cholesterol affected a rate-limiting step required for toxin trafficking into the ER (lag phase = 31 ± 4 versus 48 ± 7 versus 37 ± 5 min, means ± S.E., n = 7-8, control versus cholesterol-depleted versus cholesterol-replete monolayers). Monolayers treated with the inactive cholesterol analogues cholestane-3,5,6-triol or 25-hydroxycholesterol exhibited only partial or no recovery (Fig. 2B, columns 5 and 4 versus control columns 1 and 6). Thus, the effect of 2OH-CD on CT-induced Cl secretion was fully reversible and specific to cholesterol. Nonspecific effects of 2OH-CD on G_M1 receptor density, if any, cannot explain these results. Taken together, these initial studies indicate that CT action on T84 cells, presumably because of toxin entry into the cell or trafficking retrograde into Golgi cisternae and ER, depends on membrane cholesterol.

View larger version (17K): [in this window] [in a new window]   Fig. 1.   Effect of 2OH-CD on CT-induced chloride secretion. A, chloride secretory response induced by CT in T84 monolayers treated or not treated with 2OH-CD. Monolayers pretreated for 1 h at 37 °C with 16.5 mM 2OH-CD (squares) or buffer (circles) were exposed to 20 nM CT at time = 0 min (open symbols). Toxin-induced chloride secretion was measured as a short circuit current (mean ± S.D. of duplicate monolayers, representative of n = 8 independent experiments). Monolayers not exposed to CT (closed symbols), either treated or not with 2OH-CD, were treated with VIP ~15 min prior to the end of the time course. B, effect of 2OH-CD on the lag phase of the chloride secretory response induced by apical 20 nM CT (means ± S.E., n = 8 independent experiments in duplicate, * indicates p < 0.05). C, effect of treatment with 2OH-CD on the maximal Isc induced by apically applied CT (shaded bar) or VIP (solid bar) (means ± S.E., n = 8, * indicates p  0.01). Cells were treated as described in panel A.

View larger version (14K): [in this window] [in a new window]   Fig. 2.   Effect of 2OH-CD on GM1 receptor density and reversibility of inhibition by treatment with exogenous cholesterol. Binding of CT to T84 monolayers treated (shaded bar) or not treated (solid bar) with 2OH-CD (means ± S.E., n = 8, p  0.05). B, cholesterol, but not cholestane-3,5,6-triol or 25-hydroxycholesterol, fully reversed the effect of 2OH--CD on CT-induced Cl secretion. T84 cell monolayers were treated for 1 h in the absence () or presence (+) of 2OH-CD at 37 °C for 1 h, washed, and then treated with either buffer alone (lanes 1 and 3) or 2OH-CD (lane 2) containing 20 nM CT for the remainder of the time course or with 0.2 mM amounts of the indicated oxysterol (25-OH chol, 25-hydroxycholesterol; cholestane, cholestane-3,5,6-triol). Results are normalized to the maximal Isc induced by VIP under identical conditions (means ± S.E., n = 4, * indicates p  0.05 compared against lanes 1 and 6).

To show that the effect of 2OH-CD on CT action was not explained by a general inhibitory effect of cholesterol depletion on membrane dynamics, we utilized anthrax EdTx toxin. Unlike CT, the enzymatic subunit of EdTx translocates across the limiting membrane of the acidic endosome to reach the cytosol, where it then acts inside the cell to increase intracellular cAMP (25). Anthrax EdTx, however, can only enter polarized T84 cells by binding basolateral membrane receptors. Because basolateral membranes appeared resistant to cholesterol depletion by 2OH-CD, we utilized the more potent cholesterol binding compound m-CD. One-hour incubations with 4 mM m-CD depleted ~55% of total cellular cholesterol with minimal effects on monolayer resistance (TER = 80% control values). Initial studies showed that, in contrast to 2OH-CD, m-CD inhibited by 70 and 30% the peak Isc induced by apically and basolaterally applied CT, respectively (Fig. 3D, solid bars (not treated) versus open bars (m-CD-treated)). The time course and lag phase for toxin-induced Isc were also affected (Fig. 3, A and C, solid bars (not treated) versus open bars (treated)). Inhibition of CT-induced Isc by m-CD was fully reversed by subsequent treatments with cholesterol, indicating specificity for cholesterol depletion (Fig. 3D, shaded bar). In contrast to CT, however, treatment of T84 monolayers with m-CD had no detectable effect on the cAMP-dependent Isc induced by anthrax EdTx or on the time course of EdTx action (Fig. 3, B and solid bars (not treated) versus open bars (m-CD-treated) in C and D). Thus, movement of EdTx from its binding site on the cell surface into the basolateral acidic endosome was not affected by m-CD. These data indicate that the inhibitory effects of 2OH-CD and m-CD on CT-induced Cl secretion cannot be the result of nonspecific effects on membrane dynamics in general.

