Plant Physiol. (1988) 87, 420-4260032-0889/88/87/0420/07/$01 .00/0

Glycoprotein Synthesis in Maize Endosperm CellsTHE NUCLEOSIDE DIPHOSPHATE-SUGAR: DOLICHOL-PHOSPHATE GLYCOSYLTRANSFERASES

Received for publication September 14, 1987 and in revised form January 4, 1988

WALTER E. RIEDELL' AND JAN A. MIERNYK*Seed Biosynthesis Research Unit, United States Department of Agriculture, Agricultural Research Service,Northern Regional Research Center, Peoria, Illinois 61604

ABSTRACT

Microsomal membrane preparations from maize (Zea mays L., inbredA636) endosperm cultures contained enzymes that transferred sugar moie-ties from uridine diphosphate-N-acetylglucosamine, guanosine diphos-phate-mannose, and uridine diphosphate-glucose to dolichol-phosphate.These enzyme activities were characterized with respect to detergent andpH optima, substrate kinetic constants, and product and antibiotic in-hibition constants. It was demonstrated by mild acid hydrolysis and highperformance liquid chromatography that the products of the N-acetyl-glucosamine transferases were N-acetylglucosamine-pyrophosphoryl-dol-ichol and N,N'-diacetyl-chitobiosyl-pyrophosphoryl-dolichol and that theproduct of the mannose transferase was mannosyl-phosphoryl-dolichol.A large proportion of the products of the glucose transferase activity wasstable to mild acid hydrolysis. However, the proportion that was labilewas identified as glucosyl-phosphoryl-dolichol. Rate zonal sedimentationand isopycnic banding in linear sucrose density gradients in the presenceof 1 millimolar ethylenediaminetetraacetic acid indicated that the glyco-syltransferase activities were located in the endoplasmic reticulum. Theglycosyltransferases were not solubilized by 500 millimolar KCI or bysequential washes with tris-(hydroxymethyl)aminomethane and water,treatments that release peripheral membrane proteins. Solubilization wasachieved with low concentrations of Triton X-100. When sealed micro-somal vesicles were incubated with trypsin for 30 minutes in absence ofdetergent, the activity of N-acetylglucosaminyl-transferase was substan-tially reduced, while the activity of the glucosyl-transferase was somewhatreduced. Activity of the mannosyl-transferase was resistant to inactivationby incubation with trypsin unless Triton was present.

The oligosaccharides of N-linked glycoproteins are assembledfrom lipid-linked sugar intermediates by the membrane-boundenzymes of the dolichol pathway (8). Formation of the lipid-linked intermediates is achieved by the transfer of GlcNAc, Man,and Glc from nucleotide donors to lipid carriers (31). In theinitial step of the pathway, GlcNAc-1-P is transferred from UDP-GlcNAc to dolichol-P, forming GlcNAc-PP-dolichol. A secondGlcNAc is then donated by UDP-GlcNAc, forming N,N'-diace-tylchitobiosyl-PP-dolichol (16). Mannose residues which makeup the heptasaccharide core are transferred directly from GDP-Man to this disaccharide-lipid (4). The four outer Man residuesare donated by Man-P-dolichol, synthesized from GDP-Man anddolichol-P. Finally, three terminal Glc-residues are transferredfrom Glc-P-dolichol, synthesized from UDP-Glc and dolichol-P,forming the lipid-linked oligosaccharide Glc3Man,GlcNAc2-PP-

I Present address: United States Department of Agriculture, Agri-cultural Research Service, Northern Grain Insects Research Laboratory,RR No. 3, Brookings, SD 57006.

dolichol (24).Studies with mammalian cells and yeast (16, 32) have shown

that the enzymes of the dolichol cycle are associated with theER. The assembly of Man,GlcNAc2-PP-dolichol is thought totake place on the cytoplasmic surface of the ER. Subsequently,this oligosaccharide is translocated to the lumen of the ER whereadditional Man- and Glc-residues are transferred from lipid-car-riers, forming the final tetradeccasaccharide-PP-lipid (21, 33).The oligosaccharide is then transferred en bloc from the lipidcarrier to the nascent polypeptide in a cotranslational event (21).The first steps of oligosaccharide processing (e.g. removal ofterminal glucose residues and, in mammalian cells, at least onemannose residue) also occur within the ER (16). Subsequentsteps of oligosaccharide processing and maturation take place inthe Golgi apparatus and vacuolesMaize endosperm cells cultured in vitro secrete acid hydrolases

into the culture medium (27). All of the hydrolases appear tobe glycoproteins. These cells offer an excellent system with whichto examine protein glycosylation and its potential role in thesecretory process. As an initial step we have characterized theNDP2-glycose: dolichol-P glycosyltransferases and determinedtheir subcellular localization.

MATERIALS AND METHODS

Reagents. New England Nuclear3 provided the radioisotopesUDP-[6-3H]GlcNAc (20.4 Ci/mmol), GDP-[1-3H]mannose (10.9Ci/mmol), and UDP-[1-3H]glucose (10.4 Ci/mmol). Dolichol-phosphate, unlabeled sugar nucleotides, BSA, trypsin, soybeantrypsin-inhibitor, and tunicamycin were from the Sigma Chem-ical Company. Buffers were from Research Organics, Inc. Se-pharose-4B was from Pharmacia. Genzyme Corporation suppliedthe (+ )-1-deoxynojirimycin. Ecoscint scintillation fluid was fromResearch Diagnostics. BioRad supplied the reagents for proteinquantitation. All materials were of the highest grade commer-cially available.

