An early response regulatory cluster induced by low temperature and hydrogen peroxide in seedlings of chilling-tolerant japonica rice

Transcriptional regulation is an important aspect of the complex genetic and biochemical networks involved in plant responses to low temperature. In plants like Arabidopsis that evolved to withstand freezing by cold acclimation (CA), the transcriptional regulatory network is defined by the interaction between a number of cold-responsive transcription factors and their cognate cis-elements in the promoters of a suite of downstream target genes [1–4]. The downstream target genes, which are activated in concert not only by low temperature but also by other related environmental (dehydration, high salinity) and chemical (abscisic acid or ABA) signals were referred to as COR for co ld r egulated, rd for r esponsive to d esiccation, lti for l ow t emperature i nduced or kin for c old-in ducible [5–8]. Evidence supporting the important roles of COR genes in cellular protection against the common biochemical perturbations caused by low temperature and dehydration via protein chaperoning (e.g., LEA proteins), membrane stabilization (e.g., COR15a) and osmotic adjustment (e.g., Δ¹-pyrroline-5-carboxylate synthase) has been established by transgenic analysis [9–11].

Molecular dissection of the promoters of several COR genes of Arabidopsis (e.g., rd29A/lti78/cor78, rd17/cor47, cor15a, cor6.6/kin2) indicated that they are regulated in an ABA-dependent or ABA-independent pathway under cold, dehydration and high salinity conditions. However, their cold-induced expression is largely through the ABA-independent pathway mediated by highly conserved promoter cis-elements that function as binding sites for transcriptional activators [6, 12]. For example, ABA-responsive expression of rd29A/lti78/cor78 requires a highly conserved promoter motif PyACGTGGC called ABA-Response Element (ABRE). On the other hand, low temperature, dehydration and high salt- inducible expression of rd29/lti78/cor78 requires a 9-bp sequence TA(C/G)CGACAT called C-r epeat/D ehydration-R esponsive E lement (CRT/DRE) that is not independently activated by ABA alone [8]. Many Arabidopsis genes with ABRE-like and CRT/DRE-like motifs in their promoters have been identified and these genes have been characterized at least in terms of their stress-responsive expression [10, 13].

The transcription factor(s) that interact with the ABRE (AREB or ABF) are bZIP-type proteins that belong to a specific group (Class A) of stress-related bZIP proteins [14, 15]. The activators of CRT/DRE-containing genes belong to a plant-specific group of AP2/EREBP-type DNA-binding proteins referred to as CBF1/DREB1b, CBF2/DREB1c, CBF3/DREB1a, CBF4, DREB2A and DREB2 [12, 16, 17]. Overexpression of either CBF1/DREB1b or CBF3/DREB1a in Arabidopsis resulted in strong and coordinated activation of the target COR genes, and subsequently CA-independent tolerance freezing and dehydration [18, 19]. Thus, the CBF/DREB regulon was the first example and most prominent regulatory cluster associated with cold stress in plants.

More recently, global gene expression profiling experiments revealed that a large number of Arabidopsis genes are responsive to low temperature. Of the more than 500 candidates so far identified, about 60% and 40% were upregulated and downregulated during CA, respectively [13, 20]. Analysis of the transcriptome of CBF-overexpressing Arabidopsis revealed that the CBF/DREB regulon is a large cluster comprised of about a hundred CBF/DREB-responsive genes that includes both the known COR genes as well as many newly identified members. Potential sub-regulons defined by RAP2.1 and RAP2.6 transcription factors were also identified and presumed to function under the larger CBF/DREB regulon [10, 13]. Furthermore, transcriptome data also revealed that the low temperature genetic network involves other regulatory clusters in addition to the CBF/DREB regulon. An example of this is the ZAT12 regulon, which is comprised of more than 20 downstream target genes. Candidate cognate enhancers in the promoters of ZAT12-target genes have also been identified. Potential target sites of this transcription factor contain the core motif CATTG [1, 20].

Although the precise hierarchical organization of the low temperature genetic networks is yet to be defined, it appears that CBF/DREB and ZAT12 pathways (and perhaps other unidentified pathways) constitute a 'super regulon' that configure the overall low temperature stress tolerance mechanisms in Arabidopsis. These pathways are further fine-tuned by additional components (e.g., HOS1, LOS1 and LOS2) either by positive or negative regulation [21, 22]. Related studies involving a global survey of metabolic changes also showed that the CBF/DREB regulon plays a prominent role in configuring the biochemical status of Arabidopsis during exposure to low temperature [11, 13, 20]. Given the evolutionary conservation of CBF/DREB regulon between temperate and warm-season plants, all these findings established the prominent role of this pathway in the cold stress global regulatory networks of plants in general.