View larger version (32K): [in this window] [in a new window]   Fig. 3.   Cholesterol depletion does not affect toxicity of anthrax edema toxin. A, m-CD inhibits CT-induced Isc. Representative time course shows the effect of treatment with m-CD (closed symbols) on Isc induced by 20 nM apical CT (squares) or 1 nM basolateral CT (circles) (mean ± S.D. in duplicate, n = 4 or more independent experiments). The cAMP agonist VIP was added to control monolayers 15 min before the end of the time course. The time course of VIP-induced Isc is offset for visual clarity. B, m-CD has no detectable effect on EdTx-induced Isc. Representative time course shows the effect of m-CD (closed symbols) on anthrax EdTx-induced chloride secretion. EdTx (12 nM) was added to the basolateral surface of some monolayers at t = 0 min (squares). VIP was added to control monolayers (triangles) 15 min before the end of the time course. C, effect of m-CD on lag phase of CT and EdTx. Panel shows m-CD-treated monolayers (open bars) and control monolayers (solid bars). Data are expressed as means ± S.E. (n = 6, p  0.05). D, m-CD inhibits the peak Isc induced by CT but not by anthrax EdTx. Panel summarizes the effect of m-CD on peak chloride secretion induced by apical and basolateral CT and by anthrax EdTx. Following the protocol outlined above, monolayers were pretreated for 1 h with 4 mM m-CD (open bars) or buffer alone (closed bars) and then exposed to 20 nM apical CT (n = 10), 1 nM basolateral CT (n = 8), or 12 nM basolateral EdTx (n = 6). In some experiments monolayers pretreated with m-CD were allowed to recover in the presence of cholesterol-m-CD complex before exposure to 20 nM apical CT (shaded bar). Results are expressed as mean fraction of peak Isc (mean maximal toxin-induced Isc/maximal VIP-induced Isc) ± S.E. (n = 4, * indicates p < 0.05).

Cholesterol depletion reduced the protein content of raft fractions isolated from T84 cells by 50% (Fig. 4B), but did not preferentially displace the CT-G_M1 from lipid raft microdomains. In some experiments, treatment with cyclodextrin displaced a fraction of CT into detergent-soluble membrane domains (Fig. 4F), but always the CT-G_M1 complex remained enriched in lipid rafts (Fig. 4 (A, C, and E) and Table I). Even in monolayers treated for 24 h with 2OH-CD, all of the detectable CT-G_M1 complex continued to fractionate with lipid rafts (Fig. 4A). The GPI-anchored 5'-nucleotidase (CD73) was also not displaced from lipid rafts in T84 monolayers as assessed by enzymatic assay (data not shown). Similar results were obtained in monolayers treated with 2OH-CD for only 2 h. These conditions were the same as those used for physiologic studies described above. Unlike CT, however, the function of CD73 that acts enzymatically at the cell surface was not affected by cholesterol depletion, as CD73 continued to convert 5'-AMP to adenosine at maximal rates (104 ± 4 versus 99 ± 1% of maximal 5'-AMP-induced Isc, control versus cholesterol-depleted, n = 2). Although both CT and CD73 were retained and enriched in raft fractions after cholesterol depletion, some proteins were not. Two other raft-associated proteins, caveolin-1 (an intrinsic membrane protein oriented toward the cell cytosol) and pp60^src (a cytoplasmic lipid-anchored kinase), were selectively solubilized (Fig. 4 (E and F) and Table I). Thus, two cytoplasmically oriented proteins were preferentially displaced from raft fractions by cholesterol depletion, but the extracellularly oriented CT-G_M1 complex and GPI-anchored CD73 remained enriched.

