Genome-wide analysis of rice ClpB/HSP100, ClpC and ClpD genes

Heat stress threatens future prospects of increased grain production in crops. Rice (Oryza sativa) is most important world food crop. The production of rice is getting severely affected with increases in mean global temperature. According to estimates, yield of rice declines by 10% for every 1°C increase in growing period minimum temperature in the dry season [1]. Processes like spikelet fertility, grain quality and yield processes in rice are considered to be especially sensitive to heat stress [2]. For breeding heat tolerant rice, it is important that molecular components that underlie the heat shock response in this species are understood [3]. Microarray profiling data have shown that heat stress response in maturing tomato microspores involves heat shock proteins, ROS scavengers, hormones and sugars [4]. Understandably, the major molecular changes in rice plants as affected by heat stress need to be worked out. This is especially relevant since rice has emerged as a model plant species of the group monocots due to its small genome size, availability of large collection of full-length cDNAs (FL-cDNAs) and for the fact that the whole genome of this plant species is completely sequenced. This crop has attracted a great deal of efforts for the elucidation of gene functions; completion of genome sequencing in rice has paved way for comprehensive functional characterization of genes, transcription factors, signaling components and promoters [5, 6]. The completed rice genome sequence has been used for the characterization of a large number of gene families involved in diverse processes and pathways. However, with almost 42000 genes [7], several of them unknown, there are ample proteins still left to be characterized in this important crop species. In recent years, comprehensive details on heat shock regulated rice HSP20, HSP70, HSP90 and HSF gene families have been reported [8–11]. HSP100 is a major heat-regulated protein family in diverse organisms. Across the living systems, common features of HSP100 chaperone action include transient interactions with non-native protein species, in the prevention of aggregation and promotion of correct folding and assembly, or in unfolding for translocation or targeting to proteases. Singla and Grover [12] showed that homologues of yeast HSP104 protein are expressed in heat shocked rice seedlings. It was subsequently established that apart from heat, rice HSP100 expression is developmentally-controlled as seeds and developing embryos of rice show high constitutive levels of this protein [13, 14].

HSP100 proteins belong to ClpB family [15–17]. Clp ATPases maintain quality of cellular proteins by performing the function of molecular chaperones and energy dependent proteases. Clp (C aseinol ytic P rotease) system was first identified as a heat shock inducible, multicomponent, ATP-dependent protease complex able to hydrolyze casein [18, 19]. Subsequent studies showed that the Clp system can hydrolyze numerous other proteins and peptides in both aggregated and non-aggregated forms [20] Clp ATPases fall within the AAA⁺ superfamily of ATPases associated with a substantially broader range of biological processes [21, 22]. Class I ATPases (ClpA, ClpB, ClpC, ClpD) have two ATP binding domains and class II Clp ATPases (ClpM, ClpN, ClpX, ClpY) have one ATP binding domain [23]. Clp proteins are localized in various cellular organelles in plants [24]. Basically, Clp system members include three non-homologous gene families: ClpABCXY, ClpP and ClpQ. ClpACX members (but not ClpB) facilitate the activity of ClpP and some, such as ClpA [25] and ClpX [26], can function as independent chaperones in roles analogous to those of DnaK and DnaJ proteins. In contrast, ClpP proteolytic subunit exhibits low levels of peptidolytic activity. Further, when ClpP is complexed with ClpA, ClpC or ClpX, active holoenzymes which are able to cleave denatured proteins are formed. ClpAP, ClpXP and HslUV proteases are similar in design to the eukaryotic 26 S proteasome, with the ATPase subunits guarding the entrance to the proteolytic chamber [27, 28]. Clp proteases in bacterial and eukaryotic systems have been implicated in vital cellular processes such as sporulation, DNA replication, protein turnover, stress tolerance and acclimation and regulation of gene expression [29]. ClpB is different from ClpA, ClpX and HslU as it does not associate with peptidase subunits. The function of ClpB is also distinct from that of other Clp ATPases: this protein is not involved in protein degradation [30], instead it disaggregates and reactivates strongly aggregated proteins [31–33]. The aggregation reversing activity of ClpB requires cooperation with the HSP70/HSP110 chaperone machinery [34].

