Proc. Nat. Acad. Sci. USA Vol. 72, No. 7, pp. 2520-2524, July 1975 Biochemistry

Interaction of toxin and membrane GM1 of small intestine (mucosal receptors/[3HJGMl ganglioside/ Vibrio cholerae sialidase/) JAN HOLMGREN*, IVAR LONNROTH*, JAN-ERIC MANSSONt, AND LARS SVENNERHOLMt * Institute of Medical Microbiology and t Department of Neurochemistry, University of G~teborg, Goteborg, Sweden Communicated by Saul Roseman, March 31, 1975

ABSTRACT Ganglioside GM, was isolated from the ther evidence for the function of small intestinal mucosa of man, receptor GM, is indicated pig, and beef and amounted by the finding that incubation of fat cells with GM, in- to 0.1, 2.0, and 43 nmol per g fresh weight, respectively. creased These differences in GMI content were associated with a the toxin binding ability of the cells and also the li- quantitatively differing ability of the mucosal cells to bind polytic response to the toxin, which suggested that the exog- . Human cells bound about 15,000 toxin mole- enous ganglioside was incorporated as functional receptor in cules when saturated with the toxin, porcine cells 120,000, the and bovine (13). cells 2,600,000 molecules. The association con- In the present study, we investigate the binding properties stant (KA) of the cholera toxin binding was, for cells of all of cholera to three species, about 109 liters/mol. Exogenously added GM1 toxin intestinal mucosa cells of various species, ganglioside was incorporated in intestinal mucosal cells as including man, and examine how these properties relate to well as in intact rabbit small bowel. The increment in GM1 the ganglioside pattern of the mucosa, in particular to the was associated with a correspondingly increased number of content of GM1 ganglioside. We further directly demonstrate binding sites for cholera toxin, whereas KA was unchanged. incorporation of GM, ganglioside in intestinal cells and ana- GM! incorporation increased the sensitivity of the rabbit the effects of small bowel to lyze such incorporation on both binding of the diarrheogenic action of cholera toxin. toxin to the on Vibrio cholerae sialidase hydrolyzed isolated intestinal di- cells and the diarrheogenic action of cholera and trisialogangliosides to GM,. However, the enzyme did toxin on the intact small bowel. It is well known that GM1, not change the ganglioside pattern of intestinal mucosa, had which by itself is resistant to sialidase, can be generated in very little influence on the number of toxin binding sites on vitro from more complex by the action of V. intestinal cells, and. did not alter the sensitivity of the small cholerae sialidase The that such bowel to (1). possibility hydrolysis of the diarrheogenic action of the toxin. membrane-located di- and could occur These results demonstrate a relationship in the intestinal trisialogangliosides mucosa between the GM, ganglioside concentration, the as a contributory pathogenic event in cholera infection has number of binding sites for cholera toxin, and the sensitivity been recognized (9, 10). Consequently, we tried to evaluate to the biologic action of the toxin. Thus, the study strongly the ganglioside-converting effect of V. cholerae sialidase on supports the concept that the GM, ganglioside is the intesti- intestinal ganglioside preparations, cells, and intact small nal binding receptor for cholera toxin. bowel mucosa as well as the possible influence of enzyme ac- tion for and of cholera Diarrhea in cholera is due to the action on the small intestine binding diarrheogenicity toxin. of a protein exotoxin produced by the infecting Vibrio chol- erae (2). The isolated cholera toxin, molecular weight 84,000 MATERIALS AND METHODS (3), consists of two types of noncovalently linked subunits, one heavy (H) and several light (L) ones (4-6). The complex Cholera Toxin and Toxoid. Culture filtrate of V. cholerae of L subunits gives a rapid, tight binding of toxin to the (lot 4493 G) and highly purified toxin (lot 0572) were re- outer membrane of mammalian cells, which is followed by