Enzymes that destroy blood group specificity. I. Purification and properties of alpha-L-fucosidase from Clostridium perfringens.

Abstract An α-l-fucosidase was purified approximately 600-fold from Clostridium perfringens; it is free of other glycosidases present in the crude extract. The fucosidase occurs in multiple "isozymic" forms with molecular weights greater than 200,000. It is an α-(1 → 2)-specific l-fucosidase with action on oligosaccharides and glycoproteins, but with no action on simple methyl or nitrophenyl fucosides. Action on the H-specific hog submaxillary glycoprotein results in the loss of the serological specificity. The enzyme has a pH optimum of 6.0, with a Km of 0.175 mm and its activity is enhanced by the presence of salts, the most effective being calcium chloride.


SUMMARY
An a-L-fucosidase was purified approximately 600-fold from Clostridium perfringens; it is free of other glycosidases present in the crude extract. The fucosidase occurs in multiple "isozymic" forms with molecular weights greater than 200,000. It is an oc-(1 + 2)-specific L-fucosidase with action on oligosaccharides and glycoproteins, but with no action on simple methyl or nitrophenyl fucosides.
Action on the H-specific hog submaxillary glycoprotein results in the loss of the serological specificity.
The enzyme has a pH optimum of 6.0, with a Km of 0.175 mM and its activity is enhanced by the presence of salts, the most effective being calcium chloride.
The literature is replete with reports of enzymes that degrade glycoproteins with blood group activity (l-3).
Most of the potent sources of these enzymes are complex mixtures of glycosidases and proteinases, and in many cases are capable of destroying more than one blood group activity. Evidence has accumulated suggesting that the most active enzymes which destroy blood group activity are glycosidases. There is need, therefore, for pure glycosidases of known substrate specificity to confirm the role of the nonreducing terminal sugar in determining blood group specificity (l-3), and secondly, as a corollary (3) to establish unequivocally the mechanism of enzymatic inactivation of the serological specificity. For instance, it has been shown that both N-acetylgalactosamine deacetylase and the oc-N-acetylgalactosaminidase of Clostridium tedium can destroy blood group A activity (4). In most cases the purification of the enzymes was followed by the loss of serological activity on incubation with blood group active glycoproteins (3). A fair degree of enzymatic purification was obtained with this assay, but it has distinct limitations. These include ( from a number of different enzymatic reactions thereby making interpretation of effectiveness of purification difficult in a mixture of such enzymes, and (c) the loss of blood group activity alone gives no clue as to the mechanism involved. Synthetic phenolic glycosides have been widely employed by a number of investigators for the detection and isolation of glycosidases.
Unfortunately, as will become apparent in this report, and as has been observed by others, the enzymes that degrade blood group substances usually have no action on these simple substrates (5).
The reason for the lack of extensive purification of enzymes that degrade blood group substances, therefore, is that of methodology, namely the paucity of available techniques for the determination of the free sugar specifically in the presence of other free sugars and of the same sugar glycosidically bound. With this limitation in mind, a method was recently developed for the determination of free fucose in the presence of the glycosidically bound sugar which was not significantly affected by the presence of other free sugars (6). This assay was used successfully for the screening of fucosidases in a number of biological extracts. The fucosidase from Clostridium perfringens was chosen for further purification because of its interesting specificity.
A preliminary report of this investigation has already appeared (7). EXPERIMENTAL (8) were prepared by modifications of procedures described elsewhere (9). They had the following analytical values: A, 8.9% fucose and 20.1 y0 N-glycolyl neuramic acid; H, 9.4 and 20.8%, respectively. Purified porcine gastric mucin was prepared according to the procedure of Morgan and King (10). This is a mixture of glycoproteins obtained from a pool of stomachs from hogs of A and H phenotypes.
The buffers used in the determinations of the pH profiles were prepared as follows: (a) citrate phosphate (18), 0.4 M Na2HP0+ x ml, and 0.2 M citric acid (10-x) ml were mixed in varying proportions to give solutions of different pH; (b) acetate-Verona1 (19), 2 ml of a solution 0.143 M with respect to sodium acetate and to sodium diethylbarbiturate was mixed with 0.8 ml of 1.45 M lXaC1, and the pH was adjusted with 0.1 N HCI in a final volume of 10 ml. All pH measurements were made at room temperature.

