Synthesis of (1 -* 3), (1 4)-,8-D-glucan in the Golgi …l 3)-D-Glc oligosaccharides were synthesized in proportions similar to those foundin purified MG.Activated charcoal addedduring

Proc. Natl. Acad. Sci. USAVol. 90, pp. 3850-3854, May 1993Plant Biology

Synthesis of (1 -* 3), (1 -> 4)-,8-D-glucan in the Golgi apparatus ofmaize coleoptiles

(cell wail polysaccharides/in vitro synthesis/Zea mays/plant development)

DAVID M. GIBEAUT AND NICHOLAS C. CARPITA*Department of Botany and Plant Pathology, Purdue University, West Lafayette, IN 47907-1155

Communicated by Hans Kende, January 19, 1993

ABSTRACT Membranes of the Golgi apparatus frommaize (Zea mays L.) were used to synthesize in vitro the (1 ->

3), (1 -3 4)-fi-D-glucan (MG) that is unique to the cell wall ofthe Poaceae. The MG was about 250 kDa and was separatedfrom a much larger (1 -* 3)-fi-D-glucan (caDose) by gel-permeation chromatography. Diagnostic oligosaccharides, re-leased by a sequence-dependent endoglucanase from Bacillussubtilis, were separated by HPLC and GLC. The trisaccharidefi-D-Glcp-(I -- 4)-fi-D-Glcp-(1 -- 3)-D-Glc, the tetrasaccharide[13-D-Glcp-(1 4)12-j3-D-Glcp-(1 -l 3)-D-Glc, and longer cel-lodextrin-(l 3)-D-Glc oligosaccharides were synthesized inproportions similar to those found in purified MG. Activatedcharcoal added during homogenization enhanced synthesis ofMG, presumably by removing inhibitory compounds. TheGolgi apparatus was determined as the site of synthesis by acombination of downward and flotation centrifugations onsucrose step gradients. The rate of synthesis did not reachsaturation at up to 10 mM UDP-Glc. Chelators completelyabolished synthesis, but synthase activity was restored byaddition of either MgCl2 or, to a lesser extent, MnCl2. Synthesiscontinued for well over 1 h; addition of KOH to raise the pHfrom 7.2 to 8.0 during the reaction increased the rate ofsynthesis, which indicates that a transmembrane pH gradientmay facilitate synthesis of MG.

(1 -- 3), (1 -* 4)-,S-D-Glucans (MGs) are cell wall polysac-charides that are made at specific developmental stages ofplants in the Poaceae family. MGs are abundant in the wallsofthe endosperm cells in grass caryopses and can account for>70% of the total cell wall material (1). It is thought that thisglucan is a causal agent in lowering serum cholesterol inhumans (2). In growing maize seedlings, MG increases inabundance specifically during cell expansion of coleoptilesand leaves (3, 4) and only decreases in amount after growthceases (5). While MG accumulates during seedling develop-ment in maize, radioactivity incorporated into the glucan invivo turns over rapidly, constantly returning glucose to thenucleotide-sugar pool (4). These observations of the metab-olism of MG during cell expansion indicate that this poly-saccharide may play an important role in the vegetativegrowth of grasses.The grass MGs are homopolymers of glucose and differ

from other mixed-linkage glucans, such as lichenin, by theirprecise unit structure. The unit structure became evident bydigestion ofMG with (1 -+ 3), (1 -* 4)-p-D-glucan endo-(1 -*4)-/3-D-glucanase (E.C. 3.2.1.73) purified from Bacillus sub-tilis. This sequence-dependent glucanase cleaves a (1 -k4)-,3-D-glucosyl linkage only if preceded by a (1 3)-pD-glucosyl linkage (6). Cleavage of water-soluble barley MG bythis enzyme results primarily in the trisaccharide P-D-Glcp-(l-- 4)-/-D-Glcp-(1 -k 3)-D-Glc (G4G3G) and the tetrasaccha-

ride [,3-D-Glcp-(1 -* 4)h2-P-D-Glcp-(1 -+ 3)-D-Glc (G4G4G3G),indicating that the polymer is composed of cellotriosyl andcellotetraosyl units connected by single (1 -- 3)-linked glu-cosyl units (7, 8). The remaining undigested material iscomposed of longer stretches of cellodextrin-(l -3 3)-D-Glc.

