JOURNAL OF BACTERIOLOGY, OCt. 1984, p. 67-75 Vol. 160, No. 10021-9193/84/100067-09$02.00/0Copyright © 1984, American Society for Microbiology

Effect of Castanospermine on the Structure and Secretion ofGlycoprotein Enzymes in Aspergillus fumigatusALAN D. ELBEIN,* MICHAEL MITCHELL, AND RUSSELL J. MOLYNEUX

Department of Biochemistry, University of Texas Health Science Center, San Antonio, Texas 78284; and Natural ProductsChemistry Research Unit, Western Regional Research Center, Agricultural Research Service, U.S. Department of

Agriculture, Berkeley, California 94710

Received 1 March 1984/Accepted 4 July 1984

Aspergillusfumigatus secretes a number of glycosidases into the culture medium when the cells are grown in amineral salts medium containing guar flour (a galactomannan) as the carbon source. At least some of theseglycosidases have been reported to be glycoproteins having N-linked oligosaccharides. In this study, weexamined the effect of the glycoprotein processing inhibitor, castanospermine, on the structures of the N-linkedoligosaccharides and on the secretion of various glycosidases. Cells were grown in the presence of variousamounts of castanospermine; at different times of growth, samples of the media were removed for themeasurement of enzymatic activity. Of the three glycosidases assayed, I-hexosaminidase was most sensitive tocastanospermine; and its activity was depressed 30 to 40% at 100 ,ug of alkaloid per ml and even more at higheralkaloid concentrations. On the other hand, ,I-galactosidase activity was hardly diminished at castanosperminelevels of up to 1 mg/ml, but significant inhibition was observed at 2 mg/ml. I-Galactosidase was intermediate insensitivity. Cells were grown in the presence or absence of castanospermine and labeled with [2-3H]mannose,[6-3H]glucosamine, or [1-3H]galactose to label the sugar portion of the glycoproteins. The secreted glycopro-teins were digested with pronase to obtain glycopeptides, and these were identified on Bio-Gel P-4 (Bio-RadLaboratories). The glycopeptides were then digested with endoglucosaminidase H to release the peptide portionof susceptible structures, and the released oligosaccharides were reisolated and identified on Bio-Gel P-4. Theoligosaccharides from control and castanospermine-grown cells were identified by a combination of enzymaticand chemical studies. In control cells, the oligosaccharide appeared to be mostly Man8GlcNAc andMan9GlcNAc, whereas in the presence of alkaloid, the major structures were Glc3Man7GlcNAc andGlc3Man8GlcNAc. These data fit previous observations that castanospermine inhibits glucosidase I.

Aspergillus spp. produce a number of extracellular glyco-sidases, all of which appear to be glycoproteins. Includedamong these enzymes are oa-mannosidase (39), ,-N-acetyl-hexosaminidase (25), P-glucosidase (30), a-glucosidase (31),a-galactosidase (2), a-fucosidase (17), 3-galactosidase (1), I-mannosidase (9), cellulase (16), and so on. Several of theseenzymes have been highly purified, and carbohydrate analy-sis has demonstrated the presence of mannose and N-acetylglucosamine (GlcNAc) as the major sugars. Since theglycosylation and secretion of several of these glycoproteinswere shown to be inhibited by the antibiotic tunicamycin(34), it seems likely that these enzymes contain N-linkedhigh-mannose chains (22).

In animal cells, the oligosaccharide chains of the N-linkedglycoproteins are biosynthesized via a lipid-mediated path-way whereby the sugars GlcNAc, mannose, and glucose aretransferred to dolichyl-phosphate to form a Glc3Man9Glc-NAc2-pyrophosphoryl-dolichol (7, 37). This lipid-linked sac-charide is the donor of oligosaccharide to protein to form theN-linked glycoprotein, i.e., Glc3Man9GlcNAc2-protein (15).Once this glycoprotein has been formed, the oligosaccharidechain may undergo a number of processing reactions to giverise to either high-mannose, hybrid, or complex types ofoligosaccharides (33). The initial processing reactions in-volve the removal of all three glucose residues. Thus,glucosidase I, a membrane-bound enzyme that is located inthe rough endoplasmic reticulum, removes the terminalal,2-linked glucose (3, 4, 10, 26), whereas glucosidase II,another membrane-bound glucosidase that may also be in

* Corresponding author.

the endoplasmic reticulum, removes the two remaining al,3-linked glucose units (35, 43, 44). These reactions give riseto a MangGlcNAc2-protein that may be the direct precursorto the high-mannose glycoproteins. Or, this oligosaccharidemay be further trimmed by the removal of some mannoseresidues to give other, shorter high-mannose structures or toeventually give rise to hybrid and complex structures (15).One useful technique to study biosynthesis and function of

the oligosaccharide portion of the glycoprotein is throughthe use of inhibitors that either prevent glycosylation of theprotein or modify the structure of the oligosaccharide (8). Anexample of the latter type of inhibitor is the plant alkaloidcastanospermine (14). We have found that this alkaloid is afairly specific inhibitor of glucosidases (32), and that itinhibits the glycoprotein-processing enzyme glucosidase I(29). Thus, when influenza virus is raised in kidney cells inthe presence of castanospermine, the viral glyco-proteins contain oligosaccharides of the compositionGlc3Man7_9GlcNAc2, ratherthan the typical high-mannose andcomplex chains found in this hemagglutinin (29). Sincecastanospermine inhibits normal processing in animal andplant cells (H. Hori, Y. T. Pan, R. J. Molyneux, and A. D.Elbein, Arch. Biochem. Biophys., in press), it was ofinterest to determine what effect it would have on theoligosacccharide structure of the Aspergillus spp. glycopro-teins, and whether it would alter the secretion of theseenzymes. In this paper we describe the results of thesestudies.