Methods
The following methods were employed: total fucose with a IO-min heating period (20); free fucose (6) ; total sialic acid by modification of the Svennerholm procedure (21) with I-butanol instead of isoamyl alcohol for the development of color (22) ; free sialic acid by the thiobarbituric acid procedure (23) ; protein by absorbance at 280 nm (24) or by the microbiuret procedure (25).
Electrophoresis in polyacrylamide gel slabs was performed by a modification of the procedure of Davis (26) with the use of a 10.5% standard gel in borate buffer, pH 9. The electrophoretic separations were carried out at 4-6" with 50 ma per slab (27).
The substrate employed for the routine tests was a purified glycoprotein isolated from hog submaxillary glands with blood group H-specificity (8). Incubations were conducted at 37" for 15 min, and the reactions were stopped by heating the mixtures for 1 min at 100".
A 400~~1 aliquot was then quantitatively transferred to Conway units for the determination of fucose released (6). When the substrate contained both fucose and sialic acid and it was desired to assay for both the fucosidase and sialidase activities, the same incubation mixture was used with a final volume of 550 ~1. Of this, 400 ~1 were used for the determination of released fucose (6) and 100 ~1 for released sialic acid (23). Control incubation mixtures contained all the components with the deletion of either substrate or enzyme.
A unit of fucosidase (or sialidase) activity was defined as the amount of enzyme that released 1 pmole of fucose (or sialic acid) per hour from the hog submaxillary glycoprotein with H-specificity.
Specific activity was expressed in terms of micromoles of sugar released per hour per mg of protein.

Purijkation of Fucosidase
Growth of Cells-C. perfringens, Type 33-48, obtained from Mr. George S. Fearnehough (Department of Microbiology, University of Michigan), was maintained and grown in cooked meat medium (Difco). For the preparation of enzyme the organism was grown in a Todd-Hewitt broth medium of the following composition (grams per liter): Todd-Hewitt broth powder (Difco), 35; K~HPOI. 3Hz0, 2.36; NaCI, 2.5; glucose, 1.5; cysteine HCl.lHzO, 0.05. The glucose and cysteine HCl were each autoclaved separately from the broth and both added to the medium just before the inoculation.
A 0.1% inoculum from the meat broth stock culture was used and growth maintained at 37" for 72 hours. The cultures were grown in 2-liter flasks containing 1.5 liters of medium.
At the end of 72 hours, the cultures were chilled to 4" and the cells were removed by centrifugation in a refrigerated Sorvall or Sharples centrifuge.
All subsequent steps in the purification were carried out in the cold room at 4". Ammonium Sulfate Fractionation-The culture supernatant solution (94.2 liters) was adjusted to 80% saturation (516 g of solid ammonium sulfate per liter). After 16 hours at 4", the precipitate was collected by centrifugation in a Sharples ultracentrifuge at 17,000 rpm.
The pellet was dissolved in 5,500 ml of phosphate buffer.
The resulting solution could be lyophilized or stored frozen for at least 2 years without loss of activity.
To effectively separate the fucosidase from the sialidase, it was necessary to refractionate with solid ammonium sulfate collecting the precipitates between narrow concentration differences of ammonium sulfate.
The best fraction was obtained between 52.3% saturation (307 g per liter) and 54.5% (322.4 g per liter) with a fucosidase to sialidase ratio of 15 (Table I).
Only this fraction was utilized for further purification of the fucosidase, despite Issue of April 10, 1970 D. Aminoj' and K. Furukawa the low yield, since we have found it extremely difficult to separate the fucosidase from the sialidase in subsequent, steps.
