The plant Golgi apparatus

Paul Dupree *, D. Janine SherrierUniversity of Cambridge, Department of Biochemistry, Tennis Court Road, Cambridge CB2 1QW, UK

Received 8 December 1997; accepted 23 December 1997

Abstract

The plant Golgi apparatus has an important role in protein glycosylation and sorting, but is also a major biosyntheticorganelle that synthesises large quantities of cell wall polysaccharides. This is reflected in the organisation of the Golgiapparatus as numerous individual stacks of cisternae that are dispersed through the cell. Each stack is polarised: the shape ofthe cisternae and the staining of the membranes change in a cis to trans direction, and the cisternae on the trans side containmore polysaccharides. Numerous glycosyltransferases are required for the synthesis of the complex cell wall polysaccharides.Microscopy and biochemical fractionation studies suggest that these enzymes are compartmentalised within the stack.Although there is no obvious cis Golgi network, the trans-most cisterna or trans Golgi network often buds clathrin-coatedand sometimes smooth dense vesicles as well. Here, vacuolar proteins are sorted from the secreted proteins andpolysaccharides. This review highlights unique aspects of the organisation and function of the plant Golgi apparatus.Fundamentally similar processes probably underlie Golgi organisation in all organisms, and consideration of the plant Golgispecialisations can therefore be generally informative, as well as being of central importance to plant cell biology. ß 1998Elsevier Science B.V. All rights reserved.

Keywords: Golgi apparatus; Polysaccharide; Dictyosome; Secretion; Plant

1. Organisation of the Golgi apparatus

The plant Golgi apparatus is composed of manysmall stacks of cisternae, sometimes known as dic-tyosomes. The number of stacks and their distribu-tion within the cell is dependent on the cell type.Maize root cap cells, which are actively secretinglarge amounts of mucopolysaccharides, contain be-tween 300 and 600 Golgi stacks per cell [1] but thereare many fewer, on average 25, in the apical meri-stem cells of the hairy-willow herb Epilobiumhirsutum [2] (and references within). The stacks are

usually dispersed throughout the plant cell cytoplasmsingly or in small groups. This distribution can beillustrated dramatically by immuno£uorescence la-belling with the monoclonal antibody JIM84 (Fig.1) [3^6]. This antibody recognises a glycoproteinepitope present in the Golgi, and also in the plasmamembrane of some species [5]. The dispersed distri-bution of the stacks may result from continualstreaming through the cytoplasm, but a direct visual-isation in a living cell to demonstrate this movementhas not yet been achieved. A further consequence ofcytoplasmic streaming may be a loss of a ¢xed rela-tionship between the Golgi and the ER.

In cells with localised sites of cell wall growth orintensive secretion, Golgi stacks can become concen-trated to speci¢c subcellular regions. Within tip

0167-4889 / 98 / $19.00 ß 1998 Elsevier Science B.V. All rights reserved.PII: S 0 1 6 7 - 4 8 8 9 ( 9 8 ) 0 0 0 6 1 - 5

* Corresponding author. Fax: +44 (1223) 333345;E-mail : p.dupree@bioc.cam.ac.uk

BBAMCR 14336 5-8-98

Biochimica et Biophysica Acta 1404 (1998) 259^270

growing cells, like root hairs and pollen tubes, Golgibodies are localised in a region a small distance awayfrom the growing tip [7^9]. Golgi-derived vesiclescontaining cell wall polysaccharides concentratewith other vesicles in the extreme tip of the growingroot hair where most of the other organelles are ex-cluded [7]. However, an association of the Golgi withgrowing regions of the cell is not absolute. It is alsoprobable that vesicles produced by Golgi stacks arelatively long distance from the wall can be targetedto the site of cell growth.

Movement of secretory vesicles away from theGolgi stacks can be inhibited by cytochalasin B, sug-gesting that actin micro¢laments have a role invesicle movement [10] and may also have a role indistributing Golgi stacks throughout the cytoplasm.When microtubules are disrupted, vesicles do migrateaway from the Golgi stack [10], but the site of vesiclefusion may be altered, especially in tip-growing cells[7].

Plant cells divide by the de novo formation of aseparating cell wall between daughter cells. Forma-tion of the cell plate requires intensive secretion of

Golgi-synthesised cell wall polysaccharides. Thephragmoplast, a structure containing micro¢lamentsand microtubules, spatially orients and positions thenew cell wall by directing Golgi-derived vesicles tothe cell plate [11]. The vesicles align and then fuse ina regulated and characteristic manner [12]. A poten-tial speci¢c t-SNARE that might regulate the fusionprocess has recently been identi¢ed by analysis of acytokinesis-defective Arabidopsis embryogenesis mu-tant, knolle [13]. Although Golgi stacks are excludedfrom the direct area of the forming cell plate, theymay associate with the phragmoplast [14,15] and atlater stages become more closely associated with thematuring cell plate [12]. The new plasma membrane,also derived from the Golgi-derived vesicles, acquirestypical plasma membrane characteristics as soon asthe vesicles fuse, including a¤nity for the electronmicroscopy (EM) stain osmium ferricyanide [14],and the ability to synthesise callose [12].

The process of cell division highlights a signi¢cantdi¡erence in behaviour of the Golgi apparatus inplants and animals. To ensure partitioning of mem-branes during cytokinesis of mammalian cells, secre-tion stops and the Golgi apparatus vesiculates, re-forming after mitosis (see the chapter by Shimaand Pepperkok in this issue). In contrast, in plantcells the Golgi apparatus must be intact and func-tional for cytokinesis to occur. Consequently, themechanism to partition Golgi stacks between thedaughter cells in plants may be unique. It may relyon a dispersed distribution in the cytoplasm beforephragmoplast formation, or an interaction with cy-toskeletal elements [11,14].

