Institute of Medical Microbiology and Department of Neurochemistry. Uniersity of G6teborg, GCteborg,. Su eden. Received for publication 26 March 1973.
Vol. 8. No. 2 Printed in U.S.A.
INFECTION AND IMMUNITY. Aug. 1973, p. 208-214 Copyriaht (3 1973 American Society for Microbiology
Tissue Receptor for Cholera Exotoxin: Postulated Structure from Studies with GM, Ganglioside and Related Glycolipids .J. HOLMGREN. 1. LONNROTH. ANM) L. SVENNERHOLM Institute of Medical Microbiology and Department of Neurochemistry. Uniersity of G6teborg, GCteborg, Su eden Received for publication 26 March 1973
By a double-dif'fusion precipitation-in-gel technique. isolated cholera toxin as well as its natural toxoid were shown to be fixed and precipitated by the ganglioside G., but not by any of the related glycolipids G,,3, G\,,, G,11-GlcNAc, Gj)ja, GDib, G,,, globoside, G \, and tetrahexoside-GIcNAc. 'ITwenty-t'ive nanograms oft'G Xl was enough to give a precipitation line with 1.2 ag of toxin, whereas about 50 ng was required with this amount of toxoid. GMl also inactivated the toxin in the ileal loop as well as in the intradermal models in rabbits. A 1:1 molar ratio of' ganglioside to toxiIn was found limiting. e.g,.. 100 pg of G \ could inactivate 5 ng, (about 50 blueing doses) of isolated toxin. G, I inactivated crude toxin (culture f'il rate) with the same etf'iciency as isolated toxin, and the inactivating capacity of GX,, was unatf'ected by mixing with other gangliosides, indicating the specificity in the reaction between G\, and toxin. The other glvcolipids tested did not inactivate toxin except G,,ia and G , which did so with approximatelN 1,000 times less etf'iciency than GIl,. This identit'ied the portion Gal GaINAc Gal -- as the critical regyion in Gl, for toxin fixation. and it is p)ostulated NAN that this mav be the tissue receptor structure t'or the cholera toxin.
Vibrio cholerae bacteria produce a diarrhoeogenic exotoxin which has been isolated and characterized (5, 11; I. Lonnroth and J. Holmgren, 19713, J. Gen. Microbiol., in press). 'I'he initiating event in its toxic action in diverse tissues including the gut appears to be the attachment to a cell membrane receptor with the subsequent activation of adenyl cyclase (14). It was recently tl)served that a crude ganglioside mixture preparation could inactivate the cholera toxin (17) which opened the possibility that the tissue receptor could be a ganglioside substance in this mixture. Direct evidence for the ganglioside nature of' the receptor was provided b)y Holmgren et al. (8a) who demonstrated fixat ion as well as inacti\vation of' cholera toxiIn by a pure ganglioside. G .\ . In the present study these properties of' G., are analyzed in more detail with special reterence to the speciticity and at't'initv in the reactions. Comparative studies are done with structurally related substances, and the binding to natural cholera toxoid is analyzed. The crit ical p)ort ion of' G\, with respect to f'ixation of' toxin is 208
identif'ied, and it is postulated that it may be identical to the tissue receptor for cholera toxin. MATERIALS AND METHODS Cholera toxin and toxoid. Isolated cholera toxin (choleragen) was prepared under contract for the National Institute of Allergy and Inf'ectious Diseases (NIAID) by R. A. Finkelstein and obtained via R. Northrup. The lot number was 1071. Isolated natural cholera toxoid (choleragenoid) was a gitt from R. A. Finkelstein. The isolation procedures have been described in detail (5). In a tew experiments culture f'iltrate f'rom V. cholerae was employed. The f'iltrate (lot 4493G) had been obtained lyophilized from NIAID via J. R. Seal. The materials used were dissolved in 0.1 M tris(hydroxymethyl)aminomethane( Tris) -hydrochloride buftfer, pH 7.5., containing 0.2%' gelatin for stabilizing purposes.