Many cold-responsive genes containing CRT/DRE-like promoter elements have also been identified in plants that do not acclimate to freezing, indicating that various cold stress response regulatory modules are highly conserved even between chilling-insensitive (temperate) and chilling-sensitive (warm-season) plants [23–26]. However, despite the current models of the Arabidopsis CA regulatory networks and the identification of functional homologs of CBF/DREB transcription factors and their corresponding regulatory modules in rice and other warm-season plants, the precise composition and hierarchical organization of the component pathways of the global networks in this group of plants is yet to be defined [27, 28].

Indica and japonica rice cultivars exhibit wide contrast in terms of their sensitivity to chilling. At the seedling stage, most japonica cultivars survive continuous exposure to as low as 10°C for up to 7 days better than most indica cultivars. Having a complete genome sequence and a variety of expression profiling platforms, the japonica rice is an ideal model to assemble the various component pathways of the cold stress regulatory networks of chilling-sensitive plants. Parallel assembly of the homologous networks involved in regulating the transcriptome in chilling-insensitive (Arabidopsis) and chilling-sensitive (rice) plant models will contribute to the understanding of the evolutionary changes that led to such differences in low temperature sensitivity between the two groups of plants. To contribute to this goal, we performed a semi-global survey of the chilling stress transcriptome of rice using a cDNA microarray representing about 6,000 seedling-expressed genes whose ESTs were enriched in a cold stress subtracted library [29]. The goal of this initial survey was to identify the early components of the network through the identification of a set of co-regulated genes during the initial exposure to 10°C.

The results of the survey presented here revealed an early response regulatory cluster that is potentially independent of the CBF/DREB regulon perhaps activated via reactive oxygen species (ROS). Although many ROS inducible genes have been identified in Arabidopsis [30], our knowledge of how these genes are organized into regulatory clusters lags behind our understanding of the hierarchical organization of the CBF/DREB and other known regulons. Here we present evidence of an ROS-mediated regulatory module that functions as an early component of the chilling stress response pathway in japonica rice based on integrative genomics analysis.

Chilling-induced physiological imbalance leads to elevated levels of reactive oxygen species (ROS) in plant cells [42–45]. In particular, H₂O₂ is generated rapidly under stress conditions either by enzymatic means (plasma membrane NADPH oxidase, cell wall peroxidase, amine oxidase) or by the normal metabolic routes in the chloroplast (Mehler reaction) and mitochondria (electron transport and photorespiration). Steady state levels depend on the balance between synthesis and degradation, which is facilitated by the ROS-scavenging system of the cell [44]. With a subset of experimentally validated chilling- and H₂O₂-responsive genes, we inferred possible regulatory relationships among a group of genes that were induced in coordinated manner within a short time window during the initial stages of exposure of rice seedlings to 10°C. We hypothesized that the early response transcriptome might involve reactive oxygen species (ROS) via ROS-mediated gene expression defined by the as1/ocs-like promoter elements. Microarray data supports an elevated ROS level in rice seedlings during such period by virtue of the upregulation of ROS scavengers such as catalase, ascorbate peroxidase, glutathione S-transferase, and superoxide dismutase. Transcript levels of these genes were significantly higher than the control levels based on the combined results of microarray, northern blot and RT-PCR analyses (data not shown). Additionally, our current data indicated that a mechanism for effective reduction of the rate of the Fenton reaction is activated during the initial 24 hours based on chilling- and H₂O₂-induced expression of a ferritin gene (Os11g01530). The Fenton reaction is responsible for the conversion of H₂O₂ to more potent hydroxyl radicals via a Fe²⁺-requiring chemical reaction. Thus, the functional significance of the observed induction of ferritin expression can be explained in terms of its function in Fe²⁺ sequestration, which would then limit the rate by which unsuccessfully scavenged H₂O₂ (which could occur during periods of oxidative burst) would be converted to hydroxyl radicals [46].

