Plant Cell Physiol.42(3): 292–300 (2001)JSPP © 2001

da

etin

all,

tedm-ll.ativeanndviarseddndhere-ell,

re

f abyle-the

nies

her,ig-6,

ilyndatic

e,theis-ll

Endoxyloglucan Transferase is Localized both in the Cell Plate and in theSecretory Pathway Destined for the Apoplast in Tobacco Cells

Ryusuke Yokoyama and Kazuhiko Nishitani1

Biological Institute, Graduate School of Science, Tohoku University, Sendai, 980-8578 Japan

;Intracellular trafficking of enzymes responsible for

constructing and modifying the cell wall architecture inplants is mostly unknown. To examine their translocationpathways, we employed an endoxyloglucan transferase(EXGT), a key enzyme responsible for forming and rear-ranging the cellulose/xyloglucan network of the cell wall.We traced its intracellular localization in suspension-cultured cells of tobacco bright yellow-2 by means of greenfluorescent protein-fusion gene procedures as well as byindirect immunofluorescence. During interphase the pro-tein was extensively secreted into the apoplast via the endo-plasmic reticulum–Golgi apparatus network, whereas dur-ing cytokinesis, the protein was exclusively located in thephragmoplast and eventually transported to the cell plate.These results clearly indicate commitment of EXGT pro-tein to the construction of both the cell plate and the cellwall. This study also visualized the process of phragmo-plast development at a level of vesicle translocation in theliving cell.

Key words: Cell plate — Cell wall — Endoxyloglucan trans-ferase — Green fluorescence protein — Secretion — Tobacco.

Abbreviations: ER, endoplasmic reticulum; EXGT, endoxyloglu-can transferase; GFP, green fluorescent protein; LS, Linsmaier andSkoog medium.

Introduction

The cell division in plants is accomplished by formationof a cell plate, which begins at the late anaphase when Golgiapparatus-derived vesicles associated with phragmoplastmicrotubules are delivered to the equatorial plane of the divid-ing cell. These vesicles fuse to each other to form a membranenetwork. The assembly of the membrane network gives rise toa disc-shaped immature cell plate, which enlarges centrifugallyand eventually fuses with the mother cell wall, thereby separat-ing daughter cells (Samuels et al. 1995).

On the other hand, the cell enlargement process, whichoccurs during the interphase of the cell cycle, is achievedthrough architectural changes of the existing cell wall, which iscomposed of crystalline cellulose microfibrils embedded in amatrix of amorphous polysaccharides (Carpita and Gibeaut

1993). Cellulose microfibrils are polymerized and crystallizeby the cellulose-synthesizing complex located at the plasmmembrane (Giddings et al. 1980, Arioli et al. 1998, Kimuraal. 1999), whereas matrix polysaccharides are synthesizedthe Golgi apparatus and secreted into the apoplast or cell wspace via membrane trafficking machinery (Bolwell 1988Moore et al. 1991, Zhang and Staehelin 1992). Newly secrecellulose microfibrils and matrix polysaccharides are assebled and integrated into the framework of the existing cell waThese processes are considered to be mediated by cooperactions of various types of enzymes, such as endoxylogluctransferases (Ito and Nishitani 1999), glucanases (Høj aFincher 1995) and expansin (Cosgrove 2000), all secretedGolgi vesicles. Secretory pathways for the matrix polymeduring cell division and cell enlargement have been explorby biochemical (Kakimoto and Shibaoka 1992) animmunofluorescence methods (Moore et al. 1991, Moore aStaehelin 1988, Northcote et al. 1989, Brummell 1990). On tother hand, despite their importance in plant cells, the section processes of the individual enzymes responsible for cwall construction and modification are still poorly understoodand their regulatory systems remain elusive.

Endoxyloglucan transferase (EXGT) is responsible forearrangement of the cellulose/xyloglucan framework in thcell wall of seed plants. The enzyme catalyzes transfer olarge segment of one xyloglucan molecule and theremediates molecular grafting between the polysaccharide mocules, a molecular process essential for rearrangement ofcell wall architecture (Nishitani and Tominaga 1991, Nishitaand Tominaga 1992, Okazawa et al. 1993). Many homologuof this enzyme have been found and characterized. Togetthese homologues constitute a large multi-gene family desnated as xyloglucan-related proteins (XRP) (Xu et al. 199Nishitani 1997, Yokoyama and Nishitani 2000).

In this study, we used a member of the XRP gene famas a representative of the cell wall construction enzyme, atraced its secretory pathway during a cell cycle. We found ththis EXGT was secreted into the apoplast via the endoplasmreticulum (ER)–Golgi apparatus pathway during interphasand that the same protein was transported exclusively tocell plate, i.e. not to the apoplast, during cytokinesis. Thstudy is the first to visualize the trafficking process of a protein that is involved in both the cell plate formation and the cewall construction in the apoplast.

