Purification and Properties of an a-L-Fucosidase from Rat Epididymis*

SUMMARY An cu-L-fucosidase has been purified from rat epididymis by fractionation of extracts with ammonium sulfate and subsequent chromatography on ion exchange celluloses. It is a glycoprotein with a molecular weight of 210,000 to 220,000 and dissociates in the presence of mercaptoethanol and guanidine hydrochloride or sodium dodecyl sulfate. The dissociated material behaves as two components with molecular weights of approximately 47,000 and 60,000, thus suggesting two pairs of dissimilar subunits. A preliminary amino acid and carbohydrate composition is reported. Other glycosidases of rat epididymis, /3-D-N-acetylglucosaminidase, and /?-D-galactosidase have been partially purified from the same extract. All were active against a variety of glycoprotein and glycopeptide substrates. The purified fucosidase preparation is stable, is essentially free from other glycosidases with the exception of small amounts of N-acetylglucosaminidase and it catalyzes the hydrolysis of p-nitrophenyl-oc-L-fucoside. Studies with glycopeptides


Purification
and Properties of an a-L-Fucosidase from Rat Epididymis* (Received for publication, June 24, 1971) ROBERT B. CARLSEX~ AND JOHN G. PIERCE From the Department of Biological Chemistry, UCLA School of Medicine, Los Angeles, California 90024 SUMMARY An cu-L-fucosidase has been purified from rat epididymis by fractionation of extracts with ammonium sulfate and subsequent chromatography on ion exchange celluloses. It is a glycoprotein with a molecular weight of 210,000 to 220,000 and dissociates in the presence of mercaptoethanol and guanidine hydrochloride or sodium dodecyl sulfate. The dissociated material behaves as two components with molecular weights of approximately 47,000 and 60,000, thus suggesting two pairs of dissimilar subunits. A preliminary amino acid and carbohydrate composition is reported. Other glycosidases of rat epididymis, o!-D-mannosidase, /3-D-N-acetylglucosaminidase, and /?-D-galactosidase have been partially purified from the same extract. All were active against a variety of glycoprotein and glycopeptide substrates.
The purified fucosidase preparation is stable, is essentially free from other glycosidases with the exception of small amounts of N-acetylglucosaminidase activity, and it catalyzes the hydrolysis of p-nitrophenyl-oc-L-fucoside. Studies with glycopeptides show quantitative release of fucose by the purified enzyme from the tryptic glycopeptide of the hormone-specific chain of bovine luteinizing hormone and from a peptide of horse immunoglobulin G. A general method has been developed for quantitative determination by gas-liquid chromatography of the monosaccharides released by enzymic hydrolysis of glycoproteins and glycopeptides.
The >tructurrs of the complex oligosaccharides found in glycoproteins are of considerable h&rest.
Studies of structure hare been hindered, however, by the limitations of chemical methods such a:: partial acid hydrolysis, mcthylation, and oxidation with periodate.
Recently, a number of enzymes which degrade these oligoanccharides have been isolated (l-9).
Such enzymes are of great utility in structural studies as thq-provide both a means for sequential degradation of the oligosaccharides and informa-* This illvcstigntioll U-M support,ed by Ullited States Public Health Service Itesc:rrch Gralrt CA-2290 and Training Grant GA1 361 fronl the N:ltiotl:ll Institutes of Health.
1 Present address, I)rpartment of Biochemistry, State University of New XTork. Buffalo, New York 14214. tion regarding anomeric configurations.
The rat epididymal enzymes in crude extracts have been extensively studied in their action on synthetic substrates by Levvy and RiIcAllan (lo), Conchie and Hay (II), Snaith and Levvy (12), Conchie et al. (13),and Findlay et al. (14), and one, the oc-mannosidase, has been prepared free of contaminating glycosidic activities (12). Removal of fucose should be an early step in structural studies of glycoprotein carbohydrates as it occurs as a nonreducing terminal sugar (15), but, of the glycosidases of principal utility in such studies, a-fucosidasesl are the least common.
