II. Cytoskeleton and development of the stomatal complexes000
III. Cytoskeleton and stomatal cell shaping000
IV. Stomatal pore formation000
V. Substomatal cavity formation000
VI. Stomatal complex morphogenesis in mutants000
VII. Cytoskeleton dynamics in functioning stomata000
VIII. Mechanisms of microtubule organization in stomatal cells000
Microtubules (MTs) and actin filaments (AFs) form highly organized arrays in stomatal cells that play key roles in the morphogenesis of stomatal complexes. The cortical MTs controlling the orientation of the depositing cellulose microfibrils (CMs) and affecting the pattern of local wall thickenings define the mechanical properties of the walls of stomatal cells, thus regulating accurately their shape. Besides, they are involved in determination of the cell division plane. Substomatal cavity and stomatal pore formation are also MT-dependent processes. Among the cortical MT arrays, the radial ones lining the periclinal walls are of particular morphogenetic importance. Putative MT organizing centers (MTOCs) function at their focal regions, at least in guard cells (GCs), or alternatively, these regions either organize or nucleate cortical MTs. AFs are involved in cell polarization preceding asymmetrical divisions, in determination of the cell division plane and final cell plate alignment and probably in transduction of stimuli implicated in stomatal complex morphogenesis. Mature kidney-shaped GCs display radial AF arrays, undergoing definite organization cycles during stomatal movement. They are involved in stomatal movement, probably by controlling plasmalemma ion-channel activities. Radial MT arrays also persist in mature GCs, but a role in stomatal function cannot yet be attributed to them.
The stomatal complexes are epidermal structures of the aerial organs of higher plants and very rarely of the young primary roots (Christodoulakis et al., 2002). They exhibit a structural symmetry and an integrated function, properties that are the outcome of a precise developmental sequence and a complicated differentiation. Each of them consists of two highly differentiated cells, the GCs, which border an epidermal intercellular space- the stomatal pore- and in many plants are surrounded by one or more subsidiary cells (SCs; Fig. 1a). The GC pair and the included stomatal pore are defined as a stoma. The cells, which surround a stoma and differ in size, shape, arrangement and structure from the typical epidermal cells, are defined as SCs.
The stomata are capable of sensing a multitude of environmental signals to adjust properly the stomatal pore in order to regulate gas exchange in and out of the plant organs. According to Raschke (1979), the stomata function like ‘turgor–operated valves’. The GCs have been equipped with a specific wall structure and a unique physiology, properties that become functionally coordinated in stomatal movement (Ketellapper, 1963; Zeiger et al., 1987; Wilmer & Fricker, 1996). Changes in turgor induce alterations in GC shape and bring about opening and closing of the stomatal pore.
The GCs display two highly specialized forms: the kidney-like and the dumbbell-like one (Fig. 1b,c). The kidney-shaped GCs dominate among most plant families, while the dumbbell-shaped ones are only found in Poaceae and Cyperaceae. The anticlinal wall between the GCs is termed the ventral wall (VW), while those adjacent to the surrounding cells are the dorsal or lateral ones (Fig. 1a). The periclinal walls of the GCs are designated as external and internal ones (Fig. 1a).
The stomatal complexes have been the subject of intense research work by cell biologists, because they are model systems to examine the role of cytoskeleton in: cell shaping, the establishment of cellular polarity and the control of the plane of cell division in multicellular systems. Several aspects of MT and AF involvement in morphogenesis of stomatal complexes have been reviewed so far (Hepler, 1981; Palevitz, 1981a, 1982, 1991; Sack, 1987; Cleary, 2000). Besides, the stomata have proved to be an intriguing and important research object for plant physiologists (Hetherington, 2001; Schroeder et al., 2001).
This article attempts an overall review of the existing information on the organization and role of MTs and AFs during morphogenesis and function of stomatal complexes. Recently, the research interest has been focused on the cytoskeleton of mature GCs since the emerging evidence reveals that cortical radial AF arrays and probably cortical radial MT arrays are involved in stomatal movement (Hwang et al., 2000).
II. Cytoskeleton and development of the stomatal complexes
1. Developmental processes
The stomatal complexes are generated by one or more differential divisions, which are usually asymmetrical and produce the guard cell mother cell (GMC) and the SC(s), and by a symmetrical division, which yields the GCs. There are three ontogenetical types of stomatal complexes: the mesogenous in which the SCs have a common origin with the GCs, the perigenous where the SCs are derived from protodermal cells other than those cutting off the GCs, and the mesoperigenous in which the SCs are of mixed origin. In the perigenous stomatal complexes, the GMC induces asymmetrical divisions in some or all the adjacent cells (subsidiary cell mother cells; SMCs) that separate SCs (Stebbins & Jain, 1960; Stebbins & Shah, 1960, Fig. 2). In Vigna sinensis, which forms mesogenous stomata, the early formed leaf and hypocotyl stomata are large in size and lack SCs. These stomata have the ability to induce oriented divisions in their neighbouring cells, generating numerous SCs (Galatis & Mitrakos, 1979, 1980; see also Christodoulakis et al., 2002).
In mesogenous stomatal complexes, the number of the SCs seems to depend on the developmental stage of the protoderm. For instance, in very young leaves of Vigna sinensis (Galatis & Mitrakos, 1979) and Arabidopsis thaliana (Zhao & Sack, 1999) the stomata lack SCs or display one SC, while in more mature leaves the number of the SCs increases. Light (Kazama & Mineyuki, 1997) and increasing CO2 levels (Boetsch et al., 1996) induce an increase in the number of SCs in Cucumis sativus hypocotyls and Tradescantia virginiana leaves, respectively. Although the term stomatal cells strictly refers to the GCs, we use this term here in a broad sense to define all the cells involved in development of stomatal complexes.
2. Cytoskeletal markers of polarization in stomatal cells
The accumulation of cytoplasm, the migration of the nucleus and the organization of a preprophase MT band (MT-PPB) at the polar end of the cell manifest structurally the establishment of polarity in the asymmetrically dividing stomatal cells. The first two phenomena are visible in large cells, while the eccentric disposition of the MT-PPB is usually the only consistent premitotic marker of the polarized region in stomatal cells dividing asymmetrically (Galatis, 1974; Galatis & Mitrakos, 1979). Well-organized MT-PPBs precede stomatal asymmetrical divisions in the plants examined so far (Gunning, 1982; Mineyuki, 1999).
Although in some plant cell types the PPB position seems to be determined by the nucleus (Mineyuki et al., 1991, Mineyuki, 1999), in the asymmetrically dividing stomatal cells there is convincing evidence that it is defined by intercellular morphogenetic stimuli. This view is supported by the facts that: (a) the MT-PPB usually appears before nuclear migration to the polarized end of the SMCs of grasses (Galatis et al., 1983a, 1984b) and GMC progenitors of Allium cepa (Mineyuki & Palevitz, 1990) and (b) the displacement of the nucleus from the polarized end of the grass SMCs by centrifugation (Pickett-Heaps, 1969a; Galatis et al., 1984a) and cytochalasin B treatment (Cho & Wick, 1990) does not affect the MT-PPB assembly laterally to the GMC. However, the strongest evidence that local transcellular morphogenetic stimuli define the PPB position has been derived from the study of single SMCs of three Triticum species, being laterally in contact with two GMCs. When these SMCs are simultaneously induced to divide by the GMC pair, two MT-PPBs are assembled, one laterally to each GMC (‘double polarized SMCs’; Galatis et al., 1983a; Fig. 3a). Moreover, in bi-spaced binuclear SMCs of Triticum vulgare and Zea mays produced by caffeine treatment one MT-PPB is assembled laterally to the inducing GMC (Pickett-Heaps, 1969b; Apostolakos & Galatis, 1987). Therefore, the number and the position of the MT-PPBs is not related to the number of the nuclei, but is defined by the morphogenetic stimuli functioning in the dividing stomatal cells, a phenomenon also confirmed in other plant cell types (Manandhar et al., 1996a,b).
Fragmentary but important information is available on the mode of formation of the unique MT-PPBs in grass SMCs. These MT-PPBs are curved and line a limited region of the cell cortex transversely to the preceding interphase cortical MT arrays (Fig. 7b). They are formed by sets of antiparallel MTs, initiated by MT foci residing at the cell edges made by the periclinal walls with the anticlinal one facing the stomatal row, where putative MTOCs possibly function (Cho & Wick, 1989; Wick, 1991a). In colchicine-affected SMCs, the wall regions adjacent to the MT-focal sites are locally thickened in the absence of MTs, an observation revealing the polarizing nature of the underlying cortical cytoplasm (Galatis et al., 1984a).