It has been established that H₂O₂ plays a very important role in mediating signal transduction in response to both abiotic and biotic stresses in plant cells. H₂O₂ diffuses rapidly from its site of synthesis within subcellular microdomains depending on concentration and it can transmit intracellular signals to the nucleus by oxidizing various upstream components of the signaling pathway including protein kinases (MAPK cascades), protein phosphatases, transcription factors and membrane-bound calcium-channels. The end response is change in gene expression [47–49]. Although many ROS-inducible genes have been identified in Arabidopsis [30], direct involvement of an ROS-mediated regulatory module in the low temperature transcriptional networks has not been studied to the same level as the CBF/DREB regulon had been scrutinized in Arabidopsis.

We assembled a hypothetical model of the ROS-mediated regulon and we propose a hypothesis on its potential relationship with other major regulon(s) involved in low temperature response of the chilling-tolerant japonica rice. Our working hypothesis described in this model is that gene expression during the initial 24 hours at 10°C can be defined (at least partially) in terms of a regulatory cluster (transcription factor-target gene module) that appears to be independent of the CBF/DREB and ABA-mediated regulons (Figure 8). The downstream components of this regulatory module (pathway-1 in Figure 8) contain either the prototype as1/ocs element, which is a known target of TGA-type bZIP transcription factors, or the related W-box that is known to bind to TGA-type bZIP and WRKY transcription factors or both. These elements are known to be activated in response to various conditions that trigger the production of excess ROS (particularly H₂O₂) including biotic, abiotic and xenobiotic stresses. Genes that contain these promoter elements have also been shown to be induced by exogenously supplied H₂O₂ and other chemical signals such as salicylic acid and auxin [31, 38, 39, 50–52].

We also confirmed that a representative subset of 'early response' genes containing as1/ocs-like promoter elements were responsive to exogenous H₂O₂. Based on the percentage of occurrence and enrichment level of the as1/ocs-like elements, we predict that about 80% of the 'early response' genes that we identified in the semi-global microarray survey are probably responsive to exogenous H₂O₂ hence potential members of this regulatory cluster. The specific ferritin gene (Os11g01530) that was profiled in this study was an interesting example of a potential target of regulation by H₂O₂. We have shown that this ferritin gene not only contains as1/ocs-like promoter elements but it was also induced by exogenous H₂O₂. Ferritin is also directly related by function to H₂O₂ metabolism by virtue of its role in the regulation of the Fenton reaction [46]. In addition to ferritin, many of the genes that we identified in this group have also been shown to be responsive to exogenous H₂O₂ in Arabidopsis [30]. The downstream target cluster of the proposed ROS-mediated regulon also includes a number of transcription factors such as OsMyb4 and an EREBP1-type protein, suggesting the possible occurrence of other secondary regulatory clusters under this regulon. This layered organization appears to be consistent with the fact that the ROS-mediated regulon is an 'early response' mechanism that may be functioning as an initial trigger to a more complex transcriptional network involved in cellular defenses against the physiological injuries caused by chilling.

We hypothesized that a bZIP-type protein is the most likely component of the ROS-mediated pathway upstream to the as1/ocs-like element-containing genes based on the known fact that this family of elements is involved in the recruitment of TGA-type bZIP factors in the promoters of target genes [40]. Based on the composition of the 'early response' regulatory cluster that we have assembled, this likely regulator is a novel Class-I bZIP protein (ROS-bZIP1) that is distinct from the family of known stress-related bZIP transcription factors (Class A). This class of bZIP proteins is quite diverse in function that includes those that are involved in the regulation of vascular development. Recently, some of the members of this class have also been shown to be involved in the development of symptoms to the rice tungro disease [15, 53, 54]. The significance of this classification with respect to our proposed role of ROS-bZIP1 is unclear at this point but can be explained with better understanding or confirmation of the function of ROS-bZIP1 during chilling stress.