Xephadex G-76 Treatment-The fraction, precipitated between 52.3 and 54.5% ammonium sulfate saturation, was dissolved in 200 ml of the phosphate buffer, and 196 ml were applied to a Sephadex G-75 (particle size 40 to 120 ~.r) column (8.4 X 75 cm) previously equilibrated with phosphate buffer. The enzymes were then eluted with the same buffer at a flow rate of 10 ml per hour, and 20-ml fractions were collected for protein and enzyme assays. Both fucosidase and sialidase activities appeared in the first of the three major peaks of material absorbing at, 280 nm ( Fig. 1 progressively increasing concentrations up to 4% dissolved in phosphate buffer and at a flow rate of 75 ml per hour. Fractions of 30 ml each were collected for protein and enzyme assays. The stepwise elution pattern obtained from Cy is shown in Fig. 2. The major fucosidase activity was eluted at 1 to 1.25% ammonium sulfate (w/v).
This treatment removed the last traces of sialidase activity (Table II)  and invariably there is some loss of activity (Table III).
Other methods of concentrating the enzymes were attempted but these are not as reproducible nor do they give as good recovery of the enzymes.

Purity of Fucosidase
Gel Ebctrophoresis-The course of fractionation at the various stages of purification (Table III) was followed by electrophoresis on polyacrylamide gels (26, 27) and stained for protein (Fig. 3).
The glycosidases in the various fractions were shown histochemically on the acrylamide gels by a modification of the formazan staining technique for reducing sugars (32), devised specifically to meet the requirements of a substrate with a very large molecular weight (33). The substrate was incorporated in the acrylamide prior to gelation. By use of this modified technique with the H-specific hog submaxillary glycoprotein as substrate, it was possible to show a number of formazan-staining bands in both the crude and the highly purified enzyme fractions (Figs. 4 and 5).
Since only fucose could be detected in the reaction products of the incubation (see below), the presence of several bands would imply that we have multiple isozymic forms of fucosidase. The following experiment was carried out in order to establish this. The crude extract and pure enzymes were run in parallel. One section of each was stained for glycosidases while the remainder of the slab was sliced horizontally into l-mm sections. Each I-mm section was then incubated separately with more substrate overnight, and the reaction products tested chromatographically for the sugars released (28)(29)(30) and serologically for the loss of H activity.
Of the glycosidases detected in the crude extract, galactosidase, sialidase, and fucosidase, only fucosidase appeared in the purified preparation.
The principal fucosidase activity occurred in Slices 7 to 9, (Fig. 5), with a simultaneous maximal H-destroying FIG. 6. Diagrammatic presentation of paper chromatographic results of fucosidase action on H-specific hog submaxillary glycoprotein.
The paper was stained with alkaline silver nitrate reagents.

activity.
Minor fucosidase and H-destroying activity occurred in Slices 11 and 12 and 19 and 20 with trace activities in 92, d7 and 88.
Action on Phenyl-and oand p-Nitrophenyl GlycosidesThe following phenyl, o-nitrophenyl or p-nitrophenyl glycosides were tested: (Y-and fl-n-glucosides, /3-n-glucuronide, o(-and P-n-galactosides, P-n-xyloside, a-n-mannoside, Q-and O-n-N-acetylgalactosaminides, and o(-and P-n-N-acetylglucosaminides. The reaction mixture consisted of 0.6 unit of the fucosidase from the crude or pure enzyme, 5 pmoles of CaC12, 130 pmoles of ammonium sulfate, and 0.5 pmole of the substrate in a total volume of 0.5 ml. These were incubated for 15 min and 16 hours at 37", and heated in a boiling water bath for 1 min to stop the reaction. Blue dextran (a---0) was estimated by extinction at G25 nm. Fumarase (CI---0) was determined by change of the absorbance at 300 nm. Lactic acid dehydrogenase (a---A) was assayed according to procedure of Neilands (46).
For the determination of hydrolysis of the p-and o-nitrophenyl glycosides, 0.5 ml of 0.97 N NazC03 was added to the incubations, mixed well, and spun to remove the precipitated salt. The extent of hydrolysis was determined photometrically, with pnitrophenol as the standard.
With the phenyl glycosides, 0.5 ml of 1 N Folin reagent was added to the incubation mixture and spun down.
To 0.3 ml of the supernatant was added 1.7 ml of 0.4 M Na&O, and the resulting blue color after 30 min at room temperature was compared at 650 nm with phenol as the standard.