Two from bacterial sources have recently been prepared free of contaminating glycosidases (1, 2). Both appear to be specific for a-(1,2) linkages and do not release fucose from synthetic substrates but,, to date, no ol-fucosidases with a broader specificity have been purified, although partial purification of a kidney fucosidase has been described (16,17). The present report describes the purification of a fucosidase from rat, epididymal tissue which catalyzes the hydrolysis of fucoside linkages in several glycopeptides.
Epididymal tissue was chosen because it is rich in a-fucosidase (lo), normally present onl-at low levels in many sources of glycosidic enzymes.
Further, the epididymal fucosidase is active against glycoprotein oligosaccharides, as first noted in this laboratory, and has a broad specificity (10,18,19), in marked contrast to the narrow specificity of the bactrrial cr.fucosidases. Emphasis has been placed on the separation of the a-fucosidase from contaminating activities and on its purification. The separation and partial purification of the other major ylycosidases of rat epididymis, cr-mannosidase, /M-acetylglucosaminidase, and P-galactosidase, and their activity against a variety of glycoprotein and glycopeptide substrates, are also reported. The solution was allowed to stand at 4" for 3 hours, then centrifuged at 0" for 30 min at 29,000 x g. The precipitate was discarded and the supernatant brought to 50% of ammonium sulfate saturation by addition of solid ammonium sulfate and allowed to stand at 4" for 20 hours.
It was then centrifuged at 0" for 30 min at 29,000 x g. The precipitate was dissolved in 2.0 ml of 0.1 M sodium citrate, pH 6.0, 0.02yc in NaN3 to yield the 35 to 50% ammonium sulfate fraction.
The supernatant was brought to 60% of ammonium sulfate saturation by addition of solid ammonium sulfate and allowed to stand at 4" for 20 hours. It was then centrifuged at 0" for 30 min at 29,000 x g. The precipitate was dissolved in 2.0 ml of 0.1 M sodium citrate, pH 6.5, 0.02% in NaN3 to yield the 50 to 60$Zc ammonium sulfate fraction.
Chromatographic Methods-Further fractionation of the 35 to 50% ammonium sulfate fraction was obtained by chromatography on CM-cellulose (Whatman CM 32) and subsequent chromatography on DEAE-cellulose (Whatman DE 32) of the third fraction eluted from CM-cellulose.
Because the enzymes are heat stable, chromatography on both CM-and DEAE-celluloses can be conveniently carried out at room temperature. Details are given in the legend to Fig. 1. Pooled fractions from CMcellulose were adjusted to pH 6 with 15 N NH,OH, brought to SOY, saturation with ammonium sulfate, and the resulting suspensions centrifuged at 0" for 30 min at 29,000 X g after standing at 4" for 24 hours. The precipitates were dissolved in 0.1 M sodium citrate, pH 6.0, 0.02% in NaN3. For recovery of material after chromatography on DEAE-cellulose, pooled fractions were adjusted to pH 6 with glacial acetic acid, brought to 95% saturation with solid ammonium sulfate, and the resulting suspensions centrifuged at 0" for 30 min at 33,000 x g after standing at 4" for 24 hours. The fucosidase was again dissolved, for storage, in the citrate buffer, pH 6.0; the other fractions were dissolved in 0.1 M sodium acetate, pH 5.5. Molecular Weight Determinations 012 Pur$ed cy-.r-Fucosidase-The molecular weight was determined by high speed sedimentation equilibrium in a three-channel interference cell in the Spinco model E ultracentrifuge, equipped with interference optics, by the meniscus depletion method of Yphantis (29). Half and whole fringes were read. Protein concentrations were from 0.36 to 0.80 mg per ml and all determinations were made at 20". A value of 0.728 ml per g, based on the composition, was calculated for the partial specific volume.
The native material was run in 0.1 M sodium citrate, pH 6.0, 0.02% in NaN3 at a rotor speed of 12,590 rpm. To determine the subunit molecular weight the sample was run in 0.035 M sodium barbital-O.065 M NaCl, pH 7.2, 6 M guanidine-HCl, and 0.57, in P-mercaptoethanol. The initial rotor speed was 36,000 rpm, slowed to 30,000 rpm after 22 hours.