In Arabidopsis, ClpB proteins have been divided into 3 classes according to cytoplasmic, chloroplastic and mitochondrial isoforms. Cytoplasmic ClpB is constituted by At1g74310, chloroplastic ClpB by At5g15450 and mitochondrial ClpB by At2g25140 [17]. Much of the work on ClpB proteins has been carried out on AtClpB-cyt isoform [35–38]. We isolated and sequenced rice ClpB-cyt gene (GenBank accession no. AJ316025) and showed that rice ClpB-cyt cDNA complements yeast hsp104 deletion and loss of thermotolerance trait in yeast [39]. Genetic analysis has shown that ClpB-cyt protein is essential for heat tolerance in plants. Mutants of Zea mays and A. thaliana plants under-expressing their respective ClpB-cyt/HSP100 proteins are observed to lack both basal as well as induced thermotolerance [35–37]. Queitsch et al [38] produced transgenic A. thaliana plants by modifying level of AtClpB-cyt/HSP100 protein. Transgenic plants in the latter study survived as high as 45°C (1 h) temperature stress as they showed vigorous growth after the removal of stress. Katiyar-Agarwal et al [40] over-expressed AtClpB-cyt/HSP100 protein in rice plants. The transgenic rice lines showed re-growth in the post-high temperature stress recovery phase while the untransformed plants could not recover to the similar extents. Keeler et al [41] reported cloning of a chloroplastic HSP100/ClpB gene from Phaseolus lunatus. The corresponding protein was seen to be localized to chloroplast. The accumulation of ClpB/HSP100 in this plant species was tightly correlated with heat acclimation, suggesting that the chloroplastic ClpB/HSP100 may be an important protein for acquiring thermotolerance. Chloroplastic ClpB from Lycopersicon esculentum has been shown to be important in acquired thermotolerance [42]. Antisense suppression of LeHSP100/ClpB in tomato plants made the plants more heat sensitive. Lee et al [17] noted that Arabidopsis knockout ClpB-p plants turn yellow on heat treatment. ClpB-p mutant plants showed a distinct change in chloroplast morphology upon subjecting plants to high temperature stress: chloroplasts were irregular to roundish in shape and were 21.5% of the size of normal chloroplasts. Lee et al [17] reported that At2g25140 (ClpB-m) was localized to mitochondria based on experiments in which transgenic Arabidopsis plants produced with signal peptide of At2g25140 fused with GFP showed expression of the reporter gene in mitochondria. This group further noted that the ClpB-m transcript was up-regulated during heat shock and the level of induced transcript for ClpB-m was lower as compared to AtHSP100 and ClpB-p transcripts. ClpB-m knockout Arabidopsis plants grew in comparable way to the wild type plants both during induced and acquired thermotolerance.

Plant ClpC proteins, which are the homolog of the E. coli ClpA proteins, have been noted in chloroplast stroma of several plant species [43–45]. ClpCs are considered to be highly conserved proteins among different species [46]. ClpC of Bacillus subtilis has been found not only to be involved in the removal of misfolded and aggregated proteins but also controls through regulated proteolysis, several key steps of development [47]. Two nearly-identical ClpC isomers exist in Arabidopsis. ClpC has been seen to be associated with ClpP in the stroma [48, 49] in an ATP dependent manner [50]. Sjogren et al [51] reported that clpC1 mutant Arabidopsis plants display a retarded growth phenotype and leaves with a homogeneous chlorotic appearance. This mutant also exhibited fewer photosystem I and photosystem II complexes. ClpC in Synechococcus elongatus has been noted to possess ATPase activity as well as function as a molecular chaperone without the need of additional chaperones or adaptor proteins [52]. It is suggested that ClpC functions in part as a housekeeping chaperone in vivo, protecting unfolded, newly synthesized (or recently imported as in chloroplasts) polypeptides from aggregation. ClpD proteins differ from ClpCs by specific signature sequence and by its differential expression characteristics [23, 44]. ClpD in Arabidopsis has previously been referred to as ERD1 [44, 53, 54] and SAG15 [55]. Weaver et al [56] reported that the ERD1 protein declines while its mRNA increases during senescence in Arabidopsis.