In our initial attempts to assay for the synthesis ofMG withmembrane preparations from maize coleoptiles, we observedtwo labeled compounds that migrated close to G4G3G andG4G4G3G in thin-layer chromatographs. These may havebeen mistaken for glucan products when glucan synthase wasused as a marker for Golgi apparatus (9). Like glucanoligomers, these unknown compounds were water soluble,but only slightly soluble in cold 80% (vol/vol) ethanol.Because of the uncertainty resulting from formation of theseunknown products and the predominant synthesis of callose,we concurrently developed the analytical tools necessary toidentify not only the linkage structure of the product but alsothe oligosaccharide unit structure. Chromatographic separa-tion and detection of radioactivity in the oligomers releasedby the endoglucanase provided us with the diagnostic toolsfor unequivocal identification of the products of synthesis invitro by isolated Golgi apparatus from labeled nucleotide-sugar substrates.

Prior to the experiments reported here, the analytical toolswe developed revealed that very little MG was made com-pared to the water-soluble, but only slightly ethanol-soluble,glucosides. Because we suspected that these products mightbe glucosylated flavonoids, we added activated charcoal tothe homogenization medium. We found that the charcoal notonly greatly reduced formation of the water-soluble gluco-sides but also resulted in a marked enhancement of synthaseactivity, which finally allowed definition of some reactionrequirements for the synthesis in vitro of MG.

MATERIALS AND METHODSPlant Material and Preparation of Membranes. Coleoptiles

were excised from etiolated 3-day-old maize seedlings (Zeamays L., FR 1141 x FR 33; Illinois Foundation Seeds Inc.)grown at 29°C. Homogenization, centrifugation, and sucrosegradient centrifugation were performed as described (9). Forremoval of flavonoids or other phenolic compounds, 4 g of16-mesh activated charcoal (Sigma; 16-40 mesh) was sprin-kled on 40 g of coleoptiles before grinding. Some of the finecharcoal was removed by low-speed centrifugation. Theremaining very fine charcoal pelleted in the ensuing stepgradient, but the interface where endoplasmic reticulum andGolgi apparatus were enriched was devoid of charcoal.

Radioactive Cell Walls. For use as standards, radioactivecell walls were obtained by placing the cut ends of about 50

Abbreviations: GPC, gas proportional counter; HPAE, high pHanion-exchange; MG, (1 -- 3), (1 -- 4)-/-D-glucan; G4G3G, 3-D-Glcp-(1 -* 4)-P3D-Glcp-(1 -. 3)-D-Glc; G4G4G3G, [3-D-Glcp-(l -4)]1-3-D-Glcp-(l -- 3)-D-Glc; EIMS, electron impact MS.*To whom reprint requests should be addressed.

3850

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Proc. Natl. Acad. Sci. USA 90 (1993) 3851

coleoptiles in 5 ,uCi (1 Ci = 37 G) of D-[U-14C]Glc for 6 h atlaboratory conditions of light and temperature. The coleop-tiles were rinsed with water and extracted extensively withhot 80% ethanol. The walls were purified as described (4).Glucan Synthase Reactions. Reactions were performed at

25°C with freshly isolated membranes (50-100 ,g of protein).Reactions mixtures contained 0.5 ,uCi of UDP-D-[U-14C]Glc,20 mM KCl, 1.08 M sucrose, and 10 mM Hepes/1,3-bis[tris(hydroxymethyl)methylamino]propane (pH 7.6) in afinal volume of 1 ml. UDP-Glc, MnCl2, CaC12, and MgCl2were used as indicated. Reactions were stopped by additionof ethanol to a concentration of 80% and heating to 105°C inan oven for 15 min. The vials were cooled to room temper-ature and then centrifuged for 15 min at 3000 x g. The pelletwas washed three more times with 80% ethanol after heating,cooling, and centrifuging. This treatment removed all resid-ual UDP-Glc substrate and labeled glucosides.