In SMCs of the grasses Secale cereale (Cho & Wick, 1990, 1991) and Zea mays (Gallagher & Smith, 1999; Smith, 2001) and in the lateral SMCs of the monocotyledon Tradescantia virginiana (Cleary, 1995; Cleary & Mathesius, 1996; Cleary, 2000), apart from the MT-PPB and the AF-PPB, distinct cortical AF patches appear beneath the plasmalemma adjacent to the wall region shared with the GMC (Fig. 4a). They consist of thick AF bundles perpendicular to the epidermis. These bundles enter the cortical cytoplasm abutting the SMC periclinal walls and mark the polar region of the SMCs before mitosis. ‘Double polarized’ SMCs have also been found in Tradescantia virginiana, where two AF patches are localized, one laterally to each GMC (Cleary & Mathesius, 1996). In the same plant, AFs accumulate on the transverse anticlinal wall of the GMC progenitors closest to the polarized nucleus. By contrast to the MT-PPB and the AF-PPB, the AF patches persist in mitosis and cytokinesis and are more resistant to cytochalasin B treatment than the fine endoplasmic and cortical AFs (Cho & Wick, 1990, 1991; Kennard & Cleary, 1997). In the plasmolyzed SMCs of Tradescantia virginiana the plasmalemma adjacent to the PPB region and to the cortical AF patches is not detached from the cell wall, a phenomenon implying the existence of a relatively strong connection between the cell wall-plasmalemma-cytoskeleton in these regions (Cleary, 2001).
The cortical AF patches may form to protect plasmalemma regions being under mechanical stresses. In such conditions the animal cells form AF gatherings to reinforce the stressed plasma membrane regions (Ingber, 1997; Frame & Sarelius, 2000; Ko & McCulloch, 2000), a phenomenon also observed in plasmolyzed leaf cells of Chlorophyton comosum (Komis et al., 2002a). The AF patches in SMCs line the wall region shared with the lateral walls of the adjacent GMCs that elongate appreciably (Cho & Wick, 1989; Cleary & Hardham, 1989; Kennard & Cleary, 1997, see also section III.1.1 and Fig. 2). This elongation may trigger AF patch formation, a hypothesis supported by the following findings: first AF patches line both sides of the common wall between SMC and GMC, while the young GMCs lack them (Cho & Wick, 1990, 1991). Second the AF patches are retained in the young SCs of Tradescantia virginiana (see Fig. 13 in Cleary & Mathesius, 1996) and Zea mays (our unpublished data). At this stage the adjacent lateral GMC walls continue to elongate (see Figs 12 and 13 in Cleary & Mathesius, 1996). Third in plasmolyzed SMCs of Tradescantia virginiana, the AF patches extend into neighbouring cortical regions, which have been detached from the cell wall (Cleary, 2001).
The correct positioning of the nucleus and the proper orientation of the spindle axis are premitotic events essential for the correct placement of the new cell wall. In the monocotyledon stomatal complexes examined so far, the nuclear position at the polar end of the GMC progenitors is controlled by transcellular morphogenetical gradients functioning along the stomatal rows, while in SMCs by stimuli emitted by the GMCs (Stebbins & Jain, 1960; Stebbins & Shah, 1960; Croxdale, 1998). In SMCs of Tradescantia virginiana, the polar nuclear migration occurs at the G1 stage of the cell cycle (Kennard & Cleary, 1997). In the ‘double polarized’ SMCs (see section II.2) the nuclear migration is controlled by stimuli emitted by both the GMCs. When they are equal in strength, the nucleus occupies a position at the mid-distance between the GMCs (Fig. 3a), but where they are unequal, it is placed close to one GMC (Galatis et al., 1983a, 1984b; Kennard & Cleary, 1997). In the latter case, the preprophase/prophase nucleus often forms an angular MT-flanked protrusion directed towards the stronger inducing GMC (Galatis et al., 1983a).
The induction of asymmetrical divisions producing the GMCs in the hypocotyl protoderm of Arabidopsis thaliana is probably a hormone-dependent process (Saibo et al., 2003). Besides, the stimulus, which is emitted by the GMCs and induces asymmetrical divisions in the grass SMCs might also be a hormone-like substance (Stebbins & Jain, 1960; Stebbins & Shah, 1960; Pickett-Heaps & Northcote, 1966). The SMCs of Tradescantia virginiana polarize at the G1 stage of the cell cycle and remain polarized for 22 h, while the GMCs stay at the G1. SMC polarization during G1 stage and mitosis may be induced by different signals emitted from the GMC or some ‘master signal’ dictates both processes (Kennard & Cleary, 1997). The stimulus in grass SMCs persists for a relatively long period (Pickett-Heaps, 1969b; Apostolakos & Galatis, 1987). Observations supporting the chemical nature of the stimulus have been made in incompletely divided caffeine-affected SMCs of grasses. Some of them, which display a SC wall having a minute gap, are re-induced to divide and assemble a MT-PPB at the site of the previous MT-PPB, inside the incomplete SC (Apostolakos & Galatis, 1987). In these SMCs, the nucleus(i)- and later the one mitotic spindle pole are stabilized close to the SC wall gap through which the putative inductive stimulus possibly entered the larger cell compartment (Apostolakos & Galatis, 1987).
Apart from the above, the possibility that local mechanical stresses applied on the asymmetrically dividing stomatal cells by their neighbours may trigger cell polarization and/or define the PPB position should be also considered (Green et al., 1970; Galatis & Mitrakos, 1979; Pickett-Heaps et al., 1999). For instance, in grasses, the elongating GMCs (Figs 2, 7a–c) may exert mechanical stresses on the adjacent SMC wall region (see section III.1.1). They may induce local differentiation in the cell wall and the adjacent plasmalemma generating asymmetries on the SMC surface, triggering mechanisms establishing polarity in it (Fowler & Quatrano, 1997). A cortical asymmetry is revealed in the plasmolyzed polarized SMCs of monocotyledons, where the protoplast remains attached on the cell wall region shared with the GMC (Stebbins & Jain, 1960; Stebbins & Shah, 1960; Cleary, 2001). The stresses could activate stretch-activated channels in the plasmalemma, causing the influx of ions, which could be involved in SMC polarization (Kennard & Cleary, 1997; Cleary, 2000). Kennard & Cleary (1997) presented convincing evidence that mechanical stresses induce polarity in SMCs. They applied pressure by a fine needle on individual SMCs of Tradescantia virginiana opposite to the adjacent GMC and found that the nucleus moves towards the site of pressure application (Fig. 5).
Regardless of the nature of the inductive stimulus, between GMC and the SMCs, between the former and the SCs as well as between the young GCs and SCs, information is probably exchanged to ensure their coordinated function and differentiation. AFs might be implicated, not only in the transduction of the inducing stimulus polarizing the SMCs (see section II.3), but also in cell-to-cell communication through the cell wall and then across plasmalemma or through plasmodesmata. The lateral walls of Zea mays GMCs display numerous plasmodesmata arranged in primary pit fields (Galatis, 1982). The acto-myosin system is probably involved in the function of plasmodesmata. Actin, myosins and other actin-related proteins have been localized in the plasmodesmata of several plants. Moreover, reliable experimental data suggest that the acto-myosin system regulates the communication between neighbouring cells through plasmodesmata (Crawford & Zambryski, 1999; Heinlein, 2002; Volkmann et al., 2003).
Study of dividing SMCs of grasses showed that the mechanism functioning in the PPB region guides or attracts the cell plate only when its edges reach a certain distance from it (Galatis et al., 1984a,b). This is possible when the spatial organization of the cytokinetic protoplast is proper, that is the PPB region is not masked by large organelle(s) and the cell plate grows on a plane almost parallel to the PPB region. Otherwise, a portion of the cell plate may diverge and fuse with the parent walls far from the PPB region, forming triangular SCs or SCs with more complicated shapes (Galatis et al., 1984a,b). These conditions are established during SMC polarization by: first the placement of the nucleus close to the inducing GMC; second the proper organelle disposition; third the stabilization of the one mitotic spindle pole close to the GMC; and fourth the alignment of the mitotic spindle axis more or less transversely to the PPB plane (Galatis et al., 1983a, 1984a,b).
In vivo study of Tradescantia virginiana SMCs revealed that the anchored spindle pole acts as a pivot point for the spindle. The other spindle pole swings around during cell division placing the one edge of the cell plate in close proximity to the PPB region. In these SMCs, realignment mechanism(s) function to correct cell plate arrangement, when one edge of the cell plate is in close proximity to the PPB region (Cleary, 2000; Pickett-Heaps et al., 1999).
Factors affecting the plane of SMC division in normal Triticum spp. protoderm are the size of the SMC and the arrangement of its transverse walls in relation to the GMC, and mainly the function of two or more polarizing stimuli from different directions (Galatis et al., 1983a, 1984b). It is interesting that the shape of the atypical SCs in Triticum spp. often changes in space. In median paradermal planes, where the SC nucleus masked the PPB region, the SCs display a triangular form, while in external planes where the PPB region was approachable they exhibit the typical lens-like shape (Galatis et al., 1984b). Moreover, in caffeine-produced bi-spaced SMCs of Zea mays, atypical SCs are formed because wall strips prevent the cell plate from meeting the PPB region (Apostolakos & Galatis, 1987, Fig. 6). When in these cells the cell plate edge approaches the PPB region, it is guided to form a SC wall portion inside the wall strips of the aborted cell (Fig. 6e,f).
