Eur. J. Biochem. 218, 645-649 (1993) 0 FEBS 1993

Kinetics of endoglycoceramidase action toward cell-surface glycosphingolipids of erythrocytes Makoto ITO, Yuko IKEGAMI and Tatuya YAMAGATA Laboratory of Glycoconjugate Research, Mitsubishi Kasei Institute of Life Sciences, Tokyo, Japan

(Received June 4/August 30, 1993) - EJB 93 1318/4

As shown in the preceding paper [Ito, M., Ikegami, Y., Tai, T. & Yamagata, T. (1993) Eur: J . Biochem. 218, 637-6431, endoglycoceramidase (EGCase; EC.3.2.1.123) was found to hydrolyze the cell-surface glycosphingolipids (GSLs) of erythrocytes without any damage to other cell mem- brane components. This paper represents the kinetics of EGCase action toward cell-surface GSLs of erythrocytes. Without activator or detergents, cell-surface GSLs were found to be hydrolyzed by EGCase I1 very slowly at pH 7.0. The initial reaction velocity of EGCase I1 under this condition was 0.038 pmol . min-' . mu-' for horse erythrocyte cell-surface GM3 and 0.032 pmol . min-' . mu- ' for guinea pig erythrocyte cell-surface Gg,Cer. The addition of activator protein (60 pM), which stimulates EGCase I1 in the absence of detergents, increased the initial reaction velocity of the enzyme 61 6-fold for cell-surface GM3 and 468-fold for Gg,Cer, while no increased hemolysis was observed with the addition of activator. However, even in the presence of the activator, the cell- surface GSLs were very resistant to hydrolysis by EGCase I1 compared to GSL vesicles (or micelles) under same conditions. In contrast to the activator, Triton X-100 (0.4%, masdvol.) not only stim- ulated the enzyme activity but also solubilized erythrocyte GSLs into detergent micelles, inducing further increment of the enzyme reaction velocity. The apparent K,,, and V,,,, values of EGCase I1 were calculated from the Lineweaver-Burk plot as 47 pM and 35 pmol min-' mu- ' for horse eryth- rocyte cell-surface GM3 and 44 pM and 27 pmol . min-' . mu-' for guinea pig erythrocyte cell- surface Gg,Cer, at pH 7.0 in the presence of activator at a concentration of 60 pM.

Endoglycoceramidase (EGCase) is the enzyme which cleaves the linkage between oligosaccharides and ceramides of various glycosphingolipids (GSLs) [ l , 21. Recently, it was extensively purified and found to consist of three molecular species each with its own specificity (EGCases I, I1 and 111) [3] . The preceding paper in this journal [4] clearly demon- strates that EGCase I1 can specifically hydrolyze the intact erythrocyte cell-surface GSLs with the assistance of its acti- vator protein [5 , 61 without any damage to other cell mem- brane components. The purpose of this paper is to clarify the kinetics of EGCase I1 toward both neutral and acidic GSLs on the cell surface using erythrocytes as a model because no new synthesis of GSLs was expected. This may provide fundamental information regarding the specificity of this enzyme, as well as when using the enzyme to analyze GSL functions.

Correspondence to M. Ito, Laboratory of Marine Biochemistry, Faculty of Agriculture, Kyushu University, 6-10-1 Hakozaki, Higashi-ku, Fukuoka, Japan, 812

Abbreviations. GSLs, glycosphingolipids; EGCase, endogly- coceramidase; HPAE-PAD, high-performance anion-exchange chro- matography with pulsed amperometric detection; NaCYP,, phosphate- buffered saline; NeuCc, N-glycolylneuraminic acid; NeuAc, N-ace- tylneuraminic acid ; LcCer, Galpl,4Glcpl ,l'Cer; Gg,Cer, GalNAcp- 1,4Gal~1,4Glc~I ,1 'Cer ; Gg,Cer, Gal~l,3GalNAc~1,4Gal~l,4Glc~- 1 ,l'Cer. Abbreviations for gangliosides follow the nomenclature sys- tem of Svennerholm [12].

Enzyme. Endoglycoceramidase (EC 3.2.1.1 23).

