An Arabidopsis thaliana drought-tolerant mutant, altered expression of APX2 (alx8), has constitutively increased abscisic acid (ABA) content, increased expression of genes responsive to high light stress and is reported to be drought tolerant. We have identified alx8 as a mutation in SAL1, an enzyme that can dephosphorylate dinucleotide phosphates or inositol phosphates. Previously identified mutations in SAL1, including fiery (fry1-1), were reported as being more sensitive to drought imposed by detachment of rosettes. Here we demonstrate that alx8, fry1-1 and a T-DNA insertional knockout allele all have markedly increased resistance to drought when water is withheld from soil-grown intact plants. Microarray analysis revealed constitutively altered expression of more than 1800 genes in both alx8 and fry1-1. The up-regulated genes included some characterized stress response genes, but few are inducible by ABA. Metabolomic analysis revealed that both mutants exhibit a similar, dramatic reprogramming of metabolism, including increased levels of the polyamine putrescine implicated in stress tolerance, and the accumulation of a number of unknown, potential osmoprotectant carbohydrate derivatives. Under well-watered conditions, there was no substantial difference between alx8 and Col-0 in biomass at maturity; plant water use efficiency (WUE) as measured by carbon isotope discrimination; or stomatal index, morphology or aperture. Thus, SAL1 acts as a negative regulator of predominantly ABA-independent and also ABA-dependent stress response pathways, such that its inactivation results in altered osmoprotectants, higher leaf relative water content and maintenance of viable tissues during prolonged water stress.
Abiotic stresses including drought, temperature extremes, salinity, excessive light and combinations thereof are major limiting factors in crop yields in many parts of the world (Boyer, 1982; Araus et al., 2002). Plants can respond to drought by three different mechanisms: (i) escape, where the plant stops growth and promotes flowering to produce seed before death; (ii) avoidance, where the stomata are closed to prevent water loss or water uptake is increased (Jones, 2007); and (iii) tolerance or acclimation, which we define as the ability of a plant to subsist and grow in conditions of limited water under which wild-type plants cannot. In agriculture, tolerance is the better target as it can allow the maintenance of yield. Tolerance involves the accumulation of antioxidants, osmolytes and chaperones that protect cell components and changes in the root-to-shoot ratio to maximize water absorption (Sharp and Davies, 1985; Reddy et al., 2004).
To date, much of the research on Arabidopsis thaliana (Arabidopsis) has focused on the molecular response to sudden changes in osmotic potential or osmotic shock, largely initiating a stress avoidance response that is measured over a period of hours. Multiple signaling pathways are activated in response to osmotic shock (reviewed in Shinozaki and Yamaguchi-Shinozaki, 2007). Slower, and arguably more physiologically relevant, soil-based experiments occur over 1–2 weeks and initiate some avoidance responses, but also activate tolerance responses (Kawaguchi et al., 2004; Verslues et al., 2006). However, soil-based experiments require controls to ensure that all plants are exposed to the same stress (Verslues et al., 2006). While many components of these stress response pathways have been isolated, the exact order, interactions and outcomes of the stress networks in physiologically relevant conditions are unknown.
The levels of inositol phosphate molecules are regulated both by synthesis from PIP2 (Munnik et al., 1998) and catalysis by phosphatases, including 5-phosphatases (5Ptases) and inositol polyphosphate 1-phosphatases, which hydrolyze the 5′ position phosphate and 1′ position phosphate, respectively (Gillaspy et al., 1995; Quintero et al., 1996; Berdy et al., 2001). SAL1 is an inositol polyphosphate 1-phosphatase that was originally isolated from Arabidopsis for its ability to complement a salt-sensitive yeast strain (Quintero et al., 1996). The protein is bifunctional, as it also exhibits 3′,(2′),5′-bisphosphate nucleotide phosphatase activity, being highly specific for 3′-polyadenosine 5′- phosphate (PAP) (Quintero et al., 1996; Gil-Mascarell et al., 1999). Recently, SAL1 was shown to be a repressor of post-transcriptional gene silencing (PTGS), as it degrades the exoribonuclease inhibitor PAP (Dichtl et al., 1997; Gy et al., 2007). The fry1-1 (fiery) mutant allele of SAL1 has higher basal levels of IP3 and increased expression of some stress response genes, but paradoxically it is reported to have increased sensitivity to salt, cold and drought stress (Xiong et al., 2001).
The Arabidopsis mutant alx8 was isolated in a screen for elevated expression of the antioxidant enzyme ascorbate peroxidase 2 (APX2) under low light (LL) and high light (HL) (Rossel et al., 2004). The alx8 mutant also had increased tolerance to drought and altered leaf morphology (Rossel et al., 2006). Positional cloning identified alx8 as a mutation in the SAL1 gene. Here we resolve aspects of the SAL1 paradox by demonstrating that intact soil-grown fry1-1 and alx8 are drought tolerant. We investigate the processes by which SAL1 may impact on stress signaling pathways and propose that the acquired drought tolerance in alx8 plants could be due partially to an increase in some sugars, osmoprotectants and antioxidants and up-regulation of stress-inducible genes.
alx8 contains a point mutation in inositol polyphosphate 1-phosphatase, SAL1
alx8 (Col-0) was crossed with Landsberg erecta. The F1 was wild-type in leaf phenotype, development and APX2 expression (data not shown), confirming alx8 as recessive. Coarse mapping of 53 mutant F2 individuals with 22 primers indicated that alx8 was linked to chromosome five (Figure 1a). Fine mapping of 4300 F2 individuals was undertaken with two markers, MUB3 and MMI9, flanking the 617-kb region. Genotyping and phenotyping of recombinants eliminated all but seven genes on the MBM17 bacterial artificial chromosome (BAC), one of which was SAL1 (At5g63980; FRY1/HOS2). A line with a T-DNA insertion in SAL1, salk_020882, phenocopied alx8 and all F1 progeny of alx8 x salk_020882 had altered leaf morphology and delayed development.
