Interaction of Cholera Toxin With Rat Intestinal Brush Border Membranes RELATIVE ROLES OF GANGLIOSIDES AND GALACTOPROTEINS AS TOXIN RECEPTORS*

Rat intestinal brush borders contained a small number (3-4 pmol/mg of protein) of high &nity (10"' M) binding sites for choleragen. Following extraction with chloroform/methanol solutions, the delipidated membranes lost 99% of their capacity to bind '2SI-choleragen but retained their ability to bind '251-Ricinus communis agglutinin 1 (RCA 1). When analyzed by thin layer chromatography, the major gangliosides in the lipid extract migrated differently than galactosyl-N-acetyl-galactosaminyl-[N-acetylneuraminyl]-galactosylgluco- sylceramide (GI) and were neuraminidase-sensitive. Using a sensitive assay, however, small amounts of neuraminidase-resistant gangliosides with mobilities similar to were detected. These latter gangliosides were labeled with 'H when isolated from membranes treated with galactose oxidase and NaB'H4. Incorpo- ration of 'H into these gangliosides was reduced, how-ever, when the membranes were first exposed to cho- leragen. In contrast, choleragen did not protect membrane galactoproteins from being labeled. When membranes containing

Using a sensitive assay, however, small amounts of neuraminidase-resistant gangliosides with mobilities similar to were detected. These latter gangliosides were labeled with 'H when isolated from membranes treated with galactose oxidase and NaB' H4. Incorporation of 'H into these gangliosides was reduced, however, when the membranes were first exposed to choleragen. In contrast, choleragen did not protect membrane galactoproteins from being labeled. When membranes containing bound '2SI-choleragen were treated with neutral detergents, over 85% of the bound toxin was extracted. When the solubilized membranes were analyzed by sucrose density gradient centrifugation, the labeled toxin sedimented to the same region of the gradient as did choleragen and choleragen-GMl complexes and no higher molecular weight forms indicative of toxin-galactoprotein complexes were detected. Membranes labeled by the galactose oxidase/NaB'& method were incubated with choleragen, solubilized as above, and treated with antitoxin antibodies; the immune complexes were absorbed out with fixed Staphylococcus aureus and analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis. Less than 1% of the 'H associated with the membranes was immunoprecipitated, some of which corresponded to galactoproteins. Electrophoretograms of brush border membranes were analyzed directly, or after electrophoretic transfer of the proteins to nitrocellulose sheets, for iodinated RCA I or choleragen binding. The galactoproteins were readily detected with RCA I but not with choleragen. Choleragen, however, did bind to the lipid- containing region of the electrophoretograms. We conclude that the predominant choleragen receptors in rat intestinal brush borders are gangliosides with the characteristics of G M~.
Choleragen, an enterotoxin produced by Vibrio cholerae, is responsible for the clinical manifestations of the disease cholera (1). The toxin binds to specific receptors on the luminal surface of the intestinal mucosal cell and activates adenylate cyclase (2-5). Choleragen can activate adenylate cyclase in other veterbrate cells and its mechanism of action has for the most part been elucidated (6, 7 ) . The ganglioside G M~' has been implicated as the receptor for the toxin (7,(9)(10)(11)(12)(13). The ability of G M I to be taken up by GMl-deficient cells and thus sensitize the cells to choleragen has clearly demonstrated that GMI can function as a toxin receptor (7,(14)(15)(16). Less is known about the role of G M~ as the receptor in intestinal mucosal cells. Gangliosides isolated from intestinal mucosa blocked the action of choleragen (10). A correlation between GMI content and toxin binding has been demonstrated in the small intestine of several species including man, and prior treatment of rabbit intestine with G M~ increased the amount of toxin binding as well as the diarrheogenic action of the toxin (17). Recently, Morita et al. reported the presence of glycoproteins in rat intestinal membranes that bind choleragen (18). We thus decided to re-examine the relative roles of G M~ and glycoproteins as receptors for choleragen in rat intestinal brush border membranes.