MATERIALS AND METHODS

Materials

P-Hexosaminidase from Penicillium sp. was a generous gift from Dr K. Yamamoto (Kyoto University, Japan); GM3(NeuGc) and 4-O-acetyl-GM3(NeuGc) were kindly provided by Dr H. Higashi (our laboratory) and Gg,Cer by Dr Y. Hirabayashi (Riken, Japan). NeuAc-lactose and siali- dase (neuraminidase type I1 from Vibrio cholera) were ob- tained from Sigma, GM3(NeuAc) from Bachem, Triton X- 100 from Pierce, precoated silica gel 60 HPTLC plates from Merck, fresh difibrinated horse and guinea pig blood from Nippon Bio-Test Laboratories (Japan).

EGCase I1 and activator assay

EGCase I1 activity was assayed using purified Gg,Cer as the substrate in the presence of Triton X-100 as described [3] . One milliunit (mu) enzyme was defined as that amount capable of catalyzing the hydrolysis of 1 nmol substrate/min. In this paper, the amount of activator in the reaction mixture is indicated by the concentration of the purified activator pro- tein. Isolation of EGCase I1 [3] and activator I1 [5] from Rhodococcus sp., and preparation of 27.9-kDa activator [6] were described in previous quoted papers. The 27.9-kDa acti- vator was used instead of the native form for all experiments in this study, and is referred to simply as 'activator'.

646

Treatment of erythrocytes with EGCase I1 Fresh defibrinated blood samples from horse and guinea

pig were centrifuged at 2000Xg for 10 min, and the plasma and buffy coat were removed. Erythrocytes were collected and washed three times with 20mM potassium phosphate pH 7.0 made isotonic with 0.85% NaCl (phosphate-buffered saline, NaClP,). A reaction mixture consisting of a 10% (by vol.) erythrocyte suspension and an appropriate amount of EGCase I1 and activator protein in 100 pl NaClP, pH 7.0 was incubated at 37°C for the time indicated. In control ex- periments, EGCase I1 or activator or both were omitted from the reaction mixture. In some experiments, Triton X-100 at a concentration of 0.4% (mass/vol.) was used for EGCase stimulation in place of the activator.

Determination of GSL hydrolysis of erythrocytes

Determination of GSL hydrolysis of erythrocytes was conducted using two different methods. Method I is the de- termination of oligosaccharides released from cell-surface GSLs by high-performance anion-exchange chromatography with pulsed amperometric detection (HPAE-PAD, Dionex) [7]. Method I1 is the determination by high-performance thin- layer chromatography (HPTLC) of GSLs remaining in cells after enzyme treatment. Following treatment of intact eryth- rocytes with enzyme, the reaction mixture was centrifuged at 2000Xg for 5 min. Oligosaccharides of the supernatant and GSLs of cells were separately analyzed by methods I and 11, respectively.

Method I. 70 pl of the supernatant was withdrawn, and 10 pl internal standard (100 nmol/ml, NeuAc-lactose for horse erythrocytes ; cellotriose for guinea pig erythrocytes) was added. The mixture was partitioned with 500 pl chloro- form/methanol (2: 1, by vol.) and the upper layer was dried under a stream of nitrogen gas, dissolved in 25 pl distilled water, and analyzed for oligosaccharides by HPAE-PAD. For the gradient, eluent A was 100 mM NaOH, while eluent B contained 100 mM NaOH with 0.5 M sodium acetate. The gradient for separating NeuGc-lactose was a linear increase in eluent B (10-75% in 20min). The retention time of NeuGc-lactose was 14.3 ? 0.2 min and could clearly be sepa- rated from the internal standard of NeuAc-lactose whose re- tention time was 9.0 ? 0.2 min on a column (4.6X250 mm) of Dionex Carbopac PA-1 anion-exchange resin with a flow rate of 1 ml/min at room temperature. For determination of Gg,-oligosaccharides from guinea pig erythrocytes, the gradient was a linear increase in eluent B (0-10% in 20 min). The retention time of Gg,-oligosaccharides was 8.9 ? 0.5 min, while the internal standard cellotriose was eluted at 17.1 20.5 min. Detection was carried out with PAD, using a gold working electrode and triple-pulse amperometry. The chromatographic data were plotted using a Shimadzu Chromatopac C-R4A integrator (Shimadzu, Japan).

