Action of Arthrobacter ureafaciens Sialidase on Sialoglycolipid Substrates MODE OF ACTION AND HIGHLY SPECIFIC RECOGNITION OF THE OLIGOSACCHARIDE MOIETY OF GANGLIOSIDE

A new bacterial sialidase (N-acetylneuraminate gly-cohydrolase, isolated from the culture fil- trate of Arthrobacter ureafaciens was characterized in detail with respect to its action on sialoglycolipids. Strong electrolytes had a reversible inhibitory effect on the action of the enzyme on brain gangliosides in accordance with Debye-Htickel effect of ionic environ- ment on ionic activity, and resulted in an acidic shift and a broadening of the pH optimum. Both ionic and non-ionic detergents markedly enhanced the enzymic activity on the gangliosides, and caused an acidic shift on the pH optimum of this enzyme.

with Debye-Htickel effect of ionic environment on ionic activity, and resulted in an acidic shift and a broadening of the pH optimum. Both ionic and non-ionic detergents markedly enhanced the enzymic activity on the gangliosides, and caused an acidic shift on the pH optimum of this enzyme. Sulfhydryl groups seemed to be involved in its active site. This enzyme had a highly specific action on sialidase-resistant ganglioside Gl)ll, showing about loo-fold higher activity on GM1 than Clostridium perfringens sialidase, the only sialidase so far reported to cleave the lipid substrate in the presence of bile salts. In the absence of detergents, the activity of A. ureafaciens sialidase on GNll was very low. Ganglioside Gnrl in either the monomeric or micellar form was hydrolyzed to asialo-GM1 by A. ureafaciens sialidase most efficiently in the presence of sodium cholate of about three times the Gnrl molar concentration. The presence of detergents increased both the K,,, and V,, values for ganglioside Gnrl. The oligosaccharide prepared from GD~~ by ozonolysis was cleaved well by this sialidase in the absence of detergents, and no detergent was found to affect the hydrolysis. The K,,, value for the sugar substrate was about two orders of magnitude greater than that for the corresponding lipid substrate.
It is suggested that the hydrophobic ceramide moiety increases affinity of the lipid substrate to the enzyme, but inhibits hydrolysis of the substrate, possibly due to its hydrophobic interaction with hydrophobic portions of the enzyme molecule (resulting in lower K,,, and V,, for lipid substrates).
This inhibition may be released by detergent due to formation of mixed micelles of sialoglycolipid and detergent molecules. It is also indicated that recognition of the specific saccharide structure of Gnrl by individual sialidases is essential for release of the resistant sialyl residue, and that A. ureafaciens * This work was supported in part by a grant from the Ministry of Education, Science and Culture, Japan. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
0 Present address, The Third Department of Internal Medicine, Faculty of Medicine, University of Tokyo, Bunkyo-ku, Tokyo-113, Japan. sialidase seemed to have an isoenzymic or oligomeric structure.
Sialidases of microbial origin have been known for over 30 years (l-3), and they are frequently employed for structural studies on various naturally occurring sialo-compounds, such as sialoglycoproteins, sialoglycolipids, and oligo-and polysaccharides. Recently, their use in biochemical and immunological researches on physiological, polymorphic functions of cell membrane systems (4-13) has increased, since it has been found that they affect the structure and net charge of constituents containing sialic acid by removing terminal sialyl residues, principally resulting in alteration of intermolecular and intercellular forces (14). Among the known microbial sialidases, those from Vibrio cholerae (15-17) and Clostridium perfringens (18-23) have been most extensively characterized in terms of enzymological and biological properties. Recently, a new bacterial sialidase was isolated in a highly purified state from the culture filtrate of a non-pathogenic bacterium, Arthrobacter ureufaciens (24, 25). This enzyme preparation was reported to liberate sialic acid from substrates containing sialyl residues with (w,2-3, a,2-6, and a,2-8 linkages, and to show no activities of contaminating enzymes such as proteases, sialic acid aldolase or other glycosidases (25). In preliminary studies (26), we found that in the presence of detergents this enzyme efficiently hydrolyzed monosialo-ganglioside Ghll,' which, like GMZ, has usually been found to be resistant to various sialidases of viral, bacterial, and mammalian origin (18, [27][28][29][30][31][32][33][34]. The only exceptional reports in this respect are that the resistant sialyl residues of gangliosides GM1 and Grvlz could be cleaved enzymatically in the presence of bile salts (20, 21) and that a lysosomal sialidase of mammalian origin could hydrolyze ganglioside GM*, although this enzyme seems to be very labile, and it is unknown whether it can hydrolyze Ghll (35,36). This paper reports the precise mechanism of enzymatic hydrolysis of sialoglycolipid substrates (purified brain gangliosides) by A. ureczfaciens  showed that A. ureafaciens sialidase cleaved the sialyl residue of ganglioside Ghll in the presence of detergent (Fig. 2). As shown previously in our preliminary report (26), the production of asialo-GM1 by the enzyme was enhanced most effectively by sodium cholate. The kinetic properties of the enzyme on GM-~ and oligosaccharide from GM~ were compared with those of other bacterial sialidases, and especially C. perfiingens sialidase. Fig. 8 shows the effects of various detergents on the initial reaction velocities of the enzymes from A. ureafaciens and C. perfringens.
A. ureafaciens sialidase was markedly stimulated by addition of bile salts, sodium cholate being the most effective. Nonionic detergents also stimulated this sialidase. With equal units of enzyme in terms of activity on sialyl-lactose, the specific activity of A. ureafaciens sialidase on ganglioside Ghll was about 100 times that of C.perf~-ingens sialidase in the presence of sodium cholate (  ureafaciens sialidase was about 4.8 with an optimal concentration of sodium cholate (Fig. 9B). This value was very similar to that of the enzyme for hydrolysis of purified brain gangliosides in the absence of detergents. Addition of a larger amount of sodium cholate caused some inhibition and a slight acidic shift of the pH optimum ( Fig. 9B), possibly due to increase in the concentration of Na+ ion in the reaction mixture with increase in the concentration of sodium cholate. Fig. 10 shows that the cleavage of ganglioside GM~ by A. ureafaciens sialidase was greatly activated by detergents and especially bile salts, and that the degree of activation by detergents depended on their chemical structure and concentration, sodium cholate being most effective at about 1 mg/ml (2.32 mu) with about 0.9 mM Grvrl as substrate. Sodium deoxycholate was similarly effective in activating the sialidase. Detergents caused much less activation of the C. perfringens enzyme than of A. ureafaciens sialidase (Fig. 1OB).
Next, the mechanism of activation with sodium cholate was investigated in more detail. Fig. 11A    with a maximum at a certain concentration of lipid substrate (Fig. 11A). These results indicate that enzymatic release of sialyl residue from GM1 was affected by the relation -between the concentrations of detergent and GM1. Thus, as shown in Fig. llB, maximum activation of A. ureafaciens sialidase by sodium cholate was obtained at a molar ratio of detergent to Ghll of about 3.0.