Tissue Culture. The suspension cultures were originally de-rived from 6- to 10-d-old maize (Zea mays L., inbred A636)endosperm. The medium, culture conditions, and transfer regimehave been previously described (27).Enzyme Assays. Glucan synthetase I and IDPase were assayed

by the method of Green (13). To determine the latent incrementin IDPase activity, the initial enzyme activity was subtracted fromthe total activity after 3 d of storage at 4°C (28). Other markerenzymes were assayed by established procedures (26).

Glycosyltransferase activites were assayed by measuring theconversion of radioactive sugars from nucleoside diphosphate-

2 Abbreviation: NDP, nucleoside-diphosphate.3Names of vendors are included for the benefit of the reader and do

not imply endorsement or preferential treatment by the United StatesDepartment of Agriculture.

420

MAIZE ENDOSPERM GLYCOSYLTRANSFERASES

sugars into products soluble in chloroform:methanol 2:1, v/v).Solvent was removed from 20 ,g of dolichol-P by evaporationwith a stream of N2-gas. A solution (450 Al final volume) con-taining 0.015% Triton X-100, 50mM TES (pH 7.0), 10 mM MgCl2plus 50 or 100 Al of membrane preparation was added. Reactionwas initiated by addition of 50 ,u of 60 ,mM NDP-sugar (200,000dpm). After mixing, the assays were conducted at 30°C. Underthese assay conditions, the reactions were linear for at least 20min. Reactions were stopped by the addition of 2.5 ml chloro-form:methanol (2:1) and, after mixing, the lower organic phasewas removed. The aqueous phase was reextracted, the combinedorganic phases were pooled and washed with chloroform:methanol:H2O (3:48:47, v/v/v), then the lower organic phase wasremoved for liquid scintillation spectrometry. When inhibitorsof the enzyme activities were tested, they were preincubated inthe assay mixture for 1 min prior to addition of the NDP-sugar.It was necessary to add 2 mm DTT and 20 /LM deoxynojirimycinto assay mixtures in order to reliably assay Glc-transferase ac-tivity.

Product Analysis. Mild acid hydrolysis of the products of theglycosyl transferase reactions was performed after pooling thewashed chloroform:methanol 2:1 fractions from four separatereaction mixtures. One ml was removed for scintillation count-ing. The remaining extract was dried under a stream of N2,dissolved in 1 ml 10 mm HCl in 50% propanol, and then incu-bated at 100°C for 20 min. After cooling, the hydrolysate wasmixed with 1 ml chloroform, and the phases were separatedbefore removal of aliquots for scintillation counting. The aqueousphase after mild hydrolysis was further analyzed by HPLC witha Bio-Rad Sugar-Pak I column at 90°C, using 0.1 mm CaNa2-EDTA as the mobile phase, at a flow rate of 0.5 ml min-'.Homogenization and Centrifugation. Culture medium was re-

moved by filtration through two discs of Whatman No. 1 paperunder reduced pressure. The tissue was then washed with 60 mmsucrose containing 50 mM CaCl2. Fifteen g fresh weight of tissue,10 to 14 d after transfer, was gently homogenized for 5 min witha mortar and pestle in 30 ml of ice-cold homogenization bufferconsisting of 100 mM TES (pH 7.5), 1 mM EDTA, 500 mMsucrose, and 0.1% (w/v) BSA. All further manipulations wereat 0 to 4°C. The homogenate was filtered through 2 layers ofcheesecloth, and the filtrate was centrifuged for 10 min at 2,000g(all g-forces were determined from rmax values of the respectiverotors) using a JS-13.1 rotor in a Beckman J2-21 centrifuge. Theresulting pellet was discarded, and the supernatant was furthercentrifuged for 10 min at 10,000g. The resulting membrane pelletwas carefully resuspended in a small volume of homogenizationbuffer, and the supernatant was further centrifuged for 1 h at100,000g using an SW-28 rotor in a Beckman L8-70M preparativeultracentrifuge. The supernatant was removed and saved, andthe resulting microsomal membrane pellet was carefully resus-pended in a small volume of homogenization buffer.

Organelle Isolation. Ten ml of 2,000g supernatant was loadedonto a 1.6 x 20 cm column of Sepharose-4B (17), equilibrated,and run with homogenization buffer as the mobile phase. Two-ml fractions were collected and assayed. The turbid vesicle-con-taining fractions eluting in the void volume were combined andlayered over a 24-ml linear gradient of 20 to 50% (w/w) sucrosein 50 mM TES (pH 7.5), 2 mM DTT, 1 mm EDTA, and 0.1%BSA, over a 2-ml cushion of 60% (w/w) sucrose. Gradients werecentrifuged for 14 h at 70,000g in an SW-28 rotor. One-ml frac-tions were collected and analyzed.

