VI. Stomatal development in dicotyledonous plants 12
VII. Mutants in stomatal development 14
VIII. Control of Stomatal Development 14
IX. Cuticle composition affects stomatal development 14
X. The HIC – HI gh Carbon dioxide gene 15
XI. Fatty acid elongases 17
XII. The cuticle: an alternative signalling medium? 17
XIII. Trichome development 18
XIV.Cuticle composition affects trichome development 19
XV. Cuticle composition affects pollen germination 20
XVI. Conclusions 20
Studies of Arabidopsis wax biosynthesis mutants indicate that the control of cell fate in the aerial epidermis is dependant upon the synthesis of the waxy cuticle that overlies the epidermal layer. Several cer mutants, originally isolated as wax deficient, not only affect cuticular wax composition but also exhibit large increases in stomatal numbers. Stomatal numbers are also affected in hic mutant plants, but despite HIC encoding a putative wax biosynthetic enzyme the hic phenotype of increased stomatal numbers is more subtle, and only seen at elevated CO2 concentrations. This suggests that environmental effects on stomatal number may be mediated through cuticular wax composition. Other putative wax biosynthetic genes, FDH and LCR, have effects on the number of trichomes that develop in the epidermis, indicating that trichome development may also be affected by cuticle composition. Thus signals from the cuticle may influence how trichome and stomatal numbers in the epidermis are determined. Wax components could be the developmental signalling molecules, or could be the mediating medium for such signals, stimulated by environmental cues, which affect epidermal cell fate.
The aerial surfaces of plants are almost entirely covered with a waxy cuticle. Being the outermost layer of the plant and the first line of contact with the environment, the cuticle has several functions. The main function is as a barrier. The cuticle protects the tissue beneath from mechanical damage by the elements (Baker & Hunt, 1986), or from insects (Walker, 1988; Eigenbrode, 1996) and acts as the primary defence against pathogens (Carver et al., 1996). The cuticle is a site of communication with insects or microbes either to repel or attract depending on their interaction with the plant. The cuticle can also protect the photosynthetic tissues from excess light by reflecting light (Vogelmann, 1993), and acts as a barrier against loss from the plant (Kerstiens, 1996a). The cuticle is virtually impermeable to polar molecules, such as water, CO2 and apoplastic solutes. This means that there is little uncontrolled water, or cellular content, loss from aerial tissues, but it also means that gas exchange between photosynthesising tissues and the atmosphere is limited (Schreiber et al., 1996). Controlled water loss and gas exchange therefore has to take place through stomatal pores.
II. Cuticle structure
The cuticle is not a simple homogenous layer deposited on the external surface of the epidermal cells. The cuticle changes and matures as the leaf matures. Immature, unexpanded leaves are covered with a thin procuticle that is made up solely of amorphous, nonlamellate waxes. As the leaf expands the cuticle thickens and lamellae appear as more constituents are added (Jeffree, 1996). Lamellation of the procuticle occurs as cutin and polysaccharides are laid down in layers, and this becomes the cuticle proper. Above the cuticle proper waxes are deposited and this becomes the epicuticular wax layer. As more wax is deposited in the epicuticular wax layer, wax crystals are formed above a thin amorphous layer of waxes. This is not true of all species, Arabidopsis thaliana accession C24 has no discernable wax crystals structures on its leaves, but does form crystal structures on stems (Jenks et al., 1995). Wax crystal structure is dependant on wax composition, which is determined by various environmental (Prior et al., 1997) and developmental conditions (see below). Beneath the epicuticular wax layer and the lamellar polysaccharide and cutin layer, two thick cutin layers are deposited (Jeffree, 1996), the internal and external cuticular layers. The stages of cuticle development described above are illustrated in Fig. 1. Polymerised cutin has a large pore size, so contributes little to the hydrophobicity of the cuticle. The primary function is more likely to be structural and to provide rigidity to the tissues beneath. Waxes are also embedded within the polymerised cutin matrix. As the cuticle develops the primary cell wall beneath becomes incorporated into the cuticle proper and the secondary cell wall develops. This secondary cell wall is more fibrous than the primary cell wall and contains lipophilic globules (Jeffree, 1996). Wax deposition continues as the leaf expands and the cuticle develops, increasing the overall wax load and thickness (Riederer & Schneider, 1990; Riederer & Markstädter, 1996). The cuticle matures alongside the leaf. Once the leaf has finished expanding the cuticle is no longer added to, unless damaged by wounding, and lasts the lifetime of the leaf (Jeffree, 1996).
III. Cuticular waxes
Waxes are an important component of the cuticle. Wax is a general term for the very long chain hydrocarbons that are found embedded within the cuticle and in the crystalline epicuticular wax layer. Complex mixtures of very long chain hydrocarbons and their derivatives make up the cuticular waxes that cover the surfaces of all plants (Kolattukudy, 1996; Post-Beittenmiller, 1996) as well as being components of seed lipids. These long chain hydrocarbons are produced from C16:1 and C18:1 fatty acid precursors that are produced in the cellular plastids. The fatty acids are elongated extraplastidially, by the addition of two-carbon moieties donated by malonyl-CoA, to produce very long chain fatty acids (VLCFAs), that is, fatty acids with more than 18 carbon atoms. VLCFAs are then modified by reduction to aldehydes, then decarbonylation to produce odd number chain length alkanes, two oxidation reactions to produce secondary alcohols and ketones; by acyl-reduction to produce aldehydes and then reduction to produce primary alcohols; and by β-ketoacyl-elongation to produce β-diketones and other acyl esters (Lemieux, 1996). The principle components of Arabidopsis leaf epicuticular waxes are very long chain alkanes (C29, C30 and C33) followed by primary alcohols (C26, C28 and C30) (Jenks et al., 1995).
IV. Cell patterning in the epidermis
The plant epidermis is a single cell layer consisting of a number of specialised cell types. The most common cell type is the relatively unspecialised epidermal pavement cell. The patterning of other cell types, such as trichomes and guard cells, within the epidermal cell layer appears to be controlled by a number of mechanisms. Figure 2 shows an example of the Arabidopsis leaf epidermis illustrating these different cell types of the epidermis.