MATERIALS AND METHODS

Materials. p-Nitrophenyl-glycosides were used as sub-strates for the various glycosidases and were purchased from

67

on August 6, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

68 ELBEIN, MITCHELL, AND MOLYNEUX

Sigma Chemical Co., St. Louis, Mo. [2-3H]Mannose (25 Ci/mmol) and [1-3H]galactose (12 mCi/mmol) were obtainedfrom Pathfinders Laboratories, St. Louis, Mo., and [6-3H]glucosamine (19 Ci/mmol) was from New England Nucle-ar Corp., Boston, Mass. Castanospermine was isolated in0.3% yield from the seeds of Castanospermum australe andwas crystallized from ethanol (14). Bio-Gel P-4 (200 meshand -400 mesh) were purchased from Bio-Rad Labora-tories, Richmond, Calif. Guar flour was obtained fromGeneral Mills Chemicals Inc., Minneapolis, Minn. Endo-,B-N-acetylglucosaminidase H (Endo H) was from Health Re-search Inc., Albany, N.Y., and jack bean ac-mannosidasewas from Sigma.Growth conditions. Aspergillus niger was grown at 30°C in

a liquid medium that has the following composition (in gramsper liter): KH2PO4, 2; (NH4)2SO4, 1.4; urea, 0.3; CaCl2, 0.3;MgSO4, 0.3; mannose, 0.1; yeast extract, 0.05; and guarflour, 5.The fungus was maintained on agar slants of the above

medium. For innoculation of flasks, a loop of spores wasremoved from the slant and dispersed in 2 ml of steriledistilled water. Samples of this suspension were pipettedinto 125-ml flasks containing 25 ml of the liquid medium.Various amounts of castanospermine, sterilized by filtration(filters from Millipore Corp., Bedford, Mass.), were addedto the flasks as indicated below. Radioactive sugars werealso added to some flasks to label the glycoproteins. Theflasks were placed on a rotary shaker and allowed toincubate for up to 144 h. Samples of the medium wereremoved at various times and examined for the activity of anumber of glycosidases.

Assay of glycosidases. Each sample of the medium (i.e.,various time points and various castanospermine concentra-tions) was assayed for the activity of several differentglycosidases. The reaction mixtures for these assays con-tained the following components in a final volume of 0.4 ml:2 p.mol of the appropriate p-nitrophenyl glycoside, 10 ,umolof sodium acetate buffer (pH 5.0), and various amounts ofthe medium. Several samples of medium were selected thatgave linear responses of activity. Incubation times wereusually 30 min at 37°C, but an appropriate time was selectedthat was in the linear range of enzyme activity. At the end ofthe incubation, the reaction was stopped by the addition of2.5 ml of 0.4 M glycine buffer (pH 10.4), and the amount ofliberated p-nitrophenol was measured at 410 nm.

Preparation of radioactive glycopeptides. As indicatedabove, flasks containing various amounts of castanosper-mine and control flasks were innoculated with variousradioactive sugars to label the glycoproteins. In these experi-ments, the entire contents of the flask were removed at theindicated time, usually 120 h, and filtered to remove thecells. The filtrate (about 20 ml) and the cell wash (10 ml)were combined and concentrated in either of the followingways. In some cases, the filtrates were concentrated toabout 2 ml with an Amicon filtration apparatus with a UM 10filter. The 2-ml concentrate was then placed in a tube, and 10ml of ice cold acetone was added. The mixture was allowedto stand overnight at -20°C, and the precipitate was harvest-ed by filtration, dissolved in 2 ml of water, and dialyzedovernight against several liters of 25 mM Tris buffer (pH7.5). In the other case, the filtrates were lyophilized. Thedried material was dissolved in 2 ml of water and dialyzedagainst several liters of 25 mM Tris buffer (pH 7.5). Thisbuffer was used for two reasons. First of all, these fungisecrete cellulases, and dialysis against buffers of low pH mayresult in dissolution of the dialysis bag. Second, the Tris

buffer is the appropriate buffer for the next step, whichinvolves the proteolytic enzyme pronase.

After dialysis for 24 h, the contents of the dialysis bagswere removed and placed in screw-capped tubes. Onemilliliter of pronase solution (5 mg of enzyme per ml in 50mM Tris buffer [pH 7.5] containing 5 mM CaCl2) was addedto each tube, and the mixtures were incubated for 24 h at37°C. At the end of that time, another 1 ml of pronasesolution was added, and incubations were continued foranother 24 h. After the incubation, 2 ml of 25% trichloroace-tic acid was added to each tube, and the mixtures wereplaced in an ice bath for 30 min to precipitate the protein.The protein was removed by centrifugation and discarded,and the supernatants were extracted four or five times with

B - Hexosaminidase

1.26

1. 10 tJ /1gm

0.4

Q0.6 B-Gal. mgml

U~~~~~~

Z 8j- CONT.0.8 B -Gal.

04 0

0

uj0.6 2-g/l-04 3

FIG. 1. Effect of castanospermine on glycosidase activity inmedia. Cells were grown in various concentrations of alkaloid; at thetimes shown, samples of the medium were removed, and theactivities of the various glycosidases were measured by using theappropriate p-nitrophenyl glycoside.

J. BACTERIOL.

on August 6, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

GLYCOPROTEIN ENZYMES IN A. FUMIGATUS 69

ethyl ether to remove the trichloroacetic acid. The aqueouslayers were then concentrated to dryness and separated onBio-Gel P-4 columns.

Gel filtration of radioactive glycopeptides and oligosaccha-rides. The radioactive glycopeptides were separated on a1.5- by 150-cm column of Bio-Gel P-4 (200 mesh). Thecolumn was calibrated with a variety of standard oligosac-charides, and the radioactive materials were run under thesame conditions. Samples were eluted with 0.3% acetic acid,and 1.5-ml fractions were collected. Samples of every otherfraction were removed for the determination of radioactiv-ity. The radioactive peaks were pooled and concentrated to asmall volume. The peaks were then digested with Endo H(see below), and the products of this reaction were rechro-matographed on the same Bio-Gel P-4 column.

