Transgenic Phenotypic Differences of Grapevine Circular RNA Between Arabidopsis and Grapevine Callus Imply the Species-dependent Functions of Vv-circPTCD1

1. Introduction

Noncoding RNAs have increasingly crucial functions for growth and development and stress response in plants [1,2]. Circular RNAs (CircRNAs), which has not been found the coding ability in plants yet, are single strand non-coding RNA that are processed by backing-splicing, in which a downstream 5׳ splice site is linked by 3′,5′- phosphodiester bond to an upstream 3′ splice site in a reverse order crossing one exon or exons [3]. Owing to the development of high-throughput sequencing technology and associated analytical tools, thousands of circRNAs have been identified in various plants and were deposited in associated database [4,5]. At present, it has been reported that a total of 171,118 circRNAs from 21 plant species have been collected in PlantcircBase [6]. The biological functions of circRNA have been well investigated in mammals. Animal circRNAs have generally been suggested to function as miRNA sponges for decaying endogenous miRNA [7], as templates for translation [8,9], as mediator for regulating assembly and activation of AMPK complex under stress [10], as enhancer for cis-regulation of their host genes [11]. In plants, previous studies have focused on identification and annotation of putative circRNAs as well as function investigation. Until now, many studies have provided direct evidences that circRNAs shown important functions in plants by overexpression and gene editing technologies. Overexpression of circRNA derived from PSY1 (Phytoene Synthase 1) and PDS (Phytoene Desaturase) in tomato, respectively, causes color variation [12]. In Arabidopsis, circSEP3 from exon 6 of SEPALLATA3 can regulates alternative splicing of its host genes through R-loop formation with homological genomic region. Overexpression of this circRNA results in floral organ abnormality [13]. CRISPR-Cas9 strategy has been applied to remove Os06circ02797 processing by editing the flanking sequence in rice, resulted in a rapid growth phenotype and higher chlorophyll A/B content [14]. In Populus tomentosa, overexpression of Circ_0003418 derived from a RING-type E3 ligase gene XBAT32 deteriorates the callus thermotolerance [15]. Although some studies have been performed, functional investigation on circRNAs in plants is still very limited. Therefore, additional researches are further conducted.

Grapevine (Vitis spp.) is the most economically important fruit crops cultivated worldwide, for the purpose of producing table grape, dried fruits, and as the raw materials for juice and wine making [16]. The growth, development and flavor of berries are strongly influenced by various abiotic stresses, such as high temperature, water deficit, and salt during viticulture 17,18,19]. Previous findings suggested that high temperature influences grapevine photosynthesis, florescence and veraison [20,21,22]. In grapevine, the serine/arginine-rich splicing factors (SR) are generally phosphorylated on the condition of high temperature, which results in the disturbance of alternative splicing of pre-mRNA [23,24]. The mechanisms involved in grapevine tolerance to high temperature are poorly understood, although the heat shock factors (HSFs)-mediated regulatory networks have been well deciphered in other plants [25,26,27]. Drought is another important environment factor limiting grapevine growth and productivity, and studies have shown that the photosynthesis of vegetative organs, berries size and yield, and accumulation of secondary metabolites are greatly affected by the water deficit [28]. The abscisic acid (ABA) is induced and plays a key role for response to water deficit by regulating stomatal movements, as well as the signaling pathway has been well studied in plants [29]. Furthermore, the long non coding RNAs (LncRNA) and microRNAs (miRNA) are also serves as novel regulators of the plant response to drought stress in Arabidopsis [30], Brassica napus L [31], and grapevine [32]. Salt stress is another factor which is mainly due to the hyperosmotic conditions of soil solution with increase of Na⁺ and Cl^-, causing the deficiency of the absorption of water and nutrients in plants [33]. In grapevine, previous studies found that many stress-induced genes, such as VvWRKY2, VvNAC17, and VvASMT1, perform functions for enhancing salt and osmatic stresses tolerance in ectopic overexpression [19,34,35]. Although the above findings greatly uncover the molecular mechanisms responding to abiotic stress in grapevine, the function of circRNAs also need to be further explored.

