Transcription co‐activator Arabidopsis ANGUSTIFOLIA3 (AN3) regulates water‐use efficiency and drought tolerance by modulating stomatal density and improving root architecture by the transrepression of YODA (YDA)

Summary

One goal of modern agriculture is the improvement of plant drought tolerance and water‐use efficiency (WUE). Although stomatal density has been linked to WUE, the causal molecular mechanisms and engineered alternations of this relationship are not yet fully understood. Moreover, YODA (YDA), which is a MAPKK kinase gene, negatively regulates stomatal development. BR‐INSENSITIVE 2 interacts with phosphorylates and inhibits YDA. However, whether YDA is modulated in the transcriptional level is still unclear. Plants lacking ANGUSTIFOLIA3 (AN3) activity have high drought stress tolerance because of low stomatal densities and improved root architecture. Such plants also exhibit enhanced WUE through declining transpiration without a demonstrable reduction in biomass accumulation. AN3 negatively regulated YDA expression at the transcriptional level by target‐gene analysis. Chromatin immunoprecipitation analysis indicated that AN3 was associated with a region of the YDA promoter in vivo. YDA mutation significantly decreased the stomatal density and root length of an3 mutant, thus proving the participation of YDA in an3 drought tolerance and WUE enhancement. These components form an AN3–YDA complex, which allows the integration of water deficit stress signalling into the production or spacing of stomata and cell proliferation, thus leading to drought tolerance and enhanced WUE.

Introduction

Drought is a major abiotic stress that greatly affects plant growth and limits agricultural productivity. Crop drought tolerance is due to many factors. Among these factors, stomatal density is the leading determinants of plant drought tolerance because crop water loss directly involves stomata (Yoo et al., 2010; Yu et al., 2008).

The genetic pathway of stomatal development has been well known. Different EPIDERMAL PATTERNING FACTOR‐LIKE (EPFL) family members mediate different ERf roles, with CHAL‐related family members dedicated to regulate growth processes and EPF1/2‐related members specialized to modulate stomata (Abrash et al., 2011). EPF‐LIKE9 (EPFL9/Stomagen) can facilitate stomatal differentiation from internal tissues (Torii, 2012). Very recently, biochemical evidences have indicated that the ERECTA family is a primary receptor for EPF ligands and TMM regulates the activity of ERECTA‐family proteins (Lee et al., 2012). Both the genetic and biochemical findings reveal a crucial function of TMM in preventing stomatal precursors from auto‐inhibiting themselves from signals they produce (Torii, 2012). YODA (YDA), a MAPKK kinase gene, acts genetically downstream of TMM (Bergmann et al., 2004). The bHLH SPEECHLESS (SPCH) is phosphorylated by activated MPK3/6 by YDA, resulting in SPCH inactivation and suppression of a signalling pathway that negatively regulates the basal pathway of stomatal lineage, which is necessary to achieve a balance between pavement and guard cells in the leaf epidermis (Bergmann et al., 2004; Lampard et al., 2008; Wang et al., 2007). SDD1 encoding a subtilisin‐like Ser protease negatively regulates stomatal development (Von Groll et al., 2002). Thus, cell–cell signals from meristemoids through secreted ligands, membrane receptors and mitogen‐activated kinase cascades are very important for proper stomatal patterning and spacing (Bergmann et al., 2004; Geisler et al., 2000; Hara et al., 2007; Hunt and Gray, 2009; Shpak et al., 2005; Sugano et al., 2010).

While stomata development has been well understood, only a few reports focused on studying a relationship between stomata density and water‐use efficiency (WUE). The developmental programme of determining stomatal density is enough plastic for enabling plants to modulate transpiration in response to changing environments (Casson and Gray, 2008). GT2‐LIKE1 (GTL1) regulates water‐use efficiency via modulating stomatal density by transrepression of SDD1 (Hara et al., 2007; Hunt and Gray, 2009; Sugano et al., 2010). Overexpression of STOMAGEN, a positive regulator of stomatal density, can produce plants with a two‐ to threefold greater stomatal density than the wild type, and photosynthetic carbon assimilation (A) in these plants was enhanced via 30% because of greater CO2 diffusion into the leaf blades, but not changes in photosynthetic carboxylation capacity (Tanaka et al., 2013). The resultant 30% enhancement in A was at the expense of transpiration, which was double that observed in the wild‐type plants, leading to a 50% decrease in WUE. ER mutations lead to lower WUE, which was associated with a number of measurable phenotypes, including enhanced stomatal density and reduced photosynthetic capacity and mesophyll development (Masle et al., 2005).While negatively acting mitogen‐activated protein kinase (MAPK) signalling controls stomatal initiation in Arabidopsis, it is largely unknown whether and how MAPK signalling is involved in regulation of WUE.