Topographical Orientation of the Glycosyl-Transferases. Roughmicrosomal membrane vesicles, prepared wth homogenizationbuffer containing 1 mM MgCl2 in place of the EDTA, wereresuspended in homogenization buffer at a final protein concen-tration of 2.5 mg ml- 1. Three-ml aliquots were placed into eachof two test tubes. Triton X-100 was added to one tube to a final

concentration of 0.01%. After brief mixing, trypsin was addedto each tube to a final concentration of 30 Ag mg protein- '. Thetubes were incubated on ice and aliquots were removed at theindicated times. A 10-fold excess of soybean trypsin inhibitorwas immediately added to each aliquot prior to assay of enzymeactivity. In parallel with the isolation of microsomal vesicles forenzyme assays, a tissue sample was homogenized in the presenceof 10 ,uCi of 3H-inulin. Some of the inulin was entrapped duringthe formation of vesicles by fragmentation of the endomembranenetwork. Release of radioactivity from the vesicles can serve asa sensitive measure of membrane integrity during in vitro pro-teolysis. After treatment, the vesicles were sedimented in theultracentrifuge, and the proportion of radioactivity in the su-pernatant and pellet was quantitated by liquid scintillation spec-trometry. Treatment of the vesicles with Triton X-100 releasedall of the entrapped inulin. Using this method, it was found thatstability of the isolated maize vesicles during in vitro manipu-lation was always greater than 90%.Enzyme Solubilization. Microsomal membrane fractions were

resuspended in 100 mm TES (pH 7.5). One-ml aliquots (1.5 mgmembrane protein) were placed into centrifuge tubes, and TritonX-100 or KCl was added to the indicated final concentrations.The tubes were gently shaken on ice for 20 min. Membraneswere subsequently pelleted by centrifugation at 200,000g usinga TLA-100.2 rotor in a Beckman TL-100 ultracentrifuge. Mem-brane pellets were resuspended in buffer prior to assaying forenzyme activity.Other Analytical Methods. Sucrose concentrations were meas-

ured with a Bausch and Lomb refractometer. Absorbance at 280nm was measured spectrophotometrically using 100 Al aliquotsplus 900 Al H20. Protein was quantitated by the method ofBradford (2), using bovine y-globulin as the standard. Ecoscintscintillation fluid and the dpm program of a TM Analytic MarkIII liquid scintillation spectrometer were used for measurementof radioactivity. Analysis of enzyme kinetic data was by themethod of Garland and Dennis (12), which employs iterativecurve fitting by nonlinear regression.

RESULTS

Characterstics of the glycosyltransferase were determined fromtotal microsomal membrane preparations from maize endospermsuspension cultures (Table I). The pH optima of all three enzymeactivities were near 7. The Km values for nucleoside diphosphatesugars were in the micromolar range. Divalent cations, with Mg2+> Mn2 + >> Ca2 + > Co2 + > Zn22, were required for maximalactivities of all three enzymes. The Km values for dolichol-P werein the Ag/ml range (dolichol is a mixture of homologs of slightlydifferent size, requiring the expression of Km values in massrather than in molar concentration). A detergent was also re-quired for maximal transferase activities. Optima for Triton X-100 were from 0.01 to 0.015%.The antibiotic tunicamycin (commercial tunicamycin is also a

mixture of homologs) was a potent inhibitor of the incorporationof GlcNAc into glycolipid (Table I). Higher concentrations oftunicamycin also inhibited the incorporation of Man and Glc intoglycolipids. Uridine and guanosine nucleoside mono- and di-phosphates also inhibited glycosyltransferase activities. For theGlcNAc-transferase, the Ki for UMP was lower than the Ki forUDP. In contrast, Ki values for nucleoside-diphosphates werelower than Ki values for nucleoside monophosphates for theother two transferases.The lipid-soluble products of the GlcNAc- and Man-transfer-

ases were labile to mild acid hydrolysis (Table II). However, aconsiderable proportion of the products of the Glc-transferasereaction was stable under these conditions and remained in theorganic phase. The aqueous samples after mild acid hydrolysiswere analyzed by HPLC in order to characterize the released

421

Plant Physiol. Vol. 87, 1988

Table I. Some Characteristics of the Maize Endosperm Microsomal Nucleoside Diphosphate:Dolichol-Phosphate Glycosyltransferases

Kinetic constants were derived from initial-rate studies analyzed by iterative curve-fitting by nonlinearregression (12).

UDP-GlcNAc GDP-Man UDP-GlcTransferase Transferase Transferase

Optimum (Triton X-100) (%) 0.010 0.015 0.015pH optimum 7.5 7.0 6.5Km, NDP-sugar (,tm) 11.5 ± 1.4 1.73 ± 0.12 5.76 ± 0.89Km, Mg2+ (mM) 1.33 ± 0.07 1.31 + 0.23 0.71 ± 0.09Ki, Dolichol-P (,ug/ml) 1.70 ± 0.22 3.04 ± 0.54 0.58 ± 0.05Ki, Tunicamycin (,ug/ml) 0.014 ± 0.002 44.5 ± 2.8 37.9 ± 5.8K,, UDP (/xM) 535 ± 22 NDa 146 ± 22K1, UMP (,uM) 405 ± 59 ND 657 ± 54

K&, GDP (,uM) ND 18 ± 1 ND

Ki, GMP (/aM) ND 235 + 10 ND

a Not determined.