In grapevine, according to our previous study, a total of 8, 354 circRNAs have been identified by the whole transcriptome analysis, and a Vv-circATS1 spliced from glycerol-3-P acyltransferase was proved to improve cold tolerance in Arabidopsis [36]. In present study, a circRNA derived from the second exon of the pentatricopeptide repeat domain-containing protein 1 (PTCD1, VIT_211s0016g03830) was verified based on the grapevine circRNA dataset, named Vv-circPTCD1. Pentatricopeptide repeat (PPR) proteins is a large gene family in plants, which are characterized by 2 to 15 tandem repeats of 30-40 amino acid length motif. Some PPR proteins have been supposed to play roles in organellar RNA metabolism, organ development as well as stress response [37]. Overexpression of Vv-circPTCD1 deteriorated the tolerance to heat, salt, and PEG-mediated osmotic stresses in Arabidopsis. While, similarly, the linear fragment identical with the circRNA sequence also did. However, the Vv-circPTCD1 overexpressed callus mass of ‘Thompson Seedless’ were failure to display the phenotypes consistent with Arabidopsis. Those results implied that the functions of Vv-circPTCD1 potentially were sequence-dependent and not conserved among plants.

2. Results

2.1. Identification and verification of Vv-circPTCD1

Based on the previous published circRNA dataset of ‘Muscat Hamburg’ [36], a circRNA_4363, back-spliced from the second exon of a pentatricopeptide repeat (PPR) proteins family gene PTCD1, was identified and renamed Vv-circPTCD1 in this research. The PTCD1 encodes a protein with 636 amino acid residues length and contains the 9 tandem PPR domain (Figure 1A). In Arabidopsis genome, meanwhile, two orthologs (AT5G21222 and AT5G25630) were characterized and phylogenetic analysis showed that AT5G21222 and PTCD1 (VIT_211s0016g038030) were clustered into one clade with 96% confidence (1000 replicates) (Figure 1B). The Vv-circPTCD1 were further verified by cloning and sequencing as well as the back-splicing site was AG/GA (Figure 1C). Divergent and convergent primers were designed to amplify the target fragment form DNA and leaf cDNA, result showed that the fragment with the back-splicing site was only detected in cDNA templates (Figure 1D).

2.2. Nonconservative back-splicing of Vv-circPTCD1 in plants

To investigate whether the Vv-circPTCD1 was conserved in plants, we analyzed the physical gene structure of PTCD1 in plants, including monocots and dicots. The result showed that similar physical gene structure of PTCD1 orthologs was observed (Figure S1), as well as a CDS (coding sequence) back-spliced into Vv-circPTCD1 in grapevine was found and it was highly conserved by multiple sequence alignment among plants (Figure S2). It was reasonably hypothesized that this circRNA orthologous with Vv-circPTCD1 maybe also conserved in plants. Therefore, the divergent primer pair was tentatively designed to clone the orthologs in Arabidopsis and N. benthamiana, although the non-specific amplification was found in N. benthamiana, while the Vv-circPTCD1 orthologs were failure to cloned (Figure S3). Those results suggested that the back-splicing of Vv-circPTCD1 potentially was species-dependent.

