The DEAD-box RNA helix SHI2 represses expression of salt-inducible genes and modulates splicing of cold-responsive genes. SHI2 participates in transcription and post-transcriptional processes in response to stress conditions.
Plants can perceive and respond to stress signals through changes in gene expression patterns. Plant stress responses can be executed at multiple regulatory nodes including transcription, post-transcription, post-translation, and epigenetic modifications (Haak et al., 2017). Many genes involved in stress response and stress adaptation are stress-inducible genes that are commonly repressed under normal conditions but transcriptionally activated by stresses. Previous studies have identified some cis-elements and trans -acting proteins that are key to the transcriptional activation of stress-inducible genes (Yoshida et al., 2014). However, the molecular mechanism by which these genes are repressed under normal conditions is not well understood.
To better understand the regulation of inducible genes under normal and stress conditions, we developed a screening system and isolated a series of shiny (shi ) mutants with increased expression of the salt-inducible luciferase (LUC) reporter gene under normal and stress conditions (Jiang et al., 2013). Characterization of a few of these mutants has shown that the SHI1–SHI4 complex represses gene transcription possibly by preventing mRNA capping and transition from transcription initiation to elongation (Jiang et al., 2013), while SHI5/HDA6 binds specifically to the promoter regions of some key defense genes, causing histone deacetylation and repression of stress-inducible gene expression (Wang et al., 2017). These studies have suggested that some of the inducible genes are likely to be repressed under non-inducible conditions by a ready-for-transcription repressor complex consisting of general transcription factors, chromatin-modifying enzymes, and protein components for co-transcriptional processes such as 5' capping, mRNA splicing, and polyadenylation.
Post-transcriptional mRNA splicing has been implicated in gene regulation in plant abiotic stress response (Mazzucotelli et al., 2008). However, the involvement of 5' capping and alternative polyadenylation (APA) in plant stress response has rarely been reported. The 5' capping is a process in which a 7-methyguanylate (m7 G) cap is co-transcriptionally added onto the 5' end of each RNA polymerase II-transcribed mRNA and serves in multiple biological functions: these include its association with the cap-binding complex (CBC) to mediate efficient pre-mRNA splicing, 3' end processing, mRNA exporting, and translation initiation (Flaherty et al., 1997; Daneholt, 2001; Pabis et al., 2013; Ramanathan et al., 2016). APA generates non-canonical mRNA isoforms and affects the fate of transcripts and the functions of the protein. APA plays a role in maintaining the proper growth and development of plants (Simpson et al., 2003; Hunt, 2014) and has been implicated in several human diseases (Shi and Manley, 2015; Chang et al., 2017).
DEAD-box RNA helicase, which derived its name from the conserved motif (Asp-Glu-Ala-Asp, or DEAD), is involved in various molecular processes such as transcription, pre-mRNA splicing, ribosome biogenesis, RNA transport, translation initiation, organelle gene expression, and RNA degradation (Rocak and Linder, 2004; Cordin et al., 2006). In yeast, three DEAD-box RNA helicase proteins SUB2, PRP28, and PRP5 are required for splicing (Staley and Guthrie, 1998). In higher eukaryotes, the RNA helicase P68 is a component of the spliceosome and essential to the constitutive and alternative splicing (Guil et al., 2003), while the RNA helicase P72 is only required for alternative splicing (Hönig et al., 2002). The Arabidopsis genome has a large family of DEAD-box proteins consisting of 58 members (Mingam et al., 2004). Several Arabidopsis DEAD-box RNA helicases have been implicated in plant abiotic and biotic stress responses by affecting specific RNA processing events (Gong et al., 2005; Kant et al., 2007; Kim et al., 2008, Guan et al., 2013). RNA helicase LOS4 regulates mRNA export and plays an essential role in cold stress response in Arabidopsis (Gong et al., 2005), while RCF1, a DEAD-box RNA helicase, is crucial for cold-responsive gene regulation and cold tolerance (Guan et al., 2013). The STRS1 and STRS2 RNA helicases suppress the expression of stress-responsive transcription, and mutations in these two genes rendered the mutant plants more tolerant to salt, heat, and osmotic stresses (Kant et al., 2007). Despite much progress, the mechanisms of DEAD-box RNA helix family proteins in regulating the plant response to various stresses remain largely unknown.
