Structures of the Oligosaccharides of the Glycoprotein Coded by Early Region E3 of Adenovirus 2

(1)

0022-538X/81/110440-10$02.00/0

Structures of the Oligosaccharides of the Glycoprotein Coded

by Early Region

E3

of Adenovirus

2

ROSALIND KORNFELD' ANDWILLIAM S. M. WOLD2*

Division ofHematologyandOncology, Washington University School of Medicine,' and Institute for

Molecular

Virology,

St.Louis

University

Medical

Center,2

St.

Louis,

Missouri63110

Received27April 1981/Accepted23June1981

Early region E3 of adenovirus 2 encodes a glycoprotein, E3-gp25K, that is a

good model with whichto study structure-function relationships in

transmem-brane glycoproteins. We havedetermined thestructuresof theoligosaccharides linked to E3-gp25K. The oligosaccharideswere labeled with [2-3H]mannose in

adenovirus 2-early infected KBcellsfor 5.5 h (pulse)orfor 5.5 h followedbya

3-h c3-hase (pulse-chase). E3-gp25Kwasextracted andpurified bychromatography onDEAE-Sephacel in 7 Murea,followedby gel filtrationonacolumn of Bio-Gel

A-1.5m in 6 M guanidine hydrochloride. An analysis of the purified proteinby sodiumdodecylsulfate-polyacrylamidegelelectrophoresis indicated that itwas

>95%pure.Theoligosaccharides wereisolatedbypronasedigestion followedby

gel filtration on a column of Bio-Gel P-6, then by digestion with

endo-,8-N-acetylglucosaminidaseH, followed by gel filtrationonBio-GelP-6, and finallyby

paper chromatography. The pulse sample contained equal amounts of Man9GlcNAc and Man8GlcNAc and small amounts of Man7GlcNAc and

Man6GlcNAc. The pulse-chase sample had predominantly Man8GlcNAc and much less Man9GlcNAc, indicating that processing of the Man9GlcNAc to

Man8GlcNAc had occurred during the chase period. Thus, Man8GlcNAc is the major oligosaccharide onmatureE3-gp25K. Thestructuresofthese oligosaccha-rides were established by digestion with a-mannosidase, methylation analysis,

andacetolysis. The oligosaccharides found had typical high-mannosestructures

thathave been observed in other membrane and solubleglycoproteins, and the branching patternsandlinkages of the mannoseresidues of Man9GlcNAcwere

identical tothose of the lipid-linkedGlc3Man9GlcNAc2 donor. Thus, adenovirus 2 infection (early stages) apparently does not affect the usual cellular

high-mannose glycosylation pathways, and despite being virus coded, E3-gp25K is

glycosylated in thesame manner as atypical mammaliancell-codedglycoprotein.

Two glycoproteins, which we will designate E3-gp25K and E3-gp22K,aresynthesizedduring

earlystagesof infectionof human cellsby ade-novirus 2 (Ad2) (9, 10, 20). (gp25K and E3-gp22K have been referred to as E19K and

E17.5K by other workers [19-21, 24-26]. We have used the designations gp25K and gp22K because theyareglycoproteins thatmigrate in our sodium dodecyl sulfate [SDS]-gel system

withapparentmolecularweightsof25,000 [25K]

and22K,asjudgedbytherelative mobilities of

protein standards. Since glycoproteinsgenerally

migrate more slowly in SDS-gels than do

non-glycosylated proteins of comparable molecular weight, thetruemolecularweights of E3-gp25K

andE3-gp22Kwillbelessthan 25K and22K.) E3-gp25K and E3-gp22K have indistinguish-ablepeptide mapsand, thus, arehighly related

in their primary structures (19, 25; Q. Kapoor and W. Wold, unpublished data). Cells contain much more gp25K than gp22K.

E3-gp25K is localizedontheplasma membrane (4,

19, 20) and is complexed with the major

trans-plantation antigen of the host cell (15).

E3-gp25K has been purified to homogeneity, and antisera have been prepared (20; Kapoor and

Wold, unpublished data).

Four lines of evidenceprovethatE3-gp25Kis

coded by early region E3 (map position 76 to

86). First, it isimmunoprecipitated byantisera

against Ad2-transformed cell lines that retain E3(1, 10, 11, 23,27). Second,it isnotsynthesized incellsinfectedbyAd2+ND1, which hasa

dele-tionbetweenmappositions80.3 and 85.5(19, 24,

25).Third,it is translated in vitro from

r-strand-specificandE3-specificmRNA (18,19). Fourth,

440

on November 10, 2019 by guest

http://jvi.asm.org/

(2)

the partial N-terminal amino acidsequencehas been aligned with the DNAsequence in E3 (8, 19).

Aspects of the structure of gp25K can be

predicted from the DNAsequence (8, 18). The primarytranslation product is 159amino acids

long, and it containsaputative N-terminal

hy-drophobic signal sequence of about 17 or 18

amino acids (18). The C-terminus contains a

hydrophilic tail of15amino acids,which is

pre-ceded by a hydrophobic domain of 23 amino acids. Thereare twopotential sites for N-glyco-sylation, Asn-Val-Thr, located in the N-terminal portion of the molecule. By analogy with other membrane glycoproteins that have a similar

type of primary structure (21, 22), E3-gp25K

maybe oriented in themembrane such that the glycosylated N-terminal portion of the protein protrudesfrom thecell, the hydrophobic domain

traverses the membrane, and the hydrophilic tailextends into the

cytoplasm

and anchors the protein in the membrane (18).

E3-gp25Kwas first showntobe glycosylated by

labeling

with

[3H]glucosamine

(9, 10, 19,20) or

[2-3H]mannose

and by binding to

concana-valin AorLens

culinaris

lectins (10, 20).

Gly-cosylation

isinhibited by tunicamycin (19)orby

2-deoxyglucose

or

glucosamine

(25). When translated in

vitro, glycosylation

occursonly in thepresenceofamicrosomal fraction (18, 19).