Here, ANGUSTIFOLIA3 (AN3) encodes a homologue of the human transcription co‐activator and functions as a focal regulator of water stress tolerance and WUE by a mechanism that involves transcriptional repression of YDA, a MAPKK kinase gene, and regulation of stomatal density and transpiration. Thus, the coupling of AN3 (as a developmental and stress integrator) signalling to MAPK activity triggers the regulation of WUE and drought stress. Moreover, AN3 expression is down‐regulated through dehydration, establishing a potential paradigm for how the environment affects stomatal development to decline transpiration under low water availability conditions. Furthermore, an3 presented improved root architecture, with longer primary roots and more numerous lateral roots, which further enhances an3 tolerance to drought stress.

Results

Arabidopsis an3 plants display improved drought tolerance

We noticed accidently that when wild‐type (Col‐0) plants turned dry on the leaf surface, the an3‐4 mutant still showed green and extending leaf blades. Therefore, we speculated that the an3‐4 mutant was highly resistant to drought stress. We then repeated this experiment. The an3‐1 mutant [stock no. CS241 at the Arabidopsis Biological Resource Center (ABRC); the entire coding region was deleted] and the an3‐4 mutant (has different deletions that span the AN3 locus compared with an3‐1 (Horiguchi et al., 2005) were used. Both an3‐1 and an3‐4 belong to the homozygous mutant ethylmethylsulfone (EMS) (Horiguchi et al., 2005).

an3 and wild‐type plants were grown for 15 days in soil and the subsequently subjected to water withholding for 19 days (Figure 1a). Wild‐type plants showed leaf rolling and their leaves became purple, whereas an3 plants remained turgid with their leaves remaining green (Figures 1a, d). These results indicated the better survival capability of an3 plants under low relative soil water content (SWC) than the wild‐type plants (Figure 1b). For example, at 13% SWC, 100% and ~10% of the an3 mutants and wild‐type plants survived, respectively (Figures 1b, c). The enhanced water deficit survival of an3 plants was always closely associated with their capacity to maintain higher leaf relative water content (RWC) than wild‐type plants at ~13% SWC (Figure 1c).

Figure 1

an3 plants are less sensitive to drought stress. (a) Drought tolerance of the 15‐day‐old an3 and wild‐type plants under identical water deprivation and rewatering periods. Samples were photographed at different time points: 0 days, 5 days to 19 days, and 19 days after water deprivation at 3 days after rewatering. Relative SWC indicates the SWC over the SWC at day 0 of withholding water and is the average of three pots. (b) Survival rate in an3 and wild‐type plants. Survival rates of the plants were determined in (a) after rewatering, and the values are the mean ± SD of three independent experiments (***P < 0.001). (c) an3 and 20‐day‐old plants were grown under normal condition and exposed to water deficit stress by withholding water. Leaf RWC (mean ± SD of three independent experiments; **P < 0.01) and relative SWC were determined. (d) Leaf blades of an3‐4 and wild‐type (Col‐0) grown on soil with withholding water for 12 days. Magnifications of all images are the same. (e) Anthocyanin content in leaves of an3‐4 and wild‐type plants with or without drought treatment for 12 days in (d). Error bars represent SD for three independent experiments. FW, fresh weight.

Another indicator of stress sensitivity in plants is the accumulation of anthocyanin in the leaf blades (Li et al., 2008). The anthocyanin levels in wild‐type seedlings after withholding water for 10 days were 12.8 µg g⁻¹ fresh weight, which was approximately four times as much as that of the an3. This result indicated that wild‐type plants were more sensitive to drought stress (Figures 1d, e). Wild‐type and an3 plants were grown in the same pot (see Figure S1A–C) to further characterize drought tolerance. After 10 days without watering, the an3 mutant plants showed only mild drought stress symptoms, whereas wild‐type plants exhibited severe drought symptoms. This test unambiguously confirmed the improved drought tolerance of the an3 mutant.

Using detached shoots as materials, COR15A, which is a marker gene of dehydration response (Baker et al., 1994), was up‐regulated and AN3 was down‐regulated under dehydration stress (see Figure S2 online). This finding was consistent with results from the publicly available Arabidopsis eFP Browser Microarray Database (http://www.bar.utoronto.ca/efp/cgi-bin/efpWeb.cgi), where AN3 is down‐regulated by dehydration treatment. AN3 is expressed when sufficient water is available but is down‐regulated by water deficiency.

an3 mutations improve WUE by reducing transpiration

To better explore the mechanism of the high maintenance of leaf water status in an3 mutants under low soil moisture conditions, daily water loss was measured. Our findings showed that daily water loss in an3 mutants was significantly reduced, thus probably increasing the capacity of an3 seedlings to maintain high leaf RWC (Figure 1c) and high tolerance against water deficit stress (Figures 1a, b). The daily water loss decrease (see Figure S3A) was neither correlated with the decrease in shoot dry weight (see Figure S3B) nor with the total leaf area (see Figure S4). All of the data above suggest that decreased stomatal conductance, which cause decreased water loss, did not lead to a concomitant reduction in biomass accumulation (see Figure S3B). Consequently, an3 mutants presented high integrated WUE (biomass/water use; see Figure S3C), thus indicating that AN3 was a negative regulator of WUE.

Reduced daily water loss in an3 mutants was likely a result of low transpiration. To demonstrate this speculation, the gas exchange (water and CO₂) of the fully expanded leaf blades of an3 mutants was assayed using an infrared gas analyser. Leaf blade transpiration was ~44% lower in an3 mutants than in wild‐type plants at saturating light levels (see Figure S3D). Stomatal conductance was lower in an3 mutants (see Figure S3F), thus indicating that the decreased transpiration was due to decreased water loss. Net CO₂ assimilation rates in an3 mutants and wild‐type plants were insignificantly different (see Figure S3H). Consequently, an3 mutant plants revealed enhanced instantaneous WUE (CO₂ assimilation/transpiration) (see Figure S3E), which led to reduced transpiration (see Figure S3D). Vapour pressure deficit (VPD) was also analysed. The results showed that VPD was insignificantly different between all measurements (see Figure S3I), thus indicating that the low transpiration and stomatal conductance in an3 mutant plants were not caused by different VPDs. Internal CO₂ concentration (ci) in an3 leaf blades was lower than that of wild‐type plants (see Figure S3G). This result could be attributed to the decrease in CO₂ flux from the air to the substomatal cavity because of the low stomatal conductance in an3 mutants.

Stomatal densities were reduced on an3 leaf epidermal layers

Stomatal conductance was considerably lower in an3 mutant plants than in wild‐type plants. This result may be caused by the altered stomatal aperture, density and size. However, the stomatal aperture did not differ in the leaf abaxial surfaces among an3 mutants, wild‐type plants under abscisic acid (ABA) treatment and well‐watered plants (see Figure S5B). Moreover, the seed germination of an3 mutants and wild‐type plants responded to ABA similarly (see Figure S5A), thus implying that AN3 was not involved in regulating ABA responsiveness and that an3 drought tolerance was not caused by ABA‐mediated stomatal closure.

Fully expanded leaves in an3 plants had fewer stomata than those in wild‐type (Figure 2a–c, e). an3‐1 and an3‐4 plants were ~49% and ~47% lower leaf abaxial stomatal density than wild‐type plants, respectively. This reduction was mainly caused by the large pavement cells (Horiguchi et al., 2005), thus reducing cell density (Figure 2e). This result was consistent with that leaf blade transpiration was ~44% lower in an3 mutants than in wild‐type plants at saturating light levels (see Figure S3D). Furthermore, on the leaf abaxial surface, stomatal precursor cells, such as meristemoids or guard mother cells, were detected in an3 leaf blades but not in wild‐type plants (Figure 2a–c, f). However, on the leaf adaxial epidermis, stomatal precursor cells were not observed (data not shown). AN3 strongly expressed on the stomata of leaf abaxial epidermis (Figure 2d) but not on the stomata of the leaf adaxial epidermis (data not shown). This phenomenon led to reduced stomatal indexes (stomata number per total number of epidermal cells; Figure 2g) and stomatal densities (Figure 2e) in an3 plants. Large an3 pavement cells are due to compensation, wherein a decrease in cell number causes an increase in mature cell size (Ferjani et al., 2007). This compensation‐induced cell enlargement is mostly independent from endoreduplication, and the exact mechanism is unknown (Ferjani et al., 2007). Although stomatal density was reduced, stomatal size increased in an3 plants (Figures 2a–c, h). The inverse relationship between stomatal size and density has been well described previously (Franks and Beerling, 2009; Yu et al., 2008). The results indicated that AN3 regulated the development of both stomata and pavement cells and affected stomatal density. Thus, AN3 ultimately modulated plant tolerance to drought stress.