In this study, we identify and characterize the SHINY2 (SHI2) gene encoding a DEAD-box RNA helicase in Arabidopsis. The shi2 mutant is hypersensitive to ABA, low temperature, and LiCl treatment. Our results indicate that SHI2 is involved in pre-mRNA splicing, 5' capping, and 3' processing, and plays an important role in the repression of salt-inducible genes and modulation of cold-responsive gene splicing.
The Arabidopsis (Arabidopsis thaliana) accessions Col-0 and Landsberg erecta (Ler) were used in this study. The transgenic line with the expression of the firefly LUC reporter gene driven by the promoter of a salt-inducible gene SOT12 (At2G03760) in the Col-0 background (SOT12:LUC) was used as the wild type. The shi2 mutant characterized in this study contains a single copy of the SOT12:LUC transgene in the Col-0 background and is one of the shiny mutants, shiny2 , isolated as previously described (Jiang et al., 2013). The T-DNA insertion mutant line with a T-DNA insertion in the At1G20920 gene (SALK_146567) was obtained from the Arabidopsis Biological Resource Center (ABRC). The homozygous mutant of the T-DNA line was identified by PCR-based genotyping and used for genetic complementation.
For all plate-based assays/experiments, the Arabidopsis seeds were surface-sterilized, planted on the agar plates, and then incubated at 4 °C for 2 d before being transferred to a growth chamber set to normal growth conditions of 22 °C and a 16 h/8 h light to dark cycle. For the cotyledon greening assay, ~50 seeds were sown on half-strength Murashige and Skoog (1/2 MS) medium with 0.6% agar supplemented with different concentrations of ABA. A green seedling was defined as having two obviously green cotyledons. For the primary root growth assay, Arabidopsis seeds were planted on 1/2 MS medium with 1.2% agar and allowed to grow in a vertical position for ~5 d. The seedlings were then gently transferred onto plates containing the medium supplemented with different concentrations of chemicals and allowed to grow vertically for another 10 d.
The shi2 mutant was crossed with the Ler ecotype, and the resulting F1 was self-pollinated to generate F2 progeny. The 5-day-old plate-grown F2 seedlings were evaluated for their luminescence levels, and individuals with higher luminescence were selected and used for map-based cloning of the SHI2 gene. For the molecular complementation assay, the ORF of the SHI2 (At1G20920) sequence was first cloned into the pENTR1A gateway entry vector, followed by recombining the ORF sequence into the destination vector pEarleyGate100 using the Gateway LR Clonase II Enzyme Mix (Invitrogen, Carlsbad, CA, USA). The resulting construct was introduced into Agrobacterium tumefaciens GV3101 and then transferred to the shi2 mutant using the Agrobacterium -mediated floral dip transformation method (Clough and Bent, 1998). Primers used for cloning are listed in Supplementary Table S1 at JXB online. For genetic complementation, the T-DNA mutant allele was crossed with shi2, and the F1 seedlings were subjected to LUC imaging.
To determine the subcellular localization of SHI2, the SHI2 gene in the pENTR1A:SHI2 entry vector was recombined into pMDC43 to generate an SHI2–green fluorescent protein (GFP) fusion construct using the Gateway LR Clonase II Enzyme Mix (Invitrogen). The construct was transformed into Arabidopsis Col-0 wild type by the floral dip method (Clough and Bent, 1998). The GFP signals in the roots of transgenic plants were examined by using the Olympus IX81 inverted laser scanning confocal microscope system at an excitation of 488 nm and emission of 525 nm.
For promoter–β-glucuronidase (GUS) analysis, a 3527 bp promoter fragment of SHI2 upstream of the translation initiation codon ATG was cloned into pCAMBIA1381Z. The construct was then introduced into Arabidopsis Col-0 by using the floral dip method, and homozygous transgenic T2 plants were identified, stained with X-Gluc staining buffer (10 mM Tris pH 7.0, 10 mM EDTA, 0.1% Triton X-100. and 2 mM 5-bromo-4-chloro-3-indolyl-β-d-glucuronic acid) for 12–24 h at 37 °C, and examined for the expression of the SHI2 gene in different tissue types of Arabidopsis plants.