1 Corresponding author: E-mail, nishitan@mail.cc.tohoku.ac.jp; Fax, +81-22-217-6700.

292

Secretory pathways of endoxyloglucan transferase 293

ic

rese

ol,

ngeti-

iatee

ionped

csa

tsu,ceeg,r acro-n-fo-ro-

FPedas ar.c)p-

oomeseaic

oly

Materials and Methods

GFP-fusion gene constructsWe amplified the coding region ofEXGT-A1 (Okazawa et al.

1993) by PCR with two oligonucleotides, 5�-CGT CTA GAT GACTGT TTC TTC ATC-3� and 5�-CAT GGA TCC TGC GTC TCT GTCCCT-3�. The primers introducedXbaI and BamHI restriction sites atthe 5� and 3� end of the coding region, respectively (underlined). Theamplified DNA fragment was digested withXbaI and BamHI andinserted into theXbaI/BamHI sites of pBluescript II SK+ (Stratagene,La Jolla, CA, U.S.A.) to generate the pBluescript-EXGT-A1. ABamHI-EcoRI fragment containing the sequence for the plant adaptedgreen fluorescence protein (sGFP) and the nopaline synthase (Tnos)(Chiu et al. 1996) was inserted into theBamHI/EcoRI site in thepBluescript-EXGT-A1 to generate an EXGT-GFP fusion gene con-struct. A DNA fragment containing the EXGT-GFP fusion gene andTnos terminator was finally cloned into theXbaI/EcoRI restriction siteof pBI121 (Jefferson et al. 1987) to generate pBI-35S-EX-GFP, whichcontained a cauliflower mosaic virus 35S promoter, a coding sequencefor the EXGT-GFP and the Tnos terminator. To construct a codingsequence for a GFP fusion protein of the signal peptide of EXGT-A1and GFP, we amplified a DNA fragment containing the cauliflowermosaic virus 35S promoter and the signal sequence using an oligonu-cleotide 5�-CAA TGG ATC CGC GTG GAG GAA TAG CCA-3� as aprimer and the pBI-35S-EX-GFP as the template. The amplified frag-ment was digested withHindIII and BamHI and cloned into theHindIIIand BamHI restriction sites of pBI-35S-XE-GFP. The full lengthsequence encoding the EXGT-A1 protein in the plasmid was replacedwith a short sequence encoding a signal peptide consisting of 24 aminoacid residues and 4 amino acid residues derived from the two restric-tion enzyme sites, to generate pBI-35S-SP-GFP. The pBI 35S-GFP,which was used as the control vector, was prepared by replacing theuidA structural gene in the pBI121 with the cDNA encoding the sGFP(Jefferson et al. 1987). Each of the three plasmids was transformedinto Agrobacterium tumefaciensLBA4404 by using the freeze-thawmethod. Four-day-old suspension-cultured cells of tobacco BY-2 weretransformed by coculturing withA. tumefaciensLBA4404 harboringeach of the three types of plasmids for 2 d at 25�C without shakingaccording to the procedure described by Matsuoka and Nakamura(1991). Transformed cells were selected on a Linsmaier and Skoog(LS)-agar medium containing 100�g ml–1 kanamycin and 500�g ml–1

carbenicillin. Transgenic calli, which usually emerged 2 to 3 weeksafter the inoculation, were transferred to new LS-agar medium con-taining the two antibiotics. Calli grown on the selection medium weretransferred to a liquid LS medium containing the two antibiotics andincubated on a rotary shaker. Transformed cell lines expressing GFPand EXGTsp-GFP were subcultured weekly, whereas those expressingEXGT-GFP were subcultured at longer intervals due to their slowerproliferation.

Synchronization of tobacco BY-2 cellsSuspension cultured tobacco BY-2 cells were maintained in mod-

ified LS-medium (Nagata et al. 1981) and synchronized essentially asdescribed by Nagata et al. (1992). A 7-day-old cell culture was diluted2 : 55 and incubated in the presence of 5�g ml–1 aphidicolin for 24 h,followed by release from the inhibitor by washing with 1 liter of freshmedium by filtration. After the release from aphidicolin, cells synchro-nized at each stage of the cell cycle were subjected to analysis byimmunofluorescence procedures and GFP-fusion procedures.

Immunofluorescence localizationTobacco BY-2 cells were fixed in 3.7% formaldehyde in PMEG

buffer solution consisting of 100 mM 1,4-piperazinediethanesulfonacid (PIPES), 2 mM magnesium sulfate, 10 mM (ethylene glycol-O,-O�bis(2-amino-ethyl)-N,N,N�,N�-tetraacetic acid (EGTA) and 2% glyc-erol at pH 6.8 for 12 h at 4�C. After being rinsed with the PMEGbuffer, fixed cells were transferred to a poly-L-lysine-coated slideglass, followed by air-drying for 30 min. Cells on the slide glass wedigested in a solution containing 1% Cellulase Y-C and 1% PectolyaY23 in the PMEG buffer for 5 min followed by treatment with aphosphate-buffered saline containing 1% Nonidet, 0.4 M mannit5 mM EGTA, 0.5 mM phenylmethylsulfonyl fluoride, and 1�g ml–1

leupeptin. Immunofluorescence microscopy was performed followia standard method using affinity-purified anti-Vigna EXGT IgG as thfirst antibody and fluorescein isothiocyanate (FITC)-conjugated anrabbit goat IgG as the second antibody. DNA was stained with 4�,6-diamidino-2-phenylindole (DAPI) at a concentration of 10�g ml–1 for5 min at room temperature. Photographs were taken using approprfilters for FITC and DAPI fitted to an epifluorescence microscop(DMRXP; Leica, Heerbrugg, Switzerland).