Digestion of Reaction Products with B. subtilis Endogluca-nase as a Specific Assay for MG. The washed and driedproducts of the glucan synthase reactions were finely dis-persed in water and digested with a B. subtilis endoglucanasepreparation (4). Digestions were overnight on a shaker atroom temperature and terminated by addition of ethanol to80% and heating to 105°C in an oven. The mixtures werecooled to room temperature and centrifuged, and a portion ofthe supernatant was assayed by liquid scintillation spectros-copy. High pH anion-exchange (HPAE) HPLC was used toverify that the only radioactivity released by the enzyme wasassociated with the discrete oligosaccharides ofMG. Greaterthan 90% of the remaining insoluble radioactivity was in (1 --

3)-linked Glc and was reported as callose.HPAE-HPLC. The digestion products were separated with

a Carbo-Pac PAl anion-exchange column and a DionexBioLC and detected with a pulsed amperometric detector(Dionex). The two elution solutions were 0.5 M NaOH(solution A) and 0.5 M NaOH/0.5 M NaOAc (solution B).The flow rate was 1.0 ml/min, and the column was equili-brated in solution A for at least 10 min. Samples in water wereinjected through a 100-,l injection loop, and the followingchanges in the gradient (only the percentage of solution B isshown) were linear between the times shown in min: t = 0,0%; t = 21, 12%; t = 31, 35%; t = 41, 75%; t = 44, 100%; t= 64, 100%. Samples were sometimes spiked with oligosac-charides digested from MG of barley to verify mass andradioactivity retention times. In-line detection of radioactiv-ity in the effluent was with a Beckman model 171 radioisotopedetector with high-resolution output plumbed as described(10). The effluent was mixed with scintillation solution at aflow rate of 3.0 ml/min. The scintillation solution was amixture of 3.75 liters of Ecolume (ICN), 1.30 liters of2-propanol, and 0.09 liter of glacial acetic acid. The signalsfrom both pulsed amperometric and radioactivity detectorswere processed with a Beckman model 406 analog to digitalconverter and Beckman SYSTEM GOLD software.

Gel-Permeation Chromatography. All of the endoplasmicreticulum-Golgi apparatus fraction obtained from 40 g ofcoleoptiles was collected in a volume of 8 ml. The reactionvolume of 16 ml contained 0.250 mM UDP-Glc, 25 ,uCi ofUDP-D-[U-14C]Glc, 20 mM KCI, 10 mM MgC92, 1.08 Msucrose, and 10mM Hepes/1,3-bis[tris(hydroxymethyl)meth-ylamino]propane (pH 7.6). The washing procedures were thesame as described above. Hot water-soluble polysaccharideswere obtained in 4.0 ml of 50 mM citric acid/Na2HPO4, pH7.4, containing 200 ,g of MG from barley. The supernatantwas applied to a 2.5-cm x 60-cm column of Sepharose 4B(Sigma) equilibrated in the same buffer.Linkage Analysis of the Reaction Products. Polysaccharides

were partly methylated with n-butyllithium and methyl iodideand then hydrolyzed in trifluoroacetic acid. Monosaccha-rides were reduced with NaB2H4 and acetylated as described

(4). These partly methylated alditol acetates were dissolvedin methanol/water (40:60, vol/vol) and partitioned againstCCL4 to remove phthalate esters and other impurities asdescribed (11). A Hewlett Packard 5840A gas chromatographwas connected to a Flo-One\Beta G Series GPC (RadiomaticInstruments and Chemical, Tampa, FL) as described (10).Partly methylated alditol acetates were separated by GLC ona 0.75-mm x 30-m wide-bore glass capillary column ofSP-2330 (Supelco) with the effluent split 10% to the flameionization detector and 90% to the gas proportional counter(GPC). Retention times of derivatives were determined fromsamples verified by GLC electron impact MS (EIMS) on a30-m x 0.25-mm i.d. fused-silica capillary column of SP-2330(12).GLC-GPC of Oligosaccharides. Oligosaccharides were re-

duced overnight at 4°C in 1.0 ml of 0.1 M NH40H containingNaB2H4 at 20 mg/ml. The samples were dried at 30°C with astream of nitrogen before decomposing the NaB2H4 withglacial acetic acid. The resulting borate was coevaporatedseveral times with methanol. The oligosaccharide alditolswere methylated twice as described above to improve yield.These partly methylated oligosaccharide alditols were sepa-rated on a 2-m x 0.75-mm i.d. glass capillary column ofSPB-1, 1.0-,um-thick film (Supelco). Temperature was pro-grammed from 60°C to 160°C at 5°C/min, then at 20°C/min to260°C, and finally at 25°C/min to 320°C with a 20-min hold atthe upper temperature. The effluent was split to a flameionization detector and GPC in the same proportions asdescribed above. The injector port, flame ionization detector,and transfer line were held at 350°C. The flow rate of heliumthrough the column was 60 ml/min.GLC-EIMS of Oligosaccharides. The partly methylated