Seven-day-old seedlings of the wild type and shi2 mutant grown in 1/2 MS agar (0.6%) medium under normal conditions (control) were used for cold, ABA, and NaCl treatments. Control and treated seedlings were collected for RNA extraction using the Plant RNA Purification Reagent (Invitrogen). Reverse transcription (RT) for cDNA synthesis was performed by using the TransScript One-Step gDNA Removal and cDNA Synthesis SuperMix kit (TransGen Biotech, Beijing, China). Resulting cDNAs were used as templates for quantitative PCR (qPCR) with Sybr Green qPCR SuperMix (TransGen Biotech, Beijing, China). The qPCR was performed with a CFX96 real-time PCR detection system (Bio-Rad, Hercules, CA, USA). ACT7 was used as an internal control. All the reactions were performed in triplicate. The primers used are listed in Supplementary Table S2.
For the intron retention assay, 7-day-old seedlings of the wild type and shi2 mutant were transferred to 4 °C for cold treatment of 0, 3, 6, and 12 h. The whole seedlings were then harvested for RNA extraction. To determine the intron retention in RNA samples, primers (Supplementary Table S2) were designed to amplify the partial sequence of the first intron and the adjacent exon of the target genes by using RT–qPCR as described above. The transcript levels were further normalized against the 0 h control of wild-type samples.
Different portions of the SHI2 cDNA sequence were amplified by PCR. The sequences were cloned into the pENTR1A vector and then recombined into the destination vector pGBKT7. The constructs were then transformed into yeast strain Y190. Yeast cells grown on a synthetic medium without tryptophan were subjected to β-galactosidase assay for the self-activation test.
The reporter–effector system used in this study was described previously (Ulmasov et al., 1995, 1997; Tiwari et al., 2001, 2003; Chinnusamy et al., 2003). The coding sequence for the SHI2c protein from amino acids 875 to 1167 was cloned into the reporter vector. Then both the reporter vector and effector vector (35S-GAL4 BD) were co-transformed into Arabidopsis protoplasts following the method described by Yoo et al. (2007). ARF5M (Auxin Response Factor 5 Middle region) (Tiwari et al., 2003) was used as a positive control for transcriptional activation analysis.
5' RACE was used to analyze the capping ratio of the LUC mRNAs according to the method established by Jiang et al. (2013). Briefly, 2 μg of total RNA was reverse transcribed into cDNA. The cDNA was added with a poly(A) tail at its 5' end using terminal transferase (New England Biolabs) which served as a template for a series of nested PCRs using the adaptor primer and gene-specific primers. The PCR products were then cloned into the pGEM-T Easy vector (Promega, Madison, WI, USA), and 20–40 positive clones were isolated and sequenced. The additional G at the very end of the 5' terminus of each cDNA was counted as a capped mRNA.
The Arabidopsis eIF4E–glutathione S-transferase (GST) recombinant protein was used to separate capped and uncapped mRNA from total RNA. Recombinant protein was expressed in Escherichia coli and purified. Then 20 µg of total RNA was incubated with the purified eIF4E–GST protein, and capped and uncapped mRNAs were separated as described by Jiang et al. (2013). Both capped and uncapped mRNAs were analyzed by using RT–qPCR. The primers used in this study are listed in Supplementary Table S3.
The polyadenylation sites of the mRNAs were determined by using 3' RACE as described by Jiang et al. (2013). The cDNA was reverse transcribed from total RNA using avian myeloblastosis virus (AMV) reverse transcriptase (Promega), and oligo(dT)17 primer. The cDNA was then utilized as the template for nested PCR by using the adaptor primer and gene-specific primer. PCR products were cloned into pGEM-T Easy vector. At least 20 independent clones were sequenced, and the polyadenylation sites were scored according to the sites linking with the poly(A). The primers used are listed in Supplementary Table S3.
RNA sequencing (RNA-seq) and RT–qPCR were conducted according to Shen et al. (2016). Briefly, 10-day-old wild-type and shi2 seedlings grown under normal conditions were randomly assigned into three groups: (i) control; (ii) cold treated at 4 °C for 12 h; and (iii) salt stressed in 100 mM NaCl for 6 h. Samples were harvested at the end of each treatment and total RNA was isolated using the RiboPure™ RNA Purification Kit (Invitrogen) according to the manufacturer’s protocol. Libraries were constructed, quantified, and quality checked with an Agilent 2100 Bioanaylzer, and then sequenced using an Illumina HiSeq 2500 System platform. Clean sequencing reads were mapped to the Arabidopsis genome (TAIR10 release) using the TopHat program (version 2.0.8) (Trapnell et al., 2009). Differentially expressed genes (DEGs) were identified by edgeR (version 3.4.2) (Robinson et al., 2010) with cut-off as fold change >2 and false discovery rate (FDR) <0.05.