Vibrio cholerae sialidase. The preparation used was purchased from Behringwerke, Marburg-Lahn, and had an enzyme activity of 500 international units (IU) per ml. Other proteins. Normal rabbit serum, puritied humani serum gamma grlob)ulin and puritied human
VOL. 8, 1973
TISSUE RECEPTOR FOR CHOLERA EXOTOXIN
serum albumin were also used. The two latter preparations were gifts from AB KABI, Stockholm, Sweden. Antiserum to cholera toxin. A rabbit was given two subcutaneous injections with isolated toxin. The injections, 10 ,ug each, were spaced 3 weeks apart. The antiserum was obtained from a heart puncture bleeding 10 days after the last injection. Glycolipids. The gangliosides and allied neutral glycosylceramides were isolated from human brains with the chromatographic methods recently described (19). They were analyzed for total neutral sugars, hexosamine, sialic acid, and fatty acids by gas-liquid chromatography. The positions of glycosidic bonds were determined by permethylation (6), and the partially methylated sugars were converted into alditol acetates and were analyzed by gas-liquid chromatography and mass spectrometry (1, 2). The anomeric configuration of the glycolipids was determined by the sequential hydrolysis of the oligosaccharide chain by specific glvcosidases (6, 10). Table 1 shows the neutral glycosylceramides and gangliosides used and their chemical structures. The purity of the substances in general was better than 99%. Two preparations of G,11 were used. The first one, G,,l-I, was admixtured with 0.5'7. of G\llGlcNAc. By extensive treatment with sialidase from V. cholerae the G,l,-GlcNAc was hydrolyzed to the corresponding neutral tetraglycosylceramide (10a). The intact G,1, was finally isolated with preparative thin-layer chromatography. The purity of this second preparation of G,1 (G1,,-II) was better than 99.9%. All of the used glycolipid substances were dissolved in 0.1 M Tris-hydrochloride, pH 7.5, containing 0.2% gelatin when they were tested for their reactivity with cholera toxin or toxoid. Sialidase hydrolysis of gangliosides. Fifty micrograms of ganglioside was dissolved in 50 juliters of phosphate buffer (0.05 M, pH 7.2) to which was added 5 uiliters (2.5 IU) of the V. cholerae sialidase preparation. The mixture was incubated at 37 C for 18 h, and then evaporated to dryness and extracted with chloroform-ethanol, 1: 1 (vol/vol). The lipid extract and relevant ganglioside reference substances were then tested in analytical thin-layer chromatography in silica gel with chloroform-methanol-water 60:32: 7 (vol/vol/vol) as developing solvent (10a). Double diffusion-in-gel. A sensitive double-diffusion-in-gel microplate method (17) developed for analyses of antigen-antibody precipitation reactions (12) was used to study the capacity of the glycolipids to fix and precipitate toxin or toxoid in vitro. The circular wells were filled with 25 gliters of the reactants, which were allowed to diffuse in a humid atmosphere for 4 days at room temperature. Registration of precipitation lines was done daily as well as after the plate had been stained for protein precipitates with Coomassie brilliant blue (Mann Research Lab., New York, N.Y.). Removal of nonprecipitated material and the staining procedure was done as described (9). Toxicity tests. The ileal loop technique (4) and the intradermal test (3) in rabbits were used to study the capacity of the gangliosides and the related neutral glycosylceramides to inactivate the gut and the skin effects of cholera toxin.
209
In the loop tests 8- to 12-week-old partially inbred rabbits were used, and five loops were arranged in each animal as previously described (8). Preliminary experiments showed that the mean effective dose (ED50) of the isolated toxin in these animals was approximately 0.7 jg, and 3 ,ug of the toxin regularly caused a pronounced fluid accumulation (mean of nine animals 2.8 ml of fluid per cm of gut, range 2.1-4.2 ml per cm). A challenge dose of 3 mg of toxin was therefore chosen. This amount was mixed with glycolipid, usually 5 jig, and was incubated at 20 C for 15 min in a volume of 2 ml, and then injected into a ligated loop. Each combination of materials was tested in randomized positions in three to six animals; positive (3 ug of toxin incubated in gelatin-Tris buffer) and negative (only the gelatin-Tris buffer) controls were included in each animal. In the skin tests, the rabbits were selected to weigh approximately 2 kg because our experience with sensitivity and reproducibility of the test has shown better results in these than in older animals. A blueing dose (1 BD) was found to be approximately 100 pg of isolated toxin and 1.5 jg of culture filtrate 4493G. In the experiments, usually 5 ng of the toxin, but occasionally 10 ng, incubated with from 10 pg to 500 ng of glycolipid at 20 C for 15 min was tested, each combination in 4 to 10 randomized positions in at least two animals. When culture filtrate 4493G replaced tne isolated toxin, 3 jg was used as a test dose. Positive and negative controls analogous to those described for the ileal loop tests were included in several positions in each skin-tested animal. In several animals up to 500 ng of glycolipids incubated with the gelatin-Tris buffer but with no toxin were tested for control purposes.
