Circular RNA Vv‐circCOR27 modulates thermotolerance through attenuating VvHSP90.2b‐VvHsfA7a interaction in grapevine
Yi Ren, Yuanyuan Xu, Moyang Liu, Lipeng Zhang, Yue Song, Junpeng Li, Jingjing Liu, Dongying Fan, Zhen Zhang, Juan He, Jiuyun Wu, Qian Zha, Zhen Gao, Zheng'an Yang, Chao Ma
TL;DR
A circular RNA called Vv-circCOR27 in grapevines reduces heat tolerance by blocking a key protein interaction.
Contribution
This study reveals a novel regulatory mechanism of thermotolerance in grapevines via a circRNA that inhibits a protein interaction.
Findings
Vv-circCOR27 expression is suppressed under heat stress and is lower in thermotolerant grapevine cultivars.
Overexpression of Vv-circCOR27 worsens thermotolerance by inhibiting the interaction between VvHsfA7a and VvHSP90.2b.
Vv-circCOR27 binds directly to VvHSP90.2b, leading to downregulation of small heat shock protein genes under heat stress.
Abstract
Circular RNAs (circRNAs) are single‐stranded, covalently closed RNA molecules that arise from exon back‐splicing. The identification and function investigation of circRNAs have been comprehensively explored in plants, however, the regulatory mechanisms are largely unknown.This study employed genetic transformation, circRNA pull‐down assay, ribonucleoprotein immunoprecipitation (RIP), and the circRNA trimolecular fluorescence complementation (cTriFC) system to investigate the function and regulatory mechanisms of a circRNA Vv‐circCOR27 in grapevine. Vv‐circCOR27 expression was suppressed under heat stress and was markedly lower in thermotolerant cultivars. In addition, overexpression of Vv‐circCOR27, using an expression cassette with endogenous introns inserted into circRNA‐producing exons to minimize the background level of cognate linear RNA, was shown to exacerbate thermotolerance in…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8
Fig. 9- —Earmarked Fund for China Agriculture Research System10.13039/501100010038
- —National Natural Science Foundation of China10.13039/501100001809
- —AI+ project of Shanghai Municipal Education Commission
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsCircular RNAs in diseases · Ion Channels and Receptors · Heat shock proteins research
Introduction
Back‐splicing is a specific eukaryotic splicing process whereby a downstream 5′ splice site is linked to an upstream 3′ splice site via 3′, 5′ phosphodiester bonds, determining the biogenesis of circRNAs (Chen, 2020). Owing to the lack of polyadenylated tails and inefficient back‐splicing, most circRNAs are rarely detected by classical mRNA sequencing methods (Liu & Chen, 2022; Liu et al., 2023). With the advent of the enrichment of non‐polyadenylated RNA transcripts and analytical pipelines for circRNA identification, thousands of circRNAs have been found to be widely expressed (Gao et al., 2019; Fan et al., 2020; Wu et al., 2023). In grapevine, over 8000 circRNAs are assembled, as determined by rRNA‐depleted sequencing of whole tissues (Gao et al., 2019). Although their abundance is relatively low in tissues, most circRNAs have long half‐lives because of their resistance to degradation mechanisms, implying that they have the potential to play vital roles (Enuka et al., 2016).
In plants, numerous differentially expressed circRNAs are found in a variety of tissues, at specific developmental stages, and after abiotic and biotic stress. It has been proven that circRNAs regulate both development and abiotic stress responses in plants. Examples include floral formation, fertility, and fruit color formation, as well as modulating cold, drought, and salt tolerance across various species (Conn et al., 2017; Tan et al., 2017; Cheng et al., 2018; Gao et al., 2019; Song et al., 2021; Zhou et al., 2021). In addition, recent findings imply that the biological functions of circRNAs are largely species‐dependent, potentially because the sequence and biosynthesis of circRNAs are not conserved among species (Ren et al., 2023). Nevertheless, functional studies of plant circRNAs remain very limited, given their vast number and complexity. Research has been conducted on the regulatory mechanisms of circRNAs in plants. CircRNAs can modulate alternative splicing (Conn et al., 2017) and maintenance of chromatin stability (Liu et al., 2020) by forming R‐loops, DNA : RNA hybrid structures at cognate DNA loci. Clues for the coding ability of circRNAs in plants were given by analyzing ribosome footprint profiling and mass spectrometry data in Arabidopsis and maize, which indicate that few mitochondrion‐derived circRNAs (mcircRNAs) are translated into polypeptides (Liao et al., 2022). Following the report of ciRS‐7 that acts as a miR‐7 sponge in mammals (Hansen et al., 2013), similar competing circRNA‐miRNA networks have been predicted in silico and indirectly verified by genetic and biochemical assays in plants (Liu et al., 2023). For example, co‐infiltration of Vv‐circSIZ1, which harbors a single Vv‐miR3631 site, significantly inhibited Vv‐miR3631 splicing in a luciferase reporter assay, confirming direct binding (Gao et al., 2023). CircRNAs not only require protein assistance for their biogenesis, nuclear export, and turnover (Chen, 2020), but can also bind proteins to act as scaffolds, sponges, or suppressors in mammals (Liu & Chen, 2022). However, only a few studies have focused on the ‘circRNA‐protein’ interactions in plants. A notable exception is the lariat RNA in Arabidopsis, which interacts with the DCL1/HYL1 complex to competitively inhibit HYL1 binding to pre‐miRNA (X. Li et al., 2016; Z. Li et al., 2016). Interestingly, extracellular RNAs, including circRNAs, are associated with proteins such as GLYCINE‐RICH RNA‐BINDING PROTEIN 7 (GRP7) and the small RNA‐binding protein ARGONAUTE2 (AGO2), but their roles remain unclear (Karimi et al., 2022). That result is supported by evidence that circRNAs are identified from AGO‐immunoprecipitation libraries (Capelari et al., 2019).
Heat stress universally impairs crop photosynthesis. In grapevine, it damages photosystem II, Rubisco activity, and membrane stability (Venios et al., 2020; Zha et al., 2021). The heat stress response in Arabidopsis is coordinated by heat shock factors (Hsfs), with HsfA1s as master regulators that control key modulators like HsfA2 and HsfA7 (Charng et al., 2007; Liu et al., 2011). The widely accepted model is that the activity of HsfA1s is suppressed by heat shock protein 70 (HSP70) and HSP90 binding under normal conditions, whereas heat stress triggers the competitive sequestration of these chaperones by denatured proteins, leading to HsfA1 release and activation (Hahn et al., 2011). The repressive activity of HSP90 to HsfA1s is determined by the interaction between HSP90 and the HsfA1‐specific temperature‐dependent repression (TDR) domain (Ohama et al., 2016). HSP90, a molecular chaperone essential for client protein folding, has four cytoplasmic isoforms (VvHSP90.1a, VvHSP90.1b, VvHSP90.2a, and VvHSP90.2b) in grapevine (Banilas et al., 2012; Xu et al., 2023). However, the suppression and chaperone activity of HSP90 in grapevine in response to heat stress are largely unknown. Intriguingly, in mammals, noncoding RNA such as long noncoding RNA (lncRNA) and circRNAs can directly bind to HSP90 to modulate its chaperone function or interfere with client interactions (Tang et al., 2019). A notable example is that circ‐STK40 acts as a barrier to block the interaction between HSP90 and its client in mammals (Ni et al., 2021). However, it is not clear whether circRNAs similarly regulate HSP90 in plants.