It is hence clear that plant Clps are associated with stress and developmental processes. The comprehensive analysis of rice Clp ATPases with respect to their structural and functional aspects has not been addressed to as yet. This study provides information on rice class I Clp ATPases with respect to genomic organization, regulation, protein architecture and cellular functions.

We show in this study that rice genome has 9 entries (i.e. Os02g08490, Os02g32520, Os03g31300, Os04g32560, Os04g33210, Os11g16590, Os11g16770, Os12g12850 and Os05g44340) for the class I Clp ATPase protein family. In Arabidopsis, there are 6 class I Clp ATPase genes which specify 3 ClpB proteins (At1g74310-AtClpB-cyt; At5g15450-AtClpB-m; At2g25140-AtClpB-c), 2 ClpC proteins (C1, At5g50920; C2, At3g48870) and 1 ClpD protein (At5g51070) [17]. From the comparative details, it appears that 9 rice loci specify 3 ClpB (ClpB-cyt, Os05g44340; ClpB-m, Os02g08490; ClpB-c, Os03g31300), 4 ClpC (ClpC1, Os04g32560; ClpC2, Os12g12580; ClpC3, Os11g16590; ClpC4, Os11g16770) and 2 ClpD (ClpD1, Os02g32520; ClpD2, Os04g33210). Phylogenetic analysis using rice, Populus and Arabidopsis sequences showed that class I Clp ATPase sequences are significantly conserved. The tree showed three major clades, one each corresponding to ClpB, ClpC and ClpD proteins (Figure 1). ClpB proteins were further categorized into cytoplasmic, chloroplastic and mitochondrial isoforms based on TargetP, Predotar, LOCTREE and PSORT database. Cytoplasmic ClpB members included OsClpB-cyt, At1g74310 and Pt101.72; chloroplastic ClpB included OsClpB-c, At5g15450, Pt106.69 and Pt108.53 and mitochondrial ClpB included OsClpB-m, At2g25140 and Pt98.74. Although OsClpC3 and OsClpC4 are considered members of ClpC class, these proteins appear somewhat different from OsClpC1 and OsClpC2 proteins. All ClpC proteins identified to date are localized to the chloroplast where they interact with the chloroplastic protease ClpP. From the in silico analysis, it appears that ClpC3 on the other hand may be chloroplast/mitochondria localized while ClpC4 is membrane bound protein. Recently, AAA⁺ ATPase chaperone CoxD protein in Oligotropha carboxidovorans is found to be present exclusively in cytoplasmic membrane [61]. A Clp ATPase in Glycine max has also been found to be membrane associated [62]. High temperature causes modifications in membrane functions, associated with alteration of membrane fluidity. In plant cells, membrane-based processes such as photosynthesis and respiration are especially important. Vigh et al [63] proposed that any alterations in the plasma membrane microdomains are well suited for sensing stress and re-tailoring the expression of various classes of HSPs. The chances of interaction of membrane-associated HSPs with the proteins involved in signaling increases as these molecules are also concentrated in the membrane rafts. ClpC4 protein of rice may have such an implication.

Importantly, OsClpC2 contained 691 nucleotides long intron in the 5'UTR and further harbored five upstream ORFs (uORFs) in the 5'UTR. The role of these uORFs in regulation, if any, needs to be worked out. Both OsClpD1 and OsClpD2 genes towards their 5' ends showed a s terile a lpha m otif (SAM) domain containing protein (Os02g32530 and Os04g33220): MSU rice gene model predicts that OsClpD1 and Os02g32530 gene pair and OsClpD2 and Os04g33220 gene pair may give rise to natural sense/antisense transcripts (NATs). Further analysis of in vivo regulation of ClpD proteins may reflect the significance of this observation. Domain analysis showed that OsClpB-cyt protein contains various motifs which are typical of HSP100 proteins. Spacer signature II motif seen in OsClpB proteins was notably absent from OsClpCs and OsClpDs (Figures 2, 3). Complex formation between ClpA and ClpX with ClpP is reportedly mediated by a helix-loop-helix motif in which the tip of the loop contains an essential IGF/L tripeptide (P-element; [59]). Weibezahn et al [64] engineered a ClpB variant termed BAP (ClpB-ClpP-loop; a variant protein in which ClpP interacting site of ClpA has been engineered into E. coli ClpB), which carried the P-element. Association of ClpP converted BAP from a refolding into a degrading disaggregase, demonstrating that this engineering brings in threading of aggregated proteins into the proteolytic chamber of ClpP. When rice class I Clp ATPase proteins were searched for the presence of IGF motif, it was seen that apart from OsClpB-m, OsClpB-c and OsClpB-cyt, all other proteins contained IGF/L tripeptide motif (Figures 2, 3). This observation may have an implication for differential function of ClpB as compared to other class I Clps. According to Kim et al [59], IGF/L tripeptide consensus is generally located between the sensor I and box VII motifs. However, OsClpC4 contained this motif towards the N-terminus. This may have some relation to the fact that unlike all other Clps, OsClpC4 is the only rice Clp that appears to be associated with the plasma membrane.

