Exocytosis and cell polarity in plants – exocyst and recycling domains


  • Viktor Žárský,

    1. Department of Plant Physiology, Charles University, Viničná 5, 128 44 Praha 2, Czech Republic
    2. Institute of Experimental Botany, Academy of Sciences of the Czech Republic, Rozvojová 263, 165 02 Praha 6, Czech Republic
    Search for more papers by this author
  • Fatima Cvrčková,

    1. Department of Plant Physiology, Charles University, Viničná 5, 128 44 Praha 2, Czech Republic
    Search for more papers by this author
  • Martin Potocký,

    1. Institute of Experimental Botany, Academy of Sciences of the Czech Republic, Rozvojová 263, 165 02 Praha 6, Czech Republic
    Search for more papers by this author
  • Michal Hála

    1. Institute of Experimental Botany, Academy of Sciences of the Czech Republic, Rozvojová 263, 165 02 Praha 6, Czech Republic
    Search for more papers by this author

Author for correspondence:
Viktor Žárský
Tel: +420 221 951 685
Emails: zarsky@ueb.cas.cz, viktor@natur.cuni.cz



  • Summary 255

  • I. Introduction 256
  • II. Donor compartments for plant exocytotic vesicles: where do they come from? 256
  • III. Mechanisms of exocytotic vesicle formation: is the truth naked? 257
  • IV. Building the tracks for exocytotic vesicles: cytoskeleton in exocytosis and cell polarization 259
  • V. Manysidedness of a plant cell: delimiting functional plasmalemma domains within a cell 260
  • VI. Vesicle tethering, docking and fusion 262
  • VII. Polarity by recycling: exocytosis, endocytosis and recycling domains in plant cells 264
  • VIII Conclusions: constitutive or regulated secretion in plant cells? 266
  • Acknowledgements 267

  • References 267


In plants, exocytosis is a central mechanism of cell morphogenesis. We still know surprisingly little about some aspects of this process, starting with exocytotic vesicle formation, which may take place at the trans-Golgi network even without coat assistance, facilitated by the local regulation of membrane lipid organization. The RabA4b guanosine triphosphatase (GTPase), recruiting phosphatidylinositol-4-kinase to the trans-Golgi network, is a candidate vesicle formation organizer. However, in plant cells, there are obviously additional endosomal source compartments for secretory vesicles. The Rho/Rop GTPase regulatory module is central for the initiation of exocytotically active domains in plant cell cortex (activated cortical domains). Most plant cells exhibit several distinct plasma membrane domains, established and maintained by endocytosis-driven membrane recycling. We propose the concept of a ‘recycling domain’, uniting the activated cortical domain and the connected endosomal compartments, as a dynamic spatiotemporal entity. We have recently described the exocyst tethering complex in plant cells. As a result of the multiplicity of its putative Exo70 subunits, this complex may belong to core regulators of recycling domain organization, including the generation of multiple recycling domains within a single cell. The conventional textbook concept that the plant secretory pathway is largely constitutive is misleading.


abscisic acid


activated cortical domain


arabinogalactan protein


adenosine triphosphatase




calcium-dependent protein kinase




dynamin-related protein


early endosome


GTPase-activating protein


GDP dissociation inhibitor


guanine nucleotide exchange factor




guanosine triphosphatase


late endosome


multivesicular body


NADPH oxidase


phosphatidic acid


protein kinase D


phospholipase C/D








PtdIns4P,5-kinase isoform 3


recycling domain


recycling endosome


receptor-like serine/threonine kinase


reactive oxygen species


secreted green fluorescent protein


tip growth apparatus


trans-Golgi network


TGN-to-plasma membrane carrier


xyloglucan endotransglycosylase

I. Introduction

Exo- and endocytosis in plant cells should be understood as inseparable phases of the same dynamic secretory process (Battey et al., 1999). Surprisingly, we still know less about plant exocytosis than about transport to the vacuole or endocytosis (Rojo & Denecke, 2008). Foresti & Denecke (2008) recently noted: ‘Solving the mystery of how secreted proteins reach the plasma membrane remains a formidable challenge for the plant field’. We thus may need to speculate more often than usual for this type of review.

Here, we focus on the last exocytosis step between donor compartments and the plasmalemma. Details on endomembrane compartment relationships, cellular and molecular machineries of the ‘core’ secretory pathway and other relevant phenomena not covered here can be found in recent reviews, including whole special issues of journals [Plant Physiology 147(4), 2008; Current Opinion in Plant Biology 11(6), 2008]. However, we must include some aspects of endocytosis, important in the context of cell polarity regulation and in the proposed concept of recycling domains (RDs). We apologize to the authors of many relevant reports not covered because of space limitations.

We use the terms ‘exocytotic vesicle’ and ‘secretory vesicle’ as synonyms denoting any membrane container competent to fuse with the plasmalemma in vivo. Specifically for exocytotic carriers from the trans-Golgi network (TGN) to the plasmalemma, we adopt the previously proposed term ‘TGN-to-plasma membrane carriers’ (TPCs; Bard & Malhotra, 2006).

II. Donor compartments for plant exocytotic vesicles: where do they come from?

Exocytotic containers vary in size, appearance and contents, indicating possible diverse compartment origins. The largest secretory compartments may be multivesicular bodies (MVBs; Février & Raposo, 2004) or whole trans-cisternae of the Golgi apparatus in some unicellular algae. A typical exocytotic vesicle diameter ranges between 60 and 150 nm; in Arabidopsis pollen tubes, it is c. 180 nm, whereas, in root hairs, it is only c. 70 nm (Ketelaar et al., 2008). Chara has two classes of exocytotic vesicle – 200 nm light and 180 nm dark (Limbach et al., 2008). Even budding yeast has two types of 100 nm secretory vesicle – those containing the Bgl2 glucantransferase and plasma membrane proteins and those carrying periplasmic invertase. Although Bgl2 vesicles travel directly from TGN to the plasmalemma, invertase is transported through endosomal compartments (Harsay & Schekman, 2002).

Which plant endomembrane system compartments produce exocytotic vesicles? The answer depends on the cell type; however, TGN is clearly not the only source. The structural complexity of angiosperm tissues, as well as the molecular complexity of, for example, the plant RabA guanosine triphosphatase (GTPase) subfamily (Rutherford & Moore, 2002), suggests multiple pathways to the cell surface. The diversity of the Arabidopsis plasmalemma SNAREs is consistent with at least three exocytotic pathways (Uemura et al., 2004). Different pathways are used for secreted proteins versus cell wall polysaccharides, and two closely related syntaxins, SYP121 and SYP122, participate in two independent exocytotic pathways in tobacco (Leucci et al., 2007; Rehman et al., 2008).

1. The Golgi and trans-Golgi network (I, II)

Exocytosis from TGN is considered the default: cargos without specific sorting signals are secreted. In addition, in plants, soluble cargos travel to the apoplast by default, as documented, for example, for secreted green fluorescent protein (sec-GFP) (Batoko et al., 2000) or the Clv3 peptide (Rojo et al., 2002). However, at least some plasmalemma proteins are possibly sorted at the TGN (Bard & Malhotra, 2006; Wang et al., 2006; Foresti & Denecke, 2008).

RabE/Rab8 GTPases may characterize a Golgi/TGN compartment with direct exocytotic connection to the plasmalemma (Zheng et al., 2005; pathway II in Fig. 1), although they may also alternatively function together with RabA in pathway I (see below; Woollard & Moore, 2008).

Figure 1.