A wild-type copy of the SAL1 gene and promoter, pCAM-SAL1, was used to complement alx8 and wild-type plants. For alx8, two individual T1 transformants with wild-type leaf morphology were identified. Their progeny segregated for kanamycin resistance and leaf phenotype with over 100 T2 complemented individuals having wild-type leaf morphology, timing of flowering, and development (Table 1). Wild-type plants transformed with this construct showed no change in leaf morphology, APX2 expression or development (Table 1).
Table 1. Complementation of alx8 and the wild type with SAL1
+/+ indicates that the phenotype is indistinguishable from wild type. For alx8+ pCAM-SAL1 145 plants from two independent transformations were analyzed. For Col-0+ pCAM-SAL1 plants from 11 independent transformations were analyzed.
Sequencing of the alx8 gene, 2 kb of promoter and 3′ untranslated region (UTR) from 10 mutant plants revealed a single nucleotide polymorphism of guanine to adenine at the 1226th base pair of the At5g63980.1 genomic sequence (TAIR accession: 4010730406) (Figure 1b). The progeny of the fourth generation of backcrossed alx8 were also screened for the point mutation using derived cleaved amplified polymorphic sequence (Neff et al., 1998) markers to confirm the leaf phenotype was co-inherited with the point mutation and not another locus.
The alx8 mutation results in an amino acid change of glycine to aspartic acid at the 217th amino acid of the SAL1 protein (TAIR accession: 4010745380). In silico protein modeling against the known structure of the yeast homolog of SAL1, known as HAL2, showed that the amino acid substitution in alx8 localizes in a conserved internal β-sheet of unknown function.
The SAL1 protein is absent in alx8 and fry1-1
The recombinant protein for the wild-type SAL1 gene (rSAL1) and the mutated alx8 gene (rALX8) were produced as fusions to polyhistidine-tagged ubiquitin in Escherichia coli. Their calculated masses (∼50 kDa) corresponded to the expected sizes based on the amino acid sequence plus the tag. Although many different induction conditions were attempted, only the rSAL1 fusion protein was successfully purified in the soluble fraction as one band (Figure S1 in Supporting Information). This protein was assayed for phosphatase activity against PAP as described (Murguia et al., 1995). Recombinant SAL1 fusion protein showed similar PAP phosphatase enzymatic activity (16.6 ± SE 3.65 μmol PAP h−1 (mg protein)−1, n =3) to that previously reported (Xiong et al., 2001). The authentic protein obtained after cleavage of the tag was used to generate antibodies.
Western blot analysis was performed to determine the abundance of the SAL1 protein in wild-type and SAL1 mutant plants. A single 37-kDa band, whose size corresponded to that of the authentic recombinant protein rSAL1, was detected in the soluble protein extract from leaves of Col-0 and C24, but not in alx8 or fry1-1 leaves (Figure 1c). Additionally, SAL1 protein was not present in a urea-based total protein extract of alx8 but was in Col-0 (Figure 1c).
Drought tolerance of plants with alx8, salk_020882 and fry1 mutations in SAL1
alx8 has previously been shown to be drought tolerant in soil-based experiments (Rossel et al., 2006) whereas fry1-1 was reported as being drought sensitive in response to osmotic shock (Xiong et al., 2001). Hence, we tested fry1-1 and salk_020882 for soil-based drought tolerance by withholding water from mature vegetative plants for 18 days. In three independent experiments with at least five plants per experiment fry1-1, salk_020882 and alx8 were more drought tolerant than the respective C24 and Col-0 wild-types (Figure 2a). To quantify the extent of the tolerance, a measure of plant viability using chlorophyll fluorescence was undertaken as described (Woo et al., 2008). The results demonstrate that alx8 plants survive drought for 40–50% longer than Col-0 (Table 2).
Table 2. Viability of alx8 and wild-type during drought
Viability of plants during drought was assayed by measuring Fv/Fm. The day on which Fv/Fm is <33% of the controls, the plant is deemed unviable and will not recover if rewatered (Woo et al., 2008).
13 ± 3.0
19 ± 1.1
In a different experiment, leaf relative water content (RWC) was measured after 12 days of drought (Figure 2b). As expected, the leaf RWC of alx8 did not change significantly during drought treatment, while the leaf RWC of the wild-type decreased significantly. Leaf water potential, or the chemical potential of water divided by the partial molar volume, was calculated for the same plants using a thermocouple psychrometer (Figure 2c). The higher water potential of alx8 plants grown under drought conditions correlated with the maintained turgor while water-stressed wild-type plants had a significantly lower leaf water potential.
To ensure that all plants were being exposed to the same water stress, the relative water content of the soil was calculated. There was no significant difference in the water content of the soil between genotypes throughout the experiment (Figure 2d), indicating that water was lost at similar rates from the soil whether via plant transpiration or evaporation. Moreover, approximately 80% of the water was lost in the first week of drought, resulting in values of soil water potential lower than −800 kPa for the reminder of the treatment (Figure 2e).
The drought tolerance of SAL1 mutants was tested at different developmental stages and ages to ensure it was not a result of the developmental delay or smaller size of the plants (Table 3, Figure 3). alx8 was more drought tolerant than Col-0 wild-type at 2, 4, 6 and 8 weeks old. alx8 was also more drought tolerant than Col-0 wild-type when: both plant types had six leaves; both were mature and vegetative; and both were just starting to flower (Table 2, Figure 3a,b). When the plants were just commencing to flower all lines were of roughly the same rosette area and size, yet after 9 days of withholding water, alx8 and salk_020882 were both more turgid and green than wild-type (Figure 3a), although pots lost water at similar rates (Figure 3c). Similarly, alx8 and fry1-1 plants grown to similar vegetative stages were more drought tolerant than wild-type (Figure 3b). alx8 is also more drought tolerant than Col-0 when it has two plants per pot versus one for Col-0 (Figure S2) (Rossel et al., 2006). Finally, Col-0 plants of varying sizes are equally susceptible to drought (Figure 3d).
Table 3. Drought tolerance at various ages and stages of development
Plants grown to the same chronological age or the same developmental stage were subjected to drought as described in Figure 2 and scored as more tolerant than wild type (yes) or not different from wild type (no). The number of replicates for each mutant tested at a given stage is indicated.