Patterning of cells within a two-dimensional layer could be controlled by several possible mechanisms. The precursors of the specialised cells could be placed within the pavement cell layer as the leaf develops and be in place before expansion, and thus mature cell fate would occur independently of the cells around them. Alternatively, precursor cells placed in the developing leaf could mature differently depending on their position within the expanding leaf surface, and thus their ultimate cell fate would be positionally dependant and would require communication with the cells around them. These two different mechanisms are referred to as the cell lineage mechanism and the position dependant or cell-to-cell signalling mechanism, respectively.
V. Stomatal development
Stomata are found in the epidermal cell layer over the aerial surfaces of plants (Fig. 2b). They allow controlled water loss and gas exchange between photosynthesising tissues and the atmosphere that would otherwise be prevented by the impermeable waxy cuticle. Both these functions are gated by the opening and closure of the stomatal pore brought about by turgor changes in two sister guard cells.
Stomatal development in dicots occurs slightly differently from that in monocots. In monocots stomatal initial cells form close to the leaf apex and as the leaf expands longitudinally, the stomatal initial divides asymmetrically once, producing two oblong cells, one larger than the other. The larger daughter cell is positioned basally within the file of cells. The smaller daughter cell, positioned apically, then undergoes an equal division to form the guard cell pair (Larkin et al., 1997). In dicots however, the situation is more complex. In dicots stomatal development occurs throughout the whole leaf, throughout leaf expansion, so stomata at all stages of development occur at the same time in the same leaf epidermis (Croxdale, 2000). This leads to more questions about how all these developing stomata are coordinated. In dicots stomatal initials are known as meristemoids, because they are cells that have retained meristematic characteristics, even though they are not part of the meristem and are found to be dividing in the body of the leaf. Recent results suggest that signals from the cuticle may affect stomatal development in dicots and this will be discussed in more detail here.
VI. Stomatal development in dicotyledonous plants
In dicots, although there is no discernable pattern of stomatal position over the whole leaf, there is a very high degree of order in stomatal positioning at a cellular level (Croxdale, 2000). If stomatal positioning within the epidermis were completely random, then periodically stomata would be found placed next to each other. In most plants this is extremely rare, and stomata are always separated from other stomata by at least one pavement epidermal cell. This spacing of stomata may occur to prevent the diffusion shells of each stoma overlapping, so that the maximum gas exchange is gained with the least number of stomata, and also to allow for the large turgor pressure changes that are necessary to bring about alterations in stomatal aperture. However, the true physiological rationale behind the single cell spacing pattern has yet to be identified. In the 1940s, Bünning proposed that this spacing pattern in dicots, was due to mature guard cells producing a diffusible inhibitor that prevented surrounding cells from developing into stomata (Bünning & Sagromsky, 1948) and this mechanism has remained unproven. More recently the mechanism behind this two-dimensional spacing pattern has been extensively studied in A. thaliana.
Meristemoid cells are found randomly distributed throughout the epidermis (Geisler et al., 2000) and undergo a highly specialised set of cell divisions within the epidermal layer before they terminally differentiate into a pair of sister guard cells that will function as the stomatal complex. These cell divisions, it has been proposed, cause the spacing of stoma, by ensuring that each stoma is surrounded by a complement of subsidiary epidermal cells (Serna & Fenoll, 1997; Croxdale, 2000; Geisler et al., 2000). When developing into a stomatal complex the meristemoid undergoes a series of unequal divisions (Paliwal, 1967), usually between one to three divisions, but may be as many as five. Each unequal division produces two cell types: a large daughter cell which goes on to become a pavement epidermal cell and a smaller daughter cell that retains the meristemoid cell fate. The meristemoidal daughter cell can then go on to perform further unequal divisions and produce more epidermal cells (Paliwal, 1967). These unequal divisions are also carefully orientated so that pavement cells are always placed between the meristemoid and any other stoma or stomatal precursors (Serna & Fenoll, 1997; Geisler et al., 2000).
Several research groups have studied this sequence of highly specialised cell divisions that go into forming stomatal patterning on leaves, in different A. thaliana accessions. Two slightly different cell division mechanisms have become apparent. In the C24 accession of A. thaliana most stomata are part of clonally related anisocytic stomatal complexes (Serna & Fenoll, 2000a). Meristemoids in this accession usually divide asymmetrically three times. Each time a meristemoid divides unequally the larger epidermal daughter cell is placed on the outside of the forming cluster so that by the end of the division cycle there is a single meristemoid surrounded by three pavement epidermal cells (Fig. 3a–d). The elder two pavement cells are much larger than the youngest (Fig. 3d). At this point the meristemoid differentiates into a guard mother cell (GMC), shown by change from a triangular to an oval shape. The GMC then undergoes an equal cell division to produce the two sister guard cells that gate the stomatal pore (Paliwal, 1967; Pant & Kidwai, 1967) and complete the formation of what is known as an anisocytic complex (Fig. 3e, f). The daughter cells produced from the unequal meristemoidal divisions can also retain their meristemoidal potential before terminally differentiating into a pavement cell. These cells, usually the youngest and smallest pavement cell, can go on to form satellite stomatal complexes by following the same pattern of several polarised, unequal cell divisions, followed by one equal cell division that the original meristemoid underwent (Fig. 3g–i) (Serna & Fenoll, 2000a).
This work on the C24 accession suggests a mechanism for stomatal patterning that is position independent. If the stereotyped divisions that the meristemoid undergoes to form an anisocytic complex are controlled purely by the segregation of internal cell fate determinants during asymmetric divisions (such mechanisms studied in yeast and Drosophila are reviewed by Amon, 1996 and Jan & Jan, 2000, respectively) then this would create a stomatal pattern where stoma were always surrounded by a complement of clonal pavement cells without the need for external signals. Therefore, once a meristemoid has embarked on a stomatal cell fate it could follow it to differentiation independently of the rest of the leaf.
