The Arabidopsis STRESS RESPONSE SUPPRESSOR DEAD-box RNA helicases are nucleolar- and chromocenter-localized proteins that undergo stress-mediated relocalization and are involved in epigenetic gene silencing
French Associates Institute for Agriculture and Biotechnology of Drylands, Jacob Blaustein Institutes for Desert Research, Ben-Gurion University of the Negev, Midreshet Ben-Gurion, Israel
DEAD-box RNA helicases are involved in many aspects of RNA metabolism and in diverse biological processes in plants. Arabidopsis thaliana mutants of two DEAD-box RNA helicases, STRESS RESPONSE SUPPRESSOR1 (STRS1) and STRS2 were previously shown to exhibit tolerance to abiotic stresses and up-regulated stress-responsive gene expression. Here, we show that Arabidopsis STRS-overexpressing lines displayed a less tolerant phenotype and reduced expression of stress-induced genes confirming the STRSs as attenuators of Arabidopsis stress responses. GFP–STRS fusion proteins exhibited localization to the nucleolus, nucleoplasm and chromocenters and exhibited relocalization in response to abscisic acid (ABA) treatment and various stresses. This relocalization was reversed when stress treatments were removed. The STRS proteins displayed mis-localization in specific gene-silencing mutants and exhibited RNA-dependent ATPase and RNA-unwinding activities. In particular, STRS2 showed mis-localization in three out of four mutants of the RNA-directed DNA methylation (RdDM) pathway while STRS1 was mis-localized in the hd2c mutant that is defective in histone deacetylase activity. Furthermore, heterochromatic RdDM target loci displayed reduced DNA methylation and increased expression in the strs mutants. Taken together, our findings suggest that the STRS proteins are involved in epigenetic silencing of gene expression to bring about suppression of the Arabidopsis stress response.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
RNA helicases are ubiquitous proteins that are found in both prokaryotes and eukaryotes (Rocak and Linder, 2004). The largest RNA helicase family comprises the DEAD-box RNA helicases of which there are 58 members in the Arabidopsis genome. They consist of a helicase core that contains at least 12 characteristic motifs at conserved positions that are involved in processes central to helicase activities, including ATP hydrolysis and double-stranded RNA-unwinding (Banroques et al., 2011; Linder and Jankowsky, 2011; Putnam and Jankowsky, 2013). One motif contains the characteristic Asp–Glu–Ala–Asp (DEAD) sequence from which this class of RNA helicases obtains its name. N-terminus and C-terminus extensions are thought to play roles in substrate specificity and/or regulation of the helicase activity (Aubourg et al., 1999; Banroques et al., 2011; Linder and Jankowsky, 2011). DEAD-box RNA helicases are involved in many aspects of RNA metabolism, often as part of supramolecular complexes where they function to remodel RNA and ribonucleoprotein structure via RNA duplex unwinding activity, promotion of duplex formation and displacement of proteins from RNA (Jankowsky et al., 2001; Fairman et al., 2004; Chamot et al., 2005; Bowers et al., 2006; Putnam and Jankowsky, 2013). In addition, they may act as RNA clamps, providing nucleation centers that function as assembly platforms for large ribonucleoprotein complexes (Ballut et al., 2005; Nielsen et al., 2009; Putnam and Jankowsky, 2013).
Growing numbers of plant DEAD-box RNA helicases have been characterized, and research has revealed their roles in ribosome biogenesis (Gendra et al., 2004; Chi et al., 2012; Hsu et al., 2014), pre-mRNA splicing (Guan et al., 2013), exon-junction complex functions (Koroleva et al., 2009), mRNA export (Gong et al., 2005; Kammel et al., 2013), nonsense-mediated mRNA decay (Yoine et al., 2006a; Shi et al., 2012), chloroplast/mitochondrial intron splicing (Kohler et al., 2010; Asakura et al., 2012), and siRNA-mediated gene silencing (Kasschau et al., 2007). Furthermore, it has been shown that these helicases play roles in diverse biological processes that include plant germline development (Schmidt et al., 2011), female gametogenesis (Liu et al., 2010), pollen tube guidance (Shimizu et al., 2008), seed size (Yoine et al., 2006b), cell division and cell growth (Vain et al., 2011), seedling development and growth (Kanai et al., 2013), sugar signaling (Yoine et al., 2006a; Hsu et al., 2014), apoptosis (Li et al., 2011), and pathogen infection (Huang et al., 2010).
Several DEAD-box RNA helicases have been associated with abiotic stress responses (Gong et al., 2002, 2005; Vashisht and Tuteja, 2006; Kant et al., 2007; Kim et al., 2008; Li et al., 2008; Liu et al., 2008; Chung et al., 2009; Koroleva et al., 2009; Macovei et al., 2012; Guan et al., 2013; Tuteja et al., 2013). STRESS RESPONSE SUPPRESSOR1 (STRS1) and STRS2 are two such RNA helicases that were identified in a functional genomics screen (Kant et al., 2007, 2008). STRS gene expression is rapidly down-regulated by multiple stresses and by abscisic acid (ABA), while mutations in either gene cause increased tolerance to salt, osmotic and heat stresses, as well as enhancement of stress-induced gene expression. Taken together, the findings suggest that STRS1 and STRS2 are part of the ABA-independent and ABA-dependent stress regulatory networks and function to attenuate the stress-induced expression of transcriptional activator genes.
In this study, we show that overexpression of either the STRS1 or STRS2 gene led to reduced expression of stress-induced transcriptional activators and conferred a phenotype of diminished stress tolerance that is consistent with the role of STRS1 and STRS2 as attenuators of gene expression. STRS promoter:reporter gene experiments revealed that a cis-acting element(s) in the STRS transcribed region is required for proper organ-specific expression. GFP–STRS fusion proteins and fluorescence in situ hybridization (FISH) assays demonstrated that both STRS1 and STRS2 proteins have peculiar sub-nuclear localization at the nucleolus and at chromocenters with a lower level in the nucleoplasm. Moreover, the STRS proteins exhibited rapid relocalization in response to multiple abiotic stresses that was reversible upon stress removal. We further show that the STRS proteins exhibited mis-localization in specific gene-silencing mutants and possess RNA-dependent ATPase and RNA-unwinding activities. Heterochromatic target loci of the RNA-dependent DNA methylation pathway displayed decreased DNA methylation and increased expression in the strs mutants; these findings suggest a role for the STRSs in epigenetic gene silencing.
Overexpression of STRS1 or STRS2 confers a phenotype of diminished stress tolerance and reduced expression of stress-responsive transcriptional activators
We generated transgenic lines that overexpressed STRS1 and STRS2 (STRS–OX lines) under the control of the CaMV 35S promoter in the strs mutant background, to further strengthen the contention that the STRSs are negative regulators of stress responses. At least four T3 homozygous transgenic lines that expressed a single copy of either STRS1 or STRS2 were assessed by real-time polymerase chain reaction (PCR) for the level of STRS overexpression (Table S1). All transgenic lines exhibited higher expression of the STRS genes compared with wild-type (WT) plants (Figure S1a,b). We selected the STRS1–OX line 6 and the STRS2–OX line 2 for further study because both lines displayed persistent overexpression under salt stress (Figure S1c,d).
Because the strs mutants are tolerant to stress, we expected that the STRS–OX lines would display reduced tolerance. The germination rate of WT and mutant lines under stress was compared with the respective germination rate under control conditions. Figure 1(a) shows that seeds of both STRS–OX lines indeed exhibited a greater salt-stress-mediated reduction in germination (e.g. approximately 85% under 150 mm NaCl) than WT (67% reduction). Seeds of the STRS–OX lines were also less tolerant than WT seeds to heat stress during germination (Figure 1b). Accordingly, a 10 h heat stress treatment of 45°C led to a 64 and 67% reduction in germination index of STRS1–OX and STRS2–OX lines, respectively, whereas the WT germination index only fell by 44%.
The strs mutants have been previously shown to display up-regulated expression of stress-responsive genes that encode transcriptional activators (Kant et al., 2007). Although the expression of the transcription factor genes AtMYC2, DREBA1/CBF3, DREB2A and the DREB/CBF target gene RD29A was induced by salt stress in all plant lines examined (Figure 2a), in STRS–OX plants induction of gene expression was reduced significantly compared with WT; peak gene expression at 6 h in the STRS–OX lines was only about 50% of WT. Expression of the housekeeping gene ACTIN2 (ACT2) was unaffected by stress treatments in STRS–OX or WT plants, a finding that suggests a specific role for the STRS proteins in regulating stress-induced gene expression. Similarly, heat-induced expression of genes that encode heat shock proteins HSP101, HSF4 and HSF7 (Prandl et al., 1998; Hong and Vierling, 2000; Queitsch et al., 2000; Kant et al., 2007) was specifically reduced in the STRS-OX lines (Figure 2b). Taken together, our results reaffirm our hypothesis that the STRS proteins function as negative regulators of the Arabidopsis response to multiple abiotic stresses.
Upstream and transcribed regulatory regions determine organ-specific STRS1 and STRS2 expression
In order to examine organ-specific STRS expression, we generated transgenic Arabidopsis plants that expressed the uidA (GUS) reporter gene under the control of STRS upstream DNA. The STRS promoters are unusually small (approximately 200 bp upstream intergenic region), so plants were produced that expressed GUS under the control of 211 or 856 bp (STRS1) and 241 or 1104 bp (STRS2) upstream DNA. Several independent lines of T3 homozygous, single copy proSTRS1:uidA and proSTRS2:uidA plants were stained histochemically for GUS activity. No GUS staining could be detected in any organ of the proSTRS:uidA lines except in pollen grains (Figures 3a–e and S2). The ‘promoterless’ GUS line showed no staining of any part of the plant (Figure 3f), this result suggested that GUS staining in the anthers was specific to the putative STRS promoters.
The observed proSTRS:uidA expression did not agree with previous qPCR analysis showing STRS expression in leaves, roots, stems and inflorescences (Kant et al., 2007). Several Arabidopsis DEAD-box RNA helicases require both the promoter region and elements within the gene body (e.g. 5′-UTR, introns) for proper expression (Mingam et al., 2004). In Arabidopsis WT plants that express full-length genomic STRS–uidA fusions and driven by native STRS upstream DNA fragments (proSTRS:gSTRS–uidA lines), GUS activity could be observed in all organs examined (Figures 3g–n and S2), while constitutively high GUS expression was observed in a proCaMV 35S:uidA line (Figures 3p,q and S2). These findings thus confirm previous qPCR results (Kant et al., 2007) and indicate that a regulatory element(s) within the STRS transcribed region is required for proper organ-specific expression.
The STRS proteins exhibit peculiar sub-nuclear localization
Inspection of the speckled proSTRS:gSTRS–uidA staining pattern suggested sub-cellular STRS localization in the nuclei (Figure 3r,s). We therefore generated transgenic Arabidopsis plants that express pro35S:GFP–STRS in a WT background. At least four homozygous, single copy T3 lines were produced and two lines that exhibited constitutive overexpression under control and salt stress conditions of either GFP–STRS1 or GFP–STRS2 RNA and protein, were selected for further study (Figure S3a–c). Notably, the STRS proteins were still functional when fused to green fluorescent protein (GFP) as demonstrated by reduced tolerance of pro35S:GFP–STRS seeds during germination under either salt or heat stress, similar to that of the STRS–OX lines (Figures 1 and S3d,e).
Figure 4(a,e) shows that the GFP signal in protoplasts transformed with the respective pro35S:GFP–STRS plasmids was localized to the nucleolus with a weaker signal in the nucleoplasm. When GFP was expressed alone, it was observed all over the protoplast (Figure 4i–k) thereby confirming that the nucleolar/nucleoplasmic-localized GFP signal was due to fusion with the STRS proteins. Protoplasts that expressed pro35S:GFP–STRS were also stained with 4′,6-diamidino-2-phenylindole (DAPI), which stains the nucleus blue but does not stain the nucleolus. No overlap of DAPI staining and GFP signal was observed (Figure 4b–d,f–h), supporting nucleolar STRS localization. pro35S:GFP–STRS transgenic plants exhibited the same nucleolar/nucleoplasmic STRS localization (Figure 4l–q). Finally, co-localization of STRS–GFP proteins with the nucleolar marker protein FIBRILLARIN 1 (FIB1) fused to red fluorescent protein (pro35S:FIB1–RFP) confirmed STRS nucleolar localization (Figure 4r–w). It is well accepted that use of the 35S promoter for GFP localization studies does not affect protein localization (Tian et al., 2004). Nevertheless, we employed the knowledge gained from our GUS studies, which showed that an element in the STRS transcribed region is required for proper regulation of STRS gene expression to generate a proSTRS2:gSTRS2–GFP construct. Transient expression in protoplasts demonstrated the same nucleolar/nucleoplasmic localization as the 35S:STRS2–GFP construct (Figure S4).
The STRS proteins exhibit stress-mediated nucleolar–nucleoplasmic relocalization
The nucleolus is highly dynamic and there is a continuous exchange of nucleolar proteins with the nucleoplasm and cytoplasm (Shaw and Brown, 2011). To examine whether the STRSs also exhibit relocalization from the nucleolus in response to stress, 7-day-old pro35S:GFP–STRS seedlings were transferred from Murashige and Skoog (MS) medium to fresh MS medium that was supplemented with 200 mm NaCl. Approximately 30 min after transfer, fluorescence was no longer observed in the nucleolus and was distributed uniformly throughout the nucleoplasm (Figure 5a). However, no relocalization of the GFP–STRS signal was observed when seedlings were transferred from MS control plates to fresh MS control plates (Figure 5a, right panel); this result confirmed that nucleolar–nucleoplasmic relocalization had occurred in response to salt stress. Salt stress-mediated co-localization of GFP–STRS with nucleoplasmic-localized RFP–eIF4AIII and DAPI-stained nuclei (Koroleva et al., 2009) provided further confirmation of salt stress-mediated STRS relocalization to the nucleoplasm (Figure S5).
To capture the kinetics of STRS relocalization, 10-day-old pro35S:GFP–STRS seedlings were transferred from control MS plates to a coverslip chamber, which was flushed with either liquid MS or liquid MS supplemented with 200 mm NaCl. Relocalization of the STRS–GFP signal could be observed within 2 min after application of salt stress and indicated rapid STRS relocalization from the nucleolus to the nucleoplasm (Figure 5b). Re-flushing with fresh liquid MS led to a gradual recovery to the nucleolus. Accordingly, 7 min after re-flushing, the GFP–STRS signal re-localized into nucleoplasmic speckles. No recovery of GFP–STRS1 into the nucleolus was observed at this time, although some recovery of GFP–STRS2 could be seen. By 14 min after re-flushing, both GFP–STRS signals could be seen in the nucleolus in addition to nucleoplasmic speckles. These speckles disappeared gradually over the next 16 min until, by 30 min, the GFP–STRS signals resembled the control cells. Continuous salt stress for 1 h did not lead to reappearance of the GFP–STRS signal in the nucleolus (Figure 5c); this finding demonstrated that STRS relocalization back from the nucleoplasm to the nucleolus is part of the recovery process from stress. Notably, fibrillarin, a component of a nucleolar small nuclear ribonucleoprotein (Reichow et al., 2007), remained in the nucleolus in both unstressed and salt-stressed seedlings (Figure 5d); this finding excluded the possibility of a stress-induced breakup of the nucleolus and affirmed that relocalization is an inherent feature of STRS proteins. Furthermore, the fast nucleolar–nucleoplasmic shuttling suggested that the STRS proteins are loosely bound at nucleolar sites. This suggestion was confirmed by fluorescence recovery after photobleaching (FRAP) assays (Zemach et al., 2006), which showed that the GFP–STRS signal in the nucleolus recovered to about 85% of its pre-bleached level in about 10 sec (Figure S6).
Nucleolar–nucleoplasmic STRS relocalization was also induced following ABA application, osmotic stress and exposure to heat (Figure 5e–g). However, whereas salt stress caused relocalization to the nucleoplasm within 2 min (Figure 5b), relocalization took approximately 45 min in response to ABA and osmotic stress, and 60 min in response to heat stress. Furthermore, the GFP–STRS signal re-localized transiently to a few chromosomal sites that were likely to be chromocenters (Figure 5e,f: 15, 25 or 35 min) in response to ABA or osmotic stress. Re-flushing with fresh liquid MS led to localization of the GFP–STRS protein back to the nucleolus at approximately 10, 16 or 30 min after ABA, osmotic stress or heat stress, respectively; no nucleoplasmic speckles were observed as was seen after recovery from salt stress (see Figure 5b).
The STRS proteins also localize to the chromocenters and exhibit relocalization in response to stress
Inspection of GFP–STRS fluorescence in the nucleoplasm indicated that the STRSs may also be localized to chromocenters (Figures 6a,b and S7a,b), intense DAPI-stained regions of heterochromatic DNA that are generally transcriptionally inactive (Fransz et al., 2002). Chromocentric STRS localization was confirmed by immunolabeling using anti-GFP against nuclei isolated from unstressed WT and GFP–STRS plants followed by FISH with fluorescein-12-dUTP-labeled centromeric 180-bp repeats (CEN180; Avivi et al., 2004). Figures 6(c,d) and S7(c,d) show that the GFP–STRS proteins co-localized with CEN180 repeats as well as with the nucleolus. However, no chromocenter-associated GFP signal was observed in nuclei of plants that had been subjected to 1 h of salt stress (Figures 6f and S7f). It is unlikely that this result was due to stress-mediated break-down of the chromocenters because the CEN180 signal was clearly visible in nuclei from stressed WT and pro35S:GFP–STRS1 plants (Figures 6e,f and S7e,f). However, the GFP signal accumulated in the nucleolar rim suggesting possible relocalization from the chromocenters to this sub-nuclear region.
The STRS proteins display mis-localization in specific gene-silencing mutants
The chromocentric STRS localization suggested that the STRSs may be involved in gene silencing, a notion that would be consistent with their role as negative regulators of gene expression (Figure 2; Kant et al., 2007) and their stress-mediated relocalization out of the chromocenters (Figure 6). To test this hypothesis, we obtained 17 Arabidopsis mutant plants that are defective in genes selected in accordance with their similar localization pattern to the STRSs (e.g. Pendle et al., 2005), their involvement in gene silencing, mutant plants that exhibited increased/reduced stress tolerance or displayed a high co-expression correlation coefficient with the STRSs (CBF5; Lermontova et al., 2007). Protoplasts from each mutant were transformed with the pro35S:GFP–STRS constructs to determine whether any of the mutations led to aberrant STRS localization. Details of the various mutants examined are provided in Table 1.
Table 1. Mutants employed for analysis of STRS mis-localization
Putative pseudouridine synthase
Possible function in small nuclear ribonucleoproteins (snoRNPs; Lermontova et al., 2007)
RNase III (Dicer-like)
Cleavage of double-stranded RNA to 24 nt siRNAs (Xie et al., 2004)
SNF2-like chromatin remodeling protein
De novo RNA-directed DNA methylation at CG, CNG and CNN sites (Kanno et al., 2004, 2005).
Histone deacetylase HD2-type
DNA methylation, response to salt and ABA stress (Luo et al., 2012)
Histone H3 lysine 9 methyltransferase
Histone methylation, maintainance of non-CG methylation via CMT3 (Malagnac et al., 2002; Lindroth et al., 2004)
DNA-binding, methylated histone residue binding
Histone H3-K9 methylation, RNA interference, negative regulation of flower development, vernalization response (Nakahigashi et al., 2005; Exner et al., 2009)
RNA-dependent RNA polymerase
Synthesis of double-stranded RNA in RNAi pathway (Law et al., 2010, 2011)
Recruitment of 24 nt siRNAs in RISC effector complex (Zilberman et al., 2003; Ye et al., 2012)
DNA and protein binding
DNA repair, protein degradation, photomorphogenesis (Yanagawa et al., 2004)
Suppressor of DRM1, DRM2 and CMT3 involved in CG maintenance methylation, CNG methylation, de novo DNA methylation (CG/CNG/CNN; Cao and Jacobsen, 2002; Henderson and Jacobsen, 2008)
DEAD-box RNA helicase
Constituent of exon-junction complex, mRNA processing, translation initiation and response to hypoxia (Koroleva et al., 2009)
Largest subunit of Pol IVa
Generation of 24 nt siRNAs for RNAi-mediated gene silencing (Haag and Pikaard, 2011)
RNA and protein binding
RNA and mRNA splicing via spliceosome, sugar mediated signaling. Possible exon-junction complex component (Koroleva et al., 2009; Zhang and Mount, 2009)
Possible component of U3 snoRNP, zygote cell division plane (Li et al., 2010)
STRS2 exhibited perturbed localization in three mutants, drd1, dc13 and rdr2 (Figure 7), that are part of the RNA-directed DNA methylation (RdDM) gene-silencing mechanism at repetitive heterochromatic loci (Kanno et al., 2004, 2005; Law et al., 2010, 2011; Castel and Martienssen, 2013). STRS2 localization was also perturbed in a mutant of the SUVH4/KYP gene that is involved in maintenance of histone (H3K9) methylation (Jackson et al., 2002; Malagnac et al., 2002) and may also be linked to RdDM (Castel and Martienssen, 2013). Finally, aberrant STRS2 localization was observed in a mutant of the LHP1 gene that has a role in H3K27 methylation-induced gene silencing in euchromatin (Nakahigashi et al., 2005; Exner et al., 2009). STRS1 exhibited WT localization in the drd1, dc13, rdr2, suvh4/kyp and lhp1 mutants but displayed perturbed localization in the hd2c mutant. HD2C is a histone deacetylase that is involved in ABA and salt stress responses and that interacts with HDA6, a histone deacetylase functioning in heterochromatic DNA silencing (Earley et al., 2006; To et al., 2011; Luo et al., 2012). Both STRS1 and STRS2 exhibited aberrant localization in the cbf5 mutant. CBF5 displays nucleolar and Cajal body localization, and may be involved in small ribonucleoprotein splicing functions (Lermontova et al., 2007). However, whereas STRS1 localized to nuclear bodies in cbf5, STRS2 exhibited uniform nucleoplasmic localization. Both STRS1 and STRS2 localization was also perturbed in a mutant of At3g58660 that encoded an L1p/L10e ribosomal protein. In mutants of AGO4 (part of the RdDM machinery (Castel and Martienssen, 2013)), eIF4A-III (a DEAD-box RNA helicase and part of the exon-junction complex (Koroleva et al., 2009)) and the sdc mutant (a drm1 drm2 cmt3 triple mutant lacking CHG and CHH methylation (Henderson and Jacobsen, 2008)), both STRSs displayed WT localization.
Transcriptional silencing and DNA methylation at heterochromatic loci is reduced in the strs mutants
To assess whether the STRSs function in gene silencing, we compared transcript levels of the AtSN1 and AtCopia2 transposons, and the intergenic repeat sequence solo-LTR, in WT and the strs mutants, as a measure of RdDM-mediated transcriptional silencing of these heterochromatic loci (Zheng et al., 2009). Figure 8(a) shows that AtSN1 and solo-LTR exhibited increased expression in the strs mutants compared with WT, which indicated de-repression of these loci (Xie et al., 2004; Herr et al., 2005; Huettel et al., 2006). AtCopia2 expression only displayed up-regulation in the strs2 mutant. Furthermore, digestion of genomic DNA with the methylation-sensitive HaeIII restriction enzyme followed by PCR amplification with primers specific for AtSN1 demonstrated reduced DNA methylation at this locus in the strs mutants (Figure 8b). Methylation was particularly reduced in strs2; this result was consistent with greater de-repression of transcriptional silencing in this mutant. Overall, the results support our contention that STRS1 and STRS2 may both play a role in RdDM-mediated gene silencing.
STRS1 and STRS2 are functional DEAD-box RNA helicases
STRS1 and STRS2 possess all the conserved motifs of DEAD-box RNA helicases (Kant et al., 2007), which have been implicated in RNA interference (RNAi) pathways (reviewed in Ambrus and Frolov, 2009). To determine whether these proteins possess intrinsic RNA-dependent ATPase activity, purified His-tagged STRS proteins (Figure S8a) were employed in a coupled spectrophotometric assay (Panuska and Goldthwait, 1980). In the absence of RNA and STRS protein, or with RNA alone, only a slight reduction in ATPase activity (detected as a reduction in A338) was observed, while the decrease in absorbance was higher when both RNA and STRS protein were included in the reaction (Figure 9a). Thus, both STRS1 and STRS2 possess an RNA-dependent ATPase activity.
To examine whether the STRSs exhibit RNA-unwinding activity, the ability of purified His-tagged STRS protein (Figure S8b–e) to unwind a fluorescently labeled synthetic RNA duplex was examined. The appearance of single-stranded RNA and the concomitant disappearance of the 18-/36-mer RNA duplex demonstrated that the STRSs indeed possess an ATP/Mg2+-dependent RNA-unwinding activity (Figures 9b–e and S8f). This activity was not the result of contaminating E. coli protein (Figure S9). Only 0.1 μg of STRS2 protein was required to dissociate fully the RNA duplex in 10 min, whereas 2.5 μg STRS1 protein could not complete unwinding by 40 min; this finding suggested that STRS2 may possess higher helicase activity than STRS1. We also showed that purified GFP–STRS proteins retained RNA-unwinding activity (Figure S3f,g), thereby providing further confirmation that the GFP tag does not impede STRS function. Collectively, our results establish the STRSs as bona fide RNA helicases.
STRS1 and STRS2 are negative regulators of Arabidopsis abiotic stress responses
In this study, we further supported the contention that STRS1 and STRS2 are negative regulators of Arabidopsis abiotic stress responses as evidenced by diminished stress tolerance and reduced stress-induced gene expression in STRS overexpressors. A similar reduction in germination of STRS2-/RH25-overexpressing lines under salt stress has also been reported by Kim et al. (2008). STRS1 and STRS2 are two of several negative regulators of abiotic stress responses (Xiong et al., 1999, 2001, 2002; Lee et al., 2001; Kariola et al., 2006; Seo et al., 2012; Nishiyama et al., 2013). Mutants that are defective in genes such as FIERY2, ERD15, ATPUBs and AHPs also exhibit increased stress tolerance and enhanced expression of stress-responsive transcription factors. Thus, negative regulation or attenuation of stress-induced gene expression is an important component of the plant response to stress. Indeed, constitutive activation of the stress response by overexpression of DREB1A/CBF3 leads to severe retardation of growth in unstressed plants (Liu et al., 1998); this result suggests that negative regulators are necessary to prevent over-activation of the stress response e.g. by transient stress conditions, thereby avoiding growth arrest.
We did not test the effect of STRS overexpression on genes whose expression is normally down-regulated by stress. This experiment will be important to perform in the future because it could discern whether the STRSs are involved in stress perception, in which case the STRS-overexpressing lines might be expected to exhibit higher transcript levels of genes that are normally down-regulated by stress. Conversely, results showing no effect of STRS overexpression on these genes or lower transcript levels under control conditions would suggest that the suppressor function of the STRSs only relates to transcriptional activation of up-regulated genes.
Sub-nuclear STRS localization suggests that the nucleolus and chromocenters are the sites of STRS function
Our results showed clear localization of the STRSs in the nucleolus and chromocenters under control conditions. Although the nucleolus is classically recognized as the site of ribosome assembly, it has become clear that the nucleolus is involved in numerous molecular and cellular functions, including RNA processing, siRNA mechanisms, biogenesis of ribonucleoproteins (RNP) such as the signal recognition particle and telomerase RNP, sequestration and release of specific proteins, regulation of the cell cycle, stress responses, telomerase activity, and RNA interference (Jacobson and Pederson, 1998; Wong et al., 2002; Pontes et al., 2006; Boulon et al., 2010; Shaw and Brown, 2011). Thus, of the over 4500 human nucleolus-associated proteins that have been identified by proteomic experiments, only approximately 30% have a function that is clearly related to ribosome production (Ahmad et al., 2009; Boulon et al., 2010). In plants, 217 nucleolar proteins have been identified, including a range of proteins with no function in ribosome biogenesis (Pendle et al., 2005). Thus, our data showing nucleolar STRS localization is consistent with the idea of a plurifunctional nucleolus (Pederson and Tsai, 2009).
Results from our FRAP studies that showed high mobility of the nucleolar-localized STRSs are in agreement with high levels of nucleolar/nucleoplasmic protein trafficking (Phair and Misteli, 2000; Chen and Huang, 2001; Olson and Dundr, 2005). Most striking, though, was the fast multiple stress-mediated nucleolar and chromocenter STRS relocalization. It is unclear why each stress treatment led to different kinetics of STRS relocalization from the nucleolus – by 2 min under salt stress but 45 min under osmotic stress or ABA treatment. However, down-regulation of STRS gene expression in response to stress also shows different stress-dependent kinetics whereby STRS expression reaches its nadir after 12 h drought stress, 6 h salt stress or 2 h ABA treatment (Kant et al., 2007). Nevertheless, stress-mediated STRS relocalization implies that the STRSs act in the nucleolus and chromocenters to bring about attenuation of the stress response and that early, in response to stress, they are removed from their site of function to the nucleoplasm. Furthermore, the kinetics of stress-mediated relocalization of the STRSs is faster than stress-mediated down-regulation of STRS gene expression; this finding suggests a need to rapidly inactivate the STRSs by relocalization, which is further augmented by a decrease in STRS gene expression. The fast stress-mediated relocalization of proteins such as the STRSs and eIF4AIII (which exhibits fast hypoxia-mediated relocalization from the nucleoplasm to the nucleolus (Koroleva et al., 2009)) supports evidence in other organisms that the nucleolus is a central hub in stress signaling (Boulon et al., 2010).
The STRSs are involved in gene silencing
The fact that the STRSs are negative regulators of gene expression and are localized in chromocenters – regions of condensed, transcriptionally inactive chromatin characterized by repetitive DNA sequences, increased levels of DNA and histone methylation, and decreased levels of histone acetylation (Fransz et al., 2002; Fischer et al., 2006) – led us to postulate that the STRSs may be involved in gene silencing. This hypothesis was supported by observations that: (i) the STRS proteins were mis-localized in mutants that were defective in small RNA-mediated gene silencing, such as RdDM; and (ii) heterochromatic RdDM target sequences exhibited increased expression in the strs1 and strs2 mutants while a transposon target displayed reduced DNA methylation. Interestingly, STRS1 and STRS2 localization was perturbed in different mutants, which suggests that either each STRS protein is targeted to a different protein complex acting at chromocenters and/or the nucleolus, or that both STRS proteins are targeted to the same protein complex via interactions with different partner protein(s).
Intriguingly, STRS2 mis-localized in the rdr2 and dc13 mutants but not in the ago4 mutant; this result suggests a role for STRS2 in the early steps of the RdDM pathway, linking RDR2 production of dsRNA with its subsequent cleavage into 24 nt siRNAs by DCL3 (Kasschau et al., 2007; Law et al., 2011). The fact that RDR2 and DCL3 localize to the nucleolus in a putative RNA processing center (Pontes et al., 2006) further supports a role for STRS2 in RdDM. Although the actual helicase function of STRS2 is only speculative at present, possible roles include separation of the guide strand from the passenger strand and unwinding the guide strand from its target RNA. DICER proteins also possess a helicase domain (Collins and Cheng, 2005), which is critical for their function (Liu et al., 2012a). RNA helicases involved in RNAi may also possess RNA-unwinding activity-independent functions for example in remodeling the RNA-induced silencing complex (RISC) to facilitate dsRNA loading (Ambrus and Frolov, 2009). STRS2 also mis-localized in the drd1 mutant. DRD1 is part of the DDR complex, which is required for RNA Pol V localization to chromatin during RdDM (Law et al., 2010; Zhong et al., 2012). Furthermore, DRD1 co-localizes with the RNA Pol IVa subunit to endogenous repeat loci and is concentrated in chromocenters (Pontes et al., 2006). Thus, we have a possible link between STRS2 localization in the nucleolus and in chromocenters, and its putative function in RdDM. STRS2 mis-localization in the suvh4/kyp and lhp1 mutants also provide further possible links to RdDM and to other aspects of gene silencing (Lindroth et al., 2004; Nakahigashi et al., 2005; Zemach et al., 2006; Exner et al., 2009).
Another interesting observation was STRS1 mis-localization in the hd2c mutant. HD2C is a histone deacetylase that can interact physically with the nuclear histone deacetylase, HDA6 (Luo et al., 2012). HDA6 has been implicated in RdDM and is important for establishment of transcriptionally repressive CG methylation in cooperation with METHYLTRANSFERASE1 (MET1; Aufsatz et al., 2002; Earley et al., 2010; To et al., 2011; Liu et al., 2012b). Importantly, both HD2C and HDA6 are involved in ABA and salt stress responses (Chen et al., 2010; Luo et al., 2012). Thus, STRS1 could be linked with gene silencing via involvement in repressive histone modifications (deacetylation-coupled methylation) and DNA methylation associated with stress responses.
In conclusion, we have provided evidence that the STRSs are functional DEAD-box RNA helicases that are involved in heterochromatic gene silencing, a function consistent with their specific sub-nuclear localization and with their role as negative regulators of stress-responsive gene expression. Our findings add to the growing evidence for the involvement of RdDM in the regulation of stress-induced transcription (Popova et al., 2013).
Plant materials and growth conditions
WT Arabidopsis and strs mutants (Col-0) were used to generate transgenic plants. Seeds of the following T-DNA insertion mutants (Alonso et al., 2003) were obtained from the Arabidopsis Biological Research Center (ABRC; The Ohio State University, Columbus, OH): aly4 (SALK_111903), ddb1a (SALK_055584C), eif4a-III (SAIL_174_A01), hd2c (SALK_129799C), hsp101 (SALK_099583C), 11p/110e (SALK_132996), rnps1 (SALK_123442C), rp13 (SALK_040503), yao (SALK_022234). The ago4, cbf5, dc13, ddm1, drd1, lhp1, nrpd1a, rdr2, sdc and suvh4/kyp mutants have been described previously (Vongs et al., 1993; Gaudin et al., 2001; Cao and Jacobsen, 2002; Jackson et al., 2002; Zilberman et al., 2003; Kanno et al., 2004; Xie et al., 2004; Herr et al., 2005; Lermontova et al., 2007). Seeds of FIB1–GFP (NASC ID: N799457) were obtained from the Nottingham Arabidopsis Stock Centre. For plate and soil experiments, seeds were germinated and grown in accordance with Kant et al. (2007). For gene expression experiments, 10-day-old seedlings were transferred from MS plates to soil at a density of 6–10 plants per pot.
DNA constructs and plant transformation
For primer sequences, cloning and plant transformation see Methods S1 and Table S1. All transformants employed for analysis were T3 generation plants harboring a single homozygous copy of the transgene.