2.3. The expression analyses of PTCD1 and Vv-circPTCD1 under abiotic stress

To investigate the potential functions in grapevine, we analyzed the tissue specificity and the expression patterns of Vv-CircPTCD1 and its corresponding host gene under salt, heat, and drought stress conditions, respectively. The Vv-CircPTCD1 was mainly expressed in young and mature leaves while scarcely expressed in roots, consistent with that of host gene (Figure 2A). However, the accumulation level of host gene was excessively higher than that of Vv-CircPTCD1 (more than sixty times) (Figure 2A). During salt stress, in addition to the host gene was mildly down-regulated during initial treatment, the Vv-CircPTCD1 was significantly down-regulated on the fifth day after treatment (Figure 2B, C). During drought treatment, the VvSnRK2I (VIT_207s0197g00080), a marker gene homologous with AtSRK2I/AtSnRK2.3 induced by ABA in Arabidopsis [42], was significantly induced when the soil water potential reduces to -0.8 and -1.0 MPa (Figure 2D). The expression level of PTCD1 was mildly up-regulation when the water potential was -0.4 MPa, while the expression level of PTCD1 was significantly reduced when the water potential was reduced to -0.8 and -1.0 MPa (Figure 2E). However, the expression pattern of Vv-CircPTCD1 was barely changed during drought treatment (Figure 2F). We further investigated the expression pattern during heat stress, the results showed that the heat shock factor genes VvHsfA7, as a marker of heat stress [20], were remarkable up-regulation (Figure 2G) and the expression level of PTCD1 was down-regulation with the prolongation of treatment (Figure 2H). Although the expression level of Vv-CircPTCD1 was down-regulation after 4 h treatment, however, the expression level was also slightly up-regulation after 2 h treatment (Figure 2I). Those results indicated that the Vv-CircPTCD1 expression was always fell behind that of PTCD1 and potentially play a role during abiotic stresses in grapevine.

2.4. Overexpression of Vv-CircPTCD1 in grapevine callus and Arabidopsis

To decipher the functions of Vv-CircPTCD1 in plants, the vector for Vv-CircPTCD1 overexpression (OE) was generated, meanwhile, corresponding linear fragment (Linear_PTCD1) identical with Vv-CircPTCD1 was also constructed as a control (Figure 3A). The overexpressed callus mass was verified by RT-PCR and RT-qPCR, respectively (Figure 3B, C, D, E). The expression level of host gene and flanking genes were further investigated in all OE callus mass. Among Vv-CircPTCD1 lines, the expression level of host gene was slightly down-regulated (Figure 3F), while the up-stream gene (VIT_211s0016g03850) and down-stream gene (VIT_211s0016g03820) were not significantly affected (Figure 3G, H). However, among Linear_PTCD1 OE lines, the expression level of host gene was discrepant among different lines, which is down-regulated in line 2 (Figure 3I). As well as the up-stream gene and down-stream gene were also slightly down-regulated (Figure 3J, K). In order to uncover whether the function of Vv-CircPTCD1 is conserved among plants, simultaneously, ectopic transformation was also performed in Arabidopsis (Figure S4).

2.5. Phenotype of OE lines under heat stress

To investigate whether Vv-CircPTCD1 plays a role in Arabidopsis, the four-weeks old seedlings were treated 12 h at 45℃ followed by exposure to 22 ℃ for 7 days to recovery. Results showed that, comparing with WT, the rosette leaves were significantly deteriorated in Vv-CircPTCD1 OE lines (Figure 4A,B). It was a wonder that the rosette leaves were also remarkably damaged in linear fragment OE lines, which keep an identical sequence with Vv-CircPTCD1 (Figure 4A, B). We also observed the phenotype of transgenic callus responding to heat stress, in which the callus was continuously incubated 15 d at 40℃. However, phenotypic differences were nonevent, no matter which OE callus mass (Figure 4C and Figure S5).

2.6. Phenotype of OE lines under salt stress

To further survey the function of Vv-CircPTCD1 in plants, the four days old seedlings of Arabidopsis were transplanted into normal and NaCl-containing medium to observe the variation. Results suggested that different OE lines and WT seedlings were normally and identically grow in NaCl-free medium (Figure 5A), while the survival rate of transgenic lines were lower on the NaCl-containing medium after five days compared to WT (Figure 5B, C). In grapevine callus, the OE callus mass transformed linear fragment adversely displayed albino phenotype, while the OE callus mass transformed Vv-CircPTCD1 was also did but much slighter compared to WT callus mass (Figure 5D). Those results also showed the Vv-CircPTCD1 potentially served as negative factor for salt stress and the linear fragment identical with the Vv-CircPTCD1 also plays the same roles in Arabidopsis and grapevine callus.