Sources of plant exocytotic vesicles from indisputable (filled heavy arrow) to hypothetical (broken light arrow). Marker proteins for various compartments are shown in bold. The Golgi apparatus and trans-Golgi network (TGN) can be viewed as at least three partially overlapping subcompartments – TGN/early endosome (EE)/recycling endosome (RE) (I), TGN/Golgi (II) and a CC domain (coat in red). TGN-to-plasma membrane carriers (TPCs) may arise without specific coat proteins as a result of lipid modification-induced membrane deformation (blue halo) or by dynamin-induced membrane scission (purple rings). Other possible sources of secretory vesicles include GNOM and other putative RE compartments (III), the prevacuolar compartment (PVC) [multivesicular body (MVB)/late endosome (LE); IV; IVb shows whole MVB exocytosis with exosome vesicle release to the apoplast), and ‘kiss-and-run’ vesicles that may originate from various compartments (V). Relationships with other endomembrane compartments are not shown.

Plant TGN is partially a Golgi apparatus-independent organelle, as TGN and Golgi SNARE markers are often separated (Uemura et al., 2004). The notion that plant TGN itself is an early (or recycling) endosome is now generally accepted (Dettmer et al., 2006; Robinson et al., 2008; Woollard & Moore, 2008). TGN consists of at least three partially overlapping domains, one of which, characterized by the presence of the V-adenosine triphosphatase (ATPase) subunit paralogue VHA-a1, syntaxin SYP41 and SCAMP1 (Dettmer et al., 2006; Lam et al., 2007), represents a TGN/early endosome (EE) compartment, partially overlapping with the RabA-2/3 compartment (Chow et al., 2008). TGN may function also as a recycling endosome (RE); it is unclear whether plant TGN/EE serves also as a sorting endosome (for example, Foresti & Denecke, 2008; Robinson et al., 2008; Woollard & Moore, 2008). Different TGN subdomains may harbour not only different Arf and Rab GTPases as specific membrane domain organizers (Zerial & McBride, 2001), but also specific local modifications of membrane lipid composition. The maturation of TGN-derived compartments may involve Rab conversion, that is subsequent replacement of Rab GTPase isoforms (Rink et al., 2005). Within the framework of our RD concept (see below), we may expect diversification of TGN/EE/REs, resulting in the coexistence of differentially equipped/matured TGNs within the same cell.

2. The recycling endosome (III)

The first plant RE candidate characterized at the molecular level was the Arf guanine nucleotide exchange factor (GEF) GNOM-dependent compartment participating in the constitutive recycling of PIN auxin transporters (Geldner et al., 2003), driven by clathrin-dependent PIN endocytosis (Dhonukshe et al., 2007). Two endocytotic pathways, one clathrin-dependent and the other clathrin-independent, have been documented recently in tobacco. Endocytosed nanogold particles first enter the early endosome compartment with a tubulo-vesicular structure, and pulse-chase experiments have proven that most of the internalized marker is recycled back to the plasma membrane (Onelli et al., 2008). Several different REs (with different Arfs/Rabs and their regulators) may co-exist within the same plant cell.

3. The prevacuolar compartment (multivesicular body, late endosome; IV)

The auxin influx carrier Aux1 is recycled by a mechanism distinct from the PIN1 recycling pathway, involving the SNX1 endosome characterized by the presence of sorting nexin 1 (AtSNX1; Jaillais et al., 2006), and possibly identical with the late endosome (LE) or MVB, characterized by the RabF GTPases, the PEP12 SNARE and phosphatidylinositol-3-phosphate-rich membranes (Robinson et al., 2008).

Exocytosis from LE/MVB or even lysosomes (for example, D. Li et al., 2008) is well documented in metazoans, especially in the form of whole MVB fusion with the plasmalemma, resulting in exosome vesicle release into the apoplast (Février & Raposo, 2004; Fig. 1, IVb). In plants, possible direct exocytosis from MVB to the plasma membrane is accepted as a speculative possibility (see Foresti & Denecke, 2008; Robinson et al., 2008). Plant proteins recycle to the plasmalemma mostly indirectly via TGN, employing the retromer salvage pathway for vacuolar sorting receptors (see Robinson et al., 2008; Woollard & Moore, 2008). The operation of the MVB–exosomal pathway (Fig. 1, IVb) in plants was recently indicated by Meyer et al. (2009) as participating in the formation of pathogen-induced cell wall compartments.

4. ‘Kiss-and-run’ vesicles (V)

‘Kiss-and-run’ exocytosis, immediately followed by vesicle retrieval, has been documented in plant protoplasts (Weise et al., 2000). Kiss-and-run vesicles might also be refilled with a solute cargo, such as, for example, auxin, in the cytoplasm (Baluška et al., 2005).

Different cargos might use the same endosome as early, sorting or recycling, making finite categorization of endosome compartments impossible.

III. Mechanisms of exocytotic vesicle formation: is the truth naked?

Although most, if not all, vesicles are believed to form with the aid of coat proteins, coat-like structures are observed rarely on presumed plant secretory vesicles (for example, Staehelin & Moore, 1995). The mechanisms of TPC formation belong to major ‘unsolved mysteries in membrane traffic’ (Pfeffer, 2007).

Local modifications of membrane lipid composition, such as local production of diacylglycerol (DAG), may contribute to coat-less vesicle formation (Bard & Malhotra, 2006; Helling et al., 2006). Recent evidence implicates phosphatidylinositol-4-phosphate (PtdIns4P), traditionally understood only as a precursor of PtdIns(4,5)P2, in membrane trafficking control. In plants, PtdIns4P makes up 80% of the total PtdInsP pool (see Meijer & Munnik, 2003), and was recently found in the Golgi and plasma membrane, but not in LEs (Vermeer et al., 2009).

PtdIns4P is produced from PtdIns by PtdIns4-kinases (PtdIns4K or PI4K). Two mammalian PtdIns4K isoforms, PI4KIIα and PI4KIIIβ, the latter interacting with Arf1, localize to the Golgi apparatus, together with PtdIns4P-binding proteins (Wang et al., 2003). In metazoans, PtdIns4P and DAG work in concert in TPC formation (Bard & Malhotra, 2006). The conical shape of DAG favours membrane bilayer curvature, contributing to vesicle initiation (Szule et al., 2002; Fig. 2).

Figure 2.

Role of lipid metabolism in exocytotic vesicle formation. (a) Hypothetical mechanism of trans-Golgi network (TGN)-to-plasma membrane carrier (TPC) formation, based on the mammalian model. Differences in the molecular shapes between phosphatidylinositol-4-phosphate (PtdIns4P) (formed by RabA4b-recruited type III phosphatidylinositol-4-kinases) and diacylglycerol (DAG) [formed from phosphatidic acid (PA), produced by phospholipase D (PLD)], and their phase separation within the plane of the membrane, aided by Sec14-related proteins and flippases, supports curvature formation and vesicle initiation. Dynamin might perform final scission. (b) Inhibition of PLD by n-butanol inhibits TPC production in tobacco pollen tube tips (electron micrograph by Martin Potocký, Mieke Wolters-Arts and Jan Derksen, Radbout University, Nijmegen, the Netherlands). Bars, 2 µm.

The only two known Arabidopsis PtdIns4-kinases, AtPI4Kβ1 and AtPI4Kβ2, belong to class IIIβ. Both interact with RabA4b, a GTPase controlling post-Golgi to plasmalemma trafficking in root hair tips. A double mutant lacking both genes exhibits reduced TPC formation, resulting in enlarged vacuoles and aberrant root hair growth (Preuss et al., 2006). Secretory proteins accumulate in the Golgi apparatus in cells expressing mutant rice RabGAP; this effect is relieved by RabA but not RabE overproduction. RabGAP thus may facilitate trafficking from the TGN both to the plasmalemma and the vacuole by increasing RabA recycling (Heo et al., 2005). Efficient PtdIns4K recruitment to the Golgi may require this recycling. An Arabidopsis root hair mutant (rhd4) exhibiting enlarged Golgi cisternae and major polarity defects carries a mutation in a PtdIns4P phosphatase, again suggesting an essential role for PtdIns4P turnover in Golgi organization and TPC formation (Thole et al., 2008). Thus, PtdInsP metabolism at the TGN may have a prominent role in plant TPC formation, with RabA4b serving as a major organizer of the vesicle formation domain. Related mechanisms may participate in the formation of endosome-derived exocytotic vesicles.