Only a trace of activity was detected with p-nitrophenyl-/?-Nacetyl-o-glucosaminide after 16 hours of incubation. This was very small compared to the strong reactivity shown by the crude enzyme preparation after 15 min of incubation. Sialidase-To determine the presence of sialidase, purified hog H-specific submaxillary glycoprotein (9.4% fucose and 20.8% N-glycolyl neuraminic acid) was used as the substrate. Incubations were set up in the usual manner and the fucose and sialic acid released were determined chemically (6,23). No sialic acid was detected even after 16 hours of incubation when over 95% of the total fucose was released.
The absence of sialidase was further confirmed by the chromatographic data as described under "Characterization of Reaction Product" (Fig. 6). Protease Activity-To test for protease activity, crude porcine submaxillary glycoprotein still contaminated with proteins was considered the most suitable typical substrate. The crude glycoprotein, 1 mg, was incubated with 0.05 mg of trypsin, Pronase, or 0.5 unit of fucosidase from the crude or pure preparation (Steps 1 and 6, Table III) in acetate-Verona1 buffer, pH 6.0, in a total volume of 0.5 ml.
After 8 hours at 37", 1 ml of the ninhydrin reagent? was added and the tubes heated for 12 min in a boiling water bath, cooled in ice water, and after 5 min 1 The ninhydrin reagent consists of 25 parts of 0.04% ninhydrin in 0.5 M citrate buffer. DH 5.5. mixed with 12 parts of analvtical grade glycerol.
The method described is a private communication from Dr. H. Tager, Department of Biological Chemist,ry, University of Michigan, and represents a modification of the procedure described by Lee and Takahashi (34). Under these conditions of test, trypsin released ninhydrin-reacting material equivalent to 2.3 pmole of glycine, Pronase, 2.9 wmoles; crude fucosidase, 0.6 pmole and the pure fucosidase, 0 pmole. Action on Serological Activity of Blood Group Substances-As has been observed by Stack and Morgan (35), crude Clostridium welchii culture filtrates can inactivate A, B, and O(H) blood group specificities.
We have confirmed this and, moreover, have shown that our crude C. perfringens filtrates also contain Lea-destroying activity (36). It was of interest, therefore, to compare the effect of the purified fucosidase on the various serological activities.
These results will be reported in greater detail elsewhere,2 but suffice it to state at this stage that only the Hdestroying activity could be detected in the purified preparation.
Of these components, only n-fucose would yield acetaldehyde on oxidation with periodate. This product was detected and determined in the Conway unit assay (6). Identification of fucose, as the sugar released by the enzyme, was confirmed by paper partition chromatography of the incubation mixture after the removal of salts by precipitation with 4 volumes of ethanol.
Solvent system C gave the most effective separation of all the carbohydrate components anticipated (Fig. 6). The crude enzyme preparation released N-glycolyl neuraminic acid and fucose, while the purified enzyme released only one reducing spot that migrated like fucose. Admixture of the incubation products of the purified enzyme with substrate and known n-fucose resulted in no increase in the number of spots (Fig. 6). Further characterization of the reaction product as n-fucose was made with the DPN+-dependent L-fucose dehydrogenase from porcine liver (17).
The amount of n-fucose released on incubation of H-active hog submaxillary glycoprotein with the fucosidase at 15 min and 5 hours was determined by the Conway unit method (6) and by the DPN+-dependent n-fucose dehydrogenase (41). In both cases the dehydrogenase assay accounted for 80 to 85% of the n-fucose as determined by the periodate method.
In view of the established specificity of the dehydrogenase (17), one may conclude that the enzyme from C. perfringens is an n-fucosidase.

Determination of Molecular Weight
The behavior of the cu-fucosidase on gel filtration through Sephadex G-200 was undertaken to determine whether the electrophoretically demonstrable isozymic pattern is reflected in a like number of enzymatically active molecular subunits, and also to determine the order of magnitude of molecular weight of the enzyme (42, 43). Sephadex G-200 (particle size 40 to 120 p) was allowed to swell in 0.01 M potassium phosphate buffer, pH 7.0, containing 0.025 M KCl. The deaerated suspension was packed under gravity in a column, 2.4 x 50 cm. The concentrated T-VIII, bentonite-treated fucosidase preparation, containing 40 units of the enzyme, was mixed with a number of proteins of known molecular weight and applied to the gel as a solution in 2 ml of the same buffer.