Since much is known about E3-gp25K and since it isaviralgeneproduct that issynthesized in fairly large amountsin infected cells, it isa

good

model with whichto

study

structure-func-tion relationships in transmembrane glycopro-teins.Inthiscommunication,wereportthe

com-pletestructuresof theoligosaccharides linkedto

E3-gp25K. The

oligosaccharides,

labeled with

[2-3H]mannose,

were examined after a 5.5-h

la-beling

period

(pulse)

and also after a5.5-h la-beling

period

followed by a 3-h chase (pulse-chase). At both times, all of the oligosaccha-rides were of the high-mannose type, with

Man9GlcNAc2

and

MansGlcNAc2

predominating

inthepulse sample and

ManmGlcNAc2

predom-inating in thepulse-chase sample.

MATERIALS AND METHODS Virusinfection ofcellsandglycoprotein

label-ing. Methods used for thepurificationand

quantita-tionof Ad2 stocks have been describedpreviously (6,

28). Suspension cultures of KB cellswere grown at

37°C at aconcentration of3.5 X 105cells per ml in

Eagle minimal essential medium (MEM) containing

5% horseserum. Cells from2literswerecollectedby

centrifugation,suspendedin100ml of MEM without

horse serum, and infected with500PFU of Ad2(strain

38-2) per cell. After1h,cellsweredilutedto2liters

with MEMcontaining5%horseserum.At 2.5 h

post-infection, cells were collected by centrifugation and

werewashed in warm MEM containing l/lo the normal

glucose (i.e., 200 mg/liter) and 5% dialyzed horse

se-rum. Thewashing step wasrepeated, and the cells

weresuspended in1liter of warm MEM containing 'Ao

the normalglucose, 5%dialyzed horse serum, and20

Lg of1-/-D-arabinofuranosylcytosine per ml. The

1-fi-D-arabinofuranosylcytosine wasusedtoinhibit the

transition from earlyto late stages of infection, the

resultantshutoff of E3-gp25K synthesis (20), and the

possible modification of the host cell glycosylating

machinery. At 3.5 h postinfection, 10 mCi of

D-[2-3H]mannose (New England Nuclear Corp.; 14.5 Ci/

mmol) was added. Samplestaken immediately after

theaddition of the label and at 1-h intervals indicated

that incorporation of[2-3H]mannose into

trichloro-aceticacid-insoluble materialwaslinear overthe

la-belingperiod (datanotshown).

At 9 hpostinfection, the cellsweredivided intotwo

500-mllots,pulse andpulse-chase, andthe cellswere

collected by centrifugation. Thepulse-chase cellswere

washedonce in warm MEM containing the normal

amountofglucose, 5% horse serum, and20,ugof

1-ft-D-arabinofuranosylcytosine and thenwerefurther

in-cubated in500mlof thesamemedium. Thepulse cells

were washed twice at 4°C with cold

phosphate-buffered salinelackingCa2" and

Mg2"

and thenwere

keptonice. At12hpostinfection,thepulse-chase cells

werecollected andsimilarly washed with cold

phos-phate-buffered saline. Thus, both lots of cells were

labeled for5.5hwith[2-3H]mannose, but the

pulse-chasecellswerechasedfor3h withglucose.

Extraction andpurificationof

[2-3H]mannose-labeledE3-gp25K.Cellpelletsweresuspended in9

ml of isotonic high pHbuffer (0.14 M NaCl, 1 mM

MgCl2,10 mMTris-hydrochloride, pH8.5)containing

1mM phenylmethylsulfonyl fluoride (a protease

in-hibitor)at 1 to2°C andwerelysed by the addition of

Nonidet P-40to aconcentration of 0.5%. After5min

ofgentle mixingwith a pipette, the nucleiwere

re-movedby centrifugationat200xgfor10minat4°C.

The supernatant was removed, and the pellet was

washedoncewithlysis buffer. The supernatantswere

pooledandclarifiedbycentrifugationat12,000xgfor

10 min at4°C in aSorvall SS-34 rotorand then at

82,000xgfor1hat4°Cina50Tirotor.Triton X-100

was added to a final concentration of 1%, and the

samplesweredialyzedfor2daysat4°C againstDEAE

buffer (6 M urea, 10mM NaCl, 1%Triton X-100, 1

mMdithiothreitol, 1 mMphenylmethylsulfonyl

fluo-ride,20mMTris-hydrochloride, pH8.5). Theyieldat

this stagewas 1.38x 108and8.27x 107cpmfor the

pulseandpulse-chase samples, respectively.

TwoDEAE-Sephacel columns (1.1 by 7cm) were

poured,equilibrated againstDEAEbuffer, and loaded

with the protein samples at a rate of3 ml/h. The

columns were washed with DEAE buffer until the

counts being eluted were negligible and then were

eluted with DEAE buffercontaining1 MNaCl. The

twoflow-through poolsand thetwo1MNaCl eluate

poolsweredialyzedat4°Cfor2days against several

changesof50mMNH4HCO3-0.01%TritonX-100. An

analysisof thefractionsbypolyacrylamide gel

electro-phoresisinSDS(SDS-PAGE) indicated thatmostof

theE3-gp25Kwasin theflow-through fractions,but

on November 10, 2019 by guest

http://jvi.asm.org/

(3)

442 KORNFELD AND WOLD

that it was contaminated by a number of cellular

glycoproteins.Thepulseandpulse-chaseflow-through

fractions contained 2.2 x 107 and 1.2 x 107 cpm,

respectively.

E3-gp25Kwasfurtherpurifiedby gelifitration,as

follows. The dialyzed DEAE flow-through fractions

were lyophilized, suspended in about 5ml of0.4 M

glycine, mixed with 5volumes of cold acetone, and

precipitated for 2 h at 0 to 2°C. The proteins

co-precipitated with thelargeflocculentglycine

precipi-tate that formed in the presence of acetone. The

acetone extractsthe Triton X-100(atahigh

concen-trationafter thelyophilizationstep), which mustbe

removed because it forms micelles which preclude

partitioning of proteins in the gel filtration column.