Microscopic observationThe fluorescence images of tobacco cells expressing GFP fus

proteins were observed under an epifluorescence microscope equipwith a plant GFP filter set and differential interference contrast opti(DMRXP; Leica). The microscopic images were recorded using3CCD video camera (C5810; Hamamatsu Photonics, HamamaJapan) using MacAspect software. Images of immunofluorescenwere recorded using appropriate filters for FITC and DAPI fitted to thepifluorescence microscope (DMRXP; Leica). For optical sectionintransgenic tobacco cells living in LS medium were mounted undeglass cover slip and observed under a laser scanning confocal miscope (Fluoview; Olympus, Tokyo, Japan) equipped with a kryptoargon laser and filter sets suitable for detection of fluorescein. Concal images were acquired and composed using the Fluoview micscope and Adobe Photoshop software.

Fig. 1 Schematic presentation of three gene constructs. (a) Gintroduced via the pBI-35S-GFP vector. This construct was designto express the plant adapted GFP (sGFP(S65T)) and was usedcontrol. (b) EXGT-GFP introduced via the pBI-35S-EX-GFP vectoThe full length EXGT-A1 protein is fused to the plant adapted GFP. (EXGTsp-GFP introduced via pBI-35S-SP-GFP. A putative signal petide of the EXGT-A1 (italicized) consisting of the N-terminal 24 aminacid residues and the 4 amino acid residues (roman) derived frrestriction enzyme sites are fused to the plant-adapted GFP. Thgene constructs are placed under the control of the cauliflower mosvirus 35S promoter (CaMV35S) with a nopaline synthase gene p(A) signal (Tnos).

Secretory pathways of endoxyloglucan transferase294

n-of

theP)nsi-Pwn-hllyP

–d) andXGTsp-the same

epifluo-e (d, i), and

Results

Expression of GFP fusion reporter in tobacco BY-2 cellsTo investigate the secretory pathway of apoplastic

enzymes involved in the cell wall construction in living plantcells, we prepared two types of GFP fusion reporter constructs,EXGT-GFP and EXGTsp-GFP, as shown in Fig. 1. In EXGT-GFP, the full-length EXGT-A1 polypeptide was fused to aplant-adapted GFP (sGFP (S65T)) (Niwa et al. 1999), while inEXGTsp-GFP, the first 24-amino-acid sequence of the EXGT-A1 was fused to the sGFP (S65T). To ensure high levels ofgene expression, the two fusion genes were driven by the cauli-

flower mosaic virus 35S promoter (CaMV35S). These costructs were introduced into suspension-cultured cellstobacco BY-2, and seven independent cell lines for each oftwo GFP fusion constructs and the control construct (GFwere selected and assayed for their green fluorescence inteties and growth rates. In cell lines expressing the EXGT-GFfusion protein, the fluorescence intensity was found to be loand the growth rate was severely reduced (Fig. 2c, g). By cotrast, the cell lines expressing EXGTsp-GFP exhibited higintensities of the green fluorescence and proliferated norma(Fig. 2d, h). Thus, we employed one of the EXGTsp-GFexpressing cell lines in further experiments.

Fig. 2 Comparison of GFP fluorescence and cell shape among four different tobacco cell lines. Interference contrast (DIC) images (aepifluorescence images (e–h) of individual cells of untransformed BY-2 line (a, e), cells expressing GFP (b, f), EXGT-GFP (c, g) and EGFP (d, h). Both the DIC and fluorescence images of the same cells for each cell line were recorded by a 3CCD video camera underconditions.

Fig. 3 Intracellular localization of the EXGTsp-GFP at different phases during the mitosis. Interference contrast (DIC) images (a–e) andrescence images (f–j) of a single cell expressing EXGTsp-GFP at interphase (a, f), prophase (b, g), metaphase (c, h), anaphase/telophastelophase (e, j).