oligosaccharide alditols were separated on a 2-m x 0.25-mmi.d. fused-silica capillary column of HP-1, 0.33-,um-thick film[Hewlett-Packard (HP)]. Temperature was programmed es-sentially as above, and helium was the carrier gas at 30 kPa.Chromatography was performed with an HP 5890 gas chro-matograph with detection by an HP 5971 mass selectivedetector. Data were collected and analyzed with HP CHEMSTATION software. Ions were detected from m/z 40 to 500 ata rate of one scan per 5 s.

RESULTSCharcoal was added to the grinding medium when isolatingmembranes to remove flavonoids or other phenolics that mayinterfere with the assays for glucan synthesis. In membranesprepared without charcoal, HPAE-HPLC of endoglucanasedigests revealed that most of the radioactivity was associatedwith several unknown peaks (data not shown), and only traceamounts of the oligosaccharides of MG were observed.Addition of charcoal also gave an unexpected protection ofthe synthesis of MG, and far less unidentified products wereproduced. The combination of hot ethanol washes, endoglu-canase digest, and HPAE-HPLC resulted in a specific assayfor MG.The macromolecular nature of the MG synthesized in vitro

was demonstrated with gel-permeation chromatography onSepharose 4B (Fig. 1). Digestion of a portion of each columnfraction with the B. subtilis endoglucanase showed that theMG was associated with the first included peak, and themolecular mass, based on dextran standards, was about 250kDa. Linkage analysis (Fig. 2) of the pooled fraction V gavea ratio of 3-linked to 4-linked glucosyl units of -1:3. A sharpvoid peak, which consisted of 3-linked glucosyl units, con-tained about 30% of the radioactivity recovered from thecolumn. Of the total radioactivity incorporated during thereaction incubation, only -10% was solubilized by heating,and the remaining insoluble material consisted of mostly3-linked glucosyl units.

Plant Biology: Gibeaut and Carpita

3852 Plant Biology: Gibeaut and Carpita

"-0

x

E-oC-

.D

00

I..

a)

Cu-oCC

0

0

0)-i0

;cn

iaIF-

Fraction Number

FIG. 1. Gel-permeation chromatography of f-D-glucans synthe-sized in vitro. Water-soluble polysaccharides were applied to a 60-cmx 2.5-cm column of Sepharose 4B. Fractions (4 ml) were collectedand assayed for total radioactivity (o). Radioactivity incorporatedspecifically in MG was measured by counting ethanol-soluble radio-activity released after digestion of an aliquot of each fraction with theB. subtilis endoglucanase (o). For comparison of molecular mass,water-soluble barley MG (Sigma) was chromatographed separately,and fractions were assayed for total sugar (o). Peak positions ofdextran standards (mass in kDa indicated by numbers) were eachchromatographed separately. Radioactive polymers from the MGsynthesis reactions were pooled for further analysis as follows: I,fractions 15-19; II, fractions 20-24; III, fractions 25-28; IV, fractions29-32; V, fractions 33-36; VI, fractions 37-40; VII, fractions 41-45;VIII, fractions 46-52.

Synthesis of long stretches of MG was inferred by com-parison of the endoglucanase digestion products from mate-rials synthesized in vivo and in vitro (Fig. 3). The ratio of thepeak areas ofG4G3G to G4G4G3G was the same for both massand radioactivity, -2.5:1. The glucan products of pooledfraction V from gel-permeation chromatography were di-gested with the endoglucanase, and the HPAE-HPLC chro-matogram of radioactivity was identical to the in vivo prod-

3.0 - 4-GIc A3-GO3 I

a 1.5 -

3.0 - 3-GIc1.5

CZo 30 - 3-Gbc cIt0 4-Gic

100

0 10 20 30 40Retention Time (min)

