Stomata play a pivotal role in the regulation of gas exchange in flowering plants and are distributed throughout the aerial epidermis. In leaves, the pattern of stomatal distribution is highly variable between species but is regulated by a mechanism that maintains a minimum of one cell spacing between stomata. In Arabidopsis, a number of the genetic components of this mechanism have been identified and include, SDD1, EPF1 and the putative receptors TMM and the ERECTA-gene family. A mitogen-activated protein (MAP) kinase signalling cascade is believed to act downstream of these putative receptors while a number of transcription factors including SPCH, MUTE and FAMA have been identified that control consecutive steps of stomatal development. The environment also has significant effects on stomatal development. In a number of species both light intensity and CO2 concentrations have been shown to influence the frequency at which stomata develop on leaves. Long-distance signalling mechanisms have been implicated in these environmental responses with the conditions sensed by mature leaves determining the stomatal frequency in developing leaves. Thus, changes in the environment appear to act by modulating the developmental and patterning pathways to determine stomatal frequency.
The epidermis of the aerial parts of flowering plants contains numerous stomata, which consist of a pair of guard cells flanking a microscopic pore. The sessile nature of plants means that they must constantly adapt to variations in their environment, and stomata are vital for this function. They regulate gas exchange, mainly CO2 and water vapour, with the environment allowing the plant to optimize and balance photosynthetic performance with water availability and usage (Chaerle et al., 2005). Regulation is achieved by opening and closing of the stomatal pore thereby either increasing or reducing stomatal conductance, or the rate by which water or CO2 is exchanged (Roelfsema & Hedrich, 2005). A second strategy employed by plants to adapt to the prevailing environmental conditions is to modulate the frequency at which stomata develop in new organs. Changes in frequency can be expressed as either stomatal density per unit area, or as stomatal index, the proportion of epidermal cells that are stomata. There is a growing body of data that indicates that this mechanism involves mature organs sensing the environmental conditions and then signalling to the developing primordia (Schoch et al., 1980; Lake et al., 2001; Coupe et al., 2006; Miyazawa et al., 2006).
We will discuss the latest research regarding the molecular basis of stomatal patterning and development. With an understanding of this framework we will then consider the influence of environmental factors on the pattern of development and discuss how these factors may integrate their signals into the stomatal development pathway.
II. Stomatal patterning in leaves
The mature leaf epidermis generally consists of three cell types: trichomes, stomatal guard cells and pavement cells. In Arabidopsis, trichomes are the first cells to differentiate and develop in a basipetal (from leaf tip to base) manner (Larkin et al., 1996). Stomata also develop in a basipetal manner and can form on both leaf surfaces (amphistomatous) or be restricted to one surface only (hypostomatous); their differentiation is the last aspect of leaf development (Larkin et al., 1996; Serna et al., 2002).
Stomatal patterning in the leaf epidermis is species dependent (Fig. 1). In most cases, stomatal density is greatest on the abaxial leaf surface, which may help prevent water loss since the abaxial surface is less exposed to heating (Martin & Glover, 2007). In a number of species, particularly the monocotyledonous plants, stomata and pavement cells are arranged in a regular pattern of alternating cell files (Fig. 1a,b; Croxdale, 2000), whereas in dicotyledonous plants stomatal patterning appears to be relatively random (Fig. 1d; Sachs, 1978). Few studies have examined the relationship between stomatal patterning and the arrangement of cells in the underlying layers. In the adaxial leaf surface of Arabidopsis nearly all stomata are positioned over the junction of several mesophyll cells. This positioning is significantly different from that predicted for a purely random distribution, suggesting the possibility of an inter-layer positional signalling mechanism and draws parallels with the patterning of stomata on the hypocotyl and root hairs in the root epidermis (Berger et al., 1998; Serna & Fenoll, 2000). Also, stomata rarely develop over major veins (Martin & Glover, 2007), although members of the Selaginellaceae appear to be an exception (Fig. 1c; Brown & Lemmon, 1985). In Tradescantia, the underlying vascular tissues develop in longitudinal tracts along the length of the leaf, and stomata are restricted to the intervening tracts (Croxdale, 1998). The absence of stomata over veins may help prevent water loss or relate to the fact that photosynthetically active palisade mesophyll cells are seldom found in close proximity to vascular bundles. It seems probable that stomatal patterning is influenced by internal anatomy although it is not known if there is an active signalling mechanism involved. The maize extra cell layers1 (xcl1) mutant produces extra epidermal layers and both epidermal surfaces show a significant reduction in stomata (Kessler et al., 2002). It is possible that the extra epidermal layers block or reduce a signal from the underlying cell layers.
In the epidermis, mature stomata are usually separated from each other by a minimum of one epidermal cell (Sachs, 1978), which may ensure optimal stomatal function. Guard cell activity requires water and ion exchange with surrounding cells and the subsequent changes in turgor also impose mechanical stresses on the surrounding cells. The one-cell spacing pattern also ensures that there is minimal overlap between gas diffusion shells.
In monocots and dicots an asymmetric division initiates the stomatal lineage, with the resulting daughter cells acquiring separate fates (Larkin et al., 1997). This is also observed in stem cell populations where the asymmetric division produces a new stem cell and a cell that enters a differentiation pathway. Unlike the root and shoot stem cell populations, stomatal initials have a limited ability to self renew. Generally, in monocots, the division produces a smaller stomatal initial, which is often rectangular, and a larger neighbour cell. In a cell file, this asymmetric division is orientated such that the stomatal initial forms closest to the leaf tip. In some monocot species, this stomatal initial, the guard mother cell (GMC), undergoes a symmetrical division to form two guard cells. However, stomata are often associated with subsidiary cells that form on the flanks (Fig. 1b). These subsidiary cells may develop from the stomatal initial itself or from neighbouring cells (Croxdale, 1998). The regular division pattern ensures that mature stomata within files are always separated from each other by at least one neighbour cell, with a file of pavement cells separating the stomatal files.