The elution rate was 14 ml per hour and 2.3-ml fractions of eluate were collected in a fraction collector at 4". Blue dextran 2000 (1 mg) was used to indicate the void volume, and the following proteins were used as standards: fumarase, 1.2 mg (molecular weight 185,000 to 225,000) ; aldolase, 1 mg (molecular weight 140,000 to 150,000) ; lactic acid dehydrogenase, 0.0815 mg (rabbit muscle, molecular weight 130,000 to 140,000) ; bovine serum albumin, 4 mg (molecular weight 65,000 to 70,000) ; cytochrome c (horse heart, molecular weight 12,400) (43).
The position of the elution peaks of the fucosidase, blue dextran and standard proteins was determined either by direct absorption at 280 nm (bovine serum albumin), 412 nm (cytochrome c), and 625 nm (blue dextran), or by the appropriate enzyme assay for fumarase (44), aldolase (45), and lactic acid dehydrogensse (46). Fig. 7 shows the relevant results obtained with the blue dextran, fumarase, lactic acid dehydrogenase, and the fucosidase. The broadness of the peak obtained with the fucosidase as well as the fact that it eluted faster than any of the protein standards precludes an accurate assessment of its molecular weight beyond the statement that it appears to be greater than that of fumarase, 185,000 to 225,000 (43).

Properties of cr-Fucosidase
Stability of Enzyme-The fucosidase is stable to freezing at all stages of purification, but is inactivated on repeated freezing and thawing.
Lyophilization results in inactivation of the purified preparations.
The only procedure to give satisfactory concentration of the enzyme without appreciable loss of activity is vacuum dialysis against phosphate buffer, as discussed above. Simple dialysis against distilled water results in a rapid loss of activity.
Dialysis against buffers at various pH values results in preciprtation of the enzyme at pH values below 5.5, with considerable inactivation.

Kinetic Studies
Effect of pH-The following incubation mixtures were set up in order to determine the pH profile of the enzyme: 100 ~1 of the enzyme, 1 unit; 100 ~1 of the buffer of the appropriate pH, 100 ~1 of the hog H submaxillary glycoprotein substrate containing 0.45 pmole of bound fucose in a final volume of 0.5 ml. The amount of fucose released was determined after 15 min at 37" and the results expressed in terms of units of activity as compared to that determined by the standard conditions (1 unit) in ammonium sulfate (0.26 M) and CaClz (0.01 M) previously adjusted to pH 6.0.
The results are shown in Fig. 8. Maximal activity is obtained at pH 5.8 in sodium acetate-Verona1 buffer and 6.3 in the citratephosphate buffer. Enhancement of activity and a slight increase in pH optimum to 6.0 occurs with acetate-Verona& in the presence of CaC&, 0.01 M. Under these conditions, the activity is the same as with CaClz and ammonium sulfate previously adjusted to pH 6.0 (final concentration 0.01 and 0.26 M, respectively) as used in the routine assay.

Effect of Substrate Concentration-Incubating
various concentrations of the same substrate for 15 min with the optimal concentrations of CaClz and ammonium sulfate, alone and with the enzyme, resulted in Curves A and B, Fig. 9. The substrate gives a persistent and constant "blank" which is directly proportional to the amount of substrate present. Correction for this interference results in Curve C (Fig. 9) which gives an indication of substrate inhibition at high concentrations. When plotted according to the method of Lineweaver and Burk (47) as modified by Hanes (48), the K,,, obtained is 0.175 mM.
However, this is only a tentative and an apparent K, value on account of the high substrate blank and the fact that the substrate is a high molecular weight polymer with multiple substrate sites. E$ect of Enzyme Concentration and Period of Incubation-The rate of hydrolysis of the H-specific submaxillary glycoprotein was proportional to the amount of enzyme (Fig. 10) and period of incubation (Fig. ll)  11. Effect of time on the amount of fucose released. Incubation mixtures were made up in the assay procedure described in the text and containing CaClz and ammonium sulfate at pH 6.0, 0.5 rmole of bound fucose, and 5.25 pg of enzyme. the release of fucose from the H-specific hog submaxillary glycoprotein.