View larger version (38K): [in this window] [in a new window]   Fig. 4.   Effect of cholesterol depletion on the protein composition of lipid rafts. A, the CT-GM1 complex remains raft-associated following a 24-h treatment with 16.5 mM 2OH-CD. T84 cell monolayers were pretreated with 16.5 mM 2OH-CD (upper panel) or media alone (lower panel) for 24 h at 37 °C, after which they were brought to 4 °C, washed, and exposed to 20 nM CT. Monolayers were extracted in 1% Triton X-100 at 4 °C and layered under continuous sucrose gradients for equilibrium density centrifugation. Fractions were collected, and protein content and sucrose density were determined for each fraction. Fractions were analyzed by SDS-PAGE and Western blot loaded for equal protein (0.5 µg). *, 5 ng of CT loaded as control. B, protein content of raft fractions isolated from monolayers depleted (open bar) or not depleted (solid bar) in cholesterol. Results expressed as means ± S.E., n = 8. *, p  0.05. C, treatment with 2OH-CD decreases the apparent density of lipid rafts containing CT. T84 cell extracts were prepared from monolayers pretreated with 2OH-CD for 2 h. The extracts were layered under a 15-30% continuous gradient for equilibrium density centrifugation. Fractions (0.5 ml) were collected from the top of each gradient. Protein content and sucrose density of each fraction were determined. Equal mass (0.3 µg) of each gradient fraction was run on SDS-PAGE and analyzed by Western blot for the CT B-subunit. Western blot data for gradients from control (upper panel) and 2OH-CD-treated (lower panel) monolayers are plotted against percentage of sucrose of each fraction. All soluble fractions are in 40% sucrose. The data are representative of three experiments. D, optical density of gradient fractions shown in panel B. Absolute values of band densities between the two experiments are not standardized. E, treatment with 2OH-CD depletes pp60src and caveolin-1 from lipid raft fractions. Representative Western blots of raft (R) and soluble (S) fractions from step gradients of 2OH-CD-treated (right lanes) and control (left lanes) monolayers, loaded for equal protein (0.3 µg), and analyzed for the CT B-subunit, pp60src, and caveolin-1. F, Treatment with 2OH-CD shifts pp60src, caveolin-1, and a fraction of the CT B-subunit into soluble fractions. Representative Western blots of soluble (S) fractions from step gradients of 2OH-CD-treated (+) and control () monolayers. Each lane was loaded with 15 µg (50-fold excess protein compared with lanes shown in panel E), and analyzed for the CT B-subunit, pp60src, and caveolin-1.

View this table: [in this window] [in a new window]   Table I Relative enrichment for the CT B-subunit, pp60src, and caveolin in raft fractions isolated from cells depleted in cholesterol Western blots from Fig. 4E were analyzed by densitometry. Relative enrichment for the proteins per milligram of protein was determined as the ratio of the mass of protein (measured as optical density) in raft fractions isolated from cells depleted of cholesterol compared with the mass of protein in rafts isolated from cells not depleted of cholesterol (+2OH-CD/2OH-CD, n = 3). Fractionation of 5'-nucleotidase (5'-NT) with lipid rafts was assessed by enzyme assay (n = 2).