Yes (alx8, salk_020882, fry1-1)
15, 10, 10
Yes (alx8, salk_020822)
Yes (alx8, fry1-1)
Yes (alx8, salk_020882), no (fry1-1)
5, 5, 5
Water loss of detached rosettes was measured for the four genotypes to determine whether leaf and rosette shape alters the rate of water loss, or whether stomatal control (the initial phase of water loss) or cuticular evapotranspiration (the second phase of water loss) were altered. No significant difference was detected between the four lines in either phase (Figure 3e), which is similar to earlier observations for fry1 (Xiong et al., 2001).
As both alx8 and fry1-1 lack SAL1 protein, the difference in their reported stress sensitivity is surprising. To see if this variation is due to differing ecotype or the nature of the imposed stresses, we repeated the polyethylene glycol (PEG) osmotic shock experiment (Xiong et al., 2001). alx8 and fry1 seedlings were more tolerant to PEG-6000-induced osmotic stress than their respective wild types in three separate experiments (Figures 3f and S3).
Molecular effects of the alx8 mutation
Global expression of SAL1 mutants was measured to investigate the extent of change in mRNA, in addition to the six genes reported to change in fry1 and alx8 (Xiong et al., 2001; Rossel et al., 2006). Of the approximately 24 000 gene products quantified by microarray, 1414 were significantly up-regulated and 1033 down-regulated more than two-fold in alx8 leaf tissue relative to their expression in Col-0 wild type (P ≤0.05, q ≤0.1; Figure 4a, Table S1). In fry1-1 under well-watered conditions 1099 genes were significantly up-regulated and 745 down-regulated more than two-fold relative to C24 wild type (P ≤0.05, q ≤0.1; Table S2). APX2, ZAT10 and RAP2.6 are up-regulated in both the arrays and quantitative reverse transcription-PCR (qRT-PCR) (Rossel et al., 2006). The 3.1-fold change in RAP2.6 in the arrays was not significant due to variability between biological replicates (Table S1, Xiong et al., 2001). The absence of any change in the arrays for NCED3, GST6, APX1, RD29A and DREB2A in alx8 under well-watered conditions was also consistent with published results (Rossel et al., 2006). The lack of induction of RD29A was confirmed by qRT-PCR (Figure 4b). Significantly, neither SAL1 nor any of the five SAL1 homologs were down-regulated in the arrays. One, SAL3 (At5g63990), was up-regulated 16-fold; this was confirmed by qRT-PCR where a four-fold up-regulation was observed (SI in Table S3). Quantitative RT-PCRs for SAL1 and SAL2 (At5g64000) transcripts in alx8 were marginally higher than wild-type, but this was not statistically significant (SI in Table S3).
A number of stress response genes were up-regulated in alx8, including: transcription factors such as ZAT10, ZAT12, MYC2, RAP2.6 and HB6; several early light inducible proteins, ELIPs; an aquaporin TIP5;1; stress signaling kinase SnRK2.2; stress-inducible proteins VSP1 and VSP2; and antioxidant enzymes APX2, CSD1 and CSD2 (Table 4).
Table 4. Changes in gene expression in alx8 relative to Col-0
Selected up-regulated genes (2-fold, P ≤0.05, q ≤0.1). Values are the average of three biological replicates run on separate Affymetrix ATH1 chips.
aStress inductions are from the literature or Genevestigator.
bFold change compared to A call (below background level).
Genes whose expression was up-regulated more than 25-fold were analyzed for co-expression in arrays stored in public databases using Genevestigator (Figure 4c). There was some correlation with drought, wounding, heat and cold arrays. Twelve per cent of the up-regulated genes in alx8 were ABA-inducible, but a similar number were down-regulated by ABA. There was no strong correlation with other hormone treatments, except possibly with jasmonic acid treatment.
Role of SAL1 in stomatal morphology and dynamics
Previously, alx8 was shown to have lower stomatal conductance relative to wild-type as measured by gas exchange (Rossel et al., 2006). This was not due to altered stomatal density, size and/or morphology of the alx8 stomata. The stomata in alx8 were normal in appearance and of a similar size to the wild type (Figure 5a). The stomata were not clustered as seen in mutants with altered stomatal development, neither were they located in a pit, which would decrease conductance by increasing resistance due to the boundary layer affect. The stomatal index of the abaxial and adaxial surface of alx8 and wild-type leaves was calculated using epidermal peels (Figure 5b). No difference in stomatal index on the abaxial surface was found and there was a small, but significant, decrease on the adaxial surface of alx8 leaves (P <0.01, d.f. = 48). This was confirmed by calculations of the stomatal index from scanning electron microscopy images of the abaxial surface of independently grown plants (data not shown). The stomatal aperture, as measured by epidermal peels under non-stressed conditions at four time-points during the day, was not significantly different between alx8 and wild type at all time-points and on both leaf surfaces (Figure 5c,d). The carbon-13/carbon-12 isotope ratio (δ13Cp) of leaf tissue of alx8 and Col-0 wild type was compared as an indication of water use efficiency (WUE) over the lifetime of the plant. This ratio is inversely proportional to WUE and it is dependent on both the rate of carbon fixation and stomatal conductance (Farquhar and Richards, 1984). Under well-watered conditions it was found that there was no difference in δ13Cp between the wild type and alx8 (Figure 5e).
Plant morphology, growth and development
The leaves of alx8, fry1-1 and salk_020882 were shorter and rounder that wild-type leaves, with more lobed edges; the surface was often undulated and the petioles were shorter (Figure 6a). There were changes to the internal structure of the leaf including disorganized vascular bundles, altered cell shape, less defined palisade structure, smaller chloroplasts (Figure 6b) and thicker leaves (Figure 6b,c). There was no visible difference in the cuticle thickness or structure in alx8 compared with wild type (Figures 5a and 6b). Although delayed in development and appearing smaller due to changes in leaf and rosette shape, alx8 plants can accumulate similar rosette fresh and dry mass to the wild type by 8 weeks of growth (Table 5).