EXPERIMENTAL PROCEDURES
Materials-Choleragen was obtained from Schwarz/Mann and iodinated as described (11). RCA I ' (castor bean lectin-120) was from P-L Biochemicals Inc. and was iodinated by the chloramine-T procedure in the presence of 0.2 M D-Gal to protect the binding sites; the specific radioactivity was 1.7 mCi/mg. Na'? and NaB3H4 were obtained from Amersham Corp. Galactose oxidase was from Worthington Biochemical Co. V. cholerae neuraminidase and fiied Staphylococcus aureus (Pansorbin) were from Calbiochem-Behring Corp. Protease (Pronase Type VI) was obtained from Sigma. G M~ was labeled with 3H as described previously (14). Anti-choleragen antiserum raised in a burro was a generous gift of Dr. William Habig, Isolation and Analysis of Glycosphingolipids-Brush borders (up to 10 mg of protein) were extracted with 5 ml of chloroform:methanol (12, v/v) at 37 "C for 1 h after centrifuging, the insoluble residue was extracted with 2.5 ml of the same solvent at 45 "C for 1 h, sedimented by centrifugation, and washed with 1 ml of the same solvent. The combined extracts were analyzed directly for choleragen binding components as described below or further separated (21). Briefly, the extracts were taken to dryness, desalted on Sephadex G-25, and separated into neutral and acidic fractions on DEAE-Sephadex. The neutral fraction was treated with 0.2 M NaOH in chloroform:methanol (2:1, v/v) to hydrolyze phospholipids; after neutralizing with acetic acid, the solution was partitioned and the resultant lower phase was washed once with the theoretical upper phase (22). The washed lower phase was separated by thin layer chromatography on silica gel with the solvent system chloroform:methanol:water (65254, v/v). The glycosphingolipids were detected by spraying the plates with a solution of 0.5% orcinol in ethanol: concentrated H2S04(8:2, v/v) and heating at 110 "C for 10 min. Gangliosides were isolated from the acidic fraction following alkaline hydrolysis, desalting on Sephadex G-25, and column chromatography on UNsil (21). The gangliosides were separated by thin layer chromatography, visualized with resorcinol reagent, and quantified by scanning densitometry (21).
The total lipid extract and the purified ganglioside fraction were analyzed for choleragen binding components as described elsewhere (23). Briefly, portions were spotted along with standards on plasticbacked silica gel chromatography sheets (100 pm thick from Eastman Kodak Co.). The sheets then were clamped in a sandwich chamber and developed with chloroform:methanol:2.5 M NKOH (W409, v/ v). The chromatograms were dried and overlayed with a minimal amount of ice-cold phosphate-buffered saline containing 1% polyvinylpyrrolidone and 'Z51-choleragen (2 X IO6 cpm/ml). After 2 h at 4 "C, the chromatograms were washed 4 times in excess cold phosphatebuffered saline, air-dried and exposed to x-ray film for 1-2 days.
Labeling of Brush Borders by Galactose Oxidase and NaB3H4 Treatment-Brush borders (1.5 mg of protein) were incubated with 20 units of galactose oxidase in 0.5 ml of 10 mM Tris-C1 (pH 7.4), 5 mM EDTA, and 1 mM phenylmethylsulfonylfluoride for 30 min at 37 "C. In some experiments, the membranes were incubated first with 10 pg of choleragen at 4 "C for 30 min. Then 1 mCi of NaB3H4 (7.8 Ci/mmol) was added. After 15 min at 4 "C, the membranes were washed 3 times by diluting with 10 ml of ice-cold PBS and centrifuging at 40,000 X g for 30 min. In order to more effectively label the ganglioside fraction, 10 mg of brush borders were incubated with 100 units of galactose oxidase in 2 ml of 20 m sodium phosphate (pH 7.1), 137 mM NaCl, 0.5 m~ CaC12, 0.5 m MgC12, and 1 mM phenylsulfonyl fluoride with and without prior treatment with 20 pg of choleragen. After 3 h at 37 "C, 10 ml of ice-cold PBS was added and the samples were centrifuged at 40,000 X g for 30 min. Each pellet was suspended in 1.9 ml of PBS to which 10 mCi of NaB3H4 in 0.1 ml of 10 mM NaOH was added. After 1 h at 4 "C, the membranes were washed 3 times as described above.