Secretory pathways of endoxyloglucan transferase 295

msisti-theig.rly

natrn

arlat

sidetheas

sa–-gu--edr

nd theand an

e. (e, j

Using a synchronized culture (Nagata et al. 1992) of theEXGTsp-GFP-expressing tobacco cells, we studied the intracel-lular localization of EXGT during mitosis. Figure 3 shows dif-ferential interference contrast images as well as epifluores-cence images of a cell expressing EXGTsp-GFP at variousstages of the cell cycle. At interphase (Fig. 3a, f), the fluorescentprotein was observed at the surface of the nucleus. The proteinwas also found to exist as strings, radiating towards the periph-ery of the cell. At the beginning of mitosis the EXGT proteinwas concentrated into the central region of the cell and the fluo-rescence intensity of the radiating strings was diminished(Fig. 3b, g). During metaphase, the fluorescence was localizedin the center of the cell (Fig. 3c, h). In the central region of theanaphase cell (Fig. 3d, i), fluorescent fibers were observedalong with the spindle. At this stage, an intense fluorescentband appeared at the equatorial plate. As the cell plate forma-tion progressed, the fluorescent fibers disappeared and a fluo-rescent band expanded centrifugally toward the parental cellwall (Fig. 3e, j).

To examine whether the intracellular accumulation pat-tern of the EXGTsp-GFP actually reflected the real localizationpattern of the native EXGT protein in tobacco cells, weobserved immunolocalization patterns of the native EXGT byusing a polyclonal antibody raised against a recombinant EXGTprotein (Fig. 4). The anti-EXGT antibody strongly labeled the

cytoplasmic radiating strings and ER-like membrane systelocated around the nucleus at the interphase (Fig. 4f). As mitocommenced, the labeling of the radiating strings by the anbody became less intense (Fig. 4g). During metaphase,fluorescence labeling was found around the chromosome (F4h). At telophase, the phragmoplast and cell plate were clealabeled with the antibody (Fig. 4j). Thus, the localizatiopattern of the immunofluorescence exactly coincided with thof EXGTsp-GFP, indicating that the EXGTsp-GFP reporteprotein can properly represent the intracellular localizatiopattern of the native EXGT protein in tobacco cells.

On the basis of this result, we examined the intracellullocalization of EXGTsp-GFP in living cells using a confocalaser scanning microscope. Optical sections of a single cellinterphase revealed that the fluorescence was located outthe nucleus throughout the cytoplasm. The inner surface ofplasma membrane or the cortical region of the cytoplasm wlined with the network-like tubular structure (Fig. 5e). Thistructure was found to exist throughout the cytoplasm (Fig. 5d). In control cells, in which the sGFP protein without the signal peptide was expressed, the green fluorescence was irrelarly or indistinctly distributed inside the cell, and no networklike pattern was observed (Fig. 5f–i). This indicates that thnetwork-like localization pattern of the EXGTsp-GFP reflectethe real function of the signal peptide of EXGT-A1. A simila

Fig. 4 Immunofluorescence localization of EXGT protein in tobacco BY-2 cells at different phases during the mitosis. Nuclear material aEXGT protein in a single cell of the untransformed tobacco BY-2 line at each stage of a cell division were double-stained with DAPI (a–e)indirect immunofluorescence procedure (f–j), respectively. (a, f) interphase. (b, g) prophase. (c, h) metaphase. (d, i) anaphase–telophas) tel-ophase.

Secretory pathways of endoxyloglucan transferase296

smeERede

isn

A-gi

l-thatnd--

g.

tical sec-ote the

network-like pattern has been observed in the cortical region ofcells expressing a green fluorescent fusion protein withsequences ensuring retention in the lumen of ER (Boevink etal. 1998, Boevink et al. 1999, Ridge et al. 1999). The network-like pattern was also disclosed by staining tobacco cells withDiOC�(3), a dye which specifically stains ER (Grabski et al.1993). Taken together, these facts clearly indicate thatEXGTsp-GFP is transported via the network-like ER tubules.

Effects of brefeldin A (BFA) on secretion of EXGTsp-GFPTo confirm the involvement of the ER–Golgi apparatus in

the secretory pathway of EXGTsp-GFP, we incubated the cellsin the presence of 10�g ml–1 BFA, a potent fungal toxin capa-ble of specifically inhibiting the vesicle trafficking via the ER–Golgi apparatus (Driouich et al. 1993, Boevink et al. 1998,Boevink et al. 1999, Chardin and McCormick 1999). Within1 h after application of BFA, the radiating strings diminished

and small vesicles appeared at the periphery of the cytopla(Fig. 6A). An optical section through the cortical region of thBFA-treated cells revealed that the network-shaped tubularhad begun to fuse, giving rise to a large irregularly perforatsheet (Fig. 6B). After prolonged treatment with BFA, thperforated membranous structure became disordered. Itlikely that the disorganization of the ER was caused by aover-accumulation of the green fluorescent protein due to BFinhibited membrane trafficking between the ER and Golapparatus.