Table II. Analysis of the Reaction Products of the Maize Endosperm Microsomal Glycosyltransferases byMild Acid Hydrolysis plus Solvent Partitioning

Washed chloroform:methanol fractions were dried, dissolved in 10 mm HCl in 50% propanol, then incubatedat 100°C for 20 min. After cooling, hydrolysates were mixed with an equal volume of chloroform and thephases separated before removing aliquots for liquid scintillation spectrometry. Data presented are from arepresentative experiment.

Initial CM Fraction Aqueous Phase Organic PhaseTransferase before Hydrolysis after Hydrolysis after Hydrolysis

dpm (% of initial fraction)GlcNAc 34,477 (100) 26,763 (78) 7,714 (22)Man 16,245 (100) 15,849 (98) 396 (2)Glc 51,786 (100) 13,320 (25) 38,466 (74)

carbohydrate moietes (data not shown). The products of theGlcNAc transferase reaction gave a broad peak of radioactivity,approximately 70% of which had a retention time identical tothat of N,N'-diacetylchitobiose (7.51 min). No more than 20%of the radioactivity was detected at the retention time corre-

sponding to GlcNAc (9.71 min). The products of the Man- andGlc-transferase reactions had retention times identical to thoseof Man (10.66 min) and Glc (9.51 min), respectively.A balance sheet showing the distribution of marker enzyme

activities, glycosyltransferase activities, and protein after rate-zonal sedimentation is presented in Table III. Alcohol dehydrog-enase activity was found almost exclusively in the 100,OOOg su-

pernatant fraction. Succinate dehydrogenase activity was en-

riched in the 10,OOOg pellet. Catalase activity was found equallydistributed between the 10,OOOg and 100,OOOg pellets and the100,OOOg supernatant. The activities of NADH:Cyt c reductase,latent IDPase, glucan synthetase I, and the glycosyl-transferaseswere enriched in the 100,OOOg pellet. Co-sedimentation of themajority of the glycosyltransferase activities with NADH:Cyt c

reductase, latent IDPase, and glucan synthetase I activities in-dicates that the glycosyltransferases were associated with theendomembrane system.

Rather than preparing a microsomal fraction by ultracentri-fugation, membrane vesicles were separated from the solublefraction by gel permeation chromatography (Fig. 1). The activ-ities of NADH:Cyt c reductase and Man-transferase eluted in

Table III. Enzyme Distribution in Subcellular Fractions Isolated from Maize Endosperm Cultures and Separated by Rate-Zonal SedimentationData presented are enzyme activities followed by percent recoveries in parentheses.

FractionEnzyme 10,00Og 100,000g 100,000g

Homogenate Pellet Pellet Supernatant

Alcohol dehydrogenase 17.03a (100) 0.34 (2) 0.15 (1) 16.5 (97)Succinate dehydrogenase 0.72a (100) 0.50 (69) 0.14 (19) 0.05 (6)NADH-Cyt c reductase 20.80a (100) 1.58 (8) 12.63 (61) 7.42 (35)IDPase (latent increment) 1.26b (100) 0.49 (32) 0.83 (66) 0.00 (0)Glucan synthetase I 0.35c (100) 0.09 (26) 0.22 (63) 0.05 (14)Catalase 22585d (100) 5252 (23) 9218 (41) 6937 (31)GlcNAc-transferase 1.97b (100) 0.53 (27) 1.34 (68) 0.03 (1)Man-transferase 10.91b (100) 2.05 (18) 5.34 (49) 1.46 (13)Glc-transferase 4.73b (100) 1.43 (30) 2.14 (45) 0.14 (3)Protein 75.0c (100) 8.02 (11) 7.50 (10) 65.70 (88)

a ,.mol min- l. b ,bmol h - l. c pmol h- . d Luck units. c mg fraction - 1.

422 RIEDELL AND MIERNYK

MAIZE ENDOSPERM GLYCOSYLTRANSFERASES

'5

L)EN

LLJa)

0).>a:

0)U)

a)

0C0

-c

01)

0

8-0

0

0.50.15

0.130.4

0.3

0.2

00

U8Z.

0.1

0.11

0.09

0.07

0.051.0

0.8

a)aL) Enun

D Ut(AC

U I4c

U 2o 0

.5t

u

a)ENc

01)a4aa)

Fraction Number

FIG. 1. Separation of membrane vesicle and soluble fractions of amaize endosperm cell homogenate by gel permeation chromatography.Ten ml of a 2000g supernatant was loaded onto a Sepharose-4B columnpreviously equilibrated with homogenization buffer. Two ml fractionswere collected and assayed. Peak fraction enzyme activities are: alcoholdehydrogenase, 1.24 ,umol min-', Cyt c reductase, 1.29 ,umol min-1;Man-transferase, 0.48 ttmol h-'.