Solubilization of Choleragen Receptor Complexes-In order to test the effectiveness of various detergents in solubilizing bound choleragen, brush borders were incubated with Iz5I-choleragen as described above and centrifuged at lo5 X g for 5 min in a Beckman Airfuge. The pellets were suspended in 0.1 ml of PBS containing 1% detergent and incubated at 4 "C for 2 h with occasional mixing. After adding 0.1 ml of PBS, the samples were centrifuged at lo5 X g for 5 min and the percent solubilization was determined. NP-40 (86.9%) and Triton X-100 (84.4%) were the most effective, followed by Lubrol PX (71.9%) and octylglucoside (46.1%). The fvst two detergents were used for all further experiments. The detergent-extracted components were subjected to immunoabsorption as previously described (24) or analyzed on linear sucrose density gradients.
Immunoabsorption of Gangliosides from Labeled Brush Borders-Gangliosides isolated from brush borders labeled by galactose oxidase and NaB3H4 were analyzed further by immunoprecipitation. Gangliosides from 1 mg of membrane protein were dissolved in 0.1 ml of 0.1% Triton X-100 in 25 m Tris-C1 (pH 7.4), 135 mM NaC1, 1 mM EDTA, and 3 m NaN3. Then 1 p1 of 10% bovine serum albumin and 2.5 pl of choleragen (1 mg/ml) were added to each sample. After 30 min at 4 "C, 20 pl of anti-choleragen antibodies were added. After 18 h at 4 "C, the immune complexes were absorbed by adding 0.1 ml containing 20 mg of futed S. aureus that had been washed with 0.1% Triton X-100 in PBS. After 18 h at 4 "C, the samples were centrifuged at 10, OOO X g for 2 min in a Beckman microfuge. The pellets were washed twice with 1 ml of the above solution and once with PBS. The pellets were extracted with chloroform: methanol (1:2, v/v). After being desalted on Sephadex G-25 (21), the labeled gangliosides were separated by thin layer chromatography and detected by radioscanning the thin layer chromatograms (15). Recovery of a known amount of [%]GMI (40 pmol) carried through the above immunoadsorption procedure was 86.5%.
Other Methods-Protein was determined by the method of Lowry et al. (25) with bovine serum albumin as the standard. SDS-polyacrylamide gel electrophoresis was performed according to Laemmli (26) using a slab gel apparatus (Bio-Rad Laboratories Model 220). Gels were stained directly with iodinated choleragen or RCA I according to Burridge (27) or after electophoretic transfer of proteins from the gel to nitrocellulose sheets (28). The gels or sheets were analyzed for radioactivity either by autoradiography or by slicing and counting 1mm sections. Proteins were detected by staining the gels or sheets with Coomassie blue. Tritium was detected on gels by fluorography (29).

Binding of Choleragen to Brush Border Membranes-
Binding of 1251-choleragen to rat intestinal brush borders appeared to be to a single class of high affinity sites (  (Fig. 2 A ) . Above this concentration, binding was partially inhibited, a maximal effect occurring at 1 PM; at higher concentrations of RCA I, however, there was less inhibiti~n.~ In contrast, choleragen at concentrations up to 1.2 Because choleragen is a multivalent ligand and the toxin-receptor complex does not readily dissociate (7, 11, 20), Scatchard analysis of the binding data is not appropriate (20). The smooth shapes of the binding curves shown in Fig. 1 are consistent with a single class of binding sites.
Walker et al. reported that rat intestinal microvillous membranes bound 10 fmol of choleragen per mg of protein and half-saturation occurred at 0.2 PM (3). We recalculated their data, and arrived at a lowest estimate of 560 fmol/mg of protein and 0.2 r w for halfsaturation.
Gahmberg and Hakomori also observed that RCA I has heterogenous effects on the exposure of cell surface glycolipids and the effects of RCA I were different for different glycolipids (30). Thus, the lectin by binding to galactoproteins may mask the accessibility of GMI to choleragen but at higher concentrations may cause aggregation of the galactoproteins and thus increase the exposure of G M~ to choleragen.