Theglycine-proteinpelletsweresuspended in50mM

NH4HCO3-0.01%TritonX-100, lyophilized, and

resus-pended in0.7ml of 8 Mguanidinehydrochloride-2%

2-mercaptoethanol-10 mM EDTA-50 mM Tris-hy-drochloride, pH 8.5. The recovery of counts was >95%.

Thesamples wereheatedat 560Cfor45minto

dis-sociatethepolypeptidesandwereneutralized with0.1

ml of1 M sodium phosphate, pH 6.5. Thesamples

were mixed withasmall volume ofphenolred

con-taining100 mgofsucroseandwereappliedto acolumn

(1.5by100cm) of Bio-Gel A-1.5mequilibratedin6M

guanidinehydrochloride-10mMdithiothreitol-1 mM phenylmethylsulfonyl fluoride-20 mM sodium

phos-phatebuffer, pH6.5.One-hour(3.5-ml)fractionswere

collected. All fractionswereadjustedto0.02%Triton

X-100andwere dialyzed against 50mM

NH4HCO3-0.02% Triton X-100.Each fractionwascounted,and

selected fractionswereassayed bySDS-PAGE, using

the deoxycholate-trichloroacetic acid precipitation

method (seebelow).

SDS-PAGE. SDS-PAGEwascarriedoutwith

dis-continuous 13%polyacrylamidegelsasdescribed

pre-viously (27).ThtonX-100causes"smearing"ongels;

therefore, itwasextracted fromthe proteinsamples

beforeSDS-PAGE,asdescribedbelow.Thesamples,

adjusted to0.5 ml in 1.5-ml Eppendorf tubes, were

mixed(Vortex) with10

pl

of 2% sodiumdeoxycholate

andincubated at roomtemperaturefor 15min.

Tri-chloroacetic acidwasaddedto afinal concentration of

10%, and thesampleswereincubatedat 0 to2°C for

30min.Theheavytrichloroacetic

acid-deoxycholate-proteinprecipitatethatformedwasremovedby

cen-trifugationfor4mininanEppendorfmicrofuge.The

pelletswerewashed twice with1 ml ofcoldacetone

andfinally were suspended in 20 to 50 ul of

SDS-PAGEsample buffer (27).

Preparation oflabeledglycopeptides. The re-tained fractions of the Bio-Gel A-1.5m column that

containedpurified E3-gp25K(seeFig. 1)werepooled

andlyophilized.Eachpoolwasdissolved in1.5mlof water and extracted twice with 2 ml of toluene to removeTritonX-100.The somewhat turbid solutions

wereincubatedat600Cwith7mgof pronase in2mM

CaCl2-0.1MTris-hydrochloride,pH8.0.After4h,an

additional7mgof pronasewasadded.After20h, each

reaction was boiled for 2 min and subjected to gel

filtrationon acolumn(1.5by98cm) of Bio-Gel P-6to

isolate theglycopeptides(seeFig.2below).

Preparation ofoligosaccharides.The

glycopep-tide fractions from each sample, obtained from the

Bio-Gel P-6 column, were lyophilized and redissolved

in 2.5ml of water. Onemilliliter of each glycopeptide

samplewastaken to dryness and incubated for 19 h at

370Cin 100

pl

of 50 mM citratephosphate buffer, pH

6.5, containing2mUof

endo-,8-N-acetylglucosamini-dase H(endo H) (Miles Laboratories, Inc.). The

sam-ples, diluted to1.0 ml with water, were boiled for 3

minand againsubjected to gelfiltration on Bio-Gel

P-6as shown in Fig. 2.The pooledoligosaccharide

frac-tions from the Bio-Gel P-6 column were lyophilized,

redissolved in water, spotted onWhatman no. 1 paper,

andsubjected todescending paper chromatography in

pyridine-ethyl acetate-acetic acid-water (5:5:1:3),

sol-ventI, for7days to separateoligosaccharides

accord-ing to size. The sample lanes and a lane containaccord-ing oligosaccharide standards of known size were cut in

1-cmsegmentswhich wereeluted in 1.0 ml of water. A

sample of each was counted, and eluates were pooled that corresponded to Man9GlcNAc, Man8GlcNAc, Man7GlcNAc, and Man6GlcNAc (see Fig. 3).

Structural analysis ofoligosaccharides. The

oligosaccharides were methylated by the method of Hakomori (7), and the permethylated products were

hydrolyzed in 2 NH2S04 andanalyzed by thin-layer

chromatographyasreported by Li et al. (16). Labeled

oligosaccharides werereduced with NaBH4 and

sub-jectedto acetolysis aspreviously described (2). The

Man- Mandisaccharides obtained by acetolysis were

reduced withNaBH4andsubjected tomolybdate

elec-trophoresis as previously described (2) to separate

Man 2al,2 mannitol from Man 21,3 mannitol.

Oligosaccharidesweredigested withjack bean

a-man-nosidase preparedby the method of Li and Li (17) by

incubationat370Cwith 1U of enzyme in 20

pl

of 50

mMacetatebuffer, pH 4.6, for 1.5h. The digestions

were then subjected todescending chromatography

on Whatman no. 1 paper in solvent I for 16 h to

separatefreemannosefrom Man-*GlcNAc

disaccha-ride.

RESULTS

Purity

of

[2-3Hlmannose-labeled

E3-gp25K. Figure 1A shows the Bio-Gel A-1.5m column elution profile of the labeled

glycopro-teins from thepulse-labeledcells. Fractions20,

33, 34, 35, 36,and37-41poolwere

analyzed by

SDS-PAGE. The

majority

of

E3-gp25K

was

foundinthemajorretained

peak,

fractions33to

-37 (Fig. 1B). Some E3-gp25K wasalso found

in the void volume, despite the use of 6 M

guanidine

hydrochloride

and10mM

dithiothre-itol in the column

buffer,

which attests to the

tendency of this

polypeptide

to aggregate. A small

quantity

ofE3-gp22K wasfound in frac-tions 35 to 41. The 44K

glycopolypeptide

in

fractions34 to 36ishighly relatedtoE3-gp25K,

because it has a very similar

[35S]methionine-containing

peptide

map (27;

Q. Kapoor

and W.