The differential exon expression was analyzed using DEXseq (Anders et al., 2012). This analysis calculated the number of reads for each exon in each sample. Based on these data, the expression level of a certain exon (X) from a certain gene (Y) was calculated by comparing the number of exon X containing Y transcripts with the total number of all Y transcripts. The exons differentially expressed under different conditions, and the genes containing the differentially expressed exon were then identified by comparing the exon expression level data of different samples.
For validation of the RNA-seq results by RT–qPCR, total RNA was extracted from different samples using the Plant RNA Purification Reagent (Invitrogen). A 2 µg aliquot of total RNA was reverse transcribed with random primers using AMV reverse transcriptase (Promega). The resulting cDNAs were used as templates for real-time PCR with LightCycler® 480 master mixes (Roche, Penzberg, Germany). Actin2 was used as an internal control. The primers used are listed in Supplementary Table S4. Three biological replicates were used, and the data are shown as the average ±SD of three biological replicates.
The shi2 mutant displayed no morphological or developmental defects under normal conditions but showed enhanced expression of the LUC reporter gene under both normal and stress conditions (Fig. 1). Figure 1A shows the LUC imaging of the wild-type and shi2 plants under ABA, NaCl, sorbitol, and cold treatments. Quantification of the LUC activities showed higher LUC expression in the shi2 mutant than in the wild type under normal and stress conditions (Fig. 1B). Northern blot analysis confirmed that the expression of the LUC gene was indeed induced to higher levels in shi2 than in the wild type after stress treatments (Fig. 1C).
Genetic analysis showed that all F1 seedlings had LUC activity similar to that of the wild type, and the F2 population showed a 3:1 segregation ratio of the wild type to the shi2 mutant (220:68; χ 2=0.30) for bioluminescence phenotype. This result indicated that shi2 is a recessive mutation at a single nuclear locus.
In addition to the bioluminescence phenotype (Fig. 1A), the shi2 mutant was more sensitive to ABA than the wild type during seed germination (Fig. 2A) and showed significantly more reduction in cotyledon greening (~40%) than the wild type (~98%) at 0.5 µM ABA (Fig. 2B). Furthermore, the shi2 mutant displayed severe growth inhibition under low-temperature stress (Fig. 2C). In contrast to the hypersensitive response of shi2 to cold stress, no differences were observed between shi2 and wild-type root growth under different NaCl and mannitol treatment conditions (Supplementary Fig. S1).
The shi2 mutation was first mapped to an 80 kb region on chromosome 1 between the markers F2D10 and F9H16. Detailed sequencing analysis identified a single nucleotide substitution of C to T in the gene At1G20920 (Fig. 3A). To validate that At1G20920 was indeed the SHI2 gene, we selected a T-DNA line (SALK_146567) with an insertion at the 3'-untranslated region (UTR) of the At1G20920 gene and made a complementing cross between shi2 and the T-DNA line. The resulting F1 progeny showed the same enhanced LUC imaging phenotype as that of shi2 (Fig. 3A; Supplementary Fig. S2), indicating that the T-DNA line is allelic to the shi2 mutant. Furthermore, molecular complementation showed that the shi2 mutant expressing the wild-type SHI2 gene exhibited phenotypes comparable with those of the wild type, including the LUC imaging phenotype (Fig. 3B), cotyledon greening in the presence of ABA (Fig. 3C), and sensitivity to LiCl (Supplementary Fig. S3). These results indicate that the At1G20920 is the SHI2 gene, and the C to T substitution in SHI2 is the causal mutation for the shi2 mutant phenotypes.
SHI2 was annotated to be an intronless gene, encoding a DEAD-box RNA helicase (Fig. 3A). The C to T change in At1G20920 resulted in the substitution of a positively charged polar arginine by a non-charged polar cysteine at position 863 in the last conserved motif HRxGRxGR in the SHI2 protein (Fig. 3A: Supplementary Fig. S4A).
We determined the transcript levels of some well-known stress-inducible genes in the shi2 mutant and wild type (Fig. 4). The cold-induced expression levels of all three CBF genes were markedly higher in shi2 than in the wild type after a 6 h cold treatment, and CBF2 and CBF3 genes, but not the CBF1 gene, were also induced to higher levels in shi2 than in the wild type after a 3 h cold treatment. Moreover, the expression of the CBF genes under cold was maintained at high levels for longer in shi2 than in the wild type. However, expression of several other cold-inducible genes, including Cor15A, Cor47, KIN1, KIN2, and ERD10, did not differ between shi2 and the wild type. Interestingly, after a 12 h treatment, RD29A and SOT12, whose expression was not induced by cold in the wild type, were highly induced in shi2. Under salt stress, the salt-inducible expression of the SOT12 gene was clearly higher in shi2 than in the wild type, as expected, and higher induction of KIN1, KIN2, RD29A, and RD22 in shi2 was also observed after a 12 h salt treatment. The expression levels of all the genes shown in Fig. 4 were comparable between shi2 and the wild type in response to ABA treatment. Together, these results indicate that SHI2 regulates cold and salt stress-responsive genes in Arabidopsis.
The expression of the SHI2 gene in plants was analyzed through promoter–GUS analysis, and the results showed a ubiquitous expression of SHI2 in all tissues tested (Fig. 5A). SHI2–GFP fusion analysis revealed that SHI2 is localized in the nucleus (Fig. 5B). Protein sequence analysis revealed that Arabidopsis SHI2 shares 41% similarity with the yeast Saccharomyces pre-RNA processing 5 (PRP5). Both SHI2 and PRP5 are highly conserved at the RNA helicase signature motifs (Supplementary Fig. S4A). The yeast PRP5 is a well-studied gene involved in spliceosome assembly and is required for RNA splicing in yeast cells (Dalbadie-McFarland and Abelson, 1990). SHI2 may function in Arabidopsis similarly to PRP5 in yeast. Northern blot analysis showed that the cold-inducible gene COR15A had an additional band of a larger size of mRNA transcript after cold treatment (Supplementary Fig. S5A), indicating possible mis-splicing of pre-mRNA in the shi2 mutant. Subsequently, we detected intron retention in COR15A and COR6.6/KIN2 transcripts and found that the shi2 mutant accumulated substantially more intron-retaining mRNAs than the wild type under cold stress conditions (Supplementary Fig. S5B, C). Intron retention was further analyzed by RT–qPCR, and the results showed that, in addition to COR15A and KIN2, EDR10 and RD22 genes also accumulated more intron-containing mRNA in shi2 than in the wild type after cold stress treatment (Fig. 6). These results indicate that SHI2 is required for proper mRNA splicing of some of the cold-responsive genes.
The shi2 mutant also displayed hypersensitivity to LiCl compared with the wild type (Fig. 7A). LiCl inhibits RNA processing enzymes, and yeast mutants defective in RNA splicing show hypersensitivity to LiCl (Dichtl et al., 1997). Primary root growth of the shi2 mutant was inhibited more by 14 mM LiCl than the wild type (Fig. 7B). Hence, our findings here suggest that SHI2 may be required for pre-mRNA splicing of some cold-inducible genes, possibly as an essential component of a splicing complex.
To test whether SHI2 transcriptionally regulates gene expression, we fused different regions of SHI2 with a DNA-binding domain in the yeast bait construct. Interestingly, the C-terminal region of SHI2 (SHI2c) activated the reporter gene (β-Gal ) expression in yeast cells but the other three regions did not (Fig. 8A). This result suggests that SHI2 can be associated with the transcriptional machinery through its C-terminus to modulate gene transcription in yeast. To further confirm the function of SHI2 in gene transcriptional regulation in plants, we performed transcription activation assays in plant cells as described previously (Ulmasov et al., 1995, 1997; Tiwari et al., 2001, 2003; Chinnusamy et al., 2003). The protoplasts with the SHI2c fusion protein displayed significantly higher LUC activity than those with the empty vectors (Fig. 8B). As a positive control, ARF5M, which was defined as an activation domain (Tiwari et al., 2003), also showed transcriptional activation activity in Arabidopsis. These results further support that SHI2 could function as a transcriptional regulator in plant cells.