RESULTS
Fixation and precipitation of toxin and toxoid. The gangliosides and related neutral glycosylceramides were tested with the doublediffusion-in-gel technique for their capacity to fix and precipitate cholera toxin or toxoid. It was found that, with the glycolipid amounts tested, from 2.5 ng to 10 jig, only the ganglioside
GM1 was reactive, giving a fine precipitation line with toxin as well as toxoid (Fig. la). The precipitate became visible after a diffusion time of about 40 h and increased in density during the next 2 days. Visualization was further improved by protein staining. At least 300 ng of toxin was required for a visible precipitate with GM1, but 1.2 jg gave a denser line which appeared more rapidly and this amount of toxin (and toxoid) was therefore chosen for detailed analyses. The precipitate formed between GM, and toxin or toxoid was not due to unspecific precipitation of protein. This was tested with various amounts of G,N, against a range of amounts of human serum albumin and gamma globulin as well as against normal rabbit serum. No precipitates were formed (Fig. lb).
INFECT. IMMUNITY
HOLMGREN, LONNROTH, AND SVENNERHOLM
210
TABLE 1. Gangliosides and neutral glycosylceramides used Chemical structureb
Code name,
Gal(3,1
GM3
--
4)Glc(1
-
4)Glc(1
-
1)Cer
)a,2
NAN
Gal(3, 1
GM3(NGN)
l)Cer
a,2 ) NGN
GN,2
GalNAc(3, 1
4)Gal(3, 1
4)Glc(1 -1l)Cer
(a,2 ) NAN
GM,
3)GalNAc(,B, 1 - 4)Gal(o, 1 B 4)Glc(1
Gal(,B, 1
1)Cer
I T
(a,2 ) NAN
GmI-GlcNAc
Gal(3, 1
-
4)GlcNAc(3, 1
3)Gal(o, 1
4)Glc(1
1)Cer
( a,2 NAN
GD
Gal(O, 1 3)GaINAc(3, 1 ,4)Gal(13, 1 4)Glc(l -- 1)Cer NAN
NAN
GDlb
Gal(OO, 1 - 3)GalNAc(o, 1 - 4)Gal(O, 1 - 4)Glc(1 l_)Cer NAN(a,2
GI,l
Gal(,M 1 - 3)GalNAc(,B, 1 - 4)Gal(fl, 1 - 4)Glc(1 l_)Cer a, c2
NAN
globoside
GAI
tetrahexoside-GlcNAc
8)NAN
2) )a( NAN(a, 2 -8)NAN
GalNAc(fl, 1 - 3)Gal(a, 1 - 4)Gal(,B, 1 _ 4)Glc(1 l l)Cer Gal(1, 1 - 3)GalNAc(o, 1 - 4)Gal(o, 1 4)Glc(1 1)Cer Gal(O, 1 - 4)GlcNAc(O, 1 -, 3)Gal(O, 1 -. 4)Glc(1 , 1)Cer
aAccording to Svennerholm (15). 'Abbreviations used are: NAN,N-acetylneuraminic acid; NGN, N-glycolyl-neuraminic acid; Gal, galactose; GalNAc, N-acetylgalactosamine; Glc, glucose; GlcNAc, N-acetylglucosamine; Cer, ceramide (N-acylsphingo-
sine). In the initial experiments GM,-I which was admixtured with 0.5% GMl-GlcNAc was employed. Comparative double-diffusion tests verified that GN, and not GNu 1-GlcNAc was respon-
sible for the fixation of toxin and toxoid, since the precipitation line between GMl-I and toxin or toxoid completely fused, i.e., showed "a reaction of identity," with the single line formed
211
TISSUE RECEPTOR FOR CHOLERA EXOTOXIN
VOL. 8, 1973
a
0 0
3
iD-1
~,
©
(le
1-1
(5)
b
0
(r3
between the pure GM1-II and toxin or toxoid (Fig. 2a). In contrast, isolated GMl-GlcNAc did not cause precipitation of toxin or toxoid (Fig. 2b). It is noteworthy that the precipitation lines between GM1 and toxin and toxoid, respectively, completely fused with each other (Fig. 2). The amount of GM1 (GM1-II) required to give a precipitation line in the gel-diffusion analyses when tested with 1.2 jig of toxin and toxoid was determined with dilution series (Fig. 3). Based upon repeated