To demonstrate that cholesterol depletion affected the protein content of rafts containing the CT-G_M1 complex specifically, we examined raft density. Fig. 4 (C and D) shows that CT-containing lipid rafts isolated from monolayers depleted in cholesterol fractionated in a continuous gradient in sucrose at lower apparent densities (20-22 versus 22-25% sucrose, 2OH-CD-treated versus untreated, n = 2). Soluble fractions contained no detectable CT, even when analyzed at 100-fold excess total protein (data not shown). The effect of cholesterol depletion on raft structure was also readily apparent by visual inspection of sucrose gradients. Lipid rafts prepared from monolayers treated with 2OH-CD formed two distinct light-scattering bands on sucrose gradients with one band exhibiting lighter density relative to the single more diffuse band of lipid rafts obtained from control monolayers not depleted in cholesterol. Thus, cholesterol depletion affected the apparent density and presumably the protein/lipid ratio of isolated raft microdomains containing CT. Finally, to explain why depletion of membrane cholesterol caused inhibition of CT action, we examined whether the toxin entered the cell. Fig. 5A shows that, at all time points examined, control monolayers not depleted in cholesterol internalized more CT than monolayers treated with m-CD. This is best visualized after 30 min of continuous uptake where 8-fold less CT was detected in endosomes of monolayers depleted of cholesterol (Fig. 5B). To test the idea that retrograde trafficking of CT into Golgi cisternae (and presumably ER) was also affected, we utilized a toxin variant engineered to contain the sulfation consensus motif SAEDYEYPS (36, 37). Sulfation at this site can be measured after toxin entry into Golgi cisternae.² Fig. 5C shows that sulfation of CT-B-SAEDYEYPS was strongly attenuated in monolayers depleted in cholesterol by treatment with m-CD (17.85 ± 7.3% of control sulfation, n = 2). These data provide direct evidence that endocytosis or both endocytosis and trafficking of CT into Golgi cisternae depend on membrane cholesterol.

View larger version (20K): [in this window] [in a new window]   Fig. 5.   Effect of m-CD on endocytic uptake and Golgi trafficking of CT. A, treatment with m-CD inhibits endocytosis of CT. Monolayers were pretreated with m-CD or buffer (controls) for 1 h. CT (20 nM) was bound to control monolayers at 4 °C. Some monolayers were exposed continuously to 20 nM CT at 37 °C for the indicated times and then brought back to 4 °C. Surface-bound CT was removed by incubation at pH 2.5 (acid-stripped), and total cell-associated CT was analyzed by SDS-PAGE and Western blot as described under "Experimental Procedures." Total cell lysates were loaded for equal protein and Western-blotted either for the CT B-subunit or for actin as control for protein loading. B, data collected at the 30-min time point in panel A was analyzed by densitometry. Endocytosed toxin is expressed as fraction of toxin bound at 4 °C (n = 2 independent experiments). C, treatment with m-CD inhibits trafficking of CT to the Golgi. Monolayers were pretreated with buffer (control) or m-CD and loaded with Na235SO4. Cells were then continuously exposed to 20 nM CT containing the sulfation motif SAEDYEYPS for 50 min. CT-SAEDYEYPS was immunoprecipitated and analyzed for sulfation by SDS-PAGE gels and autoradiography (upper panel). The lower panel shows a Western blot of immunoprecipitated CT B-subunit loaded for equal volume as a "loading control" for the autoradiogram. Data are representative of two experiments.

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The results of these studies show that early steps in endocytosis and retrograde trafficking of CT into Golgi cisternae of human intestinal epithelial cells depend on membrane cholesterol. Cholesterol depletion of T84 cell monolayers affected the trafficking of CT specifically, as, under identical conditions, anthrax EdTx entered the basolateral endosome and induced toxicity with no detectable loss in potency. The defect in membrane trafficking of CT caused by cholesterol depletion correlated with a loss of toxin function and with a change in the density and specific protein content of CT-associated lipid raft fractions isolated from T84 cells. Cholesterol depletion displaced some CT into soluble fractions in some experiments, but left the bulk of the CT-G_M1 complex enriched in lipid raft microdomains. These data imply that cholesterol may function to couple the CT-G_M1 complex with other membrane components of the lipid raft required for CT entry into the cell.

Cholesterol functions critically in the biogenesis of eukaryotic cell membranes and is particularly enriched in lipid raft microdomains (38, reviewed in Ref. 39). Cholesterol dictates the physical state of the lipid bilayer. In model membrane systems, increasing amounts of cholesterol will induce formation of liquid ordered (lo) domains that exhibit detergent insolubility (40). Thus, the lo phase may describe the physical state of lipid rafts in vivo. Some membrane-associated proteins depend on cholesterol (and presumably the lo phase) for association with or displacement from raft domains, and this may regulate protein function (17, 29, 41-45). Cholesterol may also affect the activity of membrane-associated proteins by direct interaction or indirectly by affecting membrane fluidity (46-48).