We have a good rationale to hypothesize that ROS-bZIP1 is a likely regulator of the as1/ocs-like element-containing genes based on several peculiar features. First, the timing of expression of ROS-bZIP1 is consistent with being an upstream regulator of a cluster of co-regulated genes. For instance, among the 'early response' transcription factors identified in this study, only ROS-bZIP1 showed an induction peak within the initial 2 hours and it was also induced by exogenous H₂O₂. The other transcription factors were either induced by chilling at a much later period (same time period as the as1/ocs-like element-containing genes) or not induced by exogenous H₂O₂ or both. Based on the differential expression analysis of the chilling-tolerant and intolerant rice cultivars, the timing of expression of the as1/ocs element-containing genes appeared to be a consequence of the early induction of ROS-bZIP1 (i.e., ROS-bZIP1 precedes the as1/ocs-like element-containing genes). Additionally, we also revisited the original EST libraries used in the assembly of the microarray and we confirmed that ROS-bZIP1 and many of the as1/ocs element-containing genes (ferritin, HMG protein, and putative ATP sulfurylase) were co-represented in the same cold stress subtracted EST library but not in the drought stress EST library, consistent with their parallel expression in the microarray. This information provided additional circumstantial evidence in support of our hypothesis that ROS-bZIP1 and the as1/ocs element-containing genes are most likely related in a regulatory context. Induction of ROS-bZIP1 by exogenous H₂O₂ indicates that this gene is a direct target of an ROS-activated upstream component. The second rationale is that ROS-bZIP1 is differentially regulated between tolerant and intolerant rice cultivars and its early and robust expression appeared to be positively correlated with chilling tolerance. It appears that timely expression of ROS-bZIP1 is one of the initial triggers leading to the activation of the early transcriptome that plays critical roles in the initial physiological adjustments and configures subsequent defenses.

We also acknowledge the possibility that co-expression of ROS-bZIP1 and as1/ocs-like element-containing genes could be a mere coincidence, and that they are in fact not directly related in the context of a regulator and downstream target module. This is a possibility that we cannot ignore completely because of the fact that the microarray that we used in the survey consisted of only about 15% of the total genes encoded by the rice genome. There is a possibility that the regulator of the as1/ocs-like element-containing genes was not represented in the miroarray. Our on-going studies that include the analysis of the chilling stress transcriptome using a 45,000-oligonucleotide microarray representing the entire set of rice genes and in vitro binding of ROS-bZIP1 to the as1/ocs-like sequence will address this issue.

ROS-bZIP1 (or any of its homologs in other plants) has not yet been characterized in terms of its potential role in chilling tolerance. With the presumed large size of the putative ROS-bZIP – as1/ocs regulatory module, functional analysis of ROS-bZIP1 offers an opportunity to reveal another important switch of cold stress gene expression that has not been previously examined in detail in Arabidopsis. A cautionary note is that at this point the predicted regulatory relationship between ROS-bZIP1 and the as1/ocs element-containing genes awaits further confirmation by promoter deletion-reporter analysis, and direct protein-DNA binding assays such as gel-shift, yeast one-hybrid or chromatin immunoprecipitation. Nevertheless, the regulatory module that we inferred through an integrative genomics approach provides a strong justification for future use of these candidate genes as prototype model for further dissection of the early response ROS-mediated gene regulon by genome-wide transcriptome analysis of transcription factor-overexpressing or knock-out lines.

We found that the upstream regions (-1,000 bp) of the Group-I 'early response' genes were enriched with potential Myb-target (MybR) motifs. ROS-bZIP1 in particular contains three CAAC-type and three other types of MybR motifs, suggesting a possibility that the next component upstream to this gene in the ROS-mediated pathway is probably a Myb-type transcription factor. Constitutively expressed transcription factors could be direct targets of ROS [55]. An H₂O₂-sensitive Myb protein is the most likely direct link of the putative ROS-bZIP – as1/ocs regulatory module to the chilling induced-H₂O₂ signal possibly by activating such protein. Evidence showed that constitutively expressed Myb proteins are activated by redox state-mediated conformational changes in their DNA-binding domains [56]. This could be one possible way by which H₂O₂ activates a putative upstream Myb protein that regulates the expression of ROS-bZIP1. Alternatively, an inactive Myb protein may be activated through phosphorylation by an upstream kinase that is sensitive to H₂O₂. Changes in intracellular Ca²⁺ concentration may be involved in this activation [57]. Either way, a bZIP-type transcription factor appears to be the most upstream component that is regulated at the transcriptional level by an ROS-activated Myb protein (pathway-1 in Figure 8).