a ceived from Dr. Carl Miller, Cholera Research Program, slower biologic action probably mediated by the H subunit NIAID, Bethesda, Md. Purified choleragenoid toxoid was a (4-7). The key action of toxin on cellular seems gift from Dr. R. A. Finkelstein, Dallas, Texas. The purified to be activation of adenylate cyclase, and in the intestine the toxin and toxoid had been isolated as described (14). Neither resulting cyclic AMP (cAMP) accumulation causes body-to- the purified nor crude toxin had demonstrable sialidase ac- lumen secretion of chloride, which leads to diarrhea (8). tivity, as tested with GD1a ganglioside as substrate. 125I-La- The monosialosylganglioside GM1 was recently found to beled toxin was prepared with Na 125I and the Chloramine bind and inactivate cholera toxin with an affinity and speci- T coupling method as described (7), and the fraction of pro- ficity which suggested that this ganglioside, believed to be a tein similar in size to native toxin was isolated by column fil- normal constituent of the plasma membrane of most mam- tration through Sephadex G-100. More than 90% of the ra- malian cells, might be the membrane binding receptor for dioactivity of this fraction showed specific binding to cells toxin (9-11). The observation that aninactive toxin deriva- (mouse thymocytes) or to the ganglioside GM, (coupled to tive, choleragenoid toxoid, which binds as efficiently as toxin polystyrene tubes or to agarose beads), and the binding to target cells (7, 12), also had toxin-like binding properties properties of the labeled toxin were indistinguishable from to the GMi ganglioside in vitro (10) supports this view. Fur- those of unlabeled toxin. The biologic activity was also unaf- fected by the labeling (permeability increase of rabbit skin Abbreviations: The nomenclature of Svennerholm (ref. 1) is used to and thymocyte cAMP generation). designate the gangliosides: GM3, monosialylgalactosylglucosylcera- . Pure GM1 ganglioside, the sialic mide; GD3, disialylgalactosylglucosylceramide; GM1, galactosyl-N- acid-free acetylgalactosaminyl-(sialyl)-galactosylglucosylceramide; GD1, disi- derivative GA1, and GD1a ganglioside were pre- alylgalactosyl-N-acetylgalactosaminylgalactosylglucosylceramide; pared from human and characterized as described GTI, trisialylgalactosyl-N-acetylgalactosaminylgalactosylglucosyl- (15). [3H]GMl ganglioside was prepared by sequential treat- ; GA1, ganglio-N-tetraose, galactosyl-N-acetylgalactosam- ment with galactose oxidase and sodium [3H]borohydride inylgalactosylglucosylceramide. (16). The radioactive ganglioside was purified by column 2520 Downloaded by guest on September 25, 2021 Biochemistry: Ho1mgren et al. Proc. Nat. Acad. Sci. USA 72 (1975) 2521 chromatography on Sephadex G-25 and on silicic acid. The medium/bovine serum albumin with different concentra- ganglioside was finally purified by preparative thin-layer tions of unlabeled toxin. Each loop was then cut open and chromatography on Silica Gel G in a solvent system of chlo- dip-washed several times in minimal medium; its radioactiv- roform/methanol/water 60:32:7 (by volume). The specific ity was determined. activity was 2.0 X 106 dpm/nmol. Biological Action of Cholera Toxin and Inhibition with Sephadex G-25 and DEAE-Sephadex A-25 were pur- Toxoid. The ability of crude toxin (V. cholerae culture fil- chased from Pharmacia Ltd., Uppsala, Sweden; Silica Gel G trate) to cause fluid secretion in the small bowel of live rab- and H from Fluka AG, Buchs, Switzerland. V. cholerae sial- bits was quantitated in the prepared intestinal loops (17). idase (500 U/ml) was purchased from Behringswerke AG, Quantitative inhibition of the diarrheogenic action of the Marburg-Lahn, West Germany. crude toxin by choleragenoid (18, 19) was studied by inject- Intestinal Preparations for Binding and Activity Stud- ing different amounts of choleragenoid in a volume of 0.5 ies. Human cells were prepared from surgically removed ml into the intestinal loops 15 min before injection of an segments of jejunum, 15-20 