Our understanding of the mechanisms controlling stomatal development has advanced significantly in the last few years owing to studies in the model dicot, Arabidopsis thaliana. Stomatal development in Arabidopsis begins with a protodermal cell converting to a meristemoid mother cell (MMC) that then divides asymmetrically to produce a smaller triangular meristemoid and a larger sister cell (Fig. 2). This initial division is known as the entry division. Further asymmetrical amplifying divisions can occur regenerating the meristemoid and a larger daughter cell. The number of amplifying divisions is ecotype dependent and is normally between zero and three. Finally, the meristemoid converts into a GMC that divides only once, symmetrically, to form the two guard cells. Stomatal morphogenesis then follows during which the guard cell walls undergo thickening and then separate to form the pore.
Both sister and daughter cells can also undergo asymmetric spacing divisions to produce new satellite meristemoids. Indeed, the vast majority of stomata originate from satellite meristemoids (Geisler et al., 2000). The spacing divisions are orientated such that satellite meristemoids form away from the original stoma or precursor (Geisler et al., 2000). By contrast the orientation of entry divisions of isolated MMC is random. If two MMC develop next to each other, their asymmetric entry divisions can result in the two meristemoids forming adjacent to each other, suggesting that isolated MMC either do not respond to, or do not issue the positional cues for orientated divisions (Lucas et al., 2006). In this situation, one of the adjacent meristemoids will either undergo an orientated spacing division, placing a daughter cell between the two meristemoids, or differentiate into a pavement cell (Geisler et al., 2000; Lucas et al., 2006). This mechanism has also been observed in monocots and shows that mistakes in the spacing pattern can be corrected (Boetsch et al., 1995; Geisler et al., 1998).
The earliest indication that a cell is about to undergo an entry, spacing or amplifying division is a polarized cytoplasm and the appearance of the preprophase band of microtubules surrounding the nucleus (Lucas et al., 2006). The spacing mechanism is not believed to involve the asymmetric allocation of signalling molecules during division. This is based on observations that cells neighbouring a meristemoid, GMC or stoma undergo correctly orientated asymmetric divisions even if they are not clonally related to the stoma or initial. Spacing is therefore predicted to involve an intercellular signal emanating from the stoma or initial that acts to orientate the division plane.
III. The genetic control of stomatal development
One of the first components of stomatal patterning and development to be identified was the TOO MANY MOUTHS (TMM) gene (Yang & Sack, 1995; Nadeau & Sack, 2002). Mutations in TMM lead to both excess and clustered stomata in the leaf epidermis. This is caused by failure to correctly orientate spacing and amplifying divisions and failure to inhibit entry divisions in cells adjacent to two or more stomata or their initials (Geisler et al., 2000). The tmm mutant also shows a reduced number of amplifying divisions with meristemoids prematurely converting to GMCs. Therefore, the primary role of TMM is to inhibit divisions in response to positional cues from stomatal lineage cells. TMM encodes a putative cell-surface leucine-rich repeat (LRR)-containing receptor-like protein and is expressed in young leaf primordia, meristemoids, sister and daughter cells and GMCs, but not mature stomata; it is also expressed in cells likely to undergo an entry division (Nadeau & Sack, 2002). TMM expression correlates with competence to divide since expression is strongest in the youngest neighbour cells, which are the most likely to divide (Berger & Altmann, 2000). Conversely, it is often found in cells adjacent to two or more stomata or their initials. Such cells rarely divide suggesting that TMM may be required to inhibit their division (Nadeau & Sack, 2002). Together the data are consistent with TMM regulating the orientation and number of amplifying and spacing divisions in response to positional signals in order to maintain correct spacing. It may also regulate the competence of cells to divide and differentiate in response to these positional cues.
TMM is related to the LRR-receptor like kinases (LRR-RLKs), which form a large family in Arabidopsis (Johnson & Ingram, 2005). Unlike the LRR-RLKs, TMM lacks a cytoplasmic kinase domain (Nadeau & Sack, 2002). However, other LRR-RLP (receptor like proteins) such as CLAVATA2 are predicted to form functional complexes with a LRR-RLK, in this example CLAVATA1 (Jeong et al., 1999). Therefore, TMM could act in conjunction with an LRR-RLK to regulate stomatal development. Candidates for this role are the ERECTA family which consists of three members in Arabidopsis; ERECTA (ER), ERL1 (ERECTA-LIKE1) and ERL2. They work together to coordinate various aspects of cell division and growth in aerial organs (Shpak et al., 2004, 2005; Pillitteri et al., 2007a). ER has also been identified as a major determinant of transpiration efficiency, in part because of changes in stomatal density (Masle et al., 2005). In leaves, these genes are initially expressed throughout the protoderm of leaf primordia. Expression of ER rapidly diminishes, but ERL1 and ERL2 expression remains in the stomatal lineage, in a manner similar to TMM. Phenotypic analysis of mutations in the ER family members indicates subtly different roles for these genes in stomatal patterning. The phenotypes indicate that ER predominantly inhibits entry divisions but may promote meristemoid differentiation: ERL1 appears to inhibit meristemoid differentiation and ERL2 regulates amplifying divisions (Shpak et al., 2005). The er erl1erl2 triple mutant has a dramatic phenotype with severe stomatal clustering in all organs examined. In all cases, the presence of a single ER family member is sufficient to ensure correct spacing.
The relationship between the ER family and TMM appears far from simple (Fig. 3b). One difficulty is that each gene appears to have different roles in stomatal development depending on the organ examined. For example, the tmm mutant has clustered stomata on leaves and yet inflorescence stems are devoid of stomata, indicating that TMM does not always act as a negative regulator (Yang & Sack, 1995). Double mutants between tmm and er or erl2 also do not produce stomata on their stems, suggesting they function in a common pathway, whereas tmmerl1 double mutants do have small numbers of stomata. Perhaps the most telling observations occur in siliques; the tmm mutant shows small clusters of stomata whereas tmmer and tmmererl2 mutants produce clusters of stomatal lineage cells but no mature stomata. Stomatal differentiation is restored in triple and quadruple mutants with erl1 (Shpak et al., 2005). One interpretation of these results is that there is competition between the proteins for a common partner. If TMM and the ER family do function together in a complex, it may be that the observed phenotypes result from both organ-specific developmental programmes, with TMM on occasion acting independently of the ER family, and temporal changes in the nature of the complex, with TMM interacting with different family members at different phases of stomatal development. On occasion, TMM also appears to inhibit ER family function, in particular ERL1. Whether TMM interacts directly with the ER-family, forming heterodimers, remains to be seen. Interestingly, a truncated ER protein lacking the kinase domain was able to interfere with the function of the native ER and other RLKs suggesting that altering the composition of a RLK complex can modulate its activity (Shpak et al., 2003). Since the ER family have broad roles throughout plant development, it is reasonable to suggest that interactions with TMM could confer specificity to stomatal development.
Recently, AGL16, a MADS box transcription factor, has been implicated in controlling amplifying divisions during stomatal development (Kutter et al., 2007). AGL16 transcript levels are regulated by MicroRNA 824 (miR824) and both are expressed in the stomatal lineage (Alvarez-Buylla et al., 2000; Kutter et al., 2007). Plants expressing a miRNA resistant form of AGL16 have a greater number of higher order stomatal complexes whilst agl16 mutants or plants overexpressing miR824 have fewer; entry divisions are not affected. LRR-RLKs have been shown to exist in complexes with cell cycle regulators, phosphatases and transcription factors which are believed to modulate receptor activity, cellular localization and may even directly regulate transcription of the LRR-RLK (Shah et al., 2002; Rienties et al., 2005; Karlova et al., 2006). AGL15, a close relative of AGL16, has been found in a complex with the LRR-RLK SERK1. The mechanism by which AGL16 affects amplifying divisions is unknown but it would be interesting to determine if AGL16 interacts with or regulates TMM or the ER-family.
A receptor-based signalling mechanism governing the divisions of the stomatal lineage implies the existence of both ligands and a downstream signalling cascade (Fig. 3a). The identification of the stomatal density and distribution 1 (sdd1) mutant, which shows increases in stomatal index and clustering, gives an intriguing insight into the nature of a potential ligand (Berger & Altmann, 2000; von Groll et al., 2002). SDD1 is predicted to regulate the number of entry and amplification divisions as well as the orientation of spacing divisions. SDD1 encodes a putative subtilisin-like serine protease (Berger & Altmann, 2000), which in animal and plant systems have been shown to activate signalling components by processing their precursors (Schaller, 2004). SDD1 is expressed in meristemoids and GMCs, but not in mature guard cells. Ectopic expression results in a decreased stomatal index and a higher incidence of arrested meristemoids and GMCs. The tmm mutation was completely epistatic to SDD1 overexpression (von Groll et al., 2002), indicating that TMM and SDD1 are likely to act in a common pathway. The SDD1 protein appears to be exported to the apoplast where it is localized to the plasma membrane. It is therefore possible that SDD1 may be involved in processing a peptide precursor of a TMM ligand.
To identify peptide regulators of plant development, Hara and colleagues (2007) overexpressed a large number of predicted small, secreted peptides. Overexpression of EPIDERMAL PATTERNING FACTOR 1 (EPF1) led in severe cases to complete inhibition of stomatal development. Weaker phenotypes showed a severe reduction in stomatal numbers but no other developmental abnormalities, suggesting a stomatal specific function. An EPF1 loss of function allele showed increased stomatal density and clustering as a result of incorrect spacing divisions, confirming a role for EPF1 in stomatal patterning. EPF1 is expressed in meristemoids, GMCs and young guard cells. Genetic analysis indicates that EPF1 acts in the same pathway as TMM and the ER family and is a strong candidate for a TMM ligand. EPF1 is predicted to have an N-terminal secretory signal that is cleaved to release the mature peptide. However, a simple model in which SDD1 processes the EPF1 precursor does not appear to correlate with the genetic data. The epf1sdd1 double mutant had an additive phenotype with regard to stomatal clustering, suggesting that they act independently and yet the genetic data is consistent with both EPF1 and SDD1 being in the same pathway as TMM (von Groll et al., 2002; Hara et al., 2007). It is possible that EPF1 is indeed a ligand for TMM, and that SDD1 processes a different, as yet unidentified ligand, or another component of the pathway, even TMM itself. SDD1-like shares a high level of identity with SDD1 and has been implicated in epidermal development (Coupe et al., 2006). It remains to be determined if EPF1 is cleaved and if this is required for its function; however, it is possible that another protease such as SDD1-like may perform the cleavage.