Cross-induction of HSP genes by diverse stresses is known [11, 13]. Zou et al [65] have recently shown that the expression of nine different OsHSP genes was affected differentially by heat, cold, salt, desiccation as well as ABA. Hu et al [8] noted an extensive overlap of transcript levels in rice HSFs and HSPs in response to different stresses. Likewise, Mittal et al [9] noted cross-induction of HSF transcripts by diverse stresses in rice seedlings. On the same line, ClpATPase genes showed stress-related transcript expression. Agarwal et al [39] reported that OsClpB-cyt/HSP100 transcript in rice seedlings is induced strongly by heat and this transcript remains unaffected in response to salt stress, water stress, low temperature stress and ABA application. Hu et al [8] noted that OsClpB-cyt was regulated by heat shock as well as drought. In this study, microarray analysis showed that OsClpB-cyt transcript was up-regulated more than 5-folds both during heat stress as well as oxidative stress. OsClpB-cyt transcript was induced after low temperature stress (Figure 4E). This study is the first report for the cold induction of OsClpB-cyt, though at low levels. Wu et al [66] reported that rice HSP100 promoter responds to desiccation stress. In the experimental conditions used in the current study (PEG treatment), we failed to note induction of OsClpB-cyt transcript in response to desiccation stress. OsClpB-c was up-regulated during heat stress by up to 5-folds and during oxidative stress by up to 1.5 folds based on microarray analysis (Figure 4C). Q-PCR data showed that as with OsClpB-cyt, the relative abundance of OsClpB-c transcripts was far more responsive during heat stress as compared to oxidative stress. OsClpB-m gene, like its counterpart gene in Arabidopsis, was found to be heat and oxidative stress up-regulated (Figure 4). Q-PCR data showed that transcript abundance of OsClpB-m was highest during heat stress among the three OsClpB genes. On the other hand, AtClpB-m showed lowest transcript abundance among its different ClpB isoforms [17]. In a study on transcriptional profiling of genes responsive to hormonal treatments in rice, HSP100 transcript was found to be specifically induced in response to ABA [67]. We observe that transcriptionally, OsClpB-m and OsClpB-c respond more to ABA than OsClpB-cyt (Figure 4F).

OsClpC1 transcripts did not undergo any change under stress conditions while OsClpC2 was seen to be up-regulated (~ 3-4 folds) under heat, cold and oxidative stress conditions as seen in microarray experiment although we did not find the same upregulation in semi-quantitative RT-PCR analysis (Figure 5C). No amplicons were detected for OsClpC3 and OsClpC4 in the semi-quantitative RT-PCR analysis. This may indicate that these loci may in fact be transcriptionally silent under the conditions tested. However, in silico analysis of publicly-available microarray database shows OsClpC3 expression in stem, culm and leaf tissues (Figure 5A). Evidence for expression of OsClpC4 is yet awaited. Shen et al [68] reported that OsClpD1 is induced by water deficit and temperature stress in vegetative tissues. We observed that this gene was up-regulated by different stresses. In microarray data, OsClpD2 showed down-regulation during heat and oxidative stresses and a marginal up-regulation in cold stress (Figure 5C).