According to Cleary (1995, 2000; see also Pickett-Heaps, 1969a) the curved growth of the cell plate in SMCs is controlled by interactions between the phragmoplast MTs and AFs with the SC nucleus. As a result, the cell plate is pulled around the nucleus. This hypothesis does not explain all cases. Often, the nascent cell plate grows on one plane and curves after its emergence from the interzonal region, when it faces the PPB This is very clear in the atypical SCs formed in the ‘double’ polarized SMCs (Fig. 3c,d, see also Galatis et al., 1983a) and in SMCs divided during centrifugation (Galatis et al., 1984a).
In some asymmetrically dividing stomatal cells, apart from the PPB region, there are other cortical sites, affecting the cell division plane. In three Anemia species that form ‘floating’ stomata, the cell wall separating the GMC has a funnel-like shape and fuses with the periclinal walls of the mother cell only. The GMC progenitors are ‘double polarized’. A circular MT-PPB abutted on the external periclinal wall predicts the sites of fusion of the cell plate with the underlying wall regions. At the same time, a limited cortical region abutted on the internal periclinal wall outlining a minute intercellular space, defines the site of the cell plate fusion with the internal periclinal wall (Galatis et al., 1986). Cortical as well as endoplasmic MTs converge on this cortical region at preprophase. Moreover, cortical regions far from the PPB region, which display MTs during preprophase, locally affect the plane of cell division in ‘protodermal’ cells of the liverwort Marchantia paleacea involved in air pore and air chamber development (Apostolakos & Galatis, 1985b,c) and in caffeine-produced bi-spaced SMCs of Zea mays (Apostolakos & Galatis, 1987). These sites guide or attract the cell plate to fuse with the underlying wall areas. Therefore, in ‘double’- and ‘multipolarized’ cells, the cell plate does not fuse with the parent wall regions predicted by the PPB only, but, following complicated but predictable paths, separates daughter cells of aberrant shape.
III. Cytoskeleton and stomatal cell shaping
1. GMC shaping
GMCs of the dumbbell-shaped stomata The newly formed GMCs of the dumbbell-shaped stomata are hexahedra with narrow lateral walls, which before symmetrical division expand appreciably (Fig. 7a–h). The advanced GMCs are usually constricted at their mid-transverse plane, while their polar ends are swollen (Figs 2 and 7). In paradermal sections, the lateral walls appear curved inwards, while the periclinal ones exhibit a slight median transverse groove.
The GMCs of grasses bear a well-organized interphase MT band lining the mid-region of the lateral and periclinal walls (Galatis, 1974, 1982; Busby & Gunning, 1980; Cho & Wick, 1989; Cleary & Hardham, 1989; Mullinax & Palevitz, 1989, Fig. 7a–d). This is established at an early interphase stage transversely to the axis of the stomatal row and exists until preprophase, when it is replaced by the MT-PPB (Fig. 7a–h). A few MTs line the transverse walls anticlinally. By contrast, the polar regions below the periclinal walls are traversed by a relatively significant number of MTs. Initially, they are randomly oriented but later are seen to converge on the interphase MT band region (Galatis, 1982; Cleary & Hardham, 1989, Fig. 7a–d). Externally to the interphase MT band, an identical CM band is deposited (Galatis, 1982). This allows the elongation of the GMCs parallel to the axis of the stomatal row but also results in the appearance of a median GMC constriction preventing the cell bulging locally. The advanced interphase GMCs of Zea mays display numerous endoplasmic MTs, which diverge from the mid-region of the interphase MT band below the periclinal walls and enter the endoplasm. These MT arrays persist in the preprophase GMCs and are related to the spatial arrangement of plastids, endoplasmic reticulum (ER) and the nucleus that becomes ellipsoidal (Galatis, 1982). The above data and the fact that colchicine affects GMC shaping (Fig. 8a) provide evidence that this process is MT-controlled.
The median constriction of the GMCs of grasses establishes conditions favouring the formation of the GC canal (Fig. 1c). Some GMCs of Zea mays affected by colchicine at an advanced interphase stage differentiate into persistent GMCs exhibiting a bone-like shape. Moreover, in the same plant the complete inhibition of VW formation by caffeine treatment does not prevent the formation of a narrow, canal-like region in the forming aberrant stomata (Galatis & Apostolakos, 1991, Fig. 8b,c). The above suggests that the grass GC morphogenesis starts at the GMC stage and that the presence of a VW is not a prerequisite for it (Galatis, 1982; Galatis & Apostolakos, 1991).
In interphase and preprophase GMCs of Secale cereale (Cho & Wick, 1990, 1991) AF bundles oriented transversely to the stomatal row axis run through the cortical cytoplasm. AF arrays resembling an interphase MT band and an AF-PPB have not been found. AF patches line the whole surface of the lateral walls in advanced interphase, mitotic and cytokinetic GMCs of the same plant. These are probably involved in stabilization of the mitotic spindle poles (Cho & Wick, 1990, 1991) or in the correct cell plate realignment during cytokinesis (Cleary, 2000). According to the current knowledge the cortical AFs are not, at least directly, involved in GMC shaping.
GMCs of the elliptical stomata Detailed information on GMC shaping of the elliptical stomata is available only for the fern Asplenium nidus (Apostolakos et al., 1997). The GMCs of this plant undergo characteristic, MT-patterned, morphogenetic changes. The advanced interphase GMCs display a round form in surface view, which in median transverse planes becomes semilobed (Fig. 9a) by a constriction at the mid-region of the internal periclinal wall and the two lateral anticlinal ones. The external periclinal wall is lined by a radial MT array (Fig. 9b), which converges on its thin mid-region (Fig. 9a), followed by deposition of similarly oriented CM arrays. The latter dictates the tangential expansion of the external periclinal wall, promoting GMC rounding. Moreover, a U-like MT bundle lines the site of future cell isthmus. Externally of this bundle a U-like CM bundle is deposited. The latter, preventing the increase of cell diameter on its plane, results in the assumption of the semilobed shape by the GMCs (see also Panteris et al., 1993b). Although in Asplenium nidus GMCs the cortical AFs form arrays similar to those of MTs (Fig. 9c; cf. Fig. 9b), they are not involved, at least not directly, in GMC morphogenesis (Apostolakos et al., 1997). In this plant, basic structural features of GCs are established at the GMC stage (Apostolakos et al., 1997; Apostolakos & Galatis, 1999). Thus, in elliptical stomata GC morphogenesis seems also to commence at the GMC stage (Galatis & Mitrakos, 1979; Galatis et al., 1982; Zhao & Sack, 1999).
The assembly of radial MT and AF systems below the periclinal walls in the advanced interphase GMCs of the elliptical stomata is a rather general phenomenon. Radial MT arrays form in Selaginella spp. (Cleary et al., 1992) and Allium cepa (Mineyuki et al., 1989) GMCs, while radial AFs line the periclinal walls of Tradescantia virginiana GMCs (Cleary & Mathesius, 1996, Fig. 4b).
In 21 Leguminosae species (Galatis & Mitrakos, 1979; Galatis et al., 1982) and in the Cruciferae species Arabidopsis thaliana (Zhao & Sack, 1999) the anticlinal GMC wall regions adjacent to the MT-PPB become locally thickened. These thickenings are distinct in the young GCs and probably set up mechanical forces promoting stomatal pore formation (Galatis et al., 1982). In particular, the prethickened ends of the GC dorsal walls as well as the VW thickenings (see section III.2.1) may serve to make these anticlinal wall regions more resistant to expansion. As in stomatal movement they possibly result in the bending of the VW, thus facilitating the schizogenous opening of the stomatal pore (see section IV). The PPB region in GMCs of Leguminosae shows numerous coated regions on the plasmalemma and numerous coated vesicles, observations suggesting the function of a local endocytotic route at this site (see also Mineyuki et al., 2003). It should be noted here that local wall thickenings emerge at particular wall sites adjacent to the PPB region in colchicine affected SMCs of Zea mays (Galatis et al., 1984a). Besides, wall strips emerge in the PPB region of some caffeine-affected SMCs and GMCs of grasses (Pickett-Heaps, 1969b; Apostolakos & Galatis, 1987; Galatis & Apostolakos, 1991). They often grow inwards and rarely completely divide the affected cell (Apostolakos & Galatis, 1987).
The asymmetrical organization of the MT arrays under the GC walls is the first structural event of GC differentiation (Galatis, 1974; Galatis & Mitrakos, 1980; Galatis et al., 1983b). Local thickenings emerge along the whole length of the mid-region of the VW lined by the MT bundle as well as at the junction of the mid-region of the periclinal walls with the VW, before the opening of the stomatal pore (reviews by Palevitz, 1981a, 1982; Sack, 1987).