Downstream of the receptor–ligand interactions, a MAP kinase signalling cascade negatively regulates stomatal development (Bergmann et al., 2004; Wang et al., 2007). Mutations in the YODA gene (YDA), a putative MAP kinase kinase kinase (MAPKKK), results in an excessive number of clustered stomata, similar to that observed in ererl1erl2 plants (Bergmann et al., 2004). In addition, in yda seedlings the division products of a meristemoid both become guard cells. Plants designed to express a constitutively active truncated version of YODA (ΔN-YDA), in which the N-terminal negative regulatory domain is removed, do not develop any guard cells, indicating that a switch to stomatal cell fate requires lack of activation of, or inactivation of YDA. Genetic analysis places YDA downstream of TMM and SDD1 though this does not discount an independent role early in protodermal development. YDA is expressed at low levels throughout the plant and is also required for cell fate decisions during embryogenesis (Lukowitz et al., 2004). Therefore, the role of YDA in the negative regulation of stomatal cell fate is likely to be specified by direct or indirect interactions with specific receptors, most likely TMM or the ER-family.
The MAP kinase signalling cascades are highly conserved with a MAPKKK signalling to downstream MAPKKs and MAPKs by serine/threonine phosphorylation (Mishra et al., 2006). Recently candidates acting in stomatal development in both the MAPKK and MAPK classes have been identified (Wang et al., 2007). The MAPKs, MPK3 and MPK6 redundantly control stomatal patterning and development. Individual mutants showed no defects; however, the use of RNA interference (RNAi) to downregulate expression of both genes resulted in excessively clustered stomata in the cotyledon epidermis, much like yda mutants. Single MKK4 and MKK5 RNAi plants showed weak stomatal clustering, but when combined the cotyledon epidermis again resembled yda. Analysis of MPK3/MPK6 and MKK4/MKK5 action was consistent with these genes acting in a common pathway with YDA to control stomatal development (Wang et al., 2007). Both MPK3 and MPK6 are also activated in biotic and abiotic stress responses (Kovtun et al., 2000; Asai et al., 2002), but it seems that activation through YDA confers specificity to the stomatal pathway. It has been shown that an Arabidopsis Ser/Thr phosphatase of type 2C (AP2C1) regulates stress responses by inactivating MPK4 and MPK6 but not MPK3 (Schweighofer et al., 2007). It is possible that components of the stomatal MAPK signalling pathway may be regulated by phosphatases, thereby adding a further level of control.
Recently, three closely related basic helix–loop–helix (bHLH) transcription factors SPEECHLESS (SPCH), MUTE and FAMA have been shown to be required for consecutive steps in stomatal developmental (Fig. 2; Ohashi-Ito & Bergmann, 2006; MacAlister et al., 2007; Pillitteri et al., 2007b). Mutations in SPCH severely affect the ability of protodermal cells to undergo entry divisions and severe spch alleles do not produce any stomata. Analysis of the weaker spch-2 allele indicates that SPCH may also promote spacing and amplifying divisions (MacAlister et al., 2007). SPCH expression coincides with the onset of post-embryonic cell division and is initially seen broadly throughout the protoderm of cotyledons and leaf primordia but gradually becomes restricted to a limited number of cells competent to undergo entry divisions; expression is absent in the latter stages of stomatal development. Increased expression of SPCH in wild-type plants leads to excess cell division and an epidermis with numerous small cells expressing TMM. In a tmm background, increased dosage of SPCH leads to excessive stomatal formation, indicating that TMM is inhibiting stomatal differentiation (MacAlister et al., 2007; Pillitteri et al., 2007b). Therefore, SPCH is both required and sufficient to initiate stomatal development and, furthermore, its expression is required for the expression of both MUTE and FAMA (MacAlister et al., 2007).
Mutations in MUTE also lead to a total absence of stomata. However, unlike spch, epidermal cells of mute seedlings can undergo entry divisions to form meristemoids. These meristemoids undergo excessive rounds of amplifying divisions before arresting without forming a GMC (Pillitteri et al., 2007b). Expression of MUTE is strongest in the youngest meristemoids and overexpression can result in the whole epidermis converting to guard cells or cells with mixed pavement and guard cell identity (MacAlister et al., 2007; Pillitteri et al., 2007b). Therefore, MUTE is required to terminate asymmetric divisions allowing the transition of the meristemoid to GMC and may limit the number of amplifying divisions (Pillitteri et al., 2007b).
The final cell fate decision, the transition of the GMC to two guard cells, is controlled by FAMA (Ohashi-Ito & Bergmann, 2006). Instead of recognizable stomata, fama mutants develop clusters of small, narrow epidermal cells. These cells express markers found in earlier stomatal lineages up to the GMC but not guard cell markers. Ectopic or overexpression of FAMA confers guard cell characteristics on a variety of cell types suggesting that FAMA not only acts to restrict cell division following the symmetrical division of the GMC but also specifies guard cell identity.
Defects in the R2R3 MYB transcription factors FOUR LIPS (FLP) and MYB88 genes also affect the GMC to guard cell transition (Lai et al., 2005). FLP is expressed in GMC and developing stomata. Weak flp mutants develop conjoined stomata whereas severe alleles contain many more cells clustered together, much like fama. These clusters contain normal and arrested stomata suggesting that FLP, unlike FAMA, is not required for guard cell differentiation. Instead, FLP is required to terminate cell division following the symmetrical division of the GMC. It performs this function along with another closely related gene, MYB88. Mutations in MYB88 enhance the defects observed in flp mutants but do not have a phenotype themselves. However, increased dosage of MYB88 is able to complement the flp-1 mutant, suggesting functional overlap.
The fama, flp and myb88 mutants are phenotypically similar and have similar expression patterns; however, both genetic and molecular evidence suggests that they act independently to direct GMC division and differentiation. A yeast two-hybrid screen indicates that FAMA may interact with two other bHLH transcription factors, bHLH093 and bHLH071, which are widely expressed (Ohashi-Ito & Bergmann, 2006). It is postulated that FAMA is required to activate guard cell differentiation while also feeding back to inhibit further cell division while FLP and MYB88 act independently to inhibit division.
IV. The cell cycle and stomatal development
Stomata develop late in leaf development and therefore the ability of cells to enter and exit the cell cycle is important for determining the number of entry, spacing and amplifying divisions. The stomatal lineage is also unusual in that it is diploid, whereas pavement and trichome cells undergo rounds of endoreduplication during which the genome replicates without mitosis. Mutant analysis strongly suggests that the cell cycle machinery is a target of the stomatal developmental pathway (Lai et al., 2005; MacAlister et al., 2007) and it has been proposed that a cell's position in the cell cycle determines its competence to form stomata (Charlton, 1990, Croxdale, 2000). To date, very few individual cell cycle components have been directly implicated in stomatal development, although new research suggests mechanisms by which fate may be linked to the cell cycle.
The key regulators of the cell cycle are the cyclin-dependent kinases (CDKs), which control the G1-S and G2-M transitions to determine when the cell embarks on DNA replication or cell division respectively. Of the seven classes of CDKs, the A- and B-type CDKs are the major regulators of the cell cycle. In Arabidopsis CDKA1 activity peaks at G1/S and G2/M whereas the B-type CDKs, which are specific to plants, peak during the G2 and G2/M transition. Cell cycle progression involves the formation of ordered complexes between the CDKs and cyclins. With 12 confirmed CDKs and at least 49 different cyclins in Arabidopsis (Vandepoele et al., 2002) there is significant scope for both complexity and redundancy. Targeted inhibition or degradation of these CDK–cyclin complexes is crucial for determining the ability of a cell to undergo further rounds of cell division or to exit the cell cycle (De Veylder et al., 2007).
CDKB1;1 has been shown to be involved in stomatal development and unlike the more generally expressed CDKA;1, is expressed in the stomatal lineage (Boudolf et al., 2004a). A reduction in CDKB1;1 activity by expressing a dominant-negative form of CDKB1;1 resulted in leaves with fewer stomata and an absence of satellite meristemoids, suggesting a role in entry and amplifying divisions. A significant proportion of the stomata developed as single abnormal cells that had the differentiated features of mature stomata but were blocked at the G2 phase of the cell cycle, before cell division. These cells had a 4C DNA content indicating that both cellular and nuclear division are independent of stomatal differentiation. Cells expressing the dominant-negative CDKB1;1 also showed increased endoreduplication, indicating that CDKB1;1 acts to repress the endocycle (Boudolf et al., 2004b). Ploidy levels themselves do not direct stomatal development since manipulation of the levels of the cell-cycle inhibitor KIP RELATED PROTEIN2 (KRP2) or the cell cycle activator cyclin CYCD3:1, alter the levels of endoreduplication but not stomatal index (de Veylder et al., 2001; Dewitte et al., 2003). It is possible that CDKB1;1 may be a direct target of SPCH and termination of divisions in the stomatal lineage by FLP and MYB88 could also result from inhibition of CDBK1;1 activity.
The D-type cyclins which predominantly regulate G1/S transitions have also been implicated in stomatal development. Mutants in both CYCD4;1 and CYCD4;2 have a reduced number of stomata in the hypocotyls, while overexpression results in an increase of hypocotyl cells expressing TMM. The phenotypes are consistent with the CYCD4 genes functioning early in stomatal development in hypocotyls; however, leaf stomatal development was not affected suggesting organ specific influences (Kono et al., 2007).
The RETINOBLASTOMA-RELATED (RBR)/E2F pathway regulates the switch to differentiation (Wildwater et al., 2005; Desvoyes et al., 2006; del Pozo et al., 2006; Wyrzykowska et al., 2006). In plants, the RBR-E2F pathway is known to regulate, in part, the G1-S and G2-M transitions. RBR negatively regulates E2F activity; E2Fa and E2Fb act as positive regulators of the cell cycle while E2Fc acts as a negative regulator. Downregulation of RBR, or overexpressing both E2Fa and DPa, which form a transcriptional activator complex that activates DNA replication, both resulted in excessive cell division in the leaf or hypocotyl. However, in neither case did this result in increased stomatal development (de Veylder et al., 2002; Desvoyes et al., 2006). This contrasts with plants overexpressing CDC6 and CDT1, which regulate the licensing of DNA origins of replication and are transcriptional targets of E2Fs (Castellano et al., 2004). Overexpressing CDC6 and CDT1 caused significant increases in the stomatal index. Therefore, modulation of RBR/E2F activity or their targets is another mechanism by which regulators of stomatal development could effect changes in stomatal pattern, with CDKB1;1, CDC6 and CDT1 being possible integration points.
It has been recently shown that CDT1 interacts with GL2 expression modulator (GEM). GEM also binds to TTG1, which regulates expression of GL2 and CAPRICE (CPC), disrupting the interaction with CDT1 (Caro et al., 2007). Both GL2 and CPC are key regulators of root hair and trichome fate and patterning (Guimil & Dunand, 2006). GEM was shown to associate with the promoters of GL2 and CPC resulting in a reduction in the active chromatin marks H3K9acK14ac and H3K9me3. These marks are post-translational modifications of the histones that package DNA and can distinguish transcriptionally active and inactive regions of the genome. Analysis of these marks in the GL2 and CPC promoters was consistent with these genes being active early in the cell cycle and repressed at G2/M (Caro et al., 2007), which correlates with CDT1 expression (Castellano et al., 2004). Chromatin marks in the GL2 promoter are reset after each cell division thereby determining the ability of a daughter cell to express GL2 (Costa & Shaw, 2006). Therefore, the data indicate that CDT1 and GEM are part of a mechanism that integrates the cell cycle with key transcriptional regulators thereby regulating cell proliferation and cell fate.