The incubations were carried out in Veronal-acetate buffer, pH 6.0, and the amount of fucose released in 15 min in the buffer alone was taken as 100% for purposes of comparison. The compounds under investigation had no effect on the assay for free fucose, unless otherwise indicated.
Sodium, magnesium, and calcium chlorides, as well as ammonium sulfate, enhance the activity of the enzyme. The relationship of enzymatic activity to the concentration of the salt was investigated in greater detail in the case of NaCl, CaC12, and ammonium sulfate, as additives to 0.009 M acetate-Verona1 buffer, pH 6.0. The results are expressed in terms of a percentage of the activity shown in the presence of buffer alone. The enhancing effect of CaClz is illustrated in Fig. 12 I Vol. 245,No. 7 with or without the addition of the appropriate fucoside (1 mM) under test.
No significant inhibition was observed. The results would also suggest that there has been no transglycosylation involved with the simple glycosides acting as acceptors.

Substrate Specificity
The following experiment was done in order to determine the extent of release of fucose from hog H-submaxillary glycoprotein. The incubation mixtures in a total volume of 0.5 ml contained 1 unit of fucosidase, 5 pmoles of CaC&, 130 pmoles of ammonium sulfate, and 1 mg of H-active glycoprotein containing 0.63 pmoles of bound fucose.
The incubation mixtures were held at 37" and at predetermined time intervals the tubes were removed, boiled for 1 min, and 400-J aliyuots removed for the determination of fucose released (6). There is a very rapid release of fucose, as shown in Fig. 13, and the reaction is essentially complete in 1 hour. The experiment was repeated with hog A-active submaxillary glycoprotein, 1 mg containing 0.48 pmole of bound fucose.
A similar rapid release of fucose occurs within 1 hour and again the reaction is essentially complete. However, in this case only 20% of the total fucose is released.
In order to determine the reason for this incomplete release of fucose, incubations were put up as indicated above and kept at 37" for 2 hours.
The course of the fucose released over the next 2 hours was followed after incubation with the following additives: (a) more substrate, 0.48 pmole of A-glycoprotein, (b) more of the pure enzyme, 1 unit, (c) addition of the crude enzyme from C. perfringens, 0.75 unit, and (d) no additives.
From the results obtained, shown diagrammatically in Fig. 14, we can infer that the pure enzyme has not been inactivated during the 2 hours of incubation, but rather that it is fully active and capable of releasing all the fucose from the fresh substrates added.
Addition of more fresh pure fucosidase releases no further amount of fucose, confirming that the reaction was essentially complete at the end of the 2 hours and implying that all the bound fucose that is susceptible to release with this fucosidase has indeed been released.
The fact that more fucose is released on incubation with the crude enzyme, from both the A-specific, and to a lesser extent, H-specific glycoproteins, suggests that these glycoproteins either contain fucose bound in yet another form (to account for the observed increment of fucose released with the crude C. perfringens), or, as is more likely in the case of the Aspecific glycoprotein, there is steric hindrance attributable to the close proximity of the terminal N-acetyl-n-galactosamine, the A-determinant, attached cr(1 + 3) to the same galactose residue to which the fucose is attached cr(1 -+ 2), or both.
The observed 20% fucose released would represent incomplete oligosaccharide chains within the glycoprotein macromolecule. Such chains have been detected chemically in the hog submaxillary A-glycoprotein (40).
It would appear from these kinetic data that the fucosidase is highly specific.
The nature of that specificity was determined by its action on a number of simple methyl and phenyl fucosides and on milk oligosaccharides of known structure. Table IV  summarizes the results of incubation for 15 min and 5 hours to reflect relative rate and extent of susceptibility to hydrolysis. The fucosidase has no action on the simple fucosides, a-or /?-, D-, or I~-, methyl-or p-nitrophenyl, pyranoside, or furanoside. Of the milk oligosaccharides tested, the most readily susceptible appear to be 2'-fucosyl lactose and lacto-N-fucopentaose I. In both cases the L-fucose is bound to galactose, in an a-(1 + 2) linkage (11). In both cases, however, the rate of release of fucose from these oligosaccharides is slower than with the large molecular weight glycoprotein substrate. The trisaccharide obtained by Katzman and Eylar (12) from porcine submaxillary glycoprotein on alkaline borohydride hydrolysis behaves in exactly the same manner as the original I-I-specific glycoprotein.