determinations it was found that 25 ng of GM1 was required to precipitate with the toxin, and approximately twice as much was required to give a line with the toxoid. However, the line with the toxoid was consistently of higher density than that with the toxin (Fig. 3). The precipitate with toxin and toxoid was, when 5- and 10-,ug amounts of GM1 were used, broader and more fuzzy than in better balanced systems with lower GM, amounts. Preliminary experiments were also done where the precipitate formed between GM, and toxin or toxoid was compared with the immune precipitate formed between toxin or toxoid and antitoxin immune serum (Fig. 4). Interaction between the precipitates was noted; with toxin a fusion of the lines was seen, possibly as an Ouchterlony type III reaction with spur formation (12), and with toxoid an inhibition of the line with GM, was visible in comparison with plates where no immune system was included (cf Fig. 4 with e.g. Fig. 2b). Inactivation of toxin. The capacity of each
a
(iH ©)
(/4)
©6 b
FIG. 1. Specific precipitation of GM, and toxin in double-diffusion-in-gel analyses. a, Toxin (basin 1) tested with GD1b (2), GDla (3), GMl-I (4), tetrahexoside-GlcNAc (5), globoside (6), and GA I (7). The toxin amount was 1.2 Mlg and the glycolipid amounts were 2.5 ,tg. b, A 0.6-,ug amount of GM, (1) tested with 1.2 Ag of toxin (2), with 1.2 Mg (3), 5.0 Mg (4), and 25 Mg (5) of human serum albumin and with 5.0 Mg (6) and 1.2 Mlg (7) of human gamma globulin. The photographs in this and the following figures are from protein-stained plates.
\
..
'I
/~~~
E /
94
4.~~~~~~~:r FIG. 2. Verification of GM, as the toxin-precipitating substance. a, Comparative double-diffusion analysis of GM ,-II (basins 1 and 3) and GM ,-I (2) tested with toxin (4) and toxoid (5); b, the same arrangement but with GM,-GlcNAc replacing GM,-II in basins 1 and 3.
212
a
INFECT. IMMUNITY HOLMGREN, LONNROTH, AND SVENNERHOLM stances (six tested animals) completely inhibited the accumulation of fluid, and GA, and GD,a which both did so in three out of four animals. More quantitative tests showed that the toxin was inhibited by lower amounts of GM1 than of GA, and GDla; 5, 1, and 0.1 ,ug but not 0.01 mg of GNuj (GMu-11) were effective, whereas only the highest of these amounts of GDia and GA, caused inactivation of the toxin. The more detailed analyses were done with the skin toxicity assay. Again it was found that the toxin was inactivated only by GM, (Gm,1-I as well as G,,,-1I), G \1 and G Dia, but not by any of the other substances in Table 1 which were tested in amounts up to 500 ng. For inactivation of 5 ng of toxin, i.e., approximately 50 BD, only 100 pg of GN,l (determined with GM,-II) was required, whereas about 1,000-fold this amount, 100 ng, was needed of GA, as well as GDla. The higher affinity of toxin for Gml, than for GA, and GDla is also illustrated in Fig. 5, where 10 ng of toxin was incubated with various amounts of the three glycolipids, whereupon active toxin was determined by skin toxicity testing of many dilutions. These determinations were compared with the activity of toxin incubated only with gelatin-Tris buffer.
..