In this study, we find that trafficking of the CT-G_M1 complex and toxin function exhibit dependence on membrane cholesterol. In some experiments, CT shifts partially into the soluble fractions, but the bulk of CT remains enriched in lipid rafts following cholesterol depletion. Abundant data indicate that sphingolipids such as G_M1, which typically exhibit long chain fully saturated sphingosine and fatty acid domains, self-assemble into lo microdomains in vitro (13, 14, 49). Such a tendency to self-aggregate may explain the strong association of the CT-G_M1 complex with lipid rafts in native cell membranes and explain why the CT-G_M1 complex remains enriched in lipid raft domains in T84 cells after cholesterol depletion. The GPI-anchored CD73 was also not displaced from lipid raft fractions after cholesterol depletion, suggesting that CD73 in T84 cells may be anchored to the membrane by a similarly long and saturated acyl domain. Alternatively, it is possible that CT may induce its own unique form of lo domain by clustering G_M1 via binding by the pentameric CT B-subunit. Clustering of G_M1 could stabilize the assembly or induce the coalescence of raft domains carrying CT just as cross-linking of the GPI-anchored folate receptor in 3T3 fibroblasts may stabilize the association between this protein and caveolae (50). Further support for this idea can be taken from studies on Shiga toxin. Shiga toxin also exhibits a pentameric subunit that binds multivalently to glycolipid receptors for retrograde trafficking into the ER and the induction of toxicity (51). Additionally, such clustering of G_M1 by binding to the B-subunit of CT (or clustering of glycolipid Gb3 by binding the B-subunit of Shiga toxin) may induce the formation of a lipid rafts with unique structure and function. Available evidence indicates that lipid rafts isolated from a single cell type exhibit heterogeneity in structure and function (52), and we have recently found that the CT-G_M1 complex fractionates with a discrete craft domain in T84 cells (5).

Even though cholesterol depletion did not displace the bulk of the CT-G_M1 complex from raft domains in T84 cells, two other cytoplasmically oriented raft-associated proteins were depleted. These results correlated with a loss of toxin function and implied to us that cholesterol may be required for stable coupling of the CT-G_M1 complex with raft domains or with other components of the lipid raft required for toxin entry into the cell, or both. Thus, we hypothesize that CT, which is anchored only to the extra-cytoplasmic surface of the plasma membrane by binding G_M1, must interact with cytoplasmically oriented membrane-associated proteins or lipids that form or communicate with the machinery for endocytosis, or sorting into Golgi, or both. Our data indicate that such interactions depend on the structure and function of lipid rafts. This idea was strengthened by two observations. First, the total protein content and the relative enrichments of caveolin-1 and pp60^src in lipid raft fractions isolated from T84 cells were altered by cholesterol depletion, and, second, the apparent density or protein/lipid ratio of CT-associated rafts was affected specifically.

The block in CT function caused by cholesterol depletion can be explained by inhibition of endocytosis and trafficking of the CT-G_M1complex into the Golgi apparatus. Our results are consistent with those previously reported by Orlandi and Fishman (17). In the human intestinal Caco2 cell line, sequestration of membrane cholesterol by chelation with filipin or cyclodextrin inhibited endocytosis of CT and blocked toxin function. Cholesterol depletion of Caco2 cells, however, did not affect the activity of diptheria toxin that likely enters host cells by clathrin-dependent mechanisms (17, 53). In the current study, we also find that treatment of T84 cells with cyclodextrins inhibited endocytosis and function of CT, but had no detectable affect on the action of anthrax EdTx that does not bind receptors localized in lipid rafts.