However, while this may be the case for C24, work done in the Columbia accession of A. thaliana, also indicates a role for cell-to-cell signalling during stomatal development (Geisler et al., 2000). In Arabidopsis plants of the Columbia ecotype, the majority of stomata have been found to be part of anisocytic complexes that contain pavement cells that are not clonally related to the stomata they surround (Geisler et al., 2000). Therefore the processes of cell division that occur in C24 plants to surround each stomata with a set of clonally related pavement cells does not proceed in Columbia plants in exactly the same way. Stomata still form in Columbia from a meristemoid after they have undergone asymmetric divisions, as they do in C24. However, Columbia meristemoids undergo less asymmetric divisions than C24 meristemoids before they are able to terminally differentiate into the GMC and the asymmetric divisions that do take place occur specifically to separate the meristemoid from other meristemoids, GMCs and mature stomata (Geisler et al., 2000). If a meristemoid is not in contact with any other cell type other than pavement cells, it can terminally differentiate without undergoing any asymmetric divisions. Alternatively, if a meristemoid is in contact with two or more meristemoids, GMCs or mature stomata (or a combination of any of the three) it will not follow the stomatal cell fate and will terminally differentiate into a pavement cell to maintain the separation between the stomatal complexes already differentiated (Geisler et al., 2000). This phenomenon has also been observed in the monocot species Tradescantia (Boetsch et al., 1995). Such work suggests that meristemoids are receiving signals from their immediate surroundings at every stage of their development into stomata. They are able to differentiate between the cell types that they are in contact with and divide in response to cells with a stomatal cell fate, but are able to stop dividing and differentiate if pavement cells already surround them, even if these pavement cells have not been produced clonally by the same meristemoid. It is likely that both the C24 and Columbia accessions of A. thaliana utilise both position dependant signals and lineage based division mechanisms to create an overall stomatal pattern; although the extent of position dependant signalling is less well studied in C24. The literature suggests that different accessions, in different environmental conditions may rely more heavily on one of the mechanism, but as both mechanisms are vital to pattern creation they will both be utilised by the plant to a greater or lesser extent.
VII. Mutants in stomatal development
Several mutants that interrupt the one-cell spacing pattern have been identified that support the idea that there are several levels of control over stomatal development. The TMM and FLP genes ensure that stomatal clustering does not occur (Yang & Sack, 1995). Large stomatal clusters are observed in tmm plants. TMM ensures that the asymmetric divisions of meristemoids are orientated correctly so that new pavement cells are placed between new and old stomata (Geisler et al., 2000). FLP acts downstream of TMM and is involved in the control of the equal division undergone by GMCs to produce two daughter guard cells. Removal of FLP function results in stomatal pairs with odd and even numbers of guard cells (Yang & Sack, 1995). A putative subtilisin-like serine protease, SDD1, has been identified that affects not only control of stomatal spacing but also overall numbers of stomata. SDD1 is likely to be involved in processing a signal peptide that is involved with the inhibition of excess stomata production (Berger & Altmann, 2000). The identification of these control mechanisms suggests that, although there are two distinct mechanisms that control stomatal spacing and stomatal numbers, that there is cross-talk between the two pathways.
VIII. Control of Stomatal Development
Stomatal development is not just controlled at the level of cell spacing. The numbers of stomata that form on an aerial organ, such as a leaf or stem, are also influenced by the environmental conditions. For example, growth under increased atmospheric carbon dioxide concentrations leads to decreases in stomatal numbers in many plant species, including A. thaliana (Woodward & Kelly, 1995; Lake et al., 2001). Indeed, the excess CO2 generated by man's activities since the industrial revolution has had an effect on stomatal development and many plant species have a reduced number of stomata in comparison to preindustrial times (Woodward, 1987). This inverse correlation between atmospheric CO2 concentration and stomatal frequency is an important finding that has impacted on many areas of research. It has been used to accurately estimate CO2 levels in past environments, the timing of ice ages and periods of greenhouse climate warming stretching back over the past 300 million years (Retallack, 2001), to account for the mass extinction of species (McElwain et al., 1999) and to help produce future global climate models (Cramer et al., 2001).
Environmental signals, such as light levels and exposure to the drought hormone ABA, have also been demonstrated to affect stomatal numbers in some species (Quarrie & Jones, 1977; Schoch et al., 1980; Franks & Farquhar, 2001; Lake et al., 2002). Therefore, it seems likely that there are at least two levels of control over stomatal development. Firstly, controls that act at the cellular level ensure the one cell spacing pattern, so that all stomata are separated from other stomata by at least one pavement cell. Secondly, environmental and developmental controls act to control the overall number of stomata that are formed. These controls could act at the level of meristemoid initiation and control the numbers of meristemoids that form in the epidermal cell layer. Alternatively, these signals could control the number of satellite stomatal complexes that form after the primary anisocytic complex has formed (Fig. 3g–i).
IX. Cuticle composition affects stomatal development
The composition and amount of waxes in the cuticle has been shown to vary depending on the environmental conditions that the plant is exposed to. For example, carnation plants grown in vitro in high humidity have decreased wax load relative to plants grown at lower humidity (Majada et al., 2001). The constituents of the cuticular waxes may also change with altered environmental conditions. The ratio of esters, primary and secondary alcohols, alkanes and ketones can vary without affecting the overall wax load. Kale and swede plants grown indoors have a greater proportion of long chain esters making up their epicuticular waxes, than similar plants grown outdoors (Shepherd et al., 1995, 1997). Plants exposed to higher temperatures during different periods of their light/dark cycle, show differences in their wax composition, depending on when the plant is exposed to higher temperatures (Riederer & Schneider, 1990). Wax morphology also alters in response to environmental conditions, such as low nitrogen levels and water stress (Prior et al., 1997) presumably due to changes in wax composition, as wax composition is known to affect structure.
Many mutants in the epicuticular wax biosynthesis pathway have been characterised, but the list is by no means exhaustive. Most wax mutants are identified through a change in the appearance of their epidermis, as without the wax bloom, wax deficient mutants have a glossy appearance (Koornneef et al., 1989). In Arabidopsis these mutants have the common title eceriferum and there are currently over 20 characterised wax mutants (Koornneef et al., 1989). Most cer mutants see a decrease in their epicuticular wax load, but the degree of this varies quite considerably from mutant to mutant (Hannoufa et al., 1993). These mutations also effect the wax composition, varying the proportion of alcohols, esters, alkanes and ketones (Hannoufa et al., 1993; Jenks et al., 1995, 1996a, 1996b; Post-Beittenmiller, 1998; Rashotte et al., 2001). Several of the genes believed to be involved in wax biosynthesis have been identified. CER1 is thought to encode a component of a fatty aldehyde decarbonylase enzyme involved in the production of odd number carbon chain fatty acids (Aarts et al., 1995). The CER3 sequence includes a nuclear localisation signal and probably encodes a factor that regulates wax production (Robbins et al., 1991; Lemieux, 1996). CER2 is also a nuclear localised protein and thought to be a regulatory protein (Xia et al., 1997). CER6 encodes a fatty acid elongase condensing enzyme and is identical to CUT1, which had been previously identified (Millar et al., 1999; Fiebig et al., 2000; Mariani & Wolters-Arts, 2000).
Early work on eceriferum mutants was carried out in barley (Zeiger & Stebbins, 1972) and it was in this work that the link between epicuticular wax production and stomatal development was first made. Zeiger and colleagues noted that in 12 cer-g mutants there were stomatal abnormalities (Zeiger & Stebbins, 1972). The main abnormalities were double and triple stomatal complexes, that is, two or three stomata in contact with each other, breaking the one-cell spacing rule. This effect on stomatal development was stronger in some cer-g mutants, with greater proportions of their stomata in abnormal complexes. More recently, in Arabidopsis a similar pleiotropic effect has been observed in two eceriferum mutants cer1 and cer6 (Gray et al., 2000). Both these mutants have large increases in their stomatal indices (the proportion of epidermal cells that are stomata).
Recent work has shown that the blue fluorescence of guard cells is greater than that of pavement epidermal cells and this is thought to be due to the epicuticular waxes covering the guard cells (Karabourniotis et al., 2001). Thus waxes covering guard cells are different from those covering the rest of the epidermal surface. Either they are thicker or have a higher concentration of wax-bound phenolics (either could account for the increase in blue fluorescence). This could be due to guard cells producing different waxes to surrounding pavement cells and this could have implications for stomatal development if guard cells and their precursors are interacting with the cuticle in a different manner to pavement epidermal cells.
X. The HIC–HIgh Carbon dioxide gene
A gene has recently been discovered in Arabidopsis that may link guard cell wax biosynthesis with the stomatal developmental response to increased atmospheric carbon dioxide (Gray et al., 2000). The hic mutation is in the C24 background accession, which normally shows a small decrease in stomatal, numbers, if any change at all, at elevated carbon dioxide conditions. However when hic plants are exposed to the same elevated carbon dioxide conditions they show an increase in stomatal density of up to 40%. Unlike the cer-g mutants in barley however, the hic stomata show no stomatal clustering and maintain their one-cell spacing rule. Therefore the HIC gene appears to affect the number of stomata that are initiated rather than maintaining the stomatal spacing pattern, and only in response to environmental variables such as CO2 concentration (Gray et al., 2000).
The HIC gene appears to be expressed exclusively in guard cells and has a high degree of homology with KCS-1, a 3-ketoacyl CoA synthase (KCS), suggesting it could perform a KCS-like function. KCS enzymes are condensing enzymes in the microsomal fatty acid elongase complex that produce the very long chain fatty acids which go to make up the cuticular waxes. Sequence analysis reveals a large family of KCS-like genes present in the Arabidopsis genome sequence. There are over 15 putative KCS genes in the Arabidopsis genome sequence (Millar et al., 1999) five of which have been studied so far (see Table 1). The expression patterns of these five KCS-like genes have been determined and reveal different KCS genes to be expressed in different cell types. As would be expected, KCS genes are expressed in wax producing cells such as seed (FAE1), epidermal cells (e.g. CER6/CUT1) and guard cells (HIC).
Table 1. Differing expression patterns and functions of the KCS-like gene family members in Arabidopsis
It is possible, because of the homology to peptides with KCS activity, that HIC is involved in guard cell cuticular wax biosynthesis. However, when the HIC gene is mutated it has no obvious effect on guard cell wax deposition (Fig. 4).
The mechanism by which HIC affects stomatal development in response to environmental stimuli such as atmospheric CO2 level is currently unknown. One model is that HIC controls by inhibition the number of satellite stomata produced by each maturing meristemoid and by removing HIC function more satellite stomata are allowed to mature. Therefore HIC expression in the guard cell may be involved in the production of a specific VLCFA-derivative inhibitor that other epidermal cell types do not make and that directly interacts with surrounding epidermal cells as part of the cuticular wax content (Fig. 5c). Alternatively there could be an, as yet unidentified, diffusible inhibitor molecule, that is produced by guard cells and diffuses through the cuticle to inhibit surrounding cells (Fig. 5a). The subsequent change to the composition of the cuticle when HIC function is removed decreases the diffusion of the inhibitor (Fig. 5b) (Serna & Fenoll, 2000b). Less satellite meristemoids would then be inhibited from following a stomatal cell fate, thus increasing the number of mature stomata that form. Alternatively, a similar model could involve the increased diffusion of an activator necessary for stomatal development in hic plants. As the activator is reaching more meristemoids this would increase the number of meristemoids that are promoted to develop into mature stomata. However this appears a less likely mechanism, as an activator would be redundant under most conditions and would require specific expression in response to stimuli such as CO2 concentration. The inhibitor model, be it with an as yet unidentified inhibitor (Fig. 5a, b), or with a VLCFA inhibitor (Fig. 5c, d), is a simpler mechanism to envisage.