COLD‐REGULATED GENE 27 (COR27) and its functional homologue COR28 regulate circadian rhythms, freezing tolerance, and flowering time in Arabidopsis (X. Li et al., 2016; Z. Li et al., 2016). AtCOR27 also acts as a negative regulator of photomorphogenesis via the COP1‐HY5 module (Li et al., 2020; Zhu et al., 2020). Herein, we characterized a circRNA (Vv‐circCOR27) that is back‐spliced from the exons of VvCOR27 in grapevine and found it fine‐tunes thermotolerance in grapevine as a negative regulator. Mechanistically, Vv‐circCOR27 physically binds with VvHSP90.2b, thereby inhibiting its interaction with VvHsfA7a to reduce the expression level of heat‐induced sHSP genes.
Materials and Methods
Plant materials and treatments
For heat stress treatment of grapevine, tissue‐cultured seedlings of ‘Thompson Seedless’ (Vitis vinifera L.) were planted under 16 h : 8 h, light : dark at 26°C. Seedlings were used for heat stress at 45°C. The maximum quantum efficiency of PS II (F v/F m) was collected, and imaging photos were taken by Imaging‐PAM software. Meanwhile, the mature leaves corresponding to the treatment were sampled and quickly frozen in liquid nitrogen and stored at −80°C.
GeneBridge analysis
The probabilistic estimation of expression residuals tool was used to filter the data (Stegle et al., 2012). And then, the data were transferred to the GeneBridge toolkit for further analysis, focusing specifically on ontology modules. Gene‐Module associations were inferred using the G‐MAD (Gene‐Module Association Determination) tool, which employs the competitive gene set test CAMERA to calculate enrichment p‐values while accounting for inter‐gene correlations (McWhite et al., 2020). For Module‐Module associations, cross‐species transcriptome data were analyzed via M‐MAD. Both Gene‐Module and Module‐Module association scores were derived through a meta‐analysis of all datasets, weighted by intra‐module gene correlation and sample size (Liu et al., 2025).
Cytoplasmic and nuclear location of Vv‐circCOR27
Nuclear‐cytoplasmic fractionation was performed as described with minor modification (Wang et al., 2011). Briefly, grapevine leaves were ground in liquid nitrogen and homogenized in cold Lysis Buffer (20 mM Tris–HCl, pH 7.5, 20 mM KCl, 2 mM EDTA, 2.5 mM MgCl_2_, 25% glycerol, 250 mM sucrose, 5 mM DTT, 1× plant protease inhibitors, 40 U·ml^−1^ RNase inhibitor). The homogenate was filtered through two layers of Miracloth (Cat. 475855; Merck, Burlington, MA, USA) and centrifuged at 1500 ** g ** for 10 min at 4°C. Centrifuged at 10 000 ** g ** for 10 min at 4°C for cytoplasmic RNA extraction with TRIzol. The nuclear pellet was washed 3–4 times with NRB1 Buffer (20 mM Tris–HCl, pH 7.5, 25% glycerol, 2.5 mM MgCl_2_, 0.2% Triton X‐100, 1× plant protease inhibitors, 40 U·ml^−1^ RNase inhibitor). Discarded the supernatant and added 500 μl NRB2 Buffer (20 mM Tris–HCl, pH 7.5, 0.25 M sucrose, 10 mM MgCl_2_, 1% Triton X‐100, 5 mM 2‐ME, 1× plant protease inhibitors, 40 U·ml^−1^ RNase inhibitor) to the pellet. Resuspended and overlayed on top of 500 μl NRB3 Buffer (20 mM Tris–HCl, pH 7.5, 1.7 M sucrose, 10 mM MgCl_2_, 0.15% Triton X‐100, 5 mM 2‐ME, 1× plant protease inhibitors, 40 U·ml^−1^ RNase inhibitor) and centrifuged at 16 000 ** g ** for 45 min at 4°C. The nuclear pellet was used to nuclear RNA extraction using TRIzol. Vv‐circCOR27 accumulation was normalized according to a previous description (Wang et al., 2006). All primer pairs used in this work were deposited in Supporting Information Table S1.
Vector construction, transient expression, GUS staining, and callus transformation
All circRNA expression vectors were constructed as previously described (Gao et al., 2019; Ren et al., 2023). Transient expression in Nicotiana benthamiana (Nb) plants and GUS staining were performed as described (Chen et al., 2020). For grapevine seedlings, an agrobacteria suspension was prepared until the OD_600_ was up to 0.8–1.0. Centrifuged, and the pellet was resuspended in MAA buffer to OD_600_ = 0.4. Uniform tissue‐cultured seedlings were vacuum‐infiltrated at −0.8 MPa for 5 min, blotted dry, and planted under 16 h : 8 h, light : dark at 26°C. Samples were collected for RT‐qPCR analysis and treated after cultivation for 5 d. The protocol of embryogenic callus induction, subculturing, and transformation complied with our previous description (Ren et al., 2023). For heat stress treatment, calli were sub‐cultured into medium for 1 wk at 26°C and then further cultured at 40°C.
In vivo crosslinking circRNA pull‐down assay
The Vv‐circCOR27 binding protein identification was performed based on the ChIRP protocol (Chu et al., 2011; Kim et al., 2017). Transgenic calli were crosslinked in 1% formaldehyde and terminated by 125 mM glycine. Five hundred micrograms of powder was lysed in Lysis Buffer (50 mM Tris–HCl, pH 8.0, 10 mM EDTA, 1% SDS, 5 mM DTT, 80 U·ml^−1^ RNase inhibitor, 1× Cocktail). After incubation of the mixture for 10 min at ice, centrifuge at 10 000 ** g ** for 10 min at 4°C. The supernatant was diluted 1 : 2 with hybridization buffer (500 mM NaCl, 100 mM Tris–HCl, pH 7.5, 10 mM EDTA, 1% SDS, 15% formamide, 5 mM DTT, 80 U·ml^−1^ RNase inhibitor, 1× Cocktail) and incubated with 100 pmol of 3′‐biotin‐TEG antisense probes (37°C, 4 h). Streptavidin‐magnetic beads (Cat. S1420S; NEB, Ipswich, MA, USA) were blocked with yeast total RNA and bovine serum albumin. Each sample was added 50 μl of washed beads, mixed at 37°C for another 2 h. Captured beads were washed five times with washing buffer (2× SSC, 0.5% SDS, 5 mM DTT, 80 U·ml^−1^ RNase inhibitor, 1× Cocktail). The sample was reverse crosslinked at 65°C and was digested by RNase A and DNase I. Eluted proteins were analyzed by Mass Spectrometry (OEbiotech, Shanghai, China).