It has also been shown that RCA I bound to GM' incorporated into artificial lipid vesicles (31). Thus, RCA I either indirectly or directly could be inhibiting the binding of choleragen to GM,. p~ had no effect on RCA I binding, whereas 50% inhibition of '251-RCA I binding was observed with 0.1 p~ of unlabeled RCA I (Fig. 2B).
Treatment of the membranes with Vibrio cholerae neuraminidase enhanced choleragen binding by over 2-fold and had a small effect on RCA I binding (Table I). Protease treatment increased choleragen binding by 27% and reduced RCA I binding by 63%. The membranes then were extracted with mixtures of chloroform and methanol and the delipidated residue was assayed for binding (Table I). Whereas choleragen binding was reduced by 99%, there was no significant effect on RCA I binding. These results suggested that the predominant receptors for choleragen were lipids and those for RCA I were proteins.
Analysis of Lipid Extracts of Intestinal Brush Borders-The lipid extract was separated into neutral and acidic lipids and gangliosides were isolated from the latter fraction as described under "Experimental Procedures." The predomi-  containing 154 mM NaCl and 6.8 m~ CaCI2 for 1 h at 37 "C. Portions were then assayed for specific 1251-choleragen and "'I-RCA I binding as described under "Experimental Procedures" except the assays were done at 4 "C. In addition, membranes were delipidated as described under "Experimental Procedures," suspended in H 2 0 with sonication, and assayed for binding.

Treatment
Specifically bound as % of control nant neutral glycolipids migrated similarly to GL-1 and GL-3 on thin layer chromatography (Fig. 3A). Others had reported that GL-la and GL-3 were the major neutral glycosphingolipids in rat intestinal epithelium (32,33). There were several minor components that were less mobile than GL-4 as described previously (32,33). The purified ganglioside fraction contained several closely migrating components with mobilities between G M~ and G M~ and a minor component which migrated between GMI and Gols on the chromatogram (Fig.  3B). In terms of lipid-bound sialic acid, there were 36 f 1.6 and 2.4 f 0.5 (n = 3) nmol/mg of protein of the major and minor gangliosides, respectively. Following treatment with neuraminidase, 95% of the gangliosides were hydrolyzed and several new resorcinol-negative bands with more rapid mobilities were observed (Fig. 3B). Further analysis of the hydrolysis products indicated that three neutral glycolipids which migrated between GL-2 and GL-3 were produced (Fig. 3C). These results are consistent with a previous study in which the predominant mucosal ganglioside was neuraminidase-sensitive and contained NeuNGc (32; see also Ref. 33).
As delipidation effectively removed choleragen receptor activity from the intestinal brush borders, we next directly assayed for choleragen binding components in the lipid extract using a sensitive method developed by Magnani et al. (23). The lipid extract was separated on silica gel-coated plastic sheets which were then overlayed with '251-choleragen; after washing, bound choleragen was detected by autoradiography (Fig. 4). The predominant toxin binding component had a mobility identical with G M~ (Fig. 4A); there was no choleragen bound at the origin as would be expected if a choleragen binding hydrophobic protein had been extracted from the membranes. Choleragen binding appeared to be specific as it was blocked by excess unlabeled toxin (Fig. 4A ). The purified ganglioside fraction also contained a choleragen binding component which migrated as G M~ and was resistant to neuraminidase (Fig. 4B). As we were unable to successfully separate G M~ containing NeuAc from that containing NeuNGc on these chromatograms, we cannot state which form of GMI is present in the membranes.
