Changes in gene expression form a key component of the molecular mechanisms by which plants adapt and respond to environmental stresses. There is compelling evidence for the role of stimulus-specific Ca2+ signatures in plant stress responses. However, our understanding of how they orchestrate the differential expression of stress-induced genes remains fragmentary. We have undertaken a global study of changes in the Arabidopsis transcriptome induced by the pollutant ozone in order to establish a robust transcriptional response against which to test the ability of Ca2+ signatures to encode stimulus-specific transcriptional information. We show that the expression of a set of co-regulated ozone-induced genes is Ca2+-dependent and that abolition of the ozone-induced Ca2+ signature inhibits the induction of these genes by ozone. No induction of this set of ozone-regulated genes was observed in response to H2O2, one of the reactive oxygen species (ROS) generated by ozone, or cold stress, which also generates ROS, both of which stimulate changes in [Ca2+]cyt. These data establish unequivocally that the Ca2+-dependent changes in gene expression observed in response to ozone are not simply a consequence of an ROS-induced increase in [Ca2+]cytper se. The magnitude and temporal dynamics of the ozone, H2O2, and cold Ca2+ signatures all differ markedly. This finding is consistent with the hypothesis that stimulus-specific transcriptional information can be encoded in the spatiotemporal dynamics of complex Ca2+ signals in plants.
Ozone pollution is recognized as being a significant threat to food production across North America, Europe and Asia resulting in the loss of billions of dollars’ worth of crops globally (Murphy et al., 1999; Ashmore et al., 2006; Emberson et al., 2009) with the annual loss of wheat, rice, maize, and soybean estimated to be US$14–26 billion in the year 2000 (Van Dingenen et al., 2009). Tropospheric ozone levels are predicted to rise by 20–25% between 2015 and 2050 and to further increase by 40–60% by 2100 if current emission trends continue (IPCC, 2007). Recent studies have shown high mean ozone concentrations across Asia (Emberson et al., 2009) with peak ozone concentrations in excess of 300 ppb recorded in China (Wang et al., 2006). Consequently, these economic losses together with the resultant threat to global food security are of increasing concern, particularly in regions of the world where the expanding economy has led to an increased emission of ozone precursors into the atmosphere. A detailed understanding of the molecular mechanisms by which plants respond to ozone is, therefore, essential in order to generate new crop varieties that are capable of growing under conditions of increased ozone pollution and thereby contribute significantly to global food security.
Calcium (Ca2+) is probably the most important second messenger in plants and is a key component in the signalling network by which plant cells respond to a diverse range of developmental and environmental signals (Dodd et al., 2010; Kudla et al., 2010; Reddy et al., 2011). An increase in cytosolic free calcium concentration ([Ca2+]cyt) is one of the earliest-observed cellular responses to ozone (Clayton et al., 1999; Evans et al., 2005) and has been shown to be essential for the induction of the gene that encodes the antioxidant enzyme glutathione-S-transferase (GST) tau 5 (Clayton et al., 1999). However, the ubiquity of Ca2+-signalling in plants raises fundamental questions regarding how response specificity is maintained. Stimulus-induced changes in plant [Ca2+]cyt take the form of transients, spikes and oscillations in [Ca2+]cyt (McAinsh et al., 1995; Allen et al., 2001; Kosuta et al., 2008). This situation has given rise to the ‘Ca2+ signature’ hypothesis – the generation of a stimulus-specific change in [Ca2+]cyt that is characteristic for a particular stress and in which information about the nature and magnitude of the stress is encoded in the spatiotemporal dynamics of the [Ca2+]cyt changes (McAinsh and Hetherington, 1998; McAinsh and Pittman, 2009). Studies of Ca2+ signalling in stomatal guard cells and symbiosis signalling in legumes provide compelling evidence that signalling information can be encoded in the frequency, amplitude, and shape of plant Ca2+ signatures (for example, McAinsh et al., 1995; Allen et al., 2001; Kosuta et al., 2008).
Several studies have implicated Ca2+ signalling in the transcriptional regulation of stress-induced genes, for example in response to touch and cold (Knight et al., 1996; Polisensky and Braam, 1996; Galon et al., 2010). Transcriptome analysis of changes in gene expression induced by artificially imposed increases in [Ca2+]cyt has also revealed 363 Ca2+-responsive genes in Arabidopsis that include genes know to be induced by abiotic stresses, which include cold, osmotic, or oxidative stress (Whalley et al., 2011). Four motifs have been identified in the promoters of these Ca2+-responsive genes, which are regulated by [Ca2+]cytin planta and for which Ca2+ is both necessary and sufficient for expression via these motifs; the C-Repeat/Drought-Responsive Element (CRT/DRE), the Site II motif, the binding site of the CaM-binding (CAMTA) transcription factor (the CAM box), and the ABA-responsive element (ABRE) (Kaplan et al., 2006; Whalley et al., 2011). The number of genes up- or downregulated by Ca2+ was dependent on the kinetics of the artificially imposed [Ca2+]cyt increase with oscillations in [Ca2+]cyt, which induced the expression of 25 times more genes than a prolonged increase in [Ca2+]cyt (Whalley et al., 2011). This finding raises the possibility that signalling information that is required to direct the expression of Ca2+-responsive genes may be encrypted in the temporal dynamics of stress-induced increases in [Ca2+]cyt.