The hic phenotype described above indicates that the way in which stomatal development is modified by environmental conditions may be linked to the wax composition of the guard cell cuticle. The changes in epicuticular wax, in wild-type plants due to environmental conditions or due to mutations affecting wax composition in eceriferum plants, are presumably affecting how epidermal cells differentiate and determine their cell fate. Therefore, epidermal cells presumably have to sense and interact with the cuticle as part of their development, in the same way that cell-to-cell positioning signals also effect stomatal development. Alternatively, it is possible that the stomatal development pathway and the wax biosynthesis pathway are functionally independent, but are subject to the same controls from environmental signals.
XI. Fatty acid elongases
The finding that HIC encodes a putative fatty acid elongase condensing enzyme (or KCS) warrants a closer examination of these enzymes, and their role in wax biosynthesis. The fatty acid elongase complex, which catalyses the formation of VLCFAs, is composed of four enzymes that together catalyse the addition of two carbon moieties, from a malonyl CoA donor, to the elongating fatty acyl chain. The four reactions catalysed by the FAE complex are a condensation, reduction, dehydration and a second reduction reaction. It appears that the first reaction, catalysed by the condensing enzyme (or KCS), is the rate limiting step of the complex as increasing the level of a condensing enzyme component in transgenic plants results in higher levels of VLCFAs being produced (Millar & Kunst, 1997). The condensing enzyme is also believed to regulate the spatial distribution of VLCFA production in the plant. Normally, VLCFAs are only produced to significant amounts in cells that produce waxes, that is cells of the epidermal layer and seeds. However, when the condensing enzyme is ectopically expressed under the control of the constitutive CaMV 35S-RNA promoter, VLCFAs accumulate in all tissues examined and result in altered organ morphology (Millar et al., 1998). This result suggests that the three other enzymes of the fatty acid elongase complex are expressed ubiquitously and that only the condensing enzyme is differentially expressed to control VLCFA production (Millar & Kunst, 1997). Furthermore, it seems that it is the condensing enzyme that controls the length of VLCFAs that are produced by the fatty acid elongase complex. For example, when the seed specific condensing enzyme gene, FAE1, is expressed in leaves the VLCFAs produced in the leaves correspond in length to those usually produced in the seed (Millar & Kunst, 1997). Together these results indicate that the expression pattern of KCS genes, not only determines where VLCFAs are produced, but is also important in determining both the quantity and length of VLCFAs produced. Thus, the expression of members of the KCS gene family is of great importance in determining the composition of the cuticular waxes. The finding that the HIC-KCS gene is expressed exclusively in the guard cells (Gray et al., 2000) adds weight to the suggestion that the composition of guard cell cuticles is different to that of other epidermal cells (Karabourniotis et al., 2001).
XII. The cuticle: an alternative signalling medium?
Plant cell walls create an additional physical barrier between cells, and much plant cell-to-cell communication is maintained through plasmodesmata. These are membrane-lined channels that traverse plant cell walls, and ensure contiguous cytoplasmic contact between cells. Plasmodesmata allow diffusion and active transport of various classes of molecules, including nutrients, low molecular weight growth regulators, transcription factors and protein signals important for development (Zambryski & Crawford, 2000). Guard cells are known to lose their plasmodesmatal connections early during their differentiation (Palevitz & Hepler, 1985), presumably due to the high turgor pressure that these cells must maintain to open the stomatal pore. Although this phenomenon has not been specifically studied in Arabidopsis guard cells the lack of plasmodesmata is illustrated by recent studies showing that Arabidopsis guard cells are cut off from the normal traffic of signals that travel via the plasmodesmata. Kim et al. (2002), produced a GFP-fusion with the intercellular developmental signalling protein KNOTTED1 (Kn1). Kn1 is trafficked though plasmodesmatal connections within cell layers, and between cell layers. Using the GFP fusion assay they showed that Kn1 is able to reach epidermal pavement cells from mesophyll cells, but that the GFP fusion is never found in guard cells. Therefore guard cells are effectively cut off from some of the normal communication routes with the cells around them. If stomata were cut off from developmental signals from the surrounding tissue, then a plausible mechanism for stomatal development would be a cell lineage mechanism, as all developmental cues would arise internally. However, as there is now evidence that stomata are affected by, and have an effect on, their surrounding tissue (Geisler et al., 2000), even though they have lost their plasmodesmata, other signalling mechanisms must also be at work. Therefore if guard cells are cut off from the normal routes of cell-to-cell communication by the loss of their plasmodesmata, there is a need to explain how these cells are maintaining communication with surrounding epidermal cells.
Guard cells could signal to pavement cells they are in direct contact with, through cells walls, perhaps via proteins such as wall associated kinases (Anderson et al., 2001). It is possible that pavement cells surrounding a guard cell could be inhibited from following a guard cell fate, by cell wall associated proteins that act as inhibitors to cells in contact with them. Studies of the hic and cer mutants suggest that an alternative signalling mechanism takes place through the medium of the waxy cuticle. Mutations in the wax biosynthetic pathway, such as cer1 and cer6, that affect the composition of the epicuticular waxes, also affect stomatal development and increase the proportion of epidermal cells that are guard cells (Gray et al., 2000). Molecular analyses of CER1 (Aarts et al., 1995) and CER6 (Fiebig et al., 2000) have shown them to encode putative wax biosynthesis enzymes. CER1 is involved in the production of long chain alkanes from long chain aldehydes and CER6 is involved in the production of C28 fatty acids. This suggests they are directly involved in the wax biosynthetic pathway, rather than in regulation of the pathway, such as is the suggested function of CER2 (Jenks et al., 1995). Therefore it is possible that VLCFA-derivatives have a direct affect on stomatal development. Our conclusion from these data is that these VLCFA-derivatives or downstream products of them may be directly involved in signalling during stomatal development, either as the signalling molecule themselves or as mediators of unidentified signalling molecules. This is supported by the notion that cuticle composition affects diffusion rates through the cuticle by affecting the structure and fluidity of the cuticular waxes (Schreiber & Riederer, 1996; Kerstiens, 1996a). However, as yet the nature and mechanism of this stomatal development signal is unknown.