Partial characterization of glycopeptides and oligosaccha-rides. Glycopeptides and oligosaccharides were sized on a

1.5- by 200-cm column of Bio-Gel P-4. The column was

calibrated with various oligosaccharide standards includingGlc3Man9GlcNAc, Glc2Man9GlcNAc, GlclMan9GlcNAc,MangGlcNAc, Man8GlcNAc, and Man7GlcNAc. Oligosac-charides were digested with a-mannosidase, and the prod-ucts were rechromatographed on Bio-Gel P-4. Oligosaccha-rides were also subjected to methylation analysis (12), andthe radioactive methylated mannose derivatives were identi-fied by thin-layer chromatography.

Enzymatic digestions. Endo H is an enzyme that cleavessome N-linked oligosaccharides between the two internalGlcNAc residues (40). The specificity of this enzyme re-

quires that the mannose residue that is linked al,6 to the ,3-linked mannose be substituted with an (x,3-linked mannose

0

I-

I-

U

0

ag

20 30 40 50 60 70 8UFRACTION NUMBER

(21). Thus, this enzyme will act on many high-mannose andhybrid chains, but not on complex structures. Digestionswith Endo H were done in 0.2 ml of 50 mM citrate buffer (pH6.5). Enzyme (10 mU) and a few drops of toluene wereadded, and the mixtures were incubated for 24 h. At thattime, another 10 mU of enzyme was added, and incubationswere continued for 24 h. Digestions with jack bean ut-

mannosidase were done in 50 mM sodium acetate buffer (pH5.0) in a final volume of 0.2 ml.

RESULTS

Effect of castanospermine on the secretion of enzymes.Aspergillusfumigatus secretes a number of enzymes into themedium when the organism is grown in a mineral saltsmedium with guar as the carbon source (30, 31). Since a

number of these enzymes appear to be glycoproteins havingN-linked high-mannose oligosaccharides (34), it was of inter-est to determine whether the processing inhibitor, castano-spermine, would have any effect on the synthesis andsecretion of these enzymes.

Various amounts of castanospermine, from 10 p.g/ml up to2 mg/ml, were added to 125-ml flasks containing 25 ml of theguar medium, and the flasks were innoculated with a spore

suspension of the organism. The flasks were placed on a

rotary shaker at room temperature for up to 144 h; every 24h, 2 ml of medium was removed and filtered. The filtrate wasexamined for the presence of a number of glycosidases.Figure 1 presents the results of one such experiment. In thiscase, we compared the activities of 1-hexosaminidase (1-N-acetylhexosaminidase), ,B-galactosidase, and ot-galacto-sidase. There was some difference in the amount of castano-

3C

2C0

'-It1i

>it

5: At

1:

C 20 30 40 50 60 70FRACTION NUMBER

FIG. 2. Effect of castanospermine of the structures of the mannose-labeled glycopeptides and oligosaccharides from secreted glycopro-teins. Cells were grown in castanospermine and labeled with [2-3H]mannose. The medium was removed at 120 h and concentrated on an

Amicon filter, and the protein was precipitated by the addition of 5 volumes of acetone. After standing overnight at -20°C, the proteinprecipitate was isolated and digested with pronase. The profiles in panel A show the elution patterns of the glycopeptides from control cells(upper), cells in 50 ,ug of alkaloid per ml (middle), or cells in 1 mg of alkaloid per ml (lower). The glycopeptide peaks were pooled, digestedwith Endo H, and rechromatographed on Bio-Gel P-4 (B). Standards are shown by arrows as follows: G6, Glc3Man9GlcNAc2; M9,Man9GlcNAc; M5, Man5GlcNAc.

A G3 M9 M5

6 CONTROL4-

2-L II

6-

4-CAST.4 50lgimI2_

I

6-

4CAST.- 1mgjml2-

I IIl%,M A_ 1, ".dt I&l& ru %

B G3 M9 M5

) CONTROLENDO H~~~~~~~I IIII

B CAST.50pg/mI

2 - ENDOH

I I II

3 CAST. 4 4 4I mg/ml

2 ENDO H

*_I I 1

VOL. 160, 1984

F%

on August 6, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

70 ELBEIN, MITCHELL, AND MOLYNEUX

0

)0-

UP

A

160 180 ZW0FRACTION NUMBER

FIG. 3. Determination of size of oligosaccharides from controland castanospermine-grown cells. Oligosaccharides released byEndo H were chromatographed on a calibrated column of Bio-Gel P-4. Standards shown are as follows: 12, Glc3Man9GlcNAc; 11,Glc2Man9GlcNAc; 10, Glc1MangGlcNAc; 9, MangGlcNAc; 8,Man8GlcNAc. Samples of each fraction were counted to determinetheir radioactive content.

spermine required to inhibit the activities of these threeenzymes, as compared with the activities in control cells.Thus, the ,-hexosaminidase (upper panel) began to appear inthe medium of control cells at about 72 to 96 h of growth, andthis activity then increased rapidly over the next 48 h.However, when 100 ,ug of castanospermine per ml wasincluded in the medium, there was a significant decrease inthe activity of this enzyme (about 40% of control values),and this decrease in activity was even more pronounced athigher concentrations of castanospermine.On the other hand, the 3-galactosidase and the cx-galacto-

sidase were much less affected by the presence of castano-spermine (middle and lower curves), although a decrease inthe activities of these enzymes was also observed. Thus, the,-galactosidase was inhibited about 20 to 25% at 1 mg ofcastanospermine per ml and about 35 to 40% at 2 mg of thisalkaloid per ml. The a-galactosidase was even less suscepti-ble to castanospermine, and levels of this enzyme werealmost the same as in control cells at concentrations ofalkaloid up to 1 mg/ml. However, at 2 mg/ml, the activity ofa-galactosidase was depressed about 25 to 30%. Castano-spermine did not inhibit the activities of any of the aboveglycosidases when added directly to incubation mixtures ofenzyme and its p-nitrophenylglycoside substrate.The decrease in activities of these enzymes could be due

to an inhibition or slowdown in the secretion of the proteins,or it could be the result of an inhibition in the synthesis of theglycoproteins. It is also possible that the decreased activitiescould be due to a more rapid turnover of the glycoproteins.Based on the time course studies, there is no reason to

believe that the alkaloid is affecting the turnover of theglycosidases or increasing the degradation.