Based on these data, however, we cannot identify the exact mechanism of endocytosis for CT in either cell type. Although we find a significant effect of cholesterol depletion on endocytosis in T84 cells, both clathrin-mediated and non-clathrin-mediated mechanisms of endocytosis have been shown to depend on membrane cholesterol (6, 19, 20, 54), and, recently, CT was shown to enter COS-7 cells by both clathrin-dependent and clathrin-independent mechanisms (55). In hippocampal neurons, CT may enter the cell in clathrin-coated pits while still bound to lipid raft microdomains (18). Our data do show, however, that inhibition of toxin endocytosis and trafficking cannot be explained by a general nonspecific effect of cholesterol depletion on membrane dynamics. Anthrax EdTx requires entry into the basolateral acidic endosome to induce toxicity, and this was not affected by depletion of membrane cholesterol. Anthrax EdTx represents an almost perfect control for our studies on trafficking of CT, as both toxins induce a Cl secretory response in T84 cells. Thus, nonspecific effects of the cyclodextrins on T84 cell function would affect equally the Cl secretory response induced by both toxins. Finally, that cholesterol depletion had no effect on the action of EdTx, which does not bind receptors located in lipid rafts, provides indirect evidence that CT may traffic retrograde into the biosynthetic pathway of T84 cells by lipid raft-dependent mechanisms.

In support of this idea, we find direct evidence that cholesterol depletion inhibits transport of CT into Golgi cisternae. Multiple trafficking steps en route from plasma membrane to Golgi cisternae, including the initial endocytosis of the CT-G_M1 complex, are likely affected. Our data fit closely with the results of recent studies showing that retrograde trafficking of Shiga toxin into the Golgi apparatus of HeLa cells also depends on association with functional lipid rafts (51). Such retrograde transport into Golgi cisternae and then ER is required for both CT and Shiga toxin function. That lipid rafts may function in protein sorting has been reported in the literature for many years (56). For example, sorting of apical proteins (15, 57), endocytosis (19), endosome to Golgi trafficking (10, 58, 59), and even Golgi structure (57) have all been shown to depend on membrane cholesterol and presumably on lipid raft structure and function as discussed above.

In summary, our results show that trafficking of the CT-G_M1 complex into the Golgi apparatus of polarized intestinal epithelial cells, and thus the induction of toxicity, depends on membrane cholesterol. We propose that cholesterol may function to couple the ganglioside G_M1-CT complex with raft microdomains and with other membrane components required for CT entry into the cell. This likely depends on lipid raft structure and function.

	ACKNOWLEDGEMENTS

We thank Margaret Ferguson Maltzaman and Heidi Wheeler for expert assistance with tissue culture, Dr. Linda Thompson for performing enzymatic assays of 5'-nucleotidase, Dr. John Collier for the gift of anthrax edema toxin, and members of the Lencer laboratory for critical reading of this manuscript.

	FOOTNOTES

^* This work was supported by National Institutes of Health Grant DK48106 (to W. I. L.); by National Research Service Award F32 DK09863, a Charles A. Janeway award from Children's Hospital, and a research scholar award from the American Digestive Health Association (all to A. A. W.); and National Institutes of Health Grant DK34854 (to the Harvard Digestive Diseases Center).The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

^¶ To whom correspondence should be addressed: Gastrointestinal Cell Biology, Children's Hospital, 300 Longwood Ave., Boston, MA 02115. Tel.: 617-355-8599; Fax: 617-264-2876.

Published, JBC Papers in Press, February 21, 2002, DOI 10.1074/jbc.M109834200

² Y. Fujinaga and W. I. Lencer, unpublished data.

	ABBREVIATIONS

The abbreviations used are: CT, cholera toxin; G_M1, ganglioside Gal1-3GalNAc1-4(NeuAc2-3)Gal1-4Glc1 ceramide; G_D1a, ganglioside NeuAc2-3Gal1-3GalNAc1-44Glc1 ceramide; 2OH-CD, 2-hydroxypropyl -cyclodextrin; m-CD, methyl -cyclodextrin; EdTx, anthrax edema toxin; VIP, vasoactive intestinal peptide; Isc, short circuit current; HBSS, Hanks' balanced salt solution; TER, transepithelial resistance; ER, endoplasmic reticulum; HRP, horseradish peroxidase; GPI, glycosylphosphatidylinositol.

	REFERENCES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES