Report

Stomatal Development and Pattern Controlled by a MAPKK Kinase

See allHide authors and affiliations

Science  04 Jun 2004:
Vol. 304, Issue 5676, pp. 1494-1497
DOI: 10.1126/science.1096014

Abstract

Stomata are epidermal structures that modulate gas exchange between a plant and its environment. During development, stomata are specified and positioned nonrandomly by the integration of asymmetric cell divisions and intercellular signaling. The Arabidopsis mitogen-activated protein kinase kinase kinase gene, YODA, acts as part of a molecular switch controlling cell identities in the epidermis. Null mutations in YODA lead to excess stomata, whereas constitutive activation of YODA eliminated stomata. Transcriptome analysis of seedlings with altered YODA activity was used to identify potential stomatal regulatory genes. A putative transcription factor from this set was shown to regulate the developmental behavior of stomatal precursors.

Pattern formation requires that the specification of individual cells be coordinated with mechanisms that create an ordered spatial arrangement of those cells. The Arabidopsis shoot epidermis consists of three major cell types: trichomes, pavement cells, and guard cells in a nonrandom arrangement. Guard cells flank a pore, the stoma, through which water vapor and carbon dioxide are exchanged between the plant and the environment. Mechanisms that control the abundance of these stomatal complexes, therefore, are important components of photosynthetic rate and water use efficiency. Guard cells are the terminal product of a lineage that arises postembryonically in the seedling epidermis. Asymmetric division of a protodermal cell generates a larger cell that is fated to become a pavement cell and a smaller cell, the meristemoid, that serves as a precursor to the guard cells. The meristemoid possesses stem-cell character and divides asymmetrically 1 to 3 times, then modifies its division potential and cell wall and differentiates into a guard mother cell (GMC) (1). The GMC makes a single symmetric division to form a pair of guard cells (2). In Arabidopsis, stomata are always separated by at least one pavement cell (1). To enforce this 1-cell spacing rule, signals from guard cells, GMCs, or meristemoids nonautonomously influence the asymmetric divisions of neighboring cells (3). Components of this signaling pathway may include the products of the TWO MANY MOUTHS (TMM) (4) and STOMATAL DENSITY AND DISTRIBUTION1 (SDD1) (5) genes that encode a predicted cell surface receptor and a serine protease, respectively. Mutations in either TMM or SDD1 lead to elevated stomatal density and frequent violations of the 1-cell spacing rule. However, neither mutation completely disrupts stomatal pattern (1), implying that other genes act in concert to establish the final pattern. Here we demonstrate that a mitogen-activated protein kinase (MAPK) kinase kinase (MAPKK) gene, YODA (YDA), plays a central role in both guard cell identity and pattern formation. YDA acts as a cell fate switch; loss of function mutations lead to too many cells adopting a guard cell fate, and constitutive activation of YDA produces plants that completely lack guard cells.

Asymmetric divisions are correlated with multiple cell fate decisions during Arabidopsis development, beginning with the asymmetric division of the zygote (6). Reasoning that the genes controlling asymmetric divisions in the stomatal lineage may be required earlier in development, we undertook a screen for altered stomatal patterning that also allowed us to recover seedling-lethal mutations (7). Among the 19 mutants recovered, six alleles of a recently described gene, YDA, were identified (8). All six yda alleles result in plants with extreme overproduction and clustering of guard cells in the epidermis of the cotyledons (Fig. 1, A and B) and hypocotyls (9). Previously, YDA was shown to regulate cell fate decisions in the zygote and the early embryo (8). yda seedlings rarely survive on soil, but those that do produce dwarfed plants with compact leaves filled with clusters of stomata (9). yda plants bolt, but produce defective flowers and are completely sterile (8).

Fig. 1.

Development of the epidermis in wild-type and yda plants. (A to H) Confocal images of stomatal development. Arrows indicate asymmetric divisions; asterisks indicate stomata. Scale bar, 20 μm. (A) Epidermis of wild-type, 7-dpg cotyledon. (B) Massive overproliferation of stomata (bracket) in 7-dpg yda epidermis. (C to H) Time course of cotyledon development. The wild type is on the left of each pair: (C) and (D), mature embryo (0 hpg); (E) and (F), 48 hpg; and (G) and (H), 72 hpg. Brackets indicate clustered stomata.

To determine the origin of the stomatal phenotypes, we recorded epidermal development from the mature embryo until 120 hours post-germination (hpg) in yda and wild-type seedlings (7). The epidermis of the mature yda embryo was indistinguishable from that of the wild type (Fig. 1, C and D), indicating that overproduction of stomata was not due to precocious development. The initial set of asymmetric divisions that produce meristemoids was roughly coincident in the wild type and yda (Fig. 1, E and F). However, by 72 hpg, it was apparent that more cells in the yda epidermis entered the stomatal pathway, and these stomata did not obey the 1-cell spacing rule (Fig. 1, G and H). Later, YDA continued to regulate divisions of meristemoids and their sisters. Meristemoids normally produce both guard cells and pavement cells. However, in yda plants, both progeny of a meristemoid often became guard cells (Fig. 1, B and H). Stomatal clusters in yda plants consist of paired guard cells that develop the radially oriented microtubule arrays typical of mature guard cells (9). This contrasts with the clusters of unpaired guard cells in four lips mutants that arise from repeated divisions of the GMCs (10), suggesting yda mutations affect the asymmetric divisions that take place before GMCs are specified.

YDA encodes a putative MAPKK kinase with highest similarity to the sterile 11/MAP/ERK kinase (STE11/MEKK) family of kinases and is expressed at low levels throughout the plant (fig. S2) (8). The yda-Y295 allele used for phenotypic analysis is predicted to be a protein null allele (7). The N terminus of many MAPKK kinases serves as a negative regulatory domain, and its removal results in constitutive activation of the kinase (8, 11). In Arabidopsis embryos, expression of N-terminal deletions of YDA (ΔN-YDA) with the native promoter resulted in phenotypes opposite to loss-of-function phenotypes (8). We analyzed the effect of these ΔN-YDA gain-of-function alleles on stomatal development. Guard cell formation was completely abolished in plants homozygous for ΔN-YDA (Fig. 2 and fig. S1). This indicates that YDA activity must be down-regulated to allow epidermal cells to enter the stomatal lineage and suggests that YDA is part of a cell fate switch.

Fig. 2.

Conversion of cell fate in epidermis of yda and ΔN-YDA seedlings. Wild-type [(B), (E), (H), and (K)]; yda [(A), (D), (G), and (J)], and ΔN-YDA [(C), (F), (I), and (L)] seedlings are shown. (A to C) Whole 7-dpg seedlings. (D to F) Confocal images of cotyledons showing the distribution of pavement and guard cells. Cell peripheries were visualized with propidium iodide (PI) or Q8::GFP (28). Autofluorescence of chloroplasts is visible as small red disks. (G to I) TMM::GFP expression (green) marks meristemoids and guard mother cells; PI (red) stains cell peripheries. (J to L) N7::GFP (28) marker indicates nuclear size and ploidy. In (K), the arrowhead points to paired diploid guard cell nuclei and the arrow to a polyploid pavement cell nucleus. Scale bar, (A) to (C), 1 cm; (D) to (L), 50 μm.

To confirm that ΔN-YDA expression leads to the choice of an alternate epidermal cell fate and does not merely interfere with progression along the guard cell differentiation program, we assayed several markers of cell identity. ΔN-YDA plants did not accumulate meristemoids or GMCs (Fig. 2 and fig. S1), nor did they express a fusion of the TMM promoter and protein to green fluorescent protein (TMM::GFP) that is normally expressed in the precursors of mature stomata (Fig. 2) (4). Instead, ΔN-YDA epidermal cells were highly crenulated (Fig. 2, F and I) and developed the nuclear morphology typical of mature polyploid pavement cells (Fig. 2L) (12).

The mutant phenotypes and identities of TMM, SDD1, and YDA suggest that they could act in a common signaling pathway, with SDD1 cleaving a ligand that is perceived by TMM and a coreceptor kinase (13). A signal transduced through this receptor pair could trigger the phosphorylation and activation of the YDA MAPKK kinase and its downstream signaling cascade. Because ΔN-YDA plants have an opposite phenotype to sdd1 and tmm plants, we tested this model genetically. Transgenic wild-type plants homozygous for ΔN-YDA never produced stomata, but plants harboring a single copy of ΔN-YDA made similar numbers of stomata as in the wild type (Fig. 3, A and B, and fig. S1). ΔN-YDA/+ suppressed the phenotypes of sdd1 and tmm. We never observed clustered stomata in cotyledons or leaves of tmm-1;ΔN-YDA/+ or sdd-1;ΔN-YDA/+ plants, and the stomatal index (guard cells per total epidermal cells) of these plants was indistinguishable from that of the wild-type (Fig. 3, A to F, and fig. S1B). The genetic analysis is consistent with YDA acting downstream of SDD1 and TMM or with YDA in an independent pathway that controls cell fate in the same uncommitted cells as TMM and SDD1. Because lineage analysis suggested that YDA was required before the creation of GMCs, we also performed genetic tests between ΔN-YDA and flp. In contrast to the results with sdd1 and tmm, ΔN-YDA/+ did not suppress the flp mutant phenotype (Fig. 3, G and H); again, this is consistent with YDA acting before the GMC is specified.

Fig. 3.

Interactions between YDA and other stomatal pattern genes in the abaxial epidermis of 7-dpg cotyledons: (A) wild type, (B) ΔN-YDA/+, (C) sdd1-1, (D) sdd1-1;ΔN-YDA/+, (E) tmm-1, (F) tmm-1;ΔN-YDA/+, (G) flp-1, and (H) flp-1;ΔN-YDA/+. Brackets indicate clusters. All images are the same magnification. Scale bar, 50 μm.

ΔN-YDA plants, which lack stomata, can survive for several weeks on sucrose media and occasionally produce small leaves with trichomes (fig. S2E), although they do not develop to reproductive maturity. The ability to generate plants completely lacking mature guard cells and guard-cell precursors afforded a unique opportunity to assess the transcriptional profile of this cell lineage and to identify new genes involved in stomatal development. As a complementary approach to forward genetic screens, we compared the gene expression profile of plants that completely lack cells in the stomatal pathway (ΔN-YDA) to plants with an epidermis that consists almost entirely of guard cells (yda). We probed Affymetrix full-genome microarrays with RNA derived from wild-type, yda, and ΔN-YDA seedlings that were 7 days post-germination (dpg). About 1800 genes exhibited a statistically significant change in expression between each mutant and its corresponding wild type (14, 15). The expression of 220 of these transcripts changed in opposite directions between the yda and ΔN-YDA backgrounds (Fig. 4A).

Fig. 4.

Microarray analysis of yda and ΔN-YDA seedlings. (A) 220 transcripts differentially regulated between ecotype C24 (replicates in lanes 1 to 3), ecotype Ler (4 to 6), ΔN-YDA (7 to 10), and yda (11 to 14). Fold expression changes are colorcoded in the bar beside the cluster. (B) Cluster of “stomatal-enriched” transcripts mirroring HIC (white) and SDD1 (green) expression; the FAMA profile is shown in blue. Genotypes are indicated above the graph. (C) Gene model of FAMA with exons indicated by boxes, the bHLH domain shaded, and the location of the SALK_100073 T-DNA indicated by the triangle. (D) An 8 week-old fama plant. (E and F) Cotyledon epidermis of 7-dpg (E) wild-type and (F) fama1-1 seedlings. In fama1-1, brackets demarcate the clusters of up to 15 very small cells that form in place of paired guard cells. Scale bar, 20 μm. (G) Pathway toward guard cell development, with the time period of action of the stomatal regulators YDA, TMM, SDD1, FLP, and FAMA indicated by bars below.

We predicted that genes required to establish guard cell identity or to enforce the 1-cell spacing rule would be up-regulated in yda and down-regulated in ΔN-YDA, coincident with the over-abundance or absence of the guard cell lineage, respectively, in these mutant backgrounds. Three cloned stomatal pattern genes, TMM, SDD1, and High Carbon Dioxide (HIC), are expressed in stomatal precursors or mature stomata (4, 16, 17) and served as controls. All three genes behaved as predicted [down-regulated in ΔN-YDA and up-regulated in yda (Fig. 4B and table S3)]. This regulation was independently confirmed by the expression of the translational TMM::GFP reporter (4) (Fig. 2, G to I) and of a transcriptional TMMpro::GFP reporter (4, 9) in the yda and ΔN-YDA backgrounds. Other epidermally expressed genes, such as those involved in the patterning of the root epidermis, did not show this transcriptional profile (fig. S5). Downstream targets of YODA are likely to be found in the 111 genes with the complementary transcriptional profile: up-regulated in ΔN-YDA and down-regulated in yda (table S4). A subset of these genes may also be involved in stomatal identity or in establishing the 1-cell spacing rule.