Figure 2

an3 plants with reduced stomatal density. (a–c) Stomata of the 5th or 6th leaf abaxial epidermis in the 30‐day old an3 and wild‐type plants. Arrows indicate stomatal precursor cells. Bars = 50 μm. (d) GUS activity in the abaxial epidermis of the third rosette leaf from 8‐day old seedlings harbouring AN3pro‐GUS was analysed by histochemical staining. Bars = 50 μm. (e–h) Comparisons of the stomatal density and cell density (e, **P < 0.01); Stomatal precursor cells (f, **P < 0.01); stomatal index (number of stomata divided by the total number of epidermal cells) (g, *P < 0.05); and stomatal size (h, *P < 0.05 for stomatal length) in the wild‐type and an3 plants, respectively. Values are the mean ± SD from at least five independent seedlings.

an3 mutant plants display altered root architecture with long primary roots and more lateral roots

Kim and Kende (2004) strongly expressed AN3 in the roots of microarray data using the Affymetrix Arabidopsis ATH1 GeneChip. The primary roots of 6‐ and 15‐day‐old an3 plants elongated faster on agar medium than those of corresponding wild‐type plants (Figure 3a, b). Primary root elongation in an3 plants increased ~40% compared with that in wild‐type plants (Figure 3a, b). an3 plants also had more lateral roots than wild‐type plants; this phenotype was more pronounced at 15 day after germination (Figure 3c). To test whether root enhancement was maintained during subsequent plant development, plants were cultivated in the soil. Figure 3a exhibited that an3‐4 plants also possessed improved root architecture than wild‐type plants. The fresh biomass of the root systems of an3‐4 was ~37% higher than that of wild‐type plants (Figure 3d) 40 days after germination (DAG). The root‐to‐shoot biomass ratios were enhanced up to approximately 30% in an3‐4 plants (Figure 3e). An improving rooting system can increase the extraction of soil moisture derived from greater depths (White and Castillo, 1992). Thus, the drought tolerance of an3 plants might be partially attributed to their improved root systems. Longer an3 primary roots were attributable to increased root cell elongation in the elongation zone (EZ; Figure 3f).

Figure 3

an3 plants have improved root architecture. (a) Primary roots of 6‐ and 15‐day old an3 and wild‐type plants (top panel) on MS medium supplemented with 1% sucrose, and 40‐day old an3 and wild‐type root systems (bottom panel). Bar = 1.0 cm. (b) Primary root elongation of the wild‐type and an3 plants grown on MS medium supplemented with 1% sucrose. Length of the primary roots was measured at the indicated time points. The values are mean ± SD of three independent experiments (n > 20). (c) Lateral roots in the wild‐type and an3 plants grown on MS medium supplemented with 1% sucrose. The number of lateral roots was counted at the indicated time points. The values are mean ± SD of three independent experiments (n > 20). (d) Root biomass in the 40‐day‐old an3 and wild‐type plants. The values are mean ± SD of three independent experiments (n > 10; *P < 0.05). (e) Root‐to‐shoot ratio in the 40‐day‐old an3 and wild‐type plants. The values are mean ± SD of three independent experiments (n > 10; *P < 0.05). (f) Root cell length in the 10‐day‐old wild‐type and an3 roots. The values are mean ± SD from at least five independent seedlings (n > 50; ***P < 0.001).

AN3 negatively modulates YDA at the transcriptional level

To detect downstream target genes, the expression levels of the genes involved in negatively regulating stomatal development were analysed by quantitative RT‐PCR. Transcripts of YDA and MPK6 were more abundant in an3‐4 than in wild‐type plants and 35Spro‐AN3/an3‐4 (35Spro‐AN3 construct in the an3‐4 background) (Figure 4a). These findings suggested that an3 plants with fewer stomata had high levels of guard cell gene expression, similar to gtl1 plants (Yoo et al., 2010). However, the expression of TMM and SDD1 was insignificantly different between an3‐4, wild‐type and 35Spro‐AN3/an3‐4 plants (Figure 4a). Moreover, AN3 transcript abundances were similar in the wild‐type, yda, sdd1, tmm and mpk6 plants (Figure 4b). Therefore, AN3 plays a stomatal development determinant role via negatively modulating YDA and MPK6 expression.

Figure 4

AN3 transrepresses YDA. (a) Expression of MPK6, YDA, TMM and SDD1 in 2‐week‐old Col‐0, an3‐4 and 35Spro‐AN3/an3‐4 plants. RNA was extracted from at least 10 independent seedlings. (b) Expression of AN3 in 2‐week‐old Col‐0, yda‐1, sdd1, tmm and mpk6 plants. RNA was extracted from at least 10 independent seedlings. Quantifications were normalized to the expression of UBQ5. Error bars represent SD for three independent experiments. (c–e) In the wild‐type background, the strong expression of AN3pro‐GUS on the root MZ but not on EZ and DZ (n > 30). (f–h) In the wild‐type background, the strong expression of YDApro‐GUS on the root EZ and DZ but not in MZ (n > 30). (i) In an3‐4 background, the strong expression of YDApro‐GUS on the root MZ (n > 20). Plants were from the same plate. Magnifications of all images are the same. (j) Intensity of GUS coloration. Quantified using Adobe Photoshop CS (Adobe Systems Inc., San Jose, CA) software, as described by Wang et al. (2011). The YDApro‐GUS expression intensity in wild‐type background was set to 1.0. All transgenic plants with GUS were subjected to GUS staining for 8 h.

To better understand the AN3 regulation of YDA expression, we observed their GUS expression in roots. In the wild‐type background, AN3 expression was detected in the root meristematic zone (MZ) but was expressed weakly in the EZ and differentiation zone (DZ; Figure 4c–e, j). In the wild‐type background, YDA was expressed strongly in EZ and DZ but was expressed weakly in MZ (Figure 4f–h, j; Bergmann et al., 2004). However, YDA was expressed strongly in the root MZ in the an3‐4 mutants than in the wild‐type background (Figure 4i, j). Given that YDA was strongly expressed in differentiated tissues and AN3 was strongly expressed in active dividing tissues. AN3 was strongly expressed in root MZ, whereas YDA was strongly expressed in root EZ and DZ. The above data indicated that AN3 negatively modulated YDA at the transcriptional level.

AN3 protein is a major factor for repressing the YDA promoter

The study of Vercruyssen et al. (2014) (Genome‐Wide Determination of AN3 Binding Sites) of motifs using RSAT peak motifs (Thomas‐Chollier et al., 2012) resulted in the identification of two significantly enriched motifs with the following peak sequences: the tgaCACGTGgca motif containing the core G‐box sequence (CACGTG) and the TCTC motif (TCTCTCTC) (Vercruyssen et al., 2014), a putative element of Arabidopsis core promoters (Yamamoto et al., 2009). We found that two TCTC motifs (TCTCTCTCTCTCTCTC and TCTCTCTCTC) were present between −800 and −900 in the YDA promoter (Figure 5a). A few potential amplicons in YDA promoter were used for chromatin immunoprecipitation ChIP analyses. To test this possibility, we used transgenic plants that express the 35Spro‐AN3‐3XGFP construct in an3‐4 mutant for ChIP analysis. Regions Y2 (−390 to −680) and Y3 (−690 to −980) primers resulted in the greatest amount of PCR product but not in region Y1 (−3 to −311) (Figure 5b). To confirm this possibility, we performed SQ‐RT‐PCR. Similar results were obtained (Figure 5c). AN3 was associated with the YDA promoter in vivo, which is required for suppressing YDA expression. At 8 days post‐germination (dpg), wild‐type plants expressed YDA pro‐GFP containing two TCTC motifs (YDA pro‐GFP/WT; 9 of 10), but 35Spro‐AN3 plants did not (YDA pro‐GFP/35Spro‐AN3; 0 of 10; Figure 5d). Similar to wild‐type plants, at 8 dpg, 35Spro‐AN3 plants expressed YDA pro‐GFP without two TCTC motifs (8 of 10) (Data not shown), thus indicating that the TCTC motif in the YDA promoter is required for AN3 to repress the YDA promoter. Thus, AN3 specially associates with the YDA promoter by two TCTC motifs and AN3‐YDA may form a signal cascade for the regulation of cell proliferation. These findings indicated that AN3 was a major factor of repressing the YDA promoter.