The organisation into discrete stacks also has theconsequence that their number must increase duringthe cell cycle, and during di¡erentiation into an elon-gating or secretory cell. The mechanism of Golgireplication has been addressed by electron micros-copic morphological studies. Garcia-Herdugo andco-workers [16] found that the number, but not thesize, of Golgi stacks in onion meristematic cells var-ied during the cell cycle, increasing dramatically dur-ing mitosis. Additionally, in a synchronous culture ofMadagascar periwinkle (Catharanthus), the size andnumber of cisternae increased shortly before andduring cytokinesis [17]. Based on electron micro-graphs of stacks, the authors suggested that the Gol-gi stacks divide during cytokinesis by the ¢ssion of

Fig. 1. Distribution of Golgi bodies within two Nicotiana rootcells shown by labelling with JIM84. The monoclonal antibodyJIM84 recognises a carbohydrate epitope on many Golgi andplasma membrane glycoproteins. Golgi bodies are dispersed sin-gly or in small groups throughout the cytoplasm. Micrographcourtesy of Dr. C. Hawes.

BBAMCR 14336 5-8-98

P. Dupree, D.J. Sherrier / Biochimica et Biophysica Acta 1404 (1998) 259^270260

the individual cisternae, moving in the cis to transdirection. However, it is di¤cult at present to ex-clude other mechanisms, including de novo forma-tion of stacks. Perhaps with the discovery of Golgiproteins that can be fused to green £uorescent pro-tein, the timing and mechanism of Golgi replicationmay be investigated more thoroughly within livingplant tissues.

2. Morphology and polarisation of individualGolgi stacks

The size and morphology of individual plant Golgistacks vary tremendously between di¡erent cell typesand species. The cisternal membranes and associatedvesicles can be from 0.5 to 2.0 Wm in diameter. Golgistacks are usually composed of three to eight cister-nae (Fig. 2) [18], and may be composed of 15 ormore in scale-secreting algal cells (Fig. 3) [19]. Never-theless, within a single cell, the cisternal number isremarkably uniform. Three-dimensional reconstruc-tions of serial EM sections through plant Golgistacks have provided a morphological view ratherdi¡erent from the common `stacked-pancake' model.A model of a single dictyosome from Aptenia cordi-folia showed that the stack and associated vesicles

have an overall spherical shape [20]. The modelsalso revealed that there can be tubular connectionsbetween adjacent cisternae, between neighbouringGolgi stacks, and possibly connections to the endo-plasmic reticulum [20,21]. A continuity of cisternaewith the ER has also been suggested in earlier studies[22].

The directional £ow of material through the Golgiapparatus is re£ected in the morphological polarisa-tion of individual Golgi stacks. There are more ve-sicular pro¢les associated with the medial region andtrans side of the Golgi than the cis cisternae. Thecisternae vary in staining characteristics and luminalthickness. The membranes and contents of the di¡er-ent cisternae stain more intensely in a cis to transdirection (Fig. 2) [18,23]. Furthermore, the thicknessof the cisternal lumen decreases in a cis to transdirection. Because this is more constant in animalGolgi stacks, it was originally believed that the di¡er-ences in plant cisternae were artefacts of chemical¢xation. However, studies of tissues preserved byhigh-pressure freeze ¢xation and freeze substitutionhave con¢rmed these di¡erences (Fig. 2) [18,24]. So-called intercisternal elements can be seen betweencisternae on the trans side of the stack in cells ofsome species (Fig. 2) [25,26], and may be involvedin maintaining stack integrity [2,27].

Fig. 2. Golgi stack and trans Golgi network (TGN) in a columella cell of a Nicotiana root cap. Morphological characteristics of thecisternae change in a cis to trans direction. Clathrin coats (arrowheads) are associated with the TGN. ER, endoplasmic reticulum. Barequals 0.5 Wm. Micrograph courtesy of Dr. L.A. Staehelin. Modi¢ed with permission from [27].

BBAMCR 14336 5-8-98

P. Dupree, D.J. Sherrier / Biochimica et Biophysica Acta 1404 (1998) 259^270 261

In cells secreting large amounts of material, thestack polarity is exaggerated. For example, inslime-secreting maize root cap cells the trans-mostcisternal rims are dilated or connected to hugevesicles which bud from the trans side of the Golgistack (Fig. 4) [28]. Similar di¡erentiation of Golgistack morphology is also seen in slime secreting cellsof root tip cells of tobacco and Arabidopsis plants[27]. These cells show beautifully how the Golgi mor-phology of di¡erent plant cells re£ects their function-al di¡erences.

In many higher plants, there is no obvious cisGolgi network (CGN). In some algae the ER andGolgi are clearly associated and the cis Golgi canbe identi¢ed (Fig. 3). However, in most higherplants, especially in cells secreting polysaccharide,there is no ¢xed association with the ER (Fig. 4)[29]. Although this might be considered importantfor transit of proteins between the two organelles,an association is not always obvious from electron

micrographs. Golgi bodies may associate moreclosely with ER under conditions of cold stresswhen polysaccharide secretion is reduced [22]. Evenin cases where the ER is nearby, membranes thatcould constitute a CGN are not often apparent(Fig. 2).

On the trans side of the stack, slightly removedfrom it, is a membrane structure referred to as thetrans Golgi network (TGN). It is less extensive thanin animal cells, and is often reminiscent of a Golgicisterna (Fig. 2). It can be distinguished by its morecomplex morphology, and by the presence of clath-rin-coated and other budding vesicles. EM sectioninghas shown that not every stack has a clear TGN[2,30] and in this case the trans-most cisterna mayfunction in its place (Fig. 4). Alternatively, the so-called partially coated reticulum (PCR), a membraneof similar morphology, could be a `displaced' TGN.Although it is not so closely associated with thestack, it can be connected to it by tubules [30], and

Fig. 3. Golgi stack in the alga Scher¡elia dubia. A membranous scale reticulum (SCR), and secretory vesicles containing scales (arrow-heads) are seen near the trans side of the Golgi body. Vesicles not containing scales (small arrows) are closely associated with the Gol-gi stack. Large arrows indicate clathrin-coated membrane pro¢les. ER, endoplasmic reticulum; M, mitochondrion; P, plastid; V, con-tractile vacuole. Bar equals 0.5 Wm. Micrograph courtesy of Prof. B. Becker. Modi¢ed with permission from [84].

BBAMCR 14336 5-8-98

P. Dupree, D.J. Sherrier / Biochimica et Biophysica Acta 1404 (1998) 259^270262

it also contains clathrin-like coat pro¢les. However,the PCR also has some characteristics of an endo-some [31,32]. Therefore it is not clear if the TGN/PCR is one and the same organelle with functionallydistinct subdomains. Alternatively there may be twoclearly distinct but morphologically similar organ-elles. At present, there are no TGN or PCR markerproteins that can be used to distinguish these possi-bilities.

Golgi morphology, vesicle tra¤cking and secretionhave been studied using Brefeldin A (BFA), a fungaltoxin from the plant pathogen Alternaria carthami.The e¡ects of BFA depend on the cell type and ex-perimental conditions [33]. BFA inhibits secretion ofglycoproteins and polysaccharides [34,35], and dis-

rupts the delivery of soluble proteins to the vacuole[36]. Inhibition of tra¤cking is not always accompa-nied by disintegration of the Golgi stack as seen inmammalian cells. In certain cases, Golgi stack mor-phology and distribution are altered dramatically byBFA and 7-oxo-BFA causing Golgi clustering andvesiculation without fusion with the ER [33,35].This further illustrates the more distant relationshipbetween the ER and Golgi in plants than seen inmammalian cells.

3. Biosynthetic functions of the plant Golgi apparatus

The plant Golgi apparatus synthesises a wide

Fig. 4. Hyper-secretory outer cap cell from a maize root ¢xed in potassium permanganate to stain endomembranes. Golgi bodies(large arrowheads) have extended secretory vesicles attached to the cisternae. Small arrowheads depict secreted mucopolysaccharides.CW, cell wall. Bar equals 1 Wm. Micrograph courtesy of Dr. H.H. Mollenhauer. Modi¢ed with permission from [1].

BBAMCR 14336 5-8-98

P. Dupree, D.J. Sherrier / Biochimica et Biophysica Acta 1404 (1998) 259^270 263

range of cell wall polysaccharides and proteoglycans,and also carries out O-linked glycosylation and N-linked glycan processing. No studies have yet ad-dressed whether glycolipids are modi¢ed in the plantGolgi. As in the animal and yeast Golgi, it is thoughtthat a speci¢c glycosyltransferase is required totransfer each sugar from a nucleotide sugar donorto an oligosaccharide acceptor [37], and the transfer-ases may work in complexes [38]. Although the ac-tivities of several of the enzymes have been exten-sively characterised [39], there are no publishedreports of the cloning of any genes encoding them.Transporters are also required to permit entry of thesugar nucleotide precursors into the lumen of theGolgi [40]. The diverse biosynthetic processes there-fore require a very large number of speci¢c enzymes.In order to simplify the recognition of the substrates,it is probable that the enzymes are subcompartmen-talised within the stack [41].

The synthesis of cell wall polysaccharides in theGolgi apparatus was demonstrated many years agoby in vivo labelling studies followed by autoradiog-raphy [42]. Only cellulose (L1,4-glucan) and callose(L1,3-glucan), are synthesised at the plasma mem-brane. The two abundant classes of Golgi-synthe-sised polysaccharides, the pectins and hemicelluloses,

can constitute between 50 and 80% of the dry weightof the cell wall [39]. Their structures are complex,containing a range of sugars in a variety of linkages.

The pectins and hemicelluloses are distinguishedby their sugar composition and consequent extracti-bility from the cell wall, the pectin being soluble inhot dilute acid, the hemicelluloses in alkali. The pec-tins are highly branched acidic polysaccharides com-posed mainly of galactose and rhamnose, but cancontain over ten other sugars [43]. A major pectinof dicotyledonous plants, rhamnogalacturonan I(RG-I), has a backbone of K1,4-linked galacturonicacid alternating with K1,2-linked rhamnose, and avariety of side chains composed principally of arabi-nose and galactose [44]. The galacturonic acid ofpectin is often partially esteri¢ed in the Golgi [9].Other pectins, including arabinans, galactans, arabi-nogalactans and polygalacturonic acid (PGA) arepresent in variable amounts in di¡erent tissues andspecies.

The hemicelluloses are a broad group of polymersincluding xylans, glucans, mannans, glucomannansand galactomannans. The principle hemicellulose ofthe primary walls of dicotyledonous plants, xyloglu-can, has a L1,4-glucan backbone with xylose sidechains further modi¢ed by galactose and sometimes

Fig. 5. Golgi stack in a root hair of Vicia faba. The tissue is immunolabelled to show xyloglucan (large gold) and methyl esteri¢edpectin (small gold) asymmetrically localised within the Golgi stack. Bar equals 0.25 Wm. Micrograph courtesy of D.J.S. and Dr. K.A.VandenBosch. Modi¢ed with permission from [9].

BBAMCR 14336 5-8-98

P. Dupree, D.J. Sherrier / Biochimica et Biophysica Acta 1404 (1998) 259^270264

fucose. In contrast, the secondary walls contain axylan with side chains of glucuronic acid.

A single Golgi stack can synthesise both pectinsand hemicelluloses (Fig. 5). The plant Golgi appara-tus therefore contains a very wide range of enzymes.The synthesis is highly developmentally regulated,and so the activities are remodelled during cell di¡er-entiation. Nevertheless, even within a single cell, sev-eral di¡erent polysaccharides are synthesised at onetime. Understanding the speci¢city of the enzymes,and the mechanism of their regulation is one of themajor challenges of plant Golgi research.

Glycosylation of the cell wall proteoglycans con-stitutes a signi¢cant proportion of plant Golgi sugartransferase activity. The cell wall arabinogalactan(AGP) and extensin superfamilies of repetitive hy-droxyproline-rich glycoproteins (HRGPs) are exten-sively glycosylated, to the extent that carbohydratecan comprise from 5% to over 90% of their mass[45]. Cell fractionation studies suggest that incorpo-ration of arabinose into HRGPs occurs predomi-nantly in the Golgi apparatus [46,47]. Consistentwith this biochemical evidence, immunolabellingwith antibody raised against hydroxyproline-arabi-nosylated extensin labels the Golgi but not the ER[41]. Hydroxyproline (Hyp) of HRGPs can be thesite of attachment of three to four arabinose resi-dues, and serine may bear a single galactose. Sincenot all Hyp residues are equally glycosylated, recentstudies have investigated the recognition motifs ofthe glycosyltransferases. In a Douglas ¢r HRGP,the peptide Lys-Pro-Hyp-Hyp-Val is always arabino-sylated at Hyp-3, mainly by a triarabinoside [48]. Ingeneral, blocks of hydroxyproline are the most highlyarabinosylated, single non-contiguous Hyp residuesthe least, but single Hyp residues are likely attach-ment sites for the larger arabinogalactan glycanstructures of AGPs.