Method 11. GSLs were extracted from cells by sonication for 30 rnin with 500 pl chlorofodpropan-2-ol/water (7/11/ 2, by vol.). Cell extracts were centrifuged at 16000 rpm for 5 min and the supernatant was withdrawn. Cells were ex- tracted with the same solvent three times. Supernatants from each extraction were combined and dried under N, gas. GSLs were analyzed by HPTLC using developing solvent system I (chloroform/methanol/water, 65/25/4, by vol.) for neutral GSLs, and system I1 (chloroform/methanol/0.25 % KCl, 5/4/ 1, by vol.) for acidic GSLs. GSLs were visualized by spray- ing the HPTLC plates either with orcinol/H,SO, reagent for

neutral GSLs or resorcinol/HCl reagent for acidic GSLs, and were determined by a Shimadzu CS-9000 chromatoscanner with the reflectance mode set at 540 nm for orcinol/HISO, staining and at 580 nm for resorcinol/HCl staining. The ex- tent of hydrolysis of GSLs by the enzyme was calculated from the decrease in GSLs after treatment of erythrocytes with the enzyme.

Measurement of hemolysis

Following treatment of intact erythrocytes with the en- zyme, the reaction mixture was centrifuged at 2000Xg for 5 min; 90 pl of the supernatant was withdrawn, transferred to a 96-well microplate, and measured spectrophotometri- cally for hemoglobin at 550 nm using a microplate photome- ter MTP-22 (Corona Electric Co., Japan). Total hemolysis was achieved after adding Triton X-100 (0.4%, mass/vol., at final concentration) to the reaction mixture. The degree of hemolysis during incubation was expressed as the percentage of total hemolysis.

Determination of K , and V,,, of EGCase I1 for cell-surface GSLs

Activity toward horse erythrocyte GM3 and guinea pig erythrocyte Gg,Cer was assayed with 1 mU EGCase I1 for 30 min at 37°C at various cell concentrations (2.5, 5 , 7.5, 10, 15, 20%, by vol., in total 100 p1 NaCl/Pi pH 7.0) in the pres- ence of 6 nmol activator. The GM3 and Gg,Cer concentra- tions of a 10% cell suspension are 94 pM and 66 pM, respec- tively. The hydrolysis of cell-surface GSLs was determined by method I and apparent K,,, and V,,,;,, of EGCase I1 were calculated from the Lineweaver-Burk plot.

RESULTS AND DISCUSSION Oligosaccharides released from the cell surfaces of eryth-

rocytes by the action of EGCase I1 were determined with HPAE-PAD (method I). As shown in Fig. 1, N-glycolylneur- aminic-acid-containing GM3- (simply indicated as GM3 in this paper) and Gg,-oligosaccharides were identified as the major products released by the action of EGCase I1 from horse (Fig. lA, b) and guinea pig (Fig. lB, c) intact erythro- cytes, respectively. GM3(NeuAc) was not detected in horse erythrocytes at all and thus N-acetylneuraminic-acid-contain- ing sialyllactose (NeuAc-lactose) was used as the internal standard for method I (Fig. lA, a). Only trace amounts of glucosylceramide and LcCer were present in both erythro- cytes, in addition to the respective major GSLs (data not shown). Since 4-0-acetyl-GM3-oligosaccharide (4-0-acetyi- NeuGc-lactose) eluted with the same retention time as that of NeuGc-lactose on HPAE-PAD, 4-0-acetyl groups may possibly be released under alkaline conditions (pH 12) during chromatography. These two molecular species of NeuGc-lac- tose can be separated from each other by HPTLC using de- veloping solvent (n-butanollacetic acidlwater, 2/1/1, by vol.) 181, and intact GM3 were also found to be separated from 4- 0-acetyl derivatives by TLC using developing solvent (chlo- roform/methanol/0.02% CaCl,, 80/35/5, by vol.) 141. 4-0- Acetyl derivatives constituted about 40% of total GM3 of horse erythrocytes used in this study. Treatment of GM3- oligosaccharides from horse erythrocytes with Vihrio siali- dase, which is inactive towards 4-0-acetyl NeuGc-lactose, reduced peak (b) on HPAE-PAD (Fig. 1A) to about 60% of

647 I I (A) Horse (B) Guinea pig

EGCase I1 EGCase II b d

d f / j Control Control

L J L I

0 10 20 0 10 20 Time (min) Time (min)

Fig. 1. HPAE-PAD showing the oligosaccharides released from erythrocyte cell-surface GSLs by EGCase I1 in the presence of activator. (A) Horse erythrocytes : (a) NeuAca2,3Galpl,4Glc (standard NeuAc-lactose, 1 nmol); (b) NeuGca2,3Galp1,4Glc from cell-surface GM3. (B) Guinea pig erythrocytes : (c) GalNAcp1,4- GalPl,4Glc from cell-surface Gg,Cer; (d) Glc~1,4Glcp1,4Glc (stan- dard cellotriose, 1 nmol). Intact erythrocytes (10 p1 packed cells) were incubated with 2 niU EGCase I1 in 100 pl NaCl/P, pH 7.0 con- taining 6 nmol activator for 1 h (for horse erythrocytes) or 1.5 h (for guinea pig erythrocytes) at 37 “C. The oligosaccharides released from cell-surface GSLs, about I nmol each, were analyzed by HPAE-PAD as described in the text. Neither EGCase nor activator was included in control experiments.