To test if SHI2 functions similarly to SHI1 and SHI4 (Jiang et al., 2013) in mRNA 5' capping and polyadenylation site selection, we quantified the capped and uncapped LUC mRNA transcripts in the wild type and shi2 mutant under both normal and stressed conditions. The results showed that among the total LUC mRNA, the wild type had ~52% capped LUC mRNA transcripts while shi2 had ~62% under normal growth conditions (Fig. 9A). Under salt stress (200 mM NaCl for 5 h), capped LUC mRNA transcripts in the wild type increased to 74% while they were significantly decreased to 46% in the shi2 mutant.
The 5' capping was further analyzed by using the Arabidopsis cap-binding protein eIF4E, as described in Jiang et al. (2013). The results showed that the shi2 mutant had a 4.28-fold increase in eIF4E-bound LUC mRNAs compared with the wild type under normal growth conditions, but was 6.7-fold lower in the shi2 mutant than in the wild type under salt stress (Fig. 9B). Uncapped LUC transcripts were also 2.91-fold higher in the shi2 mutant than in the wild type under normal growth conditions, while they were 3.3-fold lower in the shi2 mutant than in the wild type under salt stress (Fig. 9B). These results indicate that capped LUC mRNAs were more abundant in the shi2 mutant than in the wild type under normal conditions. However, the capped LUC mRNAs were significantly decreased in the shi2 mutant compared with the wild type under salt stress. Interestingly, the capping pattern of the native SOT12 revealed no differences between the wild type and shi2 mutant under both normal and salt stress conditions (Supplementary Fig. S6).
The 3' polyadenylation site selection was assessed using 3' RACE. Four termination sites in the LUC transgene were found, and the first three sites were located upstream while the fourth site was downstream of the canonical polyadenylation signal sequence AAUAAA (Fig. 10A). Under normal growth conditions, the shi2 mutant showed preferential selection for sites 2 and 3, while the wild type displayed a relatively even distribution in all four sites, with a slight preference for sites 1 and 3. Under salt stress, both the wild type and shi2 mutant favored the selection of site 3. However, the shi2 mutant showed significantly higher selection for site 1 for polyadenylation than the wild type (Fig. 10A). The termination patterns in the native SOT12 gene were more complex. Seven termination sites were found, with three sites upstream and four sites downstream of the putative polyadenylation signal AAUAAA. Termination patterns were relatively evenly distributed in the shi2 mutant and wild type under both normal and stress conditions, with slight preference for sites 5 and 6, except that the shi2 mutant had a notable decrease at site 4 after salt stress treatment (Fig. 10B).
We performed RNA-seq analysis to assess the effects of the shi2 mutation on genome-wide gene expression. The RNA-seq data were validated by RT–qPCR analysis of 16 genes. Based on the calculation of 64 data sets, the RNA-seq and RT–qPCR are significantly correlated, with r=0.8618 and P <0.001 (Fig. 11A), indicating that the RNA-seq data are reliable.
Analysis of DEGs between control and treated conditions revealed that there were more DEGs in the shi2 mutant than in the wild type after salt treatment and fewer DEGs in the shi2 mutant than in the wild type after cold treatment (Fig. 11B). These findings suggest that the shi2 mutation resulted in a stronger response to salt stress but a weaker response to cold stress. Analysis of DEGs between the shi2 mutant and the wild type showed that the shi2 mutation affected gene expression under normal growth conditions, and cold stress resulted in significantly more DEGs between shi2 and the wild type than salt stress (Fig. 11C; Supplementary Dataset S1).