Cellular localization of EXGT during cell plate formationOptical sectioning of the equatorial plate of the cell at te

ophase by a confocal laser scanning microscope revealedthe fluorescent reporter protein was accumulated in the expaing cell plate as well as in vesicles in the vicinity of the phragmoplast (Fig. 7). The median longitudinal optical section (Fi

Fig. 5 Comparison of intracellular localization of green fluorescence between cells expressing EXGTsp-GFP and GFP. Consecutive options from the center (a, f) to the periphery (e, i) of individual transgenic cells expressing EXGTsp-GFP(a–e) or GFP (f–i) are shown. Nnetwork-like ER structure throughout the cytoplasm in the cell expressing EXGTsp-GFP (a–e).

Secretory pathways of endoxyloglucan transferase 297

earntrtione,ere

les

cu-.rn

or-larll

the

7a) showed the short strings perpendicular to the cell plate nthe parent cell wall. Bright spots were found to be presearound the periphery of the growing cell plate. The shostrings and spots were clearly observed in an optical sectthrough a peripheral plane of the cell (Fig. 7d). Furthermorthe fluorescent spots to be translocated to the cell plate wobserved (arrow, Fig. 7b).

Discussion

Secretory pathway for EXGT protein during the interphaseBy using the EXGTsp-GFP reporter protein, we were ab

to visualize the pathway through which EXGT protein wa

Fig. 6 Time course effects of BFA on intracellular localization ofEXGTsp-GFP. (A) Fluorescence images of cells which had been inbated in 10�g ml–1 BFA for 0 to 24 h as indicated on individual plates(B) Confocal laser scanning microscope images of the ER netwoklocated at the peripheral region of individual cells which had beetreated with BFA as in (A).Fig. 7 Confocal laser scanning microscopic images of a transgenictobacco cell expressing EXGTsp-GFP at the late stage of cell plate fmation during cytokinesis. Consecutive optical sections perpendicuto the cell division plane from the center of the cell (a) toward the cesurface (d). The arrow in (b) indicates vesicles to be translocated tocell plate.

Secretory pathways of endoxyloglucan transferase298

is

ed.a-uss

forread

ellat

llur

theorer,anll

inall

ternevene

den

es).r-e

da-

er-reotreentfr

n-lgiac-

ofc-

secreted into the apoplast, confirming our previous predictionbased on both the primary structure of EXGT protein (Okazawaet al. 1993), and its presence in the apoplastic fluid (Nishitaniand Tominaga 1992). This finding demonstrates that theEXGTsp-GFP that contained only a 24-amino acid sequence incommon with the EXGT-A1 protein was correctly recognizedas the signal sequence destined for the initial transfer to the ER,and ultimately destined for the apoplast. Since the transcrip-tion of this reporter gene was driven by the powerfulCaMV35S promoter in our experiment, we must be extremelycareful in considering the biological implications of the tenta-tive retention of the EXGTsp-GFP in the lumen of the ER.However, our finding that the intracellular localization patternof the native EXGT protein (Fig. 4) resembled that of theEXGTsp-GFP reporter fusion protein definitely excludes thepossibility that the visible retention of the EXGTsp in the ERlumen was an artifact caused by its overexpression. On thebasis of these findings, we conclude that a certain amount ofEXGT is present in both the ER and Golgi apparatus.

Boevink et al. (1998) visualized a network of ER tubulesby infecting leaf cells ofNicotiana clevelandiiwith a viral vec-tor expressing an ERD2-GFP fusion protein. More recently,using a similar viral expression system, Boevink et al. (1999)investigated the secretory pathway of a GFP fusion proteinwith a signal peptide of sporamin. In their system, the signalpeptide-GFP fusion protein was not clearly observed in thelumen of the ER, whereas a signal peptide-GFP-KDEL fusionprotein was clearly shown to be located in the ER. By contrast,in tobacco cell lines expressing the EXGTsp-GFP fusion pro-tein, a substantial amount of the EXGTsp-GFP was found inboth the cortical network and the radiating strands of the tubu-lar ER throughout the cytoplasm. The different localization pat-terns of the two signal peptide-GFP fusion proteins might indi-cate their different rates of processing and/or traffic through thelumen of the ER in the two different expression systems. Ahigh rate of secretion would result in a low level of GFP fusionprotein in the ER lumen. Furthermore, since complete matura-tion of GFP is considered to require a few hours (Cubitt et al.1995), rapid trafficking of the ER lumen would cause retentionand secretion of an immature GFP fusion protein, which wouldnot fluoresce. Thus, the visible retention of the secretory pro-tein in the ER lumen strongly implies the existence of a mecha-nism that regulates the flow rate of ER-mediated protein trafficor the amount of protein tentatively retained in the ER lumen.Such a regulatory system would be mediated either by post-translational modification of individual secretory proteins or byany mechanism responsible for controlling vesicle trafficking.At present, we have no clue either to the possible post-translational modification or the molecular device involved inthe rate-regulation of vesicle trafficking. Clearly, such a molec-ular mechanism would be essential in the secretory system andwould form the basis for spatially and temporally regulatedintracellular trafficking of secretory vesicles.

