Stomata are vital for the adaptation of plants to abiotic stress, and in turn stomatal density is modulated by environmental factors. Less clear, however, is whether regulators of stomatal development themselves participate in the sensing or response of stomata to abiotic stress. FOUR LIPS (FLP) and its paralog MYB88 encode MYB proteins that establish stomatal patterning by permitting only a single symmetric division before stomata differentiate. Hence, flp-1 myb88 double mutants have an excess of stomata, which are often misplaced in direct contact. Here, we investigate the consequences of loss of FLP/MYB88 function on the ability of Arabidopsis plants to respond to abiotic stress. While flp-1 myb88 double mutants are viable and display no obvious aerial phenotypes under normal greenhouse growth conditions, we show that flp-1 myb88 plants are significantly more susceptible to drought and high salt, and have increased rates of water loss. To determine whether flp-1 myb88 plants are already challenged under normal growth conditions, we compared genome-wide transcript levels between flp-1 myb88 and wild-type green tissues. Unexpectedly, uninduced flp-1 myb88 plants showed a reduced accumulation of many typical abiotic stress gene transcripts. Moreover, the induction of many of these stress genes under high-salt conditions was significantly lower in flp-1 myb88 plants. Our results provide evidence for a new function of FLP/MYB88 in sensing and/or transducing abiotic stress, which is severely compromised in flp-1 myb88 mutants.
Being unable to move, plants face many environmental challenges, which include drought, high salinity and extreme temperatures. To adapt to these adverse living conditions, plants have evolved biochemical, molecular and cellular defense mechanisms. At the biochemical level, the induction of proline and late-embryogenesis-abundant (LEA) proteins protects plants from stress (Xiong and Zhu, 2002). At the molecular level, the induction of stress-responsive and stress-tolerance genes also helps plants adapt to unfavorable environmental conditions (Matsui et al., 2008). At the cellular level, closing stomata and limiting vegetative growth facilitate the survival of plants when water is limiting (Flexas and Medrano, 2002). Indeed, it is well established that by opening and closing, and therefore controlling water vapor exchange (Davletova et al., 2005; Matsui et al., 2008), stomata play a vital role in the adaptation of plants to abiotic stress conditions (Melotto et al., 2006). More recent studies have further implicated stomata in biotic interactions, allowing pathogens to use the pore to gain access to the plant (Melotto et al., 2006).
Here, we provide evidence that FLP/MYB88 participate in drought and salt stress responses via a mechanism likely to be independent of stomatal opening and closing. We show that flp-1 myb88 plants are more susceptible to drought and salt stress conditions than wild-type plants. Genome-wide expression analyses demonstrate that many stress-responsive genes are down-regulated in flp-1 myb88 double mutants under normal growth conditions. The induction of these stress-responsive genes in flp-1 myb88 is greatly compromised under drought and high salt, suggesting that FLP/MYB88 might function, directly or indirectly, as positive regulators of stress-responsive genes. These data further indicate that FLP/MYB88 might directly mediate ABA responses. Together, our results highlight a role for the FLP/MYB88 regulators in controlling abiotic stress responses.