The transcriptional profiles of 109 genes clustered with HIC, SDD1, and TMM were enriched in a high-stomata background. Of these genes, 24% corresponded to either signaling molecules (19 of 109) or transcription factors (7 of 109), both good candidates for molecules involved in the specification of guard cells (fig. S3B). In addition to genes required for guard cell specification, we predicted that transcripts for genes required for guard cell function would appear in this transcriptional cluster. The cell walls of differentiated guard cells are extensively reinforced, and correspondingly, several transcripts (11%, or 12 of 109) were annotated as cell wall modification proteins (fig. S3B).

Sequence-indexed transferred (T-DNA) insertion lines are currently available for ∼70% of the transcripts identified in the microarray experiments (18). To validate this genomic approach to identifying new stomatal development genes, we characterized the phenotype of 160 T-DNA insertions that represent 82 of 220 differentially expressed genes. Ten loci exhibited stomatal or epidermal phenotypes in at least two independent insertion lines (9). A marked phenotype was observed in lines homozygous for insertions in a putative transcription factor, At3g24140 (FAMA) (Fig. 4, C and D). FAMA exhibited a transcriptional profile similar to that of SDD1 and HIC1 (Fig. 4B). Existing microarray data revealed that FAMA expression is consistent with a role in stomatal pattern: highest in leaves (Affymetrix values 315 ± 13), lower in flowers (212 ± 17), and absent in roots (44 ± 33) (15, 19). The cotyledon epidermis of fama1-1 (SALK_100073) (18) had no recognizable guard cells. Instead, chains of small cells that rarely make obvious stomatal pores but that, like guard cells (and unlike pavement cells), often contain mature chloroplasts were intercalated among pavement cells (Fig. 4, E and F). These clusters of incompletely differentiated cells are reminiscent of, but larger than, the clusters of mature guard cells found in flp mutants (10). FLP has been proposed to limit GMC division competence cell-autonomously (1). Based on the up-regulation of FAMA in plants that overproduce stomata, it is likely that FAMA plays a similar cell-autonomous role in regulating GMC behavior.

We have shown that YDA regulates a fundamental cell fate decision in the Arabidopsis epidermis. YDA activity must be down-regulated to allow cells to enter the stomatal lineage, and its normal activity is required to maintain the balance of proliferation versus differentiation (into GMCs) in the meristemoids (Fig. 4G). In animal cell fate decisions, MAPK signaling cascades often act as molecular switches by converting small changes at the cell periphery into multiple downstream responses (20). Stomatal cell fate may also require such an amplification mechanism because only some protodermal cells are initially chosen to enter the stomatal lineage and their recruitment is not predicted by any obvious character of those cells. The continued requirement for YDA in meristemoids may be tied to a universally conserved role of MAPK signaling in controlling cell cycle progression (21) or may reflect the utility of MAPK signaling for responses to changing environmental stimuli, as best characterized in the Saccharomyces cerevisiae osmotic response (22) and in Arabidopsis hormone (23, 24) and pathogen (25) responses.

Plants respond to global climate factors such as CO2 and water and light availability by adjusting their stomatal density, stomatal index, and stomatal distribution (26). These developmental responses have complex effects on plant fitness and water use efficiency (27). We have shown the utility of the YDA system to identify new stomatal cell fate regulators. Additionally, the ability to use YDA to manipulate stomatal density in a model plant species will facilitate an analysis of the mechanisms by which stomatal density is controlled and the physiological relevance of the large variation in density found in nature.

Supporting Online Material

www.sciencemag.org/cgi/content/full/304/5676/1494/DC1

Materials and Methods

Figs. S1 to S5

References and Notes

View Abstract

Navigate This Article