In general, O-linked glycosylation of intracellularproteins seems to be more limited, but has been littlestudied. Sporamin, a sweet potato vacuolar storageprotein becomes glycosylated on hydroxyproline, andperhaps serine [49]. Sensitivity to BFA and pulsechase experiments suggest this glycosylation occursin the Golgi. Processing of N-linked glycans is sim-ilar to that in animals. A signi¢cant di¡erence is thepresence of K1,3-linked fucose (and not K1,6-linkedfucose) and/or a L1,2-xylose linked to the

Man3GlcNAc2 core. As a consequence, many plantglycoproteins are highly antigenic. There is some het-erogeneity in further processing, with glycans ofsome proteins receiving galactose and fucose toform the Lewis a antigen [50]. Sialic acid has notbeen detected in plant proteins. The activities ofthe glycosidases and glycosyltransferases have beenlocalised to the Golgi apparatus by cell fractionation[51^55].

4. Biochemical subcompartmentation of the Golgi

The complexity of the glycosylation processes inthe plant Golgi, particularly polysaccharide synthe-sis, has led to the suggestion that biochemical sub-compartmentation of the enzymes into cis, medialand trans cisternae would simplify the substrate rec-ognition and synthetic mechanisms [41]. Althoughthere are probably three distinct compartments inthe mammalian and yeast Golgi complex [56], andit is possible to de¢ne morphologically three types ofcisternae [2], the functional equivalence of these com-partments in plants is unproven. There has been nounequivocal demonstration of biochemical compart-mentation of enzymes within the plant Golgi stack.The enzymes have not been localised microscopi-cally, since there are no antibodies or clones yetavailable for these purposes. Furthermore, the ab-sence of a clear CGN and TGN in certain cells,and the morphology suggestive of a maturing stackof cisternae, have led to the suggestion that the plantstack might be a single compartment, analogous tomaturing secretory vesicles [56]. It is important toresolve this point, since it has implications for target-ing of the Golgi enzymes, for understanding themechanism of movement of proteins and polysac-charides through the stack, and for the mechanismand speci¢city of polysaccharide biosynthesis.

There is considerable indirect evidence that theenzyme activities are compartmentalised into distinctcisternae, and morphological polarity would supportthe idea. In a series of immunocytochemical experi-ments, Staehelin and co-workers, and others, haveshown that the polysaccharide products of the Golgienzymes are asymmetrically distributed across thestack. Using antisera that recognised the xyloglucanbackbone, labelling was found over the trans cister-

BBAMCR 14336 5-8-98

P. Dupree, D.J. Sherrier / Biochimica et Biophysica Acta 1404 (1998) 259^270 265

nae and TGN [18,41]. A similar distribution of label-ling was found with antibody recognising a fucose-containing epitope of the xyloglucan side chain[9,18]. Moreover, antibodies recognising epitopes ofthe pectin PGA/RG-I labelled predominantly the cisand medial cisternae in cultured sycamore cells [57].Methyl esteri¢ed pectin is found in the medial andtrans cisternae of vetch root hairs and sycamore cells(Fig. 5) [9,18]. This suggests that pectin synthesis isinitiated in the cis Golgi and methyl esteri¢ed in latercisternae, whereas the hemicellulose is synthesisedpredominantly in the trans cisternae. However, it isunclear whether all precursor and ¢nal processedpolysaccharides are quantitatively recognised bythese antibodies. These studies can only show thesteady state distribution of product, and not of theenzymes themselves. Recently, a reversibly glycosyl-ated peripheral protein RGP1, suggested to be in-volved in polysaccharide synthesis, has been shownto be localised speci¢cally to the cytosolic side ofcisternae on the trans side of the Golgi stack [58],further suggesting a speci¢c biochemical functionfor these cisternae.

Similar immunocytochemical studies using antiserarecognising complex N-linked glycans are also con-sistent with the model of subcompartmentation ofenzymatic activities. In addition to labelling ofpost-Golgi compartments, labelling for xylose-con-taining glycoproteins was highest in the medial cis-ternae, while fucose-containing glycoproteins weremainly detected in the trans cisternae [59]. Immuno-localisation of the complex N-linked glycan Lewis aepitope also revealed a late-Golgi labelling ([50]; L.Faye, personal communication). This could suggestthat xylosylation of glycoproteins occurs earlier inthe stack than fucosylation. However, labelling couldbe found even in the cis cisternae [59], and in anearlier study, little evidence for an symmetric distri-bution was found [60]. Interpretation of these studiesis a¡ected by potential recycling of glycoproteinswithin the stack.

Many studies have attempted to demonstrate sub-compartmentation of enzymes by biochemical frac-tionation. However, the only universally used markeractivity is latent UDPase, a Golgi enzyme activitythought to be required for nucleotide-sugar incorpo-ration [61]. There are no general marker activities forearly or late Golgi compartments. Homogenates of

tissues separated by density gradient centrifugationhave separated latent UDPase into two or three dis-tinct membrane fractions [62^64]. This may re£ectseparation of subcompartments of di¡erent densities.Furthermore, in experiments looking at hemicellulosesynthesis, a glucuronyl transferase and a xylosyl-transferase activity from pea epicotyl were separatedinto membranes of di¡erent densities, indirectly sug-gesting distinct subcompartmentation of the enzymes[63,65]. However, the results could also representseparation of Golgi stacks from di¡erent cell typessynthesising the di¡erent polysaccharides of the pri-mary and secondary cell walls. In contrast to theresults suggesting separation of subcompartments,the separation of the enzymes that process N-linkedglycans could not be achieved [55].

In summary, signi¢cant weight of indirect evidencesuggests that there may be subcompartmentation ofenzymes in the plant Golgi stack. The range of en-zyme activities studied, due to species and tissue spe-ci¢city, means that there are no general markers thatcan be used to distinguish potential subcompart-ments. As antibodies become available, it will be im-portant to see whether localisation of speci¢c glyco-syltransferases supports this model.