Ribonucleoprotein immunoprecipitation (RIP) in Nb plants
The RIP assay was applied to verify circRNA‐protein interaction in Nb plants as described with modifications (Marmisolle et al., 2018). The protein candidates fused with GFP and GFP were individually co‐infiltrated with Vv‐circCOR27 in Nb plants. Leaves were crosslinked with 1% formaldehyde and quenched with 125 mM glycine. Tissue powder (0.5 g) was lysed in 500 μl RIP‐Lysis Buffer (20 mM Tris–HCl, pH 7.5, 150 mM NaCl, 1 mM MgCl_2_, 1 mM CaCl_2_, 0.1% SDS, 1% sodium deoxycholate, 1% Triton X‐100, 1× cocktail, 5 mM DTT, 80 U·ml^−1^ RNase inhibitor). After vortexing and incubation for 10 min on ice, it was centrifuged at 10 000 ** g ** for 15 min at 4°C and 10 μl of supernatant was saved as input. Thirty microliters (each reaction) of Anti‐GFP mAb‐Magnetic Beads (Cat. D153‐11; MBL, Nagoya, Aichi, Japan) were washed and blocked by 5% BSA. Then the supernatant was discarded and 550 μl RIP‐DB buffer (10 mM Tris–HCl, pH 7.5, 150 mM NaCl, 0.5 mM EDTA, 80 U·ml^−1^ RNase inhibitor) and 50 μl of the extract supernatant were added. The tube was incubated overnight at 4°C. The beads were washed five times with RIP‐WB Buffer (50 mM Tris–HCl, pH 7.5, 500 mM NaCl, 4 mM MgCl_2_, 0.5% Sodium Deoxycholate, 0.1% SDS, 2 M Urea, 2 mM DTT). The RNA was extracted by TRIzol reagent. Vv‐circCOR27 enrichment was quantified by RT‐qPCR, and protein recovery was confirmed by anti‐GFP western blot (Marmisolle et al., 2018).
CircRNA trimolecular fluorescence complementation (cTriFC) assay
The cTriFC assay was performed as Li's description (X. Li et al., 2016; Z. Li et al., 2016). The MS2 sequence was inserted into the Vv‐circCOR27. The MCP protein (Notes S1) was fused with the C‐terminal region of YFP and named pXY‐MCP‐cYFP. The VvHSP90.2b and corresponding truncations were fused with the N‐terminal region of YFP and named pXY‐nYFP‐HSP90.2b (‐F1, ‐F2, ‐F3, and ‐F4). The fluorescence was observed and photographed using a microscope (DFC450C, Leica).
DNA affinity purification sequencing (DAP‐seq) assay
DNA affinity purification sequencing assay was conducted according to a previous description with minor modifications (Bartlett et al., 2017). Genomic DNA was extracted from leaves, and a sequencing library was constructed. VvHsfA7a fused to a HaloTag was expressed using the TNT® Quick Coupled Transcription/Translation System (Cat. L1170; Promega, Beijing, China) and captured onto Magne HaloTag Magnetic Beads (Cat. G7281; Promega). The beads were resuspended and incubated with the DNA library. The binding DNA segments were purified and sequenced on an Illumina platform. The clean reads were mapped to the Pinot Noir genome (PN40024) (https://phytozome‐next.jgi.doe.gov/) by BWA‐MEM software. The motif capture was conducted by MEME‐suit (https://meme‐suite.org/).
Results
Identification of Vv‐circCOR27
in grapevine
In previous work, our group has built a circRNA dataset in grapevine (Gao et al., 2019). In this study, the biological functions and mechanistic underpinnings of circular RNAs were further investigated. Subsequently, a 502‐nt circRNA, termed Vv‐circCOR27, was identified. It originates from the second and third exons of VvCOR27, an ortholog of AtCOR27 in Arabidopsis (Fig. 1a,b). Its back‐splice junction (AG/GA) was confirmed by full‐length cloning and Sanger sequencing, revealing an exon‐derived circRNA with no alternative isoforms (Fig. 1c,d). Divergent primer PCR verified its circular nature (amplified from cDNA but not gDNA) (Fig. 1e), and a nucleocytoplasmic separation assay showed its localization in both compartments (Fig. 1f).
Characterization of Vv‐circCOR27 and COR27‐related functions. (a) A diagram showing the back‐splicing of Vv‐circCOR27. DP, divergent primer; CP, convergent primer; GSP, host gene‐specific primer; FLP, primers for the full‐length cloning of Vv‐circCOR27. (b) Phylogenetic relationship between VvCOR27/28 in grapevine and AtCOR27/28 in Arabidopsis. The number represents bootstrap value. (c, d) PCR amplification of Vv‐circCOR27 (c) and Sanger sequencing (the short vertical line indicates the back‐splicing site) (d). (e) PCR amplification of Vv‐circCOR27 within cDNA and genomic DNA using CP (►◄) and DP (◄►), with a red asterisk showing the putative production of Vv‐circCOR27. (f) Detection of Vv‐circCOR27 in the cytoplasmic (Cyt) and nuclear (Nucl) fractions. (g) Functions of the COR27 gene according to G‐MAD. Values are gene‐module association scores (GMAS).
To better understand the biological function of VvCOR27, c. 30 000 datasets from more than 400 projects deposited into the GEO, ArrayExpress, and CNGBdb repositories (Table S2) were integrated to identify COR27‐related genes at the transcription (G‐MAD), transcriptional regulation (DAP‐seq), and protein interactions (CF‐MS) levels (Liu et al., 2024) (Fig. S1a). According to the connections between COR27‐related genes and other biological functions via the M‐MAD tool, these genes are involved in 612 potential biological processes, including circadian rhythm regulation, response to light quality, absence of light, and cold (Table S3; Fig. 1g). In this work, we also found that the expression level of VvCOR27 was strongly increased by cold treatment for 4 and 8 h (Fig. S1b,c). However, the expression of Vv‐circCOR27 exhibited only a transient decrease and had already recovered (Fig. S1d). Given that the transcription of the orthologous AtCOR27 gene is slightly downregulated under warm conditions in Arabidopsis (X. Li et al., 2016; Z. Li et al., 2016) and that the biological processes are involved in the response to heat (Fig. 1g), we wondered whether and how VvCOR27 and Vv‐circCOR27 are involved in response to heat stress in grapevines.