inches (38-50 cm) from the py- equal volume containing 0.1 or 1.0 mg of V. cholerae cul- lorus, of three patients. Porcine and bovine cells were pre- ture filtrate. The assay was the same as that used for the de- pared from 1.5-m segments taken immediately after slaugh- termination of toxin action. ter from the middle of the small intestines of adult animals. Isolation of Gangliosides from Human, Pig, and Bovine The gut segments were placed on ice, cut open longitudinal- Intestine. The same bowel material was used as for the iso- ly, and with Ringer's solution gently washed free from visi- lation of intestinal cells. The jejunum and ileum were evert- ble mucus. After incubation with shaking at 370 for 30 min ed over a cold rod, and the mucosa was carefully rinsed in Ringer's solution, the cell-rich fluid was poured through a from mucus with physiological saline and then scraped off nylon mesh. The passed cells were sedimented at 200 X g with a surgical knife. Great care was taken not to obtain any for 10 min and washed once with Ringer's solution. About of the muscular layer, particularly at the preparation of the 80-90% of the human cells were viable and more than 90% human jejunum. Porcine mucosa (555 g), porcine muscular of the animal cells, as judged by trypan blue exclusion and layer (2.0 kg), bovine mucosa (185 g), bovine muscular layer phase contrast microscopy. They were suspended in Eagle's (650 g), human mucosa (8-25 g; 4 samples), and human minimal essential medium supplemented with 1 mg/ml of muscular layer (54 g) were available for ganglioside isola- CaCl2 to give cell densities of 5 X 106 per ml for the human tion. To two of the samples of human mucosa 0.1 nmol of cells and 2 X 106 per ml for the porcine and bovine cells. Al- [3H]GM1 was added. The mucosa scrapings and the samples iquots of the cell suspensions were incubated at 370 for 25 of muscular layer, after homogenization in a Waring blen- min with GM1 or [3H]GM1 ganglioside in concentrations dor, were extracted with 6 volumes of chloroform/methanol ranging from 0.07-2 ,gM (0.1-3.0 ,g/ml), V. cholerae sial- 1:2 (by volume). The gangliosides, except for ganglioside idase (1-50 U/ml), or with no additive. The cells were then GM3, were isolated by partition and purified by column washed four times in minimal medium by centrifugation at chromatography on Sephadex G-25 and Silica Gel G (Sven- 200 X g for 10 min and resuspended to the same densities in nerholm, in preparation). All the ganglioside preparations, minimal medium supplemented with 0.2 mg/ml of freshly particularly those of human and pig mucosa, were still very dissolved bovine serum albumin. impure, and the gangliosides of all the six sources were fur- Small bowel segments were prepared in live, anesthetized ther purified by mild alkaline hydrolysis, chromatography 8- to 12-week-old rabbits, that had been starved for 30-48 on Sephadex G-25, DEAE-Sephadex A-25 (20), Sephadex hr. Starting 50 cm from the pylorus, about 100 cm of the gut G-25, and finally on Silica Gel H (21). The quantitative de- was ligated, and via connecting tubes to each end of this seg- terminations of gangliosides were performed on the final el- ment it was rinsed with borate/gelatin buffer, pH 7.5 (7), uate of the Silica Gel H column by the resorcinol method supplied under mild hydrostatic pressure until the outflow (22). The ganglioside pattern was determined as for brain fluid was clear. The rinsed gut was then tied off into 25- or (23). In the human mucosa, the ganglioside concentration 50-cm segments, which were incubated for 20 min with 0.5 was extremely low and total gangliosides were only sepa- ml/cm of gut of GM1 ganglioside (0.07 or 0.7 ,M in borate/ rated into three fractions: ganglioside GM1, gangliosides GD1 gelatin buffer), V. cholerae sialidase (1, 10, or 20 U/ml in + GT1, and ganglioside GD3. Ganglioside GM, and ganglios- borate/gelatin buffer with 1 mg/ml of CaC12), or, as control, ides GD1 and GT1 were hydrolyzed with V. cholerae sial- with the buffer alone or, in a few