EXGT is specifically targeted to the cell plate during cytokinesAt the beginning of mitosis, the radiating network of ER

tubules distributed throughout the cytoplasm was disconnectAs a result, the cortical domain of the ER network was seprated from the central ER domain located around the nucle(Fig. 3, 5). The segregation of the two ER-network domainseemed to result in the blockade of the secretory pathwayboth the matrix polysaccharide and enzymes, both of which arequired for construction of the cell wall outside the plasmmembrane. Although matrix polysaccharides (Kakimoto anShibaoka 1992) and certain proteins directly related to cdivision have been shown to be transported to the cell platethe end of mitosis, localization of the so-called cell waenzymes in the cell plate has not been investigated. To oknowledge, the present study is the first to demonstratelocalization of a cell wall enzyme in either the phragmoplastcell plate at an early stage of the cytokinesis. Taken togeththese facts imply the presence of a switching system that calternate the direction of the secretory pathway for cell waproteins. Such a switching system is important particularlyplants, because the plant cell needs to construct a new cell wpromptly at the end of mitosis in order to separate daughcells. Clearly, cell-cycle-dependent alteration of the membratrafficking system carries an advantage over a constitutisecretory system, in which the apoplast-directed membratrafficking continues even during cell plate formation.

Novel role of EXGT in cell division and expansionMany studies on the function of EXGT have been focuse

on the modification of the pre-existing cell wall based on thview that EXGT can mediate molecular grafting betweexyloglucan molecules, which function as load-bearing bridgamong cellulose microfibrils (Fry et al. 1992, Nishitani 1997Extensive secretion of EXGT into the apoplast during intephase supports this widely advocated role of EXGT in threconstruction of the pre-existing cell wall.

On the other hand, our finding that the EXGT is localizein the cell plate clearly indicates its essential role in the formtion of the cell plate in addition to its well-known role in theapoplast. Xyloglucan molecules have been shown to polymize in the Golgi apparatus and to exist in the cell plate (Mooand Staehelin 1988). By contrast, cellulose microfibrils are nfound in the cell plate (Kakimoto and Shibaoka 1992, Mooand Staehelin 1988). This means that the EXGT protein presin the cell plate will not mediate either splitting or formation oxyloglucan bridges among cellulose microfibrils. In othewords, the EXGT in the cell plate may play a role distinct fromthat of EXGT in the apoplast.

What, then, is the function of EXGT in the cell plate?There are two possibilities. One is that EXGT might be resposible for the processing of xyloglucans secreted by the Goapparatus-derived vesicles. Certain molecular grafting retions between xyloglucan molecules can mediate elongationxyloglucan chains (Nishitani 1997). Repetition of such rea

Secretory pathways of endoxyloglucan transferase 299

helede

reg-hd-

e-

eidn-rchice

,d

in

9)tory

es,ER

-nd

llsal

g

6)

Y.ns.

nn of

s,me

ti-

ll

man

tions induces generation of high molecular weight xyloglucansfrom xyloglucans of moderate size derived from the Golgi ves-icles (Nishitani and Tominaga 1992). Another explanation isthat the EXGT-mediated molecular grafting between the mobilexyloglucan molecules in the cell plate might be involved in theregulation of rheological properties of the cell plate by modify-ing the molecular weight distribution profile of xyloglucans.Comparative analysis of the structural features of xyloglucansin Golgi apparatus, the cell plate and the cell wall will be nec-essary to examine these hypotheses. Such a study would lead toelucidation of the novel function of EXGT in the cell plate for-mation during cytokinesis, and would thereby clarify its pri-mary action in both cell division and cell expansion.

EXGTsp-GFP as a new tool for studying regulation of thesecretory pathway

In this study, we chose EXGT as a molecular marker toexplore the secretory pathway of apoplastic enzymes becausethis enzyme is considered to play various physiological roles ina wide range of cell wall dynamics. Comparison of intracellularlocalization images of the immunofluorescence (Fig. 4) and theGFP (Fig. 3) clearly demonstrated that the EXGTsp-GFP accu-rately represents the real distribution pattern of the endogen-ous EXGT protein in tobacco BY-2 cells, indicating that theEXGTsp-GFP fusion protein is of considerable utility as a pre-cise molecular marker for the secretory pathway. In the presentstudy, cell lines expressing EXGT-GFP (Fig. 2c, g) were notused for further analysis because they proliferated at a muchlower rate than those expressing EXGTsp-GFP. The reducedproliferation rate of the EXGT-GFP lines might be due to anegative-dominant type of effect caused by the EXGT-GFPfusion protein.