There are no data to suggest that GEM, which is ubiquitously expressed, acts during stomatal development. The reorganization of chromatin marks in the GL2 locus occurs in response to positional cues following cell division (Costa & Shaw, 2006). Such a mechanism could operate in the stomatal lineage. It is possible that SPCH or other regulators of stomatal fate are targets or components of such a mechanism with positional information supplied by TMM or the ER family receptors. Therefore, while a number of the individual components of both the cell cycle and stomatal fate have been identified, it will be interesting to determine how they are integrated and if, as with root hair and trichome patterning, epigenetic factors regulate the changes in cell fate.
V. Environmental influences on stomatal pattern
Plants are able to balance CO2 intake with water loss by regulating the stomatal pore aperture and guard cell signalling is regulated by both environmental factors and plant hormones (Hetherington, 2001; Hetherington & Woodward, 2003; Chaerle et al., 2005; Israelsson et al., 2006). Changes in both CO2 and light levels have also long been known to elicit changes in stomatal numbers (Salisbury, 1927; Tichá, 1982; Woodward, 1987). In many species, the trend is for a reduction in stomatal density and index with increases in atmospheric CO2 levels both in geological time and under laboratory conditions (Woodward, 1987). In a study of 100 species it was found that three-quarters of the species analysed showed reductions in stomatal density, averaging a 14.3% reduction, when grown at elevated CO2 levels. In controlled conditions 60% of the species tested (n = 43) showed reductions in stomatal density averaging 9% when grown at elevated (700 ppm) compared to ambient (350 ppm) levels of CO2; in these controlled conditions amphistomatous species generally showed a greater difference than hypostomatous ones. It was also observed that initial stomatal density was a significant factor. Species with higher initial stomatal densities tended to be more responsive to increased CO2 levels; other factors such as habitat did not appear to significantly alter the responsiveness (Woodward & Kelly, 1995). Exceptions have been observed with either no change or an increase in stomatal index in elevated CO2, and different responses have been observed even within species (Woodward & Kelly, 1995; Case et al., 1998; Lawson et al., 2002; Driscoll et al., 2006). In Tradescanti leaves it was found that growth in elevated CO2 led to increases in the number of subsidiary cells but no change in stomatal index (Boetsch et al., 1996).
Cuvette systems have allowed developing and mature leaves to be isolated and subjected to different experimental conditions (Lake et al., 2001; Driscoll et al., 2006; Miyazawa et al., 2006). Using these systems has demonstrated that the stomatal pattern of developing leaves is influenced by the conditions experienced by the mature leaves. In Arabidopsis, growth of mature leaves at elevated CO2 resulted in a reduced stomatal index in developing leaves that were exposed to ambient CO2 levels; in reciprocal experiments where mature leaves were exposed to ambient CO2 and the developing leaves to elevated CO2, the stomatal index was increased in developing leaves (Lake et al., 2001).
The HIGH CARBON DIOXIDE (HIC) gene of Arabidopsis has been identified as having a role in modulating changes in stomatal index in response to elevated CO2 (Gray et al., 2000) and several potential quantitative trait loci have been identified in poplar (Ferris et al., 2002, Rae et al., 2006). When exposed to elevated levels of CO2, both hic and hic/+ plants showed a significant increase in stomatal index unlike the parental C24 ecotype, which showed a small decrease. Either the number of stomatal entry or amplifying divisions must be affected at elevated CO2 but the hic mutant does not exhibit abnormal clustering and the one-cell spacing rule is not affected. HIC is expressed in guard cells and shares high homology with the Arabidopsis KCS1 gene, a 3-ketoacyl CoA synthase (KCS) involved in the production of very long chain fatty acids found in the cuticular waxes. The mechanism by which HIC affects stomatal patterning in response to CO2 is unknown. However, other mutants with altered epicuticular wax profiles show defects in stomatal development. Both cer1 and cer6 mutants have increased stomatal indices (Gray et al., 2000) while wax2 and gain-of-function shine mutants have reduced stomatal indices (Chen et al., 2003; Aharoni et al., 2004). The cuticular wax layer forms a barrier that protects against various environmental stresses (Shepherd & Wynne Griffiths, 2006). Alterations in the cuticular wax profile may alter permeability to water, CO2 or another signalling compound. It is possible that a cuticular wax or an intermediate is a signalling compound that influences stomatal development. In addition, guard cells show greater blue fluorescence than pavement cells, potentially because of either thicker or different cuticular wax (Karabourniotis et al., 2001). Therefore, another possibility is that mutations in wax biosynthesis affect the absorption spectra of the tissue.