instances, with gangliote- idase, and the hydrolyzed gangliosides were semiquantita- traose GA1 (0.7 ,M in borate/gelatin buffer). Thereafter the tively estimated on thin-layer plates sprayed with resorcinol segments were rinsed as above with 200-300 ml of borate/ (22). gelatin buffer, and further ligated into approximately 4-cm The neutral moiety of the gangliosides of long test loops. bovine and porcine intestine was characterized in the fol- Binding of Cholera Toxin. Binding of toxin to the dis- lowing manner. The apparent ganglioside GM1 and the gan- persed cells was tested by means of 125I-labeled toxin mixed gliosides GD1 and GT1 were isolated by thin-layer chroma- with multiple concentrations of unlabeled toxin. The test tography. These fractions were hydrolyzed with V. cholerae procedure and the Scatchard plot calculations of the associa- sialidase at pH 6.8, and the gangliosides that migrated as tion constant, KA, and number, n, of bound toxin molecules ganglioside GM1 were subjected to acid hydrolysis with 1 M when the cells were saturated with toxin have been de- formic acid at 1000 for 1 hr (24). The neutral tetraglycosyl- scribed for thymus cells (7). Modifications were the use of a ceramide formed was analyzed with thin-layer chromatog- greater number of toxin concentrations, and testing aliquots raphy in chloroform/methanol/water 65:25:4 and 60:32:7 of 4 X 105 porcine and bovine cells as well as 1 X 106 and 5 (by volume) with neutral glycosylceramides of known chem- X 106 human cells. ical structure as reference substances. Binding of toxin to rabbit small bowel mucosa was tested Sialidase Hydrolysis of Pig Mucosa. Porcine mucosa by removing the described entire loop-divided intestinal (100 g) was hydrolyzed for 30 min at 370 with 200 units of segment, placing it on ice, and incubating the loops for 20 V. cholerae sialidase in a total volume of 200 ml of 0.05 M min with 1 ml each of 125I-labeled toxin mixed in minimal acetate buffer pH S.0 or 0.05 M Tris-maleate buffer pH 6.8. Downloaded by guest on September 25, 2021 2522 Biochemistry: Holmgren et al. Proc. Nat. Acad. Sci. USA 72 (1975) was roughly of the same magnitude (Table 1). Although the MUCOSA MUSCULARIS total ganglioside content of the mucosa differed manyfold PIG BEEF REF PIG BEEF between the species, the concentration of ganglioside GM1 differed much more. Pig had a concentration that was ap- proximately 20 times larger than that in man, and beef a concentration 21 times larger than that in pig, which means that the GM1 concentration of human intestinal mucosa was only 1/400 of that in the bovine mucosa. Human and porcine mucosa contained large concentra- . ~~GM1 tions of simple ganglioside mono- and disialosyllactosylcera- mide (GM3 and GDM). More than 90% of the value for total ganglioside of man (Table 1) was accounted for by these two gangli6sides. In the human intestine, particu- larly in the muscular layer, the glucosamine-containing gan- glioside with a neutral lactoneotetraose moiety (25) was also a large fraction. The concentration of the presumptive pre- cursors of ganglioside GM1, gangliosides GD1 and GT1, were low in human but much larger in bovine and porcine muco- sa. The ganglioside patterns of porcine and bovine intestine G (Fig. 1) differed from those of human brain, because the gangliosides contained both N-acetyl- and N-glycolylneu- raminic acid. The gangliosides that migrated slower than GM1 were isolated and hydrolyzed with V. cholerae sialidase at pH 6.8. They were completely hydrolyzed to gangliosides that migrated as ganglioside GM1 with N-acetyl- and N-gly- colylneuraminic acid. The identity of GM1 was proven by determination of its components and by acid hydrolysis with 1 M formic acid at 1000 for 1 hr. A neutral tetraglycosylcer- amide was formed which migrated as N-gangliotetraose (GA1) in chloroform/methanol/water 65:25:4 and 60:32:7 (by volume). FIG. 1. Thin-layer chromatogram of the gangliosides from the Incubation of Porcine Mucosa with V. cholerae Sial- mucosa and the muscular layers of small intestine of beef and pig. idase. Porcine mucosa was subjected to sialidase treatment Portions of 20 nmol of N-acetylneuraminic acid were analyzed, at pH 5.0 or 6.8. The concentrations of total ganglioside sial- and the plate was developed for 18 hr at 230 in propanol/water. ic acid were the same in the incubated samples as in the con- The gangliosides were visualized by spraying with the resorcinol reagent (22). REF is a total ganglioside extract of beef brain. trol samples, which indicates that no hydrolysis occurred. This was further proven by quantitative determination of