Our previous studies raised the intriguing possibility of an ozone Ca2+ signature that encodes the signalling information necessary to coordinate an ozone-specific transcriptional response (Clayton et al., 1999; Evans et al., 2005). However, there are many other environmental stresses that result in the accumulation of ROS and that also induce an increase in [Ca2+]cyt (Jaspers and Kangasjarvi, 2010). Therefore, in order to demonstrate unequivocally that the ozone Ca2+ signature encodes ozone-specific signalling information, it is important to first demonstrate that the changes in [Ca2+]cyt observed are specific to the ozone stimulus and that the downstream responses it affects do not take place in response to other Ca2+ signatures and, in particular, to those signatures generated by other oxidative stresses and ROS. Here we show that the Ca2+-dependent changes in gene expression observed in response to ozone are not simply a consequence of a ROS-induced increase in [Ca2+]cytper se. The increases in [Ca2+]cyt observed in response to H2O2, one of the ROS generated by ozone (Mehlhorn et al., 1990), and cold stress, which also generates ROS (O’Kane et al., 1996), neither of which induce the expression of a set of co-regulated ozone-induced genes, differ markedly to that of ozone in both their magnitude and temporal dynamics. These data are consistent with the hypothesis that stimulus-specific transcriptional information can be encoded in the spatiotemporal dynamics of complex Ca2+ signals in plants.
Ozone induces three key functional groups of genes
We used microarray analysis to investigate the effect of acute ozone stress (500 ppb ozone for 6 h) on gene expression in 10-day-old Arabidopsis Col-0 seedlings. Under these conditions ozone induces the formation of necrotic lesions in approximately 50% of seedlings and enhanced ion leakage, but does not lead to long-term damage or impairment of growth (Figure S1). Analysis of the 3552 probe sets that remained after filtering to remove low-expressed and invariant genes identified 1788 sets that showed a statistically significant (paired t-test, P <0.05) difference between control and ozone-treated groups (log2 expression ratios >1.0 with a false discovery rate of 0.1%). Approximately 76% (1364) of these sets showed increased expression in response to ozone whereas 24% (424) showed a decrease. Real-time reverse transcription polymerase chain reaction (RT-PCR) analysis of the expression of the 21 most highly ozone-responsive genes (20 upregulated and 1 downregulated) showed the same response to ozone as that observed in the microarray experiment in all cases, with the majority of genes having similar relative expression values determined by both methods (Figure 1).
Next, the probable identity of the cellular processes affected by acute ozone stress was investigated using MapMan (Thimm et al., 2004). Protein turnover, vesicle transport and calcium signalling were amongst the most prominent of the functionally related groups that showed a significant (P <0.01) response to ozone (Table 1). As vesicle transport was not a process that might obviously have been associated with acute ozone stress, we investigated this group further. MapMan bin 31.4 (vesicle transport) includes 120 Affymetrix probe IDs of which 12 correspond to transcripts upregulated by ozone. These 12 Affymetrix probes were used as seeds to generate a co-expression network that contained 2344 genes (Figure 2), of which 32% exhibited a mean log2 fold-change >1 in response to ozone in the microarray experiment. Further exploration of this network identified two significant highly interconnected regions that function as sub-clusters within the network (Figure 3). Of these regions, the first cluster, centred on the genes SYP121, SYP122 and SNAP33 (SNP11), is perhaps the most interesting (Figure 3a). It contains 86 ozone upregulated genes (mean log2 fold-change >1) out of a total of 99 genes (87%) in the cluster. SYP121, SYP122 (syntaxins) and SNAP33 (Qb+c SNARE) are all members of the Arabidopsis SNARE (soluble N-ethylmaleimide sensitive factor attachment protein receptor) superfamily, which function in vesicle membrane fusion (Grefen and Blatt, 2008). SYP121, SYP122 and SNAP33 are localised to exocytotic vesicles and the plasma membrane, and are generally believed to be involved in exocytosis. SYP121, the first plasma membrane-localised SNARE (originally named SYR1) to be described in plants (Leyman et al., 1999), and SNAP33 interact physically in vivo, and form complexes with additional vesicle-associated membrane proteins (VAMPs), which included VAMP721, VAMP722 and VAMP727. VAMP722 and VAMP727 are also upregulated in response to ozone and form hubs in the main network. SNAP33 also interacts physically with another syntaxin, SYP111 (KNOLLE), which functions along with the protein SEC11 (KEULE) in vesicle fusion during cytokinesis (Heese et al., 2001). SEC11 forms a hub in the second cluster identified within our co-expression network (Figure 3b), although this network includes fewer ozone-regulated genes from the microarray experiment (55 out of a total of 133 genes (41%) in the cluster).