Table 5. Fresh and dry rosette weights
Fresh weight [SD] (g)
Dry weight [SD] (g)
Whole rosettes of 8-week-old plants were measured for their fresh and dry weight. Eight plants from each genotype were analyzed. Differences between genotypes were not significant by an unpaired t-test.
Metabolomics and starch accumulation
The metabolic profiles of alx8 and fry1-1 were compared with those of their respective wild types under well-watered conditions by gas chromatography-mass spectrometry (GC-MS). All four lines had different profiles as indicated by principal components analysis (PCA), with a degree of overlap between C24 and Col-0 (Figure 7). Both mutants were clearly separated from their respective wild types by the first principal component (PC1; accounting for 47.2% of total variance) representing the largest class separation observable by PCA. Interestingly, alx8 and fry1-1 were clearly separated by the second principal component (PC2; accounting for 25.1% of total variance) while the two wild types were not.
There were significant increases in the levels of the polyamine putrescine in alx8 and fry1-1 (Table 6). This correlated with an increase in expression of the rate-limiting polyamine biosynthesis gene arginine carboxylase (ADC2) (6.43-fold) and a decrease in expression of the enzyme that converts spermidine to spermine, ACL5 (−3.90-fold). There was no significant difference in the proline abundance in alx8 relative to Col-0 despite an increase in the proline biosynthesis gene, pyrroline-5-carboxylate reductase (3.61-fold). In both SAL1 mutants there were changes in the abundance of a large number of sugars. These included decreased levels of fructose, galactose, glucose, cellobiose and a large number of unknown sugars and sugar derivatives. There was also a dramatic accumulation of unknown sugars and sugar derivatives that were present at very low levels in the wild type (Table 6). Also striking were strong decreases in the organic acids citrate, isocitrate, fumarate and malate and strong increases in a number of metabolites of unknown class (Table 6).
Table 6. Characteristic metabolic profiles of SAL1 mutants
alx8 versus Col-0
fry1-1 versus C24
Fold differences and P-values (n =5) are shown for the major (most intense) metabolite differences that were common to both alx8 and fry1-1. Unknown metabolites (enclosed in square brackets) were annotated based on the similarity of their mass spectra to reference spectra in the NIST05 mass spectral library. Retention indices (RI) are given for unknown metabolites.
Transmission electron microscopy of wild-type and alx8 leaves revealed that alx8 chloroplasts lack transitory starch granules (Figure 8a). Iodine staining confirmed the relative reduction in starch accumulation in alx8 and fry1 compared with their respective wild types in both evening and morning samples (Figure 8b). The decrease in starch correlates with an increase in expression of two β-amylases, BMY1 (9.53-fold) and BMY8 (2.31-fold), in alx8.
SAL1 is a negative regulator of drought tolerance
This study shows that the nucleotidase/phosphatase SAL1 is a negative regulator of drought response networks that alter the transcriptome, metabolome and morphology of the rosette. Three different SAL1 mutations all confer drought tolerance, thereby resolving the apparent paradox of SAL1; namely, SAL1 is a negative regulator of stress signaling, but the fry1-1 mutant did not show enhanced tolerance to the abiotic stresses tested (Xiong et al., 2001). Here we demonstrate that SAL1 does negatively regulate transcriptional and metabolic pathways and SAL1 knockouts have enhanced tolerance to drought.
The paradox did not reflect gain-of-function mutations as alx8 and fry1-1 lack detectable SAL1 protein (Figure 1c) and fry1-1 has reduced enzymatic activity on both substrates of SAL1 (Xiong et al., 2001). The alx8 mutation, which occurs in a conserved domain of unknown function, inhibits accumulation of SAL1 protein (Figure 1c), but does not inhibit SAL1 transcription. The mutated ALX8 protein produced in E. coli could not be recovered in the soluble fraction. It is not clear whether a translational blockage or a lack of protein stability in plants accounts for the absence of SAL1 in alx8 plants. Significantly, both alleles lack detectable SAL1 protein in leaves.
The contrasting reports of drought sensitivity of fry1-1 may in part reflect the different conditions used to impose water stress and as a consequence provide insight into the role of SAL1. Here intact, mature plants in soil are subjected to a gradual decrease in water availability over a period of 1–2 weeks. In the study where fry1-1 was shown to be drought sensitive (Xiong et al., 2001) leaves were detached or the seedlings were grown on media and exposed to mannitol or PEG-6000 to induce osmotic shock. Treatments with high concentrations of mannitol may result in plasmolysis of the cell rather than cytorrhysis (Munns, 2002; Verslues et al., 2006) and toxicity (Hohl and Schopfer, 1991; Verslues et al., 1998) leading to effects not related to water stress. With respect to PEG-6000, we observed decreased electrolyte leakage in SAL1 mutants in response to PEG (Figures 3 and S3), which is the opposite result to that reported by Xiong et al. (2001). Regardless of the differing results for electrolyte leakage in response to PEG, the drought tolerance in soil was highly significant (Figures 2 and 3). In this context it is worth noting that only 27 of 806 up-regulated genes were co-regulated in three water stress experiments using mannitol, filter paper or soil water deficit (Bray, 2004). Consequently, it is important to consider the nature of the imposed stress when evaluating the role of components of stress signaling networks.
SAL1 mutants have higher leaf relative water content during drought but not constitutively increased water use efficiency
To investigate the mechanisms of drought tolerance we explored the physiological and morphological changes in alx8 and fry1-1. The rate of detached rosette dehydration was the same for alx8 and wild type (Figure 2), as was reported for fry1 (Xiong et al., 2001) and similar to reports of another drought tolerant mutant over-expressing the transcription factor, AREB1ΔQT (Fujita et al., 2005). This suggests that the morphological changes such as different rosette shape, leaf shape and thickness in alx8 (Figure 6) are not required for its drought tolerance.
The alx8 plants were able to maintain constitutively higher RWC and leaf water potential than the wild type during drought at identical stress conditions, as indicated by similar pot water contents (Figure 2d). This is strong evidence that alx8 plants are not avoiding drought by taking up less water, rather they are able to maintain cell turgor at low soil water content for a longer period than the wild-type.