The gangliosides were isolated by the same technique as the ganglioside pattern before and after treatment. The pro- used for the large-scale isolation of gangliosides of porcine intestinal mucosa. Table 2. Membrane incorporation of GM, ganglioside in intestinal mucosal cells and effect RESULTS on cholera toxin binding Gangliosides in the Intestine of Man, Beef, and Pig. In Toxin molecules [3H]GM1 the intestine there were large differences in the ganglioside GM, maximally molecules concentration of the mucosa between the three species, concentra- bound per cell incorporated while the ganglioside concentration of the muscular layer Species tion* (,M) X 10-4 per cell X 10 Table 1. Concentration of gangliosides Man 0 1.5 (nmol/g fresh weight) in small intestine of man, pig, and beef 0.2 2.5 2.0 8.0 Ganglioside Pig 0 12 sialic acid GM, GD, + GT, 0.07 25 18 0.2 60 44 Mucosa 0.7 110 166 Man 5 0.1 0.3 Beef 0 260 Pig 14 2.0 4.7 0.2 290 4.7 Beef 83 43 7.8 Muscular 0.7 295 22 Rabbit 0 L.Ont layer 0.07 1.7nt Man 54 3.5 12 0.7 4.2nt Pig 21 1.2 6.4 Beef 68 19 15 * The cells were incubated at 370 for 25 min with GM, ganglioside Cerebral in minimal medium/bovine serum albumin and then carefully cortex* washed. Man 2200 438 711 t The numerical values could not be calculated since the experi- ment was performed in intestinal segments whose number of cells * Vanier et al. (21). is unknown. Downloaded by guest on September 25, 2021 Biochemistry: Holmgren et al. Proc. Nat. Acad. Sci. USA 72 (1975) 2523

E Ia b E 0.8 0

IL z -XA I z 0 0.003 0.03 0.3 0 0.3 1 ~13 CRUDE TOXI N. mg INHIBITING CHOLERAGENOID. ug FIG. 2. Effect of pretreating rabbit small intestine with GM1 CRUDE TOXIN mg INHIBITING CHOLERAGENOID ug ganglioside on the sensitivity of the bowel to the diarrheogenic ac- FIG 3. Effect of treating rabbit small intestine with V. cholerae tion of toxin (a) and to choleragenoid toxoid inhibition of the toxin sialidase on the sensitivity of the bowel to the diarrheogenic action action (b). (a) Hatched bars are mean values (+SEM) of fluid ac- of toxin (a) and to choleragenoid toxoid inhibition of the toxin ac- cumulation in four to five differently positioned small bowel seg- tion (b). (a) mean values (+SEM) of fluid accumulation in 9 to ments of different animals, preincubated with 0.7 gM GM1 for 20 12 randomly positioned small bowel segments of different animals, min and then carefully washed before challenge with the indicated preincubated with 10 U/ml of sialidase for 20 min and then washed amounts of crude toxin; open bars are corresponding values for before challenge with the indicated amounts of crude toxin; 0, cor- buffer-treated control segments. The GM1 treatment enhances the responding values for buffer-treated control segments. (b) Mean intestinal sensitivity to toxin. (b) Corresponding values of GM1- values of fluid accumulation in sialidase-treated (filled symbols) treated and buffer-treated segments exposed to the indicated and buffer-treated (open symbols) segments exposed to the indi- amounts of choleragenoid for 15 min before challenge with 0.1 mg cated amounts of choleragenoid for 15 min before challenge with of crude toxin. Choleragenoid inhibits the toxin action in the con- 0.1 (A, A) or 1.0 mg (-, 0) of crude toxin. In no instance do the trol (open bars) but not in the GMl-treated segments (hatched values for the sialidase- and buffer-treated segments differ signifi- bars). cantly (Student's test).