Wold,

unpublished

data).

Theapparent molec-ularweightestimate of44K forthis

glycopoly-peptidedoes notruleoutthe

possibility

that it

is a dimer of gp25K, because the

mobility

of

J. VIROL.

on November 10, 2019 by guest

http://jvi.asm.org/

(4)

Vo

It

A

Cpu'

1a

X1D0

10

8

6

4

2

0 10 20 30 40 5

FRACTION NO I3.5ml/fraction)

B

fl %I 5 -. ! -

W-

93K-69K- _

->l1.K _

...:5 _K-_40

13K-s ii_

41

FIG. 1. Gel filtration of[2-3Himannose

pulse-la-beledE3-gp25Kon acolumnofBio-Gel A-1.5m. The

column wasprepared and run as described in the

text. (A) Elutionprofile ofthe labeled

glycopolypep-tides. (B) SDS-protein gelwhich illustrates the

gly-copolypeptides infractions20, 33, 34, 35, 36, and

37-41pool. "M"refersto'4C-labeledproteinstandards

(NewEngland NuclearCorp.), whichwere

phospho-rylase b(93K), bovineserum albumin(69K),

ovalbu-min(46K), carbonicanhydrase (31K),andcytochrome

c(13K).

glycopolypeptidesinSDS-gelsis not an accurate

indication oftheirmolecular weights. The60K glycopolypeptide in fraction 35 may have been yet another form(e.g.,trimer)ofE3-gp25Korit may have been a contaminant. If itwas a

con-taminant,itsconcentrationwasmuchtoolowto

compromiseouranalysisof theoligosaccharides

ofE3-gp25K. Similarresults were obtained with

the glycoproteins from the

pulse-chase-labeled

sample (data notshown). Fractions 33to41 in

Fig. 1 were pooled,as were analogousfractions

from the pulse-chase-labeled sample, and the

glycopeptides were prepared bypronase

diges-tion.

Purification of the oligosaccharides of E3-gp25K. Figure 2 shows theBio-Gel P-6 col-umnprofileof theglycopeptides obtainedonthe pronasedigestofthepulse-labeled sample (solid line) andonthe oligosaccharides released from theseglycopeptides by endoHdigestion(dashed

line). Essentially thesameprofilewas obtained

for the pulse-chase sample. EndoHcleavesthe twoN-acetylglucosamine residues of the chito-biose core of high-mannose, asparagine-linked oligosaccharides, liberating an oligosaccharide withoneN-acetylglucosamine residueat its

re-ducing end and leavingoneN-acetylglucosamine

U residue attached to asparagine in the peptide. Kobata(12) hasrecently reviewedthespecificity studies thatshowthat endoH can act on

high-mannoseoligosaccharides, butnot onthe

com-plex type. In both the pulse and pulse-chase samples, all ofthe

[2-3H]mannose-labeled

oli-gosaccharidesweresusceptibletoendoH

cleav-ageand thusmustbe of thehigh-mannose-type

structure.

The oligosaccharide fractions from the

Bio-GelP-6columnsweresubjectedtopaper chro-matographytoresolve the different sized oligo-saccharides. Thepulse sample contained almost equal amounts ofoligosaccharides comigrating with authentic

MangGlcNAc

(ninemannose

res-idues linked to N-acetylglucosamine) and

Man8GlcNAc,aswell asmuchsmaller amounts

3,000

Vo Man

P'.

9

0

11 1.

I

0

II

(r

2,000k

E

1,000_

U 20 40 60 80

FRACTION

FIG. 2. Bio-Gel P-6gel filtration ofthe glycopep-tidesin the pronasedigest(0) andthe

oligosaccha-ridesreleasedfrom them after endo H digestion (0).

The Bio-Gel P-6 column (1.5 by98cm) was

equili-bratedandeluted with0.1Mammoniumbicarbonate,

and2-mlfractionswerecollected. Samples of every

fraction were counted in asolutioncontaining 4 ml

of 3a70 scintillationfluidand0.4mlofwater.

Frac-tions 32 to 50 werepooled fromtheglycopeptide peak,

andfractions49 to 60werepooled from the

oligosac-charidepeak.

on November 10, 2019 by guest

http://jvi.asm.org/

[image:4.497.50.243.77.424.2] [image:4.497.256.446.421.564.2]

(5)

ofmaterialcomigrating with Man7GlcNAc and Man8GlcNAc (Fig. 3A). In contrast, the

pulse-chasesample(Fig.3B) containedpredominantly

MansGlcNAc,

with much less Man9GlcNAc.

Upon rechromatography, each of the pooled

eluates from the first separation migrated as

single peaks coincident with Man96GlcNAc standards. The quantitative difference in

Man9GlcNAc

between thepulse and the pulse-chase samples indicated that processing of the oligosaccharidetoitsmatureform hadoccurred during thechaseperiod.

Structural studies on each oligosaccharide species from both the pulse and pulse-chase sampleswereperformedasdescribedbelow, but dataareshownonly for thepulse samples.

Re-sults were essentially identical for the pulse-chasesample oligosaccharides.

Structure of the

Man9GlcNAc

oligosac-charide. When the oligosaccharide with the same mobility as Man9GlcNAc was digested with a-mannosidase, 89% of the radioactivity

migrated upon paper chromatography as free

mannoseand 11%migratedasthedisaccharide

Man

-.1,4

GlcNAc

(data

notshown). Thus, it

containedeighta-mannoseresidues andcanbe

represented as Man8 -a-, Man- - GlcNAc.

Todetermine whichhydroxylsonthemannose residues in

MangGlcNAc

were involved in

gly-cosidic bonds, MangGlcNAc was subjected to methylation, and the methylated mannose

spe-cies released by acid hydrolysiswereseparated

by thin-layer chromatography. The resultsare

showninFig.4andaresummarized inTable 1.

MangGlcNAc

gaveriseto2,4-dimethyl, 3,4,6-tri-methyl, and 2,3,4,6-tetramethylmannosein the

ratio of 2:4:3.