Alterations in structure of mannose-labeled glycopeptidesinduced by castanospermine. To examine the effect of castan-ospermine on oligosaccharide structure, cells were grown inalkaloid and labeled with [2-3H]mannose. After growth in thelabel for 48 to 72 h, the medium was collected and concen-trated to 2 ml on an Amicon filter. The concentrate wascooled, and 5 volumes of acetone, cooled to -20°C, wasadded to precipitate the protein. After standing for 24 h-20°C, the protein was isolated by centrifugation. Theprecipitate was digested exhaustively with pronase, and theglycopeptides were examined on Bio-Gel P-4. Figure 2Ashows the elution profiles of the mannose-labeled glycopep-tides from control cells and from cells grown in severalconcentrations of castanospermine. Two peaks of radioac-tivity were seen in control and castanospermine-grown cells,but there were some significant differences between thesevarious cells. Thus, in each case (control and castanosper-mine treated), a major peak of radioactivity eluted at frac-tions 38 through 46 (i.e., near the void volume), and thispeak appeared to be similar in control and castanospermine-grown cells. However, the second, smaller peak was clearlydifferent in the presence of alkaloid. In control cells, thissecond peak was rather broad, eluting in fractions 54 through64, whereas at 1 mg of castanospermine per ml, this peakeluted earlier (fractions 48 through 56), indicating that it waslarger in size. This peak was near the Glc3Man9GlcNAc2standard. These data indicated that this alkaloid was causingchanges in the structure of the oligosaccharide chains.The glycopeptides from control and castanospermine-

treated cells were treated with Endo H and rechromato-graphed on the Bio-Gel P-4 column. This enzyme cleaveshigh-mannose oligosaccharides and glycopeptides betweenthe two internal GlcNAc residues, but does not act oncomplex chains or on certain types of high-mannose struc-tures (21). The elution profiles of the Endo H-digestedsamples are shown in Fig. 2B. In each case, the first largepeak did not shift after this enzyme treatment. However,since this peak elutes near the void volume, it may be toolarge to be able to detect the small shift that would be causedby Endo H digestion (i.e., loss of GlcNAc-peptide). Or thispeak may be resistant to this enzyme. This peak was alsoresistant to a-mannosidase, indicating that the mannoseresidues were not a linked or were blocked with othersugars. Because of its large size it seems likely that this peak1 represents the cell wall mannan or polymannan-proteinaggregates. It is not known whether this structure is part ofthe glycosidases or is secreted as a separate polymer.However, its content of radioactivity was unaffected by thealkaloid.On the other hand, peak 2 was susceptible to Endo H in

both control and treated cells. However, the new peakresulting from Endo H was different in control cells ascompared with treated cells (Fig. 2B). In control cells, thenew peak eluted with or just after the MangGlcNAc stan-dard. On the other hand, the peak in castanospermine-growncells eluted earlier and was only slightly smaller than theGlc3Man9GlcNAc standard. Thus, the alterations caused bycastanospermine must be in the oligosaccharide rather thanthe peptide portion of the glycoprotein.

Characterization of mannose-labeled oligosaccharides fromcontrol and castanospermine-treated cells. The mannose-labeled oligosaccharides released by Endo H from controland castanospermine-treated cells were chromatographed ona long calibrated column of Bio-Gel P-4 to determine theirsize. Figure 3 shows the elution profiles of these oligosac-

12 11 109 85-

4_3 CONTROL

2-

I

12. 10 8

5-4 - CAST.

3 -

2 -

II I 1_A_dt010

J. BACTERIOL.

on August 6, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

GLYCOPROTEIN ENZYMES IN A. FUMIGATUS 71

0

%-

o

U-

U

:

(N0

U

W-

I-

U0

FRACTION NUMBER FRACTION NUMBERFIG. 4. Effect of (x-mannosidase digestion on the structures of control and castanospermine-induced oligosaccharides or glycopeptides.

(B) control glycopeptides before (upper) and after (lower) Q-mannosidase digestion. (A) Castanospermine-induced oligosaccharides before(upper) and after (lower) ox-mannosidase digestion. In each case, the mannose-labeled glycopeptide or oligosaccharide was chromatographedon the Bio-Gel P-4 column and then treated with a-mannosidase and reexamined on Bio-Gel P-4. Standards shown by the arrows are asfollows: G3, Glc3MangGlcNAc; Mg, MangGlcNAc; M5, Man5GlcNAc; S, stachyose; M, mannose. Samples of each fraction were counted todetermine their radioactive content.

charides. In both the control cells and the treated cells, theoligosaccharides were not homogeneous, but represented aspectrum of sizes. That is probably not surprising since theseoligosaccharides were derived from cell-secreted glycopro-teins. Nevertheless, the oligosaccharides from control cellswere clearly of lower molecular weight than those fromcastanospermine-grown cells. The major peak in controlcells eluted near the hexose8GlcNAc areas. On the otherhand, the major oligosaccharide from castanospermine-grown cells eluted near the hexoseI0GlcNAc standard with asecond peak in the hexosegGlcNAc area. These data supportthe idea that castanospermine inhibits processing and thusprevents the trimming of sugars from the oligosaccharides.To learn more about the structures of the oligosaccharides

from these cells, they were treated with oL-mannosidase todetermine how many mannose residues could be released. Ifcastanospermine prevented the removal of glucose residuesfrom the Glc3Man9GlcNAc2-protein, this glycopeptide (oroligosaccharide) should be only partly susceptible to cx-mannosidase since the glucoses cap some of the mannosechains. Figure 4 shows the results of these digestions. In Fig.4A, the elution profile of the castanospermine-derived oligo-saccharides is shown before and after the cx-mannosidasetreatments. In this experiment, the oligosaccharides wererun on a shorter Bio-Gel P-4 column, and thus the resolutionof species is not as good as that in Fig. 3. However, theuntreated oligosaccharides eluted in a broad peak, indicatinga heterogeneous mixture from hexose12GlcNAc (G3) tohexose10GlcNAc (mostly hexose1lGlcNAc, upper profile).This oligosaccharide was only partly susceptible to a-man-nosidase digestion with the release of about 20% of theradioactivity as free mannose (lower profile). Also, after