FIG. 2. GLC-GPC of 8-D-glucans synthesized in vitro. Reactionproducts were derivatized for linkage analysis, and retention times of3- and 4-linked glucosyl units were verified with GLC-EIMS ofderivatives from purified MG. Peaks identified were 1,3,5-tri-O-acetyl-(1-deuterio)-2,4,6-tri-O-methylhexitol (from 3-Glc) and 1,4,5-tri-O-acetyl-(1-deuterio)-2,3,6-tri-O-methylhexitol (from 4-Glc).Later eluting peaks are likely undermethylated derivatives of mostly3-Glc. (A) Pooled fraction V from gel permeation (Fig. 1) showedradioactive (1 -. 3)- and (1 -+ 4)-linked glucosyl units in a ratio of

1:3. (B) Pooled fraction I from the void volume of gel permeation(Fig. 1) showed only radioactive (1 -- 3)-linked glucosyl units.Material that was not water soluble and hence not applied to the gelpermeation column was treated with the endoglucanase. (C) Theremaining insoluble material was composed of at least 90% (13)-linked glucosyl units.

10 20 30Retention Time (min)

FIG. 3. HPAE-HPLC of B. subtilis endoglucanase digests ofpolysaccharides synthesized in vitro and in vivo. The top and middletraces are the radioactivity detector responses from an in-line liquidflow cell radioactivity detector. The bottom trace is the pulsedamperometric detector response of the hydrolysis products from cellwalls of coleoptiles labeled in vivo. These traces show that the ratiosofG4G3G to G4G4G3G are the same for MG synthesized either in vivoor in vitro.

ucts. These data show that the MG synthesized in vitro wasstructurally similar to that deposited in the cell wall and thatentire G4G3G and G4G4G3G units were synthesized in vitro.

Coelution of oligosaccharide standards and radioactivityfrom the endoglucanase digests of pooled fractions III-VIfrom gel-permeation chromatography was demonstrated withGLC-GPC (data not shown). Retention times and fragmen-tation patterns from GLC-EIMS ofthe partly methylated MGoligosaccharide alditols were compared with those deriva-tives of laminaribiose and cellodextrin standards. The pres-ence of m/z 133 and 101 rather than m/z 134 and 102confirmed that the alditol (formerly the reducing end) was3-linked, and the absence of m/z 177 and 159 indicated thatthe hexose units were 4-linked (13). The ratios of the sum ofthe ion fragments from the hexose units (m/z 219, 187, and155) to those from the alditol unit (m/z 296, 236, 204, and 172)gave 2.2 for the first and 2.8 for the second eluting peaks. Wealso observed the penta- and hexasaccharides and verifiedtheir structure. Having established our assay of MG, weexamined requirements for synthesis of the MG in vitro.

Synthesis of MG was specific for UDP-Glc. GDP-Glc andADP-Glc were not substrates (data not shown). Flotationcentrifugation demonstrated that both MG and callose syn-thase activities copurified with Golgi apparatus (Table 1).Although chelators of divalent metal ions abolished thesynthesis of MG, synthesis was restored by the addition ofeither MgCl2 to a concentration of 10 mM or MnCl2 to 1 mM(Table 2), but not by CaCl2 (data not shown). Chelatorsreduced, but did not abolish, the synthesis of callose, andCaCl2 stimulated the synthesis of callose less than 2-foldcompared to about 7-fold in a plasma membrane-enrichedfraction (data not shown). The rate of synthesis of MGincreased linearly from pH 6.4 to 8.0, whereas callose syn-thesis was highest between pH 7.0 and 7.6 (data not shown).Synthesis of both MG and callose was not saturated up to 10mM UDP-Glc in the presence of either MgCl2 or MnCl2(Table 2). Synthesis of both MG and callose continued wellover 1 h, but the rate of MG synthesis began to declinethereafter (Fig. 4). Addition of KOH, but not KCI, to inducea pH jump renewed the synthesis ofMG (Fig. 4) but had noeffect on the synthesis of callose (data not shown).

DISCUSSIONOur early efforts to detect the synthesis ofMG were hinderedby synthesis of unknown glucosides, perhaps flavonoids,

Proc. Natl. Acad. Sci. USA 90 (1993)

Proc. Natl. Acad. Sci. USA 90 (1993) 3853

Table 1. Location of glucan synthase activities

InosineMG, Callose, diphosphatase, Approximate

Gradient dpm/mg dpm/mg ,umol Pi per h density,fraction protein protein per mg protein g/mlCrudeTP 278 1,233 1.2 1.05-1.07SM 876 6,052 20.1 1.08-1.11ER-GA 1286 26,916 25.8 1.11-1.16MI-PM 962 13,723 8.3 1.16-1.24