The enzyme has a very limited action on 3'-fucosyl lactose. The results obtained with the lactodifucotetraose are misleading. Lactodifucotetraose contains 2 fucose residues per molecule, only 1 of which, the (~(1 -+ 2) residue, is susceptible to hydrolysis. Hence, the 39% of the total fucose released after 5 hours would represent 78% of the cr(l + 2).linked fucosyl residues. This is of the same order of magnitude obtained with the other a(1 --t 2) fucosyl oligosaccharides.
The lack of activity on ol(l 4 4)bound fucose is evident in fucopentaose II, and reflected in the fact that it has very little action on the Lea-active glycoprotein from human ovarian cyst fluid.  (58). In a few cases there has been a limited amount of purification beyond the initial identification (51,54,55). In most studies, the substrate used was a simple nitrophenyl fucoside for ease of following the purification of the enzymes.
In only a few cases, however, was the substrate specificity of the enzyme determined. Where tested, these fucosidases had no action on the blood group active glycoproteins (5). Where enzymes capable of destroying the H-or Lea-blood group specificity of glycoproteins had been described (3,5), it was presumed that the loss of serological activity was attributable to the loss of fucose.
But since the release of more than one sugar was detectable, the complete correlation of st#ructure and serological specificity was not possible. Again, where tested, these enzymes had no action on simple glycosides, and thus their purification was greatly handicapped for the lack of a suitable rapid chemical assay (3).
In order to overcome these limitations, it was necessary to devise a specific assay to determine the free fucose in the presence of the glycosidically bound sugar (6). The purification and isolation of the specific ~y(l + 2)+fucosidase was achieved by the use of a substrate in which all the fucose is bound in that form only, namely the hog H submaxillary glycoprotein. Another valuable innovation introduced in these investigations (33) is the development of a comprehensive procedure for the histochemical detection of glycosidases adaptable to large molecular weight substrates. This is a necessary tool in the preliminary survey for detection of potential glycosidases acting on the same substrate, to follow the purification of the glycosidase under investigation, ultimately to establish its purity, and, finally, to show the presence of multiple forms of the enzyme as evidenced in this particular a-L-fucosidase. This clear cut demonstration of multiple isozymic forms in a glycosidase with well defined substrate position specificity, ~(1 + 2)+fucosidase, is very intriguing. The isozymes may quite possibly represent end-products of partial degradation of the fucosidase by amino-or carboxypeptidases present in the crude bacterial culture filtrates.
It is possible that the same type of isozymic pattern will be by guest on March 24, 2020 http://www.jbc.org/

Downloaded from
Issue of April 10, 1970 D. Aminoff and K. Furukawa detected in other glycosidases after extensive purification, as has already been suspected in the sialidases (22,59). Apart from its intrinsic interest, it does present technical difficulties in isolating a given type of glycosidase free of all other straggling isozymes of other glycosidases, and providing us the assurance that all the isozymes of the given glycosidase under investigation are indeed harvested in the final preparation.
Turning now from the enzymes to the substrates upon which they act, there are indications that perhaps not all the fucosyl residues in the complex glycoproteins studied are a!-(1 --+ 2). This is strongly suggested by the nature of the kinetic studies on the fucose released from the various submaxillary glycoproteins when treated with the crude and purified enzymes.
The data, however, could also be explained, to a certain extent, by steric hindrance caused by sugars adjacent to the potentially susceptible fucose.
Investigations are under way to resolve these problems and their significance will be discussed elsewhere (30).
The isolation of pure glycosidases provides us with valuable tools for structural studies of complex heteropolysaccharides and conjugated glycoproteins and glycolipids. Since many of these compounds have interesting and important biological properties, either in solution or at cellular membrane surfaces, these can now be studied at the molecular level correlating biological activity with structure.