The EXGTsp-GFP was used successfully to visualize thetubular network-structure of the ER due to its temporal reten-tion in the ER. Use of this fusion protein system made itpossible to visualize the vesicle trafficking during cell plateformation in a living cell. Until very recently, our understand-ing of the formation of the cell plate has largely beendependent on electron microscopic and immunocytochemicalmethods (Hepler 1982, Samuels et al. 1995). The most recentmodel of cell plate formation was reported by Samuels et al.(1995), who successfully observed the many new transientmembrane structures by use of improved cell-preservationtechniques adapted for electron microscopy. Although theirexcellent methods for visualizing the cell plate formation infixed cells revealed a wealth of information, the formationprocess of the cell plate in living cells remained largely unclari-fied. Our confocal laser scanning microscopic analysis of cellsexpressing EXGTsp-GFP proteins has clearly disclosed thedynamic aspects of cell plate formation in living cells and pro-vided explicit evidence in support of the model advanced bySamuels et al. (1995). That is, the GFP fluorescence was beingtargeted to the margins of the expanding cell plate, suggestingthat vesicles filled with cell wall polysaccharides and enzymes

for the cell plate formation were transported and fused to tmargins of the cell plate. The GFP fluorescence also reveathe position to which GFP in the ER moved to participate in thformation of vesicles, i.e. the point at which vesicles weformed. This is the first observation of membrane traffickinoccurring during the cell plate formation in living cells. Studies of living cells using EXGTsp-GFP should provide mucadditional novel information toward an enhanced understaning of the cell plate formation in terms of vesicle trafficking.

Acknowledgements

We are grateful to Drs. Toshiyuki Nagata and Seiichiro Haszawa for their gift of the tobacco BY-2 cell line and for their kindadvice on its synchronization. We also thank Dr. Yasuo Niwa for thgift of the sGFP vector. This work was supported by a Grant-in-Afor Scientific Research on Priority Areas (No. 10182102), a Grant-iAid for Scientific Research (No. 09440269), a grant from the Reseafor the Future Program (JSPS-RFTF96L00403) and by a grant of RGenome Project PR-1108, MAFF, JAPAN.

References

Arioli, T., Peng, L., Betzner, A.S., Burn, J., Wittke, W., Herth, W., Camilleri, C.Höfte, H., Plazinski, J., Birch, R., Cork, A., Glover, J., Redmond, J. anWilliamson, R.E. (1998) Molecular analysis of cellulose biosynthesisArabidopsis.Science279: 717–720.

Boevink, P., Martin, B., Oparka, K., A. Santa Cruz, S. and Hawes, C. (199Transport of virally expressed green fluorescent protein through the secrepathway in tobacco leaves is inhibited by cold shock and brefeldin A.Planta208: 392–400.

Boevink, P., Oparka, K., Santa Cruz, S., Martin, B., Betteridge, A. and HawC. (1998) Stacks on tracks: the plant Golgi apparatus traffics on an actin/network.Plant J.15: 441–447.

Bolwell, G.P. (1988) Synthesis of cell wall components: aspects of control.Phy-tochemistry27: 1235–1253.

Brummell, D.A., Camirand, A. and Maclachlan, G.A. (1990) Differential distribution of xyloglucan glycosyl transferases in pea Golgi dictyosomes asecretory vesicles.J. Cell Sci.96: 705–710.

Carpita, N.C. and Gibeaut, D.M. (1993) Structural models of primary cell wain flowering plants: consistency of molecular structure with the physicproperties of the walls during growth.Plant J.3: 1–30.

Chardin, P. and McCormick, F. (1999) Brefeldin A: the advantage of beinuncompetitive.Cell 97: 153–155.

Chiu, W.-I., Niwa, Y., Zeng, W., Hirano, T., Kobayashi, H. and Sheen, J. (199Engineered GFP as a vital reporter in plants.Curr. Biol. 6: 325–330.

Cosgrove, D.J. (2000) Loosening of plant cell walls by expansins.Nature407:321–326.

Cubitt, A.B., Heim, R., Adams, S.R., Boyd, A.E., Gross, L.A. and Tsien, R.(1995) Understanding, improving and using green fluorescent proteiTrends Biochem. Sci.20: 448–455.

Driouich, A., Zhang, G.F. and Staehelin, L.A. (1993) Effect of Brefeldin A othe structure of the Golgi apparatus and on the synthesis and secretioproteins and polysaccharides in Sycamore Maple (Acer pseudoplatunus)suspension-cultured cells.Plant Physiol.101: 1363–1373.

Fry, S.C., Smith, R.C., Renwick, K.F., Martin, D.J., Hodge, S.K. and MatthewK.J. (1992) Xyloglucan endotransglycosylase, a new wall–loosening enzyactivity from plants.Biochem. J.282: 821–828.

Giddings, T.H., Jr., Brower, D. and Staehelin, L.A. (1980) Visualization of parcle complexes in plasma membrane ofMicrasterias denticulataassociatedwith the formation of cellulose microfibrils in primary and secondary cewalls.J. Cell Biol.84: 327–334.

Grabski, S., Feijter, A.W. and Schindler, M. (1993) Endoplasmic reticuluforms a dynamic continuum for lipid diffusion between contiguous soyberoot cells.Plant Cell5: 25–38.

Secretory pathways of endoxyloglucan transferase300

in

velglu-

)pe

se,of

hi-nular

)is

iner

siss-

-

he

00)

Hepler, P.K. (1982) Endoplasmic reticulum in the formation of the cell plate andplasmodesmata.Protoplasma111: 121–133.