that without sialidase treatment (data not shown). It was also observed that Penicillium /?-hexosaminidase [9] treatment of Gg,-oligosaccharides completely abolished the peak (c) on HPAE-PAD (Fig. 1 B), producing lactose and N-acetylgalac- tosamine (data not shown). These data support the peak assignment of oligosaccharides on HPAE-PAD in this study.

Two different assay methods, one of which determines oligosaccharides released from cell-surface GSLs by HPAE- PAD (method I) and the other the GSL contents of erythro- cytes by HPTLC (method II), have been employed to esti- mate the GSL hydrolysis of erythrocytes by the action of EGCase. As shown in Fig. 2, the extent of hydrolysis of cell- surface GSLs obtained by method I closely matched that by method 11. It is also interesting that the hydrolysis velocity of horse erythrocyte cell-surface GM3 by EGCase I1 was virtually the same as that of 4-0-acetyl GM3 (Fig. 2A). The initial reaction velocity of EGCase I1 towards GM3 micelles in the presence of Triton X-100 was also confirmed to be identical to that towards 4-0-acetyl GM3 and N-acetyl- neuraminic-acid-containing GM3 (data not shown). In con- trast, Vibrio sialidase showed poorer cleavage of NeuCc-con- taining substrates when compared with the corresponding NeuAc-containing compounds [lo] : V,,,, for the hydrolysis of GM3 by Vibrio sialidase was a third that of GM3(NeuAc)

In the presence of activator at a concentration of 60 pM, cell-surface GM3 and Gg,Cer were found to be hydrolyzed time-dependently by I mU EGCase I1 under the physiologi- cal conditions (pH 7.0), as shown in Fig. 3. Virtually no hy-

[lo].

.i 40

U z r

20 t 20 t :/ I v I

“ 2 4 6 8 0- 10 20

EGCase I1 (mu) EGCase I1 (mu)

Fig. 2. Two different methods for determination of GSL hydroly- sis. (A) Hydrolysis of horse erythrocyte cell-surface GM3. (0) GM3 by method I; (0) GM3 by method 11; (0) 4-0-acetyl GM3 by method 11. (B) Hydrolysis of guinea pig erythrocyte cell-surface Gg,Cer. (0) Gg,Cer by method I ; (0) Gg,Cer by method 11. Intact erythrocytes (10 p1 packed cells) were incubated with different amounts of EGCase I1 and activator in 100 p1 NaCW, pH 7.0 for 6 h (horse erythrocytes) or 2 h (guinea pig erythrocytes) at 37°C. The concentration of activator is 1.5 nmol/mU enzyme. 100% hy- drolysis is the hydrolysis of either 9.4 nmol GM3 for horse erythro- cytes or 6.6 nmol Gg,Cer for guinea pig erythrocytes under the con- ditions described above. All results are means of duplicate determin- ations. Details are described in the text.

8.0 I I 3.0 t 1

0 1 .o 2.0 0 1 .o 2.0 Time (h ) Time (h )

Fig.3. Time course for the hydrolysis by EGCase I1 of intact erythrocyte cell-surface GSLs. (A) Hydrolysis of horse erythrocyte GM3 by EGCase 11. (0) With 1 mU EGCase I1 in the presence of activator (60 pM); (0) with 1 mU of EGCase I1 in the presence of Triton X-100 (0.4%, mass/vol.); (U) with 1 mU EGCase I1 in the absence of stimulator. (B) Hydrolysis of guinea pig erythrocyte Gg,Cer by EGCase TI. (0) With 1 mU EGCase I1 in the presence of activator (60 pM); (0) with 1 mU EGCase I1 in the presence of Triton X-200 (0.4%, mass/vol.); (0) with 1 mU EGCase I1 in the absence of stimulator. Oligosaccharides released by enzyme action were determined by method I. All results are means of duplicate determinations. See text for details.