AtClpB-m and AtClpB-c have been shown to be localized to mitochondria and chloroplast, respectively [17]. We found that OsClpB-m is localized to the same location in the cell as AtClpB-m. Likewise, we noted that OsClpB-c is localized to the same location as AtClpB-c. We thus show that OsClpB-m and OsClpB-c are localized to mitochondria and chloroplast, respectively (Figure 6). It therefore, appears that these proteins may have specific functions in cellular organelles.

Yeast expressing HSP104 are typically 100- to 1000-fold more thermotolerant than yeast lacking HSP104, thus demonstrating the critical requirement for this protein in cell survival during extreme heat stress [69]. Arabidopsis ClpB-Cyt/HSP101 has been shown to overcome the temperature sensitivity of yeast Δhsp104 mutant [38]. Similar observations were noted for the ClpB-Cyt/HSP100 proteins from soybean [70] and rice [39]. This fact prompted us to analyze ability of diverse class I Clp genes in yeast mutant complementation assay. OsClpB-cyt showed reasonably good complementation ability. The capability of OsClpB-m appeared comparable to OsClpB-cyt in conferring tolerance. Our observation actually matches to an earlier experiment done with yeast [71]. This group expressed yeast Hsp78 protein which is mitochondrial isoform of the yeast cytoplasmic Hsp104 protein, in the cytoplasm of the Hsp104 mutant yeast cells. They noted that yeast Hsp78 protein complements the cytoplasmic Hsp104 yeast mutation. This experiment indicates that the organellar and the cytoplasmic ClpBs may have some degree of redundancy in their substrate recognition. Further, the target substrate proteins of different Clps are not yet defined. In due course when target proteins are established, it should be possible to address the question of mitochondrial ClpB complementing the cytoplasmic yeast hsp104 mutation more deeply. A significant observation in this assay test was the fact that OsClpD1 restores the mutation defect of the ScΔhsp104 yeast cells to a significant extent; ScΔhsp104-pGV8-OsClpD1 showed growth comparable to that of the ScΔhsp104-ScHSP104 yeast cells. ClpD expression in Arabidopsis has previously been seen to be associated during senescence [56] but there is no study that links expression of this gene to heat stress. The detailed mechanism behind this role of OsClpD1 remains to be worked out. It should be relevant to analyze how OsClpD1 expression alters transcript profiling in Δhsp104 mutant yeast cells. It should also be relevant to generate transgenic plants with ectopically higher ClpD1 protein and analyze their stress phenotype. Weibezahn et al [64] found that the development of thermotolerance in E. coli relies on the ClpB/KJE-mediated reactivation of aggregated proteins, whereas the removal of protein aggregates by degradation does not confer thermotolerance. Similarly in yeast, Tessarz et al [72] noted that the degradation of aggregated proteins results in loss of thermotolerance. They reported that the removal of aggregated proteins by HAP and ClpP is not sufficient for yeast cells to survive a severe heat shock, but survival demands on the reactivation of aggregated proteins. It can thus be inferred that disaggregation and reactivation of proteins is essential for cell survival after extreme heat stress, whereas the degradation of aggregated proteins is insufficient. It is possible that expression of OsClpD1 does this activity more efficiently than other class I Clps and hence its expression results in more thermotolerance than noted for other rice class I Clp proteins. Finally, it needs to be mentioned that although different Clp ATPases are noted to complement yeast hsp104 mutation in this study, it is possible that variations in expression levels of rice Clps in yeast which have not been determined, may also affect complementation efficiencies.

Finally, we wish to herein draw a comparable picture of Arabidopsis and rice class I Clp ATPases. There are some major differences in these two species for the number of ClpB, ClpCs and ClpDs nuclear genes. While Arabidopsis has 2 and 1 ClpC and ClpD members, respectively, rice appears to contain 4 and 2 ClpC and ClpD members, respectively. On the other hand, number of ClpB proteins appears same (3) in both rice and Arabidopsis. Both in rice and Arabidopsis, one each of ClpB type is present in cytoplasm, chloroplasts and mitochondria. It should be relevant to extend the above observations further to analyze what implications these might have in regulating heat response in these two contrasting plant species.