Table 1. Protein turnover, vesicle transport and Ca2+ signalling are amongst the most significantly over-represented functional groups of genes that are altered transcriptionally by ozone treatment
MapMan Wilcoxon rank sum statistics (Benjamini–Hochberg corrected) filtered for P < 0.01 and sorted by most significantly over-represented higher level categories.
not assigned.no ontology.pentatricopeptide (PPR) repeat-containing protein
protein.postranslational modification.kinase.receptor like cytoplasmatic kinase VII
protein.synthesis.misc ribososomal protein
Cell. vesicle transport
signalling.receptor kinases.leucine rich repeat III
signalling.receptor kinases.leucine rich repeat VII
RNA.regulation of transcription.NAC domain transcription factor family
RNA.regulation of transcription.WRKY domain transcription factor family
RNA.regulation of transcription.SNF7
Nucleotide metabolism.deoxynucleotide metabolism
Major CHO metabolism.synthesis
major CHO metabolism.synthesis.starch
Transport.Major Intrinsic Proteins.TIP
The expression of ozone-induced genes exhibits parallel temporal dynamics and concentration dependence
Next, five of the most highly ozone-responsive genes were selected for a more detailed analysis of the regulation of gene expression by ozone. The SYP121/SYP122/SNAP33 cluster includes the gene AtNUDT7 (At4g12720), a nudix hydrolase that has been shown previously to be upregulated by ozone (Jambunathan and Mahalingam, 2006) and that functions in oxidative signalling (Ishikawa et al., 2009; Jambunathan et al., 2010). NUDT7 negatively regulates EDS1 function, which in turn is positively regulated by FMO1 (Bartsch et al., 2006). NUDT7, EDS1 and FMO1 are all upregulated by ozone. AtHIR2 (At3g01290), which encodes a band 7 family protein (members of which are commonly associated with lipid rafts, which are also involved in exo- and endocytosis; Salaun et al., 2004), is tightly clustered with NUDT7 in both the full vesicle transport network (Figure 2) and the SYP121 cluster (Figure 3a). NUDT7 and AtHIR2 were, therefore, selected for further study, along with three additional genes from the full network implicated in ROS responses; At5g27760 (hypoxia-responsive protein), At3g12740/ALIS1 (involved in Golgi vesicle transport; Poulsen et al., 2008) and At4g02380 (AtLEA5/SAG21). ROS production has been demonstrated to increase upon hypoxia (Blokhina and Fagerstedt, 2010) and has been associated with triggering endocytotic events in plants (Leborgne-Castel et al., 2008), whilst the late embryogenesis abundant (LEA)-3 group protein LEA5/SAG21 has been shown to confer oxidative stress tolerance to yeast (Mowla et al., 2006).
We observed that the transcripts of all five genes exhibited a similar dose–response to ozone with comparable temporal dynamics (Figure 4). There was no significant change in transcript abundance in response to 100 ppb ozone. In contrast, 300 ppb ozone induced a significant increase in transcript abundance that reached peak levels within 3–6 h. AtNUDT7 showed the largest increase (15-fold) in transcript abundance. This result compares with increases of 4-, 10-, 9-, and 6-fold, respectively, for ALIS1, AtLEA5/SAG21, AtHIR2, and At5g27760. Subsequently, the transcripts of all five genes decreased rapidly and returned to their original levels within 15 h post-ozone treatment. The peak in transcript abundance in response to 500 ppb ozone occurred earlier (within 1 h) and at a higher level (9- to 22-fold increase) than in response to 300 ppb. The only exception was AtHIR2, which exhibited a similar response to both 300 and 500 ppb ozone. AtNUDT7 was again the most highly induced gene and displayed maximum expression after 1 h. The parallel transcript pattern of these genes suggests that they may be co-regulated by ozone. Examination of microarray data sets available publically indicates that these five genes are also co-regulated by other stress conditions, notably in response to pathogen infection, and by some abiotic stresses (Figure S2).