XIII. Trichome development
Trichomes are cell types that protrude from the epidermal surface (Esau, 1953); this includes root hairs, thorns, stigmatic papillae and conical-papillate petal cells. However, as with stomatal patterning, most recent work on trichome patterning has been carried out in A. thaliana on aerial organ trichomes. In the case of Arabidopsis these are single celled, branched protrusions that act as a mechanical barrier for the leaf surface (Fig. 2a) (reviewed by Marks, 1997; Glover, 2000).
In common with stomata, trichomes are separated by epidermal pavement cells and are not normally found in contact with each other (Figs 2 and 3). Unlike stomata, trichomes are surrounded by a complement of approximately 10 specialised socket cells that differ from pavement cells (Larkin et al., 1996). Trichome initiation is limited to a specific basal region of developing leaves and trichomes mature in sequence as the leaf expands. This is in contrast to stomata of dicots that develop throughout the leaf, with no particular initiation zone. Also stomata use stereotypical asymmetric divisions which ensure their spacing (to a greater or lesser extent). There is no evidence that trichome development involves any similar cell lineage mechanism. Trichomes are found to be surrounded by clonal and nonclonally related cells, and the number of clonally related cells that surround a trichome can vary greatly (Larkin et al., 1996). Therefore, the mechanism that determines trichome patterning has been postulated to be fully positionally dependant, and pre-existing epidermal cells are recruited to become part of the trichome complex (as socket cells), rather than produced as part of a trichome developmental cell division program.
As plants can survive without trichomes (unlike stomata) many mutations in trichome development and patterning have been characterised. Three positive regulators, GLABRA1 (GL1), GLABRA2 (GL2) and TRANSPARENT TESTA GLABRA (TTG), and an inhibitor, TRIPTYCHON (TRY), of trichome development have been identified (Koornneef et al., 1982; Larkin et al., 1993, 1994; Hülskamp et al., 1994). It has been postulated (Schnittger et al., 1999; Larkin et al., 1996) that GL1 and TTG create a positive feedback loop in cells predisposed to becoming trichomes, and as their expression increases the cell becomes committed to the trichome cell fate pathway. TRY, on the other hand, is expressed by the committed cell and is involved in inhibiting surrounding cells from becoming trichomes. TRY must act over a radius of several cells because normally three or four epidermal cells separate trichomes (Hulskamp et al., 1994; Larkin et al., 1994). GL1 is expressed at low levels throughout the protoderm. Therefore exactly how individual cells are selected to follow the positive feedback loop, increase GL1 transcription and become committed to a trichome cell fate is unclear (Larkin et al., 1993).
As outlined above for stomata, there also appears to be more than one level of control over trichome patterning. Genes have been identified that ensure the cellular spacing pattern, such as TRY and GL1, and their disruption increases trichome number by removing the inhibition on trichomes forming next to each other and producing clusters of trichomes. Genes, such as GL2, affect trichome number and trichome morphogenesis. Trichome number is reduced in gl2 plants, and many of the trichomes do not mature fully and have fewer branches (Koornneef et al., 1982; Marks, 1997). There are other genes, such as REDUCED TRICHOME NUMBER (RTN), that are solely involved in the numbers of trichomes formed (Larkin et al., 1996). Disruption of this gene does not produce trichome clusters, as it is not involved in maintaining the spacing pattern. In normal circumstances RTN lowers the number of trichomes formed, by reducing the developmental time ‘window’ that trichome initiation can occur in. Therefore when RTN function is removed trichome numbers increase, because trichome initiation continues for longer (Larkin et al., 1996).
TTG , GL2 , GL1 and TRY were characterised by their effect on trichome number. However all four have since been shown to have pleiotropic effects on epidermis development. TTG and GL2 have been shown to affect root hair development ( Dolan, 1996 ) and stomatal development in the hypocotyl ( Berger et al., 1998 ). In ttg and gl2 plants, as well as a reduction in leaf epidermal trichome numbers, ectopic root hair growth is seen. Therefore TTG and GL2 are believed to be negative regulators of root hair cell development ( Dolan, 1996 ). They are also believed to be negative regulators of stomatal development in the hypocotyl, as ttg and gl2 plants have increased stomatal complexes in the hypocotyl ( Berger et al., 1998 ). Recent work by Bean et al. (2002 ) has shown that GL1 and TRY affect stomatal patterning in cotyledons, though these are tissues that have no trichomes. Therefore, GL1 and TRY are believed to have roles in generating general epidermal cell patterning and specifically in regulating trichome patterning ( Bean et al., 2002 ). Stomatal numbers are strongly influenced by environmental factors, but there is currently little evidence that trichome numbers are affected by environmental conditions in the same way. Changes in trichome numbers observed following ABA or water stress ( Quarrie & Jones, 1977 ) may be due to the reduction in stomata, allowing more protodermal cells to be recruited into the trichome development pathway. Recruitment of protodermal cells to various epidermal cell fates has been shown to be competitive and an increase in one cell type, such as trichomes, leads to a concomitant decrease in another cell type, such as stomata ( Glover et al., 1998 ).