5. Sorting functions of the plant Golgi apparatus

The Golgi is involved in returning escaped proteinsback to the ER, sorting of proteins and polysaccha-rides to the cell wall or vacuoles, and in organisingthe compartmentation of its own enzymes by reten-tion or retrieval mechanisms. The evidence for com-partmentation of glycosyltransferases suggests thatthey are e¡ectively maintained in the plant Golgistack. There are few reports yet of studies addressingthe question of targeting to the Golgi, because of theabsence of identi¢ed plant Golgi membrane proteins.A GlcNAc transferase I mutant cgl1 of Arabidopsiscould be complemented by expression of the mam-malian homologue [66], suggesting that targetingmay be conserved to some extent between mammalsand plants. Transgene transferase activity was con-¢ned to Golgi-enriched biochemical fractions, but itcannot be ruled out that inactive protein was presentelsewhere in the cells [66].

Since the initial electron microscopic characterisa-

BBAMCR 14336 5-8-98

P. Dupree, D.J. Sherrier / Biochimica et Biophysica Acta 1404 (1998) 259^270266

tion of the plant Golgi, there has been considerablediscussion about the mechanism of movement of pol-ysaccharides and cell wall material through the stack.The two favoured models are cisternal progression[28] and COPI (coatamer)-mediated vesicular trans-port [67], and the relative merits of the models haverecently been well reviewed [28,29,68,69]. The earliercisternal progression model is supported by observa-tions of the wide range of unicellular algae that syn-thesise scales (Fig. 3) [19,70,71]. These scales formwithin the cisternae, and are never present in Golgiassociated vesicles. Maturation of cisternae into se-cretory vesicles containing large scales can sometimesbe directly observed [72]. Although these dramaticexamples come from algae, the progression modelis also attractive in higher plants. The morphologyof the cisternae and the TGN are sometimes verysimilar. Moreover, in slime secreting cells, the mate-rial transported across the stack is not seen in asso-ciated coated vesicles (Fig. 4) [22]. However, im-provements to the original progression model areneeded to explain the maintenance of compartmen-tation of enzymes across the stack, and the regener-ation of the cis cisternae, given the spatial separationof the ER from the stack.

Escaped ER proteins must be retrieved and re-turned to the ER. The signal for retrieval of ERluminal proteins is a C-terminal KDEL or HDEL[73,74]; retrieval of membrane proteins has not yetbeen studied in plants. As discussed above, there isno clear structure like a CGN near many Golgistacks, and there is often no association of the ERand Golgi stack. The location of the retrieval com-partment is therefore a matter of some debate. Thecompartment for retrieval of the escaped ER proteinsmay not be part of the stack itself [29], but a discretecompartment with an ER association. Nevertheless,possible COPI coat pro¢les are sometimes seen onGolgi cisternae (Fig. 2), suggesting that someCOPI-mediated retrieval occurs within the stack. Aprobable receptor for retrieval of luminal ER pro-teins, aERD2p, has been identi¢ed [75], and it willbe important to determine its location in the plantcell to demonstrate the site of the retrieval compart-ment.

There are several requirements for sorting proc-esses at the trans face of the Golgi apparatus. Thevacuolar proteins are sorted from the proteins des-

tined for the plasma membrane or cell wall. Polysac-charides are also excluded from vesicles that takeproteins to the vacuoles. Furthermore, the recentdemonstration that some plant cells have two distinctvacuoles, a lytic and a storage organelle [76], raisesthe possibility that the proteins destined for theseorganelles are sorted from each other in the Golgi.There are two candidate vesicle types seen at theTGN that might be involved. Clathrin coated vesiclesare probably involved in transport of vacuolar pro-teins in some cells [77,78]. However, in the develop-ing seed storage tissues of pea cotyledons and pump-kin, smooth dense vesicles (SDVs) carrying seedstorage proteins bud from the TGN [79,80]. TheseSDVs, up to 300 nm in diameter, have no apparentcoat. It is not yet clear if these di¡erent vesicle typesare generally involved in transport to the di¡erentvacuoles. Several studies have identi¢ed peptide sig-nals that are able to target test proteins to the vac-uole [81]. These peptide signals, present as propep-tides, may be at the N or C terminus of the protein,and are poorly conserved. A probable vacuolar sort-ing receptor that recognises the N-terminal propep-tide of aleurain, a barley storage protein, has recentlybeen identi¢ed and localised by immunoelectron mi-croscopy to the Golgi stack and a putative prevacuo-lar compartment [82]. Wortmannin can inhibit thesorting of a C-terminally targeted protein, barley lec-tin, suggesting the involvement of a PI kinase in thesorting process [83]. However, the same protein tar-geted with an N-terminal sequence from sweet potatosporamin was relatively insensitive. This supports theidea that there are two di¡erent sorting pathways,and it is tempting to speculate that the di¡erent path-ways and peptide signals are related to the need totarget to di¡erent vacuole types.

6. Summary

The Golgi apparatus is central to the growth anddivision of the plant cell through its roles in proteinglycosylation, protein sorting, and cell wall synthesis.The structure of the plant Golgi re£ects the relativeimportance of these roles. The molecular tools toinvestigate these processes, such as genes encodingGolgi proteins, are now becoming available, andwe will soon be able to investigate the challenging

BBAMCR 14336 5-8-98

P. Dupree, D.J. Sherrier / Biochimica et Biophysica Acta 1404 (1998) 259^270 267

questions of compartmentation of the enzymes andthe functional relationship of the TGN and PCR,and to visualise the dynamics of the Golgi in a livingcell.

Acknowledgements

The authors thank Dr. B. Becker, Dr. C. Hawes,Dr. H.H. Mollenhauer, Dr. L.A. Staehelin and Dr.K.A. VandenBosch for providing ¢gures, Dr. L.Faye for communicating results prior to publication,and Tracy Prime and Piers Mahon for suggestionsand criticism of the manuscript.

References

[1] H.H. Mollenhauer, D.J. Morre, Structure of Golgi appara-tus, Protoplasma 180 (1994) 14^28.

[2] L.A. Staehelin, I. Moore, The plant Golgi apparatus: struc-ture, functional organization and tra¤cking mechanisms,Annu. Rev. Plant Physiol. Plant. Mol. Biol. 46 (1995) 261^288.