Loss of COW1 and SFH1, Arabidopsis homologues of the yeast phosphatidyl transfer protein Sec14p, disrupts root hair elongation, with disorganization of tip-directed PtdIns(4,5)P2, perturbation of the cytoskeleton and dispersal of secretory vesicles from the tip cytoplasm (Böhme et al., 2004; Vincent et al., 2005). The mammalian Sec14 homologue Nir2 participates in secretory vesicle formation via the regulation of DAG metabolism at the Golgi apparatus (Litvak et al., 2005, Bard & Malhotra, 2006; Fig. 2).

Phosphatidic acid (PA), produced by phospholipase D (PLD), may be another regulator of plant TPC formation – directly or as a precursor for DAG production by PA phosphatase. PA stimulates pollen tube growth, whereas PLD inhibition by n-butanol inhibits it, causing a loss of vesicles from the clear zone (Potockýet al., 2003; Fig. 2). Butanol also induces the release of the GTPase Arf1 from Golgi membranes in tissue culture cells (Langhans & Robinson, 2007). 4,4-Difluoro-5,7-dimethyl-4-bora-3a,4a-diaza-S-indacene (BODIPY® FL)-labelled PA localized, on prolonged exposure, into Golgi bodies in pollen tubes (Potockýet al., 2003). Arabidopsis mammalian-like PLD isoforms participate in the cycling of the PIN2 auxin transporter (Li & Xue, 2007), vesicle-mediated auxin transport (Mancuso et al., 2007) and root hair development (Ohashi et al., 2003).

The translocation of phospholipids between sheets of the membrane bilayer by lipid flippases may participate in vesicle formation, as documented for the yeast Drs2p enzyme (D. Liu et al., 2008). A functionally related Arabidopsis P4-ATPase/flippase, ALA3, localizes to the Golgi apparatus and participates in secretory vesicle formation (Poulsen et al., 2008; Fig. 2).

In metazoans, vesiculation of the whole Golgi complex takes place during mitosis; it can also be induced by ilimaquinone (Bard & Malhotra, 2006). This process, as well as TPC formation, requires trimeric G-proteins and a serine/threonine protein kinase D (PKD) interacting with DAG membrane domains (Bard & Malhotra, 2006). However, plant cells retain the Golgi apparatus throughout their cell cycle and apparently lack this mechanism, as we found no PKD homologues in plant sequence searches. Other lipid-modifying activities might contribute to plant TPC formation.

Membrane tube pulling by membrane-anchored microtubule motor kinesin results in the tube breaking into vesicles (Roux et al., 2006; ‘pull and cut’ model of Bard & Malhotra, 2006). The presence of dynamin and GTP was sufficient to achieve fragmentation of artificial tubular membrane protrusions into 60–80 nm vesicles (Bashkirov et al., 2008). Plant dynamin-related proteins (DRPs), known to participate in cytokinesis and endocytosis, might assist exocytotic vesicle formation (Kang et al., 2003).

It is possible that plant TPCs might also form without the assistance of coat proteins, consistent with the default plasmalemma destination of the bulk of exocytotic cargos. Sorting, however, appears to act on some plasmalemma proteins at the TGN (Bard & Malhotra, 2006). Exocytosis of the yeast chitin synthase Chs3p involves a nonconventional coat (exomer), an activated Arf1 GTPase and acidic phospholipids (Wang et al., 2006). The discovery of nonconventional coats thus cannot be excluded also for some plant TPCs.

If exocytotic vesicle-generating compartments are modelled by DRPs and actin, consortia of nonseparated vesicles connected by short tubular necks might participate in exocytosis. In vivo observations of larger exocytotic entities at growing pollen tube tips using video-enhanced microscopy indeed suggest that plant TPCs may include not only isolated vesicles but also more complex structures (Bard & Malhotra, 2006; Ovecka et al. 2008; M. Potockýet al., unpublished).

IV. Building the tracks for exocytotic vesicles: cytoskeleton in exocytosis and cell polarization

Much of our understanding of cytoskeleton involvement in exocytosis comes from studies on tip-growing cells – root hairs or pollen tubes. In these cells, actin filaments form a fine dynamic network in the extending tip, merging into thicker cables along the shank of the hair or tube. As the cell ceases to grow, cables replace the fine meshwork, suggesting a causal role of fine actin arrays in tip growth (Baluška et al., 2000; Ketelaar et al., 2003; Lovy-Wheeler et al., 2005), and the perturbation of actin dynamics often results in characteristic phenotypes (see Ren & Xiang, 2007; Cheung & Wu, 2008).

The cytoskeleton participates in long-distance organelle and vesicle movement. In opisthokonts, motor proteins aid Rab-assisted loading of vesicles onto microtubule or actin tracks (Deneka et al., 2003). Similar mechanisms probably operate also in plants. Arabidopsis class XI myosins participate in organelle trafficking, with a specific myosin function related to vesicle transport in root hair elongation (Ojangu et al., 2007; Prokhnevsky et al., 2008). A myosin-independent ‘comet’ mechanism based on F-actin polymerization might also participate in secretory vesicle transport; in plants, there is circumstantial evidence for this mechanism in endosome movement in growing root hair tips (Voigt et al., 2005). Dynamin-mediated scission of mammalian endocytotic vesicles requires actin polymerization and fine actin filaments may also coat exo- and endocytotic vesicles (Tsujita et al., 2006). Myosin-dependent transport of membrane organelles in vivo requires continuous dynamic turnover of F-actin (Semenova et al., 2008).

Microtubules are not essential for tip growth in budding yeast or filamentous fungi; however, in fission yeast, they determine the direction of tip growth via the delivery of ‘landmark’ complexes. Some of the responsible proteins have homologues in plants (Sieberer et al., 2005). In addition, in plant tip-growing cells, microtubules control growth direction rather than growth itself (Bibikova et al., 1999); microtubule disruption by oryzalin induces wavy root hair or pollen tube growth (see Sieberer et al., 2005; Gossot & Geitmann, 2007). Tubulin depletion causes ectopic root hair formation and branching (Bao et al., 2001).

A reciprocal relationship between mechanical stress and microtubule organization, long known in plants (for example, Lintilhac, 1984), has been documented recently in fission yeast, where microtubules induce actin reorganization on mechanical bending (Minc et al., 2009). Mechanical induction of new secretory domains and RDs (see below) may be a first step towards organ initiation, as demonstrated by new auxin maxima and lateral root initiation in roots subjected to bending (Ditengou et al., 2008).

In diffusely growing plant cells, cortical microtubules delimit expanding areas (see Wasteneys, 2004). Microtubules also determine the position of secretory domains in nonexpanding cells, such as Arabidopsis pectin-depositing seed coat cells, where a dense microtubule meshwork marks domains for pectin insertion (McFarlane et al., 2008). In addition, de novo localization of PIN1 at the cross-walls of root cortex cells after cytokinesis requires microtubules, whereas actin participates in the recycling of PIN1 and in the localization of AUX1 (Geldner et al., 2003; Bouttéet al., 2006; Kleine-Vehn et al., 2006). Actin bundling elicited by a mouse talin-derived fusion protein in cultured tobacco cells produced a phenotype suggesting the perturbation of auxin transport (Maisch & Nick, 2007), and some auxin transport inhibitors perturb actin (Dhonukshe et al., 2008a). Interestingly, at least two Arabidopsis actin organizers from the formin family localize to cross-walls of the root cortex (Deeks et al., 2005).