As shown in Table III Action of A. ureafaciens Sialidase on Sialoglycolipids concentration was above or below the CMC (0.85 X 10m4 M) (53), the activity of A. ureafuciens sialidase on GM~ was very low in the absence of detergents, although it could be reproducibly measured by the spectrofluorometric method of Hammond et al. (47). On addition of the optimal concentration of sodium cholate, the enzymatic activity was much higher, and the activation index was similar whether the lipid substrate GM, was in a monomeric (lower than the CMC) or mice&r (higher than the CMC) form (Table III). These results indicate that, in the absence of bile salt, there was not much difference in the susceptibilities of monomeric and micellar forms of GM1 to the action of A. ureafaciens sialidase. In contrast, Rauvala reported that, below the CMC, ganglioside GM1 became more susceptible to C. perfiingens sialidase even without added bile salts (23). The present results also seem to be incompatible with the suggestion of Wenger and Wardell (21) that detergent may cause activation by converting micellar substrate to more available forms, possibly such as smaller micelles. Fig. 12A shows the initial velocity-substrate concentration relationship for A. ureafaciens sialidase with ganglioside GM~ as substrate in the presence of a fLved concentration of sodium cholate. Higher concentrations of Gs.I~ were apparently inhibitory, but this may be partly because at higher substrate concentrations, the detergent concentration was suboptimal. The apparent K,,, and V,,, values were calculated as 7.14 X 10m4 M and 5.19 pmol/min/mg of protein, respectively ( Fig.  12A and Table V). Fig. 12B shows Lineweaver-Burk plots for the relationship between the initial reaction velocity and the GM1 concentration over a wide range of concentrations from below to above the CMC, in the absence of detergents. There is a barely detectable discontinuity of the hyperbola as the concentration of substrate passes through the CMC. This discontinuity is more evident in derivative functions, such as reciprocal substrate-velocity plots. Ganglioside micelles are very stable (60), and their size distribution is uniform, so that the effect of concentration above the CMC on the reaction velocity may be treated in terms of diffusion-related kinetics. Procedures" except that various amounts of substrate were incubated with 2.5 milliunits of the enzyme K, values were derived in terms of simple molar concentration, irrespective of the actual number of diffusing particles per unit volume. On this basis, the apparent K,,, for the lipid substrate above the CMC was 1.04 x 1O-4 M and the K, below the CMC was 1.24 X 10m5 M. There was a clear difference in the kinetic constants for the two substrate forms in the absence of detergents. Taken together, the results indicate that, in the absence of detergents, the enzyme interacts not only with monodisperse ganglioside Ghll but also with sialyl residues located in the surface of GM~ aggregates, and that the effective concentration of lipid substrate in a kinetic sense was apparently altered by the formation of micellar aggregates.
When ganglioside GM~, in which the sialyl residue is located at the nonreducing end of the sugar chain, was used as substrate, the initial velocity of A. ureufaciens sialidase increased as a continuous hyperbolic function of the substrate concentration, irrespective of whether or not the ionic detergent sodium cholate was present (Fig. 13A). Addition of ganglioside Ghll to the reaction mixture for Gw~ hydrolysis in the absence of detergent strongly inhibited the initial velocity, as shown in Fig. 13A. Ghll had an unusual influence on the typical Michaelis-Menten hyperbolic saturation curve for GMS: it caused an initial sigmoid region (Fig. 13A) somewhat different from the classical competitive kinetic form. In the Hill plot (Fig. 13B), there are two slopes, i.e. two interaction coefficients of 1.42 and 0.80, with the break between the two at about twice the CMC of gangliosides, suggesting a mixture of positive and negative cooperativity in the interaction of oligomeric enzymes on GM~ in the presence of Ghll.
For analysis of the interaction between the enzyme molecule and the hydrophilic moiety of ganglioside Ghll, the oligosaccharide was prepared from the sialoglycosphingolipid by ozonolysis (39). As shown in Table IV, the sugar substrate was found to be highly susceptible to A. ureafaciens sialidase in the absence of detergent.
It was much less susceptible to hydrolysis by C. perfringens sialidase. Detergents had no significant effect on the hydrolytic activity of either enzyme on the oligosaccharide, in contrast to their effect in activating GM, hydrolysis. This oligosaccharide, like Ghll, was not hydrolyzed by sialidase from V. cholerae or Streptococcus K 6646 either in the presence or absence of detergents ( Fig. 2 and Table IV). These findings indicate that the activities of the two sialidases on Gr,.n reflect their activities on its oligosaccharide moiety. The initial velocity-substrate concentration relationship of A. ureafaciens sialidase on the oligosaccharide from ganglioside GM1 was measured (Fig. 14). The apparent Km and Vm v alues were calculated from Lineweaver-Burk plots to be 2.58 x 10m2 M and 2.35 pmol/min/mg of protein, respectively.