The W-box motif is present in ozone-induced genes and is over-represented in the SYP121/SYP122/SNAP33 cluster
The high level of similarity in the temporal dynamics of AtHIR2, At5g27760, NUDT7, ALIS1, and AtLEA5/SAG21 expression observed in response to ozone suggests the potential for co-regulation of ozone-induced genes at the level of transcription. We used the ‘Visualization’ tool at the Athena website (O’Connor et al., 2005) to search for known transcription factor-binding sites in the 750-bp upstream regions of genes in the different groups identified by network analysis. The W-box motif (a binding site for WRKY transcription factors) was strongly over-represented (P <10−8) in the genes present in the SYP121/SYP122/SNAP33 cluster (Figure 3a) with 159 occurrences of the element present in 80 of the 98 promoters analyzed. Furthermore, AtHIR2, At5g27760, NUDT7, ALIS1, and AtLEA5/SAG21 all contained at least one W-box motif. Five genes that encode WRKY transcription factors were also present in this cluster. No other motif showed statistically significant over- or under-representation in the SYP121/SYP122/SNAP33 cluster.
An increase in [Ca2+]cyt is essential for the induction of gene expression by ozone
We have shown that Ca2+ is an important component of the signalling pathway by which plants respond to ozone (Clayton et al., 1999; Evans et al., 2005) and that an increase in [Ca2+]cyt is essential for the induction of the gene for the antioxidant defence enzyme GST tau 5 (GSTU5) by ozone (Clayton et al., 1999). Ca2+ signalling is strongly over-represented (P =4.77 × 10−8) in our MapMan analysis of ozone-responsive genes (Table 1). MapMan bin 30.3 (Ca2+ signalling), includes 209 Affymetrix probes, of which 52 are up- and six down-regulated by ozone (Table S1). Consequently, we investigated the role of Ca2+ signalling in the co-expression of our five ozone-induced genes.
Ozone (300 ppb and 500 ppb) induced a characteristic biphasic ozone-Ca2+ signature in seedlings that expressed the Ca2+ sensor aequorin. This signature consisted of a short-lived, spike-like increase in [Ca2+]cyt followed by a smaller, more prolonged elevation (Figure 5a). This finding is consistent with previous reports of ozone-induced increases in [Ca2+]cyt (Clayton et al., 1999; Evans et al., 2005). The total increase in [Ca2+]cyt (Figure 5c), together with the size of the first (Figure 5d) and second (Figure 5e) phase of the increase, varied significantly with ozone concentration (300 < 500 ppb). The timing of the second phase also varied significantly between 300 and 500 ppb ozone peaking after 1643 ± 83 sec and 947 ± 157 sec, respectively. LaCl3, which is known to inhibit plasma membrane Ca2+ channels (Knight et al., 1996; Figure 5b), caused a significant reduction in the magnitude of the first phase of the 300 and 500 ppb ozone-Ca2+ signature (Figure 5d) and implied a role for Ca2+ influx in the generation of this phase of the response. The second phase of the 300 ppb ozone-induced increase in [Ca2+]cyt was completely abolished by LaCl3 (Figure 5e) and significantly reduced at 500 ppb ozone.
Our previous studies suggested that it is the second phase of the ozone-Ca2+ signature that encodes the signalling information that is necessary to direct ozone-induced gene expression (Clayton et al., 1999). Therefore, next, we examined whether the ozone-induced increase in [Ca2+]cyt is required for the co-expression of our set of five ozone-induced genes. For this experiment, we focussed on the response to 300 ppb ozone due to the consistent effect of LaCl3 on the second phase of the 300 ppb, but not the 500 ppb, ozone-Ca2+ signature (Figure 5). LaCl3 inhibited the induction of AtHIR2, At5g27760, NUDT7, ALIS1, and AtLEA5/SAG21, together with the gene for the antioxidant defence enzyme GSTU24, by ozone (Figure 6). There was no significant difference in relative transcript abundance between inhibitor-treated ozone-exposed seedlings and water-treated air-exposed controls, which indicated a complete abolition of ozone-dependent gene expression. These data provide strong evidence that the ozone-induced increase in [Ca2+]cyt is required for the induction of the set of five genes by ozone.