mannosidase treatment, the larger oligosaccharide becamemuch more homogeneous and mostly migrated in the hex-oseloGlcNAc area. This suggests that most of the mannoseresidues that were released were derived from the hex-ose12GlcNAc and hexose1 1GlcNAc species (i.e.,Glc3Man9GlcNAc and Glc2Man9GlcNAc). In other studies,we have found that the Glc3Man9GlcNAc is relatively resis-tant to (x-mannosidase, and the release of mannose from thatoligosaccharide is very slow (29; Hori et al., in press).The effect of ox-mannosidase was also determined on the

mannose-labeled structures from control cells. However, inthis study we used the control cell glycopeptides rather thanthe oligosaccharides. Figure 4B compares the profiles ofthese glycopeptides before and after Q-mannosidase diges-tion. In this case, the untreated glycopeptide migrated in arather broad area, emerging before, with, and after theGlc3Man9GlcNAc standard. However, after treatment witha-mannosidase, the large-molecular-weight radioactive peakcompletely disappeared and was replaced by a radioactivepeak in the mannose area as well as one or two radioactivepeaks eluting near the Man5GluNAc standard. At least oneof these peaks is probably the Man1GlcNAcGlcNAc-pep-tide, since a GlcNAc residue migrates like 2.1 hexoses onBio-Gel P-4, and the peptide would probably be equal to 1 ormore hexoses. Thus, it seems likely that the control cellglycopeptides are almost completely susceptible to x-man-nosidase and probably have little, if any, glucose.

Further characterization of the mannose-labeled oligosac-charides was done by methylation analysis. Both the EndoH-released oligosaccharides from control cells and fromcastanospermine-treated cells were subjected to completemethylation. After complete acid hydrolysis, the methylated

VOL. 160, 1984

on August 6, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

72 ELBEIN, MITCHELL, AND MOLYNEUX

R 15

,} 10

,. 52I-

Q15a

cc10.

10 20 30DISTANCE FROM ORIGIN (cm)

FIG. 5. Methylation analysis of control and castanospermineoligosaccharides. Oligosaccharides were methylated by the 14ako-mori procedure (12); after complete methylation and isolation, theoligosaccharides were subjected to complete acid hydrolysis. Theradioactive methylated mannose derivatives were identified by thin-layer chromatography in benzene-acetone-water-ammonium hy-droxide (10:200:3:1.5). Standards were prepared by methylation ofyeast mannan and ovalbumin and are as follows: 2,4, 2,4-dimethyl-mannose; 3,4,6, 3,4,6-trimethylmannose; 2,4,6, 2,4,6-trimethylman-nose; 2,3,4,6, 2,3,4,6-tetramethylmannose.

mannoses were identified by thin-layer chromatography.Figure 5 compares the radioactive methylated mannoses ofcontrol cells with those of castanospermine-grown cells. Inthe control cells (upper profile), the expected methylatedmannoses, i.e., 2,3,4,6-tetramethylmannose, 3,4,6-trimeth-ylmannose, and 2,4-dimethylmannose, were observed, indi-cating the presence of terminal mannose, 2-substituted man-nose, and 3,6-substituted mannose. The approximate ratio ofradioactivity in these three species was 1:0.47:0.76. This isclose to the expected for a Man78GlcNAc oligosaccharide.Since we cannot be certain that all of the mannose residuesare equally labeled, one might expect some departure fromthe theoretical value. Onrthe other hand, the lower profile ofFig. 5 shows the identification of methylated mannoses inthe alkaloid-derived oligosaccharide. The distinguishingcharacteristic in this case was the presence of a smallradioactive peak of 2,4,6-trimethylmannose that was absentfrom control oligosaccharides. A mannose substituted in the3 position is strongly suggestive of 9ligosaccharides contain-ing glucose. Thus, as expected, this oligosaccharide alsocontained terminal, 3,6-substituted, 2-substituted, ahd 3-substituted mannose residues in the approximate ratio of1.0:0.83:0.44:0.37. The lower than expected radioactive con-

tent in 2-substituted mannose may be due to unequal labelingin the mannose residues.To be certain that the castanospermine-induced oligosac-

charide contained three glucose residues, A. fumigatuswas grown in castanospermine (1 mg/ml) and labeled with [1-3H]galactose. The glycopeptides were isolated, digestedwith Endo H, and reisolated on Bio-Gel P-4. The oligosac-charide was then methylated, and the methylated glucoses

were identified. Three radioactive peaks were observedcorresponding to 2,3,4,6-tetramethylglucose, 2,4,6-trimeth-ylglucose, and 3,4,6-trimethylglucose (data not shown).Since three glucose derivatives were detected, the oligosac-charide must be a Glc3Man7_9GlcNAc.

Effect of castanospermine on glucosamine-labeled glycopep-tides. Since the N-linked oligosaccharides also contain glu-cosamine, we examined the effect of castanospermine on theincorporation of [3H]glucosamine into the secreted glycopro-teins. Cells were grown in several concentrations of thealkaloid and labeled with [3H]glucosamine. The secretedproteins were isolated and digested with pronase, and theliberated glycopeptides were chromatographed on Bio-GelP-4. Figure 6A shows the profiles obtained from control cells(upper) and cells grown in 1 mg of castanospermine per ml.More of the radioactive glucosamine was found in thesecond peak, although peak 1 was still labeled. This isadditional suggestive evidence that peak 1 represents man-nans with a much higher mannose content (relative toglucosamine) than the typical N-linked oligosaccharides ofthe glycoproteins (as seen in peak 2).