Considering the consistency of coalignment between MT and CM arrays in stomatal cells, recent data confirming the MT involvement in CM orientation (Baskin, 2000, 2001; Burk & Ye 2002; Zhong et al., 2002) and the effects of MT depolymerization in GC shaping (Palevitz & Hepler, 1976), it is assumed that: first the cortical MT arrays function as a prepattern of the kidney-shaped GC morphogenesis, defining the CM orientation at every stage of the process. The kidney-like as well as the dumbell-like GC shape is the outcome of a continuous interplay between the growing protoplast and the properly reinforced anticlinal and periclinal walls, which accurately define the position and the direction of their expansion. Second the radial MT arrays below the periclinal walls are the key elements of the kidney-shaped GC morphogenesis. The radial CMs, which are deposited externally to them, allow a tangential expansion of the periclinal walls forcing the GC to assume a kidney-like shape. Third the paired radial MT arrays are implicated in the deposition of local wall thickenings.
Stomatal elongation is highly coordinated, both temporarily and positionally, between the GCs, between the GCs and the adjacent SCs and between the stomatal complex and the surrounding leaf cells. When the leaf elongation is inhibited or the MTs are disintegrated, the kidney-shaped grass GCs fail to assume a dumbbell-like shape (Fig. 11a). This is also the case for Zea mays stomata at the tip of mature coleoptiles, which remain elliptical, because in this region the elongation is diminished (Fig. 11b). Moreover, the cyperaceous stomata grown in the field are short and display radial CMs in the periclinal walls, while those developed in the glasshouse are more elongated, assume a rather typical dumbbell-like shape and show axial CMs in the canal region (Mishkind et al., 1981). Thus, the elongation is essential for the dumbbell-shaped GC morphogenesis.
In summary, it is reasonable to assume that the acquisition of a temporary kidney-like shape by GCs is a prerequisite for the dumbbell-like GC morphogenesis, that is the radial CM systems are of critical importance for the shift from the kidney-like to dumbbell-like pattern of GC morphogenesis.
3. Subsidiary cell shaping
The available information on SC shaping is limited to the dumbbell-shaped stomata. Tubulin immunolabeling revealed definite shifts of MT organization in SCs of grasses at particular stages of stomatal development (Cho & Wick, 1989; Cleary & Hardham, 1989; Palevitz & Mullinax, 1989; Palevitz, 1991). The young SC displays transiently radial cortical MT systems below the periclinal walls, converging on the junction of their mid-region with the lateral wall of the GMC (Fig. 7g–j). In each SC, the sites of convergence of the two opposite radial MT arrays are paired with an anticlinal MT bundle (Palevitz & Mullinax, 1989). Therefore, all four cells of a young dumbbell-shaped stomatal complex display paired radial MT systems (Fig. 7i,j). If these MT systems are followed by deposition of similar radial CM arrays in the periclinal SC wall, the SC is prevented from elongating, but is forced to assume temporarily a lens-like shape in order to codifferentiate with the GCs.
As it has been already noted, during GC elongation, the MTs under the mid-region of the periclinal SC walls become transverse to the stomatal axis. Such a CM orientation allows the coextension of the SC with the GCs. Finally, the cortical MTs in SCs become axially oriented as in Avena sativa (Palevitz & Mullinax, 1989, Fig. 7p) or remain transverse like in Lolium rigidum (Cleary & Hardham, 1989, Fig. 7o). The above show that the morphogenesis of the lens-shaped SCs in grasses is highly coordinated with that of the GCs.
IV. Stomatal pore formation
One of the main tasks of GC morphogenesis is stomatal pore formation, a phenomenon concomitant to GC shaping that follows the deposition of local wall thickenings at the mid-region of the VW and at its junctions with the periclinal walls. This starts from the external and/or the internal periclinal walls and proceeds inwards (Stevens & Martin, 1978; Galatis & Mitrakos, 1980; Galatis, 1980; Galatis et al., 1983b; Sack & Paolillo, 1983c; Sack, 1987). Stomatal pore formation involves two processes: the weakening of the middle lamella of the VW, and the application of mechanical forces to disrupt the periclinal walls and to separate the VW partners of the GC pair at the stomatal pore site.
The mechanical forces that locally separate the GCs are generated in the course of the assumption of the kidney shape by the GCs. They are applied to the thickened region of the VW and act as follows: the radial CM arrays allow the tangential expansion of the periclinal walls and force the dorsal walls to become curved. The local wall thickenings along the stomatal pore region prevent the deformation of the VW at its mid-region and promote the local separation of the VW partners, after the disruption of the overlying periclinal wall regions. Afterwards, the separated VW partners move further apart from each other, to complete the stomatal pore (Galatis & Mitrakos, 1980). The same mechanism functions in the dumbbell-shaped stomata where the stomatal pore opening keeps pace with the temporal acquisition of a more or less kidney-like shape by the GCs (Galatis, 1980; Palevitz, 1981a, 1982; Palevitz & Mullinax, 1989). The forces creating the stomatal pore are generated by the increase of the protoplasm and/or of the turgor. K+ and Cl− ion levels increase substantially in Phleum pratense and Zea mays GCs before stomatal pore formation (Palevitz, 1981a, 1982).
In the moss Funaria hygrometrica (Sack 1983c, 1987) and in the fern Azolla spp. (Busby & Gunning, 1984) cuticular precursors are deposited in the middle lamella at the region where the stomatal pore opens. This accumulation probably weakens the binding properties of the middle lamella, facilitating VW separation (Sack, 1987). In these cases, stomatal pore formation also follows the deposition of the radial CM arrays in the periclinal walls, keeping pace with GC shaping.
A different mode of stomatal pore formation functions in the ferns Adiantum capillus-veneris (Galatis et al., 1983b), Anemia mandioccana (Zachariadis et al., 1997), Asplenium nidus (Apostolakos & Galatis, 1998, 1999) and probably in Polypodium vulgare (Stevens & Martin, 1978). In these ferns the phenomenon commences in very young GCs at the mid-depth of the VW by the local movement of the adjacent plasmalemmata, apart from each other, while the periclinal walls remain intact. The VW at this stage is a thin membranous diaphragm, which does not display detectable middle lamella and wall materials and contains callose. Thus, a rudimentary ‘internal’ stomatal pore is formed, which broadens towards the periclinal walls (Fig. 12a). The final stomatal pore opening is achieved by disruption of the periclinal wall portions covering the ‘internal’ stomatal pore (Zachariadis et al., 1997; Apostolakos & Galatis, 1998, 1999).
The initiation of the ‘internal’ stomatal pore coincides with the organization of the anticlinal MT bundles along the middle of the VW and the colocalization of AFs in the same sites (Zachariadis et al., 1997; Apostolakos & Galatis, 1998, 1999, Fig. 10b,d). Treatment of Anemia mandioccana and Asplenium nidus stomata with anti-MT drugs (Zachariadis et al., 1997; Apostolakos & Galatis, 1998) and taxol (our unpublished data) inhibits stomatal pore opening. Therefore, the MTs in the above ferns seem to be directly involved, by an uknown mechanism, in the separation of the plasmalemmata during the internal opening of the stomatal pore. This mechanism, at least in Asplenium nidus, is Ca2+-dependent. The mycotoxin cyclopiazonic acid, which interferes with the establishment of the ER-dependent Ca2+ gradients, inhibits the internal stomatal pore formation in Asplenium nidus (our unpublished data; Fig. 12b). Afterwards, forces generated during GC shaping disrupt the periclinal wall remnants covering the ‘internal’ stomatal pore. Therefore, the cortical MTs are directly involved in stomatal pore formation in some ferns and indirectly in the majority of the plants via MT implication in GC morphogenesis. The AFs do not seem to play a particular role in this process in angiosperms, since cytochalasin B does not inhibit stomatal pore formation (Palevitz, 1980).
The stomata form a pore even when a typical VW is absent. In the caffeine-formed aberrant stomata of Zea mays (Galatis & Apostolakos, 1991) and in the cyd1 mutants of Arabidopsis thaliana (Yang et al., 1999) wall thickenings are deposited at the free end of one of the VW strips. In these aberrant stomata a modified mechanism of stomatal pore formation functions. The wall thickenings display electron dense material, which is degraded to form a rudimentary nonfunctional stomatal pore (Galatis & Apostolakos, 1991). The tendency to form a pore is astonishing in caffeine-formed aberrant stomata, which completely lack a VW. In these stomata, wall thickenings are deposited on the mid-region of the periclinal walls, which often grow inwards to form a continuous wall column surrounded by MTs (Galatis & Apostolakos, 1991, Fig. 13a,b). Electron dense materials are deposited in the wall column (Fig. 13a,b), the dissolution of which results in the formation of a rudimentary pore (Fig. 13c). These findings support a function of VW formation in bringing into communication the two periclinal walls to form the stomatal pore.