'. So/
iA
_
.I
FIG. 3. Comparison of minimal amounts of GM . required for precipitation of toxin and toxoid. a, A 1.2-jig amount of toxin (basin 1) tested with G.U.-II in the amounts of 160 ng (2), 80 ng (3). 40 ng (4), 20 ng (5), 10 ng (6), and 5 ng (7). b, The same arrangement but with toxoid replacing toxin in basin 1.
FIG. 4. Comparative double-diffusion analysis of the precipitate formed between antiserum to toxin (basins 1 and 3) and toxin (4) or toxoid (5), and that formed between these two bacterial proteins and G, (2).
10
8-
c
&0-j-, a-
0..e
-"
*o
t
of the glycolipids listed in Table 1 to cause inactivation of cholera toxin was tested in the rabbit ileal loop and intradermal systems. In the ileal loop model, a fixed amount, 5 ,ug, of the glycolipid was tested with a regularly diarrhoeogenic amount, 3 jug, of isolated toxin. Each combination was studied in at least three animals. Only three of the substances caused a change in the fluid accumulation as compared to control loops given only the toxin. These were GN,u(preparation I as well as II) which in all in-
0~~~~
0
tI
11
M -S-A,.M T 7 0
I
.03
.1
I
.3
I
3 9 28 GLYCOSYLCERAMIDE, ng 1
83
250
750
FIG. 5. Inactivation of toxin by GMI (), GDI. (A), and GAR (-) determined as remaining skin activity of a mixture of 10 ng of the toxin and various amounts of the glvcosylceramides.
TISSUE RECEPTOR FOR CHOLERA EXOTOXIN
VOL. 8, 1973
It was ascertained that the capacity of GM, to inactivate cholera toxin was the same whether GM, was added alone to the toxin or mixed with the same, 10-fold or 100-fold amounts of another ganglioside of no or low inhibitory activity (tested with GM2, GDla, GDIb, and GT). Moreover, it was found that 100 pg of GM, (GM,-II) inhibited the skin toxicity of 3 ,g of V. cholerae culture filtrate, whereas, e.g., 50 ng of G N¶2, GDla, GDlb, or GT caused no inhibition. This shows the toxin specificity of G,,l since approximately 99.99% of the culture filtrate is of nonexotoxin nature. It was assessed that none of the gangliosides and other glycosylceramides employed in the present study caused any skin reactions in the tested amounts, i.e., up to 500 ng. The binding of GM, to toxoid was quantitatively studied. One volume (0.5 ml) of toxoid in different concentrations was mixed with one volume of a GM1 solution, 4 ng per ml, and 15 min later one volume of a solution of toxin, 100 ng/ml, was added. The toxicity of the mixture was tested in the skin of rabbits. The results are seen in Table 2. Toxoid effectively bound GN¶l, and, although the preincubation design favored the toxoid, it appears that the affinity of this protein for GM, was not much different from that of the toxin. The influence of sialidase from V. cholerae on Gnia and GM, was also tested at a pH similar to that in the lower ileum. It was found that the toxin-inactivating capacity of G Dla increased about a 1,000-fold, but that of GM1 was unaffected by the sialidase treatment. This effect apparently depended upon the enzymatic hydrolysis of GDla to GM, and free sialic acid; thin-layer chromatography analyses showed that this conversion was almost complete TABLE 2. Consumption of GMI by toxoid assessed by the subsequent addition of toxin and testing for
toxicity Amt in mixture (ng)a
Toxicity
Toxoid
GM
Toxin
(skin test)
330 100 33 10
0.13 0.13
3.3 3.3
3+b 3+
0.13
3.3 3.3 3.3 3.3 3.3
2.5+ 2.5+ 2+ 1+
3.3 1.0 0
0.13 0.13 0.13 0.13
-
a Refers to the amounts in the 0.1 ml injected in each skin position. bMean of four positions in two rabbits, where the toxicity was recorded from 3+ to -; a 2+ reaction corresponds to the effect of 1 BD of toxin.
