Processing of Carbohydrate Units of Glycoproteins CHARACTERIZATION OF A THYROID GLUCOSIDASE*

The role of glucose-containing dolichyl pyrophos- phate-linked oligosaccharides in protein glycosylation by thyroid microsomes described an accompany- ing oligosaccharide transfer a of glycoprotein investigation

Glucose was found to be essential for oligosaccharide transfer but a loss of this sugar from the newly synthesized glycoprotein was noted shortly thereafter.
The results of the present investigation demonstrate the occurrence in calf thyroid microsomes of an enzyme which can selectively release glucose from the dolichollinked oligosaccharide (K,,, = 7 x 10T6 M) or from the oligosaccharide after transfer to endogenous protein. This glucosidase was less active towards the free oligosaccharide and towards substrate from which the peripheral mannose residues had been removed by (Ymannosidase treatment. Enzyme activity was solubilized by Triton X-100, was maximal between pH 6.0 and pH 7.0, and was unaffected by EDTA. Marked stimulation of glucose release was observed in the presence of the antichaotropic ions, SO'-and POZ2-. The specificity of the enzyme as well as its neutral pH optima and low K,,, value distinguished it from other thyroid glucosidases. A large number of LY-and P-glucosides were ineffective as inhibitors although an a-linked glucose tetrasaccharide from nigeran manifested some inhibitory action.
About 85% of the glucose could be released from the oligosaccharide-lipid by the glucosidase without scission of any other bonds and after such treatment the oligosaccharide portion became much more susceptible to a-mannosidase digestion.
Since the glucosidic linkages in the oligosaccharide were resistant to cleavage by glucosidases of known anomeric specificities, chromic acid oxidation studies were performed and indicated that these bonds are (Y-in configuration. Information obtained from the action of the thyroid glucosidase as well as periodate oxidation studies suggested that the glucose residues of the oligosaccharidelipid are attached to the polymannose portion primarily in the form of an a-Glc-(1 + 3)-Glc disaccharide although some lipid-saccharide molecules apparently contain single glucose residues and a small number have glucose as a component of an cu-Man-(1 -+ 3)-Glc disaccharide.
Studies with thyroid microsomes under conditions permitting transfer of oligosaccharide from lipid car-* This work was supported by Grant AM 17477 from the National Institutes of Health. A preliminary report of this work has been published (1). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ To whom correspondence should be addressed at the Elliott P. Joslin Research Laboratory, One Joslin Place, Boston, Mass. 02215. rier to protein, as well as under conditions where this transfer is inhibited, indicated that the glucose released in this system originated from both the oligosaccharide-lipid donor and the glycoprotein product. This suggests the possibility that thyroid glucosidase, in addition to initiating processing of the oligosaccharide after its transfer to protein, may also control the level of glucose-containing oligosaccharide-lipids available for protein glycosylation.
We have previously reported that thyroid slices, as well as those from a number of other tissues, actively synthesize a dolichyl pyrophosphate-linked oligosaccharide which contains glucose in addition to mannose and N-acetylglucosamine residues (2)(3)(4). From pulse-chase studies carried out in slices (2) and more recently from investigations with particulate enzyme, described in an accompanying paper (5) with the thyroid cell-free system indicate that the glucose is essential for effective transfer reactions to take place and appears to play a specifying role in this glycosylation step (5). These studies, as well as investigations carried out in a number of other laboratories (6-8), now strongly suggest that the glucose-containing oligosaccharidelipids function as the physiological donors in the en bloc transfer of carbohydrate to protein which represents the initial step in the biosynthesis of the asparagine-linked saccharide units.
Although glucose is still attached to the mannose-N-acetylglucosamine oligosaccharide moiety after transfer from the lipid carrier to nascent protein, this sugar has not been found as a component of asparagine-linked carbohydrate units of mature glycoproteins (9). This paradox prompted us to postulate that glucose may be enzymatically cleaved from the protein-linked oligosaccharide as part of a series of modifying reactions which lead to the typical N-glycosidically-bound saccharide moieties (2).