Vv‐circCOR27
and host gene VvCOR27 are suppressed by heat stress
To investigate the expression pattern of VvCOR27 and Vv‐circCOR27 under heat stress, 5‐wk‐old seedlings were subjected to 45°C treatment for 2 h. The phenotypic change was not obviously observed (Fig. 2a). The F v/F m value significantly decreased at 0.5 h and 1 h but was restored at 2 h (Fig. 2a,b). The malondialdehyde (MDA) rapidly accumulated after heat treatment (Fig. S1e). Under heat stress, both VvCOR27 and Vv‐circCOR27 were transcriptionally repressed (Fig. 2c,d), a finding that is consistent with the observation in Arabidopsis (X. Li et al., 2016; Z. Li et al., 2016). Vv‐circCOR27 was predominantly expressed in leaves (Fig. S1f). Notably, the expression level of Vv‐circCOR27 is significantly higher in thermosensitive cultivars such as ‘SN45’ and ‘SD25’ than in thermotolerant cultivars such as ‘SJ109’ and ‘Z11’ (Fig. 2e). These data implicated a potential role of both Vv‐circCOR27 and its host gene VvCOR27 in thermotolerance regulation.
The phenotypes of grapevine and expression pattern of Vv‐circCOR27 under heat shock. (a) Representative photographs (upper) of ‘Thompson Seedless’ seedlings treated at 45°C. Chlorophyll fluorescence imaging of leaves (below) was observed. (b) F v/F m value fluctuation under 45°C treatment. The horizontal line in the boxplot represents the median, the box limits represent the upper and lower quartiles, the whiskers denote 1.5× interquartile range, and the outliers represent abnormal values. (c, d) The expression pattern of VvCOR27 (c) and Vv‐circCOR27 (d) under heat treatment. (e) The expression level of Vv‐circCOR27 among cultivars with different thermotolerances. All cultivars were treated at 45°C for 2 h and the thermotolerance of cultivars was evaluated by Chl fluorescence imaging and F v/F m values, TT, thermotolerance; TS, thermo‐sensitivity. Differences are compared using a t‐test (, P < 0.05; **, P < 0.01; **, P < 0.001; ns, no significant difference) (b–d); lowercases indicate significant differences at P < 0.05, as determined by Duncan's multiple range test (e). Closed circles represent the measured values (mean ± SD).
The relative abundance of Vv‐circCOR27
is improved by extra intron insertion
A strategy to purposefully overexpress circRNA in grapevine was developed in our group (Gao et al., 2019). However, the biological effects of cognate linear RNA along with the generation of circRNAs are unable to be ignored (Fig. 3a). To differentiate the functions of Vv‐circCOR27 from those of cognate linear RNA, we developed an ‘intron‐retention’ strategy by extra intron insertion within exons that are back‐spliced (Fig. 3b,h). Notably, the designed site (AG/GA) of intron insertion invariably obeyed the splicing rules based on the U2 spliceosome (Notes S1), which is responsible for circRNA processing (Chen, 2020).
*‘Intron‐retention’ strategy and the relative abundance of mature circRNA. (a) A schematic diagram of vector construction for circRNA expression and splicing. (b, h) A diagram of intron insertion within the exon region. (c, i) PCR verification of the expression and splicing of Vv‐circCOR27 (c) and circRNA_2407 (i) with different intron insertions in Nb plants. (d–g) RT‐qPCR analysis of the relative expression level of Vv‐circCOR27 with different intron insertions (‘cDNA’, ‘gDNA’, ‘2 introns’, and ‘3 introns’) in Nb plants. (j, k) RT‐qPCR analysis of the relative expression level of circRNA_2407 with intron‐free insertion (j) and native DNA (k). Differences are compared using a t‐test (**, P < 0.01; **, P < 0.001; ns, no significant difference, n = 3). DP, divergent primer; CP, convergent primer. Closed circles represent the measured values (mean ± SD).
Initial constructs using cDNA (without intron) or native gDNA (one intron) both produced Vv‐circCOR27 in Nb plants (Fig. 3b,c). They also yielded high levels of cognate linear RNA, as detected by convergent primers (CP), significantly exceeding circRNA (DP) levels (Fig. 3d,e). To suppress linear RNA production, another one or two introns (two introns and three introns) were artificially inserted into the back‐spliced exons. This did not affect circRNA splicing (Fig. 3b,c), but intron insertion arrests the production of cognate linear RNA when two introns are inserted (Fig. 3f), and eventually, almost all nascent transcripts are processed into mature Vv‐circCOR27 when two extra introns (3 introns) are artificially inserted (Fig. 3g).
To assess the universality of this approach, we focused on circRNA_2407 (624‐nt), which is generated from exon 8 to exon 13 of VIT_206s0009g02170. Expression vectors were constructed using either an intron‐free CDS or the native genomic DNA (ranging from exon 8 to exon 13 with 5 introns) (Fig. 3h). Both strategies consistently generated only one isoform of circRNA_2407 in Nb plants (Fig. 3i). As expected, intron deletion clearly enhanced the generation of cognate linear RNA (Fig. 3j), whereas intron‐rich native gDNA ensured the proportion of mature circRNA_2407 (Fig. 3k). Overall, we suggest that redundant intron retention likely prolongs the splicing of cognate linear RNA and increases the likelihood of circRNA processing.
Vv‐circCOR27
plays a negative role in the response to heat stress
To investigate the potential functions of Vv‐circCOR27 and VvCOR27 in response to heat stress in grapevines, we overexpressed Vv‐circCOR27 in ‘Thompson Seedless’ calli using the ‘intron retention strategy’ (three introns) to minimize linear RNA production (Fig. 3b,g). As a control, to convincingly eliminate the functional disruption of linear RNA, we separately overexpressed only the overlapping linear RNA. Consequently, Vv‐circCOR27 and linear RNA were successfully overexpressed in calli (Vv‐circCOR27‐OE and Linear_COR27‐OE) (Fig. S2a,b,f). The transcript level of the host gene VvCOR27 and the upstream neighbor gene (VvPKFP) (Fig. S2k) was slightly upregulated in the Vv‐circCOR27‐OE lines, a trend not observed in linear RNA controls (Fig. S2c,d,g,h). Conversely, the transcript level of the downstream gene VvCOQ10 was obviously decreased in the Vv‐circCOR27‐OE and Linear_COR27‐OE lines (Fig. S2e,i,k). These results indicate that Vv‐circCOR27 can influence the expression of its host gene and adjacent genes.
Given that the host gene was inhibited by heat stress (Fig. 2c) and upregulated in Vv‐circCOR27‐OE calli (Fig. S2c), we further tested whether VvCOR27 plays a role in the response to heat stress, and VvCOR27 was successfully introduced into calli (OE‐3, OE‐4, and OE‐6) (Fig. S2j). Under heat stress (45°C), VvCOR27‐OE calli exhibited a phenotype similar to WT, whereas Vv‐circCOR27‐OE calli showed severe browning (Fig. 4a–c). We further developed a transient overexpression system using tissue‐cultured seedlings of ‘Thompson Seedless’ (Fig. 4d,e). We further confirmed that heat shock injury was significantly alleviated in both VvCOR27‐OE and empty vector (EV) seedlings compared to Vv‐circCOR27‐OE seedlings (Fig. 4f–h). In Arabidopsis, the thermotolerance of AtCOR27 mutants (cor27‐1 and cor27‐2) was also tested (X. Li et al., 2016; Z. Li et al., 2016). The damage to rosette leaves of the cor27‐1 and cor27‐2 seedlings was similar to that of the Col‐0 plants under variable periods of heat stress (Fig. S3a–e). These results indicated that Vv‐circCOR27 plays a negative role and that the host gene VvCOR27 may be functionally inactive in the response to heat stress.