flp-1 myb88 plants show enhanced sensitivity to drought and salt stress
FLP (At1g14350, MYB124) and MYB88 (At2g02820) encode atypical R2R3-MYB proteins, which participate in Arabidopsis stomatal development (Yang and Sack, 1995; Lai et al., 2005). The loss of both FLP and MYB88 function (flp-1 myb88) leads to the formation of stomatal clusters (Lai et al., 2005), a situation that has clearly been selected against during plant evolution. Because clustering might compromise stomatal function, we hypothesized that these mutants are more susceptible to drought. Indeed, this is what we observed when seedlings or mature plants were subjected to water deficiency (Figures 1 and S1). While flp-1 myb88 plants overall display no obvious visual defects under normal growth conditions (Figure 1a), drought imposed by withdrawal of water for 4 days had a significant impact on flp-1 myb88 compared with wild-type plants (Figure 1b). Eventually, flp-1 myb88 plants died 8 days after withdrawal of water, whereas wild-type plants survived, despite showing symptoms of water deficiency (Figure S1). Similarly, the treatment of young seedlings with 50 mm NaCl for 3 weeks had a significantly more dramatic effect on flp-1 myb88 plants than on wild-type plants grown under identical conditions (Figure 1c). While flp-1 myb88 plants turned yellow with clear signs of dying, wild-type plants appeared green and healthy. In contrast, we observed no obvious difference between 2-week-old wild-type and flp-1 myb88 seedlings grown on MS plates (Figure 1d). To explore further whether these mutant phenotypes result from the loss of FLP/MYB88 function, flp-1 myb88 mutant plants were transformed with a construct harboring the FLP cDNA (cFLP) driven by 8.7 kb of the FLP promoter (pFLP). This construct was shown to be sufficient to complement the stomatal clustering phenotype of flp-1 as well as of flp-1 myb88 mutants (E. Lee and F. Sack, unpublished data). The exposure of flp-1 myb88/pFLP:cFLP transgenic plants to salt stress failed to induce any susceptibility phenotype, consistent with a complementation of the stress sensitivity (Figure S2).
Water loss assays using leaves detached from flp-1 myb88 or wild-type plants showed that the double mutant lost water significantly faster than the wild type (Figure 1e). To investigate the basis for this rapid water loss, we measured stomatal apertures in flp-1 myb88 and wild-type detached leaves with either water or ABA. In normal conditions, no significant differences in stomatal aperture were found for correctly patterned stomata between flp-1 myb88 and wild-type plants. However, ABA treatment induced significantly greater aperture closure in the wild type than in flp-1 myb88 (Figure S3), suggesting that the ABA response of stomata in flp-1 myb88 is impaired. Taken together, these results suggest that flp-1 myb88 mutants are more susceptible to abiotic stress than the wild type, and that this increased sensitivity probably involves ABA.
Genome-wide expression in wild type versus flp-1 myb88
To probe how FLP/MYB88 might participate in the sensing or response to abiotic stress, genome-wide gene expression analyses were conducted using the Arabidopsis ATH1 Affymetrix microarrays representing 22 810 genes, with RNA extracted from green tissues from 10-day-old flp-1 myb88 or wild-type seedlings, grown under normal conditions. We found similar gene expression patterns from three biological replicates for the wild type and flp-1 myb88 and the replicate results were highly reproducible (Figure S4). Stringent statistical analyses on the biological triplicates resulted in the identification of 521 genes that were significantly (P <0.05 and |fold change| >2) and differentially expressed between mutant and wild-type plants. Of these 521 genes, 194 were significantly lower in expression levels in flp-1 myb88 compared with the wild type, while 327 genes were significantly higher in flp-1 myb88 than in wild-type plants (Tables S1 and S2).
Using the Munich Information Center for Protein Sequences (MIPS) Gene Ontology (GO) classification, we investigated whether genes that were differentially expressed between flp-1 myb88 and the wild type were enriched in particular functional categories when compared with the whole genome. Among those with lower expression in flp-1 myb88, genes involved in ‘cellular communication/signal transduction’ (P =8.19 × 10−4), ‘cell rescue, defense and virulence’ (P =2.55 × 10−18), ‘interaction with the environment’ (P =3.40 × 10−13) and ‘systemic interaction with the environment’ (P =1.84 × 10−10) were found to be very significantly over-represented (Figure 2a). In contrast, we only observed a few functional categories that were over-represented among the genes with higher expression in flp-1 myb88, and none of those corresponded to stress genes (Figure S5). Interestingly, however, many of those genes are associated with processes likely to be involved in FLP/MYB88 control of stomatal development, such as the GO categories of ‘cell cycle’, ‘plant hormone responses’, ‘tissue pattern formation’ and ‘cell fate’ (Figure S5). These results indicate that FLP/MYB88 affect gene expression in two ways: FLP/MYB88 negatively control genes associated with stomatal development (Xie et al., 2010) and positively regulate genes related to stress conditions.