Figure 5

AN3 transrepresses YDA. (a) Schematic of ChIP analysis on the YDA promoter loci and three amplicons. (b, c) ChIP analysis; Enrichment of particular YDA chromatin regions with anti‐GFP antibody or anti‐HA antibody (As a negative control) in 35Spro‐AN3‐3XGFP transgenic plants as detected by Q‐PCR analysis (b) and SQ‐PCR analysis (c). Quantifications were normalized to the expression of UBQ5. Error bars represent SD for three independent experiments. (d) YDApro‐ GFP in WT(Col‐0) (top panel) and 35Spro‐AN3 (bottom panel) 9cotyledons. Magnifications are the same. White arrows point to GFP‐positive nuclei. Bars = 50 μm.

yda mutants have shortened primary roots and enhanced stomatal density because of suppressed cell elongation

The 10‐day‐old roots elongated slower in yda than in the corresponding Ler plants (Figure 6a, b). This result agrees with the observation of Lukowitz et al. (2004). Similarly, yda lateral root densities decreased (Figure 6a, c). The shortened yda primary roots were attributable to the suppressed root cell elongation in EZ (Figure 6e). yda leaf blades had high stomatal densities because of the reduced cell size caused by the suppressed cell extension (Figures 6d, f; Lukowitz et al., 2004). Moreover, YDA was a negative regulator of stomatal development and the loss of YDA function presented enhanced stomata index, thus leading to high stomatal density (Bergmann et al., 2004). Thus, high stomatal density in yda leaf blades was caused by both reduced pavement cell size and enhanced stomata index. Therefore, YDA participates in the an3 drought tolerance caused by altered root systems and stomatal density.

Figure 6

yda root architecture is altered. (a) Primary roots in 15‐day‐old yda and Ler plants on MS medium supplemented with 1% sucrose. Bar = 1.0 cm. (b) Primary root elongation in the Ler and yda plants grown on MS medium supplemented with 1% sucrose. Values are mean ± SD (n > 20; **P < 0.01). (c) Lateral roots in the yda and Ler plants grown on MS medium supplemented with 1% sucrose. Values are mean ± SD (n > 20; **P < 0.01). (d) Stomata of the 5th or 6th leaf abaxial epidermis in the 30‐day old yda‐1 and Ler plants. Bars = 20 μm. (e) Root cell length in yda and Ler plants. Roots of 10‐day old Ler and yda plants were photographed, and the length of the root cells in the EZ was measured using Image J Microsoft. Values are mean ± SD of at least five independent roots. (n > 50; **P < 0.01). (f) The cell size of the 5th or 6th leaf abaxial epidermis in the 30‐day old yda‐1 and Ler plants (n > 60; *P < 0.05).

AN3 generally acts upstream of YDA in stomatal and root development regulation

To test whether AN3 acted genetically upstream of YDA in root development, double mutant analysis was performed by combining an3‐4 with yda‐1, which produced long and short primary roots, respectively. Given that an3‐4 (Col) and yda‐1 (Ler) have different genetic backgrounds, genetic analysis in a mixed Col and Ler background was performed after crossing an3‐4 to yda‐1. The an3‐4yda‐1 was confirmed by RT‐PCR (Figure S6). The yda‐1 plants significantly diminished the long root phenotype of an3‐4 (Figure 7a, b). The results indicated that AN3 acted upstream of YDA in root development.

Figure 7

AN3 acts genetically upstream of YDA in regulating stomata and root development. (a, b) Genetic interactions of AN3 with YDA. All genotypes are in a mixed Col/Ler background for 15 day old (n > 15; **P < 0.01). Bar = 1.0 cm for (a). (c) Confocal images of the cotyledon epidermis of E1728 an3‐4 and E1728 an3‐4 ∆N‐YDA (n > 10). Bar = 10 µm. (d) Different stomatal index in indicated leaf (number of stomata divided by the total number of epidermal cells, n > 5; **P < 0.01).