the void volume of the column, coincident with a peak of A2,-absorbing material. Alcohol dehydrogenase activity, coincidentwith a second peak of A280-absorbing material, eluted within theincluded volume of the column.Membrane vesicles prepared by Sepharose-4B chromatogra-

phy in the presence of 1 mM EDTA were resolved by centrifu-gation on linear 20 to 50% sucrose gradients for 14 h at 70,000g(Fig. 2). The first two protein peaks in the gradient correspondto peak activities of NADH:Cyt c reductase (1.11 g/cm3) andIDPase (1.13 g/cm3). The protein peaks at a density of 1.18 and1.24 g/cm3 represent mitochondria (succinate dehydrogenase)and peroxisomes (catalase) (data not shown). Glycosyltransfer-ase activities were coincident with the peak of antimycin A-insensitive NADH:Cyt c reductase activity, suggesting an ERlocalization of these enzymes.Microsomal membrane preparations were washed by the Tris-

water-Tris method of Dallner (5) to remove peripheral proteins.This washing procedure did not release the glycosyltransferaseactivities from the membranes, nor did incubation with 0.05 to0.5 M KCl (data not shown). The glycosyltransferases were sol-ubilized by Triton X-100 (Fig. 3), however, solubilization wasaccompanied by a 30% loss of total (soluble + particulate) ac-tivity in each case.Microsomal membrane vesicles prepared in the presence of 1

mM MgCl2 were treated with trypsin or Triton plus trypsin (Fig.4). Treatment of the vesicles with trypsin alone inactivated 75to 80% of the Cyt c reductase and 60 to 65% of the GlcNAc-transferase activities within 30 min, but only 10 to 30% of theGlc- and Man-transferase activities. A brief pretreatment of thevesicles with Triton X-100 had no effect upon the extent of in-activation of Cyt c reductase or the GlcNAc-transferase, butsubstantially increased the proportions of Glc- and Man-trans-ferases that were inactivated.

0.6

0.4

0.2

0.01.0

0.8

0.6

0.4

0.2 j

0.0 L0 5 10 15 20

Fraction Number

1.44

x1.42 .o

C

a)1.40 '

00

1.38 W4)0

1 .36

a)0

0una,a)

01 (Ac0

U au a)

tn0

)asU

(L) 0L.

0)

UacI~L.

<'IC)

C5 24 4

cn

a)tnaa)C:0

U

x

25 30

FIG. 2. Separation of maize endosperm culture organelles by iso-pycnic banding in linear sucrose gradients. Microsomes were separatedfrom soluble components by gel-permeation chromatography, then lay-ered onto a linear 20 to 50% (w/w) linear sucrose gradient. Peak fractionenzyme activities are: Cyt c reductase, 0.216 ,umol min-1; IDPase, 0.646,umol h -'; GlcNAc-, Man-, and Glc-transferase, 0.004, 0.220 and 0.017,tmol h- , respectively.

DISCUSSION

The pH optima and kinetic characteristics of the maize en-dosperm glycosyltransferases were similar to those described forthese enzymes from other plant sources (11, 18, 19, 30). Theantibiotic tunicamycin, which resembles a transition-state reac-tion intermediate, inhibits the first enzyme of the dolichol path-way blocking the formation of GlcNAc-PP-dolichol (19). Lowconcentrations of tunicamycin inhibited maize endosperm UDP-GLcNAc:dolichol-P GlcNAc-1-P transferase. Higher concentra-tions (40 ,tg/ml) also inhibited the transfer of Man and Glc fromGDP-Man and UDP-Glc to chloroform:methanol soluble prod-ucts. Our observations are similar to some of those by Elbein etal. (7), but are in contrast to those of Ericson et al. (10) andothers of Elbein et al. (7), who reported that tunicamycin at 5to 500 u.g/ml had no effect on the synthesis of mannosyl-P-dol-ichol. The basis of tunicamycin inhibition of the maize Man- andGlc-transferase activities is unknown.

In mammalian systems, the rate at which new lipid-linkedoligosaccharides are synthesized is proportional to the rate atwhich preassembled oligosaccharide chains are transferred toprotein (16). It has been suggested that the availability of dol-Pfor lipid-linked oligosaccharide synthesis could be rate-limiting(15). Lipid-linked oligosaccharides are assembled from lipid-linkedintermediates in a sequential manner starting with the synthesisof GlcNAc-PP-dolichol. Reactions leading to the assembly ofMan-P-dolichol and Glc-P-dolichol could compete with those

423

JO.0

RIEDELL AND MIERNYK

80

FIG. 3. Solubilization of glycosyltransferaseactivities from microsomal membrane prepara-

tions of maize endosperm cultures. Membraneswere mixed with Triton X-100 at the indicatedfinal concentrations, the particulate material re-

moved by centrifugation, and the supernatant

fractions assayed for enzyme activities.