2.7. Phenotype of OE lines under drought stress

To comprehensively explore the function of Vv-CircPTCD1 in Arabidopsis, the four weeks old seedlings of Arabidopsis were stopped watering. Results showed that the transgenic lines were much more susceptible and withering (Figure 6A). The four days old seedlings were further transplanted into normal and PEG-containing medium, and found that different transgenic lines and WT seedlings were normally and identically grow in normal medium (Figure 6B), while the growth of rosette leaves were obviously inhibited in OE lines but the growth of root barely did (Figure 6C, D, E). Those results suggested that the Vv-CircPTCD1 and corresponding linear fragment also play negative roles during drought stress in Arabidopsis. In grapevine callus, oppositely, the OE callus mass of transgenic callus mass significantly displayed resistance to osmotic stress mediated by PEG-8000, while the WT has become brawn after 30 days incubation (Figure 6F). We further counted the growth index of callus mass, and found that the growth rate of transgenic callus was significantly higher than that of WT (Figure 6G and Figure S5). Those results implied that the function of Vv-CircPTCD1 potentially is species-dependent.

3. Discussion

CircRNAs are widespread in plants. Most circRNAs are generated by backing-splicing from the exon or exons of genic primary mRNA, for example 94.5% are derived from the coding genes in rice and the remaining were intergenic circRNAs [43]. In addition, the mitochondrion genome is also proved to generate circRNAs, called mitochondrion-encoded circular RNAs (mcircRNAs) [44]. Here, we identified a circRNA spliced from the second exon of PTCD1, a PPR family genes, in which it was spliced based on canonical GU/AG splicing signal [36] (Figure 1C). In general, hundreds of this superfamily proteins encoded by the PPR family genes located in the nuclear and organelle genome are found in plants and characterized by multiple tandem arrays of 35 residues PPR domain [45]. Previous finding proved that the PPR proteins, served as RNA binding protein, affect cytoplasmic male sterility, embryogenesis, seed development, and stress [46,47,48,49]. Interestingly, the physical gene structure of PTCD1 orthologs are similar, as well as the sequence of the CDS back-spliced into Vv-circPTCD1 are highly conserved among plants (Figure S1 and S2). However, the verification failure of corresponding circRNA in Arabidopsis and N. benthamiana imply that the circPTCD1 is not naturally conserved among plants. No matter what mammals and plants, a factually small proportion of circRNAs are supposed to be conserved, such as the overlap of 26% circRNAs in both human and mouse based on the syntenic region of genome [8]. In plants, only 8.7% circRNAs are genomic conservation in dicotyledonous plants, even that are less than half in Oryza genus [43]. Taken together, the Vv-circPTCD1 is not evolutionarily conserved, although the host genes are conserved. And this result partially supports the idea that circRNAs result largely from splicing errors [50,51].

We also successfully constructed the artificial vector containing a reverse complementary sequence pairs to generate the circRNA in Arabidopsis and grapevine callus according to the previous strategy, although the natural back-splicing of Vv-circPTCD1 is not conserved (Figure 3A) [36]. The intronic complementary sequences have been found in the flanking sequence of highly expressed circRNA in eukaryotes. In mammals, the exons flanked by the orientation-opposite Alu elements, a shorted interspersed elements in primates, are preferentially circularized [52]. In Populus tomentosa, the miniature inverted repeat transposable elements (MITEs) are enriched in circRNA flanking regions and served as cis-regulatory factor to significantly regulate the exon circularization [15]. Furthermore, previous finding supposed that the back-splicing of circRNA is also executed by the canonical spliceosome, a highly conserved precursor mRNA splicing mechanism in eukaryotes [53]. Additionally, the flanking intron sequence is crucial adjective for precise splicing and circularization of one exon and exons, which has been proved by the heterologous investigations [36,54]. Therefore, the inverted repeats and flanking sequence are commonly contributed to the fidelity and expression level of Vv-CircPTCD1 in Arabidopsis and grapevine callus (Figure 3B, C and Figure S4). Therefore, the artificial “stem-loop” structure mediated by the reverse complementary sequence greatly contributes to the back-splicing of Vv-CircPTCD1 in heterogeneous expression, in which the “stem-loop” brings the splice sites into close proximity to each other [55].

Additionally, previous investigations uncovered that circRNAs are incorporated into chromatin by R-loop structure interacting with the nuclear DNA to regulate the chromatin stabilization or alternative splicing [13,56]. The host gene and neighboring genes are potentially influenced thorough overexpressed or deleted circRNA. In rice, a neighboring gene is significantly up-regulated at circRNA mutant lines [14]. In this study, the Vv-CircPTCD1 has slightly negative effect on the host gene, and barely do on the neighboring genes after overexpression of Vv-CircPTCD1 (Figure 3F, G, H). In P. tomentosa, overexpression of Circ_0003418 significantly increased the expression abundance of its parental gene [15]. Therefore, the circRNA potentially have the ability to function through disturbing the expression of host or neighboring genes in plants.

The expression pattern of circRNAs is also determined by tissue and environment factors, and even genotype [15,57]. We found that the Vv-CircPTCD1 focused in this research were preferentially expressed in young and mature leaves, which is consistent with the expression pattern of corresponding host gene, while the expression abundance of Vv-CircPTCD1 greatly less than 2% that of host gene (Figure 2A). In addition to some circRNAs in animals (almost 10%) [8], the circRNAs barely hold a higher expression level with respect to the linear counterpart and are largely obeyed the expression pattern of host gene in plants, such as grapevine [36]. In this study, the expression level of PTCD1 is significantly inhibited during drought stress, while the expression of Vv-CircPTCD1 is slightly decreased (Figure 2D, E, F), indicating it plays a role in grapevine responding to drought stress. In addition, the expression level of Vv-CircPTCD1 is remarkably decreased during salt stress after 5 days, while the host gene never occurred (Figure 2B, C). Those results imply that the Vv-CircPTCD1 plays a role during abiotic stress, in which the expression of circRNA potentially is controlled on the condition of specific abiotic stress, at least during the processing or degradation pathway [58].

The hypothesis of “splicing error” is indeed uncontradictory with the biochemical activities and biological functions of circRNAs. According the previous idea, the biogenesis of circRNAs and the regulation of their abundance are managed by various mechanisms, such as cis- or trans-elements, export and turnover of circRNAs [3]. In this study, although the expression of Vv-CircPTCD1 is related to the salt and drought stress, the mechanism is also unclear. For investigation of the biological function, a problem with their circular conformation and sequence overlap with linear cognate mRNAs and incomplete cyclization is difficult to overcome [53]. The strategy to evade this problem is to transform the linear fragment identical with circRNA as a control. Generally, the linear fragment scarcely functions in plants [13,15,36]. In present research, the linear counterpart is also transformed as a control (Figure 3D, E and Figure S4). Interestingly, identical stress-induced phenotype is observed between Vv-CircPTCD1 and its linear counterpart OE lines (Figure 4, Figure 5 and Figure 6). It is putative that (1) the incompletely cyclized transcript plays roles identical with linear counterpart; (2) the Vv-CircPTCD1 functions through sequence-dependent style. We observed that only the phenotype induced by salt stress is conserved between Arabidopsis and grapevine callus (Figure 5), and that is distinctly inconsistent during the treatment of heat and PEG (Figure 3 and Figure 6). Coincidentally, although the cognate sequence generated Vv-CircPTCD1 are greatly conserved among multiple species (Figure S2), while the biogenesis is not conserved (Figure S3). Those results imply that the biological functions of circRNA are potentially species-dependent in plants.