Although the amounts of GMl in the brush borders were too small to quantitate by chemical means, we were able to estimate their amounts. Previous studies had shown that G M~deficient rat glial C6 cells took up G M~ from the culture medium in proportion to its concentration and bound 1251choleragen in proportion to the amount of GMI taken up (34). C6 cells were suspended in medium containing increasing amounts of G M~ or brush border gangliosides, incubated for 90 min at 37 "C, washed extensively, and assayed for choleragen binding. Using this sensitive assay, we estimated that rat  . (18), a number of glycoproteins were labeled by exposing the brush borders to galactose oxidase and NaB"H4 (Fig. 5b); there was very little incorporation of 'H in the absence of galactose oxidase (Fig. 5a). The labeled glycoproteins appeared to be those that bound Iz5I-RCA I (Fig. 5e); as expected, 0.2 M Gal totally inhibited RCA I binding to these galactoproteins (not shown). Prior exposure of the brush borders to a saturating amount of choleragen, however, did not appear to have any effect on the labeling of these glycoproteins (Fig. 5c). The labeled brush borders were extracted and analyzed for labeled glycosphingolipids. Using the mild conditions of labeling (see "Experimental Procedures"), the ganglioside fraction was not effectively labeled. Using the more exhaustive conditions, substantial amounts of 3H were incorporated into the ganglioside fraction (Fig. 6). In the absence of galactose oxidase, some label was incorporated into the major gangliosides of the brush borders (Fig. 6A). These gangliosides also were labeled when the membranes were exposed to galactose oxidase and the addition of choleragen had no effect on their labeling (Fig. 6A). In contrast, gangliosides with mobilities similar to G M~ were labeled only with enzyme treatment and labeling was substantially reduced by prior treatment of the membranes with choleragen. To further clarify the nature of these labeled gangliosides, portions were dissolved in 0.1% Trition X-100, incubated with choleragen, and subjected to immunoadsorption. The predominant labeled ganglioside isolated by this procedure corresponded to G M~ and its labeling was effectively blocked by choleragen (Fig. 6B). The major neutral glycosphingolipids were labeled by the galactose oxidase/NaB3H4 procedure and prior exposure of the brush borders to choleragen had no effect on their labeling (data not shown).
Analysis of Solubilized Choleragen-Receptor Complexes on Sucrose Density Gradients-As indicated under "Experimental Procedures,'' Triton X-100 and NP-40 effectively extracted '251-choleragen bound to intestinal brush borders. When centrifuged on a linear 5-4055 sucrose gradient, the detergent-extracted radioactivity sedimented in a symmetrical peak slightly ahead of '251-choleragen (data not shown). When brush borders containing bound iodotoxin were extracted with Up to 50% of the choleragen bound to cultured cells including Triton X-100 and NP-40 depending on the extraction conditions.

GMl-deficient cells treated with Grm is resistant to extraction by
Some of the GMI taken up by GMl-deficient cells also is resistant to detergent and prior treatment of the cells with choleragen increases this amount of Ghll (J. Hagmann and P. H. Fishman, unpublished detergent and applied directly to the gradient without an initial centrifugation to remove insoluble material, some of the label also sedimented through a 60% sucrose cushion to the bottom of the tube (Fig. 7 A ) : Most of the label sedimented slightly faster than '251-choleragen and the same as iodotoxin incubated with Gm in detergent. Results similar to those observed in Fig. 7 A were obtained whether the brush borders were incubated first with 2,20 or 50 nM '251-choleragen or whether the detergent extracts were sedimented on 5-20, 5-30 or 5-40% linear sucrose gradients. There was no evidence of soluble choleragen-receptor complexes greater than 1 0 0 , daltons. We were able to confirm the stability of toxin-receptor complexes as follows. Rat glial C6 cells were incubated with [3H ]GMl (16,34), then with and without choleragen and then solubilized with Trition X-100. After centrifuging at 500 X g for 10 min,6 the extracts were sedimented on linear sucrose gradients (Fig. 7 B ) . In extracts from cells not exposed to toxin, all of the remained at the top of the gradient. Prior addition of choleragen to the cells caused the [3H]G~1 to sediment in the same position as detergent-treated iodotoxin-G M~ complexes. Similar results were obtained when the [3H] GMl-treated cells were exposed to the B component of choleragen or when the detergent extract of the cells was incubated with choleagen prior to centrifugation on the gradient (data not shown).