Functional investigation of Vv‐circCOR27 and VvCOR27. (a) The phenotype of Vv‐circCOR27‐OE and VvCOR27‐OE calli at 26°C and 45°C, respectively. (b, c) The gray value measured by ImageJ software of calli at 26°C (b) and 45°C (c). (d) The schematic diagram of transient overexpression of grapevine seedlings. (e) Representative relative overexpression level of Vv‐circCOR27, VvCOR27, and Linear_COR27 based on the transient overexpression method in grapevine seedlings. (f–h) The phenotypic differences (f), the content of MDA (g) and H2O2 (h) of leaves from grapevine seedlings before and after heat shock (45°C). Scale bars represent 1 cm. Differences are compared using a t‐test (, P < 0.05; *, P < 0.01; ns, no significant difference). Closed circles represent the measured values (mean ± SD).
The linear transcript overlapping with Vv‐circCOR27
enhanced thermotolerance in plants
We further examined the thermo‐sensitivity of Linear_COR27‐OE calli and found that Linear_COR27 significantly enhanced thermotolerance (Fig. 5a–c). To further validate this finding, Linear_COR27 was also transiently overexpressed in grapevine seedlings (Fig. 4d,e), and the result of heat treatment indicated that the Vv‐circCOR27‐OE seedlings were obviously deteriorated, but the heat shock injury of Linear_COR27‐OE seedlings was obviously alleviated compared to EV seedlings (Fig. 5d–f). Interestingly, allogenic overexpression of Linear_COR27 in Arabidopsis also enhanced its thermotolerance (Fig. 5g–i). These results suggested that although Vv‐circCOR27 negatively modulates thermotolerance in grapevines, the linear fragment overlapping with Vv‐circCOR27 oppositely enhances the thermotolerance.
Thermotolerance investigation of Linear_COR27 overlapping with Vv‐circCOR27. (a) The phenotype of Vv‐circCOR27‐OE and LinearCOR27‐OE calli at 26°C and 45°C. (b, c) The gray value of grapevine calli at 26°C (b) and 45°C (c). (d–f) The phenotypic differences (d), the content of MDA (e) and H2O2 (f) of seedling leaves before and after heat shock. Closed circles represent the measured values (mean ± SD). (g–i) The overexpression of Linear_COR27 in Arabidopsis (g), phenotype (h), and leaf damage rate (i) in response to heat shock (mean ± SD, n = 3). WT and Linear_COR27‐OE lines were grown at 22°C, exposed to 40°C for 8 h, and then returned to 22°C for a 1 wk recovery. Differences are compared using a t‐test (, P < 0.05; **, P < 0.01; **, P < 0.001; ns, no significant difference).
Here, the unusual thermotolerance of the Linear_COR27‐OE was notable, as the linear transcript overlapping a circRNA is typically considered a non‐functional control (Gao et al., 2019). We hypothesized that the Linear_COR27 potentially encodes a functional product contributing to heat stress tolerance. To test this, we predicted the ORFs hidden in Linear_COR27, and three ORFs (ORF1, ORF2, and ORF3) were identified (Fig. S4a). All ORFs (with termination codons) were fused to a GUS reporter and transiently overexpressed in Nb plants (Fig. S4a). Strong GUS activity was detected for ORF2 and ORF3 fusions (Fig. S4b,e,f). By contrast, GUS activity for ORF1 and VvCOR27 was significantly diminished, suggesting that activity for translation from ORF1 and VvCOR27 cistrons blocks GUS expression (Fig. S4c,d). Therefore, we surmised that the ORF1 hidden in Linear_COR27 possibly encodes a protein that potentially contributes to thermotolerance in both grapevine and Arabidopsis.
RNA‐binding proteins are enriched by circRNA pull‐down
In mammals, circRNA can function by binding RNA‐binding proteins or acting as scaffolds for protein interactions (Liu & Chen, 2022). To investigate how Vv‐circCOR27 negatively regulates heat stress response in grapevines, we performed an in vivo crosslinking circRNA pull‐down assay to identify its binding proteins in calli (Fig. 6a). Mass spectrometry analysis identified 753 proteins, including translation‐related factors such as 40S/60S ribosomal protein, RNA helicase, and translation initiation factor (Table S4). Gene Ontology (GO) analysis revealed enrichment in ribosome (10.3%), spliceosome (4.0%), and RNA degradation (1.9%) (Fig. 6b). Notably, RNA‐binding proteins characterized by RNA‐binding domains such as the K‐Homology (KH) domain and RNA‐recognition motif were detected in this work (Table S4). Protein interaction network prediction further highlighted three functional clusters related to RNA binding, DNA binding, and catalytic activity (Fig. 6c). These results confirm the feasibility of our approach for screening circRNA‐binding proteins.
Identification of Vv‐circCOR27‐binding proteins. (a) The workflow and principle of an in vivo crosslinking circRNA pull‐down assay. Formaldehyde crosslinking of calli, followed by pull‐down with biotinylated antisense oligos and streptavidin beads, with subsequent LC‐MS analysis. (b) The GO molecular function of proteins identified in the circRNA pull‐down assay. The top 21 GO terms were presented, a.a., amino acid. (c) The protein interaction network of the proteins identified in the circRNA pull‐down assay. The protein interaction was predicted by the STRING platform (https://cn.string‐db.org/) and visualized by Cytoscape software (https://cytoscape.org/).
Vv‐circCOR27
physically binds to VvHSP90.2b
To minimize interference from host gene‐derived proteins and crosslinked complexes, we further predict the possibility of binding between Vv‐circCOR27 and protein candidates using the RPISeq tool (Table S4) (Muppirala et al., 2011). The protein candidates were fused with GFP and transiently co‐expressed with Vv‐circCOR27 in Nb plants to test their binding using a circRNA pull‐down assay. The assay confirmed that Vv‐circCOR27 was significantly enriched by biotin‐tagged antisense probes (Fig. 7a,b), and a direct interaction with a cytoplasmic HSP, VvHSP90.2b, was identified (UniProt: F6I581) (Fig. 7c) (Banilas et al., 2012). To further verify the binding between Vv‐circCOR27 and VvHSP90.2b, we performed a RIP assay after the transient overexpression of Vv‐circCOR27 and VvHSP90.2b‐GFP in Nb plants. The RT‐qPCR result showed that, compared with GFP, Vv‐circCOR27 was significantly enriched by VvHSP90.2b‐GFP (Fig. 7d,e). These results indicated that Vv‐circCOR27 directly binds to VvHSP90.2b in vivo.