The induction of ozone-induced gene expression is not a general response to ROS-induced [Ca2+]cyt increases
Many stimuli that induce an increase in [Ca2+]cyt also result in the accumulation of ROS and oxidative stress (Jaspers and Kangasjarvi, 2010). Consequently, in order to test whether the Ca2+-dependent induction of AtHIR2, At5g27760, AtNUDT7, ALIS1, AtLEA5/SAG21 was simply a response to a ROS-induced increase in [Ca2+]cytper se, the ability of two ROS-inducing stimuli, H2O2 (Mehlhorn et al., 1990) and cold (O’Kane et al., 1996), to induce the expression of our set of five co-regulated ozone-induced genes was examined. H2O2 (Figure 7a) and cold (Figure 8a) both induced a biphasic increase in [Ca2+]cyt, which was consistent with previous reports of H2O2- and cold-induced increases in [Ca2+]cyt (Knight et al., 1996; Rentel and Knight, 2004). However, the kinetics of the H2O2- (Figure 7b) and cold- (Figure 8b) Ca2+ signatures differed markedly from that of ozone (Figure 5c–e). The total [Ca2+]cyt elevation over 1 h induced by H2O2, cold, and ozone differed significantly (269 ± 17, 511 ± 24, and 207 ± 15 μm, respectively) as did the magnitude of the individual phases of the response. The peak of the first phase of the cold-Ca2+ signature was approximately five times higher (1.20 ± 0.06 μm) than that of the H2O2- and ozone-induced [Ca2+]cyt increase (0.22 ± 0.06 and 0.15 ± 0.01 μm, respectively), although there was no difference in their timing (Figure 9). The second phase of the cold-Ca2+ signature was well defined; [Ca2+]cyt increased rapidly and peaked at 0.22 ± 0.01 μm after 298 ± 19 sec. In contrast, the second phase of the H2O2-Ca2+ signature was poorly delimited, peaking at 0.11 ± 0.01 μm after 555 ± 36 sec, whilst ozone induced a more gradual increase that, although smaller (0.07 ± 0.01 μm), was considerably longer in duration peaking at 1643 ± 83 sec (Figure 9).
Genes that encode the H2O2-induced antioxidant enzyme GSTU5 (Figure 7c) and the cold-regulated transcription factor CBF2 (Figure 8c) were induced significantly by H2O2 (3.3- to 3.7-fold) and cold (>200-fold), respectively, to levels comparable with those results reported previously (Medina et al., 1999; Rentel and Knight, 2004). In contrast, treatment with H2O2 or cold for either 3 h or 6 h had little effect on the expression of the ozone-induced genes; only slight (≤2-fold) increases in the expression of At5g27760, NUDT7, ALIS1 and AtLEA5/SAG21 (≤2-fold; Figure 7c) and NUDT7 (1.4-fold; Figure 8c) were detected in response to treatment with H2O2 or cold for 3 h, respectively. This result compares with a 4- to 15-fold induction after 3 h and a 6.5- to 15-fold induction after 6 h in response to ozone (Figure 2).
A gene expression module implicated in vesicle transport regulated by ozone and other stresses
Several authors have reported previously on the effects of ozone on the transcriptome of Arabidopsis (e.g. Miyazaki et al., 2004; Mahalingam et al., 2005; Tosti et al., 2006; Ludwikow and Sadowski, 2008; Blomster et al., 2011). These studies have identified various biological processes and signalling pathways that are altered under ozone stress (reviewed by Ludwikow and Sadowski, 2008), and many of these processes, such as cell wall modification, photosynthesis, anthocyanin metabolism, protein turnover and jasmonate and ethylene signalling are also revealed by our analyses. We also identified genes involved in vesicle transport as affected significantly by ozone in our experiments. As far as we are aware, this association has not been observed before. We focussed on this group of genes to derive a network of stress-responsive genes in which we identified highly interconnected clusters of ozone-regulated genes (Figure 3). One of these clusters is centred on the genes SYP121, SYP122 and SNAP33. SYP121-SNAP33-VAMP722 SNARE complexes appear to be important for exocytotic secretion events, which form part of a non-host resistance response during Arabidopsis–powdery mildew interactions, and it has been suggested that they may be involved in the secretion of antimicrobial secondary metabolites (Kwon et al., 2008). SYP122 is partially redundant with SYP121 and is likely to play a similar role. Humphry et al. (2010) recently used a bioinformatics approach to identify genes that are co-expressed with SYP121, SNAP33 and VAMP722, and that are important for powdery mildew resistance. The resulting collection of genes overlaps substantially (>50%) with those present in the ozone-regulated SYP121/SYP122/SNAP33 cluster identified here, which suggests that a similar network of genes may perform similar functions during different stress responses, including both ozone and pathogen defence. Interestingly, there is evidence that part of the secretory activity of the SYP121/SYP122/SNAP33/VAMP722 complex during fungal infection may be callose deposition (Shimada et al., 2006). The production of callose is also a well characterised response to ozone (e.g. Schraudner et al., 1992; Bussotti et al., 2005).