In this report, we present evidence that thyroid contains a microsomal enzyme which can selectively release glucose from its attachment to the mannose-N-acetylglucosamine oligosaccharide. The properties of this glucosidase are described and suggest that it may indeed carry out the fist step in a series of processing reactions of the kind which have been proposed for the vesicular stomatitis virus G glycoprotein (8,10,11 (3). After the mannosidase incubations, the digested oligosaccharide-linid was nurified bv DEAE-cellulose chromatography while enzyme-treated oligosaccharide was isolated by filtration on a Bio-Gel P-4 column (3). Oligosaccharide-protein was prepared by enzymatic transfer of 14Clabeled oligosaccharide from oligosaccharide-lipid to endogenous thyroid particle protein.
For this purpose the 14C-labeled oligosaccharidelipid was incubated with the 215,000 X g thyroid particle fraction for 5 min at 25°C under the conditions specified in the accompanying paper (5). The delipidated radiolabeled oligosaccharide-protein from several such incubations was suspended in 0.45% Triton X-100 for use in the glucosidase assays. The a-mannosidase-treated 14C-labeled oligosaccharide-lipid was also transferred to endogenous thyroid particle protein by incubation under the same conditions to yield a mannosidase-treated oligosaccharide-protein.

RESULTS
Assay of Glucosidase Activity-Incubation of the Triton extract of calf thyroid microsomes with 14C-labeled oligosaccharide-lipid resulted in the selective release of radiolabeled glucose as shown in a typical radioscan of a chromatogram used in the assay (Fig. 1, lower). Since the oligosaccharidelipid employed as substrate contained almost 3 times as much radiolabel in its mannose residues as in the glucose, the absence of free mannose in this chromatographic assay served as a sensitive index for excluding the presence of even small levels of mannosidase activity. The radiolabeled component present at the origin represents oligosaccharide split from the lipid during the deproteinization step and could also be observed when the substrate was incubated without enzyme ( Fig. 1, upper). Solubilization of Particulate Glucosidase-Incubation of thyroid microsomal particles with various concentrations of Triton X-100 indicated that maximal release of glucose from oligosaccharide-lipid was achieved at a detergent concentration of 0.45% (v/v) (Fig. 2). Treatment of the microsomal fraction with this concentration of Triton brought about 40% of the protein into solution and yielded a supernatant with a specific enzyme activity about 3 times greater than that of the unextracted membranes. This T&on-solubilized fraction was used to study in detail the properties of the microsomal glucosidase. AS far as could be ascertained, the glucosidase activity in the Triton supernatant was qualitatively similar to that in the whole microsomal pellet and moreover was free of mannosidase action towards oligosaccharide-lipid, an activity which was occasionally detected after prolonged incubations with the unextracted membranes. Properties of Glucosidase-The release of glucose by the Triton-solubilized microsomal enzyme increased with the length of incubation (Fig. 3) and the amount of protein (Fig.  4). After 3 h of incubation, about half of the glucose in the substrate was cleaved while 73% was in the free form at 6 h ( Fig. 3). More prolonged incubations (greater than 24 h) led to the liberation of up to 85% of the glucose in the oligosaccharide-lipid.
Glucosidase activity was maximal between pH 6.0 and 7.0 and fell off sharply in the acidic range (Fig. 5).
Some other properties of the enzyme are shown in Table I. Enzyme activity was unaffected by the presence of EDTA and was not influenced by the addition of manganese or magnesium. Glucose, even in high concentration, was not inhibitory. While perchlorate, a chaotropic ion, eliminated enzyme activity, a remarkable stimulation was observed with the antichaotropic ions (24), sulfate and phosphate (Table I). Maximal activation of the glucosidase with (NH&SO., was observed at a concentration of about 1.0 M (Fig. 6), while phosphate buffer at pH 7.0 brought about a similar peak stimulation at 1.5 M which represents an HP04'-concentration of 0.92 M. Indeed, because of this observed enhancement of enzyme activity, ammonium sulfate (0.95 M) was incorporated into the standard enzyme assay. Stimulation of glucosidase by (NH&SO4 was also observed when the whole microsomal pellet was used as enzyme in the incubations.
The effect of increasing concentrations of oligosaccharidelipid on glucosidase activity is shown in Fig. 7. From this data a K,,, value of 7.1 X lo-" M was calculated.