Another interesting feature of the as1/ocs-like element-containing genes is the significant occurrence of MybR motifs (different types combined) in their promoters. For instance, all five genes studied in detail by qPCR including those that were confirmed to be induced by exogenous H₂O₂ (OsMyb4, EREBP1-type transcription factor, HMG protein, ferritin, and putative ATP sulfurylase) contain different types of MybR motifs. However, unlike the as1/ocs and W-box elements, which are very closely related (nearly identical in sequence), there was no obvious consensus MybR motif found in the promoters of this group of genes. Nevertheless, it is likely that the as1/ocs-like element-containing genes might also be regulated through a secondary route, perhaps involving a possible MybR element and a constitutively expressed but ROS-activated Myb similar to the proposed regulation of ROS-bZIP1. In this case, the chilling- and H₂O₂-induced expression of this group of genes may be the result of the interaction between as1/ocs-like and certain MybR elements. The slight variation in expression profiles among the Group-II genes examined may be attributable to the potential interaction between as1/ocs-like and certain MybR elements. Alternatively, since all the as1/ocs-like element-containing genes examined by qPCR had significant basal expression level (constitutive level) but highly induced by chilling, the occurrence of certain MybR motifs may be related to a possible modulator function or potential determinants of the basal expression of this specific group of genes.

A number of genes that belong to the proposed ROS-bZIP – as1/ocs regulatory module also contain CRT/DRE-like motifs in their promoters. An example of this is ferritin (Os11g01530), which has been reported as a putative CBF/DREB target [24]. Previously, we also reported a sequential activation of CBF/DREB homolog and CRT/DRE-containing gene (OsLti6a) in CT6748-8-CA-17 during the initial 24 hours at 10°C [23]. Both ferritin and OsLti6a had genotype-specific expression profiles (CT6748-8-CA-17 vs. INAP12) that were very similar to the patterns of the representative Group-II genes. Therefore, ferritin and OsLti6a represent a group of genes that are potentially regulated by both an ROS-responsive bZIP and CBF/DREB transcription factors (pathway-2 in Figure 8). This result suggests a potential link between the ROS-mediated and CBF/DREB regulons through common downstream target genes. Promoter motif distribution data also showed that less than 40% of the 'early response' genes identified in this study contain both as1/ocs element and CRT/DRE in their promoters (at least within the -500 region), suggesting that the overlap between these two regulons is probably not that extensive. Comparative microarray analysis of transgenic lines overexpressing ROS-bZIP1 and CBF/DREB is currently underway to address this hypothesis.

ABA signaling has known linkages with ROS-mediated signaling [41, 49]. However, based on our current results it does not appear that the putative ROS-bZIP – as1/ocs regulatory module is directly linked to ABA-mediated pathway at least by virtue of the lack of detectable ABA-mediated induction of representative as1/ocs-like element-containing genes. Alternatively, it is possible that the ROS-mediated regulon is upstream to ABA in the signaling pathway. More detailed analysis of the effect of exogenous ABA on a larger set of confirmed H₂O₂-responsive genes is required to validate these hypotheses. Genome-wide microarray survey is also currently underway to address this hypothesis.

Rice is sensitive to even mild cold stress (chilling) particularly at the early stages of seedling establishment. However, japonica cultivars are generally more tolerant than most indica cultivars. Studies aiming to dissect the regulatory networks that define the mechanism by which the responses of a chilling-sensitive plant like rice might differ from the responses of chilling-insensitive plants are just in the early stages. Based on the analysis of a subset of candidate genes, we hypothesized that the early response transcriptome has an important role in the short-term defenses, which are crucial to the survival of a chilling-sensitive plant like rice under above-freezing temperature conditions.

In this study, we have established the rationale that ROS is involved in the early branches of the chilling stress response transcriptional network of rice. We have identified the components of a putative ROS-mediated regulatory module by an integrative analysis of genomics data. We propose that ROS regulates an important component of early response transcriptome by mediating the rapid induction of a group of as1/ocs-like element-containing genes, hence ROS-dependent regulon. We also hypothesized that this regulon is probably independent of the CBF/DREB or ABA-mediated regulons in rice. The putative ROS-bZIP – as1/ocs regulatory module is currently being validated by biochemical and genetic tests. Given the rapidly growing functional genomics resources for rice, our next step is also to validate the putative regulatory module described in this report with the use of mutants and overexpression lines. Physiological experiments are currently underway to compare the intracellular ROS levels between japonica (chilling tolerant) and indica (chilling intolerant) rice seedlings during the initial 24 hours of chilling.