To localize the Man

.1,6

Man linkages,

MangGlcNAc

wasreduced with NaBH4and

sub-jectedtoacetolysis (which preferentially cleaves

Man

--16,

Manlinkages).Fragmentsobtained

12 3 4 5

MgM8 M7 M6 M5

400

A

200

E

D-)

B

10 20 30 40 50

cm

FIG. 3. Preparativepaperchromatography of oli-gosaccharides released by digestion with endo H.

Thechromatogramsweredevelopedandanalyzedas

described in the text. (A) Pulse sample; (B)

pulse-chasesample. Thearrowsshow themigration posi-tionsof authenticMan9GlcNAc-Man5GlcNAc

oligo-saccharides. Fractions corresponding toeachpeak

werepooledandanalyzed further.

r Man8

5 10 15

cm

FIG. 4. Thin-layer chromatography ofthe

methyl-atedmannosespeciesderivedfromthepooled

oligo-saccharidefractionsshown inFig.3. Thethin-layer platesweredevelopedandanalyzedasdescribedin

thetext.Thearrowsatthetopdenote themigration positions ofthefollowingauthenticmethylated man-nose species: (1) 2,3-dimethyl, (2) 2,4-dfmethyl, (3)

3,4,6-trimethyl; (4) 2,4,6-trimethyl, and(5) 2,3,4,6-tet-ramethylmannose.

E

a-u

200[

1600 1200

E

a- 800

-U

on November 10, 2019 by guest

http://jvi.asm.org/

[image:5.497.290.430.262.579.2] [image:5.497.66.243.307.590.2]

(6)

VOL.40,1981OLIGOSACCHARIDES OF THE E3 GLYCOPROTEIN OF Ad2 445

TABLE 1. Methylation ofpulse-labeled oligosaccharidesandacetolysisfragments

No. ofmannoseresidues0a

Sample 2,3,4,6-Tetramethyl 2,4,6-Trimethyl 3,4,6-Trimethyl- 2,4-Dimethyl mannose mannose niannose mannose

Intactoligosaccharides

Man9GIcNAc 3.2 0 3.8 2.0

MansGlcNAc 3.2 0 2.9 1.9

Man7GlcNAc 2.8 0.4 2.1 1.7

Man6GlcNAc 2.8 0.2 1.2 1.8

Acetolysis fragmentsofMan9GIcNAc

Man--.Man --.Man 1.1 0.9 1.0 0

Man4GlcitolNAc 1.1 0.9 2.0 0

0Thesumof theradioactivityin

all

speciesofmethylatedmannosein eachsampleinFig.4wassetequalto

thenumberofmannoseresidues in thatsampleand the counts per minute permannoseresiduewerecalculated

foreachsample.This valuewasthen usedtodeterminethe number ofmannoseresidues of everymethylated species.

had the same mobilities as authentic

Man4GlcitolNAc, Man3,

and Man2

(Fig.

5A).

[image:6.497.254.445.262.519.2]

These results

suggested

thestructure shown in

Fig. 5A,

where thedashed lines indicate

cleavage

sites at

al,6

linkages.

Thetwo

larger fragments

werecharacterized further

by methylation

(Ta-ble

1). Man2

was shown to be linked

al,2

by

subjecting

the materialtoNaBH4 reduction

fol-lowed

by

high-voltage

electrophoresis

in

molyb-date

buffer,

which

separates

Man 2,

man-nitol from Man al3 mannitol

(data

not

shown).

The results establish that

Man9GIcNAc

con-tains two

branching

mannose residues substi-tutedat0-3and

0-6,

giving

riseto

2,4-dimnethyl

mannose after

methylation,

which must be linked to one another via an

al,6

linkage

as

shown in

Fig.

5Ato

produce

the

fragments

ob-served upon

acetolysis.

TheMan3

fragment

con-tainsoneterminalmannoseresidue

(giving

rise

to

2,3,4,6-tetramethyl mannose),

one residue substituted at0-2

(giving

riseto

3,4,6-trimethyl

mannose),

and one residue substituted at 0-3

(giving

rise to

2,4,6-trimethyl

mannose) which

must have

originally

been one of the branch

point

mannose residues. The other

original

branch

point

mannoseis in the

Man4GlcitolNAc

fragment,

where it issubstituted at 0-3

(giving

riseto

2,4,6-trimethyl mannose).

Also

present

in this

fragment

are oneterminalmannoseresidue and two mannose residues substituted at 0-2.

These results

provide

evidence that the

Man9GIcNAc

has thestructureshown in

Fig.

6.

Structure of the Man5GleNAc

oligosac-charide. When the

Man8GIcNAc

oligosaccha-ride was

digested

with

a-mannosidase,

88% of the

radioactivity

was releasedasfree mannose,

and 12% had the

chromatographic mobility

of

Man

-~---P,

a

GIcNAc,

indicating

the presence of

1201

60~

1 2 3 4

E

0. 1u

M M

GlcitodNAc

160-C MMajor

80-M ,m

GlcitoINAc

D

,M\ XY

~ ~ ~ m

Gc.tolNAc

FIG. 5. Paper chromatography of the fr-agments producedby acetolysis oftheoligosaccharides. The structures attherightindicate thestructuresdeduced

fr-om

the data, with the dashed lines showing the

location of the al,6 linkages that are

preferen-tiaily

cleaved by acetolysis. (A) Man9G1cNAc, (B) Man8GlcNAc, (C) Man7GlcNAc, and (D)

Manr4-GleNAc. Thearrowsatthetop denote themigration

positionsofthefollowingstandardoligosaccharides:

(1)Man4GlcitolNAc,(2) Man3, (3) Man2,and(4) man-n-ose.

sevena-mannose residues. As indicated in

Fig.

4 and Table 1,

methylation

of the

oligosaccha-ride gave rise to

2,4-dimethyl,

3,4,6-trimethyl

and

2,3,4,6-tetramethyl

mannosein the ratio of

40,

1981

on November 10, 2019 by guest

http://jvi.asm.org/

(7)

Man Man Man

al a2al al,2

Man Man Man

al,21 xl

,3\\

/l,6

Man Man

al,3 X\..4

//al

,6

Man

Ia

G1lcNAc

MangGlcNAc

Man Man

al,2 Ial,2

Man Man Man

al,21

al,3\\

Sal,6

Man Man

al,3

\*.