Abiotic stress tolerance assays
Salt stress and basal thermotolerance germination assays as well as analysis of stress-responsive gene expression were performed in accordance with Kant et al. (2007) except that plants were exposed to 45°C for heat-induced gene expression. Four replicate plates were used per treatment. Stress germination assays were repeated independently at least three times while gene expression experiments were repeated twice. Statistical analysis was performed with Tmev software (Saeed et al., 2003).
Quantitative real-time PCR
Isolation of total RNA, preparation of cDNA, primer design, qPCR and data analysis were performed in accordance with Kant et al. (2006) except that reactions contained 5 μl PerfeCTa® SYBR® Green Fast Mix® (Quanta Biosciences, Gaithersburg, MD, USA), 40 ng cDNA and 100–500 nm of gene-specific primer in a final volume of 10 μl (see Methods S1). See Table S1 for primer sequences.
Histochemical staining of GUS activity
GUS activity was detected by histochemical staining of 2.5-week-old plant material in accordance with Vitha et al. (1995). Following staining, tissue was stored in 50% glycerol at room temperature before being mounted in 100% glycerol on microscopic slides. Tissue was photographed using a Nikon Eclipse TE2000-U microscope with a digital camera DXM1200F or Nikon SMZ1500 stereomicroscope with a digital sight DS-L1 (Nikon Instruments, Melville, NY, USA).
Sub-cellular localization of GFP/RFP fusion proteins
For transient expression, leaves from 2.5-week-old Arabidopsis WT plants and various gene-silencing mutants were used for isolation and transformation of protoplasts as described (molbio.mgh.harvard.edu/sheenweb/protocols_reg.html). Fluorescence signals were examined 12–16 h after transformation. Protoplasts were subsequently stained with 4′,6- diamidino-2-phenylindole (DAPI). For analysis of transient transformation of hydroponically grown roots (Figure S4), and analysis of GFP localization in stable transgenic plants see Methods S1. For localization in chromocenters, nuclei of young leaves from 14-day-old soil-grown pro35S:GFP–STRS lines under control conditions or after 1 h treatment with 200 mm NaCl were isolated in accordance with Fass et al. (2002). GFP immunolabeling followed by FISH assay was performed as described (Avivi et al., 2004). Slides were mounted in Vectashield solution (Vector Laboratories, Inc., Burlingame, CA, USA) for fluorescence detection. For numbers of cells counted in each experiment see Figure S10.
Fluorescence recovery after photobleaching (FRAP)
FRAP assays were performed for nucleolar-localized STRSs proteins using a ×63 oil immersion objective mounted on a Zeiss LSM 510 confocal microscope (Carl Zeiss, Oberkochen, Germany). GFP images were obtained using an excitation wavelength of 488 nm while fluorescence signals were detected using a 505–550 nm band-pass filter. A background fluorescence image was generated at a maximum speed of 150 msec using 1% laser power for 2 sec before the bleach. A circular area of diameter 3 μm (7 μm2 area) was photobleached with 100% laser power for 0.1 sec and the rate of fluorescence recovery of the bleached region was immediately monitored by imaging scans using 4% of laser power at 150 msec intervals until no further recovery of the signal was observed (approximately 40 sec). The relative fluorescence intensity was normalized to the non-bleached signal after subtraction of the background signal. Each experiment was repeated three times and comprised an average of 18 cells per experiment. Statistical significance was determined by Student's t-test.
DNA methylation analysis
DNA methylation level of the AtSN1 locus was detected by chop-PCR. Genomic DNA was digested with the DNA-methylation-sensitive HaeIII restriction enzyme and used for PCR amplification of target DNA sequences. See Table S1 for primer sequences and Methods S1 for full chop-PCR protocol.
ATPase and RNA-unwinding activities
His-tagged STRS constructs were transformed into E. coli strain BL21 (DE3; Promega, Madison, WI, USA) for expression and purified on an affinity FPLC-NiNTA column (His Trap HP; GE Healthcare Life Sciences, Little Chalfont, Buckinghamshire, UK). STRS ATPase activity and RNA-unwinding activity were determined as described by Iost et al. (1999) and Garbelli et al. (2011), respectively.
We thank Inna Lermontova (Leibniz Institute of Plant Genetic and Crop Plant Research) for cbf5 seeds, Jim Carrington (Donald Danforth Plant Science Center) for dc13, rdr2, ago4 seeds, David Baulcombe (University of Cambridge) for sdc seeds, Marjori Matzke (Gregor Mendel Institute of Molecular Plant Biology) for drd1, nrpd1a and nrpd1b seeds, Ricarda Jost (University of Western Australia) for CaMV 35S:uidA seeds, Peter Shaw (John Innes Centre) for FIB1–GFP and RFP–eIF4AIII clones, and Yuval Eshed (Weizmann Institute of Science) for the pRITA vector. We also thank Noah Isakov (Ben-Gurion University) for anti-GFP antibody. We express our great appreciation to Ruth Shaked for excellent technical help, Gil Eshel and Yana Kazachkova for statistical help and Fayek Negm for help with the ATPase assay. This work was supported by the Israel Science Foundation (Grant No. 959/11 to S.B.) and by the Shirley and William Fleischer Family Foundation.
Conflict of Interest
The authors do not have any conflict of interest to declare.