Electrophoresis-Electrophoresis of the native hormone was carried out at pH 9.5 in Tris-phosphate buffer with a glycinerunning buffer as described by Liao et al. (20). For the 5.6% acrylamide gel, pH 6.6, the concentrations of all reagents except the ammonium persulfate were reduced appropriately. The molarity of the Tris buffer was 0.095. Enzyme activity was IOcated after electrophoresis at pH 6.6 by cutting the gel into segments; each segment was macerated and allowed to stand 100 ~1 of citrate buffer, pH 6.6,0.1 M, for 15 min at 37". Substrate was added (100 ~1 of 3 rnlr p-nitrophenyl fucoside) and the solution again allowed to stand for 15 min. Full color was then developed by addition of alkali.
The enzyme was also subjected to electrophoresis on 57, polyacrylamide gels in the presence of O.lyO SDS according to the method of Weber and Osborn (30). Electrophoresis was at 6 ma per tube (27 volts) for 75 hours at room temperature. Standards run concurrently for molecular weight determination were pepsin, ovalbumin, bovine serum albumin, glutamic dehydrogenase, Escherichia coli /3-galactosidase, and thyroglobulin.
Two determinations were made with the cr-fucosidase; one with 20 pg of material and the other with 30 pg.

Hydrolysis of Glycoprotein and Glycopeptide Oligosaccharides
with Rat Epididymal Glycosidases-Hydrolyses for 72 hours were carried out with a crude enzyme mixture (35 to 50% ammonium sulfate fraction) and the time course of hydrolysis was determined with the purified cu-fucosidase (Fraction DEAE-III).
From 25 to 75 nmoles of protein or peptide sample were used per single hydrolysis or per lOO-~1 aliquot for time study hydrolyses.
These amounts were the sum of three equal additions of enzymes at 0, 24, and 48 hours. For hydrolyses with the purified a-fucosidase sufficient enzyme was added so that 0.5 unit of a-fucosidase was present per 100~~1 aliquot.
The total amount of enzyme was added at 0 hour. The substrate was dissolved in a solution containing the internal standard needed in the subsequent sugar determinations by gas-liquid chromatography (20 pg of i-inositol per 25 ~1 in glassdistilled water) so that 20 pg of inositol were present per hydrolysis or per aliquot.
The enzyme solution was then added and the mixture diluted, if necessary, to 100 ~1 per hydrolysis or per aliquot.
Hydrolyses were at 37" with the enzymes added and dilutions made in 0.1 M sodium citrate, pH 6.0, 0.02%,. in NaNa.
Isolation and Determination of Monosaccharides Released from Glycoproteins and Glycopeptides by Glycosidases-To isolate the released monosaccharides the hydrolysate was diluted with water to about 15 ml in an Amicon model lo-PA ultrafiltration cell and ultrafiltered through a PM-10 membrane, 10,000 molecular weight cutoff (Amicon Corp., Lexington, Mass.), to remove most of the protein.
The retentate was discarded and the ultrafiltrate concentrated to about 1 ml by rotary evaporation, then passed through three columns, 0.5 x 5 cm, stacked on top of one another, each with a glass wool plug at the bottom.
The columns were washed three times with 1.0 ml of water each time.
This treatment desalts the mixture and removes any glycopeptide substrate or remaining protein.
The eluate and washings were pooled, concentrated by rotary evaporation, and hydrolyzed in 1 ml of 2.0 N HCl in a sealed, evacuated tube at 100" for 2 hours to deacetylate any N-acetylhexosamines released. The hydrolysate is dried by rotary evaporation, redissolved in 0.5 ml of water, and passed through about 200 mg of Dowex 50-X12 (200 to 400 mesh) (H+ form) to remove the free hexosamines which are later eluted from the resin with 3 ml of 6 N HCI and analyzed on an amino acid analyzer.