YDA protein eliminated the N‐terminal fragment (∆DN–YDA), and the cotyledon epidermis of transgenic plants expressing ∆DN–YDA produced no stomata (Bergmann et al., 2004). We used a construct expressing ∆DN–YDA fused to a chemical‐inducible promoter and then transformed it into an3‐4 plants. The an3‐4 and an3‐4∆DN–YDA stomata were marked using a mature guard cell‐specific green fluorescent protein (GFP) marker, E1728. Stomatal phenotype analyses were conducted for E1728 an3‐4 and E1728 an3‐4∆DN–YDA. Results showed that E1728 an3‐4∆DN–YDA failed to produce stomata, unlike E1728 an3‐4 (Figure 7c, d). Thus, ∆DN–YDA dramatically reduced the an3 mutant stomata index.

Discussion

AN3 is a small gene family member in Arabidopsis and is known as GRF‐INTERACTING FACTOR1 (AtGIF1) (Horiguchi et al., 2005). AN3 is implicated in cell proliferation regulation, adaxial/abaxial patterning in leaf primordia and establishment of cotyledon identity (Horiguchi et al., 2005; Kanei et al., 2012; Kim and Kende, 2004; Lee et al., 2009). Moreover, AN3 is also involved in secondary metabolism (Meng, 2014).

AN3 is a positive regulator of stomatal development as a transrepressor of YDA

YDA transcripts were up‐regulated in an3 seedlings, thus indicating that AN3 is required for the proper expression of YDA. AN3 was associated with chromatin YDA promoter; thus, AN3 is a transcriptional repressor of YDA. Genetic findings revealed that E1728 an3‐4∆DN–YDA failed to produce stomata, unlike E1728 an3‐4, thus indicating that the repression of YDA expression by AN3 can dramatically reduce the an3 mutant stomata index.

AN3 encodes a human transcription co‐activator homologue, that is, the synovial sarcoma translocation protein (SYT) (Clark et al., 1994; Crew et al., 1995; Thaete et al., 1999) and is a putative transcription co‐activator (Horiguchi et al., 2005). SYT contains a conserved N‐terminal domain called the SYT N‐terminal homology (SNH) domain, which is involved in protein–protein interactions (de Bruijn et al., 2001; Kato et al., 2002; Nagai et al., 2001; Thaete et al., 1999). This domain appears to be functionally important for AN3 because the an3‐2 mutation eliminates a conserved amino acid residue in the SNH domain (Horiguchi et al., 2005). While SNH of AN/GIF1 can interact with GRF1 for regulating leaf development (Kim and Kende, 2004), our results showed that stomatal density in grf1‐Knock Out (KO) mutants was similar to that of the wild‐type plants (data not shown), suggesting GRF1 is not interacting target gene of AN3/GIF1 for regulating stomatal development. A similar function was reported for numerous other proteins, that is, FLOWERING LOCUST (FT), GIGANTEA (GI), FLAVIN BINDING, KELCH REPEAT, F‐BOX1 (FKF1) and SHORT HYPOCOTYL UNDER BLUE1 (SHB1) (Sawa et al., 2007; Wigge et al., 2005; Zhou et al., 2009). Thus, AN3 probably acts as a cofactor, may interact with other unknown proteins that are YDA promoter‐specific transcription factors, and achieves regulation over the YDA gene transcription during stomatal development regulation (Figure 8a).

Figure 8

A proposed model illustrating drought tolerance by the AN3‐YDA signalling pathway. (A) AN3 is recruited to the promoters of YDA by protein X and regulates the expression of key genes such as YDA that are required for stomatal development. (B) AN3 negatively regulates WUE by positively regulating stomatal density and transpiration through transrepression of YDA. AN3 is expressed in water‐sufficient conditions to facilitate stomatal development and is down‐regulated under water deficit to reduce transpiration, which results in drought tolerance. Water‐sufficient conditions improved root systems and resulted in significantly improved survival during drought. Solid lines indicate direct regulation, whereas dotted lines indicate either indirect regulation or regulation in an unknown manner.

AN3 modulates stomatal development by YDA transrepression to regulate WUE

The stomatal lineage must respond to intralineage signals as well as systemic signals, such as hormones and environmental signals, for example, stress signalling for optimizing the number of stomata to the state of the plant and the environment. Drought signalling (such as relative humidity and water deficit) is an important factor. In this work, we found that AN3 is a negative regulator of drought tolerance, and AN3 was an important integration point between stress signalling and stomatal development.