V. Substomatal cavity formation
Substomatal cavity initiation precedes stomatal pore formation. These processes are substages of a major process, stomatal morphogenesis, and keep pace with the intercellular space formation in the subepidermal tissues. Although it has been reported that the substomatal cavity formation may result from the lysis of some cells below the stomata (DeChalain & Berjak, 1979), there is conclusive evidence from the study of various plants that it is the result of a gradual schizogenous process (Laroche et al., 1976; Steven & Martin, 1978; Galatis, 1980, 1982; Sack & Paolillo, 1983a,b,c; Galatis et al., 1986; Apostolakos et al., 1997). Mechanical forces generated from the differential expansion between the GMC and the underlying cells, seem to play the major role in the substomatal cavity formation, although some local lytic activity on the middle lamella of the separating walls, at least at their junctions, cannot be excluded (Sack, 1987).
Substomatal cavity initiation in three species of Anemia (Galatis et al., 1986) and in Asplenium nidus (Apostolakos et al., 1997), keeps pace with the morphogenesis of the GMCs or their progenitor cells. In Anemia spp., the interphase GMC progenitors undergo a peculiar polarization expressed on its shape. The internal periclinal wall at the polar end of the cell is locally detached from the walls of the adjacent cells, while the cell bulges slightly outwards. Thus, a minute intercellular space is initiated at the polar end of the cell, which gradually develops into a substomatal cavity. The initial detachment of the walls is the consequence of the shaping of the GMC progenitor only. Cortical MTs as well as endoplasmic MTs converge on the internal periclinal wall region bordering the minute intercellular space (Galatis et al., 1986).
Similarly to Anemia spp., in Asplenium nidus minute intercellular spaces open between protodermal and subprotodermal cell layers below the proximal polarized end of the GMC progenitors, which gradually develop into substomatal cavities (Apostolakos et al., 1997). The formation of the median constriction in Asplenium nidus GMCs (see section III.1.2) promotes the further wall detachment between the GMC and the subepidermal cells, contributing to the broadening of the substomatal cavity. Inhibition of GMC morphogenesis with anti-MT drugs blocks the initiation and development of the substomatal cavities (Apostolakos et al., 1997).
The assumption of the lobed form by the subepidermal cells also contributes to the development of the substomatal cavities. The morphogenesis of the lobed mesophyll cells in grasses (Jung & Wernicke, 1990; Apostolakos et al., 1991; Wernicke & Jung, 1992) and ferns (Panteris et al., 1993c) is accurately controlled by a system of highly organized MT bundles. They control the deposition of identical CM systems externally to them, which in turn define the positions of the cell contrictions. The lobed cell morphogenesis is always coupled with intercellular space formation.
In Asplenium nidus the broadening of the substomatal cavity is the result of the temporal and positional coordination of the GMC morphogenesis with that of the underlying cells. Frequently, the U-like MT bundle defining the position of the GMC constriction (see section III.1.2) is opposite to a MT bundle defining the position of a cellular constriction in the subepidermal cell (Apostolakos et al., 1997). This synchronous and opposite formation of constrictions between the GMC and the subprotodermal cells contributes to the substomatal cavity development. The position of the U-like MT bundles, in the differentiating semilobed protodermal cells of Adiantum capillus-veneris, seems to be determined by the subprotodermal cells (Panteris et al., 1993b, 1994). Therefore, the coordination of morphogenesis between protodermal and subprotodermal cell layer resulting in substomatal cavity formation should be a more general phenomenon among plants.
VI. Stomatal complex morphogenesis in mutants
The morphogenesis of stomatal complexes in mutants has been studied in Zea mays and Arabidopsis thaliana. In the first plant the mutations dcd1, dcd2, pan1 and brk1 disturb cell plate arrangement during asymmetrical divisions but not of the symmetrical ones. In particular, the dcd1 and dcd2 mutations affect the divisions in the GMC progenitors and in the SMCs, producing GMCs and SCs that are atypical in size and form, while pan1 and brk1 affect the divisions in SMCs forming atypical SCs (Gallagher & Smith, 1999, 2000; Smith, 2001).
In the asymmetrically dividing GMC progenitors and SMCs of dcd1 and dcd2 mutants the one anticlinal edge of the cell plate fuses with the parent wall at the PPB site, while the other diverges and meets the parent walls at unpredicted sites or remains free in the cytoplasm. In this way, triangular or more abnormal daughter cells are formed or the cytokinesis is not completed (Gallagher & Smith, 1999). In the double-mutants dcd1/dcd2, apart from the above-described cell plate disarrangements, the cell plate in SMCs may fuse with parent wall regions far from the PPB. Thus, lens-shaped cells not adjacent to GMCs or circular cells, appearing free inside SMCs in a paradermal view, are produced (Gallagher & Smith, 2000).
The structural polarization of the GMC progenitors and SMCs in dcd1, dcd2 and dcd1/dcd2 mutants is not affected. In particular: first the nucleus is placed at the polar end of the cell, which is marked by a cortical AF patch; second the MT-PPB and the AF-PPB appear in the expected position; third the one pole of the spindle is linked with AFs with the cortical AF patch; and fourth the phragmoplast-cell plate system is connected with the PPB region and the cortical AF patch by AFs (Gallagher & Smith, 1999, 2000; Smith, 2001). Therefore, in the mutants as in the wild type, all the cellular conditions are fulfilled for the cell plate to meet the parent walls at the predetermined position. According to the above authors, the discordia mutations affect the AF-based mechanism, which guides the margins of the expanding cell plate to fuse with the parent walls at the PPB region. This conclusion is based on: first the similarity of the effects of mutations on cell divisions to those produced by a cytochalasin D treatment in the wild type; and second the finding that cytochalasin D treatment in dcd1 and dcd2 mutants increases the number of the atypically dividing SMCs (Gallagher & Smith, 1999).
The cell plate misalignments in SMCs of the discordia mutants resemble those in some SMCs of wild type Triticum spp. seedlings (Galatis et al., 1984b). In the latter, the PPB region is unable to guide the expanding cell plate edges completely, since it is locally covered by larger organelles or the diverging cell plate edge grows far from it (see section II.5). Therefore, in these cells the AF systems involved in the final cell plate orientation seem to be functional because in planes where the developing cell plate approaches the PPB region it grows towards it (Galatis et al., 1984a,b). As in Triticum SMCs, the disturbance of the SMC division plane in discordia mutants may also be due to an inappropriate spatial organization of the dividing protoplast, as the result of an improper cell polarization.
In pan1 and brk1 mutants, SMC polarization is completely inhibited or significantly disturbed (Gallagher & Smith, 2000; Smith, 2001; Frank & Smith, 2002). This view relies on the following observations: first that many SMCs of the brk1 mutants lack AF patches; second that the nuclear migration towards the inducing GMC is inhibited in many SMCs of both the above mutations; and third that about 25% of SMCs of the pan1 mutants do not form a PPB at the expected position laterally to the inducing GMC, but obliquely or transversely to the long SMC axis. In these cells, the phragmoplast-cell plate system traverses the cell diagonally or transversely, forming equal or unequal daughter cells, which, according to the criteria used by the authors, do not display features of SCs (Gallagher & Smith, 2000; Frank & Smith, 2002). Frank & Smith (2002) found that the BRK1 gene encodes a novel 8KD protein that may be involved in the AF-dependent cell polarization.
However, in SMCs of the pan1 and brk1 mutants the nuclear behaviour, the PPB position, and the orientation of the mitotic spindle may predict a symmetrical and not an asymmetrical division. Although in grasses the cells of the rows, which function as SMCs, have completed their division cycle, they rarely undergo symmetrical divisions (Galatis et al., 1983a, 1984b). In the latter case, they are not competent to form a SC. Therefore it is possible that pan1 and brk1 mutations affect the time-course of the asymmetrical and symmetrical divisions in protodermal cell rows involved in stomatal development.
In Arabidopsis thaliana the ‘cytokinesis defective’ (cyd1) mutation affects the symmetrical division of the GMCs, disturbing cytokinesis. In cyd1 mutants aberrant stomata are formed, that in paradermal view display one or two incomplete VW strips opposite to one another or even they lack them completely (Yang et al., 1999). The free end of one of the VW strips becomes locally thickened and a stomatal pore is initiated. The effects of cyd1 mutation mimic those induced by caffeine treatment in Zea mays stomata (Galatis & Apostolakos, 1991). Like caffeine, cyd1 mutation may disturb cell plate stabilization, thus leading to partial or total inhibition of cytokinesis in GMCs. All the other phenomena of GC morphogenesis are carried out normally in the atypical stomata, even in the complete absence of a VW.
