The morphogenesis of lobed plant cells in the mesophyll and epidermis: organization and distinct roles of cortical microtubules and actin filaments



The morphogenesis of lobed plant cells has been considered to be controlled by microtubule (MT) and/or actin filament (AF) organization. In this article, a comprehensive mechanism is proposed, in which distinct roles are played by these cytoskeletal components. First, cortical MT bundles and, in the case of pavement cells, radial MT arrays combined with MT bundles determine the deposition of local cell wall thickenings, the cellulose microfibrils of which copy the orientation of underlying MTs. Cell growth is thus locally prevented and, consequently, lobes and constrictions are formed. Arch-like tangential expansion is locally imposed at the external periclinal wall of pavement cells by the radial arrangement of cellulose microfibrils at every wall thickening. Whenever further elongation of the original cell lobes occurs, AF patches assemble at the tips of growing lobes. Intercellular space formation is promoted or prevented by the opposite or alternate, respectively, arrangement of cortical MT arrays between neighboring cells. The genes that are possibly involved in the molecular regulation of the above morphogenetic procedure by MT and AF array organization are reviewed.


Plant cell morphogenesis is the outcome of cell wall shaping in concert with cell growth and differentiation. Plant cell growth requires precise morphogenetic control to produce a shape appropriate enough for the functionality of each tissue for the rest of cells’ life. Morphogenetic pattern is imposed by the organization of the cytoskeleton in various plant cell types, either diffusely growing (for reviews see Lloyd, 1987; Seagull, 1989; Williamson, 1991; Mathur & Hülskamp, 2002; Smith, 2003) or exhibiting tip growth (Kropf et al., 1988; Hepler et al., 2001; Mathur & Hülskamp, 2002; Smith, 2003). Similarly, it is the cytoskeleton that controls morphogenesis in parenchymatic tissues such as the mesophyll and epidermis of the leaves, the cells of which exhibit a rather lobed shape.

According to direct observations on control as well as drug-treated plant material, lobed cell morphogenesis results from the organization of cortical microtubules (MTs) and the concomitant local reinforcement of the cell wall. However, recent studies on various mutants defective in cell morphogenesis emphasize the primary role of actin filaments (AFs) in lobe formation (for reviews see Mathur & Hülskamp, 2002; Smith, 2003; Wasteneys & Galway, 2003). This article reviews the existing information on the implication of the cytoskeleton in lobed epidermal and mesophyll cell morphogenesis and proposes a comprehensive mechanism in which MTs and AFs play distinct roles. This consideration also includes the genetic and molecular regulation of morphogenesis, as has been recently revealed by the study of mutants.

Morphogenesis of lobed mesophyll cells

Mesophyll morphogenesis includes the formation of cylindrical cells of palisade parenchyma and lobed aerenchyma-like cells of spongy parenchyma and of the mesophyll of other plants such as grasses, conifers and ferns. Since mesophyll has to carry out photosynthesis, cell shaping is coordinated with the opening of a vast reticulum of intercellular spaces to facilitate the exchange of gases.

The mechanism of lobed cell morphogenesis was first studied in epithem cells of hydathodes in Pilea cadierei leaves (Galatis, 1988): cortical MTs form ring-shaped bundles, which interconnect and create a reticulate scaffold of MT bundles. Local thickenings are deposited on the cell wall, copying the pattern of MT bundles. The cellulose microfibrils (CMFs) in these wall thickenings are parallel to the underlying MTs. This elaborate cell wall reinforcement surrounds the protoplast like a ‘swaddling band’ that defines areas where cell growth is locally prevented. As a result, the cell bulges in the areas between the local wall thickenings while it remains constricted at the latter, forming lobes. As the MT bundles and the concomitant local cell wall thickenings are usually oppositely arranged between neighboring cells, intercellular spaces open at the sites of cell constrictions during cell growth.

The mechanism described is also the one that controls the morphogenesis of lobed mesophyll cells in the leaves of the grasses Triticum aestivum (Jung & Wernicke, 1990; Wernicke & Jung, 1992) and Zea mays (Apostolakos et al., 1991; see also Fig. 1a) the dicot Nigella damascena (Wernicke et al., 1993) and the fern Adiantum capillus-veneris (Panteris et al., 1993a,c). The same mechanism is also responsible for spongy mesophyll morphogenesis (unpublished) and for the formation of ‘infolded’ mesophyll cells of Pinus sylvestris (Hoss & Wernicke, 1995). In most of the cases mentioned, the MT and cell wall reinforcement pattern is opposite between neighboring cells, producing an extended reticulum of intercellular spaces.

Figure 1.

Confocal-laser-scanning microscope (CLSM) images of lobed mesophyll cells of Zea mays after immunostaining of microtubules (MTs) (a) and visualization of actin filaments (AFs) with fluorescent phalloidin (b). Cortical MTs form ring-like bands (a), while AFs (b) exhibit diffuse arrangement. Bar, 20 µm.

In A. capillus-veneris leaflets, where the mesophyll is rather an aerenchymatic tissue, a second stage of growth occurs, as some of the cell lobes elongate and form lateral branches, which further widen the intercellular spaces in the mesophyll (Panteris et al., 1993a). During this stage, a ring of transverse cortical MTs is organized at the base of these cell lobes, followed by the deposition of an identical ring of CMFs in the cell wall, establishing new axes of cell growth (see Fig. 3d; Panteris et al., 1993a). At the same time, AF patches line the domes of elongating lobes (Fig. 2a). Patches of cortical AFs assemble in developing stomatal complexes (Galatis & Apostolakos, 2004), where they might be correlated with some protection of the plasmalemma against mechanical stresses, as is the case for animal cells (Ingber, 1997; Frame & Sarelius, 2000; Ko & McCulloch, 2000), and also in plasmolysed leaf cells of Chlorophytum comosum where they reinforce detached regions of the plasmalemma (Komis et al., 2002). The AF assembly might also be correlated with the targeted transport of cell wall materials (Baskin & Bivens, 1995) and/or wall loosening factors. Some kind of local tip-like growth of the cell lobes cannot be excluded during this stage (Fu et al., 2002; Smith, 2003).

Figure 3.

Drawing presenting the successive stages of mesophyll (a–d) and pavement (e–h) cell morphogenesis. Microtubules (MTs) are shown in green, cellulose microfibrils (CMF) in blue and actin filament (AF) patches in red. In (a–d) cell wall thickenings, with CMFs parallel to MTs, are not shown. (a) Young mesophyll cell; the cortical MTs form interconnecting ring-like bundles. (b) As mesophyll cells grow, lobes and constrictions are created. New MT bundles also form. (c) Further mesophyll cell growth leads to formation of additional lobes. (d) Microtubule rings form at the base of growing mesophyll cell lobes, while AF patches (red lines) assemble at their domes. New axes of cell growth (arrows) are established. Apart from the AF patches, AFs exhibit diffuse pattern of organization and are not depicted. (e) Cortical MTs form bundles that line the anticlinal walls and radial MT arrays at the junctions of the external periclinal with the anticlinal walls in a young pavement cell. (f) Later morphogenetic stage of a pavement cell. Local cell wall thickenings with CMFs parallel to the underlying MTs are deposited. Arch-like tangential expansion (arrows) is imposed on the external periclinal cell face. (g) Lobes and constrictions are initiated in this morphogenetic stage of pavement cells, while AF patches (red lines) assemble at the domes of growing cell lobes. (h) Paradermal view of an epidermal area during cell lobe formation and expansion. Note that cortical MT arrays, wall thickenings with CMFs and AF patches are alternate between neighboring pavement cell resulting in a ‘jigsaw-puzzle’ top view.

Figure 2.

Visualization of actin filaments (AFs) with fluorescent phalloidin in lobed mesophyll (a) and pavement (b) cells of Adiantum capillus-veneris. Apart from the overall diffuse organization, AFs form patches (arrows in a and b) that line the dome of growing lobes. The arrowhead in (b) points to the external periclinal cell face. Bar, 20 µm.

In several of the above plants, experimental disruption of MTs with colchicine (Apostolakos et al., 1991; Panteris et al., 1993a,c) or oryzalin (Wernicke & Jung, 1992) severely affects morphogenesis, further proving the role of cortical MT organization in the morphogenetic process. In the presence of these drugs, mesophyll cells do not form lobes (Apostolakos et al., 1991; Wernicke & Jung, 1992; Panteris et al., 1993a,c) while opening of intercellular spaces in the mesophyll is greatly reduced (Apostolakos et al., 1991; Panteris et al., 1993a,c).

In T. aestivum mesophyll cells, AFs have also been observed to form bundles, coincident to the MT bundles (Jung & Wernicke, 1991). In these cells, cytochalasin D (CD) treatment prevented MT bundle formation although AFs did not totally disappear, resulting in abnormal mesophyll morphogenesis (Wernicke & Jung, 1992). The above finding suggests an indirect role for AFs in mesophyll cell morphogenesis by controlling MT organization. However, similar observations have not been repeated in the mesophyll of other plant species. In mesophyll cells of Z. mays, AF organization does not follow the pattern of MT bundling (Fig. 1b; cf. Fig. 1a) and CD treatment does not affect mesophyll cell morphogenesis (unpublished). In addition, brick mutants of Z. mays, although defective in AF organization, display normally lobed mesophyll cells, compared with the mesophyll of wild-type leaves (Frank & Smith, 2002; Frank et al., 2003). Therefore, further research is needed to elucidate the possible role, if any, of AFs in the control of mesophyll cell morphogenesis.

In conclusion, cortical MT bundles are the major and very efficient tool for the morphogenesis of lobed mesophyll cells by the following procedure: MT bundle organization; cell wall reinforcement; control of local cell growth; formation of lobes and constrictions; opening of intercellular spaces (Fig. 3a–c). As is considered to be the rule for plant cells (Baskin, 2001), in the lobed mesophyll cells the control of CMF orientation by that of cortical MTs is pivotal. For the mesophyll of A. capillus-veneris especially, MT rings and AF patches form at the base and the dome, respectively, of every initial cell lobe that elongates further to form a branch (Fig. 3d).

Morphogenesis of ordinary epidermal cells

The ordinary cells of leaf epidermis, also named ‘pavement’ cells, exhibit in several plant species a wavy (sinuous) contour of their anticlinal walls (Esau, 1965; Mauseth, 1988). Rectangular or polygonal protodermal cells become wavy as they grow and differentiate, following a highly coordinated sequence of morphogenetic steps, in which the cytoskeleton is implicated. Although pavement cell protrusions are not quite similar to mesophyll cell lobes, the term ‘lobe’ is also applied for simplicity and uniformity.

Cortical MTs have been considered to be the cytoskeletal component that initiates the morphogenetic mechanism of wavy pavement cells, their role been first described in detail in the leaves of Vigna sinensis (Panteris et al., 1993b): while the anticlinal walls of young pavement cells are still straight, cortical MTs organize into anticlinal bundles under them (Fig. 3e,f). In every cell, the anticlinal MT bundles interconnect through MT bundles that line the internal periclinal cell wall. In addition, every anticlinal MT bundle terminates at the junction of the anticlinal with the external periclinal cell wall to a radial MT array (Fig. 3e,f). The MTs of these radial arrays fan under the external periclinal cell wall (Fig. 3e–g). As a result of MT organization, the cell wall is reinforced with local thickenings, the CMFs of which are parallel to the underlying MTs. In particular, over the radial MT arrays, cell wall pads (see Fig. 4a) with radially arranged CMFs are deposited. As the CMFs of wall pads of each cell interconnect at their lateral proximities, a frame of reinforcement is determined (Fig. 3f–h) on the external periclinal cell wall. This frame extends through anticlinal CMF bands and surrounds the whole cell periphery.

Figure 4.

Pavement cells of wild-type (a) and brick 1 (b) Zea mays plants after staining of the cell wall with Calcofluor White. The external periclinal cell wall is recognized by its intense fluorescence and is marked by asterisk in both figures. Wall pads (arrowheads in a) can be seen at the junctions of the external periclinal cell wall with the anticlinal ones in wild-type but not in the brick 1 pavement cell (compare a with b). A mesophyll cell overlaps on the external periclinal pavement cell wall in (a). Bar, 20 µm.

Similar to what happens in mesophyll cells, the local cell wall thickenings in the anticlinal walls prevent cell growth locally, resulting to the initiation of lobes and constrictions. The radially arranged CMFs of the pads, at the junctions of the anticlinal cell walls with the external periclinal one, dictate locally an arch-like tangential expansion pattern of the external periclinal wall (Fig. 3f), similar to that occurring in the morphogenesis of kidney-shaped guard cells of stomata (Galatis & Apostolakos, 2004). The combination of the precisely controlled growth at the anticlinal cell walls and the external periclinal wall results in the initial formation of the wavy contour of pavement cells, the lobes of which grow mostly in the plane of the epidermis and not towards the mesophyll. The pattern of MT organization and corresponding cell wall reinforcement are always alternate between neighboring cells, which results in the formation of a jigsaw puzzle top view of the epidermis (Fig. 3h). Consequently, as every cell lobe is indented in a neighboring cell constriction, intercellular spaces between pavement cells never form. In general, it seems that the formation of intercellular spaces between epidermal cells or the prevention of their opening is a matter of the arrangement of radial MT arrays between neighboring cells: in stomata guard cells they are opposite and the stomatal pore is formed, while in pavement cells they are alternating (Fig. 3h), preventing the opening of any intercellular space between them (Galatis & Apostolakos, 2004).

This morphogenetic process is evolutionarily conserved in several plant species (Panteris et al., 1994). The organization of radial MT arrays under the external periclinal cell wall (in certain rare cases under both periclinal walls) is a common feature, while anticlinal MT bundle organization varies between different plant species, producing a different extent of waviness in every case (Panteris et al., 1994). Since 1994, several authors have confirmed MT bundle organization at the anticlinal pavement cell faces (Wasteneys et al., 1997; Frank & Smith, 2002; Fu et al., 2002; Qiu et al., 2002; Frank et al., 2003). However, no attention has been paid to the radial MT arrays that organize at the junctions of the anticlinal walls with the external periclinal one. Nevertheless, according to the above mechanism, pavement cell morphogenesis is not just ‘attributed to alternate bundling and absence of MTs on the anticlinal cell faces’ (Wasteneys et al., 1997), but the radial MT arrays have been considered to play the primary role in early pavement cell morphogenesis (Panteris et al., 1993b, 1994). In fact, the anticlinal MT bundles play an auxiliary role, extending the waviness along the depth of the anticlinal cell walls. This is quite obvious in Cyperus papyrus leaves, where the wavy contour of pavement cells is confined almost exclusively near the external periclinal cell wall and the anticlinal cell walls remain straight at deeper levels of the epidermis. While radial MT arrays are regularly organized in these cells, the anticlinal MT bundles are quite short, extending only a little at anticlinal cell faces (Panteris et al., 1994).

Apart from the variations in MT organization mentioned in the preceding text, it is also important that in several plants, such as Begonia lucerna, waviness of pavement cells is not a general feature of every leafy structure: while leaf pavement cells possess straight anticlinal walls, petal pavement cells appear wavy, as the morphogenetic mechanism for cell waviness, although functional in petals, remains dormant in the leaves of the same plant (Panteris et al., 1994). In addition, pavement cell waviness, wherever present, varies significantly between different sites of leaves and petals in several plant species and the degree of waviness is affected by environmental conditions (Watson, 1942; Esau, 1965). Accordingly, any possible structural and/or functional advantage of a wavy cell contour is not easy to understand. Interestingly, defects in pavement cell lobing in spike 1 (Qiu et al., 2002), quasimodo 1 (Bouton et al., 2002) and scd 1 (Falbel et al., 2003) mutants of A. thaliana are combined with the presence of ‘gaps’ in the epidermis. It can therefore be suggested that waviness of the anticlinal pavement cell walls, in the plants in which it is observed, is important for maintaining the integrity of the epidermis. However, the above mutations, apart from affecting cell shape, may also interfere with the composition and adhesive properties of pavement cell walls (Szymanski, 2005). Accordingly, the ‘gaps’ between pavement cells may be a side-effect and not result from a failure in cell shaping.

Recent studies on wild-type plants, as well as on mutants defective in cell morphogenesis, revealed that assembly of cortical AF aggregations/patches (Frank & Smith, 2002; Fu et al., 2002; Qiu et al., 2002; Frank et al., 2003) at the tips of growing lobes is required for normal lobe formation in pavement cells but not in mesophyll cells. According to the data from mutants, in which such AF patches do not form, pavement cell morphogenesis is considered to be mainly the outcome of AF organization and not of cortical MTs (Frank & Smith, 2002; Fu et al., 2002; Frank et al., 2003; Smith, 2003; Wasteneys & Galway, 2003). However, in all these cases, AF patches appear at the tips of wild-type pavement cell lobes only after the latter have already been initiated (see earlier). What is more, transverse extension of the waviness along anticlinal pavement cell walls cannot be attributed to AF patches. The exact site of AF patch assembly is the junction of the external periclinal cell wall with the anticlinal ones, where maximum waviness occurs and cell lobes exhibit ‘tips’ (see earlier). As is demonstrated by the case of C. papyrus pavement cells (Panteris et al., 1994), waviness along anticlinal cell walls is determined by the length of anticlinal MT bundles. Comparison of cortical MT organization with the occurrence of AF patches leads to the assumption that the former is responsible for the initial formation of the wavy pattern, while AF patches participate in late cell lobe growth and extension. The observations described in the following text further support this opinion.

The pavement cells of colchicine-treated leaves of V. sinensis (Panteris et al., 1993b) and Z. mays (Fig. 5a), which do not contain any MTs at all, completely fail to become wavy. In addition, in spike 1 mutants of A. thaliana, the cortical MTs of pavement cells remain almost uniformly arranged (Qiu et al., 2002). Obviously, MT bundles and radial arrays are not organized and, subsequently, pavement cells do not become wavy. It can therefore be concluded that cortical MT organization is the primary and the major event in pavement cell morphogenesis: in the absence of cortical MTs, local cell wall reinforcement does not occur and, as a result, the pavement cells grow almost uniformly without any constrictions or protrusions (Panteris et al., 1993b). In pavement cells of brick mutants of Z. mays, cortical MT bundles have been observed, although these are less distinct than those in wild-type leaves (Frank & Smith, 2002; Frank et al., 2003). However, radial MT arrays are not organized and wall pads with radial CMFs are not deposited at the junctions of the external periclinal wall with the anticlinal ones, which is the rule for wild-type pavement cells (Fig. 4a; cf. Fig. 4b). It can therefore be supported that brick pavement cells fail to become wavy because they lack these radial CMF systems. This further underlines the significance of radial MT arrays and concomitant radial CMF systems in wavy pavement cell morphogenesis: anticlinal MT bundles, although present in brick mutants, seem not to be enough for pavement cell shaping. Last but not least, it seems conceivable that BRICK protein, which participates in AF assembly (Frank & Smith, 2002; Frank et al., 2003; Szymanski, 2005), might be directly or indirectly involved in the organization of the radial MT arrays in Z. mays pavement cells. However, the molecular nature of such an involvement remains unknown.

Figure 5.

Differential interference contrast (DIC) images of living epidermides. (a) Pavement cells of Zea mays treated with 0.08% colchicine for 72 h. No waviness at all can be observed in these randomly swollen cells. Vigna sinensis (b) and Zea mays (c) pavement cells treated with 20 µm cytochalasin D for 72 h. The cells exhibit wavy pattern, although not as extended as in control material. Bar, 20 µm (a), 20 µm (b), 40 µm (c).

According to the previous interpretation, the failure of pavement cells in brick mutants of Z. mays to become wavy is not primarily due to a defect in AF patch assembly. The AF patches seem to be responsible not for the initial formation but for further pavement cell lobe elongation: since such lobes do not develop in brick pavement cells, AF patches do not form. Apart from the observations in brick mutants of Z. mays, in DN-rop2 mutants for ROP2-GTPase (Fu et al., 2002), and in NAPP and PIRP mutants (Brembu et al., 2004) of A. thaliana, also showing defects in AF organization, the pavement cells become wavy but cell lobes are shorter than in wild-type plants. Similarly, pavement cells of Z. mays and V. sinensis leaves treated with CD become wavy (Fig. 5b,c) but the cell lobes are shorter than in control cells.

Summarizing the data so far, in the absence of AF aggregations/patches from the tips of cell lobes, pavement cells develop a less wavy shape. Oppositely, when radial MT arrays are not organized or when MTs are totally absent, pavement cells do not become wavy at all. The AF patches at the domes of growing lobes of pavement cells might play similar role(s) as in the mesophyll cells of A. capillus-veneris (see the section ‘Morphogenesis of lobed mesophyll cells’). It is possible that some sort of temporally and locally restricted tip-like growth occurs as pavement cell lobes elongate (Fu et al., 2002; Smith, 2003). Further experiments are required in order to prove such a possibility.

In conclusion, pavement cell morphogenesis is carried out in three distinct steps (Fig. 3e–h). First, the cortical MTs are organized into bundles and radial arrays, creating a ‘scaffold’ for local cell wall deposition (Fig. 3e). Second, the cell walls are locally reinforced, mimicking the pattern of cortical MTs. Formation of the wavy anticlinal cell wall contour is initiated as a result of this local cell wall reinforcement (Fig. 3f). Third, AF patches assemble at the tips of the initiated cell lobes, which correlate with further lobe elongation and, subsequently, further extension of the overall wavy pattern of the epidermis (Fig. 3g,h).

Imposition of morphogenetic pattern: the mesophyll–epidermis paradigm in A. capillus-veneris

In A. capillus-veneris leaflets, MT bundle organization is initiated at the mesophyll cells in the middle of the intercostal area (Panteris et al., 1993a). Once initiated, the pattern of MT bundling is transferred in a cell-to-cell relay race manner, assuring the opposite arrangement of MT bundles within the whole tissue and the concomitant opening of a sophisticated labyrinth of intercellular spaces.

In these leaflets, mesophyll cells adjacent to the vascular bundles are semilobed (Panteris et al., 1993c): the cortical MTs organize into U-shaped bundles under the cell walls facing the mesophyll, while they retain a uniform transverse arrangement at the cell wall facing the endodermis. Cell wall reinforcement copies the pattern of cortical MTs, resulting in the formation of mesophyll cells with lobes and constrictions only on their surface facing the mesophyll. It seems that these semilobed cells are subject to the induction of two different morphogenetic influences, one from the mesophyll and another from the adjacent vascular bundle. As a consequence, two different patterns of cortical MT organization are followed, resulting in this peculiar morphogenesis.

Pavement cells of normal A. capillus-veneris leaflets, although exhibiting wavy contour of their anticlinal walls, follow a morphogenetic procedure different from that of other plant species (Panteris et al., 1993c, 1994). In these cells, radial MT arrays do not organize at the junctions of the external periclinal wall with the anticlinal ones. Instead of this, U-shaped MT bundles form, lining the anticlinal cell walls and the internal periclinal one. As for semilobed mesophyll cells, it is the mesophyll that dictates the pattern of cortical MT organization in pavement cells: the pattern of MT bundle formation in pavement cells mirrors that of the MT bundles in the underlying mesophyll cells and, consequently, the MT bundles that line the inner cell wall are always opposite to those of the mesophyll cells (Panteris et al., 1993c). As a result of this pattern of MT organization, the pavement cells of A. capillus-veneris form lobes towards the mesophyll. Intercellular spaces also open between the epidermis and the mesophyll to facilitate gas exchange, while the chloroplasts of pavement cells gather in these lobes, as the epidermis is a major photosynthetic tissue in this plant (Panteris et al., 1993c). As pavement cell lobes grow further towards the mesophyll, AFs line their domes (Fig. 2b), similar to what happens in the tips of pavement cell lobes of other plant species (see the section Morphogenesis of ordinary epidermal cells) and in elongating mesophyll cell lobes of the same plant (Fig. 2a). As already mentioned, AFs in these sites seem to participate in cell lobe elongation.

While being opposite between pavement and mesophyll cells, the U-shaped MT bundles shift to a strictly alternating arrangement between pavement cell anticlinal walls. This arrangement prevents the opening of any intercellular space between pavement cells at the proximity of the external periclinal cell wall and, as a consequence, the epidermis remains integral (Panteris et al., 1993c). However, while in radial MT array-dependent morphogenesis maximum waviness occurs at the external pavement cell wall, in A. capillus-veneris semilobed pavement cells, waviness is initiated and maximized at the proximity of the mesophyll (Panteris et al., 1994).

Pavement cells over vascular bundles are neither wavy nor lobed, which is a further evidence for the role of the mesophyll in imposing the pattern of MT organization in the epidermis. In addition, in leaflets of A. capillus-veneris grown in tissue culture, which lack mesophyll and consist only of the two epidermides and the vascular bundles, pavement cells do not grow inwards to form lobes but follow the morphogenetic procedure of wavy pavement cells of other plants (Panteris et al., 1994). It seems that, in normal leaflets of A. capillus-veneris, epigenetic influence of the mesophyll is capable of altering a morphogenetic mechanism that is common in plants with wavy pavement cells (Panteris et al., 1994).

Morphogenetic influence of the mesophyll on the epidermis has also been observed in brick1 wild-type mosaic Z. mays leaves (Frank et al., 2003). Pavement brick1 cells, which overlie wild-type mesophyll cells, become wavy almost like those of wild-type plants. It seems that BRICK protein can affect cell morphogenesis as it is transferred from one cell layer to the other (Frank et al., 2003). Since pavement cells of brick mutants lack radial MT arrays at the junctions of the external periclinal wall with the anticlinal ones (see the section Morphogenesis of ordinary epidermal cells), it could be well expected that BRICK protein, transferred from the mesophyll to the epidermis, promotes the organization of such MT arrays, thus rescuing the wild-type phenotype. Intercellular trafficking of regulatory proteins, probably through plasmodesmata, from the mesophyll towards the epidermis but not in the other direction, has been shown to occur in the leaves of A. thaliana (Kim et al. 2003). While in Z. mays the influence of underlying mesophyll rescues the wavy pavement cell phenotype (Frank et al., 2003), in A. capillus-veneris the mesophyll alters pavement cell morphogenesis (Panteris et al., 1994). This can be attributed to the difference in the role of the epidermis between these two plant species: pavement cells of A. capillus-veneris, but not of Z. mays, are the major photosynthetic tissue of the leaflet, as in other ferns (see also Nasrulhaq-Boyce & Duckett, 1991), and, subsequently, their morphogenesis has to serve mainly this function apart from reassuring the integrity of the epidermis.

Molecular control of the organization of the cytoskeleton for lobed cell morphogenesis

Cortical MT organization for mesophyll and pavement cell morphogenesis requires MT reorientation, formation of MT bundles and radial MT arrays and, in certain cases, further extension of the primary MT pattern (Panteris et al., 1993a,b,c, 1994). Initiation of MT bundles and radial arrays occurs at special sites of the cortical cytoplasm (Panteris et al., 1993a,b,c). Formation of additional MT bundles in the mesophyll cells of A. capillus-veneris is initiated at the already existing MT bundles (Panteris et al., 1993a), in accordance with the statement that ‘microtubules beget microtubules’ (Lloyd & Chan, 2002). It seems that the above MT arrays do not form by the reorganization of preexisting MTs but that new MT assembly occurs. Differential expression of α-tubulin genes during lobed mesophyll cell development in Hordeum vulgare (Hellmann & Wernicke, 1998; Schröder et al., 2001) further supports this view (see also Schröder et al., 2002). These observations are consistent with recent data from studies on cortical MT organization. Although the following data originate from studies in cells with more or less uniform cortical MT distribution, they may well fit for MT bundle and radial array formation.

Recently, factors participating in MT nucleation, such as the homologue of animal EB 1 centrosomal component (Chan et al., 2003) and large γ-tubulin complexes (Drykováet al., 2003), have been detected in the cortical cytoplasm or possibly localized to the plasma membrane, respectively, of plant cells. Apart from the above observations, the presence and function in plant cells of proteins homologue to animal katanin (McClinton et al., 2001; Stoppin-Mellet et al., 2002) supports the ability for MT formation into precisely organized arrays directly in the cortical cytoplasm. In accordance, the defects in growth of A. thaliana katanin mutants botero 1, fragile fiber 2 and ectopic root hair 3 (Bichet et al., 2001; Burk et al., 2001; Webb et al., 2002) are attributed to failure in cortical MT organization. Cortical MTs have been shown to be highly dynamic (Shaw et al., 2003), which is a prerequisite for direct cortical MT array initiation and extension. Recent in vitro analysis has confirmed that cortical MTs organize by random nucleation and selective stabilization (Tian et al., 2004). Factors that stabilize cortical MTs, such as MOR 1 (Whittington et al., 2001), and/or form intermicrotubule bridges that vary in length (Chan et al., 1999; Lloyd & Chan, 2002; Yasuhara et al., 2002) may play important roles in MT bundle organization for lobed cell morphogenesis.

An 8 kDa protein, BRICK, is also required for pavement cell waviness in Z. mays (Frank & Smith, 2002; Frank et al., 2003). Its vertebrate homologue HSPC300 has been shown to regulate F-actin nucleation as a part of the ARP 2/3 complex (Eden et al., 2002). However, although BRICK-encoding genes are required for the WAVE-ARP 2/3 pathway in plants, their exact role is not yet determined (Szymanski, 2005). As already mentioned (see the section Morphogenesis of ordinary epidermal cells), apart from any involvement of BRICK in AF assembly, the failure of pavement cells in brick mutants of Z. mays to become wavy results from the lack of radial MT arrays. A possible involvement of the ARP 2/3 complex, through AF assembly, in MT organization has been suggested after observations in trichomes of DISTORTED mutants of A. thaliana (Schwab et al. 2003; Saedler et al., 2004). It is therefore tempting to support such an involvement for BRICK protein in radial MT organization in pavement cells.

Another gene that participates in wavy pavement cell morphogenesis in A. thaliana is SPIKE 1 (Qiu et al., 2002), which encodes a 207 kDa protein homologue to the CDM family of adapter proteins of animals (Qiu et al., 2002). According to this, SPIKE 1 must interact with AtROP (Qiu et al., 2002; Wasteneys & Galway, 2003), a protein involved in AF assembly (see also Fu et al., 2002). However, in spike 1 mutants of A. thaliana, the failure of pavement cells to become wavy is primarily the result of a failure in the organization of cortical MTs rather than that of AFs (Qiu et al., 2002). SPIKE 1 interaction with AtROP may control cortical MT organization through mDia-like formin homologue proteins (Palazzo et al., 2001) consistent with the recent identification of formin-like proteins in A. thaliana (Cvrčková 2000; Banno & Chua, 2000). Apart from this, although a possible control of cortical MTs on AF assembly (Wasteneys & Galway, 2003) and organization (Kobayashi et al., 1988) or vice versa (Wernicke & Jung, 1992) may occur in plant cells, this cannot be substantiated for the MT arrays and AF patches that are organized during pavement cell development. These morphogenetic tools organize at different stages of pavement cell growth, the MT arrays coming first and the AF patches following during the final stage. In Z. mays and V. sinensis plants treated with CD (unpublished), MT organization is not affected despite the disruption of AFs. In addition, the absence of AF patches in pavement cells treated with colchicine cannot be directly attributed to the absence of MTs but to the complete lack of cell lobes (see the section Morphogenesis of ordinary epidermal cells). Accordingly, cortical MT array organization for lobed cell morphogenesis does not appear to correlate with that of AF patches or vice versa. Organization of these cytoskeletal components for pavement cell morphogenesis may rather be regulated by diverse time- and site- specific organizing factors.

The view that cortical MTs control the orientation of CMFs in the cell wall has been recently disputed (Emons et al., 1992, 2002; Sugimoto et al., 2001, 2003; Himmelspach et al., 2003; Wasteneys, 2004). In particular, it has been shown that transverse orientation of CMFs in A. thaliana root cells can be established in the absence of transversely oriented cortical MTs (Himmelspach et al., 2003; Sugimoto et al., 2003). However, recent experimental studies in the same plant material support the involvement of cortical MTs in CMF patterning (Gardiner et al., 2003; Baskin et al., 2004). Especially in the case of fragile fiber 2 mutants of A. thaliana, defective growth is correlated with reduced cellulose deposition in the cell wall, as a result of defective cortical MT organization (Burk et al., 2001), and the role of cortical MTs in controlling CMF orientation is further supported (Burk & Ye, 2002). However, these data derive from cells with a more or less uniform CMF distribution, which is not the case of lobed cell morphogenesis, where patterning of cell wall reinforcement by cortical MT organization is the rule. In mesophyll as well as in pavement cells, the CMFs of local cell wall thickenings copy with accuracy the pattern of cortical MTs (Apostolakos et al., 1991; Panteris et al., 1993a,b,c, 1994). Experiments with anti-MT drugs result in alteration of cell wall patterning, further supporting the above data (Apostolakos et al., 1991; Wernicke & Jung, 1992; Panteris et al., 1993a,b,c).

The AF patches at the tips of pavement cell lobes seem to be assembled during the final expansion of the wavy pattern. Fu et al. (2002) have shown that ROP2-GTPase, involved in new AF assembly, is localized at the sites where AF patches assemble. According to studies on DN-rop 2 and CA-rop 2 mutants of A. thaliana, wavy pavement cell morphogenesis is correlated with local ROP2-GTPase function and AF assembly at lobe tips (Fu et al., 2002). In addition, several genes of A. thaliana, such as NAPP, PIRP (Brembu et al., 2004), WURM and DISTORTED 1 (Mathur et al., 2003), encode products, the animal homologues of which are involved in ARP 2/3 activation: accordingly, napp-1 and pirp-1 (Brembu et al., 2004) as well as wurm and distorted 1 (Mathur et al., 2003) A. thaliana mutants for ARP 2/3 exhibit pavement cells with reduced yet apparent waviness. Accordingly, it can be well supported that late expansion of pavement cell waviness is promoted by ARP 2/3 activation through AF patch assembly at the tips of growing cell lobes.

Finally, SCD 1, a gene required for targeted vesicle trafficking, is important for cell expansion and thus implicated in wavy pavement cell morphogenesis (Falbel et al., 2003). However, the exact role of its product as well as its possible control, if any, on the cytoskeleton remain to be elucidated (Falbel et al., 2003).

Conclusions and Further perspectives

Morphogenesis of lobed cells is the outcome of the combined but distinct control of the cortical MTs and AFs on cell wall reinforcement and differential cell growth. Cortical MTs determine the pattern of local cell wall deposition, dictating primarily the formation of lobes and constrictions, which is the rule for both the mesophyll and epidermis. Whenever a final step of further cell lobe elongation occurs, which is the case of pavement cells of several plant species as well as of growing lobes of A. capillus-veneris mesophyll cells, AFs form aggregations/patches at the tips of growing lobes. The key tools for pavement cell morphogenesis are the radial MT arrays at the junctions of the external periclinal wall with the anticlinal ones, while anticlinal MT bundles play an auxiliary role. Control of cell wall reinforcement is applied by cortical MTs, while AFs at the tips of growing lobes may play, among others, a role in local cell wall differentiation for further lobe elongation. Lobed cell morphogenesis is the outcome of precisely controlled diffuse growth, while some kind of local tip-like growth may occur during the final step of lobe elongation. Formation or complete absence of intercellular spaces relies upon intercellular cortical MT patterning, depending on the future physiological requirements of the tissue.

Significant molecular factors that control cytoskeletal reorganization have been revealed through the study of mutants defective in morphogenesis. Several genes, the products of which regulate the organization of MT (reviewed by Wasteneys & Galway, 2003) and AF (reviewed by Wasteneys & Galway, 2003; Szymanski, 2005) systems, seem to be required for the achievement of lobed cell shape (see also Mathur, 2004). However, apart from identifying these molecular factors, the very primary questions about morphogenesis have not yet been answered: What is the nature of extracellular signals for morphogenesis? How are they perceived by individual cells? How are they coordinated within tissues and organs and in what way do they regulate the cytoskeleton? Forces, tensions, hormone gradients, local alterations in plasma membrane ingredients might be considered as candidates for signaling as well as triggering rearrangements of the cytoskeleton. The product of the SPIKE gene may well be involved in the integration of extracellular signals with cytoskeletal organization (Qiu et al., 2002), yet the nature and mechanism of such an integration remains unknown. ROP2-GTPase may also participate in signaling for cytoskeletal reorganization (Fu et al., 2002), but the way this signaling is initiated has not yet been revealed. Another candidate for signaling could well be phosphatidic acid (Munnik, 2001): a possible involvement of phospholipase D (Dhonukshe et al., 2003) in initiating the reorganization of the cytoskeleton could well be expected. Phospholipase D has been identified as a MT-binding protein that may interconnect the plasmalemma and cortical MTs, possibly involved in conveying external environmental signals to MTs (Gardiner et al., 2001, 2003; Wang, 2002). Further investigations in this direction may shed light on the primary response of the cytoskeleton, which is responsible for lobed cell morphogenesis.


The authors thank Dr P. Apostolakos for creative discussions and useful suggestions on the manuscript and Dr M. Zachariadis and V. Katsaros for preparing the drawings and figures of this article. Brick mutants of Zea mays were a generous gift of Dr L. G. Smith, University of California, San Diego, CA, USA.