Detailed analysis of the RNA-seq data, as shown in Fig. 11D and E, revealed that among the 2552 DEGs with higher expression levels in the shi2 mutant after cold treatment, 1621 genes (63.5%) were due to a weaker response to cold stress in the shi2 mutant than in the wild type. Similarly, among the 1803 DEGs which showed lower expression levels in the shi2 mutant after cold treatment, 1051 genes (58.3%) displayed weaker response to cold stress in the shi2 mutant than in the wild type. Overall, among the 4355 DEGs between the cold-stressed wild type and the shi2 mutant, only a small portion (641 genes, 14.7%) of them were caused by their stronger response to cold stress in the shi2 mutant whereas 61.4% of them (2672 genes) were a result of a weak response to cold stress. In contrast, 41.8% of the DEGs identified between salt-stressed shi2 and the wild type were genes that responded strongly (346 genes out of 827), and 36.3% of the DEGs (300 genes out of 827) were a result of their weaker response. These results suggest that the shi2 mutation reduced the magnitude of the response of the cold-responsive genes. For the salt stress response, however, there is not such a clear pattern in the shi2 mutant.
To further assess the attenuation of the cold response in the shi2 mutant, 112 mitochondria-encoded genes were analyzed. In the wild type, most of the mitochondria-encoded genes were induced by cold stress (Fig. 11F). However, the number of cold-induced mitochondria-encoded genes was clearly reduced in the shi2 mutant (Fig. 11G). As a result, the expression levels of most mitochondria-encoded genes in the shi2 mutant were lower than those in the wild type after cold treatment (Fig. 11H). Together, the RNA-seq data revealed that the shi2 mutant has an overall attenuated cold stress response with fewer cold-responsive genes and a reduced magnitude of the response of these cold-responsive genes. This attenuated response to cold may account for the hypersensitivity of the shi2 mutant to low temperature conditions. Although salt stress induced more genes in the wild type and the shi2 mutant than cold stress (Fig. 11B), the number of DEGs between the wild type and the shi2 mutant under salt stress was much less than those under cold stress (Fig. 11C), indicating that the impact of the shi2 mutation is greater on the cold response than on the salt response.
The effect of the shi2 mutation on mRNA splicing was also examined by using the RNA-seq data based on the expression level of each exon. In the wild type, 376 exons (90.4%) showed a lower expression level after salt stress (fold change >2, P <0.05) and 160 exons (82.9%) after cold stress (Fig. 12A). These results indicated that salt and cold stress resulted in the removal of exons from certain mRNAs, which might be a post-transcriptional regulation to generate protein variants with distinct functions in response to these stress conditions. In the shi2 mutant, the ratio of exons expressed at a lower level (222 exons, 92.9%) and more highly expressed exons (17 exons, 7.1%) after salt stress was very similar to that in the wild type (Fig. 12A). Among them, only two were differentially expressed between shi2 and the wild type (Fig. 12B). These results indicate that the shi2 mutation has a minor effect on the mRNA splicing in response to salt stress. On the other hand, cold stress dramatically changed the ratio of exons expressed at a lower level (86 exons, 58.9%) and more highly expressed exons (60 exons, 41.1%) in the shi2 mutant, compared with 17% and 83% in the wild type, respectively (Fig. 12A). As a result, in the 239 exons that are differentially expressed between the wild type and the shi2 mutant after cold treatment, 96.7% of them (231 exons) were more highly expressed in the shi2 mutant, whereas only 3.3% of them (8 exons) were underexpressed in the shi2 mutant (Fig. 12B). These results suggest that the shi2 mutation could affect the cold stress response by retaining those exons that should be removed from the mature mRNA in response to cold stress.
In this study, we identified the DEAD-box RNA helicase SHI2 as a repressor for salt-inducible genes. SHI2 also modulates cold stress response by maintaining proper splicing of cold-responsive genes. Our results suggest that SHI2 is involved in transcription and post-transcriptional processes such as 5' capping and 3' poly(A) site selection.
A shi2 allele designated as rcf1 was previously reported to affect the expression of the LUC reporter gene driven by the cold-inducible CBF2 promoter (Guan et al., 2013). RCF1 functions in maintaining proper pre-mRNA splicing and plays a crucial role in regulating cold-responsive genes (Guan et al., 2013). Consistent with the findings of Guan et al. (2013), SHI2 was also found to be involved in pre-mRNA splicing of the cold-responsive genes COR15A, EDR10, KIN2, and RD22 (Fig. 6). Transcriptomic mRNA splicing analysis further revealed that the shi2 mutation preferentially affected mRNA splicing under cold stress conditions (Fig. 12), which supports the important role of SHI2 in cold stress response. Meanwhile, the shi2 mutant displayed a hypersensitive phenotype to LiCl (Fig. 7), which is a well-known inhibitor of RNA processing enzymes (Dichtl et al., 1997), further supporting the role of SHI2 in RNA splicing. Taken together, we propose that the RNA helicase SHI2 is involved in pre-mRNA splicing and affects cold tolerance through maintaining proper mRNA splicing of the genes important for cold response. In addition, an overall attenuated cold stress response in shi2 (Fig. 11) may contribute to its hypersensitivity to cold.