XIV. Cuticle composition affects trichome development
Recent evidence from two Arabidopsis mutants suggests that, as for stomatal development, the composition of the cuticle affects trichome development. FIDDLEHEAD (FDH) has a high degree of homology with HIC and KSC1 (Gray et al., 2000), and is also thought to encode a 3-ketoacyl CoA synthase (Yephremov et al., 1999; Pruitt et al., 2000). As 3-ketoacyl CoA synthases are involved in the production of VLCFAs, which contribute to cuticular wax composition, disruption of the FDH gene, like HIC disruption, is thought to alter the composition of the cuticular waxes. The major phenotype of the fdh mutant is postgenital organ fusion. It has been suggested by Lolle et al. (1997) that the epidermal fusion phenotype in fdh is due to a change in the permeability of the cuticle. It is proposed that small diffusible fusion-promotion molecules are exchanged between epidermal surfaces that are in contact with each other and thus promote fusion. However, Yephremov et al. (1999) suggest that it is not increased permeability to small diffusible fusion-promoting molecules that causes the increased fusion phenotype. They argue that if this was the case then more fusion phenotypes would be seen in the many different wax biosynthesis mutants identified in Arabidopsis and maize. Alternatively they argue that it is exposure of surface adhesion molecules that would normally be covered by the cuticle that promotes fusions between organs. Whatever the mechanism, both groups appear to agree that the phenotype is due to a change in the composition of the cuticular waxes. Yephremov et al. (1999), also reported an additional fdh-1 phenotype, seen in the Columbia background, which is of interest to our discussion of cell fate determination in the epidermis. In fdh-1 plants trichome numbers are dramatically lowered to approximately half their wild-type density. This reduced trichome number phenotype in fdh provides a link between cuticular wax composition and trichome development. As with stomatal development the nature of this link is unclear. The fdh phenotype is most similar to that of the effect of RNT, as there are no trichome clusters present, rather an overall decrease in trichome numbers over the whole leaf surface (Yephremov et al., 1999). Therefore it is possible that FDH, or its products, are involved in determining the length of time over which trichome initiation can occur.
A second mutant in fatty acid biosynthesis has been identified that effects trichome development, lacerata (lcr) (Wellesen et al., 2001). LACERATA has been identified as a cytochrome P450 monooxygenase gene. The LCR monooxygenase catalyses the ω-hydroxylation of fatty acids. It has been suggested by Wellesen et al. (2001) that LACERATA is involved in the formation of C16 and C18 cutin monomers. LCR is thought to esterify cutin monomers so that they can be polymerised with the carboxyl group of the next monomer (Wellesen et al., 2001). It may also produce in-chain hydroxyl groups, so there can be cross-linking polymerisation. With the removal of LCR, cutin monomers are unable to be fully polymerised and therefore it is not possible to create a continuous layer of cutin. lcr mutant plants were identified as having a number of phenotypic characteristics, including organ fusions between inflorescences and rosette leaves, distorted rosette leaves due to bulges of malformed cells in the epidermal layer, plants that were considerably smaller than wild-type and delayed senescence (Wellesen et al., 2001). The abnormalities observed in the development of the epidermal cell layer include changes in trichome development. Some lcr leaves have areas totally devoid of trichomes. The trichomes that are present appear to remain at an immature stage of development and the most mature ones are found in the centre of the lamina, along the proximodistal axis (Wellesen et al., 2001). This work suggests that removal of LACERATA function, as with FDH, decreases the trichome initiation period. In lcr, however, the effect appears to be much more pronounced with the trichomes that are initiated being prevented from reaching maturity. lcr plants share many phenotypic characteristics with fdh plants, such as organ fusion, dwarfism, reduced trichome numbers and nonstigma pollen germination. However there are some differences. lcr organ fusions are more severe in rosette leaves than in fdh, which shows the strongest fusions in inflorescences. fdh plants are also sterile, whereas lcr produce fertile flowers.
In other studies plants with altered levels of cutin have been characterised. Transgenic Arabidopsis plants expressing a fungal cutinase have a cuticle with increased permeability, which again results in a fiddlehead-like phenotype (Sieber et al., 2000). In addition to ectopic pollen germination and postgenital organ fusions these plants exhibit abnormalities in some epidermal cells. Abnormally sized epidermal cells and multilayered epidermis are found. Protrusions of epidermal cells occur at some fusion junctions with evidence of periclinal epidermal cell divisions. Trichomes and stomata however, appear to differentiate normally in these cutinase expressing these plants. This raises the possibility that many of the fiddlehead traits that are common to the over-expressing cutinase plants and fdh plants (such as postgenital fusion) are the result of increased permeability of the cuticle, and that epidermal cell fate (in this case trichome development) is influenced particularly by the composition of VLCFAs that make up the plant cuticle. This supports the idea that it is the composition of the waxes in the cuticle that play a role in determining epidermal cell fate. Not all eceriferum mutants characterised have reported altered trichome development, however, this is not an extensively studied area, and due to the range of background ecotypes used and the availability of trichomeless background lines, finding distinct trichome effects can be difficult.
Unlike stomata there has not been a wax mutant characterised that specifically affects only trichome development. Both fdh and lcr affect trichome development as part of a pleiotropic effect on plant development. However both the stomatal and trichome developmental mutants that are described above illustrate the dramatic effect that the cuticle has on the development of the epidermis, and in turn on the whole plant. By changing the composition of the cuticle plant growth can be stunted, pollen become infertile, senescence delayed, the epidermis distorted and epidermal cell fate altered.
XV. Cuticle composition affects pollen germination
The composition of the cuticle has also been shown to affect the hydration of pollen grains and germination of the pollen tube on the epidermis. Lipid components of the cuticle are involved in cell-to-cell signalling processes during pollination and there is evidence indicating that KCS gene products are involved. Several of the Arabidopsis KCS gene family appear to play a role in male fertility as mutations in these genes result in conditional male sterility (e.g. cut1/cer6) or allow nonstigma pollen germination (e.g. fdh1). There is evidence that the composition of the waxes of the pollen coat (in plants with dry stigma such as Arabidopsis) or exudates of wet stigma surfaces are involved in pollen–pistil interactions that determine whether pollination is successful. Wax deficient mutants, cer1, cer3 and cer6/cut1 that are defective in the production of VLCFAs are conditionally male sterile (Preuss et al., 1993). Pollination can be restored in these wax mutants at high humidity suggesting that sterility is due to lack of pollen hydration. In addition transgenic tobacco plants with an ablated stigma surface are unable to support successful pollination. Application of cis-unsaturated triacylglycerides, but not saturated triacylglycerides, to the stigma in either situation restores fertility, and results in seed production (Wolters-Arts et al., 1998). The presence of cis-unsaturated triacylglycerides can even promote pollen penetration of leaf tissue in a manner similar to that observed in fdh1 plants. Thus, it appears that it is the composition of the lipids and/or fatty acids of the pollen or stigma cuticle, rather than other cell-type determinants, that control whether pollination is successful. This hypothesis is supported by the finding that wax fluidity affects water (and other solute) transport across the cuticle, albeit at a very slow rate (Kerstiens, 1996a); and that it is the chain length composition of the cuticular waxes that affect wax fluidity and in turn water transport rates (Riederer & Schneider, 1990). Therefore the addition of certain types of triacylglycerides could restore rehydration of pollen by changing the cuticle composition sufficiently to allow water movement across the cuticle (Kerstiens, 1996b).