Although mitogen‐activated protein kinase (MAPK) signalling is known to have key roles in controlling Arabidopsis stomatal development, whether and how MAPK signalling is involved in WUE regulation is still unknown. Our findings indicate that drought signals might be perceived by multiple photoreceptors that converge to repress AN3. Target‐gene analysis indicated that AN3 is associated with YDA promoter for drought stress‐controlling stomatal development regulation. Therefore, these stomatal development regulating components form a AN3‐YDA complex, allowing the integration of drought stress signalling into the production or spacing of stomata and cell proliferation, which leads to improved root system and reduced stomatal density and enhanced WUE (Figure 8b).

Experimental procedures

Plant materials and growth conditions

an3‐1 (CS241) and an3‐4 (Horiguchi et al., 2005), yda‐1 and yda‐2 (Bergmann et al., 2004) mutants, and 35Spro‐AN3‐3XGFP (Kanei et al., 2012) and pER8–XVEpro–∆N‐YDA (Bergmann et al., 2004) transgenic plants were described previously.

an3‐1, yda‐1, yda‐2, mpk6 (SALK‐073907), grf1 (SALK_069339C) and sdd1 (SALK‐035559) seeds were obtained from the ABRC (Ohio State University). The pGEM‐35Spro‐AN3‐3XGFP/an3‐4 seeds were kindly provided by Professor G. Horiguchi (Rikkyo University, Japan). The an3‐4 mutant seeds were kindly provided by Prof H.G Nam (DGSIT, Korea). The pER8–XVEpro–∆N‐YDA/Col‐0 and tmm seeds were kindly provided by Prof H.Q Yang (ShangHai JiaoTong University, China).

The an3 yda‐1/+ mutant was obtained from F2 seedlings of an3‐4 x yda‐1/+ that had narrow rosette leaf phenotype of plants grown in white light (the phenotype of an3‐4 homozygous mutants; Horiguchi et al., 2005) and dramatically shortened roots (the phenotype of yda‐1 heterozygous mutants; Lukowitz et al., 2004). The an3 yda‐1 mutant identified from offspring (F3) of an3 yda‐1/+ (F2), which had clustered stomata on the cotyledon in the dark (the phenotype of yda‐1 homozygous mutants; Kang et al., 2009). The above method about the double mutant construction has been described by Kang et al. (2009). E1728 an3‐4 XVE pro ∆N‐YDA and E1728 an3‐4 were obtained by cross, and the relative method has been described by Kang et al. (2009).Transgenic plants were generated using the Agrobacterium tumefaciens‐mediated floral dip method (Zhang et al., 2006).

Plants exhibiting the an3‐4 mutant phenotype [narrow rosette leaf phenotype of plants grown in white light (Horiguchi et al., 2005)] in the F2 populations were screened for pCB308R‐YDApro‐GUS expression in roots. F3 seeds were collected from those exhibiting expression, and lines expressing GUS in all F3 plants were used for subsequent analysis, as has been described by Zgurski et al. (2005). pMD111‐YDA pro‐GFP (YDA pro‐GFP containing two TCTC motif or YDA pro‐GFP without two TCTC motif) were introduced into the pCB2004‐35Spro‐AN3 and wild‐type background by Agrobacterium mediated transformation, as has been described by Lampard et al. (2008). Transformants were selected on hygromycin B for three or more generations and analysis of segregation ratios.

The seeds were subjected to 4 °C for 3 day and then sown onto solid Murashige and Skoog (MS) medium supplemented with 1% sucrose at pH 5.8% and 0.8% agar. The seedlings grown on agar were maintained in a growth room under 16/8 h light/dark cycles with cool white fluorescent light at 21 ± 2 °C. Plants grown in soil‐less media were maintained in a controlled environment growth room under 16/8 h light/dark cycles with cool white fluorescent light at 21 ± 2 °C. Relative humidity is ~60% in growth room (see Appendix S1 in details about Experimental procedures).

Acknowledgements

We thank Xiao‐Teng Cai, Yao Wang, Lin‐Hui Yu and Xiao‐Yu Guo for their technical assistance. We also thank Profs. Hong Gil Nam (DGSIT, Korea) and Cheng‐Bin Xiang (University of Science and Technology of China, China) for their support at the initial stage of this work.