Local balance between G-actin, fine F-actin arrays and F-actin cables may control exocytosis, with cortical actin acting as a barrier preventing secretory vesicles from reaching the plasmalemma and thus delimiting the secretory domain (as proposed for metazoan cells by Valentijn et al., 1999), whereas microtubules may provide a vesicle-permeable framework at the sites of exocytosis or endocytosis in plant cells (Karahara et al., 2009; McFarlane et al., 2008; Fig. 3). The concerted action of Rop GTPases (and effectors), microfilaments and microtubules defining plant epidermal cell lobes (Fu et al., 2005) is consistent with this model, which might represent an evolutionarily conserved polarity regulation module.

Figure 3.

The cytoskeleton participates in the delimitation of exocytotically (and endocytotically) active plasmalemma domains (light blue). While the cortical microtubular network (blue) is permeable for secretory vesicles and may even direct them to the sites of exocytosis, the fine actin mesh (red) acts as a barrier to exocytosis.

V. Manysidedness of a plant cell: delimiting functional plasmalemma domains within a cell

A mere look into a microscopic section of plant tissue suggests the presence of multiple exocytotic domains in a single cell. For example, a leaf epidermal cell has a ‘distal’ plasmalemma domain oriented towards the environment and a ‘proximal’ domain facing palisade parenchyma cells. Tip-growing cells exhibit dramatic surface diversity, with a pectinaceous wall at the growing tip and a callose-rich subapical nongrowing domain. A cell attacked by a fungal pathogen re-polarizes to form a callose, lignin and suberin-rich cell wall papilla adjacent to the invading appressorium (Schmelzer, 2002).

Local differences in cell wall and plasmalemma protein composition have been well documented by immunodetection techniques (see, for example, Knox, 2008). Examples include the polarized localization of PIN and AUX1 carriers, glycosylphosphatidylinositol (GPI)-anchored protein COBRA and cuticle protein BODYGUARD (see, for example, Schindelman et al., 2001; Kurdyukov et al., 2006; Wisniewska et al., 2006). We use the term ‘activated cortical domain’ (ACD) to denote plasmalemma domains poised to or actually performing exocytosis (and endocytosis), not necessarily related to cell growth (Fig. 4).

Figure 4.

Schematic view of activated cortical domain (ACD) establishment. The activation of Rop guanosine triphosphatase (GTPases), for example, by receptor-like serine/threonine kinases (RLKs) via a plant-specific Rop nucleotide exchanger protein (PRONE-GEF) is central to the assembly of an ACD. Rop controls the local activity of lipid-modifying enzymes [which generate a phosphatidylinositol-4,5-bisphosphate (PtdIns(4,5)P2)-enriched membrane domain that can be recognized by the Exo70 exocyst subunit], mediates the assembly of the exocyst via the effector Icr1 interacting with Sec3, and controls cytoskeletal organization. Inhibition of Rop, for example, by GTPase-activating proteins (GAPs) or GDP dissociation inhibitors (GDIs, GEFs), and lipid modification by phospholipase C (PLC), restricts lateral spreading of the secretory domain. Activated Rops and RLKs are sequestered into lipid rafts, thus promoting signalling interactions. Expansin- and xyloglucan endotransglycosylase (XET)-mediated cell wall loosening takes place adjacent to the ACD. PM, plasma membrane.

1. Rop guanosine triphosphatases and related signalling pathways in the definition of the activated cortical domain

Local activation of Rho family GTPases (Rac, Rho or Cdc42 in opisthokonts, Rop in plants), organizing both the cytoskeleton and the secretory pathway, is central to eukaryotic cell polarization; unfortunately, we know little about plant Rop effectors and regulators (see Ridley, 2006; Kost, 2008; Yalovsky et al., 2008). Receptor-like serine/threonine kinases (RLKs), a large family of versatile signalling proteins whose localization, at least in some cases, requires Rop-regulated exocytosis (Lee et al., 2008), and a novel class of Rop GEFs (PRONE-GEF; Berken et al., 2005), which are activated by RLKs (Zhang & McCormick, 2007), are prime candidates for local regulators of Rop activity and ACD formation. The multiplicity of angiosperm RLKs may provide a means for the formation of diverse ACDs during development or in response to external signals, including interactions with pathogens. The recently discovered adaptor protein ICR1, which interacts with GTP-charged Rop and is able to recruit the exocyst subunit Sec3 (see below), is important for ACD initiation (Lavy et al., 2007); the same protein is recruited into pollen tube tips by a Rop-dependent mechanism (S. Li et al., 2008; Fig. 4).

The overexpression of Rops in tip-growing cells often causes tip depolarization (for example, Fu et al., 2001; Cheung et al., 2003). Wild-type, constitutively active and dominant-negative Rop mutants all elicit tip swelling to a varying extent, and some Rops inhibit tip growth when overexpressed (Kost, 2008; Yalovsky et al., 2008). Microinjected nonhydrolysable analogues of GTP or GDP cause swelling in pollen tubes (Eliáš et al., 2001). This suggests either a requirement for Rop cycling between the GTP and GDP-bound states, or involvement of multiple pathways (perhaps employing different Rops or Rop effectors) in the maintenance of tip-focused growth.

Negative regulators of Rop may control the spread of ACDs. Mutation of the Arabidopsis Rop-associated GDP dissociation inhibitor (GDI) SCN1 results in multiple, short, swollen root hairs per trichoblast (Carol et al., 2005), and the loss of either GDI or a GTPase-activating protein (GAP) causes pollen tube swelling (Klahre & Kost, 2006; Klahre et al., 2006).

Metazoan Rhos induce actin cables in stress fibres, filopodia and lamellipodia, partly by direct binding to formins (see Ridley, 2006). However, angiosperm formins contain no known GTPase-interacting motifs, although moss and lycophyte formins possess a candidate Rho-binding domain (Grunt et al., 2008). Thus, plant Rops may control actin indirectly or via other effectors, such as SCAR/WAVE proteins regulating Arp2/3-mediated actin nucleation (see Deeks & Hussey, 2005; Mathur, 2005) or a PtdIns-P kinase (Kost et al., 1999, see below). The first known Rop effectors were the small CRIB domain-containing RIC proteins (Wu et al., 2001). Arabidopsis RIC3 and RIC4 are responsible for ‘translating’ oscillations in apical [Ca2+] into periodical growth rate changes, with the accumulation of secretory vesicles on RIC4-driven F-actin assembly and their simultaneous discharge on RIC3-induced calcium-regulated actin disassembly (Gu et al., 2005; Lee et al., 2008).

The Rop/RIC module is also involved in diffuse cell expansion (Panteris & Galatis, 2005). AtROP2 controls the development of lobed epidermal pavement cells through two complementary pathways, one stimulating actin assembly via RIC4 in outgrowing lobes and the other inducing RIC1-dependent microtubule reorganization in indentations of the adjacent cell, whilst locally inhibiting ROP2 (Fu et al., 2005). Antagonistically acting RICs may partly explain the diversity of Rop-associated phenotypes.

Local cell wall relaxation [involving expansins or xyloglucan endotransglycosylases (XETs)] might serve as an example of mechanical stress inducing new ACDs, resulting in the initiation of root hair bulges on rhizodermal trichoblasts or ectopic leaf initiation at the apical meristem (Reinhardt et al., 1998; Baluška et al., 2000; Vissenberg et al., 2001).

2. The role of membrane lipid composition and modifications

Eukaryotic plasma membranes exhibit lateral heterogeneity based on cholesterol-enriched lipid microdomains (detergent-resistant membranes or rafts) organized by membrane proteins (Salaun et al., 2004; Grossmann et al., 2007). Arabidopsis sterol mutants indicate the involvement of membrane sterols in endocytosis, exocytosis and cell polarization (Souter et al., 2002; Willemsen et al., 2003; Men et al., 2008). GTP-bound type I Rops are sequestered into detergent-resistant membranes as a result of hydrophobic modification by reversible covalent binding of palmitic or stearic acids in addition to standard C-terminal geranylation (Sorek et al., 2007; Yalovsky et al., 2008).