In contrast to attenuated cold response of the shi2 mutant, the shi2 mutation displayed a minor effect on mRNA splicing in response to salt stress but increased the number of salt-responsive genes (Figs 4, 11). This suggests that the regulation of salt-responsive genes by SHI2 is unlikely to be due to mRNA splicing, but is more likely to be executed at the transcription level. The shi2 mutant was isolated based on increased expression of the LUC reporter gene driven by the salt-inducible AtSOT12 promoter. Both LUC and AtSOT12 are intronless genes. Therefore, the repression of these genes by SHI2 is clearly independent of its function in mRNA splicing. The C-terminus of SHI2, which does not include the RNA helicase domain, could function as an activator in both yeast and plant cells (Fig. 8), suggesting that SHI2 may interact with and enhance the binding of the general transcription machinery at the promoter region. Both 5' RACE and RT–qPCR experiments revealed that capped LUC mRNAs were more abundant in the shi2 mutant than in the wild type under normal growth conditions (Fig. 9), which resembles our previously reported shi1 and shi4 mutants (Jiang et al., 2013). The SHI1–SHI4 complex represses gene transcription by preventing mRNA capping and transition from transcription initiation to elongation (Jiang et al., 2013). Thus, SHI2 may also function as a gene repressor by affecting mRNA capping and transcription initiation.
Inducible gene expression is generally believed to be activated by transcription factors binding to the promoter cis -elements, which recruits the general transcription machinery to promote transcription. However, how inducible genes are repressed under non-inducible conditions remains elusive. Our findings suggest that a repressor complex may include the general transcription machinery and other factors such as SHI2 at the promoter to repress the transcription of the inducible genes. Several genome-wide analyses of RNA polymerase occupancy in yeast (Radonjic et al., 2005), human (Guenther et al., 2007), and Drosophila (Muse et al., 2007; Zeitlinger et al., 2007; Hendrix et al., 2008) indicate that most protein-coding genes, including those that are inactive but inducible by environmental stimuli or developmental signals, are occupied by a transcription initiation complex. These studies suggest that the inducible genes are repressed but poised for transcription under non-inducible conditions. Our previous studies on SHI1–SHI4 (Jiang et al., 2013) and SHI5/HDA6 (Wang et al., 2017), together with the results of this study, suggest that, under normal growth conditions, the inducible promoters such as the AtSOT12 promoter may be occupied by a repression complex. This complex may comprise not only the general transcription machinery, but also proteins or enzymes modulating RNA polymerase II (e.g. SHI1 and SHI4) and proteins involved in co-transcriptional processes, such as the splicing factor SHI2. Therefore, we propose that, under normal growth conditions, such a repression complex is berthed at the promoters of stress-inducible genes but poised for transcription. Upon stress, changes to the components within the repressor complex could be modulated by either transcriptional regulation or post-translational modifications of the components, leading to the release of the repressor proteins and the recruitment of the activator proteins into the complex, hence changing the repressor complex into an activation complex for stress-induced gene expression. We postulate that the repressor complex and activation complex share components for general transcription and co-transcriptional processes, thereby ensuring the stress-inducible genes are repressed under normal growth conditions but poised for transcription and can be rapidly induced in response to stresses.
The ability of SHI2 in activating gene expression and the effects of shi2 mutation on 5' capping, mRNA splicing, and polyadenylation site selection provide compelling evidence that SHI2 is probably associated with transcription machinery that regulates transcription of stress-inducible genes under normal and stressed conditions. In our future work, identification of the SHI2-associated proteins will help our understanding of the SHI2-associated repressor complex for stress-inducible gene regulation.
This work was partially supported by the US Department of Agriculture National Research Initiative project 2007-35100-18378 to HS and the National Natural Science Foundation of China grants 31270316 and 31328004 to WY. We thank Dr Jiafu Jiang and Dr Shiming Han for technical assistance.