[3] D.E. Evans, P.J. Clay, S. Attree, L.C. Fowke, Visualisationof Golgi apparatus in methacrylate embedded conifer em-bryo tissue using the monoclonal antibody JIM84, CellBiol. Int. 21 (1997) 295^302.

[4] J. Henderson, B. Satiat-Jeunemaitre, R. Napier, C. Hawes,Brefeldin A-induced disassembly of the Golgi apparatus isfollowed by disruption of the endoplasmic reticulum in plantcells, J. Exp. Bot. 45 (1994) 1347^1351.

[5] D. Horsley, J. Coleman, D. Evans, K. Crooks, J. Peart, B.Satiat-Juenemaitre, C. Hawes, A monoclonal antibody,JIM84, recognizes the Golgi apparatus and plasma mem-brane in plant cells, J. Exp. Bot. 44 (1993) 223^229.

[6] B. Satiat-Jeunemaitre, C.R. Hawes, Redistribution of a Gol-gi glycoprotein in plant cells treated with Brefeldin A, J. CellSci. 103 (1992) 1153^1166.

[7] E. Schnepf, Cellular polarity, Annu. Rev. Plant Physiol. 37(1986) 23^47.

[8] R.W. Ridge, Freeze substitution improves the ultrastructuralpreservation of legume root hairs, Bot. Mag. Tokyo 101(1988) 427^441.

[9] D.J. Sherrier, K.A. VandenBosch, Secretion of cell wallpolysaccharides in Vicia root hairs, Plant J. 5 (1994) 185^195.

[10] H.H. Mollenhauer, D.J. Morre, Cytochalasin B, but notcolchicine, inhibits migration of secretory vesicles in roottips of maize, Protoplasma 87 (1976) 39^48.

[11] C.R. Schopfer, P.K. Hepler, Distribution of membranes andthe cytoskeleton during cell plate formation in pollen mothercells of Tradescantia, J. Cell Sci. 100 (1991) 717^728.

[12] A.L. Samuels, T.H. Giddings, L.A. Staehelin, Cytokinesis intobacco BY-2 and root tip cells : a new model of cell plateformation in higher plants, J. Cell Biol. 130 (1995) 1345^1357.

[13] W. Lukowitz, U. Mayer, G. Juergens, Cytokinesis in theArabidopsis embryo involves the syntaxin-related KNOLLEgene product, Cell 84 (1996) 61^71.

[14] P.K. Hepler, Endoplasmic reticulum in the formation of thecell plate and plasmodesmata, Protoplasma 111 (1982) 121^133.

[15] W.G. Whaley, H.H. Mollenhauer, The Golgi apparatus andcell plate formation ^ a postulate, J. Cell Biol. 17 (1963) 216^221.

[16] G. Garcia-Herdugo, J.A. Gonzalez-Reyes, F. Garcia-Navar-ro, P. Navas, Growth kinetics of the Golgi apparatus duringthe cell cycle in onion root meristems, Planta 175 (1988)305^312.

[17] S. Hirose, A. Komamine, Changes in the ultrastructure ofGolgi apparatus during the cell cycle in a synchronous cul-ture of Catharanthus roseus, New Phytol. 111 (1989) 599^605.

[18] G.F. Zhang, L.A. Staehelin, Functional compartmentationof the Golgi apparatus of plant cells, Plant Physiol. 99 (1992)1070^1083.

[19] M. Melkonian, B. Becker, D. Becker, Scale formation inalgae, J. Electron. Micro. Tech. 17 (1991) 165^178.

[20] J. Lockhausen, U. Kristen, W. Menhardt, W.J. Dallas,Three-dimensional reconstruction of a plant dictyosomefrom series of ultra-thin sections using computer image proc-essing, J. Microscopy 158 (1990) 197^205.

[21] U. Kristen, J. Lockhausen, D.G. Robinson, Three dimen-sional reconstruction of a dictyosome using serial sections,J. Exp. Bot. 35 (1984) 1113^1118.

[22] H.H. Mollenhauer, D.J. Morre, W.J. Vanderwoude, Endo-plasmic reticulum-Golgi apparatus associations in maizeroot tips, Mikroskopie 31 (1975) 257^272.

[23] S.N. Grove, C.E. Bracker, D.J. Morre, Cytomembrane dif-ferentiation in the endoplasmic reticulum-Golgi apparatus-vesicle complex, Science 161 (1968) 171^173.

[24] P.J. Moore, L.A. Staehelin, Immunogold localization of thecell-wall-matrix polysaccharides rhamnogalacturonan I andxyloglucan during cell expansion and cytokinesis in Trifoliumpratense L.; implications for secretory pathways, Planta 174(1988) 433^445.

[25] H.H. Mollenhauer, An intercisternal structure in the Golgiapparatus, J. Cell Biol. 24 (1965) 504^511.

[26] F.R. Turner, W.G. Whaley, Intercisternal elements of theGolgi apparatus, Science 147 (1965) 1303^1304.

[27] L.A. Staehelin, T.H. Giddings, J.Z. Kiss, F.D. Sack, Macro-molecular di¡erentiation of Golgi stacks in root tips of Ara-bidopsis and Nicotiana seedlings as visualised in high pres-sure frozen and freeze substituted samples, Protoplasma 157(1990) 75^91.

[28] E. Schnepf, Golgi apparatus and slime secretion in plants:the early implications and recent models of membrane traf-¢c, Protoplasma 172 (1993) 3^11.

BBAMCR 14336 5-8-98

P. Dupree, D.J. Sherrier / Biochimica et Biophysica Acta 1404 (1998) 259^270268

[29] C. Hawes, B. Satiat-Jeunemaitre, Stacks of questions: howdoes the plant Golgi work?, Trends Plant Sci. 1 (1996) 395^401.

[30] S. Hillmer, H. Freundt, D.G. Robinson, The partially coatedreticulum and its relationship to the Golgi apparatus in high-er plant cells, Eur. J. Cell Biol. 47 (1988) 206^212.

[31] M.E. Galway, P.J. Rennie, L.C. Fowke, Ultrastructure ofthe endocytotic pathway in glutaraldehyde-¢xed and high-pressure frozen/freeze-substituted protoplasts of white spruce(Picea glauca), J. Cell Sci. 106 (1993) 847^858.