Høj, P.B. and Fincher, G.B. (1995) Molecular evolution of plant beta-glucanendohydrolases.Plant J.7: 367–379.

Ito, H. and Nishitani, K. (1999) Visualization of EXGT-mediated moleculargrafting activity by means of a fluorescent-labeled xyloglucan oligomer.Plant Cell Physiol.40: 1172–1176.

Jefferson, R.A., Kavanagh, T.A. and Bevan, M.W. (1987) GUS fusion :�-glucuronidase as a sensitive and versatile gene fusion marker in higher plants.EMBO J.6: 3901–3907.

Kakimoto, T. and Shibaoka, H. (1992) Synthesis of polysaccharides in phragmo-plasts isolated from tobacco BY-2 cells.Plant Cell Physiol.33: 353–361.

Kimura, S., Laosinchai, W., Itoh, T., Cui, X., Linder, C.R. and Brown, R.M., Jr.(1999) Immunogold labeling of rosette terminal cellulose-synthesizing com-plexes in the vascular plantVigna angularis. Plant Cell11: 2075–2086.

Matsuoka, K. and Nakamura, K. (1991) Propeptide of a precursor to a plant vac-uolar protein required for vacuolar targeting.Proc. Natl. Acad. Sci. USA88:834–838.

Moore, P.J. and Staehelin, L.A. (1988) Immunogold localization of the cell wall-matrix polysaccharides rhamnogalacturonan I and xyloglucan during cellexpansion and cytokinesis inTrifolium pratenseL.; implication for secretorypathways.Planta178: 353–366.

Moore, P.J., Swords, K.M.M., Lynch, M.A. and Staehelin, L.A. (1991) Spatialreorganization of the assembly pathways of glycoproteins and complexpolysaccharides in the Golgi apparatus of plants.J. Cell Biol.122: 589–602.

Nagata, T., Nemoto, Y. and Hasezawa, S. (1992) Tobacco BY-2 cell line as the“Hela” cell in the cell biology of higher plants.Int. Rev. Cytol.132: 1–30.

Nagata, T., Okada, K., Takebe, I. and Matsui, C. (1981) Delivery of tobaccomosaic virus RNA into plant protoplasts mediated by reverse-phase evapora-tion vesicles (liposomes).Mol. Gen. Genet.184: 161–165.

Nishitani, K. (1997) The role of endoxyloglucan transferase in the organizationof plant cell walls.Int. Rev. Cytol.173: 157–206.

Nishitani, K. and Tominaga, R. (1991) In vitro molecular weight increasexyloglucans by an apoplastic enzyme preparation from epicotyls ofVignaangularis. Physiol. Plant.82: 490–497.

Nishitani, K. and Tominaga, R. (1992) Endo-xyloglucan transferase, a noclass of glycosyltransferase that catalyzes transfer of a segment of xylocan molecule to another xyloglucan molecule.J. Biol. Chem.267: 21058–21064.

Niwa, Y., Hirano, T., Yoshimoto, K., Shimizu, M. and Kobayashi, H. (1999Non-invasive quantitative detection and applications of non-toxic, S65T-tygreen fluorescent protein in living plants.Plant J.18: 455–463.

Northcote, D.H., Davey, R. and Lay, J. (1989) Use of antisera to localize calloxylan and arabinogalactan in the cell plate, primary and secondary wallsplant cells.Planta178: 353–366.

Okazawa, K., Sato, Y., Nakagawa, T., Asada, K., Kato, I., Tomita, E. and Nistani, K. (1993) Molecular cloning and cDNA sequencing of endoxyloglucatransferase, a novel class of glycosyltransferase that mediates molecgrafting between matrix polysaccharides in plant cell walls.J. Biol. Chem.34: 25364–25368.

Ridge, R.W., Uozumi, Y., Plazinski, J., Hurley, U. and Williamson, R.E. (1999Developmental transitions and dynamics of the cortical ER of Arabidopscells seen with green fluorescent protein.Plant Cell Physiol.40: 1253–1261.

Samuels, A.L., Giddings, T.H. Jr. and Staehelin, L.A. (1995) Cytokinesistobacco BY-2 and root tip cells: a new model of cell plate formation in highplants.J. Cell Biol.130: 1345–1357.

Xu, W., Campbell, P., Vargheese, A.K. and Braam, J. (1996) The ArabidopXET-related gene family: environmental and hormonal regulation of expresion.Plant J.9: 879–889.

Yokoyama, R. and Nishitani, K. (2000) Functional diversity of xyloglucanrelated proteins and its implications in the cell wall dynamics in plants.PlantBiol. 2: 598–604.

Zhang, G.F. and Staehelin, L.A. (1992) Functional compartmentation of tGolgi apparatus of plant cells.Plant Physiol.99: 1070–1083.

(Received September 25, 2000; Accepted December 14, 20