VII. Cytoskeleton dynamics in functioning stomata
1. Actin filaments
The mature kidney-shaped GCs are unique in that they are able to change the organization of their cortical cytoskeleton in response to external environmental factors, inducing stomatal movement. In stomata opened under white light, the cortical AFs below the periclinal walls are radially arranged around the stomatal pore (Fig. 14), an arrangement resembling that of MTs (Hwang et al., 2000). By contrast, in stomata closed in darkness or by other stimuli, the radial AFs are disintegrated (Kim et al., 1995; Eun & Lee, 1997, 2000; Hwang et al., 2000; Hwang & Lee, 2001). Long cortical and subcortical AF bundles displaying various orientations replace the radial AF arrays (Hwang & Lee, 2001; Fig. 14). These results have been mainly derived from Commelina communis, using AF immunolocalization or AF staining by rhodamine-phalloidin in fixed cells or by microinjection of fluorescein isothyocyanate-phalloidin into living GCs. The changes in cortical AF organization during stomatal movement have been also confirmed in vivo in Arabidopsis transgenic plants (WT/GFP-mTn line) expressing a GFP fusion protein targeted to actin (Dong et al., 2001; Lemichez et al., 2001).
Radial AF arrays have been also found in open stomata of Vicia faba and Nicotiana plumbaginifolia, but in these plants they are not so obvious compared to Commelina communis. Radial-more or less-AF arrays have also been observed in GCs of Selaginella spp. (Cleary et al., 1993), Tradescantia virginiana (Cleary & Mathesius, 1996) and in Arabidopsis thaliana (wild plants, mutants and plants transformed with a GFP-mouse talin reporter; Kost et al., 1998; Eun et al., 2001).
The diverse responses of the radial AF arrays to opening stimuli provide evidence that their organization in functioning GCs is not a result of stomatal movement. Thus: first circadian clock-induced opening of stomata is accompanied by the assembly of radial cortical AF arrays (Eun & Lee, 1997); second fusicoccin, a proton pump activator inducing excessive stomatal opening, inhibits the formation of radial AFs and disrupts the existing ones (Eun & Lee, 2000); and third hypotonic treatment of Vicia faba GCs, which promotes stomatal opening, induces AF disintegration (Liu & Luan, 1998). Besides, abscisic acid (ABA) treatment of Commelina communis and Arabidopsis thaliana, which induces stomatal closure, disrupts radial AF arrays rapidly (Eun & Lee, 1997; Hwang & Lee, 2001; Lemichez et al., 2001).
Cytochalasin D treatment, which disrupts the radial AF arrays in GCs (Kim et al., 1995; Eun & Lee, 2000), promotes the stomatal pore opening, which is induced by white light and by circadian clock and enhances stomatal closure in the presence of ABA, under low external CO2 and high K+ concentration in the medium, conditions in which the stomatal closure does not easily occur. Besides, the AF-stabilizing agent phalloidin inhibits stomatal opening induced by white light and fusicoccin or circadian clock; it also inhibits the ABA induced stomatal closure (Hwang et al., 2000). These results further support the role of AFs as a signal mediator during stomatal movement and show that in the presence of cytochalasin D the GCs become more responsive to the stimuli, while phalloidin has the opposite effect.
AFs are probably implicated in stomatal movements by regulation of activity and possibly the arrangement of ion channels in the GC plasmalemma (Kim et al., 1995; Hwang et al., 1997, 2000). Cytochalasin D, which promotes the light-induced stomatal opening, potentiates the inward K+ current in GC protoplasts, leading to the enhancement of stomatal opening. By contrast, phalloidin inhibits both light-induced stomatal opening and inward K+ current, resulting in inhibition of stomatal movement (Hwang et al., 1997, 2000). According to Liu & Luan (1998), the disruption of the radial AF arrays under hypotonic conditions, which promotes stomatal opening, is accompanied by the increase of K+ in the GC protoplasts. Therefore, AF disintegration appears to accelerate the increase of GC volume by activation of an inward K+ current. According to Hwang et al. (1997, 2000), the radial AF arrays probably play a negative regulatory role in stomatal movement. When the radial GC AFs are disrupted, the plasmalemma ion channels become less stable and more channels become ready to respond to the hyperpolarized plasmalemma potential (Fig. 15b). The intact radial AFs stabilize the channels in the closed state and, as a result, the channels become less responsive to plamalemma hyperpolarization (Fig. 15a). Hwang & Lee (2001) expressed the view that in a similar way, the AF bundles in the GCs of closed stomata might help to maintain their closed state. Besides, the AF disintegration in response to opening stimuli might allow stomata to open readily and reorganize AFs into the radial pattern.
The changes of AF organization in GCs may affect the inward K+ current by actin-binding proteins or by regulation of some signal mediators. It has been reported that some plant K+ channels display domains for binding to the cytoskeletal elements (Fox & Guerinot, 1998). Moreover, many signaling molecules, such as protein kinases, phospholipases, and G-proteins seem to be associated with AFs in other plant cell types (Putnam-Evans et al., 1989; Ibarrondo et al., 1995). In Arabidopsis thaliana, central components of the ABA-mediated stomatal closure process are probably the small GTP-binding proteins AtRaC1 and AtADF1, which can operate as negative regulators in AF reorganization during stomatal closure induced by ABA (Lemichez et al., 2001; Dong et al., 2001; Eun et al., 2001).
The second messenger Ca2+ probably mediates the signal that reorganizes AFs in moving stomata. It is well known that Ca2+ initiates changes in the stomatal aperture (Assmann, 1993; Evans et al., 2001; Hetherington, 2001; Schroeder et al., 2001) and regulates the activity of many actin-binding proteins (Forscher, 1989). Hwang & Lee (2001) found that the cytosolic Ca2+ levels, protein kinase and protein phosphatase activities mediate the ABA induced AF reorganization in GCs of Commelina communis (Fig. 14). Moreover, Hwang & Lee (2001) suggested that the radial AF disruption involves activation of staurosporine-sensitive kinase(s), whereas the radial AF array formation involves the activity of phosphatase(s) sensitive to calyculin A (Fig. 14).
The recent finding that phosphatidylinositol 3- and 4-phosphate control the opening and closing of the stomatal pore in Vicia faba and Arabidopsis thaliana (Jung et al., 2002), suggests that the phosphoinositide system is involved in the AF reorganization during stomatal movement. Finally, it should be noted that AFs may also regulate the plasmalemma recycling taking place during stomatal movements, by their implication in the exocytotic and endocytotic GC processes (Hwang et al., 2000).
Considering all the above information it can be concluded that the AF dynamics play a regulatory role in movements of the elliptical stomata by modulation of the plasmalemma K+ channel in the GCs. Further work is needed to understand completely the whole mechanism. The AF dynamics should be also examined in functioning dumbbell-shaped stomata.
According to Fukuda et al. (1998), changes of MT organization in GCs of Vicia faba are involved in diurnal stomatal movement (Fig. 16). In the morning, the open stomata display well-organized radial MT systems, but they are broken down in the afternoon and the night when the stomata close. This MT cycle was also confirmed in living GCs of Vicia faba microinjected by fluorescent tubulin (Yu et al., 2001). Recently, Fukuda et al. (2000) described changes in α-tubulin and β-tubulin contents during the diurnal cycle in Vicia faba stomata. Both tubulins were abundant in the morning when the stomata open and display radial MT arrays, but were almost undetectable at midnight when the stomata are closed and display disintegrated MT arrays. It is concluded that the dynamic diurnal changes in GC MT organization and stomatal movement in Vicia faba may be, at least partly, regulated by de novo synthesis and decomposition of tubulin molecules in GCs. These hypotheses are important but should be further substantiated.
ABA, which induces stomatal closure, disrupts the MTs in GCs of Vicia faba but not in the epidermal cells (Jiang et al., 1996; Huang & Wang, 1997a). When the closed stomata open in response to light, the MTs in GCs assume the radial organization. Fukuda et al. (1998), using the anti-MT drug propyzamide, observed MT disintegration and an inhibition of the increase of the stomatal aperture size in early morning, whereas taxol treatment, which stabilizes the MTs, suppressed the decrease of the stomatal aperture in the evening. Yu et al. (2001) presented similar results. Taxol and the MT-depolymerizing drug oryzalin suppressed the light-induced stomatal opening and the dark-induced closure. In addition to the above, the anti-MT drugs colchicine and amiprophos-methyl inhibited the light and fusicoccin-dependent stomatal opening in Tradescantia virginiana and Vicia faba (Couot-Gastelier & Louguet, 1992; Huang & Wang, 1997b; Huang et al., 2000).
Regarding MT function in stomatal movement, the following hypotheses have been made: first the MTs have a regulatory role in stomatal movemement similar to that of AFs (see section VII.I), by controlling the activity of K+ and Ca2+ channels (Huang & Wang, 1997a,b; Zhou et al., 1999; Yu et al., 2001); and second at the beginning of each diurnal cycle, the radial MT arrays are involved in the deposition of radial CM arrays in the periclinal walls of the GC, which in turn are responsible for the opening of the stomatal pore (Fukuda et al., 1998, 2000). The radial CM arrays are disorganized during the closing of the stomatal pore. This view, which is based on the finding that the mature GC walls are actively metabolized and turnover to alter their mechanical properties when the stomata are opened or closed (Kondo & Maruta, 1987; Takeuchi & Kondo, 1988a,b), needs further documentation.