We further confirmed the microarray data using RT-qPCR (reverse-transcription quantitative PCR) on biological triplicates, and validated statistically significant differences (P-value <0.05 based on a two-sided t-test) in the expression of 22 out of 29 genes tested (Figure 2b). Among the genes validated as expressed at lower levels in flp-1 myb88, compared with the wild type, were WRKY40 (At1g80840), ERF6 (At4g17490), ZAT10 (At1g27730), ZAT12 (At5g59820) and DREB2A (At5g05410) (Figure 2b). Taken together, our results indicate that genes involved in abiotic stress responses are expressed at significantly lower levels in flp-1 myb88 mutants, compared with the wild type. These findings are difficult to reconcile with a model in which flp-1 myb88 plants are constitutively stressed because of defective opening/closing of stomata. Rather, our results suggest that in flp-1 myb88 plants the basal expression levels of stress genes are significantly lower than in the wild type.
To establish whether this is a consequence of abnormal stomatal patterning in flp-1 myb88 mutants, we investigated the expression of WRKY40, ERF6, ZAT10, ZAT12 and DREB2A in another Arabidopsis stomatal mutant, tmm-1. Loss of TMM gene function leads to a clustered stomatal phenotype as a consequence of the misplaced asymmetric divisions near existing stomata (Nadeau and Sack, 2003). Because the expression of stress-responsive genes did not differ significantly between wild-type and tmm-1 plants (Figure 3a), it is thus unlikely that the low level of expression of stress-responsive genes in flp-1 myb88 mutants is primarily due to abnormal stomatal patterning.
flp-1 myb88 mutants are unable to respond to drought stress
To determine whether the sensitivity to abiotic stress conditions of flp-1 myb88 plants is a consequence of a decreased ability to sense, and hence respond to, stress, young flp-1 myb88 and wild-type seedlings grown on MS plates were subjected to either drought conditions (20% polyethylene glycol, PEG), 300 mm NaCl, or were left untreated as a control (for details see Experimental Procedures). After 4 h, the material was collected, RNA extracted and gene expression assayed by RT-qPCR and normalized to PDF2/PP2A, a gene shown to be unaffected by these conditions (Berkowitz et al., 2008). WRKY40, ERF6, ZAT10, ZAT12 and DREB2A were randomly selected from the stress-related genes in the microarrays that showed significantly lower expression in flp-1 myb88, compared with the wild type, and that were then validated by RT-qPCR (Figure 2b). Consistent with their role in the response to abiotic stress (Matsui et al., 2008), the expression of WRKY40, ERF6, ZAT10, ZAT12 and DREB2A was very significantly induced in wild-type plants by NaCl (Figures 4a and S6; green color corresponds to wild type) and by drought (Figure 4b). In contrast, none of these genes was significantly induced by high-salt conditions or drought in flp-1 myb88 (Figures 4a, 4b and S6, red color corresponds to flp-1 myb88). Again highlighting that these effects are a consequence of the absence of FLP/MYB88 function, pFLP:cFLP expression in flp-1 myb88 mutants restored the normal induction of gene expression by salt for many of these genes (Figure S6). Our results suggest that the low expression of abiotic stress response genes in flp-1 myb88, and their lack of inducibility upon high salt or drought, are probably responsible for the increased susceptibility of flp-1 myb88 to abiotic stress.
The NAC019 regulatory gene is an immediate direct target of FLP/MYB88
To investigate whether FLP/MYB88 directly control abiotic stress genes, we generated antibodies against recombinant MYB88 protein which cross-react with FLP. As we recently described, these antibodies were used in chromatin immunoprecipitation (ChIP) experiments followed by hybridization of Arabidopsis whole-genome tiling arrays (ChIP-chip) (Xie et al., 2010). The only abiotic gene differentially expressed between wild-type and flp-1 myb88 plants (Table S1) that was also reproducibly FLP/MYB88 enriched in ChIP-chip experiments was NAC019 (At1g52890). NAC019 was previously shown to control abiotic stress responses, including dehydration, ABA and high salt. Moreover, NAC019 over-expression significantly induces stress-responsive genes (Tran et al., 2004), indicating that it is a positive regulator of abiotic stress responses. Using RT-qPCR, we confirmed the microarray results showing that NAC019 expression is significantly higher (two-sided t-test, P =0.04) in the wild type compared with flp-1 myb88 plants (Figure 5a). The direct binding of FLP/MYB88 to a region upstream of the NAC019 transcription start site was confirmed by ChIP-PCR, comparing the enrichment of the NAC019 promoter between immunoprecipitated chromatin obtained from wild-type or flp-1 myb88 seedlings (Figures 5b and S7). These results demonstrate that FLP/MYB88 directly activate NAC019 gene expression, and suggest that the regulation of other abiotic stress genes might be indirect, probably mediated through NAC019.