213
( > 99.9%), whereas the GM1 was not changed by the sialidase treatment. DISCUSSION The initial event in the pathogenesis of cholera seems to be the binding of the cholera exotoxin to the mucosal surface, for which it has specific affinity (13). Recognition of the structure of the mucosal receptor is important, both to facilitate a better understanding of the mode of action of the toxin and to provide clues for a rational prophylactic-therapeutic approach on the basis of competitive inhibition. Our finding in a foregoing communication (8a) and confirmed in this study that a pure ganglioside, GmNl, fixed as well as inactivated cholera toxin suggests that this ganglioside constitutes or contains the receptor structure. Both the diarrhoeogenic- and the capillary permeability-increasing activities of the toxin were inhibited by GM1, which indicates that the initiating events in these toxin manifestations are identical. The quantitative analyses performed with the skin toxicity assay indicated that the pure GM, was about 500-fold more effective than the mixed ganglioside preparation used by van Heyningen et al. (17); 100 pg of GM1 inactivated approximately 50 BD of toxin as compared to 25 ng of their preparation for 20 BD. Thus, one weight unit of GM1 could inactivate up to approximately 50 weight units of toxin, which corresponds to 1: 1 molar proportions since the molecular weight of G,NI is 1,600 and that of toxin 84,000. This high affinity of isolated toxin for pure GM, was combined with high specificity, since the reactivity was similar when culture filtrate was used as a crude toxin containing approximately 99.99% nontoxin material or when GM1 was mixed with up to 100-fold of other structurally closely related glycolipids. Thus, with respect to affinity and specificity in the inactivation of cholera toxin, GM, had the properties expected of an isolated receptor substance. The double-diffusion variant of precipitationin-gel elaborated by Ouchterlony for immunological analyses (12) proved useful also for the study of the fixation of cholera toxin by GM-. In similarity with the bioassays one weight unit of GM, could precipitate 50-fold as much of toxin. Also, the natural cholera toxoid, but not the other proteins tested, precipitated with GM, and a "reaction of identity" was noted with the line formed with the toxin. This indicates that the receptor-combining site of the toxin is present also in the toxoid. This conclusion was verified in experiments where consumption of free GM, by toxoid was assessed by the subsequent addition of toxin, the toxicity of which was then
HOLMGREN, LONNROTH, AND SVENNERHOLM
214
INFECT. IMMUNITY
This investigation was supported by grant no. assayed. Moreover, these analyses showed that B72-16X-3383 and B72-40P-3592 from the Swedish Medical the affinity of G,N for toxoid was not much Research Council and by the Walter, Ellen and Lennart different from that for toxin, so affinity differ- Hesselman Foundation for Scientif'ic Research. ences for the tissue receptor do not easily seem to explain the different toxicity of the two LITERATURE CITED proteins. The data fit well with the observation 1. Bjorndahl, H., C. G. Hellerqvist, B. Lindberg, and S. that toxoid also binds specifically to the muSvensson. 1970. Gas-liquid chromatography and mass cosal surface (13) and suggest that the combinspectrometry in methylation analysis of polysaccharides. Angew. Chem. Int. Edit. 9:610-619. ing site of the toxin is located in the L subunits shared with the toxoid, rather than in the 2. Bj6rndahl, H., B. Lindberg, and S. Svensson. 1970. Gas-liquid chromatography of partiallv methylated toxin-specific H subunit (I. L6nnroth and J. alditols as their acetates. Acta Chem. Scand. Holmgren, 1973, J. Gen. Microbiol., in press). 21: 1801-1804. Experiments are under way to clarify this as 3. Craig, J. P. 1965. A permeability factor (toxin) found in cholera stools and culture filtrates and its neutralizawell as to analyse in more detail the possibilities tion by convalescent cholera sera. Nature (London) indicated in the present study to use the double207:614-616. diffusion technique for comparative investiga- 4. De, S. N. 1959. Enterotoxicity of bacteria-free culture tions of the receptor-binding and antibodyftiltrates of' Vibrio cholerae. Nature (London) 183:1533-15:34. binding regions of the toxin. R. A., and J. J. LoSpalluto. 1970. Production From the results with the different structur- 5. Finkelstein, of highly purified choleragen and choleragenoid. J. ally well-defined glycolipids used it is postuInf'ect. Dis. 121( Suppi) :S62-S72. lated that in