Stomatal patterning also responds to changes in light intensity. In general, an increase in light intensity results in an increase in stomatal index and mature leaves again determine the response (Schoch et al., 1980; Lake et al., 2001; Thomas et al., 2003; Coupe et al., 2006). In cowpea (Vigna sinensis), shifting between high and low light intensities showed that the light intensity perceived by the mature leaves during a 6-d period before unfolding of a new leaf blade determines the stomatal index of the new leaf (Schoch et al., 1980). Similarly, in tobacco, shading of mature leaves while exposing developing leaves to high-intensity light resulted in a 12.7% decrease in stomatal index of the developing leaves compared with controls exposed only to high intensity light. Conversely, shading of developing leaves in combination with exposure of the mature leaves to high intensity light gave a 24.2% increase compared with the controls (Thomas et al., 2003). The effect of photoperiod on stomatal index was also examined in cowpea using night breaks (Schoch et al., 1984). Plants were grown in 9 h of light and 15 h darkness. If low-intensity blue or far-red light was given instead of darkness, the developing leaves had a reduced stomatal index whereas low-intensity red light caused an increase in index. The same responses were observed if only a 30-min far-red or red light treatment was given at the start of the dark period. The far-red effect was reversed if immediately followed by a 30-min red light treatment. Light is perceived by various photoreceptors (Franklin et al., 2005), stomatal movements are regulated by both blue and red light (Shimazaki et al., 2007). It would be interesting to determine if photoreceptor-mediated signalling, light-regulated stomatal movements or changes in photoperiod mediate the stomatal index responses observed by Schoch and colleagues (1984).
Genes regulating this response to changes in light intensity have not been identified, although the response has been analysed using transcriptomics (Coupe et al., 2006). In these experiments, developing leaves were exposed to high-intensity light whereas mature leaves were shaded, and the data was compared with controls. Significant alterations in gene expression were found in both the shaded mature leaves and the developing leaves. For example, a number of auxin-related genes were upregulated specifically in the mature shaded leaves while the developing leaves had, amongst others, increases in both SDD1 and SDD1-like. This is consistent with previous data showing that overexpression of SDD1 causes a decrease in stomatal index, which correlates with the reduced stomatal index of shaded leaves. However, SDD1 is not required for the responses to light, although this may be due to redundancy (von Groll et al., 2002).
While other environmental factors are known to affect stomatal pore aperture very little is known about their effects on stomatal development. Water stress resulting from a reduced soil-watering regime caused a reduction in stomatal index in Caltha palustris and wheat (Salisbury, 1927; Quarrie & Jones, 1977). However, drought conditions did not alter the stomatal index in groundnut (Clifford et al., 1995). Increased humidity resulted in a reduction of the stomatal index of Scilla nutans leaves (Salisbury, 1927). Arabidopsis grown in closed containers showed increased stomatal density, although the stomatal index was not reported (Serna & Fenoll, 1997). Interestingly, a number of stomata were present as clusters, though it is not possible to determine if this was a result of increased humidity or other variables resulting from growth in the closed conditions.
Little is known about the role of long-distance signalling in stomatal development including how and where the environmental signals are perceived. For example, it is not clear whether the stomatal lineage cells or another cell type perceives the environmental stimuli. There is a large body of research regarding light perception and signalling (Chen et al., 2004; Franklin et al., 2005) and although stomatal movements are regulated by light (Shimazaki et al., 2007) little is known about how the light-signalling components regulate stomatal development. The mechanism by which plants perceive CO2 is still unclear, although it may be through metabolism (Long et al., 2004). A potential component of the CO2 sensing pathway in stomata is HIGH LEAF TEMPERATURE 1 (HT1) which encodes a putative protein kinase (Hashimoto et al., 2006). HT1 is part of a pathway that controls stomatal movements in response to lower than ambient CO2 concentrations. While HT1 is expressed in guard cells, unlike hic mutants there are no changes in stomatal density in ht1 plants although this was only determined at ambient CO2 levels at which hic mutants also show no difference.
There is also the question of where the long-distance signal is generated and how it is transported to developing leaves. At present no grafting experiments have been performed and so it is unclear if mature leaves actually generate and transmit a signal. Wherever a signal is produced it is likely that it must enter a long-distance transport stream such as the phloem or xylem or elicit the transport of a mobile signal, as may be the case in wound signalling with systemin and jasmonic acid (Wasternack et al., 2006). A cuvette and shading system was used to manipulate CO2 levels, vapour pressure deficit and irradiance of lower leaves in poplar and the effects on stomatal development in new leaves was analysed (Miyazawa et al., 2006). Factors such as net photosynthesis and transpiration rate per unit area did not correlate with changes in stomatal index, whereas the stomatal conductance of mature leaves was highly correlated. It is known that increases in CO2 concentrations reduce stomatal conductance (Ainsworth & Rogers, 2007) whereas light positively influences stomatal conductance (Shimazaki et al., 2007). Therefore, this may be a critical parameter in the long-distance control of stomatal development. If this is the case it would predict that other factors that influence stomatal conductance, such as partial soil drying, humidity and plant hormones (see later; Sobeih et al., 2004; Bunce, 2006) should also influence the stomatal index of developing leaves.