ABA sensing is impaired in flp-1 myb88 mutants
One of the main functions of the plant hormone abscisic acid (ABA) is to regulate plant water balance and tolerance to osmotic stress. Under drought or osmotic stress conditions, ABA synthesis increases and its degradation decreases (Jakab et al., 2005). Treatments with ABA also result in significant alterations of gene expression (Matsui et al., 2008). To determine whether ABA bypasses the inability of flp-1 myb88 mutants to efficiently respond to drought or salt stress, we investigated the expression of WRKY40, ERF6, ZAT10, ZAT12 and DREB2A in wild-type and in flp-1 myb88 mutant seedlings upon ABA treatment. Similar to what was observed for the NaCl treatments, ABA efficiently induced the expression of these genes in the wild type but not in the flp-1 myb88 mutants (Figure 4c). These results suggest that the stress-sensing deficiency of the flp-1 myb88 mutants is probably downstream of ABA. Consistent with this hypothesis, ABA levels in wild type and flp-1 myb88 mutant seedling shoots are not significantly different, either under normal growth conditions or after the NaCl treatments (Figure 6a). As expected, there is also no difference in ABA levels between salt-treated tmm-1 and wild-type plants (Figure 3b).
Abscisic acid is also an inhibitor of seed germination (Nambara and Marion-Poll, 2005). To test the hypothesis that ABA response is impaired in flp-1 myb88 mutants, we investigated whether FLP/MYB88 affect seed germination. Within 2 days, most of the flp-1 myb88 and wild-type seeds showed significant germination. However, the flp-1 myb88 seeds germinated significantly faster than the wild type (Figure S8a). To rule out possible differences in pre-existing ABA levels in the seeds, we determined the ABA content in wild-type and flp-1 myb88 seeds. Similar to our data for seedlings (Figure 6a), ABA levels were not significantly different between wild-type and flp-1 myb88 seeds (Figure 6b). Furthermore, when seeds of both genotypes were germinated in different concentrations of ABA, flp-1 myb88 seeds germinated significantly faster than the wild type, within the first 60 h in the presence of 1 or 2 μm ABA (Figure S8b). However, in 10 μm ABA, there were no significant differences in the rates of germination between wild-type and flp-1 myb88 seeds (Figure S8b). Thus, excess ABA appears to swamp the sensitivity differences between wild-type and flp-1 myb88 seeds. These results indicate that the ABA response of flp-1 myb88 mutants is altered compared with the wild type.
Our results indicate that the Arabidopsis stomatal patterning regulators FLP and MYB88 participate in the response to abiotic stress. First, normally grown flp-1 myb88 plants express abiotic stress genes at significantly lower levels than wild-type plants. In addition, high-salt treatments, which in wild-type plants strongly induce stress genes including WRKY40, ERF6, ZAT10, ZAT12 and DREB2A, have little or no effect in flp-1 myb88. FLP and MYB88 thus appear to act in a pathway that senses abiotic stress. Because we identified NAC019 as an immediate direct target of FLP/MYB88, it is likely that these MYB regulators control the abiotic stress-response pathway at least in part by regulating NAC019 gene expression. In addition, stress sensing might be dampened by the arrested development of many stomata in flp-1 myb88 double mutants.