Over the last decade information from a variety of mutants in A. thaliana, and other species such as barley and maize, have provided useful insights to the control of epidermal cell differentiation. Direct observation of the developing epidermis (Schnittger et al., 1999; Geisler et al., 2000) and the use of molecular methods (Serna & Fenoll, 2000a; Larkin et al., 1996), in wild-type and mutant plants, has shown that cell-to-cell signalling is important in the development of stomatal and trichome patterning within the epidermis. These cell signals ensure the maintenance of trichome and stomatal spacing and prevent clusters of each cell type. Mutant analysis has shown disruption of both trichome and stomatal spacing patterns to be possible, not only by disrupting genes that are specifically involved in maintenance of the spacing pattern, but also by disrupting genes that have a less specific developmental functions.
Stomata and trichome numbers have also been shown to be regulated. This regulation does not affect the spacing of the individual trichomes and stomata at the cellular level. Rather it appears to be acting on a whole plant scale as part of tissue patterning, by either affecting the number of precursor cells placed within the developing epidermis (for trichomes and stomata), the length of time in which cell fate can be initiated (for trichomes) or the number of specialised cells a particular precursor cell can go on to form (for stomata). This has been illustrated by the discovery of genes such as TTG and GL2, which have whole plant developmental effects and affect epidermal cell development in the root, hypocotyl and leaf. Stomatal numbers are strongly influenced by environmental factors, but there is currently little evidence that trichome numbers are affected by environmental conditions in the same way. Where they are, this may be due to a wider mechanism that allows more cells to be recruited to the trichome cell fate pathway because they are not required for the stomatal cell fate. However as trichome initiation ends before stomatal initiation begins, the mechanism by which this pathway would function is unclear.
Light levels, atmospheric CO2 concentration, humidity and drought signals all have an effect on the level of stomatal development. As stomata are intrinsically linked to the rate of CO2 uptake and fixation, and water loss, it is unsurprising that stomatal numbers should be sensitive to the environmental conditions that affect these central concerns of the plant. This suggests that there are several levels of control over stomatal development: control over the spacing of individual cells, to ensure that stomata are separated by at least one epidermal pavement cell; control of stomatal numbers as part of tissue patterning during development; and control of stomatal numbers by environmental conditions. These different levels of control must feed through synergistically to produce the final numbers of stomata found on mature leaves.
The analysis of plant cuticles has shown that their wax composition can vary in response to environmental conditions. The study of plants with wax biosynthetic pathway mutations has revealed a possible link between cuticle wax composition and stomatal development. Several wax biosynthesis mutants, such as cer1 and cer6/cut1 in A. thaliana, have an altered stomatal phenotype as well as having altered wax compositions. Both cer1 and cer6/cut1 plants have an increased stomatal density at ambient atmospheric CO2, whereas the hic mutant displays an increased stomatal number only at elevated CO2 concentrations. hic is a mutant in a putative guard cell fatty acid elongase condensing enzyme, which suggests a role for the cuticular wax composition in the environmental control of stomatal development. Other mutants with altered cuticle composition, such as fdh and lcr have made a link between cuticle composition and trichome numbers. LCR and FDH may also have an effect on stomatal development, but this has not been investigated. A mutant that specifically affects trichome number in response to environmental conditions has yet to be identified so a direct link between trichome numbers and environmental conditions cannot be made.
This link between cuticle wax composition and epidermal cell fate determination raises the possibility of the existence of a novel signalling pathway, that acts through the medium of the cuticle. Epidermal cells are intrinsically linked to the cuticle as the external cell walls of epidermal cells become integrated into the cuticle as it develops. Stomatal guard cells loose their plasmodesmatal connections early in their development and therefore must utilise other signalling pathways to communicate with surrounding cells. The identification of signalling pathways that function directly through cells walls, such as wall activated kinases (WAK) (Anderson et al., 2001), suggest that plant cells utilise many different signalling pathways to communicate with their surrounding cells. It is possible that epidermal cells, and particularly guard cells, may use the cuticle as a continuous signalling medium that all epidermal cells are in contact with. The cuticle is not a completely crystalline structure. There are areas of wax fluidity that molecules are able to diffuse through. This could be the case in the immature procuticle that covers the epidermis during development, as epicuticular wax crystals only begin to appear on the surface of more mature cuticles (Jeffree, 1996). Plants maintain a mutualistic flora on their aerial surfaces by the slow release of nutrients from the cells beneath. Therefore it is entirely possible that signalling molecules could pass through the cuticle too and that this process is affected by the composition and fluidity of the cuticular waxes. Alternatively, it is possible that components of the waxes themselves, such as individual VLCFA species derivatives, could be acting as the developmental signalling molecules. What is clear is that cuticle and epidermal development are highly integrated and the nature of their interaction is complex.
We gratefully acknowledge the help of Mr J. Proctor with microscopy, Mr R. Adams with figure drawing and Dr G. Holroyd for the gift of hic and C24 plant material used for SEM work. We thank Prof. FI Woodward for his critical reading of the manuscript. This work is funded by BBSRC and NERC.