The neutral sugars are converted to their alditol acetates and analyzed by gas-liquid chromatography by the method of Kim et al. (31).

RESULTS
Purijication of Glycosidases-The procedure results not only in purification of the a-fucosidase but separates the other activities reasonably well. The results are summarized in Table I. The fucosidase, essentially free of the other glycosidase activities, except for N-acetylglucosaminidase activity, was found to be in the last fraction eluted from DEAE-cellulose.
Based on the activity of the homogenate, a 180-fold purification was achieved (12% yield of fucosidase activity).
Addition of 0.1 M NaCl to the extraction buffer greatly improves the yield of a-mannosidase as reported by Snaith and Levvy (12). The yield of a-mannosidase is also sensitive to the pH of the extraction buffer, the optimum being at about pH 6. The yields of the other enzymes are not appreciably affected by these parameters in the pH range 5 to 6. The step of heating at 60" precipitates approximately half of the protein present in the 37" homogenate.
Adjustment of the pH to 5.25 prior to heating at 60" is a compromise which gives the best yields for all four enzymes. It was also found that adjustment of the pH to 6.3 improved the ammonium sulfate fractionation and increased the amount of a-mannosidase precipitating between 50 and 60% saturation.
In order to separate the a-fucosidase, &V-acetylglucosaminidase, and P-galactosidase, the 35 to 50% ammonium sulfate fraction is chromatographed on CM-cellulose. The results are shown in Fig. la. The /%galactosidase was recovered from Fraction (XV-1 and was contaminated with about 12% cr-fucosidase and small amounts of the other enzymes (based on activity against nitrophenyl glycoside substrates). Fraction CM-II contained the ol-fucosidase and P-N-acetylglucosaminidase and was rechromatographed on DEAE-cellulose. The cY-mannosidase activity is largely lost during the chromatography on CM-cellulose but that remaining (less than 10yc) also emerges in this fraction. The results of the chromatography on DEAE-cellulose are shown in Fig. lb. It can be seen that, as measured by absorbance at 280 nm, most of the protein emerged with the solvent front (Frac-tion I) which also contained the P-N-acetylglucosaminidase contaminated with a-mannosidase and fl-N-acetylglucosaminidase contaminated with cY-mannisodase and ,&galactosidase. The protein peak corresponding to the fucosidase was just detectable by the absorption meter used (Gilson UV-280 IF) but coincided with the enzymic activity.

Inactivation of Contaminating
Activities-Each enzyme preparation obtained by chromatography is contaminated with varying amounts of other activities.
Therefore, inactivation of the contaminants was attempted. The best treatment found thus far for each enzyme and the results obtained are summarized in Table II. Treatment of the P-galactosidase for 2 hours at 37" in 0.1 M sodium citrate, pH 3.25, 0.02% in NaN3 (see Fig. 2a) is sufficient to inactivate most of the contaminating activities with 60% recovery of P-galactosidase activity.
The P-N-acetylglucosaminidase is less sensitive to inactivation by cupric acetate than are the other enzymes and treatment with 5 mM cupric acetate in 0.1 M sodium acetate, pH 5.5, 0.02% in NaN3 (2 mM copper concentration) at 25" overnight inactivates much of the contaminating activity in Fraction DEAE-I with no loss of /3glucosaminidase activity.
At pH 6 the a-fucosidase is more heat stable than are the other enzymes (see Fig. 2b). Heating of this enzyme at 70" for 15 min in 0.1 M sodium citrate, pH 6.0,0.02% in NaN3 inactivates much of the contaminating activities remaining in the preparation, particularly /3-N-acetylglucosaminidase, while 90yc of the a-fucosidase activity is recovered.
The cr-mannosidase is obtained in the 50 to 60% ammonium sulfate fraction and although contaminating a-fucosidase was easily inactivated by treatment with 5 IIIM ZnCh in acetate buffer, repeated attempts to inactivate the other activities were unsuccessful. While it would seem feasible to inactivate both /3-galactosidase and fl-N-acetylglucosaminidase with good recovery of a-mannosidase by heating at 70" in citrate buffer at pH 6.0 or 6.5 (see Fig. 2b), attempts to do so with the undiluted enzyme in both citrate and acetate buffers resulted in almost complete loss of cr-mannosidase activity as well as of the other activities.