Figure 6A also demonstrates that peak 2 in the controlcells was of lower molecular weight than that seen incastanospermine-grown cells, since it eluted in later frac-tions. This is shown more clearly in Fig. 613, where theglycopeptides have been digested with Endb H and thenrechromatographed on the Bio-Gel P-4 columns. As in thecase of mannose-labeled glycopeptides, peak 2 was suscepti-ble to Endo H as shown by the change in migration aftertreatment. Thus, the oligosaccharide released from controlcells migrated near the MangGlcNAc standard, whereas thatfrom castanospermine-treated cells was larger and migratednear the hexose1lGlcNAc standard. The glucosamine-la-beled oligosaccharide from control cells was mostly suscep-tible to a-mannosidase digestion as demonstrated by theappearance of most of the radioactivity in a peak migratinglike Man-GlcNAc. The oligosaccharide from castanosper-mine-grown cells, on the other hand, was only slightlysusceptible to a-mannosidase, and its migration was onlyaltered by a few fractions, indicating the removal of only oneor two mannose residues (data not shown). These results aresimilar to those observed with the mannose-labeled glyco-peptides and oligosaccharides.

DISCUSSIONThe results reported in this paper show that the tetrahy-

droxyoctahydroindolizine castanospermine inhibits theprocessing of the oligosaccharide chains of the glycoproteinenzymes secreted by A. fumigatus. Thus, cells grown in thepresence of this alkaloid produce N-linked oligosaccharidesthat are larger than the oligosaccharides of normal cellglycoproteins. Based on our partial characterization and thereported structures for the various processing intermediates,the oligosaccharide(s) produced in the presence of castano-spermine appear to be mostly Glc3Man1wGlcNAc2 struc-tures, whereas those found in control cell glycoproteins aremostly Manv9GlcNAc2 structures. Thus, the oligosaccha-rides from normal cells were almost completely susceptibleto a-mannosidase digestion and released most of the radioac-tivity as free mannose indicating the absence of blockingglucose residues. On the other hand, the oligosaccharidesfrom castanospermine-treated cells only released one or twomannose residues by a-mannosidase treatment, suggestingthat some of the branches were capped by glucose residues.These data were also confirmed by methylation analysis ofthe [3H]mannose-labeled and [3H]glucose-labeled oligosac-charides.

STANDARDS 2A 2 -3,4,6 2.3Ao6I I .1

CONTROL OLIGO.

I

CAST. OLIGO.

I IfIa

J. BACTERIOL.

on August 6, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

GLYCOPROTEIN ENZYMES IN A. FUMIGATUS 73

0a.O

I-

3-.

U

0

10 20 30 40 50 60 70 80FRACTION NUMBER FRACTION NUMBER

FIG. 6. Effect of castanospermine on the structures of the glucosamine-labeled glycopeptides and oligosaccharides from secretedglycopeptides. Protocol for this experiment was as described in the legend to Fig. 2, except that [6-3H]glucosamine was used as the label. (A)Glycopeptides from control and castanospermine (1 mg/ml)-grown cells. The glycopeptides were digested with Endo H and rechromato-graphed to give the profiles in B.

The glycoproteins produced in the presence of castano-spermine were still secreted from the cells, as shown by thepresence of various enzymatic activities in the culture medi-um (Fig. 1). However, some inhibition in secretion wasobserved, since the activities in the media were decreasedwith increased castanospermine concentration. Interestinglyenough, the various glycosidases did not show the same

sensitivity to the alkaloid. Thus, the P-hexosaminidase was

the most sensitive of the glycosidases tested, and its activitywas depressed by 30 to 40% at 100 ,ug of alkaloid per ml. Onthe other hand, a-galactosidase activity was not greatlyaffected at alkaloid concentrations of up to 1 mg/ml, anddecreases in activity were only seen at 2 mg/ml or higherconcentrations. We tested the effect of castanosperminedirectly on the enzymatic activity of the ,-hexosaminidase,the a-galactosidase, and the ,-galactosidase by adding vari-ous amounts of alkaloid to the assay mixtures. No inhibitionof enzymatic activity was observed. Thus, the decrease inactivity in the culture media must be attributed to loweramounts of the specific enzymes, or to less active enzyme inthe media.There are several possible explanations to account for the

inhibition of secretion observed in these studies. Sincecastanospermine inhibits the processing glucosidases andprevents the removal of glucose residues fron the N-linkedglycoproteins (28), the glucose-containing glycoprotein maybe recognized only poorly by the secretory mechanism of thecell. In fact, there is some precedence for this idea. A recentstudy by Lodish and Kong (24) compared the effects ofseveral processing inhibitors on the secretion of a number ofglycoproteins by human hepatoma HepG2 cells. Deoxynojir-imycin, also an inhibitor of glucosidase I (20), reduced therate of secretion of the glycoproteins al-antitrypsin and cx1-

antichymotrypsin, but had only marginal effects on thesecretion of other glycoproteins. Equilibrium density gradi-ent centrifugation indicated that this al-antitrypsin and (xl-antichymotrypsin accumulated in the rough endoplasmicreticulum in the presence of deoxynojirimycin. The authorssuggested that the movement of the protein from the roughendoplasmic reticulum to the Golgi required that the N-linked oligosaccharides be processed to at least MangGlc-NAc2 and that glucose residues on these oligosaccharidesmight retard or prevent their movement. Thus, the resultswith these hepatoma cells are quite analogous to thosedescribed here with Aspergillus sp., and the explanation forthe reduction in glycosidase activities in the media could be areduction in the rate of secretion.There are several other compounds that have also been

reported to retard or inhibit the intracellular transport ofnewly synthesized glycoproteins. For example, monensin isa carboxylic acid ionophore that collapses the proton gradi-ent by the electroneutral exchange of a proton for a monova-lent cation (preferably sodium) across a membrane. Thus,monensin causes a rapid dilation of the Golgi elements andblocks the transport of secreted proteins to the extracellularspace (18, 36, 41, 43). Chloroquine and ammonium chlorideare weak bases that become protonated after entering theintralysosomal space. A primary consequence of this actionis the ability of either agent to raise the intralysosomal pHand disrupt the targeting of newly synthesized lysosomalenzymes to the lysosome (13, 42). Since castanospermineand deoxynojirimycin are also weak bases, it is possible thatthey could also act as lysosomotropic drugs and alter intra-cellular pH.

It is clear from a number of studies that the carbohydrateis not always necessary for protein secretion. For example,

AG43 M.9

GIcN optCONTROL

C . l m I I I I

h ICAST. lmg/ml

ioa I I I I I

2

0

1

t2

0.