The expression of co-regulated ozone-responsive genes is dependent on the ozone Ca2+ signature
From the co-expression network constructed around ozone-responsive vesicle transport genes, we selected five for further analysis. Our data show that expression of these genes is co-regulated in response to ozone and that an ozone-induced change in [Ca2+]cyt is essential for this response (Figures 5 and 6). Furthermore, this transcriptional response is not induced by the [Ca2+]cyt changes stimulated by other oxidative stresses and ROS applied here, such as H2O2 (Figure 7) or cold (Figure 8). Interestingly, the expression of SYP121 which is a hub in the SYP121/SYP122/SNAP33 network is also not up-regulated by cold (Leyman et al., 1999). These data provide strong support for the Ca2+ signature hypothesis (McAinsh and Hetherington, 1998; McAinsh and Pittman, 2009) and suggest the capacity for stimulus-specific signalling information to be encoded directly in the spatiotemporal kinetics of stress-induced changes in [Ca2+]cyt or through their interaction with sensor proteins that differ in their affinities for both Ca2+ and their target proteins (Dodd et al., 2010; Kudla et al., 2010; Reddy et al., 2011). This finding, therefore, raises the intriguing possibility that the Ca2+ signature plays an important role in the molecular machinery by which plants distinguish between, and adapt to, changing environmental conditions through the differential regulation of appropriate stress-induced genes.
The present study provides additional support for the importance of the second phase of the ozone Ca2+ signature in directing ozone-induced gene expression (Clayton et al., 1999; Figures 5e and 6). The kinetics of stress-induced changes in [Ca2+]cyt vary markedly, however, and this opportunity is therefore only one of several for encrypting stimulus-specific information within plant Ca2+ signals. For example, cold-shock induces a rapid, transient increase in [Ca2+]cyt (Knight et al., 1996; Kiegle et al., 2000), whereas drought, salt stress and prolonged cooling all result in a biphasic increase in [Ca2+]cyt (Plieth et al., 1999; Kiegle et al., 2000), reminiscent of that stimulated by ozone (Clayton et al., 1999; Evans et al., 2005), but which differ in their temporal dynamics. The kinetics of changes in [Ca2+]cyt induced by multiple stresses working in concert also differ markedly from those of the individual stresses (Hetherington et al., 1998; Tracy et al., 2008); whilst the kinetics of the cold-Ca2+ signature vary throughout the day and highlight the potential for the circadian modulation of Ca2+ signals (Dodd et al., 2006). Furthermore, computer simulations of cold-induced increases in guard cell [Ca2+]cyt suggest that [Ca2+]cyt increases measured from populations of cells are likely to represent the summation of the cold-induced changes in single-cell [Ca2+]cyt (Dodd et al., 2006). This finding suggests that there may also be discrete roles for individual cell types and/or organelles in the detection and response of plants to different environmental stresses through the generation of cell-type/organelle-specific Ca2+ signatures (McAinsh and Pittman, 2009). These studies, therefore, illustrate the potential for the integration of signalling information from multiple stresses acting in concert through the generation of complex Ca2+ signatures that enable plants to formulate the optimum physiological response to the specific environmental conditions to which they are exposed.
Coupling Ca2+ signatures to transcription
The ability of cells to decode complex Ca2+ signatures and to transduce the encrypted signalling information to downstream effectors will depend on the complement of Ca2+-binding proteins that function as Ca2+ sensors present in the cell (Dodd et al., 2010; Kudla et al., 2010; Reddy et al., 2011). These proteins are typically encoded by multiple genes, many of which are differentially regulated by stresses (DeFalco et al., 2010). Our analyses show Ca2+ signalling is one of the main functional groups over-represented amongst the genes that are altered transcriptionally by ozone (Table 1), and that these include eight Ca2+-dependent protein kinases (CDPKs), 10 calmodulin (CaM) genes, and five CaM-binding proteins that exhibit at least a two-fold increase in expression in response to ozone (Table S1). This group also contained several genes that are stress-induced in a Ca2+-dependent manner, such as the touch-inducible CaM, CAM2 (Ito et al., 1995), and TCH2 and TCH3 (Polisensky and Braam, 1996). The SYP121 and SYP122 proteins, encoded by genes that form hubs in our ozone-regulated vesicle transport network, are rapidly phosphorylated after elicitation with Flg22 (a bacterial flagellar peptide that functions as a pathogen-associated molecular pattern (PAMP) peptide in plants), most likely by Ca2+-dependent, CaM domain protein kinases (Nüshe et al., 2003; Benschop et al., 2007). Two such CDPKs, CPK4 and CPK24, are also present in the SYP121/SYP122/SNAP33 cluster (Figure 3).