Glycosidase Stabilit&--The stabilities of the four enzymes with respect to pH were determined at 37 and 70" on the several fractions and the results are shown in Fig. 2 Aliquots of 25 ~1 of the effluent were assayed for a-fucosidase (---) and P-,I;\;-acetylglucosnminidase (--).
Units indicated are milliunits. In b the absorbance at 280 nm is given by (. . This experiment was carried out with both the isolated enzyme fractions and with the 35 to 50% ammonium sulfat,e fraction; the results for the isolated fractions arc given in Fig. 3 and show that all but the mannosidase retain good proportions of their activity with the glucosaminidase the most stable. The results with tht 35 to 50% ammonium sulfate fraction mere essentially the same except that the ar-fucosidase retained 80':; of its activity after 200 hours and the P-Nacetylglucosaminidase and fl-galactosidase retained 950/,. The instability of the ac-mannosidase is in agreement with the findings of Conchie and Hay (11) and &with and Levvy (12). The nietal conccntra-Salts of heavy metals, -Ig+, Hg++, CL?, Cd++, Co++, Xi++, and Pb++, are generally quite effective inhibitors of the oc-fucosidase and oc-mannosidase in the presence' of ncctate except fol Co++ in the case of the ar-fucosidase.
The P-g:llactosidasc and /3-N-acetylglucosaminidase are inhibited only by -\g+, Jig++ and Cu++ and the P-X-acctylglucosaminidase is somewhat resistant to inhibition by CU+T. Zinc acetate inhibits the cr-fucosidase greater than 90% but has little or no effect on the other enzymes. Aminoff and Furukawa have reported that the cr-(1 ,2)-fucosidase from Clostridium perjringens is likewise inhibited by zinc (2). $fect of Metals and Chelators-The enzymes (35 to 50% ammonium sulfate fraction) were incubated at room temperature for 20 hours in both acetate and citrate buffers at pH 6 containing No significant inhibition was observed with the chelators either singly or in combination even if the enzyme preparation was (10) and Conchie and Hay (11) except that the latter authors redialyzed against acetate buffer containing 25 InM EDTA.
These port a pH optimum of 5.0 for cY-mannosidase. The pH optimum results, as they concern the a-mannosidase, appear to conflict of the /3-N-acetylglucosaminidase has not been previously rewith the data of Snaith and Levvy (12). However, their data ported. are quite compelling, suggesting that the native enzyme binds Isoelectric Points-A crude enzyme fraction precipitating bezinc very tightly but that they may have labilized their enzyme tween 0 and 50% of ammonium sulfate saturation was subjected preparation with respect to the zinc by treatment with pyridine to isoelectric focusing by the method of Vesterberg and Svensson or acetone.
A control enzyme-ampholine solution a stored at room temperature during the electrofocusing run retained full activitv of all the enzymes.
The low vield of cx-mannosidase recovered is probably due to instability at its isoelectric points (see Fig. 2 Gel electrophoresis of the native material was difficult. As shown in Fig. 6b, the protein did not readily penetrate a 7% polyacrylamide gel at pH 9.5 and may have dissociated in the process. In 3% gels at pH 9.5 no staining was observed, indicating that the protein was not fixed in gels of this porosity.
Successful runs PH were obtained in 5.6YC gels at pH 6.6, although an artifact of the Units in the plot are normalized. electrophoretic system appeared. After 2 hours of electrophoresis, penetration of the protein was again limited to the top 10 mm of the gel, but after 3 hours results as given in Fig. 6c were consistently found.
One protein component was seen, indicated by the arrow, and enzyme activity was found exclusively in the segments of the gel containing this band. The sharp band seen directly below the protein and in the control is an artifact of the system.a After runs of 2 hours, enzyme activity also coincided with the protein band at the top of the gel.