GPI-anchored proteins, including plant cell wall arabinogalactan proteins (AGPs, Morel et al., 2006), are characteristic inhabitants of lipid microdomains. AGPs might be responsible for the lateral differentiation of plant plasmalemma, partly via the anchoring of detergent-resistant microdomains to the cell wall. Indeed, some plant membrane microdomains are nonmotile, as shown for the KAT1/SYP121 markers (Sutter et al., 2007; Grefen & Blatt, 2008). This may contribute to plasmalemma domain specification, as illustrated by the distinct localization of the COBRA GPI protein to the lateral cell wall domain, required for correct cell expansion (Schindelman et al., 2001). However, not all plant GPIs are polarized; the Arabidopsis SKU5 protein covers more or less evenly root cell surfaces, yet mutants suffer from root polarity defects (Sedbrook et al., 2002). Thus, even the plasmalemma of a single cell may harbour diverse membrane microdomains which may combine to form distinct, but possibly overlapping, plasmalemma sectors or ACDs with different recycling kinetics (documented for yeast by Grossmann et al., 2007). As Rops, RLKs, SNAREs, ion transport channels, GPIs and other proteins involved in exocytosis and polarity signalling are all enriched in plant detergent-resistant membranes (Morel et al., 2006), these membrane microdomains may facilitate crowding of cell polarity regulators and enhance their mutual interactions (Yalovsky et al., 2008; Fig. 4).

Phosphoinositides (PPIs) also act as local membrane domain organizers and regulators. PPI metabolism is localized, with various kinases and phosphatases active at distinct compartments (see Thole & Nielsen, 2008). PPIs bind to specific protein motifs (for example, PH, PX, FYVE, etc.) and additional binding sites are often required to engage membrane-resident proteins (Lemmon, 2008). In tip-growing plant cells, PtdIns(4,5)P2 is localized to the apex and spreads throughout the plasma membrane on growth cessation (for example, Kost et al., 1999; Vincent et al., 2005; Ischebeck et al., 2008; Kusano et al., 2008; Stenzel et al., 2008). PtdIns4,5-kinase isoform 3 (PtdIns4P-5K3), a key enzyme producing PtdIns(4,5)P2, is preferentially expressed in root hairs, and localized to the periphery of their apex, possibly associated with the plasmalemma or exocytotic vesicles. Mutants exhibit reduced root hair growth, whereas overexpression leads to multiple hairs per trichoblast and the loss of cell polarity, i.e. disturbances of ACD regulation (Kusano et al., 2008; Stenzel et al., 2008). Analogous observations were made for PtdIns4P-5K4/5 in pollen, whose loss inhibits pollen germination and pollen tube growth (Ischebeck et al., 2008; Sousa et al., 2008). In growing pollen tubes, both proteins localize to a distinct tip plasmalemma domain, overlapping with the distribution of PtdIns(4,5)P2; overexpression induces tip branching and large plasma membrane invaginations, again suggesting a direct role of PtdIns4P-5Ks in ACD regulation. Negative regulators are important to restrict PtdIns(4,5)P2 to a distinct plasmalemma domain. In pollen tubes, this is achieved by the localization of PtdIns(4,5)P2-consuming PLC to the border of the ACD just behind the growing tip (Dowd et al., 2006; Helling et al., 2006). Plants have multiple PtdIns-P kinases producing PtdIns(4,5)P2 (five isoforms are co-expressed in Arabidopsis pollen; Žárskýet al., 2006; Ischebeck et al., 2008). Mutant phenotypes point to a differential effect, implying the existence of different PtdIns(4,5)P2 pools (Ischebeck et al., 2008; Sousa et al., 2008). Multiple PtdIns(4,5)P2 pools with distinctive fatty acids in the same cell were observed on stress treatment (König et al., 2008), implying their possible participation in defining distinct ACDs (see Section VII).

So far, we have treated ACD almost as an isolated plasmalemma domain. However, the establishment and function of an exocytotically active ACD involves dynamic membrane recycling. In Section VII, we propose a general concept of a ‘recycling domain’ (RD) as a dynamic entity connecting the ACD with a specific subset of recycling compartments/endosomes.

VI. Vesicle tethering, docking and fusion

The ACD may be viewed as a docking platform for exocytotic vesicles. In opisthokonts, distinct phases of vesicle tethering and docking at the target membrane, preceding SNARE complex formation and vesicle fusion, have been described. Tethering links the vesicle at a distance of more than one-half of its diameter from the target membrane, whereas docking sensu stricto holds the two membranes within a bilayer's distance (< 5–10 nm). In plants, such resolution has not yet been achieved; therefore we shall refer to both tethering and docking as tethering sensu lato. The last phase of vesicle fusion has been exhaustively reviewed recently, focusing especially on SNARE function (for example, Lipka et al., 2007).

Vesicle tethering is an important targeting step, mediated either by long tethering proteins or by multisubunit tethering complexes (Cai et al., 2007). In opisthokonts, exocytotic vesicle tethering involves the octameric exocyst complex, originally discovered in yeast as effector of the Rab GTPase Sec4 (TerBush et al., 1996), and later found also in animals (for a review, see Hsu et al., 2004; Cai et al., 2007; Wu et al., 2008). The exocyst localizes exocytotic vesicles to specific plasmalemma domains (ACDs), facilitating early events of polarized secretion. We have recently described the angiosperm exocyst complex using genetic, biochemical and cytological methods (Hála et al., 2008), confirming the presence of homologues of all eight subunits (Sec3, Sec5, Sec6, Sec8, Sec10, Sec15, Exo70, Exo84). Surprisingly, Exo70 forms an extremely large family of paralogues in land plants (Eliáš et al., 2003; Synek et al., 2006). T-DNA insertions into the Exo70A1 gene, encoding the most abundant Exo70 isoform among the 23 Arabidopsis paralogues, resulted in a discernible phenotype with a semi-dwarf stature, reduced apical dominance, failure of polar growth of root hairs and stigmatic papillae, and delayed senescence and lateral root initiation (Synek et al., 2006). Mutants in SEC5, SEC6, SEC8 and SEC15a show defective pollen germination and pollen tube growth, and Sec6, Sec8 and Exo70A1 colocalize at growing tobacco pollen tube tips (Cole et al., 2005; Hála et al., 2008). A maize mutant (rth1) in the Sec3 subunit was discovered independently in a forward screen for root hair defects (Wen et al., 2005). All of these observations strongly implicate the plant exocyst in cell polarity and morphogenesis regulation.

Although the mechanism of exocyst action remains somewhat unclear (for example, Wu et al., 2008; Songer & Munson, 2009), the Exo70 and Sec3 subunits obviously provide a crucial activated Rho-dependent landmark for assembling the rest of the complex on the plasmalemma (Boyd et al., 2004). The complete exocyst probably assembles on arrival of the remaining subunits carried by vesicles to the target membrane marked by the Exo70/Sec3 landmark, which is delivered to the ACD independently from other exocyst subunits and possibly also from actin (Novick et al., 2006). However, an alternative model has been proposed, involving the local activation of preassembled exocyst by the Rho GTPase (Wu et al., 2008). In mammals, insulin-induced delivery of the GLUT4 transporter to the plasmalemma depends on binding between Exo70 and Rho GTPase TC10 (Inoue et al., 2003).