N

V)

QJ)

60

40

20

Plant Physiol. Vol. 87, 1988

0.0 0.1 0.2 0.3 0.4

a)(tnCo

L.

co

L-

0

z

0

0

Co

L.Q)4-Coc

0

L-

o

0

0

0.5

[Triton X-100] (%)

0.8

0.6

* 0.4

<0.2a1)ENC=L 1.0a)

0.8 _-

0.6 _

0.4 _

0.2

0.00 5 10 20 30

Minutes

a)cnC,,

LO)0) CD

c 0 D

4*aU 5'0 C

< O

FIG. 4. Topography of the maize endosperm microsomal glycosyl-transferases. Rough microsomal vesicles prepared from maize endo-sperm cultures were incubated on ice with 30 ,ug trypsin mg microsomalprotein-'. At the indicated times a 10-fold excess of soybean trypsininhibitor was added to aliquots of the reaction mixtures which were thenassayed for enzyme activity. Zero-time enzyme activities were: NADHCyt c reductase, 1.8 ,tmol min- mg protein- 1; GlcNAc-. Man-, andGlc-transferases, 0.2, 0.7, and 0.3 ,tmol hr- I mg protein- 1, respectively.

reactions leading to the formation of GlcNAc-PP-dolichol forthe limited amount of dolichol-P. Recently, Kaushal and Elbein(19) have reported that GDP-Man and UDP-Glc inhibited theGlcNAc-1-P transferase from membrane preparations of cul-tured soybean cells. Increased amounts of dolichol-P in the assaymixture relieved these inhibitions, indicating that the competi-tion for dolichol-P by the Man- and Glc-transferases inhibitedthe GlcNAc-1-P transferase. If, however, the transferases aresubject to product inhibition in vivo, as indicated by the results

presented in Table I, then the competition for dolichol-P mightbe less important.Most of the glycolipids produced in the glycosyltransferase

reactions were labile to mild acid hydrolysis, indicating attach-ment of the carbohydrate to an a-saturated polyprenol (30). Anexception was approximately 75% of the Glc-transferase reactionproducts. Typically, most of the Glc that is transferred fromUDP-Glc to lipid-acceptors in plants is transferred to sterols (9).The carbohydrate-sterol linkage is stable to mild acid hydrolysis(30). Microsomal membrane preparations contain a large pro-portion of the total sterols found in maize tissue (20), suggestingthat the product of the Glc-transferase reaction found in theorganic phase after mild acid hydrolysis might be steryl glucosids.Approximately 25% of the lipid-soluble products of the Glc-transferase reaction were labile to mild acid hydrolysis. The en-zymic transfer of Glc from UDP-Glc to dolichol-P required di-valent cations and was inhibited by EDTA. Transfer of Glc fromUDP-GIc to sterol does not require divalent cations (30) andpresumably is not inhibited by EDTA. In this report, activity inthe presence of 10 mm EDTA and in the absence of exogenousdolichol-P was subtracted from the activity requiring 10 mM MgCl2and 10 ,ug dolichol-P.

Mild acid hydrolysis of the products of the Man- and Glc-transferase reactions released radioactive water-soluble com-pounds which chromatographed as single peaks. The retentiontimes of these peaks were identical to authentic mannose andglucose, respectively. Similar analysis of the products of theGlcNAc-transferase reaction yielded two radioactive compoundswhich were not well resolved by our chromatographic system.Integration of the peaks at 7.51 and 9.71 min indicated thatapproximately 70% of the radioactivity was associated with N,N'-diacetylchitobiose, while no more than 20% was GlcNAc. Aproduct of the first GlcNAc-transferase reaction is GlcNAc-PP-dolichol, the substrate for the second GlcNAc-transferase. It isnot then surprising that the reaction mixture from the maizetransferases contained the products of both enzymes. Similarresults were obtained by Brett and Leloir (3) and by Lehle et al.(23), although in both cases GlcNAc predominated. The char-acterization of the GlcNAc-transferase presented in Table I isactually the sum of UDP-GlcNAc:dolichol-P GlcNAc-1-P-trans-ferase plus UDP-GlcNAc:dolichol-PP-GlcNAc GlcNAc-trans-ferase activities. Further experiments with individual, purifiedenzymes will be required to distinguish the contribution of eachactivity to the observed sum.

Having partially characterized the glycosyltransferases, andtheir association with the endomembrane system of the maize

424

II . 0 k qk-----O-l

MAIZE ENDOSPERM GLYCOSYLTRANSFERASES

endosperm cells, we wished to examine in detail their subcellularlocalization. The GlcNAc-, Man-, and Glc-transferase activitiesof maize endosperm cultures were coincident with the ER marker-enzyme NADH-Cyt c reductase when stripped microsomes wereresolved on linear sucrose gradients and were well separated fromthe Golgi marker IDPase. There was no activity associated withthe mitochondrial or peroxisomal fractions. Glycosyltransferasesof the dolichol-cycle were coincident with ER marker enzmes inpeas (6, 29), mung beans (22), and castor oil seeds (25). Likewise,in mammalian cells and yeast the enzymes of the dolichol cycleand N-glycosylation are located in the ER (15, 32).A low level of Glc-transferase activity coincident with the Golgi

apparatus marker IDPase was observed when gradient fractionswere assayed in the absence of exogenous dolichol-P and in thepresence of EDTA. This activity probably represents UDP-Glc:sterol Glc-transferase. In other plant systems the glycosyl-transferases involved in sterol-glycoside and polysaccharide bio-synthesis were located within the Golgi apparatus (1, 6).The classical methods of Dallner (5) and several more recent