The enzyme was subjected to electrophoresis in polyacrylamide gels in the presence of 0.1 $ZO SDS. With 20 pg of material, two major bands were seen. Molecular weights of 49,700 and 53,700 were calculated. These results substantiate those found in the ultracentrifuge with regard to the subunit nature of the enzyme. At a higher concentration, 30 pg of material, one broad band was seen with an indicated molecular weight of 53,900. In each gel a very faint band was also observed with a molecular weight of about 130,000. The results are shown in Fig. 6~.
Although the amounts of the fucosidase available were insufhcient for detailed chemical studies, the centrifugal and electrophoretic studies indicated that the enzyme preparation was homogeneous or nearly so. Thus, the results of a preliminary analysis of the amino acid and carbohydrate content are shown in Table III. The material is glycoprotein in nature (0.6% carbohydrate) and it is relatively rich in leucine and histidine as compared to many proteins.
Hydrolysis of Glycoprotein and Glycopeptide Oligosaccharides with Rat Epididymal GlycosidasesA variety of substrates are hydrolyzed with the 35 to 50% ammonium sulfate fraction for 72 hours at 37". The results are given in Table IV. It should be 3 An artifactual band formed 45 min after electrophoresis began in the control and 15 to 30 min later in the sample tubes. Thus the artifact moved further in the controls.
It was yellow in color before staining with the protein stain. The fucosidase solution (0.6 mg per ml) contained no trace of yellow color. The artifact band was not eliminated by addition of mercaptoethanol (lo/, a freshly purchased supply (Eastman) of acrylamide or the substitution of 4% glycerol for the 40% sucrose present in the sample solution when introduced on to the gel. The same reagents at pH 9 to 9.5 do not result in production of the artifact.
FIG. 6. a, polyacrylamide gel electrophoresis of the purified cr-fucosidase in the presence of SDS. Electrophoresis was for 7) hours at 6 ma per tube (27 volts) in a 5% a&amide gel containing 0.1% SDS. One gel (left), with 20 pg of a-fucosidase, shows two major bands with apparent molecular weights of 49,700 and 53,700. The second gel (right), with 30 pg of cr-fucosidase shows a single broad band with an apparent molecular weight of 53,900. A faint band (arrow) with a molecular weight of 130,069 was also observed in each gel. 5, polyacrylamide gel electrophoresis of the purified or-fucosidase. Electrophoresis was for 1, 3, and 6 hours (left to right) at 2.5 ma per tube in a 7% acrylamide gel. Gel buffer was 0.125 M Tris-phosphate, pH 9.0; running buffer was 0.012 M sodium glycinate, pH 9.5. Each gel contained 60 pg of protein.
c, electrophoresis of the native fucosidase at pH 6.6 for 3 hours at 2 ma per tube in a 5.6% acrylamide gel. Gel buffer was 0.095 M Trisphosphate, pH 6.6; running buffer was 0.012 M sodium glycinate, pH 6.6. On the left is a control with the enzyme preparation omitted.
On the right, 30 fig of enzyme were subjected to electrophoresis. The sharp band in both gels is an artifact (see text); the arrow indicates the single band of protein observed in the sample.

III
Amino acid and carbohydrate composition of purified cu-jucosidase The calculation of glucosamine was based on assumed 90% recovery.   Time studies of the hydrolyses were carried out with the purified a-fucosidase (Fraction DEAE-III) on the LH-/3 peptide and the horse Fc peptide with the results shown in Fig. 7. In the case of the LH-/3 peptide no sugar other than fucose was released and removal of this sugar was complete in 200 hours, thus showing that all the fucose is present in a nonreducing terminal position and is a-l linked.
In addition to fucose, N-acetylglucosamine and galactose were released from the horse Fc peptide. The amounts of these sugars released were surprisingly large considering that these activities, as measured against synthetic substrates, were only 3.8 and 0.25a/ respectively, of the fucosidase activity.
The N-acetylglucosamine released represented about one-third of the total (1.3 of 4.0 residues).