5:I.

!

:

1

VOL. 160, 1984

1[

on August 6, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

74 ELBEIN, MITCHELL, AND MOLYNEUX

in mouse myeloma tumor cells, 2-deoxyglucose preventedthe incorporation of glucosamine, mannose, and galactoseinto secreted protein while allowing the incorporation ofleucine to proceed at 40% of the normal rate. The proteinthat was secreted under these conditions was shown to bethe nonglycosylated form of K46. Thus, in this case, glyco-sylation was not necessary for secretion, although the ab-sence of carbohydrate did appear to retard intracellulartransport and export from the cell (6).A number of studies have also been done with tunicamy-

cin, an antibiotic that prevents N-glycosylation of proteins(8, 27, 37). In several of these studies, the nonglycosylatedproteins were still secreted from the cells or functionednormally (19, 28, 38), whereas in other cases secretion didnot occur (5, 11). A plausible explanation for these variationsand one for which some evidence has been gathered suggeststhat at least one role for the carbohydrate is to help todetermine or maintain the conformation of the protein (23).Since the carbohydrate is added during polypeptide synthe-sis, it may have a great influence on the protein conforma-tion, depending, of course, on the amino acid composition ofthe protein. That is to say, carbohydrate may be essential ininfluencing the conformation of some proteins, but not ofothers. Since castanospermine apparently does not inhibitglycosylation, but causes alterations in the final oligosaccha-ride structure, it may be possible to correlate subtle changesin structure with alterations in function. The changes insecretion observed with the glycosidases may be an examplethat needs further examination.

ACKNOWLEDGMENTThis study was supported by Public Health Service research grant

HL-17783 from the National Institutes of Health.

LITERATURE CITED1. Akasaki, M., M. Suzuki, K. Funakoshi, and I. Yamashina. 1976.

Characterization of P-galactosidase from a special strain ofAspergillus oryzae. J. Biochem. (Tokyo) 80:1195-1200.

2. Ayda, S., and A. D. Elbein. 1977. Glycoprotein enzymes secret-ed by Aspergillus niger: purification and properties of a-galacto-sidase. J. Bacterol. 129:850-856.

3. Burns, D. M., and 0. Touster. 1982. Purification and character-ization of glucosidase II, an endoplasmic reticulum hydrolaseinvolved in glycoprotein biosynthesis. J. Biol. Chem. 257:9991-10,000.

4. Chen, W. W., and W. J. Lennarz. 1978. Enzymatic excision ofglucosyl units linked to the oligosaccharide chains of glycopro-teins. J. Biol. Chem. 253:5780-5785.

5. Cox, G. S. 1981. Synthesis of the glycoprotein hormone a-subunit and plancental alkaline phosphatase by Hela cells:effects of tunicamycin, 2-deoxyglucose and sodium butyrate.Biochemistry 20:4893-4900.

6. Eagon, P. C., and E. C. Heath. 1977. Glycoprotein biosynthesisin myeloma cells. Characterization of nonglycosylated immuno-globulin light chain secreted in the presence of 2-deoxy-D-glucose. J. Biol. Chem. 252:2372-2383.

7. Elbein, A. D. 1979. The role of lipd-linked saccharides in thebiosynthesis of complex carbohydrates. Annu. Rev. Plant Phy-siol. 30:239-272.

8. Elbein, A. D. 1984. Inhibitors of the biosynthesis and processingof N-linked oligosaccharides. Crit. Rev. Biochem. 16:21-49.

9. Elbein, A. D., S. Ayda, and Y. C. Lee. 1977. Purification andproperties of a 3-mannosidase from Aspergillus niger. J. Biol.Chem. 252:2026-2031.

10. Grinna, L. S., and P. W. Robbins. 1980. Substrate specificitiesof rat liver microsomal glucosidases which process glycopro-teins. J. Biol. Chem. 255:2255-2258.

11. Hackman, S., A. Kulczycki, Jr., R. G. Lynch, and S. Kornfeld.1977. Studies on the mechanism of tunicamycin inhibition ofIgA and IgE secretion by plasma cells. J. Biol. Chem. 252:4402-

4408.12. Hakomori, S. 1964. A rapid permethylation of glycolipid and

polysaccharide catalyzed by methyl sulfinyl carbanion in di-methyl sulfoxide. J. Biochem. (Tokyo) 55:205-208.

13. Hasilik, A., and E. F. Neufeld. 1980. Biosynthesis of lysosomalenzymes in fibroblasts. J. Biol. Chem. 259:4937-4945.

14. Hohenschutz, L. D., E. A. Bell, P. J. Jewess, D. P. Leworthy,R. J. Pryce, E. A. Arnold, and J. Clardy. 1981. Castanosper-mine, a 1,6,7,8-tetrahydroxyoctahydroindolizine alkaloid fromthe seeds of Castanospermum australe. Phytochemistry 20:811-814.

15. Hubbard, S. C., and R. Ivatt. 1981. Synthesis and processing ofasparagine-linked oligosaccharides. Annu. Rev. Biochem.50:555-583.

16. Hurst, P. L., J. Neilsen, P. A. Sullivan, and M. G. Shepherd.1977. Purification and properties of a cellulase from Aspergillusniger. Biochem. J. 165:33-41.

17. Iwashita, S., and F. Egami. 1973. a-D-fucosidase from Aspergil-lus oryzae: characterization of a-D-fucosidase with a-D-galacto-sidase activity. J. Biochem. (Tokyo) 73:1217-1222.

18. Johnson, D. C., and M. J. Schlesinger. 1980. Vesicular stomatitisvirus and Sindbis virus glycoprotein transport to the cell surfaceis inhibited by ionophores. Virology 103:407-424.

19. Keller, R. K., and G. D. Swank. 1978. Tunicamycin does notblock ovalbumin secretion in the oviduct. Biochem. Biophys.Res. Commun. 85:762-768.

20. Kilker, R. D., Jr., B. Saunier, J. S. Tkacz, and A. Hercovics.1981. Partial purification from Saccharomyces cerevesiae of asoluble glucosidase which removes the terminal glucose fromthe oligosaccharide Glc3Man9GlcNAc2. J. Biol. Chem.256:5299-5303.