4. Materials and methods

4.1. Plant materials and treatments

For drought treatment of grapevine, 1-year-old ‘Muscat Hamburg’ cuttings were planted in a mixture of loam and sand (1:1, v/v) in flowerpot in a greenhouse. Cuttings with four well-developed leaves were drought stress treated. The well-developed leaves were collected when the soil water potential was 0 MPa, -0.2 MPa, -0.4 MPa, -0.6 MPa, -0.8 MPa, and -1.0 MPa, respectively. For heat stress treatment of grapevine, the cuttings were treated at 45 ℃, and the leaves were collected at 0 h, 1 h, 2 h, and 4 h, respectively. For salt stress, the cuttings were planted in a chamber (25 ℃) and watered the 300 mM NaCl solution, and the mature leaves were sampled at 0 d, 1 d, 2 d, 3 d, and 5 d. All samples were quickly frozen in liquid nitrogen and then stored at -80 ℃. All treatments were carried out in three biological replicates.

Arabidopsis ecotype Col-0 was cultured in chamber under 16 h light/8 h dark conditions with a light intensity of 600 mmol m^-2 s^-1 at a temperature of 22°C. For heat stress treatment of Arabidopsis, 6 weeks-old WT and transgenic plants were exposed to heat stress at 45◦C for 12 h followed by exposure to 22◦C for 7 days to recovery. For drought stress treatment, 6-weeks-old WT and transgenic plants were stopped watering until the substrate is completely dry, and observed the phenotype. The PEG-8000 treatment (the water potential was -0.7 MPa) was performed according to the Verslues׳s descriptions [38]. And the 4 days old seedlings were transferred into the plate containing PEG-8000 and cultured in chamber for another 5 days. For salt treatment, the plates containing 1/2 MS salts supplemented with 300 mM NaCl were prepared and 4 days old seedlings were incubated. The survival rates were counted after 5 days.

4.2. Validation of circRNA and RT-qPCR

The total RNA of Arabidopsis and grapevine callus was extracted by TRIzol reagent (Sangon Biotech, China). And the total RNA of grapevine leaves was extracted by CTAB method. In brief, 200 mg sample are powered and transferred into RNase-free 2 mL tube. Added quickly 900 μL pre-heated (65 ℃) CTAB buffer (2% CTAB, 100 mM Tris-HCl, 20 mM EDTA, 2 M NaCl, 2% PVP-40, pH8.0) and 20 μL β-mercaptoethanol. Mixed the mixture robustly and then bathed at 65℃ for 15 min. Extracted lysate with 900 uL of chloroform: isoamylol (24:1, v/v) by vortexing for 30 s and centrifuged at 4 ℃ for 10 min at 12, 000 rpm. Repeated two times with chloroform: isoamylol extraction. Added 20 μL NaAC (pH5.2) and 60 μL pre-cooled (-20℃) absolute alcohol, and bathed 10 min in ice. And added volume chloroform: isoamylol for vortexing. After centrifuged at 4℃ for 10 min at 12,000 rpm, added 150 μL 10 M LiCl to 450 μL supernatant and mixed. The RNA was precipitated at 4℃ for 6 to 8 h, and harvested by centrifugation at 4℃ for 10 min at 12,000 rpm. Carefully removed the supernatant, and washed two times by 70 % pre-cooled (-20℃) alcohol. The pellet was dried and resuspended by 30 μL DEPC-treated ddH₂O.

The total RNA was transcribed to cDNA using random primers by a FastKing RT Kit (TIANGEN, China) in accordance with the manufacturer’s instructions. To confirm the grapevine circRNAs predicted, divergent primer was designed by Primer 5 (Table S1). PCR was as follows: 94°C for 3 min; 40 cycles at 94°C for 30 s, 56°C for 15 s, and 72°C for 20 s; and then 1 cycle at 72°C for 5 min. For PCR, 2×Taq master mix (Vazyme, China) was used. The PCR products were separated by agarose gel electrophoresis and then purified. Sanger sequencing was further performed to verify the back-spliced junction sites. RT-qPCR (real-time quantitative PCR) analysis was conducted to evaluate the expression levels of circRNAs and linear counterparts using Talent qPCR PreMix (TIANGEN, China) with a qTOWER³ Real-Time PCR Detection System (Analytikjena, Germany). Gene expression levels were calculated by the 2^-∆∆Ct method [39].