variants, including rate-zonal sedimentation, rate-zonal sedi-mentation in discontinuous Ficoll gradients, and flotation in lin-ear and discontinuous sucrose gradients, were used in attemptsto fractionate total maize ER membranes into rough (ribosome-studded) and smooth vesicles. Despite evaluating a variety ofmarkers, we were unable to obtain more than a twofold enrich-ment in any of the rough-ER preparations (data not presented).There was a similar enrichment in the activities of the glycosyl-transferase, but we feel that the results were too equivocal fora definitive fine localization. It should be noted, however, thatsimilar results by others (14) have been interpreted as demon-stration of an exclusively rough-ER localization.Treatments intended to remove adsorbed, peripheral, and cis-

ternal membrane protein from the maize endosperm microsomalmembrane preparations did not solubilize the GlcNAc-, Man-,or Glc-transferase activities. However solubilization of glyco-syltransferase activities was achieved with low concenrations ofTriton X-100. These observations indicate that the glycosyltrans-ferases are integral membrane proteins.The topographical orientation of the glycosyltransferases was

investigated by treating intact and detergent-disrupted micro-somal vesicles with trypsin. Under conditions where membraneintegrity was maintained, trypsin treatment resulted in a reduc-tion in the activities of NADH:Cyt c reductase and GlcNAc-transferase. Permeabilization of the vesicles with Triton did notgive any increase in protease sensitivity. Mammalian Cyt c re-ductase is an integral membrane protein with the active site facingthe cytosol. Based upon our results, it appears that this is alsothe case for the maize endosperm reductase. The protease sen-sitivity of the GlcNAc-transferase is similar to that of Cyt creductase, suggesting that this enzyme is also oriented with theactive site facing the cytosol. This conclusion is in agreementwith that of Snider et al. (33) for the GlcNAc-transferase frommammalian microsomal membranes. Activity of Man-transfer-ase, and to a lesser extent Glc-transferase, was resistant to trypsininactivation in the absence of detergent. The protease sensitivityof the Man-transferase was greatly increased by membrane dis-ruption while that of the Glc-transferase was only slightly in-creased. These results are consistent with the proposal that theactive sites of the Man- and Glc-transferase face the lumen ofthe ER. Similar results were obtained when pepsin or subtilisinwas used rather than trypsin.

Overall, our observations concerning the protease sensitivityof the maize endosperm glycosyltransferases agree with the pro-posal that the Man3 5GlcNAc2-PP-dolichol oligosaccharides facethe cytosol whereas the Glc, 3Man6 9GlcNAc2-PP-dolichol oli-gosaccharides face the lumen of the ER during assembly in vivo(reviewed in Ref. 21). Maize endosperm glycosyltransferases are

integral membrane proteins localized in the endoplasmic retic-ulum. This cellular localization of the enzymes that assemblelipid-linked intermediates used to build the oligosaccharide sidechains of glycoproteins is consistent with reports that glycosy-lation occurs on nascent polypeptides in the lumen of the rough-ER (21, 24, 32).Maize endosperm microsomal preparations contain more than

one enzyme that transfers Glc from UDP-Glc to chloro-form:methanol soluble products. This interfering activity may beUDP-Glc:sterol Glc-transferase. Kinetic characterization of theGlc-transferase that synthesizes Glc-phosphoryl-dolichol wascomplicated by the presence of the interfering enzyme(s). Partialpurification and characterization of the UDP-Glc:dolichol-P Glc-transferase is in progress and will be the topic of a future com-munication.

Ackntowledgments-Thanks are due J. F. VanMiddlesworth for assistance withthe HPLC analyses and Drs. A. H. C. Huang and J. Nagahashi for their constructivecomments on the manuscript. The excellent technical assistance of C. Kessler isgratefully acknowledged.

LITERATURE CITED

1. BOWLES DJ. L LEHLE. H KAUSS 1977 Glucosylation of sterols and polypren-olphosphate in the Golgi apparatus of Phaseolus aureus. Planta 134: 177-181

2. BRADFORD MM 1976 A rapid and sensitive method for the quantitation ofmicrogram quantities of protein utilizing the principle of protein-dye binding.Anal Biochem 72: 248-254

3. BRETT CT. LF LELOIR 1977 Dolichol monophosphate and its sugar derivativesin plants. Biochem J 161: 93- 101

4. CHAPMAN A. K FUJIMOTO. S KORNFIELD 1980 The primary glycosylationdefect in Class E Thy-1-negative mutant mouse lymphoma cells is an inabilityto synthesize dolichol-P-mannose. J Biol Chem 255: 4441-4446

5. DALLNER G 1974 Isolation of rough and smooth microsomes-General. MethodsEnzymol 31: 191-201

6. DURR M. DS BAILEY, G MACLACHLAN 1979 Subcellular distribution of mem-brane-bound glycosyltransferases from pea stems. Eur J Biochem 97: 445-453

7. ELBEIN AD. J GAFFORD, MS KANG 1979 Inhibition of lipid-linked saccharidesynthesis: Comparison of tunicamycin, streptovirudin. and antibiotic 24010.Arch Biochem Biophys 196: 311-318

8. ELBEIN AD 1979 The role of lipid-linked saccharides in the biosynthesis ofcomplex carbohydrates. Annu Rev Plant Physiol 30: 239-272