[32] M.A. Tanchak, L.R. Gri¤ng, B.G. Mersey, L.C. Fowke,Endocytosis of cationised ferritin by coated vesicles of soy-bean protoplasts, Planta 162 (1984) 481^486.

[33] B. Satiat-Jeunemaitre, C. Hawes, G.A.T.T. (a general agree-ment on tra¤c and transport) and Brefeldin A in plant cells,Plant Cell 6 (1994) 463^467.

[34] A. Driouich, G.F. Zhang, L.A. Staehelin, E¡ect of BrefeldinA on the structure of the Golgi apparatus and on the syn-thesis and secretion of proteins and polysaccharides in syc-amore maple (Acer pseudoplatanus) suspension-cultured cells,Plant Physiol. 101 (1993) 1363^1373.

[35] A. Driouich, A. Jauneau, L.A. Staehelin, 7-dehydrobrefeldinA, a naturally occurring Brefeldin A derivative, inhibits se-cretion and causes a cis-to-trans breakdown of Golgi stacksin plant cells, Plant Physiol. 113 (1997) 487^492.

[36] L. Gomez, M.J. Chrispeels, Tonoplast and soluble vacuolarproteins are targeted by di¡erent mechanisms, Plant Cell 5(1993) 1113^1124.

[37] D.M. Gibeaut, N.C. Carpita, Biosynthesis of plant cell wallpolysaccharides, FASEB J. 8 (1994) 904^915.

[38] P. Albersheim, A. Darvill, K. Roberts, L.A. Staehelin, J.E.Varner, Do the structures of cell wall polysaccharides de¢netheir mode of synthesis?, Plant Physiol. 113 (1997) 1^3.

[39] C. Brett, K. Waldron, Physiology and Biochemistry of PlantCell Walls, 2nd edn., Chapman and Hall, London, 1996.

[40] A. Orellana, G. Neckelmann, L. Norambuena, Topographyand function of Golgi uridine-5P-diphosphatase from peastems, Plant Physiol. 114 (1997) 99^107.

[41] P.J. Moore, K.M. Swords, M.A. Lynch, L.A. Staehelin, Spa-tial organization of the assembly pathways of glycoproteinsand complex polysaccharides in the Golgi apparatus ofplants, J. Cell Biol. 112 (1991) 589^602.

[42] D.H. Northcote, J.D. Pickett-Heaps, A function of the Golgiapparatus in polysaccharide synthesis and transport in theroot cap cells of wheat, Biochem. J. 98 (1965) 159^167.

[43] E. Zablackis, J. Huang, B. Mueller, A.G. Darvill, P. Al-bersheim, Characterisation of the cell-wall polysaccharidesof Arabidopsis thaliana leaves, Plant Physiol. 107 (1995)1129^1138.

[44] M. McNeil, A.G. Darvill, P. Albersheim, Structure of plantcell walls. X. Rhamnogalacturonan I, a structurally complexpectic polysaccharide in the walls of suspension-culturedsycamore cells, Plant Physiol. 66 (1980) 1128^1134.

[45] J.P. Knox, Developmentally regulated proteoglycans andglycoproteins of the plant cell surface, FASEB J. 9 (1995)1004^1012.

[46] M.C. Gardiner, M.J. Chrispeels, Involvement of the Golgiapparatus in the synthesis and secretion of hydroxyproline-rich cell wall glycoproteins, Plant Physiol. 55 (1975) 536^541.

[47] R.J. Owens, D.H. Northcote, The location of arabinosyl:hy-droxyproline transferase in the membrane system of potatotissue culture cells, Biochem. J. 195 (1981) 661^667.

[48] M.J. Kieliszewski, M. O'Neill, J. Leykam, R. Orlando, Tan-dem mass-spectrometry and structural elucidation of glyco-peptides from a hydroxyproline-rich plant-cell wall glycopro-tein indicate that contiguous hydroxyproline residues are themajor sites of hydroxyproline O-arabinosylation, J. Biol.Chem. 270 (1995) 2541^2549.

[49] K. Matsuoka, N. Watanabe, K. Nakamura, O-Glycosylationof a precursor to a sweet potato vacuolar protein, sporamin,expressed in tobacco cells, Plant J. 8 (1995) 877^889.

[50] A.C. Fitchette-Laine, V. Gomord, M. Cabanes, J.C. Michal-ski, M. Saint Macary, B. Foucher, B. Cavelier, C. Hawes, P.Lerouge, L. Faye, N-Glycans harboring the Lewis a epitopeare expressed at the surface of plant cells, Plant J. 12 (1997)1411^1417.

[51] M.J. Chrispeels, UDP-GlcNAc-glycoprotein GlcNAc-trans-ferase is located in the Golgi apparatus of developing beancotyledons, Plant Physiol. 78 (1985) 835^838.

[52] A. Sturm, H. Kindl, Fucosyltransferase activity and fucoseincorporation in vivo as markers for subfractionating cu-cumber microsomes, FEBS Lett. 160 (1983) 165^168.

[53] S.M.D. Ali, T. Mitsui, T. Akazawa, Golgi-speci¢c localiza-tion of transglycosylases engaged in glycoprotein biosynthe-sis in suspension-cultured cells of sycamore (Acer pseudopla-tanus L.), Arch. Biophys. Biochem. 251 (1986) 421^431.

[54] K.D. Johnson, M.J. Chrispeels, Substrate speci¢cities of N-acetylglucoaminyl-, fucosyl-, and xylosyltransferases thatmodify glycoproteins in the Golgi apparatus of bean cotyle-dons, Plant Physiol. 84 (1987) 1301^1308.

[55] A. Sturm, K.D. Johnson, T. Szumilo, A.D. Elbein, M.J.Chrispeels, Subcellular localization of glycosidases and gly-cosyltransferases involved in the processing of N-linked oli-gosaccharides, Plant Physiol. 85 (1987) 741^745.

[56] I. Mellman, K. Simons, The Golgi complex: in vitro veri-tas?, Cell 68 (1992) 829^840.

[57] M.A. Lynch, L.A. Staehelin, Domain-speci¢c and cell-typespeci¢c localization of two types of cell wall matrix polysac-charides in the clover root tip, J. Cell Biol. 118 (1992) 467^479.