//al

,6 Man

GIcNAc

Man8GlcNAc

Man al,21

Man Man Man

al,2| al,3\ /al,6

Man Man

al,3 /al ,6

Man

GlcNAc

A (major)

Man Man

al 21 |al,2

Man Man

al 2 1xl 3

Man Man

al,3\ /l,6

Man G1cAc

B

Man

|al,2

Man Man Man

al,2 al,3\ / l,6

Man Man

al ,3 'Nl\ /cal,6 Man

Ia

G1lcNAc

(mi nor) C

Man7GlcNAc

Man Man Man

al,21 al,3

\

/cl ,6

Man Man

al,3 \ /l611 Man

GlcNAc

Man

al

.21

Man Man

al,21

cl,3\

Man Man

al,3 \ /al,6

Man

GlcNAc

(minor) (major)

Man6GlcNAc

FIG. 6. Structuresproposedfor theoligosaccharides of E3-gp25K.

2:3:3. Acetolysis of Man8GlcNAc produced two

major fragments corresponding to Man4-GlcitolNAc and Man2(Fig.5B).Molybdate

elec-trophoresis (data not shown) resolved the

re-duced Man2 into two peaks, one comigrating

with Man

012

mannitol and the other

comi-gratingwith Man

al,3,

mannitol in the ratio of1.5:1.0.Thestructureexpectedtoyield these fragments isshown in Fig. 5B and Fig. 6. The

excessMan

al2

mannitolobtained

probably

reflectssomeunderdegradationintheacetolysis reaction, in which only a single cleavage

oc-curred, giving rise to Man

a1,2,

Man and

Man6GlcitolNAc.

The latter wouldaccountfor

theslowest-moving peakonthechromatogram inFig.5B. Themethylationandacetolysisdata

indicate that

MansGlcNAc

has the structure

proposedinFig.6.

on November 10, 2019 by guest

http://jvi.asm.org/

[image:7.497.52.472.49.533.2]

(8)

Structure of the

Man7GlcNAc

oligosac-charide. WhenMan7GlcNAc wastreated with

a-mannosidase,

85% of the

radioactivity

was

re-leasedasfreemannoseand 15%comigrated with Man

1,4,

GlcNAc

disaccharide,

indicating the

presence of six a-mannose residues. The

methylation products

shown in

Fig.

4and Table

1 and the

acetolysis

cleavage

pattern shown in Fig.5C both indicate thepresenceofmorethan

oneisomeric

Man7GlcNAc

structure inthis

oli-gosaccharide.

Acetolysis

produced major

frag-ments

corresponding

to

Man4GlcitolNAc, Man2,

and free mannose; this is consistent with the

structureand

cleavage

sites shown in

Fig.

5C for the major

Man7GlcNAc species. However,

sig-nificantamountsof

fragments corresponding

to

Man3GlcitolNAc

andMan3werealsoproduced, andmolybdate

electrophoresis

revealed that the Man2 contained both Man al,3- Man

(ex-pected

from the

major

species)

and Man

al,2 Man.Toaccountfor theMan3

fragment

andfor the

production

of

2,4,6-trimethyl

man-nose and less thantwo full residues of

2,4-di-methylmannosefrom

Man7GlcNAc,

wepropose

thataminor

species

with the Bstructure occurs

along with the major A structure

(Fig. 6).

To

accountfor the Man3GlcitolNAc

fragment

and

the Mn al,2 Man

fragment,

we alsopropose

that another minor

species

with theCstructure

shownin

Fig.

6alsooccursin

Man7GlcNAc.

Structure of the

Man^GlcNAc

oligosac-charide.

When the

Man6GlcNAc

was treated with

a-mannosidase,

84% of the

radioactivity

wasreleasedasfreemannoseand16%migrated with Man

1,4-

GlcNAc. This indicates five

a-mannose residues.

Methylation

of

Man6-GlcNAc gave the results shown in

Fig.

4 and Table 1,and

acetolysis

produced the fragmen-tationpattemshown in

Fig.

5D. The

fragments

corresponded

to

Man3GlcitolNAc, Man2,

and free mannose and fit the structure with the

cleavage

sites shown in

Fig.

5D.Allof theMan2

wasshown tobe Man

a1,3,

Manby molybdate

electrophoresis;

thus, the structure of the vast

majority of the Man6GlcNAc

oligosaccharide

mustbe as shown inFig.6forthemajor species.

This oligosaccharide should give rise to 2,4-di-methyl, 3,4,6-tri2,4-di-methyl, and2,3,4,6-tetramethyl

mannoseintheratio 2:1:3 andproduceno

2,4,6-trimethyl

mannose.In

fact,

the

methylation

re-sults

(Table

1) indicate thepresenceof0.2 resi-dueof2,4,6-trimethylmannose, 0.2 residue more

3,4,6-trimethyl

mannosethanexpected,and 0.2 residueless

2,4-dimethyl

andtetramethyl man-nose than expected; this suggests that 20%

of

Man6GlcNAc

has thestructure shown in Fig. 6 asthe minor species. This structurewould

pro-duce a Man4GlcitolNAc fragment upon

aceto-lysis, and inspection of Fig. 5D shows a small

radioactive peak at that position.