The molecular pathways by which Ca2+ signatures direct the expression of Ca2+-responsive genes are key to the mechanism by which plants exhibit a stimulus-specific transcriptional response to environmental stresses (Dodd et al., 2010; Kudla et al., 2010; Reddy et al., 2011). Recently, four cis-acting elements have been identified in the promoters of Ca2+-responsive genes, CRT/DRE, Site II, CAM box, and ABRE, for which Ca2+ is both necessary and sufficient for expression via these motifs (Kaplan et al., 2006; Whalley et al., 2011). Interestingly, Blomster et al. (2011) identified the ABRE element as being enriched significantly in genes that are regulated at late time points by ozone. It is, therefore, likely that the interaction of Ca2+/CaM with transcription factors that regulate these motifs, and other putative Ca2+-regulated cis-acting elements, may play a central role in the remodelling of the transcriptome in response to environmental change. Network analysis of ozone-regulated genes reveals that the W-box binding motif for WRKY transcription factors (Eulgem and Somssich, 2007) is strongly over-represented in the promoters of genes in the SYP121/SYP122/SNAP33 sub-cluster from our stress-related vesicle transport network, with all five of our co-regulated ozone-induced genes containing at least one W-box motif. In addition, five WRKY genes appear in this cluster. Several other studies have also identified the W-box as over-represented in Arabidopsis ozone-responsive genes (Mahalingam et al., 2005; Tosti et al., 2006; Wrzaczek et al., 2010; Blomster et al., 2011). Importantly, several members of the WRKY family are known to bind CaM in a Ca2+-dependent manner (Park et al., 2005; Popescu et al., 2007) and interactions of CaM with MYB transcription factors have also been reported (Yoo et al., 2005). Furthermore, evidence is beginning to emerge that stress-induced changes in [Ca2+]cyt may regulate gene expression at the post-transcriptional level (Popescu et al., 2007; Whalley et al., 2011). Taken together, these data highlight the extent of the connectivity between the regulation of transcription and Ca2+-mediated stress responses (Kudla et al., 2010; Reddy et al., 2011).
Our results shed light on the molecular mechanisms by which stimulus-specific signalling information is encoded in stress-induced Ca2+ signatures and on how it is decoded and transduced to the transcription machinery. The potential for interplay between sensor proteins that differ in their affinities for both Ca2+ and their target proteins (Dodd et al., 2010; Kudla et al., 2010; 1; Reddy et al., 2011) and that are differentially expressed, both temporally and spatially, depending on the physiological address of cells (McAinsh and Hetherington, 1998), provides a dynamic and flexible system for decoding variations in the spatiotemporal kinetics of stress-induced changes in [Ca2+]cyt and organelle Ca2+ and for integration of information about multiple stresses perceived concurrently. The emergence of multiple Ca2+-regulated cis-acting elements and Ca2+-/CaM-binding transcription factors as candidates for coupling Ca2+ signatures to transcription contributes additional components to the Ca2+ signalling toolkit by which cells formulate an appropriate physiological response that is dependent on the specificity of their interactions and the interconnections present within the Ca2+-signalling network.
Plant material and growth conditions
Seeds of Arabidopsis thaliana ecotype Col-0 and RLD1, the latter constitutively expressing the apoaequorin gene (Knight et al., 1991), were sterilised with ethanol and sown in Petri dishes on Murashige and Skoog medium (Sigma-Aldrich, http://www.sigmaaldrich.com/, UK) that contained 0.8% (w/v) agar, at approximately 50–100 seeds per dish. Seeds were maintained at 4°C in the dark for 4 days and then transferred to an AR36L3 Arabidopsis growth cabinet (Percival, http://www.percival-scientific.com/, USA). Seedlings were grown at 20 ± 2°C day/18 ± 2°C night, with a photosynthetic photon flux density of 130 μmol m−2 sec−1, and a 16 h photoperiod. All experiments were performed on 10-day-old seedlings. To reconstitute aequorin, seedlings constitutively expressing apoaequorin were incubated in 4 μm coelenterazine (Calbiochem; CN Biosciences, Nottingham, UK) in the dark at 20 ± 2°C for a minimum of 16 h. When appropriate, seedlings were incubated in either 10 mM lanthanum chloride (a plasma membrane channel blocker [Clayton et al., 1999]) for 1 h, or distilled H2O (control) immediately prior to treatments being applied.