By contrast to the above, Assmann & Baskin (1998) observed in Vicia faba that the stomata opened in light or closed in darkness displayed typical radial MT systems in GCs. Measuring of the stomatal aperture showed that neither colchicine nor taxol affected the responses tested in Vicia faba stomata. The authors conclude that the MTs are not needed for GC function. Eun & Lee (1997), dealing with MT organization in open and closed stomata of Commelina communis, reached the same conclusion. They observed that the radial organization of cortical MTs in illuminated GCs is not affected by incubation of epidermis with ABA or under darkness. In addition, in transgenic plants of Arabidopsis thaliana (WT/MAP4-GFP line) expressing a GFP fusion protein targeted to tubulin, the radial MT organization does not change in stomata closed by ABA (Lemichez et al., 2001).
Recently, Marcus et al. (2001) investigated in vivo the possible implication of GC MTs in stomatal function, using stomatal aperture assays with different MT inhibitors. GCs of Vicia faba expressing in high levels the MT-binding green fluorescent fusion protein (GFP::MT binding domain) are not able to open in response to all the major environmental stimuli triggering stomatal opening. In addition, stomata treated with the anti-MT drugs oryzalin, propyzamide and trifluralin do not open under the same environmental conditions. The inhibitory conditions caused by the GFP::MT binding domain and the anti-MT drugs were reversed by fusicoccin. These data, and particularly the fact that the chimeric protein GFP::MBD, which binds to MTs and disrupts their function, inhibits stomatal opening, support the idea that the mechanism of stomatal movement is MT-dependent, at least in Vicia faba. It is hypothesized that the MTs are required for stomatal opening somewhere upstream to the ionic events (i.e. H+ efflux and K+ influx) that lead to stomatal opening and that the radial MT arrays might participate in the signal transduction via MAPs (Marcus et al., 2001).
Finally, it should be noted that most of the existing information on the cortical MT reorganization during stomatal movement has been derived from the study of one plant, Vicia faba, and that there are contradictory data in different plants. Therefore, the phenomenon should be examined in more plant species, including those possessing dumbbell-shaped stomata.
VIII. Mechanisms of MT organization in stomatal cells
1. General remarks
The determinant aspect of MT cytoskeleton in developing stomatal complexes seems to be the formation of distinct cortical radial MT arrays that line transiently the periclinal walls of the GMCs, SCs and GCs of grasses and permanently the GCs of the elliptical stomata. They appear in critical morphogenetic stages of stomatal complexes. In GCs and in grass SCs the sites of their convergence are paired with an anticlinal MT bundle forming a united MT system. The young stomatal complexes of grasses display four paired radial MT arrays arranged in a strikingly symmetrical pattern (Fig. 7i,j) with the future stomatal pore site at its ‘epicenter’ face (Palevitz & Mullinax, 1989; Palevitz, 1991). The paired radial MT arrays are arranged oppositely in reference to the ‘epicenter’ face, which, according to the above authors, provides a bidirectional transverse signal controlling MT organization on either side. They are the tools by which the shaping of the GCs is coordinated with that of SCs.
The paired radial MT arrays are not restricted to stomatal cells but they are common in differentiating epidermal cells forming sinuous anticlinal walls. They control their morphogenesis defining the deposition of identical CM arrays on the underlying walls (Panteris et al., 1993a, 1994). When the undulation is limited to the external part of the anticlinal wall, the radial MT arrays are connected with anticlinal MT bundles lining that wall region (Panteris et al., 1993b, 1994). At the foci of the radial MTs, local wall thickenings emerge, comparable to those in young GCs (Panteris et al., 1993a,b, 1994). According to Frank & Smith (2002), although the MTs are present, the AFs play the major role in morphogenesis of the sinuous anticlinal walls in Zea mays epidermis, forming patches at the tips of their lobes. However, the morphogenesis of the sinuous epidermal cell walls in Arabidopsis thaliana is a MT-dependent process (Qiu et al., 2002).
Therefore, the involvement of paired radial MT arrays in the morphogenesis of the sinuous epidermal cells and stomatal cells is a rather general phenomenon among plants. In GCs, the paired radial MT arrays are oppositely arranged, while in the epidermal cells they are alternated. The opposite arrangement is crucial for the stomatal pore formation (see section IV), while the alternated one in the epidermal cells prevents the formation of intercellular spaces and leads to the undulation of the anticlinal walls (Panteris et al., 1993a,b, 1994).
The radial MT arrays persist in mature GCs and at least in Vicia faba seem to undergo disorganization-reorganization cycles during stomatal movement (see section VII.2). As far as we know, the mature GC is the only plant cell type provided with such a dynamic cortical MT cytoskeleton. The organization of the MT arrays in differentiating and mature stomatal cells may be the result of the activity of putative cortical MTOCs or of self-ordering MT processes. These hypotheses will be discussed below.
2. MTOC concept
It has been suggested that in GMCs of grasses putative cortical MTOCs operate in a definite succession at the edges formed by the anticlinal and the periclinal walls, as well as at the mid-region of the latter (Galatis, 1982; Cho & Wick, 1989; Cleary & Hardham, 1989). They have been considered to be involved in the formation of: first the interphase MT-band; second the MT arrays lining the periclinal walls; third the endoplasmic MTs converging on the mid-region of the periclinal walls; and fourth the MT-PPB. In preprophase grass SMCs, cortical MTOCs may also function at the sites of MT convergence (Cho & Wick, 1989; Wick, 1991a). In addition, in GMC progenitors of Anemia spp. (Galatis et al., 1986) and in GMCs of Asplenium nidus (Apostolakos et al., 1997) putative cortical MTOCs may operate in the cytoplasm lining the detached regions of the internal periclinal walls, implicated in the organization of distinct cortical and endoplasmic MT arrays. Despite the above mentioned, it is possible that the putative MTOCs in GMCs, SCs and GMC progenitors represent MT nucleation sites only. However, in the case of GCs there are serious indications that the cytoplasmic sites of MT convergence not only nucleate the cortical MTs but also determine the pattern of their organization.
The first view was initially based on TEM observations on nascent dumbbell- and kidney-shaped GCs, which revealed that numerous MTs, some of which form foci, diverging from the junction of the mid-region of the VW with the periclinal walls (Galatis & Mitrakos, 1980; Galatis, 1980; Galatis et al., 1983b; Busby & Gunning, 1984). MT foci were also detected below the periclinal walls of young GCs using high voltage electron microscopy (Palevitz, 1981b). Tubulin immunolabeling in young kidney- and dumbbell-shaped GCs provided further arguments in favour of the idea that putative MTOCs operate in the cytoplasm at the junctions of the VW with the middle of the periclinal walls (Cleary & Hardham, 1989; Cleary et al., 1993; Apostolakos & Galatis, 1999). This view was also supported by experimental work. In GCs recovering from an oryzalin or high-pressure treatment, the MTs reappear around the pore at the junction of the mid-region of the VW with the periclinal walls (Cleary & Hardham, 1990). Moreover, in aberrant Zea mays stomata formed by caffeine treatment, typical radial MT systems form below the periclinal walls even in the complete absence of a VW (Galatis & Apostolakos, 1991).
If the above hypothesis is true, the MTOCs in GCs continuously provide the cell cortex with MTs, which may be separated from MTOCs by the action of a katanin-like protein. In animal cells, katanin is localized in centrosomes, where it seems to cut MTs free of their nucleation sites at the minus ends (McNally et al., 1996; Quarmby, 2000). In GCs these MTs could move along pre-existing MTs, via motor proteins or more probably migrate across the GC cortex by a hybrid treadmilling mechanism (Shaw et al., 2003) to form the paired radial MT systems as well as the MTs below the dorsal GC walls. Alternatively, the MTOCs may provide the GC cortex with MT-initiating factors, that is short MT segments possessing the minus ends, also separated by the action of a katanin-like protein (Wasteneys, 2002; Lloyd & Chan, 2002). The MT-initiating factors could also be transported along pre-existing MTs via motor proteins towards the GC faces to form the MT arrays. Thus, the radial MTs seem to dictate the pattern of MT organization on the periclinal and dorsal GC faces. The anticlinal MT bundles may organize by interaction between MTs emerging from the opposite MTOCs.
The pattern of MT organization in differentiating and functioning kidney-shaped GCs as well as in differentiating dumbbell-shaped GCs is very stable. The MTs in each radial array probably share the same polarity with the minus ends at the focal sites to maintain the integrity of the MT arrays. The MT-converging centers that form in the mitotic spindle and the phragmoplast link MT minus ends to control the predominant direction of elongation and shortening of the MT arrays (Lloyd & Chan, 2002; Wasteneys, 2002). In addition, the anticlinal MT bundle pairing the radial MT arrays, might consist of two MT subsystems, the minus ends of which should be at the sites of MT convergence, while the plus ends function at the opposite sites where the MTs are interdigitated.