4.3. Vector construction

All expression vectors were constructed based on the pHB binary plasmids. The construction of circRNA expression vector was conducted based on previous published strategy [36]. For linear RNA expression vector, the linear fragment derived from the same sequence with circRNA was cloned into pHB vector which was digested by BamH І and Xba І. The constructed plasmids were individually transformed into Agrobacterium tumefaciens GV3101 strains and incubated at 28℃ for 2-3 d on the LB medium supplemented with 25 mg·L^-1 rifampicin (Rif) and 50 mg·L^-1 kanamycin (Kan). A single clone was incubated at 200 rpm and 28℃ for 8-10 h in LB liquid medium containing 25 mg·L^-1 Rif and 50 mg·L^-1 Kan. The presence of the corresponding plasmid was verified by PCR. The positive clones were added glycerol and stored at -80℃.

4.4. Arabidopsis transformation and verification

Arabidopsis transformation was performed according to floral dip method [40]. T1 seedling were selected by spraying of Glufosinate ammonium 10% solution (dilute 1000×) (Sangon Biotech, China). The positive transgenic ones were transferred into pots and confirmed by PCR detection using the gene primers. For circRNAs confirmation, the cDNA was used as templates for PCR and RT-qPCR reaction and the splicing site was verified by further sequenced.

4.5. Callus transformation of ‘Thompson Seedless’ and treatment

The embryogenic callus was induced from floral explants of ‘Thompson Seedless’ according to a previously published protocol [41]. The callus mass was monthly sub-cultured in MSTP medium (MS base salts, 20 g·L^-1sucrose, 1 mg·L^-1 TDZ, 2.2 mg·L^-1 picloram, pH5.8) in dark condition at 26℃. For callus transformation, the A. tumefaciens containing vector was inoculated in 20 mL LB liquid medium and incubated until OD₆₀₀ reached 1.0. After centrifugation for 5 min at 6000 rpm and 25℃, the medium was discarded. The pellet was resuspended and the OD₆₀₀ was adjusted to 0.4 with MS medium (0.1 mM acetosyringone, pH5.8). The bacterial mixture was incubated with the embryogenic callus for 30 min at 40 rpm in 50 mL conical flask, and was then transferred onto filter paper to fully remove and evaporate excess liquid. Then, the callus was collected and transferred to solid MS medium (0.1 mM acetosyringone, 8g·L^-1 agar, pH5.8) and co-cultivated in dark condition at 26 °C for 3 days. After 3 days, the callus was transferred to solid MSTP medium containing 50 mg·L^-1 hygromycin, 200 mg·L^-1 Timentin. Petri dishes were monthly sub-cultured in the dark for the induction of transgenic callus. The expression of circRNA in transgenic callus was confirmed by PCR and RT-qPCR using divergent primer pairs.

For heat stress treatment, the callus mass was sub-cultured into MSTP medium for one week at 26 ℃ and then further cultured at 40 ℃ for 15 days. The callus mass was sub-cultured into MSTP medium supplemented with 300 mM NaCl for salt stress at 26 ℃. For PEG-8000 treatment, the plates were prepared according to above description, of which the base salt was replaced by MSTP medium supplemented with 15 g·L^-1 agar on solid layer but the liquid layer was the MSTP medium without sucrose and agar. Then the callus mass was sub-cultured into the PEG-containing plates and cultured at 26 ℃ for one month.

5. Conclusions

According to this result, Although the cognate counterpart mRNA is greatly conserved among plants, the biogenesis of circRNA generated from them potentially is not. Furthermore, the biological effects of circRNA are largely species-dependent.