Immunoadsolption of Brush Border Components-Putative choleragen receptors were isolated from galactose oxidase/NaB3H4-labeled brush borders by immunoadsorption (24). The labeled membranes were extracted with NP-40 and choleragen was added to the detergent extracts followed by anticholeragen antibodies. The immune complexes then were absorbed to fured S. aureus, washed extensively with detergent, solubilized in SDS and analyzed by SDS-polyacrylamide gel electrophoresis (Fig. 8). The major galactoproteins of the were incubated with 20 nM 1Z51-choleragen in 0.2 ml for 30 min at 4 "C, centrifuged, and extracted with 1% Triton X-100 as described under "Experimental Procedures." Binding and extraction were done in the presence of 1 m~ phenylmethylsulfonyl fluoride. The detergent mixture (118,000 cpm) was layered on top of a 5-ml 5-3056 sucrose gradient with a 60% sucrose cushion. All sucrose solutions were in 0.1% Triton X-100. After centrifuging for 18 h at 114,000 X g(av), fractions were collected from the bottom of the tube and counted for Iz5I. The small pellet at the bottom of the tube contained 13,600 cpm (11.5% of that applied to the gradient). B, rat glial C6 cells were incubated with 0.5 p~ [ 3 H ] G~~ for 1 h at 37 "C (34), washed, incubated with (0) and without (0) 50 n~ choleragen for 30 min at 4 "C, washed, and scraped from the culture dishes. The cells were extracted with 1% Triton X-100 as described under "Experimental Procedures." After centrifuging at 500 X g for 10 min, the extracts were layered on 5-40% sucrose gradients which were centrifuged as above for 16 h. brush border were effectively extracted by NP-40 (compare Fig. 8a with 86). Galactoproteins of similar molecular weight to the major galactoproteins were adsorbed in the presence of choleragen (Fig. sd) although there was some nonspecific adsorption (Fig. 8c). Only 0.24% of the radioactivity in the NP-40 extracts was specifically absorbed. Recoveries of choleragen-receptor complexes by this method were extremely high as shown by the fact that [3H]G~l added to the detergent extracts was quantitatively and specifically adsorbed to S. aureus in the presence of toxin and antitoxin. Similar results were obtained when brush borders were incubated with choleragen and washed prior to detergent extraction or when no [3H]G~l was added to the extracts.
Detection of Choleragen and RCA I Receptors on SDS-Polyacrylamide Gel Electrophoretograms-Intestinal brush borders were solubilized in SDS and subjected to SDS-polyacrylamide gel electrophoresis. When the gels were incubated with 1251-choleragen according to Burridge (27), tbe labeled toxin only bound to that region of the gel where the lipids migrated (Fig. 9). Binding was specific as it was blocked by excess unlabeled toxin. Choleragen did not bind to other regions of the gel where the glycoproteins migrated. Identical results were obtained in several additional experiments including ones where the proteins were electrophoretically transferred to nitrocellulose sheets (28). With this latter technique, there was no choleragen binding at all as the lipids were not effectively transferred to the sheets (inset, Fig. 9). The proteins and galactoproteins were effectively transferred as shown by the staining with Coomassie blue and the specific binding of '251-RCA I.
In one experiment with the direct Burridge procedure, we did observe binding of '251-choleragen to the major glycoproteins as well as to all other proteins including the molecular weight markers. The labeled choleragen preparation used for this experiment had a large proportion of radioactivity that did not bind to membranes and we suspect that this material nonspecifically absorbed to the various proteins on the gel.

DISCUSSION
Our results clearly demonstrate that rat intestinal brush borders contain a small number of high affinity binding sites for choleragen. These sites are resistant to proteases and neuraminidase but are extracted from the membranes with organic solvents. In contrast, the brush borders contain a large number of lower affinity binding sites for RCA I which are protease-sensitive and not extracted with organic solvents. Although we were unable to chemically detect G M~ in the lipid extracts, we were able to detect choleragen binding components in the lipid extracts that had the same mobility as G M~ by thin layer chromatography. Using this very sensitive technique developed by Magnani et al. (23), we were able to show Receptors for Cholera that these components were resistent to neuraminidase treatment which is consistent with a GMl-type structure. We also have provided indirect evidence that the brush borders contain sufficient G M~ to account for a l l of their choleragen binding capacity. As NeuNGc is the predominant species of lipid-bound sialic acid (32, 33), this form of hl may be present; previous studies have demonstrated that choleragen can use G M~ containing NeuNGc as a receptor (34).