*Confirmation of binding between Vv‐circCOR27 and VvHSP90.2b. (a–c) Verification of the binding between Vv‐circCOR27 and VvHSP90.2b through circRNA pull‐down assay. Vv‐circCOR27 was co‐expressed with GFP (negative control) and VvHSP90.2b‐GFP (2b‐GFP) in Nb plants, respectively. Enrichment of Vv‐circCOR27 by antisense (AS) and sense (S) tags was verified by RT‐qPCR (a) and PCR (b), and VvHSP90.2b was confirmed by western blotting (c). circ: amplification of Vv‐circCOR27; Rcr: the full length of Vv‐circCOR27 elongated by rolling‐circle. (d, e) Verification of the interaction between Vv‐circCOR27 and VvHSP90.2b through RIP assay. GFP or VvHSP90.2b‐GFP (2b‐GFP) was co‐expressed with Vv‐circCOR27, respectively. GFP was immunoprecipitated and detected by western blotting (d), while enrichment of Vv‐circCOR27 was quantified by RT‐qPCR (e). (f) Schematic diagram of 6×MS2 insertion into the Vv‐circCOR27 vector. The 6×MS2 (green) is cloned from pTSK108‐6×MS2 (Li et al., 2016) and inserted into the first exon of Vv‐circCOR27 (Supporting Information Notes S1). Blue: flanking sequence; red: exon; gray line: intron. (g) PCR detection of Vv‐circCOR27 (502‐nt) and 6×MS2‐Vv‐circCOR27 (947‐nt) in Nb plants. (h) Schematic diagram of the cTriFC assay. (i) Fluorescence imaging of interaction between VvHSP90.2b truncation (F4) and Vv‐circCOR27 in Nb plants. Bar, 100 μm. Differences are compared using a t‐test (**, P < 0.01; **, P < 0.001). Closed circles represent the measured values (mean ± SD).
The interaction between Vv‐circCOR27 and VvHSP90.2b in vivo was further verification based on the MS2‐MCP system (X. Li et al., 2016; Z. Li et al., 2016). A 6×MS2 tag was inserted into the first exon of Vv‐circCOR27 (named 6×MS2‐Vv‐circCOR27) (Fig. 7f), which largely minimized the impact on the secondary structure of Vv‐circCOR27 (Fig. S5a,b). The PCR result showed that the insertion of 6×MS2 has no influence on the back‐splicing of Vv‐circCOR27 (Figs 7g, S5c). On the basis of these results, we designed a TriFC assay (Fig. 7h). Moreover, to determine which domain binds to Vv‐circCOR27, VvHSP90.2b was truncated to the HATPase_c domain (F1), NurA domain (F2), HSP90 domain (F3), and C‐terminus (F4) (Fig. S6a). We found that the MCP protein directly interacts with VvHSP90.2b (FL), F1, and F3 domains in vivo (Fig. S6b,c). Notably, the fluorescence imaging was only observed in the F4 domain, which was co‐expressed with 6×MS2‐Vv‐circCOR27, indicating that the C‐terminus of VvHSP90.2b is responsible for binding with Vv‐circCOR27 (Fig. 7i). Taken together, these results confirmed the binding capability between Vv‐circCOR27 and VvHSP90.2b in vivo.
Vv‐circCOR27
represses the expression of sHSPs by attenuating the VvHSP90.2b‐VvHsfA7a interaction in response to heat stress
Given that HSP90.1a/b was proven to inhibit the transcriptional activation of HsfA1a through the formation of a protein complex. We investigated whether Vv‐circCOR27 binding to VvHSP90.2b modulates thermotolerance by affecting its interaction with VvHsfA proteins. We found that VvHSP90.2b interacted with VvHsfA1a and VvHsfA7a, respectively (Fig. 8a–c). And using luciferase complementation imaging (LCI) and bimolecular fluorescence complementation (BiFC) assays, we observed that co‐expression of Vv‐circCOR27 significantly reduced the interaction between VvHSP90.2b and VvHsfA7a, but not VvHsfA1a, compared to the empty vector control (Fig. 8d,e), suggesting that Vv‐circCOR27 interfered with the interaction of VvHSP90.2b and VvHsfA7a in vivo.
Disturbance of the interaction between VvHSP90.2b and VvHsfA7a. (a–c) Verification of VvHSP90.2b (90.2b) and VvHsfA1a/7a (1a/7a) interaction using bimolecular fluorescence complementation (BiFC) assay (a), luciferase complementation imaging (LCI) assay (b), and yeast two‐hybrid (Y2H) assay (c). (d, e) Test of Vv‐circCOR27 interference with the interaction of VvHSP90.2b and VvHsfA1a/7a in Nb plants using LCI assay (d) and BiFC assay (e). In LCI and BiFC assays, VvHSP90.2b was fused with nLuc (90.2b‐nLuc) or nYFP (nYFP‐90.2b) and VvHsfA1a/7a was fused with cLuc (cLuc‐1a/7a) or cYFP (1a/7a‐cYFP); Vv‐circCOR27 was expressed based on ‘intron retention’ strategy and empty vector (EV) was used as control. In Y2H assay, the full‐length (2b‐FL) and F4 fragment (2b‐F4, Supporting Information Fig. S6) of VvHSP90.2b were used to test. GFP, green fluorescent protein; BF, bright field; AD, pGADT7 vector; BD, pGBKT7 vector; EV, empty vector.
The potential effects of VvHsfA7a and its targets are not yet comprehensively excavated in grapevines. To address this, we conducted a DAP‐seq assay. The analysis showed that VvHsfA7a binding peaks were primarily located in distal intergenic regions (47.8%), introns (25.44%), and exons (14.06%) (Fig. 9a). A total of 9587 (11.7%) binding peaks shared with two biological replicates were assigned to the promoter region (≤ 2 kb) (Fig. 9a; Table S5). De novo motif analysis revealed a binding motif (CNNGAANNTTC) (named motif_1) (Fig. 9b). GO functional enrichment of the putative genes associated with motif_1 revealed that the targeted genes of VvHsfA7a were related to heat‐temperature response, ion transport, development, and metabolic processes (Fig. S7a).
The expression of sHSP genes and proposed model for the Vv‐circCOR27 regulatory mechanism. (a) Distribution of the enriched VvHsfA7a DAP‐seq fragments. (b) The motif_1 enriched based on VvHsfA7a binding sites. (c) Identification of HSP family members based on the motif_1. (d) The expression level of sHSPs was verified in calli at 45°C for 3 d. (e) Identification of motif_1 in the promoter of differentially expressed sHSPs and yeast one‐hybrid (Y1H) verification of the interaction between VvHsfA7A and this motif in both the promoter of sHSP18.2c (psHSP18.2c, WT) and its artificial mutant (Mut); AbA, Aureobasidin A. (f) Proposed model for Vv‐circCOR27 regulatory mechanism. At 26°C, VvHSP90.2b and VvHsfA7a maintain a lower expression level and abundant Vv‐circCOR27 sequesters VvHSP90.2b. On the condition of 45°C, the expression level of Vv‐circCOR27 is repressed, whereas VvHsfA7a is significantly induced. Sufficient VvHSP90.2b, likely through its chaperone activity, sustains the transcriptional activation of VvHsfA7a, thereby enhancing the expression of sHSPs. Differences are compared using a t‐test (, P < 0.05; **, P < 0.01; **, P < 0.001; ns, no significant difference). Closed circles represent the measured values (mean ± SD). A proposed genetic locus is shown using cylinders (genes, by color) and blunt arrows (unknown regions). Solid orange arrows indicate promotion (Bloost); dashed arrows indicate inhibition (Down); solid black arrows indicate no effect (Hold).