portions of di- and trisialogangliosides were unchanged after treated intestinal cells or small bowel segments were also sialidase hydrolysis. compared. Only a slight enhancing effect on the number of Binding of Cholera Toxin to Small Intestine. By means intestinal binding sites for toxin, much less than that caused of '25I-labeled toxin in mixture with unlabeled toxin, specif- by GM1 treatment of the cells, was found after the sialidase ic saturable binding was demonstrated of toxin to dispersed treatment. Thus, the increment in toxin binding was for the small bowel mucosal cells of human, porcine, and bovine or- human, porcine, and bovine cells only about 20, 40, and igin. The association constant, KA, in the binding to the dif- 10%, respectively, even when the enzyme concentration was ferent cell types was similar, about 109 liters/mol, but the very high (40-50 U/ml), and for the intact rabbit mucosa number of receptors per cell differed much between the about 30% (10 U/ml of sialidase). The KA values for toxin- species. When saturated, the human cells bound, on the av- binding to sialidase-treated and untreated cells or segments erage, 15,000 toxin molecules per cell, the porcine cells were similar. 120,000 molecules, and the bovine cells 2,600,000 molecules Biologic Action of Toxin on Small Intestine. Influence (Table 2). of GMl-Ganglioside Incorporation and V. cholerae Sial- Specific binding of toxin to intact rabbit small bowel mu- idase. The sensitivity to the diarrheogenic action of crude cosa was also demonstrated. Since the number of cells per in- toxin (V. cholerae culture filtrate) was compared for buffer- testinal loop was unknown, the number of bound toxin mole- treated and GM1 ganglioside-preincubated rabbit small cules per mucosal cell could not be estimated. The binding bowel loops (Fig. 2a). It was found that the GM1 ganglioside affinity, however, was found to be similar to that observed treatment made the loops responsive tootoxin concentrations with dispersed cells of the other species. that were too low to be effective in the control loops; the GM1 Ganglioside Incorporation and Cholera Toxin maximal response, on the other hand, was unchanged. Binding. By means of [3H]GM1 it could be directly demon- Treatment of loops with the gangliotetraose GA1 did not strated that this ganglioside, by mere incubation at 370 for change the sensitivity to toxin. 25 min, can be firmly attached to intestinal cells, probably To verify that the enhanced toxin sensitivity of the GMI- by incorporation in the outer membrane. This was qualita- treated intestinal loops could be ascribed to an increased tively demonstrated with all cells tested, i.e., human, por- number of toxin-binding sites, we tested whether more chol- cine, and bovine mucosal cells, but great quantitative differ- eragenoid toxoid was required to inhibit the secretory re- ences were noted between the porcine and bovine cells sponse to subsequent toxin challenge in GMlLtreated than in (Table 2). buffer-treated loops. Fig. 2b shows that this was the case. Incubation of intestinal cells with GM1 ganglioside did not Choleragenoid completely prevented the secretory response change the KA values in the binding of toxin to the cells, but of the control loops, but in the tested amounts it had no in- led to an increase in the number of toxin molecules that hibitory action in the GM1-treated loops. Choleragenoid could bind to the cells. This would indicate an increased binds as efficiently as cholera toxin to both GM1 ganglioside number of receptors of identical qualities as the natural in vitro (10) and to mammalian cells (7, 12), which indicates ones. Table 2 shows such GM1 concentration-dependent in- that it inhibits the secretory