The drought tolerance of alx8 is not dependent on developmental stage or the size of the plants (Table 3 and Figure 3): that is, equivalently sized alx8 plants are more tolerant than Col-0 (Figure 3a,b). Finally, both SAL1 mutants lost fewer electrolytes than the wild types under high PEG concentration, suggesting there was a reduction of cellular damage under dehydration stress (Verslues et al., 2006) (Figure 3f). Although this response was unexpected based on previous results (Xiong et al., 2001), the reduced electrolyte leakage of young SAL1 mutants is consistent with all the soil-based experiments which show prolonged survival under water deficit conditions in soil.
Water use efficiency is the ratio of photosynthate produced to water used (Kirda et al., 1992; Masle et al., 2005) and is influenced by the relative rates of photosynthesis and transpiration; thus WUE need not indicate drought tolerance or vice versa (Martin et al., 1999). Gas exchange measurements indicated that alx8 had increased WUE during a short-term light response curve (Rossel et al., 2006). However, carbon isotope ratio determination indicated that WUE of well-watered alx8 and wild-type plants is similar (Figure 5e). The difference is possibly due to altered stomatal response times (data not shown) in the light response experiment (Rossel et al., 2006), which are not significant over the lifespan of the plant. Alternatively, more complex interactions between stomatal conductance and SAL1 may explain the differences. For example, there may be secondary variable effects of SAL1 on phytic acid or ABA-response pathways that alter stomatal conductance. With respect to this, genes encoding three isoforms of phospholipase D (PLD, At4g11830, At1g52570, At4g38560), which catalyzes the production of phosphatidic acid, thereby promoting closure, are up-regulated up to six-fold in alx8 (Table S1). Furthermore, the addition of putrescine can increase ABA in polyamine mutants (Cuevas et al., 2008). Further investigation into the role of SAL1 in stomatal signaling is required.
SAL1 regulates the metabolome of Arabidopsis, including the production of osmoprotectants
Increased polyamine levels due to over-expression of biosynthetic enzymes have previously been shown to induce tolerance to a range of abiotic stresses, including drought (Kurepa et al., 1998; Kasukabe et al., 2004). The level of the polyamine, putrescine, was 15.2-fold (P-value = 0.008) higher in alx8 leaves than in Col-0 leaves. Correspondingly, alx8 has a higher expression of the polyamine biosynthetic enzyme ADC2, which is normally up-regulated in response to osmotic stress (Perez-Amador et al., 2002; Urano et al., 2003). Constitutively high putrescine levels have been reported in a drought-tolerant wheat variety and in oxidative stress-tolerant variety of the weed Conyza bonariensis (Ye et al., 1997). Also, levels of putrescine in the stress-sensitive Arabidopsis mutant adc2-1 are only 25% of wild type (Urano et al., 2004). Although the exact role of putrescine in tolerance remains uncertain, it has been suggested to function as a direct or indirect antioxidant defense (Ye et al., 1997) and as a regulator of spermine and spermidine synthesis during drought (these provide an antisenescence effect at the whole-plant level), resulting in phenotypically normal plants (Capell et al., 2004). It would be of interest to decrease the production of putrescine in the alx8 background by crossing with the adc2-1 mutant or by treatment with the ADC inhibitor difluoromethylarginine (Xiong et al., 2006) in order to determine the contribution of putrescine and other polyamines to the drought tolerance of alx8 and fry1-1.
Levels of a number of sugars are also altered in alx8 leaf tissue, including large increases in a number of unidentified sugars (Table 6). These changes are inversely correlated with decreased accumulation of transitory starch in the chloroplasts of alx8 (Figure 8a,b). Sugars may play an important role in the osmotic stress response as osmoprotectants, protecting macromolecules and preventing membrane fusion (Bartels and Sunkar, 2005). The accumulation of sugars and osmoprotectants in alx8 and fry1-1 is likely to be a contributing factor to the higher water content and viability of SAL1 mutant leaves during drought. The reduced loss of electrolytes observed in SAL1 mutants during PEG treatment (Figure 3) also suggests a reduction of cellular damage under dehydration stress (Verslues et al., 2006). Osmoprotectants may limit membrane damage, or alternatively the accumulation of sugars could alter water potential, resulting in reduced water flow and electrolyte leakage.
With regard to the changes in sugar-related osmoprotectants, one of the more interesting findings was that the most highly studied of the sugar-based osmoprotectants were either unaltered or decreased in concentration in alx8 compared with the wild type. Specifically, trehalose, the accumulation of which has been shown to confer drought tolerance in transgenic Arabidopsis (Karim et al., 2007) and rice (Garg et al., 2002) plants, was unaltered in alx8 or fry1-1. Similarly, no changes were seen in the metabolite signal tentatively matched to myo-inositol (matched on the basis of mass-spectral similarity, abundance and approximate retention index match to MPIMP library, data not shown), a metabolite also implicated in osmoprotection and the biosynthesis of other osmoprotectants (Loewus and Murthy, 2000). Myo-inositol can be produced by the dephoshorylation of l-myo-inositol 1-phosphate (Loewus and Loewus, 1983). At least two myo-inositol 1-phosphate phosphatases (At1g31190 and At4g38690) were down-regulated in alx8. Perhaps more surprising was that metabolite and transcript data suggested a transcriptional down-regulation of the biosynthesis of raffinose family oligosaccharides (RFOs) – oligogalactosides of sucrose which have been implicated in desiccation tolerance of seeds (Taji et al., 2002b) and resistance to drought, cold and other stresses (Taji et al., 2002a; Nishizawa et al., 2008). Galactinol synthase (GolS) catalyzes the rate-limiting step in RFO biosynthesis: the synthesis of galactinol (the donor of galactosyl units for RFO extension) from UDP-glucose and myo-inositol (Karner et al., 2004). The drought-inducible isoform of GolS in Arabidopsis, AtGolS2 (At1g56600) (Taji et al., 2002b), was down-regulated 28- and 57-fold in alx8 and fry1-1, respectively. These decreases in AtGolS2 transcript abundance were correlated with 59% and 93% decreases in galactinol and 86% and 96% decreases in raffinose in alx8 and fry1-1, respectively. Thus, the alx8 drought-tolerant phenotype is likely to be independent of the production of galactinol and RFOs.