The kinetic parameters of A. ureafaciens sialidase are summarized in Table V. The K,,, and V,,,,, values varied with the type of substrate in terms of molecular structure and physicochemical state. The apparent Km and V,,, values for purified brain gangliosides (ganglioside mixture) were very similar to those for ganglioside GM~. In the absence of detergents, the V,,, values for ganglioside GM, in either the monomeric or micellar form was very small. The apparent K,,, for this sialoglycolipid without detergents was significantly smaller than that with detergents. In general, addition of detergent apparently increased both the Km and V,,, values for all of the complex lipid substrates, but had no effect on the values for water-soluble sugar substrates. The V,,, value for ganglioside GM~ was much lower than that for sialyllactose, whereas the K,,, values for the two substrates were very similar. Since the sugar moieties of GM3 and sialyllactose have the same structure, these results suggest that the hydrophobic ceramide moiety may restrict the cleavage of sialyl residues by some hydrophobic interactions. This restriction may be released by addition of detergents, because the V,,,,x value for GM~ was greatly increased by sodium cholate. It should be noted that the Km value for the oligosaccharide from ganglioside GM~ was about 2 orders of magnitude greater than that for the corresponding lipid substrate, irrespective of whether the latter was in a monomeric or micellar form. This also suggests that the  lipophihc ceramide moiety may be important in enhancing the affinity of the substrate for the enzyme, while it may restrict the mutual mobilities of the substrate and enzyme in the absence of detergent. DISCUSSION This work was on the enzymatic actions of a new bacterial sialidase on sialoglycolipid substrates, purified from the culture filtrate of a nonpathogenic bacterium, Arthrobacter ureafaciens (24,25). As reported previously (26), the most striking enzymatic property of this bacterial sialidase is its highly specific ability to cleave sialyl residues of ganglioside Ghll, which exists in aqueous media as micelles composed of about 225 monomers at above the CMC, and as monomeric, disperse forms below the CMC (53), and which has long been believed to be resistant to sialidases of various origins (18, [27][28][29][30][31][32][33][34]. However, as found by TLC ( Fig. 2A), this enzyme showed only slight activity on fucosyl-ganglioside G111~, in which the fucosyl residue is linked a-glycosidically to the terminal galactosyl residue of Ghll (51, 52). This finding suggests that A. ureafaciens sialidase may recognize the neighboring galactosyl-N-acetylgalactosaminyl residue, resulting in recognition of the whole structure of the saccharide moiety of ganglioside GM1, and that the terminal fucosyl residue of fucosyl-ganglioside Ghll may cause steric hindrance of the cleavage of the sialyl residue adjacent to N-acetylgalactosamine located on the galactosyl residue closest to the lipophilic, ceramide residue.
The low pK of the carboxyl group of sialic acid, which must have a strong negative charge in the physiological pH range, implies that the sialidase is strongly influenced by the ionic environment, whether the sialoglyco-compounds are watersoluble or whether they form part of superstructural aggregates. The present results indicate that the ionic strength of the medium also affects the action of A. ureafaciens sialidase on sialyl lipids (purified brain gangliosides) without markedly affecting the activity of the catalytic center, since the watersoluble sialyl compound sialyl-lactose was readily hydrolyzed under ionic conditions which minimize the availability of the lipid-bound sialyl substrate, as shown previously with the sialidases from V. cholerae (17)

Action of A. ureafaciens
Sialidase on Sialoglycolipids by the appearance of an acidic shift of the pH optimum, and then by a marked decrease of the enzymatic activity over the whole pH range with a broadening of the pH optimum, as observed previously with C. perfringens sialidase (22). The ionic environment had more influence on the new bacterial sialidase than on the C. perfringens enzyme ( Fig. 3 and Table  I). The size of aggregates of gangliosides changes with increase in concentration of electrolytes and these changes have been measured in the presence of mono-and divalent cation (60, 61). However, the inhibitory effects of strong electrolytes are probably not chiefly due to salt-induced changes in the degree of aggregation of the ganglioside substrate, because in this work the effect of the ionic environment was studied at substrate concentrations above the CMC to avoid the effect of possible phase transition of the lipid substrate from a monomeric to mice&r form. Thus, as suggested by Lipovac et al. (17) for V. cholerae sialidase, strong electrolytes may inhibit the action of A. ureafaciens sialidase on lipid substrates by screening some like-charge interaction on the enzyme that is necessary for maintenance of a conformation of the enzyme in which the catalytic center is available to sialyl residues of bulky lipid molecules and aggregates.