Analysis of motif_1 targets identified 65 HSP family genes, including 17 HSP70, 5 HSP90, and 43 small HSP (sHSP) members (Fig. 9c). We emphatically test the expression of sHSPs, which have been shown to be induced by heat in grapevine (Zha et al., 2020) (Table S6). Ultimately, 17 sHSPs were verified in grapevine callus, and more than half (9 sHSPs), that is, HSP15.7, HSP17.3, HSP17.3a, HSP17.3b, HSP18.1a, HSP18.1d, HSP18.2c, HSP18.2d, and HSP23.6, were significantly induced by heat stress, and their expression levels were lower in Vv‐circCOR27‐OE calli than in WT calli after heat treatment (Figs 9d, S7b), and the Y1H assay proved that VvHsfA7a binds to such a motif identified in the promoter of such sHSP genes (Fig. 9e). The expression levels of VvHsfA1a, VvHsfA7a, and VvHSP90.2b were also monitored in calli, revealing that the VvHsfA7a expression was upregulated 300‐ to 400‐fold in calli under heat treatment, whereas an inappreciable upregulation of VvHsfA1a and VvHSP90.2b expression was observed (Fig. S7c). Taken together, these results indicate that the Vv‐circCOR27‐VvHSP90.2b‐VvHsfA7a module plays a negative role in regulating the expression of sHSP genes.
Discussion
In Arabidopsis, AtCOR27 expression is induced by cold, repressed by blue light, and slightly downregulated at warm temperatures (X. Li et al., 2016; Z. Li et al., 2016). In this work, the transcription of VvCOR27 was also repressed by heat stress (Fig. 2c) and enhanced by cold (Fig. S1c) in grapevine, suggesting that the regulatory cascade is possibly conserved among plants. The expression pattern of Vv‐circCOR27 was consistent with that of its host gene under heat stress (Fig. 2d), but it did not follow the host gene's strong induction under cold (Fig. S1d), indicating likely posttranscriptional regulation. In eukaryotes, mature mRNAs and circRNAs are processed from a common precursor mRNA transcribed by RNA polymerase II (Chen, 2020). However, in grapevines, only a weakly positive correlation was found between circRNAs and mRNA expression level (Gao et al., 2019). In our previous work, the expression level of Vv‐circPTCD1 is unaffected by drought stress, while the expression of its host gene VvPTCD1 is significantly downregulated (Ren et al., 2023). In fact, the abundance of circRNAs is largely governed by posttranscriptional processing (Liu et al., 2023). CircRNA biogenesis often occurs in the middle exons of pre‐mRNAs; therefore, the flanking introns determine the occurrence of back‐splicing, in which introns provide indispensable binding sites for the spliceosome and specific cis‐acting elements (Song et al., 2021). Moreover, circRNA processing is mediated by RNA‐binding proteins, such as Quaking (QKI), a KH family protein, that binds to the specific motifs of flanking introns and facilitates loop formation (Conn et al., 2015). Therefore, the expression of Vv‐circCOR27 is regulated by additional undiscovered factors in the grapevine in response to temperature stress.
The biological functions of circRNAs are widely investigated by overexpression and silencing (Song et al., 2021; Zhou et al., 2021). To overexpress circRNAs in vivo, back‐splicing is triggered by inserting the circRNA‐producing genomic DNA or CDS containing the flanking intron sequence, while cognate linear RNA expression needs to be designed in parallel as a background control (Tan et al., 2017; Gao et al., 2019, 2023; Guarnerio et al., 2019). However, two barriers have been faced in our previous work: (1) the efficiency of back‐splicing is low, resulting in much lower circRNA accumulation than cognate linear RNA (Fig. 3a), and (2) circRNA and cognate linear RNA potentially play the same cellular roles owing to the consistent sequence (Ren et al., 2023). In this work, an ‘intron retention’ strategy was developed and significantly increased the accumulation of circRNAs when more than three introns were artificially inserted into circRNA‐producing CDS (Fig. 3). Interestingly, the precision of splicing was guaranteed, as the intron was cloned from grapevine and the insertion sites (AG/G) were subjected to the conserved U2 spliceosome in eukaryotes (Meyer & Staiger, 2015). It has been suggested that back‐splicing is coupled with canonical splicing and that the efficiency of back‐splicing is much lower than that of linear RNA, causing nascent pre‐RNAs to be processed mostly into linear RNA (Gao et al., 2016; Zhang et al., 2016). Nevertheless, an increase in intron number delays the canonical splicing of linear RNA and retains more time for back‐splicing, leading to hindering the nascence of cognate linear RNA.
Here, we showed that Vv‐circCOR27 overexpression enhanced the thermo‐sensitivity of grapevine calli (Figs 4, 5). Accumulating evidence suggests that circRNAs function by regulating the expression of their host genes (Gao et al., 2023). Here, we found that the expression level of the host gene was also influenced by overexpression of Vv‐circCOR27 (Fig. S2c). However, the thermo‐sensitivity was unaffected in VvCOR27‐OE calli (Fig. 4a–c), this result was confirmed by testing the thermotolerance in cor27 mutants of Arabidopsis (Fig. S3). Increasing findings suggest that circRNAs can bind to their cognate DNA locus and form an RNA: DNA hybrid (R‐loop) (Conn et al., 2017; Xu et al., 2020; Song et al., 2021). R‐loops are functional structures involved in multiple types of chromatin modifications, such as DNA hypermethylation, transcriptional activation or repression of gene loci, DNA replication, and genome stability (Xu et al., 2017; Liu et al., 2020, 2021). Nucleus‐localized EIciRNAs, a class of circRNAs that exons are circularized but the introns ‘retained’ between exons were widely identified in animals, interact with U1 snRNPs and promote transcription of their host genes (Li et al., 2015). In our work, Vv‐circCOR27 localized to the cytoplasm and nucleus (Fig. 1f), which provides a potential mechanism by which Vv‐circCOR27 regulates the expression of the host gene and neighbor genes (Fig. S2) through the formation of R‐loops or interaction with trans‐acting factors in the nucleus. However, the mechanism has not yet been elucidated in plants.