While we could find no evidence for increases in the most highly studied osmoprotectants, major accumulation was seen in a number of sugar-related compounds that could not be identified, even tentatively, by matching against public mass-spectral databases. These were the metabolites annotated as [Unknown Putative Disaccharide] eluting with retention indices of 2997.5 and 2915.4 (Table 6). Mass-spectral searching revealed a high degree of mass-spectral similarity to trehalose, suggesting that they may be unknown structural analogs of trehalose with similar properties conferring drought tolerance. Further structural elucidation work using complementary analytical technologies will therefore be required to determine the structures of these metabolites and, together with more targeted genetic experiments, may lead to the identification of novel osmoprotectant-producing pathways in Arabidopsis.
SAL1 regulates ABA-independent and ABA-dependent drought stress signaling pathways
The absence of SAL1 resulted in up-regulation of more than 1000 genes and down-regulation of 500–1000 genes under non-stressed growth conditions compared with wild-type plants. This role as a regulator of gene expression in both stress and development is reinforced by its negative role in post-transcriptional gene silencing (Gy et al., 2007) and the altered morphology and flowering time of SAL1 mutants. Just 7.6% of the 2447 genes differentially regulated in alx8 are classified as stress response genes and most of these are not inducible by ABA, which suggests not all stress response pathways are activated in alx8 under non-stressed growth conditions. For example, APX2, RAP2.6, sHSP and DREB2A are induced between 2- and 20-fold in non-stressed alx8 leaves, but high light stress (HL) results in 25- to 700-fold induction of the same genes (Rossel et al., 2006), demonstrating that other pathways that regulate these genes are stress-inducible and act synergistically with the SAL1 pathway. Second, the expression of RD22 is up-regulated by ABA (Yamaguchi-Shinozaki et al., 1992) and this induction is partially dependent on the transcription factor MYC2 (Abe et al., 1997, 2003). However, in alx8 despite an 8.2-fold increase in MYC2 mRNA, RD22 is down-regulated 3.7-fold. Likewise, there is no significant induction of ADH1 or KIN2/COR6.6, which are ABA-responsive and MYC2-regulated (Abe et al., 2003). This indicates a need for the interaction of multiple pathways in addition to the ABA-independent pathways constitutively activated in alx8 and fry1 for full activation of the stress response.
Of the known stress responsive transcription factors a small subset are significantly up-regulated in alx8 under non-stressed conditions, including two zinc finger transcription factors, ZAT10 and ZAT12. Both ZAT10 and ZAT12 are involved in the abiotic stress response pathways such as drought and high light (Rossel et al., 2002, 2007; Sakamoto et al., 2004; Davletova et al., 2005). Overexpression of ZAT10 induces expression of APX2 and 18% of the genes up-regulated in HL (Rossel et al., 2007). Correspondingly, 24.6% of the genes up-regulated by HL (Rossel et al., 2007) are also up-regulated in alx8, further emphasizing the overlap between drought and light stress response networks (Kilian et al., 2007). High levels of expression of SAL1 in the vascular tissue (Xiong et al., 2001) suggest a role in the transduction of signals during drought stress. This localization correlates with HL-induced H2O2 production, APX2 and ZAT10 expression, and systemic acquired acclimation in the bundle sheath (Karpinski et al., 1997, 1999; Fryer et al., 2003; Rossel et al., 2007).
Drought induces an increase in ABA, and there is typically a corresponding increase in expression of the rate-limiting ABA biosynthetic enzymes NCED3 and NCED1 (Iuchi et al., 2001). However, these genes are not up-regulated in alx8, nor is there down-regulation of genes involved in ABA catabolism; in fact, the catabolic enzymes CYP707A1 and CYP707A3 are up-regulated (Figure S4). Hence, the increase of ABA content in alx8 (Rossel et al., 2006) does not appear to be transcriptionally regulated. Interesting questions that remain to be answered are, as mentioned above: how does ABA change during drought in alx8 and how does SAL1 regulate stomatal signaling?
The majority of the up-regulated (66%) and down-regulated (53%) genes in fry1-1 were co-expressed in alx8 compared with wild-type. However, there were still differences in gene expression and metabolic profile between alx8 and fry1-1, which is interesting given they are both loss-of-function mutations in the same gene. C24 and Col-0 had unique, although similar, metabolite profiles and thus it is possible that ecotype differences subtly alter the role of SAL1 and the effects of its loss in each ecotype.
What are the substrate(s) of SAL1 in vivo?
SAL1 has been proposed to function in PI metabolism based on its high sequence homology with yeast inositol phosphatases, MET22/HAL2 (Glaser et al., 1993) and previous enzymatic assays of recombinant proteins (Murguia et al., 1995; Quintero et al., 1996; Xiong et al., 2001). SAL1 has been reported to dephosphorylate PAP, implicating it in sulfate metabolism and post-transcriptional gene silencing (Xiong et al., 2001; Gy et al., 2007). Furthermore, any actual role for IP3 in abiotic stress signaling in plants has been debated (Perera et al., 2008). Thus, a consideration of the possible substrate(s) for SAL1 in vivo and how the substrate(s) affect overall cellular metabolism is needed.