The question of what the long-range signal that modulates stomatal development might be remains to be determined. A change in CO2 concentrations or light affects photosynthesis and therefore it is possible that a metabolic compound may regulate stomatal development. Sugar signalling has been shown to be important for plant development and control of the cell cycle (Riou-Khamlichi et al., 2000; Rolland et al., 2006). Reduced light intensity and increased CO2 have opposite effects on net photosynthesis and sugar content of mature leaves. However, both treatments reduce stomatal development in new leaves, arguing against a photosynthetic signal (Lake et al., 2002; Coupe et al., 2006), as does the lack of correlation between photosynthetic rate and stomatal index (Miyazawa et al., 2006). A number of plant hormones (or their precursors) undergo long-distance transport (Bradford & Yang, 1980; Wilkinson & Davies, 2002; Aloni et al., 2005; Wasternack et al., 2006; Vieten et al., 2007) and some have also been linked to stomatal development (Kieber et al., 1993; Serna & Fenoll, 1996; Saibo et al., 2003; Kazama et al., 2004). The constitutive triple response 1 (ctr1) mutant of Arabidopsis is defective in ethylene signalling and leaves of the ctr1 mutant, as well as those of wild-type plants exposed to ethylene, show increased stomatal density (Kieber et al., 1993). Plants grown in the presence of the ethylene precursor, 1-aminocyclopropane-1-carboxylic acid (ACC) also showed increased stomatal density as well as increased clustering. These effects can be blocked with the ethylene biosynthesis inhibitor aminovinylglycine or silver ions, which block ethylene perception (Serna & Fenoll, 1996). In cucumber hypocotyls, a transient exposure to ethylene results in an increase in cell division and stomatal development and it has been proposed that ethylene initially suppresses cell division but extends competence for cell division upon removal and may also act on cell fate determination (Kazama et al., 2004). Stomatal development is positively regulated by gibberellins (GA) and ACC in the Arabidopsis hypocotyl. Together, GA and ACC have an additive effect on stomatal development; auxin enhances the action of GA, but does not by itself effect stomatal development (Saibo et al., 2003). Leaf development is tightly linked to plant hormones (Li et al., 2007) and fluxes in hormone delivery to developing leaves may determine the fate of undifferentiated cells in the epidermis. Growth in the presence of elevated CO2 has been shown to cause significant increases in the auxin, cytokinin and GA concentrations in leaves, but a decrease in ABA (Teng et al., 2006). In addition, exposure to gentle wind, which would be predicted to increase transpiration rate, resulted in an increase in cytokinin in guard cells and the leaf vasculature (Aloni et al., 2005). Other factors may also determine the effect of hormones in the leaf. Transpiration rate can alter the pH of xylem sap, in turn influencing the leaf apoplastic pH (Jia & Davies, 2007). This is proposed to affect the compartment to which hormones such as ABA are delivered (Slovik et al., 1995). Hormone activity is also regulated by the rate of synthesis and turnover. Expression of the cytokinin oxidase (AtCKX6) gene increases in response to a low red/far-red ratio of light leading to cytokinin breakdown and reduced cell division activity in leaf primordia (Carabelli et al., 2007). Other members of the cytokinin oxidase family are expressed in the stomatal lineage (Werner et al., 2003) and may modulate the division competence of these cells. Regardless of the nature of the mobile signal the response may depend on a number of factors such as the rate of delivery, transport to the correct cellular compartment and the rate of breakdown (or synthesis).
Where do environmental induced signals integrate into the stomatal developmental pathway? The involvement of TMM and the ER-family in stomatal development raises the possibility that they may act as receptors of a systemic signal. LRR-RLKs have been shown to bind peptides and hormones (Montoya et al., 2002; Szekeres, 2003). TMM and the ER-family are expressed throughout the protoderm of developing leaf primordia becoming restricted to the stomatal lineage at a later stage (Nadeau & Sack, 2002; Shpak et al., 2005). Therefore, they could act in fate determination of undifferentiated cells. YDA or other components of the MAP kinase cascade could also be direct targets or alternatively SDD1 or EPF1 could influence the ability of neighbour cells to enter the stomatal lineage. As outlined earlier, SDD1 and SDD1-like are upregulated in developing leaves following shade treatment of mature leaves (Coupe et al., 2006), but the sdd1 mutant responds normally to light (Schlüter et al., 2003). Unless SDD1 and SDD1-like act redundantly, this would suggest SDD1 is not the integration point in response to light. It is apparent that further work is required to determine where the signal integrates specifically into the stomatal patterning and developmental pathways.
Stomata play a vital role in the ability of land plants to balance water loss with photosynthetic performance. While it has been known for several decades that stomatal pattern alters in response to the environment (Salisbury, 1927), only now are we beginning to dissect the underlying mechanisms and yet a number of issues are still to be resolved. Much of the work presented has focused on changes in light intensity and CO2 concentrations, which has great significance given the changes in global climate (Betts et al., 2007). However, the influence of other environmental variables needs to be evaluated as well as how plants integrate these separate signals to determine a developmental response. How plants perceive these environmental changes in the first instance and how the signal is transduced must be addressed and clarified.
Studies using Arabidopsis have improved our understanding of both stomatal physiology and development. We still need to determine how TMM and the ER family function to regulate the MAPK signalling cascade and how this influences the key transcriptional regulators. We have seen from transgenic studies that all protoderm cells in young leaf primordia are competent to acquire guard cell fate. Therefore, an environmental induced signal could affect a number of key events including the decision of a protoderm cell to become a MMC, the number of entry or spacing divisions giving rise to the stomatal lineage, or the number of amplifying divisions. We need to determine how environmental signals integrate into this developmental pathway, with the MAPK signalling pathway being an attractive possibility. It should be noted that our understanding of the components of stomatal development in other species is severely lacking and may not follow the Arabidopsis model.
It is possible that stomatal conductance is one of the factors that influence long-distance responses between mature and developing organs. It will be interesting to determine if a link exists between the physiological responses of stomata to the environment and the developmental pattern in new organs. A number of stomatal response mutants exist that could be used to evaluate this (Kinoshita et al., 2001; Cominelli et al., 2005; Hashimoto et al., 2006; Xie et al., 2006, Young et al., 2006). The current data is consistent with a long-range signalling mechanism operating in environmental responses. Research from other fields suggests that a range of molecules from hormones, proteins, peptides and RNA species could be candidates for a mobile signal. It remains to be seen if different environmental signals use the same signalling molecules and how they exert their effect.
We are grateful to Dr Shona Lindsay and Michelle Somerville for allowing us to use images of stomatal peels and to Prof. Alistair M. Hetherington for useful discussions.