The inhibition of GMC division by colchicine produces aberrant stomatal cells forming local wall thickenings at the middle of the periclinal walls, comparable to those of typical GCs, in the absence of MTs (Galatis, 1982). This observation reveals that the GMC cortex at the mid-region of the periclinal walls, where cortical MTOCs may transiently function, are strongly polarized. MTOCs are reactivated in the correspondent GC regions. The establishment of this polarity is probably the primary phenomenon of the GC morphogenesis, a process that seems to commence at the stage of GMC.
According to the second view, in Allium cepa stomata the cortical cytoplasm adjacent to the mid-depth of the VW functions as a ‘planar MT-organizing zone’ (Marc et al., 1989a,b). In aberrant stomata of Allium cepa produced by CIPC, cytochalasin D and caffeine treatments Marc & Palevitz (1990) noted that the ‘planar MT-organizing zone’ not only initiates MTs, but also determines their spatial order throughout the cell cortex. γ-Tubulin has been also localized at this cytoplasmic zone (McDonald et al., 1993). However, the localization of γ-tubulin, at least in higher plant cells, is not a reliable criterion to assess the position of the MTOCs. γ-Tubulin is localized along the whole length of MTs in every cell-specific stage MT array as well as in colchicine-tubulin paracrystals (Panteris et al., 2000 and literature therein). For localization of MTOC sites in stomatal cells, more specific MTOC-markers like the Spc98p orthologs (Erhardt et al., 2002) have to be used.
Although the function of a ‘planar MT-organizing zone’ is well documented in Allium cepa stomata, it cannot explain: first the organization of radial MT arrays below the periclinal walls in aberrant grass stomata, which lack a VW (Galatis & Apostolakos, 1991); second the appearance of the radial MT arrays under the periclinal walls in Asplenium nidus GCs before the organization of the anticlinal MT bundle along the middle of the VW (Apostolakos & Galatis, 1999); and third the differences in MT organization between the periclinal faces of GCs in grasses and other plants (Cleary & Hardham, 1989; Cleary et al., 1993).
The critical shifts in the cortical MT organization in stomatal cells show that the activation of the cortical MTOCs is under strict temporal and positional control. MTOCs may be activated in stomatal cells at the sites where the cell cortex is mechanically stressed by local expansion or by the local wall detachment (Galatis, 1982; Galatis et al., 1983b, 1986; Apostolakos et al., 1997). Examples of the former case are the mid-region of the periclinal walls of the GMC (Fig. 9a) as well as the junction of the mid-region of the GC periclinal walls with the VW (Fig. 12a,b). Frequently, in these regions the cell wall appears thinner than the rest of the walls. The deposition of callose platelets in such areas of fern stomata (Waterkeyn & Bienfait, 1979; Apostolakos et al., 1997; Apostolakos & Galatis, 1999) suggests that they are probably mechanically stressed. Callose is consistently deposited at ‘wounded’ wall regions. Besides, in the protodermal cells of the liverwort Marchantia paleacea putative MTOCs function at sites, where wall detachment takes place (Apostolakos & Galatis, 1985a).
Animal cells being under mechanical stress reinforce their cortical cytoskeleton by polymerization of new elements to protect the plasma membrane. Polymerization takes place at the regions where the mechanical forces are applied (Ingber, 1997; Kano et al., 2000; Ko & McCulloch, 2000). A similar mechanism seems to function in plant cells experiencing hyperosmotic stress (Komis et al., 2002a,b). These findings support the view that the cortical cytoplasm of the stomatal cells seems to have the ability to polymerize MTs in response to mechanical stresses.
The paired radial MT arrays, by means of motor proteins, may also be involved in preferential transfer of dictyosome vesicles to the site of their convergence initiating the local wall thickenings. In this way, more CM synthases are delivered at these regions. However, the deposition of these local wall thickenings is not exclusively dependent on MTs. In colchicine-treated GMCs and GCs, comparable thickenings are deposited in the absence of MTs (Galatis, 1977, 1982). Other factors, which colocalize with MTs in stomatal cells, promote the local wall deposition, when the MTs are absent.
3. MT self-ordering hypothesis
In an alternative view, the cytoplasm at the focal regions of the radial MT arrays in stomatal cells does not nucleate MTs but is implicated in their organization only. The MTs are nucleated elsewhere in the cell cortex and by a self-ordering process involving MAPs and motor proteins organize the radial arrays below the periclinal walls as well as the MT bundles pairing them. The findings that in dividing higher plant cells perinuclear MTs and MTs nucleated around or among chromosomes are transformed by self-organization processes into bipolar prophase and metaphase spindles, respectively, support the above hypothesis. This trasformation is probably carried out by the integrated function of motor proteins and MAPs (Asada & Colling, 1997; Canaday et al., 2000; Vantard et al., 2000; Lloyd & Hussey, 2001; Reddy, 2001; Schmit, 2002; Wasteneys, 2002). The minus ends of the spindle MTs are localized at the pole regions, where they are focussed.
Two motor proteins, the C-terminal kinesin-like protein KCBP and KatA, B, C and D and the N-terminal kinesin-like protein TKRP-125, as well as two MAPs, the MAP65 (Smertenko et al., 2000) and MOR1 (Whittington et al., 2001), isolated from plants, are implicated in MT organization in dividing cells (Reddy, 2001; Hepler et al., 2002; Schmit, 2002; Wasteneys, 2002). Among them, of particular importance for MT convergence, is the KCBP, the activity of which is controlled by the Ca2+/calmodulin complex (Vos et al., 2000; Reddy, 2001; Hepler et al., 2002). The other proteins seem to participate in the formation of the interdigitations between the plus ends of the antiparalell MTs in the mitotic spindle and phragmoplast (Reddy, 2001; Lloyd & Hussey, 2001; Wasteneys, 2002). The same motor proteins are involved in the organization of the MT arrays, controlling hair cell morphogenesis in Arabidopsis thaliana (Reddy, 2001). Stress-activated mechanisms may also be responsible for activation and/or localization of motor proteins and MAPs in the cortical cytoplasm below the periclinal walls in stomatal cells.
The information presented in this review shows that the cytoskeleton controls all the major morphogenetic phenomena of stomatal complexes. The individual stomatal cells have developed impressive strategies to control their shape during the successive morphogenetic stages, which are coupled with stomatal pore and substomatal cavity formation. The position and the extent of wall deposition/expansion are regulated not only on the level of a single cell wall, but also on the level of the stomatal complex, via cortical MT arrays that determine accurately CM orientation.
Considering the highly controlled cell expansion of grass stomata, it may be assumed that, parallel to the accurate regulation of the CM orientation, some wall-loosening factor(s), possibly expansins, are preferentially forwarded to the expanding wall regions. The cytoskeletal elements directing the exocytosis of dictyosome vesicles might control such an activity. Alternatively, the cytoskeleton may be implicated in activation of wall-loosening processes in the locally expanding regions.
To perform their role the cortical cytoskeletal elements in stomatal cells form, at critical stages of morphogenesis, a variety of highly organized arrays, like the MT- and the AF-PPBs, the AF patches, the interphase MT bands, the paired radial MT arrays, the radial AF arrays, etc. The paired radial MT arrays are the key elements of GC morphogenesis. Probably, they coexist with paired radial AF arrays at least in kidney-shaped GCs. Their organization mirrors the establishment of a stable polarity in particular cortical regions adjacent to the periclinal wall. This polarity is probably the primary morphogenetic phenomenon of stomata and persists in mature ones.
Further research work is needed to find out the factor(s) inducing the above polarization, its nature and how the MTOCs are activated and/or the MT organizing proteins are localized at the sites where the radial MTs converge. In this direction, the role of protein-motors and MAPs in MT organization in stomatal cells, using proper inhibitors as well as mutants, in which the genes coding them are inactive or overexpressed, must be examined. Of particular interest is the spike 1 mutant of Arabidopsis thaliana. The products of the SPIKE 1 gene control MT organization in typical epidermal cells (Qiu et al., 2002). It could also be important to study the development of stomatal complexes in ton mutants of Arabidopsis thaliana, the meristematic cells of which lack MT-PPBs (Camilleri et al., 2002).
Moreover, the mechanism of organization and function of cortical AFs during morphogenesis of stomatal complexes, including the dumbbell-shaped ones, should be further examined. The existing information is fragmentary and often inconclusive. In particular, it should be investigated whether the cortical AF patterns are controlled by the correspondent cortical MT patterns, or vice versa as well the role of proteins, like the ADF or the Rop GTPase in AF organization. In this direction, the use of proper mutants of Arabidopsis thaliana could be useful. The overexpression of the gene producing an ADF protein disorganizes the AFs in mature GCs of Arabidopsis thaliana (Dong et al., 2001).
As far as the involvement of cortical AFs in stomatal movement is concerned, the available data substantiate adequately their participation in movement of the elliptical stomata, through the control of ion channel function in the plasmalemma. However, there is still controversy on the role of the cortical MTs in this process. Their changes during stomatal movement should be examined in more plants, including those possessing dumbbell-shaped stomata.
We want to thank the referees for their valuable comments, Dr C. Katsaros for the critical reading of the manuscript and M. Zachariadis for his assistance in preparation of the plates.