Interestingly, our results found that the cognate linear RNA enhanced calli thermotolerance (Fig. 5), which conflicts with the idea that the overexpression of linear RNA overlapping with circRNA is usually used as a background control without biological function (Gao et al., 2019; Liu & Chen, 2022). However, the circRNA and cognate linear RNA potentially function in a sequence‐dependent manner (Ren et al., 2023) or by coding (Fig. S4), highlighting the importance of the ‘intron retention’ strategy to evade the functional overlap of linear RNAs. Subsequent experiments that monitored translation in the cytoplasm suggested that pervasive translation events outside the canonical main ORF sequence occurred in eukaryotes (Wei & Guo, 2020). Numerous studies have shown that the short ORFs hidden in pri‐miRNAs are translated into regulatory peptides (miPEPs), such as miPEP165a, miPEP171d, and miPEP858a (Lauressergues et al., 2015; Chen et al., 2020; Sharma et al., 2020). Therefore, we propose that the thermotolerance of Linear_COR27‐OE calli is potentially caused by the protein translated from ORF1 hidden in linear RNA (Fig. S4).
Herein, we showed that Vv‐circCOR27 physically binds to cytoplasmic VvHSP90.2b (Figs 6, 7), expanding the regulatory mechanism of circRNAs in plants. The repression of transactivation activity of HsfA1s is caused by specific binding between HSP90/70 and the central TDR domain, which is highly conserved among HsfA1s in plants and is not characterized in Hsf members that belong to other groups (Ohama et al., 2016). These results imply that HSP90.2b possibly acts as a molecular chaperone for other Hsf members, such as HsfA7a, to maintain their activity. For example, the DNA binding activity of HsfA1 and the coactivator function of HsfB1 are repressed by HSP70, whereas the DNA binding activity of HsfB1 is stimulated by the chaperone HSP90 (Hahn et al., 2011). Therefore, we supposed that the transactivation activity of VvHsfA7a is maintained by the chaperone function of VvHSP90.2b and is weakened by competitive binding of Vv‐circCOR27.
In our work, DAP‐seq analysis and the Y1H assay confirmed the binding between VvHsfA7a and the promoter region of sHSP genes through the classical Hsf‐associated motif in grapevine (Fig. 9b,e). Therefore, on the basis of the expression pattern of sHSP family members in response to heat stress (Fig. 9d), a competitive binding model was proposed to illustrate the regulatory mechanism of Vv‐circCOR27, in which Vv‐circCOR27 binds to VvHSP90.2b and impedes the protection for VvHsfA7a, a client protein of the chaperone, downregulating the expression of sHSP genes and exacerbating the thermotolerance (Fig. 9f). It is an adaptive evolution to respond to heat stress through repressing the expression of Vv‐circCOR27 in thermotolerant cultivars of grapevine (Fig. 2d,e).
Competing interests
None declared.
Author contributions
YR performed the key experiments, prepared figures, and wrote this manuscript. YX performed the RT‐qPCR, LCI, and BiFC assays. ML performed GeneBridge analysis. LZ, YS, and JingJing Liu performed the heat treatment assay of grapevine. Junpeng Li, JW and QZ tested the Arabidopsis thermotolerance. DF, ZZ and JH performed the calli sub‐cultured and transgenic experiments. ZG annotated the circRNA in grapevine. ZY revised the manuscript. CM obtained funds, revised the manuscript, conceived, and designed the research.
Disclaimer
The New Phytologist Foundation remains neutral with regard to jurisdictional claims in maps and in any institutional affiliations.
Supporting information
Fig S1 Workflow of VvCOR27 functional analysis and the expression pattern of VvCOR27 and Vv‐circCOR27 under cold stress. Fig. S2 Overexpression of Vv‐circCOR27, Linear_COR27, and VvCOR27 in grapevine calli. Fig. S3 The mutant thermotolerance of AtCOR27 in Arabidopsis. Fig. S4 Coding capability test of linear RNA Linear_COR27. Fig. S5 Structure prediction and sequencing of 6×MS2‐Vv‐circCOR27 expression in Nb plants. Fig. S6 Visualization of binding between Vv‐circCOR27 and VvHSP90.2b. Fig. S7 Go enrichment of HsfA7a targets and expression changes of Hsf and HSP members in grapevine calli. Notes S1 All sequences used in this work.
Table S1 Primer pairs used in this work. Table S2 Summary of data resources. Table S3 Functional annotation of COR27 based on G‐MAD. Table S4 Identification of Vv‐circCOR27 binding protein using LC‐MS/MS analysis. Table S5 The VvHsfA7a targets based on DAP‐seq analysis. Table S6 Small heat shock proteins identified in this work using VvHsfA7a DAP‐seq.Please note: Wiley is not responsible for the content or functionality of any Supporting Information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Banilas G , Korkas E , Englezos V , Nisiotou AA , Hatzopoulos P . 2012. Genome‐wide analysis of the heat shock protein 90 gene family in grapevine (Vitis vinifera L.). Australian Journal of Grape and Wine Research 18: 29–38.
- 2Bartlett A , O’Malley R , Huang SS , Galli M , Nery JR , Gallavotti A , Ecker JR . 2017. Mapping genome‐wide transcription‐factor binding sites using DAP‐seq. Nature Protocols 12: 1659–1672.28726847 10.1038/nprot.2017.055PMC 5576341 · doi ↗ · pubmed ↗
- 3Capelari EF , da Fonseca GC , Guzman F , Margis R . 2019. Circular and micro RN As from Arabidopsis thaliana flowers are simultaneously isolated from AGO‐IP libraries. Plants 8: 302.31454955 10.3390/plants 8090302 PMC 6784238 · doi ↗ · pubmed ↗
- 4Charng YY , Liu HC , Liu NY , Chi WT , Wang CN , Chang SH , Wang TT . 2007. A heat‐inducible transcription factor, Hsf A 2, is required for extension of acquired thermotolerance in Arabidopsis . Plant Physiology 143: 251–262.17085506 10.1104/pp.106.091322 PMC 1761974 · doi ↗ · pubmed ↗
- 5Chen LL . 2020. The expanding regulatory mechanisms and cellular functions of circular RN As. Nature Reviews Molecular Cell Biology 21: 475–490.32366901 10.1038/s 41580-020-0243-y · doi ↗ · pubmed ↗
- 6Chen QJ , Deng BH , Gao J , Zhao ZY , Chen ZL , Song SR , Wang L , Zhao LP , Xu WP , Zhang CX et al. 2020. A mi RNA‐encoded small peptide, vvi‐mi PEP 171d 1, regulates adventitious root formation. Plant Physiology 183: 656–670.32241877 10.1104/pp.20.00197 PMC 7271809 · doi ↗ · pubmed ↗
- 7Cheng J , Zhang Y , Li Z , Wang T , Zhang X , Zheng B . 2018. A lariat‐derived circular RNA is required for plant development in Arabidopsis . Science China Life Sciences 61: 204–213.29101586 10.1007/s 11427-017-9182-3 · doi ↗ · pubmed ↗
- 8Chu C , Qu K , Zhong FL , Artandi SE , Chang HY . 2011. Genomic maps of long noncoding RNA occupancy reveal principles of RNA‐chromatin interactions. Molecular Cell 44: 667–678.21963238 10.1016/j.molcel.2011.08.027PMC 3249421 · doi ↗ · pubmed ↗