drolysis appeared to occur upon incubating the cells either with the enzyme or activator alone under these conditions (Fig. 3A, B). It should be noted that the hemolysis of both types of erythrocytes following incubation for 120 min with both EGCase I1 and activator was less than 5 % of the total hemolysis. This value corresponds to that for control experi- ments in which neither enzyme nor activator was present. It is thus evident that cell-surface GSLs can be hydrolyzed un- der physiological conditions without affecting cell viability. The addition of Triton X-100 to the reaction mixture further

648

Table 1. Initial reaction velocity of EGCase I1 for erythrocyte GSLs. Intact erythrocytes (10 pI packed cells) were incubated with I mU EGCase I1 and either 6 nmol activator or 400 pg Triton X-100 or no stimulator in 100 p1 NaClP, pH 7.0 at 37°C for 15 min. The oligosaccharides released from GSLs were determined by method I. See text for details. GSLs (9.4 nmol for NeuGc-GM3 or 6.6 nmol for Gg,Cer which correspond to GSL content of 10 pl packed horse and guinea pig erythrocytes, respectively) were incubated with 1 mU EGCase 11 as described above except for incubation time ( 5 min). All results are means of duplicate determinations.

Sample Intact erythrocytes GSL micelles or vesicles

GM, Gg,Cer GM, Gg,Cer (horse) (guinea

Pig)

pmol . min-' . mu-' ~~ -~

EGCase 0.038" 0.032" 18.5 12.3 EGCase

+ activator 23.4 15.0 968 181 EGCase

f Triton X-100 137 37.3 296 50.4 Activator 0 0 0 0

~

* Estimated from the value for the 6-h incubation with 20 mU EGCase 11.

enhanced the reaction velocity of EGCase I1 (Fig. 3A, B), although this detergent completely hemolyzed the erythro- cytes as described below.

When the reaction mixture contained activator at a con- centration of 60 pM, the extent of hydrolysis of horse eryth- rocyte cell-surface GSLs was found to be proportional to the enzyme amount up to 2 mU after a 30-min incubation. The activation of 1 mU EGCase I1 reached a plateau with 40- 60 pM activator when GSL-substrates were present either in the form of micelles or attached to the plastic plate. Thus, the initial reaction velocity of EGCase I1 toward cell-surface GSLs was examined using 1 mU EGCase I1 with activator at a concentration of 60 pM for a 15-min incubation at 37°C. Without activator or detergents, cell-surface GSLs were found to be hydrolyzed by EGCase I1 very slowly at pH 7.0. As shown in Table 1, the initial reaction velocity of EGCase I1 under this condition was 0.038 pmol . min-' . mu- ' for horse erythrocyte cell-surface GM3 and 0,032 pmol . min-' . mu-' for guinea pig erythrocyte cell-surface Gg,Cer. The addition of activator protein at a concentration of 60 pM increased the initial reaction velocity of the enzyme 616- fold for cell-surface GM3 and 468-fold for GgiCer, while no increased hemolysis was observed with the addition of activator. When Triton X-100 was used instead of the activa- tor protein at a concentration of 0.4% (masshol.), the initial reaction velocity of EGCase I1 increased a further 5.9-fold for GM3 and 2.5-fold for Gg3Cer, compared to those with the activator protein (Table 1). Even in the presence of acti- vator, GSLs of intact erythrocytes were quite resistant to hy- drolysis by EGCase I1 compared to GSL micelles (or vesi- cles), while in the presence of Triton X-100 both appeared to exhibit quite similar susceptibility to the enzyme (Table 1). This discrepancy may be due to whether or not the stimulator hemolyzed the erythrocytes. It was found that even at a con- centration of 100 pM the activator neither hemolyzed horse erythrocytes nor solubilized GM3 from cells after a 4-h incu- bation, while Triton X-100 (0.4%, masshol.) completely he-

molyzed cells and more than 90% of GM3 was recovered in the supernatant after the removal of erythrocytes by centrifu- gation. In other words, Triton X-100, but not the activator protein, solubilized virtually all GSLs into detergent micelles in which the enzyme can hydrolyze GSLs much faster than on the cell surface.

The apparent K,,, and V,,,, values were calculated from the Lineweaver-Burk plot as 47 pM and 35 pmol . min-l . mu-' for horse erythrocyte cell-surface GM3 and 44 pM and 27 pmol . min-' . mu-' for guinea pig erythrocyte cell- surface Gg,Cer in the presence of activator at a concentration of 60 pM.