Mature fry1-1 rosette plants have higher basal levels of IP3 compared with wild-type plants and an altered IP3 response upon ABA treatment (Xiong et al., 2001). However, the activity of SAL1 against IP3 and Ins(1,4)P is 4% and 34%, respectively, compared with that against PAP and 3′-phosphoadenosine-5′-phosphosulfate (PAPS) (Murguia et al., 1995; Quintero et al., 1996; Xiong et al., 2001). This could be used to circumstantially argue against IP3 being the primary substrate of SAL1 in vivo. If that were the case, the reported increase in IP3 could be a secondary effect of loss of SAL1. A proposal for this to occur via substrate feedback effects has been made (Gunesekera et al., 2007). For example, SAL1 could result in changes in other phosphatases, such as the 5Ptases. However, we did not observe any change in the transcript abundance in the microarrays for any of the 5Ptases. Alternatively, polyamines, specifically spermine and spermidine, increase the IP3 pool in animal systems by stimulating PLC (Hedeskov et al., 1991; Spath et al., 1991), PIP 5-kinase (that catalyses the production of the IP3 precursor Ins(4,5)P) (Singh et al., 1995) and decreasing the enzymatic activities of IP3 5-phosphatase (Seyfred et al., 1984). Additionally, as alx8 plants have 15-fold higher levels of putrescine this could potentially alter PI pools.
3′-Phosphoadenosine-5′-phosphate and PAPS are involved in sulfate assimilation and transfer of sulfate groups (Saito, 2004). But SAL1 may be dispensable for sulfate metabolism as SAL1 mutants grow in hydroponics with sulfate as the only source of sulfur (data not shown). 3′-Phosphoadenosine-5′-phosphate inhibits exoribonuclease (XRN) activity (Dichtl et al., 1997) and implicates SAL1 as a negative regulator of PTGS (Gy et al., 2007). If this were the case, message levels of the target genes of XRNs in alx8 may be expected to be up-regulated. Two of 14 targets of XRN4 were found to be up-regulated in the microarrays in similar ways (two-fold) in both alx8 and xrn4-5 mutants (SI in Table S1; Souret et al., 2004).
In summary, a careful and detailed evaluation of the synthesis and accumulation of PAP, PAPS and IP3 and other potential substrates of SAL1 under a range of developmental and stress treatments in wild type, alx8 and other relevant germplasm, such as other SAL homologs that have different relative specificities for PAP and IP (Gil-Mascarell et al., 1999), is required to determine the actual in vivo substrate of SAL1 and which compound is regulating drought and high light-stress signaling.
The SAL1 mutants alx8 and fry1 survive prolonged drought for 40–50% longer than wild-type plants. Whilst development is altered in these mutants this effect is temporary, as by the later stages of development the biomass of alx8 and fry1-1 rosettes is similar to their respective wild types. We hypothesize that the SAL1 protein has a key role in the negative regulation of pathways controlling morphological, physiological and molecular changes whose activation results in enhanced induction of stress networks. The evident health and growth of alx8 leaves during drought is indicative of enhanced tolerance, rather than avoidance or escape responses. Under non-stressed conditions stress response genes such as the antioxidant APX2 are constitutively up-regulated, amounts of osmoprotectants such as polyamines and sugars are elevated and ABA is accumulated. Thus, SAL1 negatively regulates drought tolerance.
Plant growth and drought stress
Arabidopsis plants were grown on metro mix soil (35% Canadian Peat Moss, 19% Perlite 500, 46% vermiculite, 1.5 g l−1 lime), vernalized for 72–96 h and fertilized with 0.5 × Hoagland’s solution once a fortnight (Hoaglands and Arnon, 1950). Growth conditions were either a light (150 μmol photons m−2 sec−1)/dark cycle of 12/12 h at 21 ± 2°C and 55% relative humidity or 8/16 h for Figures 3d, 6b–d and 7 or 16/8 h for Figure 3f. All seed was obtained from the Arabidopsis Biological Resource Center (ABRC) except alx8 (Rossel et al., 2004).
Drought treatments were applied to plants in soil with sufficient water by withholding further watering. Viability was determined by measuring leaf and soil relative water content, visual assessment, chlorophyll fluorescence and recovery after rewatering. See Appendix S1 for details.
Positional cloning and complementation
Linkage analysis was performed with the F2 progeny of a cross between alx8 and Landsberg erecta as described by Lukowitz et al. (2000). Further markers were designed around simple sequence length polymorphisms (SSLPs) and single-nucleotide polymorphisms (SNPs) found in the Cereon database (http://www.arabidopsis.org/; details available upon request). Derived cleaved amplified polymorphic sequences (dCAPS) primers were designed using dCAPS Finder 2.0 (http://helix.wustl.edu/dcaps/dcaps.html; Neff et al., 2002). For sequencing, 600-bp fragments were amplified using Phusion Taq polymerase (Finnzyme, http://www.neb.com/), A-tailed and cloned into pGem T-easy (Promega, http://www.promega.com/) as per the manufacturer’s instructions.
For complementation the T8H11 BAC was obtained from ABRC. SAL1 and 1.7 kb of the upstream genomic sequence was digested from the BAC using PstI, ligated into the binary vector pCAM2300 (CAMBIA, http://www.cambia.org/) and transformed into Arabidopsis.
SAL1 protein analysis
To produce recombinant SAL1 protein, total RNA was extracted from Col-0 and alx8 leaves using a Plant RNeasy Kit with on-column DNAse digestion (Qiagen, http://www.qiagen.com/) and used for first-strand cDNA synthesis (SuperScript II, Invitrogen, http://www.invitrogen.com/). The complete SAL1 coding sequence was amplified from the cloned Col-0 and alx8 SAL1 cDNAs with the primers SacII-SAL1 F1 (5′-CTCCGCGGTGGTATGGCTTACGAGAAAGAGC-3′) and EcoRI-SAL1 (5′-GCTCGAATTCTCAGAGAGCTGAAGCTTTCTC-3′), then cloned into the pHUE vector (Baker et al., 2005). Recombinant proteins were expressed in E. coli strain BL21(DE3) (Novagen, http://www.emdbiosciences.com/) after induction with 1 mm isopropyl β-d-1-thiogalactopyranoside (IPTG) and purified by affinity chromatography using His-Bind resin according to the manufacturer’s instructions (Novagen). SAL1 protein without the tag was assayed for phosphatase activity against PAP (Murguia et al., 1995). The anti-SAL1 rabbit polyclonal antibodies (IMVS, Adelaide, http://www.imvs.sa.gov.au/) were isolated from the IgG fraction by immunoaffinity purification. For western blots, 20 μg of leaf protein was resolved on a 4–12% gradient gel, electrotransferred to a nitrocellulose membrane, and probed with a 1:1000 dilution of purified anti-SAL1 antibodies. After washes with 0.05% (v/v) Tween PBS, blots were incubated with a 1:10 000 dilution of HRP-conjugated goat anti-rabbit IgG and developed using the SuperSignal West Femto Chemiluminescent detection kit (Pierce, http://www.piercenet.com/). See Appendix S1 for detailed protocols.