Purified*ER 967 (0.7) 24,272 (1.5) 32.1 (1.2) 1.08-1.14GA 1528 (1.0) 40,042 (2.5) 37.1 (1.4) 1.15-1.17PM 520(0.4) 6,082 (0.4) 15.6 (0.6) 1.21-1.26

Crude membranes were obtained with downward centrifugationinto a discontinuous gradient of 12%, 19%o, 27%, 37%, and 51%(wt/vol) sucrose; crude fractions of the endoplasmic reticulum andGolgi apparatus (27-37% interface) were mixed with sucrose to 47%and purified by flotation centrifugation in a discontinuous gradient oflO%, 20%b, 34%, and 40%7s sucrose (9). Membrane-enriched fractionswere identified with marker enzymes: tonoplast (TP) with nitrate-sensitive ATPase (14), smooth membranes (SM) with antimycinA-insensitive NADH-cytochrome c reductase (15), endoplasmicreticulum-Golgi apparatus (ER-GA) with antimycin A-insensitiveNADH-cytochrome c reductase (15) and total inosine diphosphatase(16), mitochondria (MI) with cytochrome c oxidase (17), and plasmamembrane (PM) with naphthylphthalamic acid binding (18) andvanadate-sensitive ATPase (19). Each assay was performed asdescribed (9). For MG synthesis reactions, MgCl2 was added to 10mM after the addition of 5 mM EDTA/EGTA. Reactions wereperformed for 30 min with the concentration of UDP-Glc at 0.5 mM.Values in parentheses are purification factors relative to the crudeendoplasmic reticulum-Golgi apparatus.*After flotation of endoplasmic reticulum-Golgi apparatus.

which are readily glucosylated with UDP-Glc as the glucosedonor (20). Flavonoids are water soluble when glycosylatedand only slightly soluble in cold ethanol (21). The unknownswere simple glucosides because only glucose and the originalcompound were detected after limited hydrolysis with glu-cosidase (data not shown). Because the glucosides are onlyslightly soluble in cold ethanol, they behave like glucanproducts and may be mistaken for polysaccharide in manyreports in which glucan synthase was used as a marker forGolgi and ethanol insolubility was the sole criterion. Synthe-sis of unknown compounds was greatly reduced when char-

Table 2. Divalent cation and substrate concentration effectsMG, nmol Glc Callose, nmol Glcper mg protein per mg protein

Addition, mM Mg Mn Mg Mn

Divalent metal0.001 0.56 0.57 884 10230.008 1.50 1.57 889 9520.040 4.73 6.04 936 5950.240 13.6 12.5 898 3361.680 24.6 13.0 821 191

10.00 28.8 7.20 450 102UDP-Glc

0.002 0.12 0.03 2.28 1.740.010 0.60 0.21 10.6 8.010.042 2.23 1.04 47.2 48.10.242 8.42 3.46 161 1251.682 11.9 8.37 645 482

10.00 18.5 11.4 863 821

Crude endoplasmic reticulum-Golgi apparatus fractions were usedfor these analyses. For the cation effects, UDP-Glc was used at 0.5mM. For the substrate concentration effects, Mg was used at 5 mMand Mn was used at 1 mM.

, 1.6

x 1.2

i' 0.8

c 0.4

a: nn0 20 40 60

Minutes80 100

FIG. 4. Time course and the effect of a pH boost on theincorporation ofglucose from UDP-Glc into MG. Standard reactionswere performed with 5 mM MgC92. A pH boost to pH 8.0 increasedthe extent but not initial rate of the synthesis of MG. The synthesisof callose was unaffected by the pH jump (not shown).

coal was added to the homogenization medium, and, moreimportantly, the synthesis ofMG was made possible.