The pH optimum for EGCase I1 was 5.5 in the presence of Triton X-100, while it was 5.0 without the detergent. As previously reported [5], the activator shifts the pH optimum to 5.5 and increases the relative activity of the enzyme at pH 7.0-7.5. The initial velocity of EGCase I1 for cell-sur- face GSLs at pH 7.0 in the presence of activator was found to be 90% for GM3 and 64% for Gg,Cer of those at pH 5.5.

As pointed out by Lampio et al., sialidase treatment of guinea pig erythrocytes increased the ratio of Gg,Cer sensi- tive to galactose oxidase [l l] . We found the same tendency using EGCase 11. Pretreatment of guinea pig erythrocytes with Vibrio sialidase increased the reaction velocity of EGCase I1 toward cell-surface Gg,Cer by 1 .s-fold. Since no gangliosides (sialic-acid-containing GSLs) are present in guinea pig erythrocytes, an increase in the velocity of EG- Case I1 should be possible by removal of sialic acid residues from glycoproteins. One possible explanation for this phe- nomenon is that the removal of sialic acid residues results in the clustering of glycoproteins, so cell-surface Gg,Cer be- comes more accessible to these enzymes.

We thank Dr K. Yamamoto for kindly supplying the P-hexosami- nidase. Thanks are also due to Drs Y. Hirabayashi and H. Higashi for providing GSL samples. This work was supported in part by a Grant-in-Aid for Scientific Research on Priority Areas (052741 06) from the Ministry of Education, Science and Culture of Japan.

REFERENCES 1. Ito, M. & Yamagata, T. (1986) A novel glycosphingolipid-de-

grading enzyme cleaves the linkage between the oligosaccha- ride and ceramide of neutral and acidic glycosphingolipids, J. Biol. Chem. 261, 14278-14282.

2. Ito, M. & Yamagata, T. (1990) Endoglycoceramidase from Rho- dococcus sp. G-74-2, Methods Enzymol. 179, 488-496

3. Ito, M. & Yamagata, T. (1989) Purification and characterization of glycosphingolipid-specific endoglycosidase (endoglycocer- amidase) from a mutant strain of Rhodococcus sp., J . Biol. Chem. 264, 9510-9519.

4. Ito, M., Ikegdmi, Y., Tai, T. & Yamagata, T. (1993) Specific hydrolysis of intact erythrocyte cell-surface glycosphingoli- pids by endoglycoceramidase. Lack of modulation of erythro- cyte glucose transporter by endogenous glycosphingolipids, Eui: J . Biochem. 218, 637-643.

5. Ito, M., Ikegami, Y. & Yamagata, T. (1991) Activator proteins for glycosphingolipid hydrolysis by endogl ycoceramidases, J. Biol. Chem. 266, 7919-7926.

6. Ito, M., Ikegami, Y., Omori, A. & Yamagata, T. (1991) Conver- sion of endoglycoceramidase-activator I1 by trypsin to the 27.9 kDa polypeptide possessing full activity, J . Biochem.

7. Hardy, M. R. & Townsend, R. R. (1988) Separation of positional isomers of oligosaccharides and glycopeptides by high-per- formance anion-exchange chromatography with pulsed

(Tokyo) 110, 328-332.

649

amperometric detection, Proc. Natl Acad. Sci. USA 85, 3289 - 3293.

8. Kitamikado, M., Ito, M. & Li, Y.-T. (1981) Isolation and charac- terization of a keratan sulfate-degrading endo-p-galactosidase from Flavobacterium kerutolyficus, J. Biol. Chem. 256, 3906 - 3909.

9. Yamamoto, K., Lee, K. M., Kumagai, H. & Tochikura, T. (1985) Purification and characterization of P-N-acetylhexosaminidase from Penicillum oxalicum, Agric. Biol. Chem. 49, 611 - 619.

10. Corfield, A. P., Veh, R. W., Wember, M., Michalski, J.-C. & Schauer, R. (1981) The release of N-acetyl- and N-glycolloyl- neuraminic acid from soluble complex carbohydrates and erythrocytes by bacterial, viral and mammalian sialidases, Biochem. J. 197, 293-299.

11. Lampio, A., Rauvala, H. & Gahmberg, C. G. (1986) Exposure of major neutral glycolipids in red cells to galactose oxidase: Effect of neuraminidase, Eur J. Biochem. 157, 611 -616.

12. Svennerholm, L. (1963) Chromatographic separatioii of human brain gangliosides, J. Neurochem. 10, 61 3 -623.