The sialidase-catalyzed reactions described here involve a water-soluble enzyme amd a lipid-soluble substrate. As expected, various detergents, whether ionic or non-ionic, stimulated the activity of A. ureufuciens sialidase on purified brain gangliosides. Detergents caused less activation of C. perfringens sialidase, in contrast to the report of Wenger and Wardell (21), although their effects were found to be dose-dependent.
More striking than the difference between A. ureafaciens and C. perfringens sialidase in their activations by detergents, was the difference in their initial velocity-pH relationships (Fig. 7): A. ureczfuciens sialidase showed an acidic shift of the pH optimum with both ionic and non-ionic detergents, whereas C. perfringens sialidase was activated by non-ionic detergents, but not ionic ones, only in the acidic pH regions and was slightly inhibited in the neutral pH range, resulting in a marked broadening of the initial velocity-pH curve. The stimulatory effects of detergents on the reaction of A. ureafaciens sialidase with purified brain gangliosides was presumably due to increase in the solubility of lipid in water or an alteration of the hydrophobic structure of the enzyme, both resulting in a marked increase of the V,,, value (Fig. 5). The acidic shift of the pH optimum by detergents was not due simply to the presence of a strong electrolyte, i.e., sodium ion, since it was also observed with non-ionic detergents, such as Triton X-100 and Tween 80. For fuller understanding of the effects of detergents, further studies are required on the hydrophobic interactions of the lipid substrate and the enzyme molecule.
Sulfhydryl groups are involved at the active site of A. ureafaciens sialidase, because its activity was inhibited by pCMPS (Table II). It is uncertain why the sialidase was strongly inhibited by pCMPS but not appreciably by other sulfhydryl reagents Similar results have been obtained on nucleoside tetraphosphate hydrolase (62): this enzyme is strongly inhibited by p-chloromercuribenzoate, only moderately by N-ethylmaleimide, and not at all by iodoacetate. The sialidase does not seem to have a serine residue in the active site, because its activity was not inhibited by DFP.