Gene expression analysis
Total RNA was extracted from mature green leaves from 5-week old soil grown plants 2 h after ‘dawn’ using the Plant RNeasy Kit as described. The quality and quantity of RNA was tested by gel electrophoresis, BioAnalyzer (Agilent, http://www.chem.agilent.com/) and/or spectrophotometry. For qRT-PCR 1 μg of RNA was reverse transcribed to make cDNA using T23V primer and SuperScriptII (Invitrogen). Two technical replicates for each of the two reverse transcription reactions were done using LightCycler480 SYBR Master (Roche, http://www.roche-applied-science.com/) and a Rotorgene3000 (Corbett Research, http://www.corbettlifesciences.com/). Results were analyzed by comparative quantification with each set of replicates averaged and normalized against a reference gene (see Appendix S1, Czechowski et al., 2005; Schmittgen and Zakrajsek, 2000) and the control (Warton et al., 2004). Primers used are listed in Appendix S1.
Microarrays were performed and analyzed using Affymetrix GeneChip® Arabidopsis Genome ATH1 Arrays (Affymetrix, http://www.affymetrix.com/), MAS5 normalization, statistical analysis and false discovery rate correction as described (Rossel et al., 2007). Three biological replicates were analyzed for each genotype. All microarray data have been deposited in the ArrayExpress database (http://www.ebi.ac.uk/microarray-as/ae/) under the accession E-MEXP-1495.
The abaxial surface of leaves was imaged by cryogenic scanning electron microscopy and thin cross-sections of resin-embedded leaf material were used for transmission electron microscopy and light microscopy. See Appendix S1 for details.
Metabolites from whole leaves were extracted and derivatized using a method modified from Roessner-Tunali et al. (2003). Derivatized metabolite samples were analyzed on a GC/MSD system (Agilent Technologies). Data files were processed using in-house MetaMiner software to carry out all peak detection, quantification, library matching, normalization, statistical analysis and data visualization. Detailed methodology is presented in Appendix S1. Analyte abundance matrices were exported from MetaMiner and chemical artifacts, internal standards and metabolite signals that were not detected in at least 80% of the replicates from at least one sample class (genotype) were removed from the dataset prior to PCA. Principal components analysis was carried out with Avadis Prophetic (Version 4.3, Strand Life Sciences, http://www.strandgenomics.com/) software with mean centering and scaling of all variables to unit variance. The input data matrix is supplied in Appendix S2.
Morphological and physiological methods
Total water potential was measured using a custom-built thermocouple psychrometer (Morgan, 1991). Leaf disks were placed into equilibration chambers for 4 h at 22°C. A Peltier cooling current was passed through the thermocouple and the electromotive force (e.m.f.) read as a needle deflection in a microvoltmeter (HR 33 Dew Point, Wescor, http://www.wescor.com/environmental/). The leaf water potential (MPa) was calculated by interpolation of the e.m.f. to a standard curve. Soil water potential was measured using a pressure plate apparatus as in Klute (1986).
Rosette leaves from the same plants were excised, the fresh weight (Fw) recorded and incubated in water for at least 4 h at 4°C in the dark. The leaves were blotted and the turgid weight (Tw) measured. Finally, leaves were dried at 80°C overnight and weighed to determine the dry weight (Dw). The relative water content (RWC) was calculated as (Jones, 2007)
For Figure 3d, plants were cultivated under 8/16 h light (100 μmol photons m−2 sec−1)/dark cycle at 21 ± 2°C and 55% relative humidity for 30 days before initiating drought treatment. Rosette size estimates were derived from fluorescence data at day 0 of drought treatment, and were plotted against the eventual drought survival times determined for each specimen. Fluorescence measurements were obtained using a CF Imager (Technologica, http://www.technologica.co.uk/).
To calculate the 13C isotope ratio (δC13), leaves from 4–5-week-old plants were harvested, dried and ground. The amount of carbon present was then measured by mass spectrometry and the carbon isotope ratio was calculated as described in Farquhar and Richards (1984).
For epidermal leaf peels, impressions of the leaf surfaces of 5-week-old plants were made using surface activated dental putty (Coltene Whaledent, http://www.coltenewhaledent.biz/). Clear cellulose varnish was applied to the impressions, mounted onto slides and viewed through a light microscope (Leica DMLB, http://www.leica-microsystems.com/). The stomatal index (SI) was calculated as SI = [S/(E + S)] × 100 where S is the number of stomata and E is the number of epidermal cells. For starch analysis, the chlorophyll was removed by boiling in ethanol and leaves were stained in Lugol’s solution for 5 min [5% (g/v) iodine, 10% (g/v) potassium iodide], then destained for 1 h in water. A precision micrometer (Mitutoyo, http://www.mitutoyo.com/) was used to measure leaf thickness.
Research was supported by the ARC Centre for Plant Energy Biology (CE0561495) and a Meat and Livestock Australia Postgraduate Award (PBW). JAL is funded by a Royal Society Research Fellowship, KJF by a University of Sheffield Scholarship, AJC by a GRDC PhD Scholarship and AHM by an ARC Australian Professorial Fellow (DP0771156).We are grateful for the assistance of Jan Bart Rossel in identifying alx8 and establishing our interest in it; David Deery, CSIRO, for soil water potential measurements; Lily Shen and Dr Cheng Huang at ANU Electron Microscopy Unit; Drs Rana Munns and Grant Cramer for discussions; Drs Rohan Baker (JCSMR, ANU) and Spencer Whitney (RSBS, ANU) for providing the pHUE vector; Drs Tony Condon and Richard Richards (Plant Industry, CSIRO) for help with leaf water potential measurements and Stephen Graham for assistance with protein analysis.