PtdIns(4,5)P2 also participates in vesicle tethering, in particular in attaching Exo70 and Sec3 to the plasmalemma. Binding of Exo70 to PtdIns(4,5)P2 is confined to a motif at the C-terminus of the rod-like molecule (He et al., 2007b), conserved in many, but not all, Arabidopsis Exo70 paralogues (Fig. 5). In addition, yeast Sec3 binds to PtdIns(4,5)P2 in the membrane (Zhang et al., 2008), but its PPI-binding site is located in the nonconserved N-terminal part of the protein and may thus represent a lineage-specific feature. In yeast, genetic evidence suggests that Sec8p, Sec10p and Sec15p exocyst subunits and the Sec9p plasmalemma t-SNARE are regulated by PtdIns(4,5)P2 (Routt et al., 2005). Although, in opisthokonts, activated Rho interacts directly with exocyst subunits (Guo et al., 2001; Wu et al., 2008), in Arabidopsis, the Rop–Sec3 interaction appears to be mediated by the adaptor protein ICR1 (see above, Lavy et al., 2007). We have observed interaction between Arabidopsis Sec3 and Exo70A1 subunits (Hála et al., 2008), which, together with the predicted ability of Exo70A1 (and some other Exo70 paralogues) to bind membrane phospholipids, suggests that plant Exo70s might serve as landmarks for exocyst vesicle targeting, coordinated with local Rop activity via Sec3–ICR1 interactions (Fig. 4).

Figure 5.

Plant Exo70 isoforms harbour C-terminal conserved PtdIns4,5P2 binding (red) and Arp2/3 complex binding (blue) sites. (a) Structural model produced by threading of the major housekeeping Arabidopsis isoform, AtExo70A1, on the empirically determined structure of mouse Exo70 (PDB:2pft) shows that the binding sites nearly overlap. (b) PROSITE pattern representation of the same binding motifs in selected yeast and metazoan Exo70 proteins and all Arabidopsis Exo70 isoforms reveals a diversity of presumed binding abilities among the plant paralogues (GenBank accessions: yeast, ScExo70, NP_012450.1; rat, RnExo70, NP_073182.1; Drosophila, DmExo70, NP_648222.3; Arabidopsis– AtExo70A1, NP_001119162.1; AtExo70A2, NP_200047.3; AtExo70A3, NP_200048.2; AtExo70B1, NP_200651.1; AtExo70B2, NP_172181.1; AtExo70C1, NP_196819.1; AtExo70C2, NP_196903.1; AtExo70D1, NP_177391.1; AtExo70D2, NP_175811.1; AtExo70D3, NP_566477.2; AtExo70E1, NP_189586.1; AtExo70E, NP_200909.1; AtExo70F1, NP_199849.2; AtExo70G1, NP_194882.2; AtExo70G2, NP_175575.1; AtExo70H1, NP_191075.2; AtExo70H2, NP_181470.1; AtExo70H3, NP_187564.1; AtExo70H4, NP_187563.1; AtExo70H5, NP_180432.2; AtExo70H6, NP_683286.2; AtExo70H7, NP_200781.1; AtExo70H8, NP_180433.1).

In multicellular Characean algae, or in the first land plants, genes encoding Exo70 underwent a series of duplications, generating three major Exo70 families found in all land plants beginning with mosses (Eliáš et al., 2003, Synek et al., 2006). Multiple Exo70 isoforms within the same cell may contribute to specific landmarks, and thus to the initiation and maintenance of specific ACDs and RDs (Fig. 6).

Figure 6.

Hypotheses on possible primary functional diversification of land plant Exo70 paralogues. The three assumed Exo70 paralogues in ancestral plants may have divided their function either among three spatial directions (or planes of division) within the multicellular body – basal–apical, radial and tangential (a), or between three kinds of cell polarity phenomena – tip growth, nonisodiametric cell expansion and protein deposition, and cytokinesis (b). Distinct Exo70 paralogues may be responsible for maintaining multiple recycling domains per cell (c).

Yeast Exo70 participates in the regulation of actin dynamics by interaction with the Arp2/3 nucleation complex via a motif almost overlapping with the PtdIns4,5P2-binding site (Zuo et al., 2006). Among Arabidopsis Exo70 paralogues, only Exo70A1 exhibits a perfect match to the mammalian Arp2/3-binding motif (Fig. 5). The diversity of C-terminal-binding motifs in plant Exo70 proteins suggests diverse protein or lipid interactions (Fig. 5).

As the exocyst is currently the only plant candidate for vesicle tethering at the plasmalemma, we suggest that Exo70 paralogues may, apart from cell type-specific functions, create a variety of exocyst complexes within the same cell, poised to specific ACDs or endomembrane destinations (Fig. 6). Together with other factors, such as a variety of GTPases, lipids, formins or SNAREs, this may be central for the establishment of the dynamic manysidedness of the plant cell.

In opisthokonts, vesicle tethering, followed by activation and formation of the SNARE complex, involves a Rab GTPase cascade. The yeast exocytotic Rab GTPase Sec4 is activated by GDP/GTP exchange, aided by the Sec2 GEF that is recruited by the upstream Golgi-based Rab GTPase Ypt31/32 (Novick et al., 2006). Active Sec4 recruits the exocyst by interacting with its Sec15 subunit, which also transiently binds Sec2 (Novick et al., 2006). Sec4 and Exo84 then attract Sro7/77, a WD40 family protein implicated in maintaining cell polarity, which, in turn, binds and regulates the t-SNARE Sec9p (Lehman et al., 1999). Arabidopsis has two putative Sro7 homologues, At5g07860 and At3g35560, both carrying N-terminal syntaxin-binding WD40 repeats and C-terminal R-SNARE domains, suggesting that they bind SNARE proteins. It is not known yet whether they also interact with Rab GTPases. The yeast Exo84 exocyst subunit binds to Sec1, whose interaction with the SNARE complex initiates membrane fusion (Novick et al., 2006). The Arabidopsis Sec1 homologue KEULE is involved in cytokinesis (Assaad et al., 2001). Exocyst-like structures in plant cytokinesis have been observed (Otegui & Staehelin, 2004; Seguí-Simmaro et al., 2004), and phenotypes of Arabidopsis exo70a1 mutants also implicate the exocyst in meristem cell division regulation (Synek et al., 2006).

Lipids participate in both vesicle fission (see above) and fusion. PA may promote membrane fusion by influencing membrane topology. Yeast PLD, encoded by Spo14p, is required for prospore membrane formation, and PA controls the localization of the SNAP25-like t-SNARE Spo20p during sporulation (Liu et al., 2007). PA production by Spo14p is also activated by the nonclassical PtdIns transfer protein Sfh5p, which, in turn, promotes PtdIns(4,5)P2 production at the plasmalemma (Routt et al., 2005). Moreover, in animal cells, PA stimulates membrane fusion governed by the t-SNARE complexes syntaxin 1A/SNAP25 and syntaxin 4/SNAP23 (Vicogne et al., 2006; Lam et al., 2008). PLD activation by the small GTPase ARF6 stimulates PtdIns(4,5)P2 production and exocytosis (Bégléet al., 2009).

In opisthokonts, PtdIns(4,5)P2 is crucial for both ATP-dependent vesicle priming and Ca2+-dependent fusion during regulated exocytosis. This involves the binding of PtdIns(4,5)P2 to several proteins, including SCAMP2, SNAP25/syntaxin 1A, exophilin 4 and synaptotagmins, which all have homologues in plants. Arabidopsis AtSYT1, a plasmalemma-localized synaptotagmin homologue, is indispensable for maintaining plasmalemma integrity under salt or freezing stress, presumably through exocytotic membrane resealing (Schapire et al., 2008). The role of [Ca2+]i gradients in plant exocytosis, and especially in tip growth, has been the subject of many reviews (for example, Hepler, 2005). As in metazoan neurones, in plant cells, calcium has been suggested to regulate the switch from constitutive to regulated exocytosis (see Homann, 2006). Calcium-dependent candidates for plant exocytosis regulators include calmodulin, calcium-dependent protein kinases (CDPKs), NADPH oxidases (NOX), actin-binding proteins (see Cole & Fowler, 2006), and annexins and synaptotagmins (Mortimer et al., 2008, Schapire et al., 2008). Dynamic proton transport loops at the tip of pollen tubes may play a prominent regulatory role upstream of calcium (Certal et al., 2008).