The Triton-solubilized glucosidase when stored at -20°C lost about 5% of its activity per month.
Substrate Specificity-A comparison of the activity of the thyroid glucosidase towards the 14C-labeled oligosaccharidelipid in its intact and modified forms indicated that the enzyme acted optimally on the native lipid-saccharide (Table II). Removal of the peripheral mannose residues by a-mannosidase treatment substantially reduced the action of the glucosidase on the lipid-bound oligosaccharide. Moreover, the lipidfree oligosaccharide either in its native or a-mannosidase- The enzyme also acted on the oligosaccharide after transfer to protein (Table II). Glycopeptides prepared from this oligosaccharide-protein demonstrated reduced activity, and protein glycosylated with an cu-mannosidase-treated oligosaccharide-lipid was an even less effective substrate for the glucosidase.
Comparison with Other Thyroid Glucosidase Actiuities-The rather unusual specificities of the glucosidase, as well as the low K, value determined for the oligosaccharide-lipid, suggested that the enzyme was distinct from various other known glucosidases. Indeed, a comparison of the K,,, values of various glucosidase activities in the Triton extract of thyroid microsomes indicated that the value for the oligosaccharidelipid was about 2 to 3 orders of magnitude lower than that for a variety of other substrates (Table III). Moreover, the ratio of glucosidase activity at neutral to acidic pH was considerably higher for the lipid-bound oligosaccharide than for the other compounds examined. The ratio of glucosidase activity at pH 7.0 to pH 4.2 when the oligosaccharide-protein was used as substrate was similar to that for the oligosaccharide-lipid (10.2 compared to 11.8).
Characterization of the Products of Glucosidase Treatment-under the conditions of the glucosidase assay, no transfer of oligosaccharide to endogenous protein took place and all the radioactivity was recovered in the chloroform/ methanol/water (10:10:3) extract. Upon DEAE-cellulose chromatography of the '%-labeled oligosaccharide-lipid after extended digestion with the thyroid enzyme, it emerged from the column primarily in a position (B) ahead of that observed for the co-chromatographed tritium-labeled oligosaccharidelipid (Fig. 8). However, a small amount of material (C) eluted in the position of the undigested compound. Peak A, which emerged from the column before start of the gradient, con- tained the sugars released by the enzyme, and paper chromatography of the material in that peak revealed that it contained only free [%]glucose ( Fig. 9A). Moreover, paper chromatography of the neutral sugars from an acid hydrolysate of the major oligosaccharide-lipid fraction (Fig. 8, Peak B) indicated that mannose alone was now present (Fig. 9B). In contrast, chromatography of a hydrolysate of the minor component (Fig. 8,Peak C) showed that it still contained 2-some glucose in addition to the mannose (Fig. 90.   1 The oligosaccharide obtained by mild acid hydrolysis of the m DEAE-cellulose-purified, glucosidase-treated oligosaccharideb lipid yielded a single peak on Bio-Gel P-4 filtration which X emerged from the column slightly later than the native oligosaccharide (Fig. 10). The oligosaccharide derived from the H a lglucosidase-treated lipid-saccharide was found to be more 0 susceptible to a-mannosidase digestion than the oligosaccha-I.
ride obtained from the undigested compound (Fig. 11). While Y a-mannosidase, as we have previously reported (3), released only a limited amount of mannose from the native oligosaccharide leaving a large residual polymer (Fig. 11, upper scan) the action of this enzyme on glucosidase-treated material O-resulted in the appearance of most of the radioactivity as free mannose or as smaller oligosaccharides which migrated away from the origin on paper chromatography (Fig. 11, lower scan). Treatment with a-mannosidase of oligosaccharides obtained after partial glucosidase digestion indicated that the amount of mannose released was directly related to the extent of prior glucose removal. in Peaks A, B, and C from DEAE-cellulose column (Fig. 8). The material in Peak A was chromatographed directly while that in Peaks B and C was hydrolyzed with acid and passed through coupled Dowex 50 and Dowex 1 columns before chromatography. Paper strips were scanned for radioactivity after chromatography was performed in Solvent System A for 5 days. The positions of migration of standard glucose (Glc) and mannose (Man) are indicated. The mannosidase digestion was carried out as described under "Experimental Procedures" on equimolaf amounts of oligosaccharides prepared by mild acid hydrolysis from untreated and thyroid-glucosidase-digested, DEAE-purified, '%-labeled oligosaccharides, respectively. The deproteinized and desalted samples were chromatographed on Whatman No. 1 paper in Solvent System B for 24 h and then scanned for radioactivity. The position of migration of standard mannose (Man) is shown. by guest on March 23, 2020 http://www.jbc.org/ Downloaded from follows: almond /?-glucosidase (pH 4.5); yeast a-glucosidases (pH 4.5 and 6.8); rice a-glucosidase (pH 6.8); fungal cy-glucosidases from A. niger (pH 5.0) and A. oryzae (pH 4.5); and hog pancreatic cY-amylase (pH 7.0).