Compounds in which N-acetylgalactosamine is linked glycosidically to position 4 of a galactose residue which bears sialic acid at position 3, such as GM~ and Ghll gangliosides, are very resistant to sialidases of various origins (l&27-34). This resistance has been ascribed to steric hindrance of the sialyl group by the substituent on the axial hydroxyl group at position 4 of the galactose residue (63). Lipovac and Rosenberg (64) proposed the existence of competitive inhibition between sialyl and N-acetylgalactosaminyl residues. However, recently, Wenger and Wardell (21) reported that this resistant sialyl residue could be hydrolyzed by C. perfringens sialidase in the presence of bile salt. As we reported previously (26), the new bacterial sialidase from A. ureafaciens hydrolyzes the resistant sialyl residue of ganglioside Ghll in the presence of detergents without prior hydrolysis of the terminal galactosyl-N-acetylgalactosaminyl residue. A. ureufuciens sialidase showed much greater activity than the C. perfringens enzyme for hydrolysis of GM~ to asialo-GM1 in the presence of detergents (Fig. 8). Of the detergents tested, bile salts and especially sodium cholate were effective in increasing the activity of the former enzyme on GM~ (Figs. 8 and 10). Since bile salts have a physiological role in lipid absorption from the gut, their physical properties have been extensively studied (65-68). The hydroxyl groups of bile salts are all on one side of the rigid cyclopentheno-phenanthrene ring and the terminal ionic group is situated at the end of a short flexible branched aliphatic chain. Thus, they are thought to have a bean-shaped molecule with a polar and an apolar face, and to form small aggregates (from dimers to octamers) in which the molecules lie back to back in water at above a critical concentration (69). At higher counter ion concentrations, larger aggregates may generally form concomitantly with decrease in the CMC, but both the CMC and the aggregation number of trihydroxy bile salts (sodium cholate) have been reported to be resistant to the counter ion concentration (66), having values of 1.3 to 1.5 X IO-' M, and 2 (dimer) to 4 (tetramer), respectively. Maximal activation by sodium cholate of A. ureafaciens sialidase with ganglioside GM~ as the substrate was demonstrated when the molar ratio of the ionic detergent to the lipid substrate was approximately 3 (Fig. 11B). Taken together, these results suggest that single aggregates of sodium cholate (from dimers to tetramers) contain one molecule of ganglioside GM1, and that the sialyl residues in these mixed aggregates are the most susceptible to A. ureafaciens sialidase.
The new bacterial sialidase was active on ganglioside Ghn even in the absence of detergent (Table III and Fig. 12B). Very recently, Rauvala (23) reported that GM~ below the CMC became susceptible to C. perfringens sialidase even without addition of bile salt. However, in our experiment, there was not much difference in the susceptibilities of monomeric and micellar forms of Gllli to the action of A. ureafaciens sialidase in the absence of bile salt, and the detergent increased the activities on both forms to almost equal degrees (Table III). These results indicate that detergents may not be essential for release of the sialyl residue from ganglioside GM~, but may play an auxiliary role in enhancing the enzymatic activity. In contrast, C. perfringens sialidase is reported to show an absolute requirement for bile salts in cleavage of ganglioside Ghll (21). Our results also indicate that the cleavages of the resistant sialyl residue from the oligosaccharide prepared from Ghll by the bacterial sialidases are well correlated with their activities on Ghll, and that detergents have no effect on hydrolysis of the oligosaccharide (Table IV). These findings suggest that the recognitions of the specific saccharide structure of ganglioside GM, by the respective enzymes are of primary importance for hydrolysis of the sialyl residue of GMl. The hydrophobic ceramide moiety of GM~ seems to restrict the enzymatic hydrolysis (the V,,, value for Ghll is very low in the absence of detergent, while the V,,, value for the oligosaccharide, des-GMurl, is high). Detergents may release this restriction possibly by modification of the interaction between the hydrophobic portion of GM~ and the enzyme molecule (the V,,,,, value for GM1 is high in the presence of detergents) (Table V).
Ganglioside GM~ had a marked effect on the enzymatic cleavage of ganglioside GM3 in the absence of detergents (Fig.  13A). With A. ureafaciens sialidase, ganglioside GM~ gave a typical hyperbolic Michaelis-Menten curve, but it gave an unusual curve in the presence of ganglioside GUI. The initial sigmoid region in the presence of Ghll shown in Fig. 13A indicates some degree of deviation from the classical competitive kinetic form. The inhibition of the action of A. ureafaciens sialidase on ganglioside GM~ by ganglioside GMI may involve some type of subunit interaction.
The Hill plot (Fig.  16) had two slopes, with a break between them at about twice the CMC of gangliosides (53), again suggesting that some oligomeric enzyme structures exhibit a mixture of negative and positive cooperativity.
This phenomenon seems to be compatible with the finding by gel filtration and polyacrylamide gel electrophoresis that the highly purified preparation of A. ureafaciens sialidase shows molecular heterogeneity, the different enzyme fractions exhibiting different substrate specificities (Fig. 1).