In tip-growing cells, a tip-focused calcium gradient may be maintained by a positive feedback loop involving calcium operating at the activated exocytotic domain: the Rop-activated NOX catalyses reactive oxygen species (ROS) production, which then stimulates Ca2+ influx carriers, bringing about calcium-induced NOX activity (Carol et al., 2005; Žárskýet al., 2006; Potockýet al., 2007; Takeda et al., 2008).

Membrane capacitance measurements have shown that calcium-independent vesicle fusion also takes place at the plant plasmalemma (Homann, 2006). Different dependences on calcium for vesicle fusion may thus contribute to the diversity of secretory pathways within a single plant cell.

VII. Polarity by recycling: exocytosis, endocytosis and recycling domains in plant cells

In opisthokonts, secretion restricted to specific plasmalemma domains is characteristic, for example, of yeast buds or the apical and basolateral membranes in animal epithelia (Mostov et al., 2003). Such polarized exocytosis has also been well documented in plants. New cell wall components are inserted into specific cortical domains during cell expansion and secondary cell wall deposition and modification, as exemplified by differential cell wall thickening in colenchyma, xylem maturation, Caspari band formation or pectin deposition in seed coat volcano cells. Cytokinesis and tip growth can be viewed as extreme examples of direct polarized cargo delivery as well, although they obviously also involve membrane recycling.

Most (but not all) plasmalemma constituents are continuously recycled even in fully differentiated plant cells. Fast and slow recycling takes place in plant cells, for example, via ‘kiss-and-run’ vesicles with a half-time of approximately 1 s and via clathrin-coated vesicles lasting c. 50 s (Homann, 2006). In metazoan synapses, an increase in [Ca2+]i correlates with the switch between fast and slow recycling; this might be similar in plant cells.

Dynamic recycling of plant membrane proteins has received attention after the discovery of the constitutive recycling of PIN auxin efflux carriers, which is directly regulated by auxin itself (Geldner et al., 2003; Paciorek et al., 2005). Meckel et al. (2004) observed constitutive internalization and recycling of vesicles carrying the KAT1 potassium channel, with kinetics quite different from PIN recycling, whereas a heterologously expressed GFP-tagged human membrane protein was not recycled, pointing to lateral differentiation of the plasmalemma with respect to selective endocytotic retrieval of specific membrane proteins (Homann et al., 2007). Abscisic acid (ABA) triggers specific KAT1 internalization with slow recycling back to the plasma membrane (Sutter et al., 2007). The effects of auxin and ABA on the kinetics of endocytosis suggest that endocytosis is a regulated step in plant RD dynamics; it is expected that recycling will also be regulated at the level of exocytosis (see below).

The polarization of plasmalemma proteins is intimately linked to endocytotic membrane recycling, as shown by Valdez-Taubas & Pelham (2003), who recognized two basic mechanisms of cell polarity. Different plasmalemma domains may be laterally separated by diffusion barriers, such as, for example, the septin ring at the neck between bud and mother cell in yeast. Alternatively, dynamic polarization may be based on slow lateral membrane protein diffusion (in cells with a cell wall) and local membrane recycling via an endocytosis/exocytosis cycle (Fig. 7). Even in yeast, the second mechanism is crucial. Yeast SNARE Snc1 is kinetically polarized by endocytosis, as the mutation of an endocytotic signal results in its even distribution on the cell surface, whereas the mere addition of an endocytosis sorting motif to the Sso1 SNARE results in its polarization (Valdez-Taubas & Pelham, 2003). Studies on pollen tubes, root hairs, the KAT1 channel and auxin carrier recycling show that this type of plasmalemma polarization is also dominant in plant cells (Paciorek et al., 2005; Cole & Fowler, 2006; Sutter et al., 2007; Kost, 2008; Yalovsky et al., 2008).

Figure 7.

Cell polarization by recycling. (a) A membrane protein deposited in a polar manner stays polar if it is either prevented from lateral diffusion by a barrier (possibly anchored to a stable structure, such as the cell wall) or removed by endocytosis [and possibly recycled back to the activated cortical domain (ACD)]. (b) A plant cell may comprise multiple recycling domains (RDs), whose topology is also influenced by cytoplasmic streaming, the position of the nucleus and the vacuole. (c) In late interphase, the preprophase band of microtubules (blue) initiates an ACD of local maximum of endocytosis at the plasmalemma. This dominant endocytosis belt might establish two new ‘global’ RDs comprising future daughter cells.

The establishment and maintenance of different plasmalemma domains of the same plant cell are largely dependent on kinetic polarization via recycling. This also implies the multiplicity of RE compartments – a feature already observed for some membrane proteins related to auxin transport (see Section II). ACDs at the plasmalemma and related recycling compartment are inseparably dependent on each others’ dynamics, thus establishing a dynamic superstructure which we propose to call the ‘recycling domain’ (RD).

The RD is a dynamic spatiotemporal entity based on the continuous reciprocal exchange of specific membrane constituents between distinct ACDs of the plasmalemma and specific compartments of the endomembrane system. A typical manysided plant cell embedded in a multicellular tissue has multiple RDs.

The best studied examples of plant RDs are those involved in the dynamic polarization of auxin carriers. Tip-growing cells provide an extreme example of a highly polarized RD (including cytoskeletal structures), in fungi described as the tip growth apparatus (TGA; Heath & Geitmann, 2000). We prefer to use this term also for the polarity machinery recently described as LENS (Cole & Fowler, 2006) in pollen tubes or root hairs to acknowledge convergent similarities between fungal and plant tip growth.

RD equilibrium is sensitive to subtle changes in the exocytosis/endocytosis ratio, affected by diverse signalling pathways. The inhibition of endocytosis by activated Rop (Bloch et al., 2005) and auxin (Paciorek et al., 2005) has been described. In addition, in metazoans, activated Rho GTPases inhibit endocytosis (see Ridley, 2006). The effects of auxin and Rop on endocytosis might be linked through the activation of Rop by auxin via the RLKs-PRONE GEF pathway (Žárský & Fowler, 2009). Opposite to auxin, ABA stimulates endocytosis of the KAT1 channel quite selectively (Sutter et al., 2007); membrane protein fluxes within RDs thus might be regulated mainly at the level of endocytosis.

There may be major differences between regulatory networks governing the dynamics of RDs in growing vs nongrowing cells. As a result of the lateral differentiation of the plasmalemma, linked to lipid microdomain formation, RDs with very different kinetics might partly topologically overlap, using alternative endocytotic and recycling routes.

A good example of endocytosis-dependent plant plasmalemma protein polarization has been published recently by Dhonukshe et al. (2008b), who demonstrated that newly synthesized PIN1 is first localized all over the plasmalemma in Arabidopsis root stele cells, and only subsequently polarizes via recycling through a late endosomal compartment. This observation pinpoints the important distinction between RD establishment and RD maintenance; in this case, the former requires endocytosis and MVB/LE-dependent transcytosis, but as soon as polar localization is established, it is maintained via GNOM RE recycling (Geldner et al., 2003; Dhonukshe et al., 2008b; Kleine-Vehn et al., 2008).

Continuous endosome (and actin)-dependent recycling, maintaining the polar localization of PIN (and partly also AUX1) auxin transporters, endows the system with sensitivity and adaptivity to external cues, explaining the almost immediate relocalization of PIN transporters on gravistimulation of roots and a concomitant shift in the ACD position (Ottenschläger et al., 2003). Polarization of the PIN2 auxin carrier is regulated by sterol-dependent endocytosis (Men et al., 2008), underpinning again RD-based recycling as the background of many polarization phenomena in plant cells, as well as the role of lipid rafts in RD establishment. The relocalization of PIN2 protein to the opposite cell domain on prolonged brefeldin A inhibition of RE function shows that the disruption of membrane recycling may result in reversible artificial RD formation (Kleine-Vehn et al., 2008).