In the specific case of the synthesis of MG, the productmust be differentiated from callose and 4-linked glucose. Notonly must 3- and 4-linked glucosyl units be demonstrated, butthe units must also be shown to be in the same polymer.Based on detection of 3- and 4-linked glucosyl units, Robin-son and Glas (22) reported the synthesis of MG in mi-crosomes obtained from cells that lack this glucan (23, 24).Callose was undoubtedly formed, but a small amount of (1 -+4)-linked glucans may have arisen from another synthaseactivity related to the synthesis of xyloglucan. Other reportsof the synthesis of MG with cellular membranes are moreconvincing (25-28), but the relative amounts of callose, (1 -+4)-(-D-glucan, and MGs reported in these studies are verydifferent from our data. Although some of the differencesmay be attributed to the different tissues used and assay withtotal membranes rather than fractionated membranes, theother analyses relied upon differential hydrolysis with threeglucan hydrolases (25-28). Linkage analysis of products andchromatography of hydrolysis products were not effectivelyused to differentiate between the synthesis of callose, (1 -*4)-3-D-glucan, and MG (25-28). Our use of a combination ofgel-permeation chromatography, linkage analysis, enzymicdigestion, HPAE-HPLC, GLC-GPC, and GLC-EIMS con-firmed that in vitro synthesized macromolecular MG is iden-tical to the cell wall polysaccharide.We observed considerable synthesis ofcallose by the Golgi

apparatus that could not be attributed solely to cross-contamination with plasma membrane. This activity wasstimulated at most 2-fold by CaC12 as compared to 7-fold inthe plasma membrane (9). Although it is well documentedthat callose is synthesized in vivo at the plasma membrane,the synthesis of callose at the Golgi apparatus has never beenruled out. At least 90% of the glucan synthesized in ourpreparations of Golgi apparatus was callose. Also, the spe-cific activity for the synthesis of callose increased whenendoplasmic reticulum was separated from the Golgi appa-ratus by flotation centrifugation, but the specific activity forthe synthesis ofMG was about the same, most likely becauseof a relative loss of the synthetic activity for MG.

Is the MG synthase sensitive to disruption in a fashionsimilar to the cellulose and callose synthase of the plasmamembrane? Some synthase activity can be maintained in thepresence of detergents (26), but attempts to purify polypep-tides from mixed membranes have led to the recovery ofonlycallose synthase after detergent solubilization (27). Perhaps,like "wounded" cellulose synthase reverting to callose syn-thase (29), MG synthase may revert to callose synthase whendisrupted. We cannot rule out the possibility that Golgi-

+KOH

additions

+KCI

control

s . I . I . I . a

Plant Biology: Gibeaut and Carpita

3854 Plant Biology: Gibeaut and Carpita

associated callose synthase activity is from cellulose syn-thase en route to the plasma membrane.

Synthesis of MG and callose was not saturated by 10 mMUDP-Glc (Table 2) and did not display first-order kinetics(data not shown). Substrate transport, activation by un-known effectors, localized pH, or transmembrane pH gradi-ents could each influence glucan synthesis. An increase inMG synthesis with increasing pH to above 8 and boosts ofactivity by pH jumps also indicate that a pH gradient mayenhance synthesis (Fig. 4). We found also that the ionophoresmonensin and nigericin inhibited MG synthesis (data notshown), indicating that a pH gradient, which is postulated tobe important for Golgi vesicle swelling and exocytosis (30,31), may be required for synthesis as well. Membrane elec-trical potentials are thought to be important for cellulosesynthesis in plants and Acetobacter (32, 33). The rate ofsynthesis of ,B-glucans could be stimulated severalfold byartificial potentials across membranes induced by a combi-nation of valinomycin and potassium ions (34). Although theplasma membrane- and Golgi-associated synthases appar-ently have different requirements, electrical and pH gradientsmay be universally important in membrane-associated syn-thesis of cell wall polysaccharides.

Activities for several other cell wall polysaccharide syn-thases, such as those for xyloglucan, galactomannan, andglucuronoxylans, have been documented (for review, seerefs. 35 and 36). All synthase activities reported to date havebeen short-lived and have low apparent Km values (10-30FM). When maize membranes are prepared under conditionswhere putative flavonoid contaminants are removed, we findthat the synthesis of MG is not saturated at even 10 mMUDP-Glc and that synthesis can be linear for over 2 h with aboost in pH.

This work was supported by Grant DE-FG02-88ER13903 from theU.S. Department of Energy. This is paper no. 13,564 of the PurdueUniversity Agriculture Experiment Station.

1. Fincher, G. B. & Stone, B. A. (1986) Adv. Cereal Sci. Technol.8, 207-295.

2. Anderson, J. W., Spencer, D. B., Hamilton, C. C., Smith,S. F., Tietyen, J., Bryant, C. A. & Oeltgen, P. (1990) Am. J.Clin. Nutr. 52, 495-499.