[58] K.S. Dhugga, S.C. Tiwari, P.M. Ray, A reversibly glycosyl-ated polypeptide (RGP1) possibly involved in plant cell wallsynthesis : puri¢cation, gene cloning, and trans Golgi local-ization, Proc. Natl. Acad. Sci. USA 94 (1997) 7679^7684.

[59] A.C. Fitchette-Laine, V. Gomord, A. Chekka¢, L. Faye,Distribution of xylosylation and fucosylation in the plantGolgi apparatus, Plant J. 5 (1994) 673^682.

[60] A.C. Laine, V. Gormord, L. Faye, Xylose-speci¢c antibodiesas markers of subcompartmentation of terminal glycosyla-tion in the Golgi apparatus of sycamore cells, FEBS Lett.295 (1991) 179^184.

[61] P. Munoz, L. Norambuena, A. Orellana, Evidence for a

BBAMCR 14336 5-8-98

P. Dupree, D.J. Sherrier / Biochimica et Biophysica Acta 1404 (1998) 259^270 269

UDP-glucose transporter in Golgi apparatus-derived vesiclesfrom pea and its possible role in polysaccharide biosynthesis,Plant Physiol. 112 (1996) 1585^1594.

[62] D.A. Brummell, A. Camirand, G.A. Maclachlan, Di¡eren-tial distribution of xyloglucan glycosyl transferase in peaGolgi dictyosomes and secretory vesicles, J. Cell Sci. 96(1990) 705^710.

[63] M.C. Hobbs, M.H.P. Delarge, E.A.H. Baydoun, C.T. Brett,Di¡erential distribution of a glucuronyltransferase, involvedin glucuronoxylan synthesis, within the Golgi apparatus ofpea (Pisum sativum var. Alaska), Biochem. J. 277 (1991)653^658.

[64] A. Grunow, B. Becker, M. Melkonian, Isolation and char-acterisation of the Golgi apparatus of a £agellate scaly greenalga, Eur. J. Cell Biol. 61 (1993) 10^20.

[65] E.A.H. Baydoun, C.T. Brett, Distribution of xylosyltransfer-ases and glucuronyltransferases within the Golgi apparatusin etiolated pea (Pisum sativum L.) epicotyls, J. Exp. Bot. 48(1997) 1209^1214.

[66] L. Gomez, M.J. Chrispeels, Complementation of an Arabi-dopsis thaliana mutant that lacks complex asparagine-linkedglycans with the human cDNA encoding N-acetylglucosami-nyltransferase I, Proc. Natl. Acad. Sci. USA 91 (1994) 1829^1833.

[67] J.E. Rothman, F.T. Wieland, Protein sorting by transportvesicles, Science 272 (1996) 227^234.

[68] R. Schekman, I. Mellman, Does COPI go both ways?, Cell90 (1997) 197^200.

[69] B.S. Glick, T. Elston, G. Oster, A cisternal maturationmechanism can explain the asymmetry of the Golgi stack,FEBS Lett. 414 (1997) 177^181.

[70] B. Becker, M. Melkonian, Intra-Golgi transport mediated byvesicles?, Bot. Acta 108 (1995) 172^183.

[71] B. Becker, B. Bolinger, M. Melkonian, Anterograde trans-port of algal scales through the Golgi-complex is not medi-ated by vesicles, Trends Cell Biol. 5 (1995) 305^307.

[72] R.M. Brown, Observations on the relationship of the Golgiapparatus to wall formation in the marine chrysophyceanalga. Pleurochrysis scher¡elii Pringsheim, J. Cell Biol. 41(1969) 109^123.

[73] J. Denecke, R. De Rycke, J. Botterman, Plant and mamma-lian sorting signals for protein retention in the endoplasmicreticulum contain a conserved epitope, EMBO J. 11 (1992)2345^2355.

[74] E.M. Herman, B.W. Tague, L.M. Ho¡man, S.E. Kjemtrup,M.J. Chrispeels, Retention of phytohemagglutinin with car-boxyterminal tetrapeptide KDEL in the nuclear envelopeand the endoplasmic reticulum, Planta 182 (1990) 305^312.

[75] H.I. Lee, S. Gal, T.C. Newman, N.V. Raikhel, The Arabi-dopsis endoplasmic-reticulum retention receptor functions inyeast, Proc. Natl. Acad. Sci. USA 90 (1993) 11433^11437.

[76] N. Paris, C.M. Stanley, R.L. Jones, J.C. Rogers, Plant cellscontain two functionally distinct vacuolar compartments,Cell 85 (1996) 563^572.

[77] S.M. Harley, L. Beevers, Coated vesicles are involved in thetransport of storage proteins during seed development inPisum sativum L., Plant Physiol. 91 (1989) 674^678.

[78] D.G. Robinson, Clathrin-mediated tra¤cking, Trends PlantSci. 1 (1996) 349^355.

[79] I. Hohl, D.G. Robinson, M.J. Chrispeels, G. Hinz, Trans-port of storage proteins to the vacuole is mediated byvesicles without a clathrin coat, J. Cell Sci. 109 (1996)2539^2550.

[80] I. Hara-Hishimura, Y. Takeuchi, K. Inoue, M. Nishimura,Vesicle transport and processing of the precursor to 2S al-bumin in pumpkin, Plant J. 4 (1993) 793^800.

[81] J.-M. Neuhaus, Protein targeting to the plant vacuole, PlantPhysiol. Biochem. 34 (1996) 217^221.

[82] N. Paris, S.W. Rogers, L.W. Jiang, T. Kirsch, L. Beevers,T.E. Phillips, J.C. Rogers, Molecular cloning and furthercharacterization of a probable plant vacuolar sorting recep-tor, Plant Physiol. 115 (1997) 29^39.

[83] K. Matsuoka, D.C. Bassham, N. Raikhel, K. Nakamura,Di¡erent sensitivity to wortmannin of two vacuolar sortingsignals indicates the presence of distinct sorting machineriesin tobacco cells, J. Cell Biol. 130 (1995) 1307^1318.

[84] B. Becker, M. Melkonian, The secretory pathway of protists^ spatial and functional organization and evolution, Micro-biol. Rev. 60 (1996) 697^721.

BBAMCR 14336 5-8-98

P. Dupree, D.J. Sherrier / Biochimica et Biophysica Acta 1404 (1998) 259^270270