About two-thirds of the galactose was also released; no release of mannose was observed. Again, release of fucose was complete in 100 hours. Work by Rothfus and Smith (33) is also consistent with the view that fucose and galactose are terminal residues in the homologous human peptide. It is evident that, while the cr-fucosidase preparation is sufficiently free of other glycosidases to be useful in cases such as the LH-/3 peptide, the small amounts of contaminating B-N-acetvlglucosaminidase (3.8% based on activitv against noted that as with many similar studies with glycosidases from other sources, release of sugars is not complete even after 72 hours. Most evident is the difference between the glycopeptides and the native glycoproteins in terms of their suitability as substrates. The oligosaccharides of the peptides are hydrolyzed fairly extensively; however, there are interesting differences in the extent of hydrolysis between the various peptides.
The I% peptides are all hydrolyzed to a similar degree, particularly with respect _ " -\ ,-I u to mannose and N-acetylglucosamine, thus indicating that their oligosaccharide structures are basically the same. The L peptide from horse immunoglobulin and the LH-@-peptide, which have very different compositions, yield results which are distinct from those of the Fc peptides.
It should be noted that the LH-fl glycopeptide contains no galactose but does contain 2 moles of N-acetylgalactosamine per mole of peptide, none of which was released.
In contrast, the native glycoproteins are generally quite resistant to glycosidase hydrolysis although galactose was released from human immunoglobulin G and about two-thirds of a residue of N-acetylglucosamine is released from each of the immunoglobulins.
Small amounts of mannose and N-acetylglucosamine were released from native LH, together with a trace of N-acetylgalactosamine.
The isolated /3 chain of LH was not hydrolyzed at all, indicating that the small amount of carbohydrate released from native LH probably originated in the (II chain. Issue of January 10, 1972 R. B. Carlsen and J. G. Pierce 31 synthetic substrates) and /3-galactosidase (0.25%) in the preparation also cause release of appreciable quantities of these two monosaccharides in other cases. Further attempts should be made to inactivate the residual contamininants. Hydrolyses with the cu-fucosidase after heating at 70" were not carried out because this preparation was still contaminated with 0.5% /3-Nacetylglucosaminidase and 0.2% fi-galactosidase.

DISCUSSION
The data concern two major aspects of the cr-n-fucosidase from rat epididymis, its physical and chemical properties and its suitability for use in studies on the structures of the oligosaccharide moieties of glycoproteins.
The epididymis has been shown to contain a number of different glycosidase activities, principally ol-fucosidase, a-mannosidase, P-N-acetylglucosaminidase, and &galactosidase.
It also contains very low levels (1 to 2y0 or less of the activity of the four principal epididymal glycosidases) of cr-and ,6-glucosidase, cu-galactosidase, and fl-mannosidase.
It is lacking /3-fucosidase, /3-xylosidase, and may have a trace of CY-Nacetylglucosaminidase activity (10, 13). Of these enzymes, the cu-mannosidase is the only one which has previously been obtained free of other glycosidase activities (12). The work described herein is the first purification of a mammalian fucosidase lvhich, of the epididymal enzymes, appears to be the most useful as an addition to the glycosidases, from other sources, which are valuable in structural studies. The nonmammalian cr-fucosidases which have been isolated are characterized as 1,2-a-~fucosidases and seem to require a fucosyl-galactose bond (1, 2); thus their specificity is very narrow.
The epididymal fucosidase has a broader specificity as it is active against both synthetic substrate (10) and against a variety of glycoprotein oligosaccharides (18,19). Further, it is active against, an oligosaccharide (from LH-/3) which contains no galactose.
In the course of purifying the cy-fucosidase, the separation and partial purification of the other major epididymal glycosidases also resulted; the latter may also be useful after further purificatiou. All are active against glycopeptide substrates (18,19,34). However, with the exception of the a-fucosidase, which has a specificity different from the other oc-fucosidases described, enzymes which are probably equally effective have recently become available from other sources in apparently pure form (3-Q). An important observation made in this study is that the epididymides, in 30% glycerol at -2O", can be stored for at least 2 years without significant losses of enzyme activity.