New secretory domain and RD initiation in response to pathogen attack provides insights into plant ACD and RD establishment (Schmelzer, 2002). This process is possibly initiated by Rop GTPase activation at the attack site, with a subsequent increase in actin dynamics (Schütz et al., 2006). The nucleus moves towards the nascent RD, i.e. to a new pole of cell polarity (Schmelzer, 2002). Similarly, the nucleus also acts as an organizer in the regulation of polar cell expansion in root hairs, where optical trapping of the nucleus inhibits tip growth. In the root hair branching mutant cow1-2, the nucleus periodically migrates between the tips of the branched hair. Tip growth activity alternates between these two tips, dependent on the proximity of the nucleus; nuclear dynamics in this cell type are exclusively dependent on the actin cytoskeleton (Ketelaar et al., 2002). Dynamic nuclear positioning within the cell thus participates in RD establishment – as is most apparent in preprophase band and fragmoplast-related RDs in cytokinesis (Van Damme et al., 2007). The preprophase band of microtubules directed by nuclear position has been shown to initiate new ACD in late interphase with a distinct maximum of endocytosis (Dhonukshe et al., 2006; Karahara et al., 2009). We propose that this local cytoplasmic membrane belt of endocytosis maximum might be dynamically regulated by membrane recycling via exocytosis in two halves of the cell delineated by the nuclear position, resulting in the formation of two global RDs encompassing future daughter cells (Fig. 7c). Gene expression regulation is a major source of ‘nuclear control’ above cell polarity, reaching, however, beyond the scope of our review. In any case, nuclear migration should result in partial polarization of de novo-synthesized mRNAs.

The concept of multiple RDs in a single plant cell not only implies the presence of several subtypes of TGN/EEs and other endosomes (RE, LE), but, despite the known mobility of endosomes, also demands spatiotemporal ‘subcompartmentalization’ of endosomes within the cell. This could be at least partly organized by the position of the nucleus, the development of the vacuole and also by domains of cytoplasmic streaming. An active role of Golgi/TGN location in directed secretion and regulation of cell polarity has been documented recently in human HeLa cells under conditions specifically disrupting Golgi positioning without causing microtubule disassembly (Yadav et al., 2009). It should be expected that, in plant cells, the pace and direction of cytoplasmic streaming, carrying multiple Golgi bodies and especially ‘semi-autonomous’ TGNs/EEs/REs, will be equally important. Actomyosin-driven cytoplasmic streaming induces the hydrodynamic flow of small molecules; more elongated/differentiated cells exhibit significantly more rapid cytoplasmic streaming (Esseling-Ozdoba et al., 2008). New RDs are established in cells exiting meristem – during the initiation of elongation and differentiation (for example, in the root tip transition zone where apical and basal RDs are formed). During the cell growth phase, the organization and dynamics of cytoplasmic streaming may participate in the formation of new RDs or the turning down of existing RDs. On exit from the elongation zone and terminal cell differentiation, RDs may become more stable.

The exocyst participates in plasmalemma recycling, as it also localizes to RE via the GTPases Rab11/RabA, and regulates animal cell polarity-related recycling (for example, Langevin et al., 2005). Our current data also suggest similar relationships for the plant exocyst (H. Toupalováet al., unpublished). The multiplicity of plant RabA GTPases (Rutherford & Moore, 2002) and Exo70 subunits may substantially contribute to the establishment and maintenance of multiple RDs in a manysided plant cell, through both recycling and the recognition of diverse targeting platforms (ACDs). The single yeast Exo70 controls the secretion of specific exocytotic vesicles (Bgl2p, see above) at the early stages of bud growth, suggesting an important general principle that ‘different exocytotic pathways can be specifically regulated at the target membrane as well as at the TGN’ (He et al., 2007a). The regulation of diverse plant ACDs and exocytotic pathways may be generally based on reciprocity between specific plasmalemma ACDs and recycling endosomal compartments, as components of dynamic RDs (Fig. 8).

Figure 8.

Schematic representation of the recycling domain (RD) concept. RDs are only partly synchronous, as they are consecutively established (or switched off) during the differentiation and elongation phases of plant cell ontogenesis. Endosomal recycling often relies on the cooperation of several endosomal compartments (for simplification, only one is shown). The scheme of RD I includes reference boxes relating to the detailed figures above.

VIII. Conclusions: constitutive or regulated secretion in plant cells?

The eukaryotic secretory pathway is generally understood as two different processes – constitutive secretion and regulated secretion. Constitutive secretion is associated with ‘housekeeping’ cellular functions (for example, metazoan collagen secretion), whereas regulated secretion features in a situation in which the exocytosis of pre-existing stored vesicles is triggered by specific signal inputs. Surprisingly, only vesicles produced in this second pathway are commonly called ‘secretory vesicles’ (see Lodish et al., 2004, p. 703). In this context, the secretory pathways of yeast and plant cells are considered to be constitutive (see, for example, Fukuda, 2008). This zoocentric concept might be wrong, as it implies that, in yeast and plants, secretion is just a ‘boring’ default. This, obviously, is not the case – even in the secondarily simplified yeast cell, secretion is highly regulated. The distinction between ‘constitutive’ and ‘regulated’ secretion is, apart from a different molecular mechanism, based on different time scales of operation – although the latter step of regulated secretion is triggered by an abrupt signal on the scale of milliseconds, ‘constitutive’ secretion is regulated on a scale of seconds and minutes. Many recently discovered examples of plant secretory pathway regulation suggest that the concept of constitutive vs regulated secretion does not fit the distinctive features of the plant secretory pathway. Thoroughly studied examples of tip growth, cytokinesis, auxin transport (including tropic responses) or stomata regulation show that plant secretion is always regulated – on different time scales, covering both slow and abrupt signal regulation of vesicle exocytosis/endocytosis, often also co-regulated by calcium, which is considered to be the major trigger for regulated exocytosis in animals. Moreover, the concept of constitutive vs regulated secretion is an oversimplification, as it considers fusion with the plasmalemma to be the most important controlling step. Thus, despite the fact that plants lack ‘secretory Rabs’sensu Fukuda (2008), we propose a perspective acknowledging that plant exocytosis is regulated, with essential control units being RDs (see above). In most plant cell types, several RDs co-exist depending on the stage of differentiation.

In addition to testing the hypothesis that diverse Exo70 paralogues might localize to different ACDs within the same cell, future research should focus on the mapping of connections between distinct ACDs and specific endosomal recycling pathways, as well as the analysis of RD differentiation in cells leaving the meristematic stage at the onset and during elongation and differentiation. This should also include a consideration of the role of endomembrane compartment motility and the organization of cortical microtubules and cytoplasmic streams during different stages of cell elongation on cell polarity. Detailed mapping of RD compartments, and the identification of endosomal pathways linked to distinct plasmalemma proteins (both for RD establishment and maintenance), has already begun, as exemplified by work on auxin transporters.

To solve the mystery of TPC formation, it may be necessary to develop plant in vitro systems for vesicle formation and fusion, analogous to those which, in combination with genetics, were crucial in unravelling the deep secrets of the secretory pathway in opisthokonts. Although genetic analysis of plant secretory pathways is flourishing, we have no such in vitro systems in plants.

A systems biology approach to the study of cell polarity is becoming more than just fashionable also in plants. Models of auxin polar transport at the cellular level, together with mathematical models of tip-growing cells (for example, de Keijzer et al., 2009), will certainly lead the field. Online resources for signalling network model development are becoming available (see Grierson & Hetherington, 2008). This approach will facilitate both the integration of the fragmented knowledge on cell polarity regulation, as well as the identification of unrecognized parameters and components.


We thank Bruno Goud (Institut Curie, Paris, France) and Marek Eliáš (Charles University, Prague, Czech Republic) for valuable comments and suggestions. Our work was supported by MSMT LC06034, MSM 0021620858 and GAAV-IAA601110916 projects and the MSMT-KONTAKT-ME841 grant (collaboration with the group of John Fowler, Oregon State University, Corvallis, OR, USA).