DISCUSSION

Glycosylation of proteins in mammalian cells

occursbyenbloctransfer ofa

Glc3Man9GlcNAc2

moiety froma

lipid-linked

oligosaccharide donor

(16). Typically, the varioustypes of

high-man-noseand complexoligosaccharides foundon gly-coproteins are generated by processing of

Glc3Man9GlcNAc2

asthenewly synthesized gly-coproteins travel from the rough endoplasmic reticulum through the Golgi region, with

re-moval of the threeglucoses

occurnng

very

rap-idly(14). In ouranalyses of theoligosaccharides released by endo H, we found that the pulse-labeled sample, which should contain newly

synthesized

andmatureE3-gp25K,gaverise to

equal

amounts of

Man9GlcNAc

and Man8-GlcNAc and much smaller amounts of

Man7-GlcNAc

and Man6GlcNAc. In contrast, the

pulse-chase-labeled sample,

which should

con-tain only mature

E3-gp25K,

gave rise to

mainly

MansGlcNAc,

with

considerably

less

MansGlcNAc,

Man7GlcNAc,

andMan6GlcNAc. This implies that

Man9GlcNAc

was processed mainly into Man8GlcNAc during the 3-h chase

period.

The branching pattern and linkage of the mannose residues that we found in the

Man9GlcNAc oligosaccharide

from E3-gp25K

are identical to those in the lipid-linked

Glc3Man9GlcNAc2 previously

described (16), suggesting that E3-gp25K is initially glycosyl-atedby thisprecursoroligosaccharide and then three

glucose

residuesaretrimmedoff.

In the case of

glycoproteins

having

high-mannose

oligosaccharides

smaller than

Mani-GlcNAc2,

there is

generally

some

degree

of

het-erogeneity

in the number ofmannose residues introduced

during

theprocessing reactions

cat-alyzed by

the

Golgi

al,2-mannosidase (26). This

typeof sizeheterogeneity has been described for theheavy-chain

oligosaccharides

ofhuman

im-munoglobulin

M

myeloma

proteins (2, 3, 5), and

inthe presentstudy, this mechanism probably

accountsfor the small amounts ofMan7GlcNAc andMan6GlcNAc

species

observed. In addition

tosize

heterogeneity,

high-mannose oligosaccha-ridescandisplay structural heterogeneityamong

oligosaccharides

of the same size. Although three

possible

Man8GlcNAc oligosaccharide

iso-merscanbeformed

by

the

cleavage

of asingle terminal mannose residue from

Man9GlcNAc,

we found only one isomer, indicating thatthe

removalof the first mannose residue from the

on November 10, 2019 by guest

http://jvi.asm.org/

(9)

E3-gp25K oligosaccharide is specific and not

random. A similar finding has been made for several otherglycoproteins (3, 5).However, since

several Man7GlcNAc isomers and two

Man6GlcNAc isomers were

found,

the

subse-quent mannose

processing

was either less

or-dered, ordifferent patterns of ordered removal occurredatseparate

glycosylation

sitesto gen-eratethesesmaller,

quantitatively

minorspecies. Theoligosaccharides ofE3-gp25Karealltypical high-mannose oligosaccharides having struc-tures identical to those found in a variety of other membrane and soluble

glycoproteins

(13). Other viral

glycoproteins,

for example, the G protein of vesicular stomatitis virus (14), are

initially

glycosylated by

the host cell with a

high-mannose

oligosaccharide,

which is then processed and convertedto a

complex-type

oli-gosaccharide

by

the addition ofouter chain

N-acetylglucosamine,

galactose,

and sialic acid

res-idues and by the addition offucosetothe core

N-acetylglucosamine.

Inthe case of

E3-gp25K,

wefoundnoevidence that suchcomplex chains

were formed, as shown by the fact that the oligosaccharideswere

quantitatively

cleavedby endo H and could be degraded

completely

to

Man g GlcNAc

by

treatmentwith

a-manno-sidase. The factorsthat determine whether the oligosaccharideat agiven

glycosylation

site on aparticular protein will remainahigh-mannose

type orbeconvertedtoacomplextypeby

gly-cosyl

transferases of the

Golgi

apparatus are not

known, but probably the

conformation

ofthe

polypeptide isonefactor.

The DNA sequence of the putativegene for E3-gp25K reveals two potential sites for

Asn-linkedglycosylation (8, 19). Although thereare nodirectdata, circumstantial evidencesuggests

that both sitesmaybe

glycosylated.

For

exam-ple,Persson etal. (20)found4mol of N-acetyl-glucosaminepermol ofE3-gp25K, which is

con-sistent with2mol of

ManxGlcNAc2.

Further, the

apparent molecular weight of nonglycosylated gp25K, obtained

by

infecting

cells in the pres-ence oftunicamycin (19, 20) or by cleaving with endo H (Q. S. Kapoor, W. S. M.Wold, and G.

Chinnadurai, Virology, inpress), isabout 4.5K to 6K smaller than glycosylated gp25K as

estimated by SDS-PAGE. Only 1 mol of

Man8GlcNAc2,

with amolecularweight of 1.7K, is

unlikely

to cause such a large difference. If there are, in fact, 2 mol of the same

MansGlcNAc2

moiety per molofgp25K and if

they are linked to the two Asn-Val-Thr sites

predictedby theDNAsequence, it is ofinterest

that there are no obvious similarities in the amino acidsproximaltothesesites(8, 19).Thus,

the signals that dictate the oligosaccharide

structure at a given glycosylation site are not

obvious from a casualinspection of the amino

acidsequenceof theglycosylation sites; perhaps

these signals are degenerate, or perhaps they

involve protein folding.

ACKNOWLEDGMENTS

Wethank Maurice Green for support and for reading the manuscript and Catherine Devine and Karen Fujimoto for excellenttechnical assistance.

Thiswork was supported by Public Health Service research grantsCA-24710 and CA-08759 and Cancer Center Core grant CA-25843from the National Institutes of Health. W.S.M.W. wassupported by Public Health Service Research Career Development award CA-00650 from the National Institutes of Health.

LITERATURE CITED

1. Brackmann,K.H.,M.Green,W.S. M.Wold,and M.

Cartas. 1980. Identification andpeptide mapping of adenovirus 2-induced earlypolypeptides isolated by two dimensionalgel electrophoresis. J. Biol. Chem. 255: 6772-6779.

2. Chapman, A., and R. Kornfeld. 1979.Structure of the high mannoseoligosaccharides of a human IgM mye-lomaprotein. I. The major oligosaccharides of the two high mannoseglycopeptides. J. Biol. Chem. 254:816-823.