3. Carpita, N. C. (1984) Plant Physiol. 76, 205-212.4. Gibeaut, D. M. & Carpita, N. C. (1991) Plant Physiol. 97,

551-561.5. Carpita, N. C. & Kanabus, J. (1988) Plant Physiol. 88, 671-678.6. Anderson, M. A. & Stone, B. A. (1975) FEBS Lett. 52, 202-

207.

7. Staudte, R. G., Woodward, J. R., Fincher, G. B. & Stone,B. A. (1983) Carbohydr. Polym. 3, 299-312.

8. Woodward, J. R., Fincher, G. B. & Stone, B. A. (1983) Car-bohydr. Polym. 3, 207-225.

9. Gibeaut, D. M. & Carpita, N. C. (1990) Protoplasma 156,82-93.

10. Bancal, P., Gibeaut, D. M. & Carpita, N. C. (1993) in Scienceand Technology of Fructans, eds. Suzuki, M. & Chatterton,N. J. (CRC, Boca Raton, FL), pp. 81-116.

11. Gibeaut, D. M. & Carpita, N. C. (1991) J. Chromatogr. 587,284-287.

12. Carpita, N. C. & Shea, E. M. (1989) in Analysis of Carbohy-drates by GLC and MS, eds. Biermann, C. J. & McGinnis,G. D. (CRC, Boca Raton, FL), pp. 157-216.

13. Kfirkkaiinen, J. (1971) Carbohydr. Res. 17, 11-18.14. Mandala, S. & Taiz, L. (1985) Plant Physiol. 78, 104-109.15. Ardies, C. M., Lasker, J. M., Bloswick, B. P. & Lieber, C. S.

(1987) Anal. Biochem. 162, 39-46.16. Wienecke, K., Glas, R. & Robinson, D. G. (1982) Planta 155,

58-63.17. Bomhoff, G. H. & Spencer, M. (1977) Can. J. Biochem. 55,

1114-1117.18. Brummer, B. & Parish, R. W. (1983) Planta 157, 446-453.19. Gallagher, S. R. & Leonard, R. T. (1982) Plant Physiol. 70,

1335-1340.20. Hahlbrock, K. & Scheel, D. (1989) Annu. Rev. Plant Physiol.

Plant Mol. Biol. 40, 347-369.21. Markham, K. R. (1982) Techniques ofFlavonoid Identification

(Academic, New York).22. Robinson, D. G. & Glas, R. (1983) J. Exp. Bot. 34, 668-675.23. Burke, D., Kaufman, P., McNeil, M. & Albersheim, P. (1974)

Plant Physiol. 54, 109-115.24. Carpita, N. C., Mulligan, J. A. & Heyser, J. W. (1985) Plant

Physiol. 79, 480-484.25. Henry, R. J. & Stone, B. A. (1982) Plant Physiol. 69, 632-636.26. Henry, R. J. & Stone, B. A. (1982) Biochem. J. 203, 629-636.27. Meikle, P. J., Ng, K. F., Johnson, E., Hoogenraad, N. J. &

Stone, B. A. (1991) J. Biol. Chem. 266, 22569-22581.28. Smith, M. M. & Stone, B. A. (1973) Biochim. Biophys. Acta

313, 72-94.29. Delmer, D. P. (1987) Annu. Rev. Plant Physiol. 38, 259-290.30. Griffing, L. R. & Ray, P. M. (1985) Eur. J. Cell Biol. 36, 24-31.31. Ledger, P. W. & Tanzer, M. L. (1984) Trends Biochem. Sci. 9,

313-314.32. Carpita, N. C. & Delmer, D. P. (1980) Plant Physiol. 66,

911-916.33. Delmer, D. P., Benziman, M. & Padan, E. (1982) Proc. Natl.

Acad. Sci. USA 79, 5282-5286.34. Bacic, A. & Delmer, D. P. (1981) Planta 152, 346-351.35. Delmer, D. P. & Stone, B. A. (1988) in The Biochemistry of

Plants, ed. Preiss, J. (Academic, New York), Vol. 14, pp.373-420.

36. Fincher, G. B. & Stone, B. A. (1981) in Encyclopaedia ofPlantPhysiology, eds. Tanner, W. & Loewus, F. A. (Springer,Berlin), Vol. 13B, pp. 68-132.

Proc. Natl. Acad. Sci. USA 90 (1993)