Further, the enzymes, either in crude extracts or as the purified fractions, may be stored in solution at 4" for 3 months with less than a 10% loss in any activity.4 The preparation of the crude extract is quite straightforward, involving only homogenization and incubation of the homogenate at 37" to release the enzymes as described by Levvy and McAllan (lo), followed by a heat step and fractionation with ammonium sulfate. Large amounts of inert protein are removed by the heat treatment.
The crude extract is then subjected to chromatography on CM-and DEAEcelluloses. There were two major difficulties in developing the chromatographic procedures; first, the enzymes are unstable be-10~ an ionic strength of about 0.05 and, secondly, the cl-fucosidase is unstable outside the pH range 5 to 7. In practice it was neces- sary to restrict the pH range for column chromatography to 5.5 to 6.5 for the best recovery of the fucosidase.
Because several of the enzymes do not bind strongly to cellulose ion exchangers under such conditions of ionic strength and near-neutral pH, the performance of the columns is fairly sensitive to alterations of conditions, although, after suitable conditions were found, the final chromatographic fractionation was simple and reproducible. Finally, except for the a-mannosidase, which is the least stable of these enzymes, much of the contaminating glycosidic activities in the individual glycosidase preparations can be inactivated. However, none of these procedures is completely effective and further work on inactivation or improvement of the chromatography is required.
The purified cr-fucosidase, while still contaminated with small amounts of other glycosidase activities, appeared to be essentially homogeneous by the criterion of gel electrophoresis at pH 6.6 (as well as by the less rigorous centrifugal criterion).
Only a single protein band was observed and the fucosidase activity coincided with the protein.
The material dissociates in the presence of 6 M guanidine or sodium dodecyl sulfate and the ultracentrifugal and electrophoretic measurements suggest that the native enzyme is a tetramer.
Further studies are needed to establish conclusively that the subunits consist of two pairs of unlike chains with molecular weights of 47,000 and 60,000 as indicated by the data. The analyses of composition show that the enzyme is glycoprotein containing 0.6Y0 carbohydrate. Other glycosidases that have been chemically characterized include P-galactosidase from E. coli (35) and the P-N-acetylglucosaminidase from Aspergilhs oryzae (36) ; the latter is also a glycoprotein.
Physical and chemical characterization of the epididymal mannosidase (12) and of the bacterial fucosidases (1, 2) have not yet been reported.
The fucosidase is primarily of interest in terms of its ability to hydrolyze glycoprotein oligosaccharides.
The obvious difference in the degree of hydrolysis of peptides as opposed to that of native glycoproteins probably results from steric hindrance, with smaller substrates strongly preferred.
This appears to be the case with many glycosidases (5,9,37).
Hydrolyses of glycopeptides with crude glycosidase preparations can also be of use in structural comparisons between oligosaccharide moieties. This is illustrated by the hydrolyses of the two types of glycopeptide obtained from horse immunoglobulin G. The amino acid compositions of these two peptides are very different while their carbohydrate compositions are almost the sa.me, differing by only a single residue of galactose.
Since these 2 carbohydrate units occur in the same molecule it is of interest to know whether or not they are structurally homologous with one perhaps containing an extra terminal galactose.
The gross differences observed between the two peptides in the amounts of mannose and Nacetylglucosamine released suggest significant structural differences between them, although it is possible that differences in amino acid sequence might have affected the hydrolysis.5 The time studies of the hydrolyses of the LH-0 peptide and horse Fc peptide demonstrate that the fucose is completely removed from both of these dissimilar peptides by the purified cu-fucosidase and so the fucose in these structures must be a-l linked.
It was surprising, however, that the traces of &N-acetylglucosaminidase and fl-galactosidase activity exhibited against 6 A paper reporting on the structures of glycopeptides from human myeloma proteins has just appeared (38). A partially purified preparation of the fucosidase was used in conjunction with other glycosidases.