3. Chapman,A., and R.Kornfeld.1979.Structure of the high mannose oligosaccharides of a human IgM mye-loma protein. II. The minor oligosaccharides of high mannoseglycopeptideI. J.Biol. Chem.254:824-828. 4. Chin,W.W.,and J. V.Maizel,Jr.1976.Polypeptides

ofadenovirus.VII. Further studies of early polypeptides in vivo andlocalization of E2 and E2Atothe cellplasma

membrane. Virology 71:518-530.

5. Cohen,R. E., and C. E. Ballou. 1980. Linkage and sequence analysis of mannose-rich glycoprotein core oligosaccharidesbyprotonnuclear magnetic resonance spectroscopy.Biochemistry 19:4345-4358.

6. Green, M., and W.S. M.Wold. 1979.Preparation of humanadenoviruses.MethodsEnzymol. 58:425-435. 7. Hakomori,S.1974.Arapidpermethylationofglycolipid

andpolysaccharide catalyzed by methylsulfinyl carb-anion indimethyl sulfoxide. J. Biochem. (Toyko) 55: 205-208.

8. Herisse,J., G.Courtois,and F. Galibert. 1980. Nu-cleotide sequence of the EcoRI Dfragment of adeno-virus 2 genome. Nucleic Acids Res. 8:2173-2192. 9. Ishibashi, M.,and J. V.Maizel,Jr.1974.The

polypep-tides of adenoviruses. VI.Early and late

glycopolypep-tides.Virology 58:345-361.

10.Jeng, Y-H., W. S. M. Wold, and M. Green. 1978. Evidence for an adenovirustype 2-coded early

glyco-protein. J.Virol. 28:314-323.

11.Johansson, K., H. Persson, A.M.Lewis,U. Petters-son, C.Tibbetts,andL.Philipson.1978.Viral DNA sequences and geneproducts in hamstercells trans-formedby adenovirus type2.J.Virol.27:628-639. 12.Kobata, A.1979. Use of endo-andexoglycosidasesfor

structural studies ofglycoconjugates. Anal. Biochem. 100:1-14.

13. Kornfeld,R., and S. Kornfeld.1980.Structureof

gly-coproteins and theiroligosaccharide units,p.1-34.In W.J. Lennarz(ed.), Thebiochemistryofglycoproteins

and proteoglycans. Plenum Publishing Corp., New York.

14. Kornfeld, S.,E.Li, and I. Tabas.1978.Thesynthesis of complex-typeoligosaccharides. II. Characterization of theprocessing intermediates in the synthesis of the complex oligosaccharideunits of the vesicular

stomati-J. VIROL.

on November 10, 2019 by guest

http://jvi.asm.org/

(10)

tis virus G protein. J. Biol.Chem.253:7771-7778. 15. Kvist,S., LOstberg,H.Persson, L.Philipson,and

P. A. Peterson. 1978.Molecular association between transplantation antigens and cell surface antigen in adenovirus-transformed cell line. Proc. Natl. Acad. Sci. U.S.A. 75:5674-5678.

16.Li, E.,L. Tabas, and S. Kornfeld. 1978. The synthesis of complex-type oligosaccharides. I. Structure of the lipid-linked oligosaccharide precursor of the complex-type oligosaccharides of thevesicular stomatitis virus G pro-tein. J. Biol.Chem.253:7762-7770.

17. Li, Y. T., and S. C. Li. 1972. a-Mannosidase,fl-N-acetyl

hexosaminidase, and B-galactosidase from Jack Bean Meal.Methods Enzymol. 28:702-713.

18.Persson,H., M. Jansson, and L.Philipson.1980. Syn-thesis and genomic site for an adenovirus type 2 early

glycoprotein.J.Mol. Biol. 136:375-394.

19. Persson,H., H. Jornvall, and J.Zabielski 1980. Mul-tiple mRNA species for the precursor to an adenovirus-encoded glycoprotein: identification and structure of thesignal sequence. Proc. Natl. Acad. Sci. U.S.A. 77: 6349-6353.

20. Persson,H.,C.Signas,andL.Philipson. 1979. Purifi-cation and characterization of an early glycoprotein fromadenovirus type 2-infected cells. J. Virol.

29:938-948.

21. Rogers, J., P.Early, C. Carter, K. Calamine, M. Bond, LHood,and R. Wall. 1980. Two mRNAs with differ-ent 3'ends encodemembrane-bound and secreted forms

ofimmunoglubulin g chain. Cell 20:303-312. 22. Rose,J. K., W. J. Welch, B. M. Sefton, F. S. Esch,

andN.C. Ling. 1980. Vesicular stomatitis virus gly-coprotein isanchored in the viral membrane by a hy-drophobic domain near the COOH terminus. Proc. Natl. Acad. Sci. U.S.A. 77:3884-3888.

23.Ross, S. R., S. J. Flint, and A. J. Levine. 1980. Identi-fication of the adenovirus early proteins and their ge-nomic mappositions. Virology 100:419-432.

24. Ross, S., and A. J. Levine. 1979. The genomic map positionof the adenovirus type 2glycoprotein.Virology 99:427-430.

25. Storch, T. G., and J. V. Maizel, Jr. 1980. Theearly

proteins of the nondefectiveAd2-SV40 hybrid viruses: the 19Kglycoprotein is coded by Ad2 earlyregionE3. Virology 103:54-67.

26. Tabas, I., andS. Kornfeld. 1979. Purification and char-acterization of a rat liverGolgi a-mannosidasecapable

of processing asparagine-linked oligosaccharides. J. Biol.Chem. 254:11655-11663.

27. Wold, W. S. M., and M. Green. 1979. Adenovirus type 2early polypeptidesimmunoprecipitated byantisera to fivelines ofadenovirus-transformedratcells. J.Virol. 30:297-310.

28. Wold,W. S.M.,M.Green,and J. K. Mackey.1978. Methods and rationale foranalysis of human tumors for nucleic acid sequences ofoncogenichuman DNA vi-ruses, p. 69-161. In E. Busch(ed.),Methods in cancer research, vol. 15. AcademicPress,Inc., New York.