Chromium Trioxide Oxidation-Since the anomeric configuration of the glucose residues in the lipid-linked oligosaccharide could not be determined by the use of glucosidases of established specificity, a CrOa oxidation procedure was employed to distinguish between (Y-and /?-glycosidic bonds (Table IV). The oxidation, when carried out on standard reduced acetylated disaccharides (25), resulted in the expected high recovery of a-linked glucose as in maltose and melibiose but in substantial destruction of the P-linked sugar as in cellobiose and lactose (Table IV).
Chromium-trioxide treatment of the reduced, acetylated, radiolabeled oligosaccharide from the thyroid-lipid intermediate resulted in a 75% yield with a ratio of ['4C]glucose to ['4C]mannose actually slightly higher than in the unoxidized control (Table IV). Since all of the mannose residues of the oligosaccharide except the most internal one are believed to be (Y in anomeric configuration (4,8,26), the glucose-to-mannose ratios observed indicate that the glucose in at least the recovered 75% of the oligosaccharide-lipid molecules are also in a-glycosidic linkages. The presence of one P-linked mannose which would be destroyed by the CrOa oxidation probably accounts for the slight increase in the ratio of glucose-tomannose radioactivity compared to the unoxidized control. The incomplete recovery of the CrOz-treated oligosaccharide may be related to the difficulty of applying this procedure to picomole amounts of radiolabeled saccharides (see "Experimental Procedures").
Periodate Oxidation Studies-Although the extensive release of glucose from the oligosaccharide-lipid indicated that this sugar is primarily located in positions external to the mannose residues, the resistance of about 15% of the glucose to enzymatic release suggested that some glucose may be located in a more internal position. To explore this possibility, periodate oxidation was performed on the oligosaccharide before and after cu-mannosidase digestion and these studies indicated that while 53% of the glucose was destroyed in the native oligosaccharide, 66% was susceptible to oxidation in the mannosidase-treated derivative. Although this difference is small, it is consistent with the substitution of about 13% of the glucose residues with peripheral mannose.
Relationship of Glucose Release to Oligosaccharide Transfer to Protein by Thyroid Microsomes-Since the specificity studies indicated that the thyroid glucosidase was capable of releasing glucose from the oligosaccharide-lipid as well as from glycoproteins derived from it, an attempt was made to differentiate between the two activities by incubating intact microsomal membranes with radiolabeled oligosaccharide-lipid under conditions where transfer of oligosaccharide to protein takes place as well as under conditions where this transfer is inhibited (Fig. 12). As reported in the accompanying paper (5), a rapid transfer of oligosaccharide from lipid carrier to endogenous protein takes place in the presence of manganese but this reaction is completely abolished when EDTA is added. While the glucose released in the presence of the cation can originate from both glycoprotein and oligosaccharide-lipid, the glucose liberated during incubation with EDTA would arise from the oligosaccharide-lipid alone. It is evident from the data presented in Fig. 12 that in this microsomal system glucose appears to originate from both potential substrates in comparable amounts as this sugar is released in the absence of glycoprotein synthesis, and its liberation is approximately doubled when transfer of oligosaccharide to protein occurs. DISCUSSION It has become apparent in recent years that glucose-containing dolichyl pyrophosphate-linked oligosaccharides are