9. ELBEIN AD, WT FORSEE 1976 Biosynthesis of steryl glucosides and acylatedsteryl glucosides in plants. In L Whitting, ed, Glycolipid Methodology,American Oil Chemical Society. Champaign, IL. pp 345-368

10. ERICSON MC, JT GAFFORD, AD ELBEIN 1977 Tunicamycin inhibits GIcNAclipid formation in plants. J Biol Chem 252: 7431-7433

11. FORSEE WT, AD ELBEIN 1973 Biosynthesis of mannosyl- and glycosyl-phos-phoryl-polyprenols in cotton fibers. J Biol Chem 248: 2858-2867

12. GARLAND WH. DT DENNIS 1977 The steady-state kinetics of glutamate de-hydrogenase from Pisuwn sativinn L. mitochondria. Arch Biochem Biophys182: 614-625

13. GREEN JR 1983 The Golgi apparatus. In JL Hall. AL Moore. eds. Isolationof Membranes and Organelles from Plant Cells. Academic Press. New York.pp 135-152

14. HORTSCH M, DI MEYER 1985 Immunochemical analysis of rough and smoothmicrosomes from rat liver. Segregation of docking protein in rough mem-branes. Eur J Biochem 150: 559-564

15. HUBBARD SC. RJ IvATT 1981 Synthesis and processing of asparagine-linkedoligosaccharides. Annu Rev Biochem 50: 555-583

16. HUBBARD SC. PW ROBBINS 1980 Synthesis of the N-linked oligosaccharidesof glycoproteins. Assembly of the lipid-linked precursor oligosaccharide andits relation to protein synthesis ini vsivo. J Biol Chem 255: 11782-11793

17. JONEs RL 1980 The isolation of endoplasmic reticulum from barley aleuronelayers. Planta 150: 58-69

18. KAUSHAL GP. AD ELBEIN 1986 Purification and properties of UDP-GIcNAc:dolichyl-pyrophosphoryl-GlcNAc GlcNAc-transferase from mungbean seedlings. Plant Physiol 81: 1086-1091

19. KAUSHAL GP. AD ELBEIN 1986 Properties of solubilized UDP-GlcNAc:dolichol-phosphate GlcNAc-1-P-transferase from soybean cultured cells. Plant Phys-iol 82: 748-752

20. KEMP RJ. El MERCER 1968 Studies on the sterols and sterol esters of theintracellular organelles of maize shoots. Biochem J 110: 119-125

21. KORNFIELD R. S KORNFIELD 1985 Assembly of asparagine-linked oligosac-charides. Annu Rev Biochem 54: 631-664

22. LEHLE L. DJ BOWLES, W TANNER 1978 Subcellular site of mannosyl transferto dolichol phosphate in Phaseolus aureus. Plant Sci Lett 11: 27-34

425

426 RIEDELL Al

23. LI.HLE L, F FARTACZEK, W TANNER. H KAUSS 1976 Formation of polyprenolmono and oligosaccharides in Plaseoluls alurelus. Arch Biochem Biophvs 175:419-426

24. LEHLE L. W TANNER 183 Polyprenol-linked sugars and glycoprotein synthesisin plants. Biochem Soc Trans 11: 568-574

25. MARRIOTT KM, W TANNER 1979 Dolichylphosphate-dependent glycosvl trans-fer reactions in the endoplasmic reticulum of castor bean endosperm. PlantPhysiol 64: 445-449

26. MIERNYK JA 1985 The isolation and characterization of nongreen plastids. IIIHF Linskens, JF Jackson, eds, Modern Methods of Plant Analysis, NewSeries, Vol I, Cell Components. Springer-Verlag, Berlin, pp 259-295

27. MIERNYK JA 1987 Extracellular secretion of acid hydrolases bv maize endo-sperm cells grown in liquid medium. J Plant Physiol 129: 19-32

28. MORRE DJ, CA LEMBI, WJ VAN DER WOUDE 1977 A latent inosine-5 -di-

D MIERNYK Plant Physiol. Vol. 87, 1988

phosphatase associated with Golgi apparatus-rich fractions from onion stcm.Cytobiologie 16: 72-81

29. NAGAHASHI J, L BEEVERS 1978 Subcellular localization of glycosyltransferasesinvolved in glycoprotein biosynthesis in the cotyledons of Pislitn sativiiin. L.Plant Physiol 61: 451-459

30. PONT LEZICA R, PA ROMERO, MA DANDERT 1976 Membrane-bound UDP-glucose:lipid glucosyltransferases from peas. Plant Physiol 58: 675-680

31. PRESPER KA. EC HEATH 1985 Assembly. transfer and processing of carho-hydrate side chains of glycoproteins. In The Enzymology of Post-transla-tional Modification of Proteins, Vol 2. Academic Press, London, pp 53-93

32. SCHEKMAN R 1985 Protein localization and membrane traffic in yeast. AnnuRev Cell Biol 1: 115- 143

33. SNIDER MD, LA SULTZMAN, PW ROBBINS 198(1 Transmembranc location ofoligosaccharide-lipid synthesis in microsomal vesicles. Cell 21: 385-392

:N]