response to toxin by occupying creases in the number of receptors for toxin on human and the mucosal GM1 receptors for the toxin. porcine intestinal mucosa cells and in rabbit small bowel In a similar manner the influence of treating rabbit intes- segments; on the GMl-rich bovine intestinal cells the effect tinal loops with V. cholerae sialidase was evaluated (Fig. 3). was much less. The sialidase-treated loops showed no different sensitivity Influence of Sialidase on Cholera Toxin Binding. The than the control loops to the diarrheogenic action of toxin toxin-binding properties of buffer-treated and sialidase- (Fig. 3a), nor did they require more choleragenoid for inhi- Downloaded by guest on September 25, 2021 2524 Biochemistry: Holmgren et al. Proc. Nat. Acad. Sci. USA 72 (1975) bition of fluid production to toxin challenge (Fig. Sb). The quired to inhibit the toxin action on the gut. These findings quantitative inhibition by choleragenoid of two different contradict a major pathogenetic role of V. cholerae sialidase doses of challenge toxin gives further support for the con- production as a toxin-receptor-creating enzyme in the intes- cept of receptor competition with similar affinity between tine. In contrast to our findings, King and van Heyningen the toxoid and toxin. (9) found with sialidase treatment an 8-fold increase of the To study a possible delayed influence of sialidase on the toxin-inactivating capacity of rabbit and bovine intestinal intestinal sensitivity to toxin, the incubation time with the mucosa scrapings; they also claimed that the ganglioside pat- enzyme in the intestine was extended to 200 min in three terns of bovine brain and intestinal mucosal scrapings did animals. This, however, did not change the sensitivity for not show any remarkable differences. Our Fig. 1 illustrates toxin of the enzyme-treated segments from that of the con- that the ganglioside pattern of the intestinal muscular layer trol segments incubated with borate/gelatin buffer. is similar to that of brain, which is not unexpected since the Finally, it was found that incubation of the intestine in gangliosides of the muscular layer are partly derived from mvo with 10 U/ml of sialidase for 20 min did not result in the nerve cells in this tissue. Contamination of the mucosal liberation of more free sialic than incubation with buffer; scrapings tested by King and van Heyningen with muscular the enzyme was fully active after the incubation, however, layer tissue might possibly explain the divergent findings as as tested with GDla ganglioside as substrate. compared with those of the present study.

DISCUSSION We thank Dr. Tore Schersten for the specimens of human small The concept, derived from the observations described in the intestine and Dr. Harald Almgren, Scan, Uddevalla, for the facili- introduction, that the GM1 ganglioside is the binding recep- ties he put at our disposal for the work with beef and pig intestine. tor for cholera toxin in the small bowel has gained much This work was supported by grants from the Swedish Medical Re- support from the present study. The occurrence of GM1 is search Council (no. 6X-3383 and 3X-627), the W.H.O., and the demonstrated in the small bowel mucosa of various mamma- Hesselman Foundation. lian species, including man, and the marked differences in the content of this ganglioside among the species are found 1. Svennerholm, L. (1963) J. Neurochem. 10, 613-623. to correspond to different numbers of toxin molecules that 2. Pierce, N. F., Greenough, W. B., III & Carpenter, C. C. (1971) can bind per intestinal cell. Further, it is shown by means of Bacteriol. Rev. 35, 1-13. that the exogenously added GM1 can be incorporat- 3. LoSpalluto, J. J. & Finkelstein, R. A. (1972) Biochim. Biophys. [3H]GMS Acta 257, 158-166. ed in intestinal mucosal cells or intact small bowel, and that 4. Lonnroth, I. & Holmgren, J. (1973) J. Gen. Microbiol. 