GM,
the
portion Gal
GalNAc
6. Hakomori, S. 1964. A rapid permethylation of glvcolipid
and polysaccharide catalyzed by methylsulfinyl carbGal, is the critical region for the fixation anion in dimethvl sulfoxide. J. Biochem. Tokvo t 55:205-208. NAN 7. Hakomori, S., B. Siddiqui, Y.-T. Li, S.-C. Li, and C. (C. inactivation of cholera toxin. Even minor Hellerqvist. 1971. Anomeric structures of'globoside and ceramide trihexoside of human erythrocytes and hamchanges in this structure severely affected the ster fibroblasts. J. Biol. Chem. 246:2271-2277. toxin-binding capacity. G,.a where the terminal 8. Holmgren, J., A. Andersson, G. Wallerstrom, and 0. Gal is linked to a NAN had approximately Ouchterlony. 1972. Experimental studies on cholera 1,000-fold-lower affinity tor toxin than G,1 and immunization. II. Evidence for protective antitoxic immunity mediated by serum antibodies as well as with G\12, where this Gal is absent, no binding of local antibodies. Infect. Immunity 5:662-667. toxin was demonstrable. The importance of the ,J.. I l,innroth, and I. Svennerholm. 197:3. NAN linked to the intemal Gal in G,NI is 8a. Holmgren, Fixation and inactivation of' cholera toxin by G,,, indicated by the observation that GA1, i.e., G.1, ganaliosides. Scand. J. Infect. Dis. 5:77-78. devoid of this NAN, had an affinity for toxin 9. Laurell, C.-B. 1972. Electroimmuno assay. Scand. J. Clin. Lab. Invest. 124 (Suppl. 29):21-39. about 1,000-fold lower than that of G,,1. 10. Li, Y.-T., and S.-C. Li. 1971. Anomeric contiguration of' It may be noteworthy that V. cholerae bactegalactose residues in ceramide trihexoside. J. Biol. ria produce sialidase, which can convert all the Chem. 246:3769-:3771. major brain gangliosides to G,,,, the NAN of l(a. Li. Y.-T.. et al. 197:3. J. Hiol. Chenm. 248:26:34-26:36. which is resistant to the enzyme (15). This 11. LoSpalluto, J. J., and R. A. Finkelstein. 1972. Chemical and physical properties of cholera exo-enterotoxin effect was confirmed in the present study with (choleragen) and its spontaneously formed toxoid GI)la, which converted to G.,l and thereby (choleragenoid). Biochem. Biophys. Acta 256:158-166. increased accordingly in toxin-fixing capacity. 12. Ouchterlony, 0. 1962. Diffusion-in-gel methods for immunological analysis II. Progr. Allergy 6:30-154. It is possible that this sialidase production may J. W., J. J. LoSpalluto, and R. A. Finkelstein. facilitate enterotoxicity by uncovering more 1.3. Peterson, 1972. Localization of' cholera toxin in vivo. .J. Infect.
mucosal receptors by such conversion to GNI l or substance containing the "active" region of this ganglioside. However, the fact that isolated toxin is effectively diarrhoeogenic indicates that bacterial sialidase is not an essential prerequia
site for toxin activity. ACKNOWLEDGMENTS We are grateful to Ouchterlony for support and for help with the interpretation of' the double-dit'fusion analyses presented in Fig. 4, and to J.-E. Mansson for assistance with the sialidase hydrolysis experiments. We thank G. Wallerstrom and A. Andersson for excellent technical assist0.
ance.
Dis. 126:617-628. 14. Pierce, N. F., W. B. Greenough III, and C. C. J. Carpenter. 1971. Vibrio cholerae enterotoxin and its mode of action. Bacteriol. Rev. 35:1-1.3. 15. Svennerholm, L. 1963. Chromatographic separation of human brain gangliosides. J. Neurochem. 10:613-623. 16. Svennerholm, L. 1972. Gangliosides, isolation, p. 464-474. In R. L. Whistler and J. N. BeMiller (ed.), Methods in carbohydrate chemistry, vol. 6, Academic Press Inc., New York. 17. van Heyningen, W. E., C. C. J. Carpenter, N. F. Pierce, and W. B. Greenough III. 1971. Deactivation of'cholera toxin by ganglioside. J. Infect. Dis. 124:415-418. 18. Wadsworth, C. 1957. A slide microtechnique tor the analysis of immune precipitates in gel. Int. Arch. Allergy 10:355-360.