Ozone fumigation, hydrogen peroxide and cold treatments, and RNA extraction and real-time RT-PCR
Ozone fumigations were performed according to Clayton et al. (1999). Petri dishes with the lids removed were exposed to ozone (100–500 ppb) or ozone-free air (control) at a flow rate of 910 ml min−1 inside a fumigation chamber within the growth cabinet. Ozone was generated and monitored using a Dasibi, model 1008-PC ozone generator (Glendale, CA, USA). Ozone-free air was produced by passing laboratory air through a charcoal/purafil filter. In both cases, the air stream was bubbled through distilled H2O before entering the chamber in order to remove any H2O2 present and to humidify the air. H2O2 treatments were imposed by spraying all the seedlings on an single plate with either 9 ml of freshly prepared 10 mm H2O2 or distilled H2O (control). Cold treatments were imposed by transferring Petri dishes to an identical Percival AR36L3 Arabidopsis growth cabinet at 4°C. Seedlings were exposed to treatments for 6 h, following which RNA was extracted from three replicate pools of approximately 100 seedlings per treatment using an RNeasy Plant Mini Kit with the optional on-column DNase treatment (Qiagen, http://www.qiagen.com/, UK). Relative transcript abundance was quantified using real-time RT-PCR. cDNA was synthesised from 5 μg total RNA using Superscript II RNase H− Reverse Transcriptase (Invitrogen, http://www.invitrogen.com/, UK) and diluted with 81 μl distilled H2O at the end of the reaction for use. Each reaction contained 1× SYBR® Green mastermix (Applied Biosystems, http://www.appliedbiosystems.com/, UK), 1 μl cDNA, 18 nm each primer and distilled H2O up to 25 μl. PCR was carried out for 40 cycles in an Applied Biosystems 7000 Sequence Detection System (10 min, 96°C; 20 sec, 96°C; 20 sec, 50°C; 4 min, 60°C). The relative amounts of cDNA were calculated by the ΔCT method using the formula 2−ΔΔCt (as recommended by Applied Biosystems) against tubulin (TUB4; At5g44340) and actin (ACT1; At2g37620). The mean variation in the expression of the internal control genes across all samples in the microarray data was 10.8%, and the primer pairs used for all genes had a calculated efficiency of 100 ± 10%. Transcript levels were expressed as the mean of these two relative expression values normalized to the control sample.
Microarray and bioinformatics
RNA labelling, hybridisation and scanning of the Genechip®Arabidopsis ATH1 genome array (Affymetrix, http://www.affymetrix.com/) were performed at the Nottingham Arabidopsis Stock Centre (NASC) as part of the GARNet service provision (Craigon et al., 2004). Six chips were hybridised in total (one for each of the ozone-treatment replicates and one for each of the control replicates). Data were normalized using dChip (Li and Wong, 2001) and expression values generated from the Affymetrix CEL files using the PM/MM difference model and log2 transformed. Raw and normalized microarray data were deposited in the Array Express database (accession number E-MEXP-342). Genes that demonstrated low expression (log2 expression values <7.0 in >50% of arrays) and genes with constant expression (log2 standard deviation <0.5 across all arrays) were excluded from further analysis. The co-expression gene network was identified using the CressExpress tool (Srinivasasainagendra et al., 2008) to query data from publically available biotic and abiotic stress microarray datasets, and was visualized using Cytoscape software (Shannon et al., 2003). Significant sub-clusters within the network were identified using the program MCODE (Bader and Hogue, 2003) as a plug-in within Cytoscape.
Ozone fumigation, hydrogen peroxide and cold treatments, and [Ca2+]cyt measurements
Seedlings in which aequorin had been reconstituted were placed individually into a glass cuvette inside a temperature controlled sample chamber at 20°C that was attached to a digital chemiluminometer (Electron Tubes Ltd, http://www.et-enterprises.com/, UK). Resting luminescence counts were recorded for a period of 5 min at 1-sec intervals prior to the application of treatments. Ozone fumigations were performed essentially according to Clayton et al. (1999). Individual seedlings were placed on strips of filter paper that had been wetted with room-temperature distilled H2O inside the luminometer cuvette. Ozone (300 ppb) or ozone-free air (control) was introduced into the cuvette at a flow rate of 100 ml min−1. H2O2 and cold treatments were performed essentially according to Rentel and Knight (2004) and Knight et al. (1996), respectively. Individual seedlings were floated in 0.5 ml room temperature distilled H2O in the luminometer cuvette. Treatments were imposed by addition of 0.5 ml of H2O2 (10 mm final concentration) or 1 ml of ice-cold distilled H2O (final temperature of 4°C maintained using the temperature controlled sample chamber), or the equivalent volume of room temperature distilled H2O (control), to the cuvette. Luminescence counts were recorded for 1 h following the application of treatments. At the end of the experiment, any remaining aequorin was discharged by the addition of a solution of 0.9 m CaCl2, 10% (v/v) ethanol to the cuvette. Calibrations were performed using an empirically derived equation (Knight et al., 1991, 1996).
Statistical analyses were performed using SigmaStat (SPSS, http://www-01.ibm.com/software/analytics/spss/). Differences in transcript abundance were assessed in a paired t-test. The effects of treatments on [Ca2+]cyt were investigated by analysis of variance, or equivalent non-parametric tests. Unless otherwise stated, data were regarded as significantly different when P-values were ≤0.05.
This work was supported by studentships from NERC (NER/S/A/2000/03401) and BBSRC (01/B1/P/07198), UK.