In green tissues of the epidermis of 10-day-old seedlings, FLP is expressed specifically in the stomatal cell lineage, especially just before and after the division that creates the stoma (Lai et al., 2005; Xie et al., 2010). Leaves of 10-day-old flp1 myb88 plants contain many stomata, with about 80% of them present in clusters (Lai et al., 2005). However, most of the genes involved in abiotic stress identified as differentially expressed between wild-type and flp-1 myb88 mutant plants, including WRKY40, ERF6, ZAT10, ZAT12 and DREB2A, are not expressed specifically in the stomatal cell lineage (Table S3) (Leonhardt et al., 2004). While it is not known whether low levels of either FLP or MYB88 are expressed in mature stomata or more widely in leaves, it is tempting to speculate that stomata might function as ‘stress signal sensing and distribution centers’. During abiotic stress, ABA signals originating in roots (Nambara and Marion-Poll, 2005) are sensed by stomata and amplified by the generation of additional ABA and perhaps other diffusible signals. We hypothesize that, in the flp-1 myb88 mutant, a sensing mechanism is impaired by the frequent presence of incompletely developed guard cells in the clusters. By this reasoning, a hypothetical non-cell-autonomous signal that normally originates from stomata would be lacking, resulting in a failure to induce abiotic stress genes. Finally, although the expression of FLP and MYB88 during stomatal development is well characterized, these genes are also expressed in roots (Lai et al., 2005). It is therefore possible that aspects of the stress-response regulation reported here are controlled by FLP and MYB88 function in tissues outside of, or in addition to, the stomatal pathway.
The biological functions of FLP and MYB88 in abiotic stress responses might provide a potential target for improving the adaptation of plants to abiotic stress as do two other R2R3-MYB proteins, AtMYB60 and AtMYB61 (Cominelli et al., 2005; Liang et al., 2005). Together with other studies (Gray et al., 2000; Wang et al., 2007; Kanaoka et al., 2008), our findings suggest additional evidence for the cross-talk between plant development and signal transduction pathways under conditions of environmental stress (biotic or abiotic).
Arabidopsis flp-1 myb88 plants were generated by introgressing a T-DNA insertion line in MYB88, SALK_068691 (insertion in exon 10; Columbia ecotypic background), into plants containing an ethyl methanesulfonate (EMS)-mutagenized flp-1 allele (Lai et al., 2005). A pFLP:cFLP construct was generated by cloning an 8.7-kb FLP promoter and FLP cDNA into a pGWB4 destination vector using the GATEWAY cloning system. Transgenic flp-1 myb88 pFLP:cFLP plants were obtained by stably transforming this translational construct into flp-1 myb88 plants (E. Lee and F. Sack, unpublished data). After surface sterilization, seeds were plated on half-strength Murashige and Skoog (MS) medium containing 1% glucose. Plates were vernalized in a cold-room for 3 days, and then grown on continuous light.
Phenotypic studies on flp-1 myb88 under drought and salt stress conditions
To check for drought responses, pot-grown 1-week old flp-1 myb88 and wild-type seedlings were grown for 4 or 8 days without watering. For salt treatment experiments, 1-week-old flp-1 myb88, flp-1 myb88/pFLP:cFLP as well as wild-type seedlings were grown in half-strength MS plates and then transferred to half-strength MS plates containing 0 mm NaCl, 50 mm NaCl or 100 mm NaCl, and grown for another 3 weeks with the agar in a vertical orientation.
Water loss measurement
Plants were grown in soil for 12 days after germination. The third and fourth leaves were excised and weighed immediately after excision, left on the bench at room temperature (22°C) and weighed every hour thereafter. Three biological replicates, consisting of third and fourth leaves excised from three separate plants, were used for each line, and the average values were used for calculating water loss over time.
Microarray hybridization and analysis
Total RNA was extracted from 10-day-old flp-1 myb88 and wild-type shoots using the Qiagen RNeasy Mini kit (Qiagen, http://www.qiagen.com/). The quantity of RNA was evaluated using a Bioanalyzer 2100 (Agilent, http://www.agilent.com/). The hybridization to an Arabidopsis Affymetrix ATH1 microarray was done at the Functional Genomics Core Facility, Nationwide Children’s Hospital Research Institute (Columbus, OH; http://www.dnaarrays.org). Three biological replicates corresponding to RNA extracted from different plants in the same plate were conducted on both flp-1 myb88 and wild-type plants.