76, such incorporation increases the number of toxin molecules 417-427. that can bind per cell as well as the sensitivity of the intes- 5. Cuatrecasas, P., Parikh, I. & Hollenberg, D. (1973) Biochemis- tine to the diarrheogenic action of toxin. try 12, 4253-4264. Large differences were found in the concentration of GM1 6. Holmgren, J. & Lonnroth, I. (1975) J. Gen. Microbiol 86, 49- ganglioside in the intestinal mucosa of man, pig, and beef. It 66. is striking that the 21-fold higher GM1 concentration in beef 7. Holmgren, J., Lindholm, L. & Lonnroth, I. (1974) J. Exp. as compared with pig mucosa was associated with a 22-fold Med. 139, 801-819. higher average number of binding sites to toxin in the bo- 8. Field, M. (1971) N. Engl. J. Med. 284, 1137-1144. 9. King, C. A. & van Heyningen, W. E. (1973) J. Infect. Dis. vine than in the porcine mucosal cell. It is also notable that 127,639-647. the naturally GM1-rich bovine cells incorporated many 10. Holmgren, J., Lonnroth, I. & Svennerholm, L. (1973) Infect. fewer [3H]GM1 molecules than the porcine cells, and paral- Immun. 8,208-214. lelly did not increase the number of toxin binding sites per 11. Cuatrecasas, P. (1973) Biochemistry 12,3547-3558. cell as much as the porcine cells. In consideration of the 12. Cuatrecasas, P. (1973) Biochemistry 12,3577-3581. complicated experimental procedures, the rather close 13. Cuatrecasas, P. (1973) Biochemistry 12, 3558-3566. agreement in the porcine cells between the number of 14. Finkelstein, R. A. & LoSpalluto, J. J. (1970) J. Infect. Dis. 121, [3H]GM1 molecules incorporated per cell and the increase in S 63-72. number of toxin molecules that were bound by a toxin-satu- 15. Svennerholm, L. (1972) in Methods in Carbohydrate Chemis- mean ratio try, eds. Whistler, R. L. & BeMiller, J. N. (Academic Press rated cell, the being 1.3, suggests that almost Inc., New York), Vol. 6, pp. 464-474. every GM1 molecule incorporated is in a position to attach 16. Radin, N. S., Hof, L., Bradley, R. M. & Brady, R. 0. (1969) one toxin molecule. Moreover, the association constant in Brain Res. 14,497-505. binding is similar, around 109 liters/mol, in the GMl-treated 17. Kasai, G. J. & Burrows, W. (1966) J. Infect. Dis. 116, 606- cells as in the untreated ones, further indicating that the in- 614. corporated GM1 molecules have identical toxin-binding 18. Pierce, N. F. (1973) J. Exp. Med. 137, 1009-1023. properties as the natural receptors. 19. Holmgren, J. (1973) Infect. Immun. 8, 851-859. V. cholerae sialidase can convert isolated di- and trisialo- 20. Yu, R. K. & Ledeen, R. W. (1972) J. Res. 13,680-686. gangliosides to the sialidase-resistant GM1 ganglioside, as 21. Vanier, M.-T., Holm, M., MAnsson, J.-E. & Svennerholm, L. shown previously (1) and in the present study. However, (1973) J. Neurochem. 21, 1375-1384. 22. Svennerholm, L. (1957) Biochim. Biophys. Acta 24,604-611. when permitted to act in vitro on intestinal cells or mucosa 23. Vanier, M.-T., Holm, M., Ohman, R. & Svennerholm, L. or in ivo in the small bowel, the enzyme did not change the (1970) J. Neurochem. 18,581-592. ganglioside pattern of the mucosa, had little influence on the 24. Holm, M., Mansson, J.-E., Vanier, M.-T. & Svennerholm, L. number of toxin-binding sites per intestinal cell, and altered (1972) Biochim. Biophys. Acta 280,356-364. neither the sensitivity of the small bowel to the diarrheogen- 25. Li, Y.-T., MAnsson, J.-E., Vanier, M.-T. & Svennerholm, L. ic action of toxin nor the amount of choleragenoid toxoid re- (1973) J. Biol. Chem. 248, 2634-2636. Downloaded by guest on September 25, 2021