21. Kobata, A. 1979. Use of endo and exoglycosidases for structuralstudies of glycoconjugates. Anal. Biochem. 100:1-14.

22. Kornfeld, R., and S. Kornfeld. 1976. Comparative aspects ofglycoprotein structure. Annu. Rev. Biochem. 45:217-237.

23. Leavitt, R., S. Schlesinger, and S. Kornfeld. 1977. Impairedintracellular migration and altered solubility of nonglycosylatedglycoproteins of VSV and Sindbis virus. J. Biol. Chem.252:9018-9023.

24. Lodish, H. F., and N. Kong. 1984. Glucose removal from N-linked oligosaccharides is required for efficient maturation ofcertain secretary glycoproteins from the rough endoplasmicreticulum to the golgi complex. J. Cell Biol. 98:1720-1729.

25. Mega, T., T. Ikenaka, and Y. Matsushima. 1972. Studies on N-acetyl-D-glucosaminidase of Aspergillus oryzae. Il. Substratespecificity of the enzyme. J. Biochem. (Tokyo) 71:107-114.

26. Michael, J. M., and S. Kornfeld. 1980. Partial purification andcharacterization of the glucosidases involved in the processingof asparagine-linked oligosaccharides. Arch. Biochem.Biophys. 199:249-258.

27. Mozraki, A., J. A. O'Malley, W. A. Carter, A. Takatsuki, G.Tamura, and E. Sulkowski. 1978. Glycosylation of interferons.Effects of tunicamycin on human immune interferon. J. Biol.Chem. 253:7612-7615.

28. Olden, K., R. M. Pratt, and K. M. Yamada. 1978. Role ofcarbohydrates in protein secretion and turnover: effects oftunicamycin on the major cell surface glycoprotein of chickembryo fibroblasts. Cell 13:461-473.

29. Pan, Y. T., H. Hori, R. G. Saul, B. A. Sanford, R. J. Molyneux,and A. D. Elbein. 1983. Castanospermine inhibits the processingof the oligosaccharide portion of the influenza viral hemaggluti-nin. Biochemistry 22:3975-3984.

30. Rudick, M., and A. D. Elbein. 1973. Glycoprotein enzymessecreted by Aspergillus fumigatus. Purification and propertiesof ,-glucosidase. J. Biol. Chem. 248:6506-6513.

31. Rudick, M. J., and A. D. Elbein. 1974. Glycoprotein enzymessecreted by Aspergillus fumigatus. Purification and propertiesof a-glucosidase. Arch. Biochem. Biophys. 161:281-290.

32. Saul, R., J. P. Chambers, R. J. Molyneux, and A. D. Elbein.1983. Castanospermine, a tetrahydroxylated alkaloid that inhib-its 3-glucosidase and ,-glucocerebrosidase. Arch. Biochem.Biophys. 221:593-597.

33. Schachter, H., S. Narasimhan, P. Gleeson, and G. Vella. 1983.

J . BACTERIOL .

on August 6, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

GLYCOPROTEIN ENZYMES IN A. FUMIGATUS 75

Control of branching during biosynthesis of asparagine-linkedoligosaccharides. Can. J. Biochem. Cell. Biol. 61:1049-1066.

34. Speake, B. K., D. J. Mailey, and F. W. Hemming. 1981. Theeffect of tunicamycin on secreted glycosidases of Aspergillusniger. Arch. Biochem. Biophys. 210:110-117.

35. Spiro, R. J., M. J. Spiro, and V. D. Bhoyroo. 1979. Processing ofcarbohydrate units of glycoproteins. Characterization of a thy-roid glucosidase. J. Biol. Chem. 254:7659-7667.

36. Srinivas, R. V., L. R. Melsen, and R. W. Compans. 1982. Effectsof monensin on morphogenesis and infectivity of Friend murineleukemia virus. J. Virol. 42:1067-1075.

37. Struck, D. K., and W. J. Lennarz. 1981. The function ofsaccharide-lipids in synthesis of glycoproteins, p. 35-73. In W.J. Lennarz (ed.), The biochemistry of glycoproteins and proteo-glycans. Plenum Publishing Corp., New York.

38. Struck, D. K., P. B. Siuta, M. D. Lane, and W. J. Lennarz. 1978.Effect of tunicamycin on the secretion of serum proteins byprimary cultures of rat and chick hepatocytes. J. Biol. Chem.253:5332-5337.

39. Swaminathan, N., K. L. Matta, L. A. Donoso, and 0. P. Bahl.

1972. Glycosidases of Aspergillus niger. III. Purification andcharacterization of 1,2-a-mannosidase. J. Biol. Chem.247:1775-1779.

40. Tarentino, A. L., and F. Maley. 1974. Purification and propertiesof an endo-,B-N-acetylglucosaminidase from Streptomyces gri-seus. J. Biol. Chem. 249:811-817.

41. Tartakoff, A. M., and P. Vassalli. 1977. Plasma cell immuno-globulin secretion. Arrest is accompanied by alterations of thegolgi complex. J. Exp. Med. 146:1332-1345.

42. Tietze, C., P. Schlesinger, and P. Stahl. 1980. Chloroquine andammonium ion inhibit receptor-mediated endocytosis of man-nose-glycoconjugates by macrophages: apparent inhibition ofreceptor recycling. Biochem. Biophys. Res. Commun. 93:1-8.

43. Uchida, N., H. Smilowitz, and M. S. Tanzer. 1979. Monovalentionophores inhibit secretion of procollagen and fibronectin fromcultured human fibroblasts. Proc. Natl. Acad. Sci. U.S.A.76:1868-1872.

44. Ugalde, R. A., R. J. Staneloni, and L. F. Leloir. 1980. Microsom-al glucosidases of rat liver: partial purification and inhibition bydisaccharides. Eur. J. Biochem. 113:97-103.

VOL. 160, 1984

on August 6, 2020 by guest

http://jb.asm.org/

Dow

nloaded from