Using the R package, microarray data were pre-processed by RMA (robust multi-array average). Expression values were computed by background correction by applying the RMA algorithm. Probe values were then normalized by quantile normalization, and further summarized by calculating the median polish algorithm using RMA. The resulting RMA values were log2-transformed. Since Q–Q plots showed that the microarray data followed a near normal distribution, two-sample t-tests in anova were used to calculate the P-values for each gene contrasting flp-1 myb88 and wild-type samples. Only probes with adjusted P-values smaller than 0.05 were selected. Of these, genes with differential expression of more than 1.0 (log2) were further considered. These genes were classified according to their putative functions using the Munich Information Center for Protein Sequence (MIPS) (http://mips.helmholtz-muenchen.de/proj/funcatDB/search_main_frame.html). Microarray data from these experiments was submitted to GEO (accession numbers GSE12911, GSM324018 and GSM324020).
Quantitative PCR experiments were performed on the cDNA obtained by reverse transcription of RNA from 10-day-old wild-type, flp-1 myb88, flp-1 myb88/pFLP:cFLP and tmm-1 (Nadeau and Sack, 2002) shoots by real-time RT-PCR using SYBR-Green chemistry. For probing gene expression under abiotic stress conditions, 10-day-old wild-type and flp-1 myb88 seedlings that were cultured in half-strength MS liquid medium were treated with 20% PEG, 100 μm ABA or 300 mm NaCl for 4 h, and the shoots were collected for RNA extraction, which was used for RT-qPCR experiments as previously described. The housekeeping gene PDF2/PP2A was used as an internal control for normalization (Besson-Bard et al., 2009). Primers used for RT-qPCR are shown in Table S4.
Wild-type and flp-1 myb88 seeds as well as 10-day-old shoots of wild type, flp-1 myb88 and tmm-1 were exposed to zero or 300 mm NaCl for 4 h and then collected. After weighing, they were ground into a fine powder using a Beadbeater (Biospec Products, http://www.biospec.com/) for 30 sec at medium power. The ABA was subsequently extracted in Eppendorf tubes by shaking in 80% methanol at 4°C for 48 h. The supernatants were then concentrated at 42°C for 1 h using a SpeedVac (http://www.thermoscientific.com). After adding 1× TBS buffer, the ABA concentration in three biological replicates was determined by the monoclonal anti-ABA antibody ELISA method using a Phytodetek ABA Immunoassay kit (Agdia Inc., http://www.agdia.com/).
Chromatin immunoprecipitation (ChIP)
Antibodies against the FLP/MYB88 proteins were generated by inoculating rabbits with Ni-NTA affinity-purified NHis6-MYB88 (Xie et al., 2010). The ChIP experiments on chromatin obtained from 10-day-old seedlings from the wild type and flp-1 myb88 mutants were performed as described (Morohashi et al., 2007), with some modifications (Xie et al., 2010). The primers used for qPCR were pNAC019-F and pNAC019-R, and PDF2-F and PDF2-R; their corresponding positions are shown in Figure S7.
To quantify germination rates, wild-type and flp-1 myb88 seeds (collected at the same time) were surface sterilized, plated on half-strength MS medium containing 0, 1, 2 or 10 μm ABA, stratified for 3 days at 4°C, and then transferred to continuous light. Germination (judged by the emergence of the radicle) was assessed and scored over time. Three biological replicates were conducted for each line, using 10 seeds per replicate.
Stomatal opening and closing
To measure how ABA affect stomatal opening and closing in wild-type and flp-1 myb88 plants, abaxial epidermis from three or four leaves was peeled off from 2-week-old wild-type and flp-1 myb88 plants, respectively, and then treated with either 10 μm ABA or water as a control for 10 min. The number of stomata either closed or opened was scored under a regular compound microscope. At least three biological replicates were conducted for each line, at least 10 normal-looking stomata per replicate.
We thank Dr Eunkyoung Lee for sharing unpublished data on the complementation of flp-1 myb88 by pFLP:cFLP, and allowing us to use the unpublished transgenic line. This work was supported by National Science Foundation grant MCB-0418891 to EG, an NSERC Discovery grant to FS, and an Excellence Graduate Fellowship from the Plant Molecular Biology/Biotechnology Program at the Ohio State University to ZX.