When brush borders were labeled by the galactose oxidase/ NaB3H4 procedure, a labeled ganglioside that corresponded to GMI was isolated from the labeled membranes. Incorporation of 3H into this ganglioside was substantially reduced when the membranes were exposed to choleragen prior to being labeled. The ability of choleragen to specifically protect GMI from galactose oxidase has been shown previously for thyroid membranes (35), cultured cells (36), and liposomes (37).
Using a variety of techniques, we were unable to convincingly demonstrate a direct interaction between choleragen and rat intestinal brush border glycoproteins. Although we used similar procedures, our results do not in general c o n f m a previous report by Morita et al. (18). In their studies, they reported the following.
(i) Glycoproteins from rat intestinal microvillus membranes could be separated from glycolipids by affinity chromatography on RCA-agararose and could bind choleragen as detected by gel filtration chromatography. Our studies indicated that choleragen did not block RCA I binding to rat intestinal brush borders and that inhibition of choleragen binding by RCA I was of low affinity and incomplete. In addition, detergentsolubilized choleragen-receptor complexes were separated according to molecular size on sucrose density gradients; there was no evidence of complexes greater than 100,000 daltons. As the smallest glycoprotein detected by Morita et al. had a molecular weight of 69,000 (18), the smallest putative toxinglycoprotein complex would have a molecular weight of over 150,OOO.
(ii) Using the Bunidge technique, intestinal glycoproteins separated by SDS-polyacrylamide gel electrophoresis bound 'z51-choleragen and binding was blocked by 200 pg of unlabeled toxin (18). In several of our experiments, we observed that the iodotoxin only bound to the dye front region of the gels where the glycolipids migrate. In one experiment, we did detect choleragen bound to membrane glycoproteins as well as other proteins, including the molecular weight markers. This unusual binding also was blocked by a large excess of unlabeled choleragen. We did find, however, that "'I-RCA I bound in a specific way to many of the brush border glycoproteins either directly on the gels or after transfer to cellulose nitrate sheets. None of the galactoproteins bound 1251-choleragen by this latter technique.
(iii) When membranes labeled by galactose oxidase/NaB3H4 were detergent-extracted and subjected to immunoadsorption by anti-choleragen antibodies and S. aureus, at least 5 glycoproteins with molecular weights from 69,000 to 132,000 corresponding to the major galactoproteins were detected (18). We also observed several glycoproteins by this same procedure but they represented less than 0.25% of the total labeled membrane components. Prior incubation of the membranes with excess choleragen, however, did not protect the glycoproteins from being labeled by galactose oxidase/NaB31-b. The possibility that a small proportion of the total membrane galactoproteins are associated with Ghll and therefore would appear to have been specifically immunoadsorbed must be considered (24).
One unusual feature of the studies by Morita et al. (18) is that most of the major glycoproteins of the intestinal membranes that are labeled by galactose oxidase/NaB31& appear Toxin in Rat Intestine to bind 1251-RCA I and '251-choleragen. In addition, their results suggest that most of the choleragen binding is to these glycoproteins and not to the glycolipids although it is unfortunate that their results were strictly qualitative and no quantitative data were presented. We believe that the strongest evidence in support of glycolipids being the predominant (if not only) receptors for choleragen is the fact that 99% of the toxin binding activity is removed from the membranes by delipidati~n.~ Although it is possible that this procedure may denature putative choleragen binding glycoproteins, one would anticipate that the binding determinants would reside in the carbohydrate portion of the molecules. In this regard, delipidation had no deleterious effects on the ability of these carbohydrate sequences to bind RCA I.
Although we used a different strain of rat and a different procedure for isolating intestinal membranes, we do not believe that these minor differences can account for the major differences between our results and those of Morita et al. (18). Our results clearly indicate that rat intestinal mucosal membranes have receptors for choleragen with the characteristics of GMl and are thus consistent with numerous other studies demonstating that G M l is the receptor for choleragen.