Functional Differences of Grapevine Circular RNA Vv-circPTCD1 in Arabidopsis and Grapevine Callus under Abiotic Stress

Circular RNAs (circRNAs) serve as covalently closed single-stranded RNAs and have been proposed to influence plant development and stress resistance. Grapevine is one of the most economically valuable fruit crops cultivated worldwide and is threatened by various abiotic stresses. Herein, we reported that a circRNA (Vv-circPTCD1) processed from the second exon of the pentatricopeptide repeat family gene PTCD1 was preferentially expressed in leaves and responded to salt and drought but not heat stress in grapevine. Additionally, the second exon sequence of PTCD1 was highly conserved, but the biogenesis of Vv-circPTCD1 is species-dependent in plants. It was further found that the overexpressed Vv-circPTCD1 can slightly decrease the abundance of the cognate host gene, and the neighboring genes are barely affected in the grapevine callus. Furthermore, we also successfully overexpressed the Vv-circPTCD1 and found that the Vv-circPTCD1 deteriorated the growth during heat, salt, and drought stresses in Arabidopsis. However, the biological effects on grapevine callus were not always consistent with those of Arabidopsis. Interestingly, we found that the transgenic plants of linear counterpart sequence also conferred the same phenotypes as those of circRNA during the three stress conditions, no matter what species it is. Those results imply that although the sequences are conserved, the biogenesis and functions of Vv-circPTCD1 are species-dependent. Our results indicate that the plant circRNA function investigation should be conducted in homologous species, which supports a valuable reference for further plant circRNA studies.


Introduction
Noncoding RNAs have increasingly crucial functions for growth, development, and stress response in plants [1,2]. Circular RNAs (circRNAs), which have not been found to have the coding ability in plants yet, are single-stranded noncoding RNAs that are processed by backing-splicing in which a downstream 5 splice site is linked by a 3 ,5phosphodiester 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 deposited in an 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 overexpression [21,36,37]. Although the above findings greatly uncover the molecular mechanisms responding to abiotic stress in grapevines, the function of circRNAs also needs to be further explored.
In grapevine, according to our previous study, a total of 8,354 circRNAs have been identified by whole transcriptome analysis, and a Vv-circATS1 spliced from glycerol-3-P acyltransferase was proven to improve cold tolerance in Arabidopsis [38]. In the 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) protein is a large gene family in plants that are characterized by 2-15 tandem repeats of 30-40 amino acid length motifs. Some PPR proteins have been supposed to play roles in organellar RNA metabolism, organ development, as well as stress response [39]. Overexpression of Vv-circPTCD1 deteriorated the tolerance to heat, salt, and PEG-mediated osmotic stresses in Arabidopsis. While similarly, the linear fragment identical to the circRNA sequence also did. However, the Vv-circPTCD1 overexpressed callus mass of 'Thompson Seedless' failed to display the phenotypes consistent with Arabidopsis. Those results implied that the functions of Vv-circPTCD1 were potentially sequence-dependent and not conserved among plants.

Identification and Verification of Vv-circPTCD1
Based on the previously published circRNA dataset of 'Muscat Hamburg' [38], a circRNA_4363, back-spliced from the second exon of the pentatricopeptide repeat (PPR) proteins family gene PTCD1, was identified and renamed Vv-circPTCD1 in this research. PTCD1 encodes a protein with 636 amino acid residues in length and contains the nine tandem PPR domains ( Figure 1A). In the Arabidopsis genome, meanwhile, two orthologs (AT5G21222 and AT5G25630) were characterized, and phylogenetic analysis showed that AT5G21222 and PTCD1 (VIT_211s0016g038030.1) were clustered into one clade with 96% confidence (1000 replicates) ( Figure 1B). The Vv-circPTCD1 was further verified by cloning and sequencing, and the back-splicing site was AG/GA ( Figure 1C). Divergent and convergent primers were designed to amplify the target fragment from DNA and leaf cDNA, and the results showed that the fragment with the back-splicing site was only detected in cDNA templates ( Figure 1D). conditions of soil solution with an increase of Na + and Cl -, causing the deficiency of absorption in water and nutrients in plants [35]. In grapevine, previous studies found that many stress-induced genes, such as VvWRKY2, VvNAC17, and VvASMT1, perform functions for enhancing salt and osmotic stress tolerance in ectopic overexpression [21,36,37]. Although the above findings greatly uncover the molecular mechanisms responding to abiotic stress in grapevines, the function of circRNAs also needs to be further explored.
In grapevine, according to our previous study, a total of 8,354 circRNAs have been identified by whole transcriptome analysis, and a Vv-circATS1 spliced from glycerol-3-P acyltransferase was proven to improve cold tolerance in Arabidopsis [38]. In the present study, a circRNA derived from the second exon of the pentatricopeptide repeat domaincontaining protein 1 (PTCD1, VIT_211s0016g03830) was verified based on the grapevine circRNA dataset, named Vv-circPTCD1. Pentatricopeptide repeat (PPR) protein is a large gene family in plants that are characterized by 2-15 tandem repeats of 30-40 amino acid length motifs. Some PPR proteins have been supposed to play roles in organellar RNA metabolism, organ development, as well as stress response [39]. Overexpression of Vv-circPTCD1 deteriorated the tolerance to heat, salt, and PEG-mediated osmotic stresses in Arabidopsis. While similarly, the linear fragment identical to the circRNA sequence also did. However, the Vv-circPTCD1 overexpressed callus mass of 'Thompson Seedless' failed to display the phenotypes consistent with Arabidopsis. Those results implied that the functions of Vv-circPTCD1 were potentially sequence-dependent and not conserved among plants.

Identification and Verification of Vv-circPTCD1
Based on the previously published circRNA dataset of 'Muscat Hamburg' [38], a circRNA_4363, back-spliced from the second exon of the pentatricopeptide repeat (PPR) proteins family gene PTCD1, was identified and renamed Vv-circPTCD1 in this research. PTCD1 encodes a protein with 636 amino acid residues in length and contains the nine tandem PPR domains ( Figure 1A). In the Arabidopsis genome, meanwhile, two orthologs (AT5G21222 and AT5G25630) were characterized, and phylogenetic analysis showed that AT5G21222 and PTCD1 (VIT_211s0016g038030.1) were clustered into one clade with 96% confidence (1000 replicates) ( Figure 1B). The Vv-circPTCD1 was further verified by cloning and sequencing, and the back-splicing site was AG/GA ( Figure 1C). Divergent and convergent primers were designed to amplify the target fragment from DNA and leaf cDNA, and the results showed that the fragment with the back-splicing site was only detected in cDNA templates ( Figure 1D). " indicated the convergent primer pairs, and " " indicated the divergent primer pairs.

Nonconservative Back-Splicing of Vv-circPTCD1 in Plants
To investigate whether 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 structures of PTCD1 orthologs were observed (Figure S1), and a CDS (coding sequence) back-spliced into Vv-circPTCD1 was found in grapevines, which was highly conserved by multiple sequence alignment among plants ( Figure S2). It was reasonably hypothesized that this circRNA orthologous with Vv-circPTCD1 might also be conserved in plants. Therefore, the divergent primer pair was tentatively designed to clone the orthologs in Arabidopsis and N. benthamiana, although non-specific amplification was found in N. benthamiana, while the Vv-circPTCD1 orthologs failed to be cloned ( Figure S3). Those results suggested that the back-splicing of Vv-circPTCD1 was potentially species-dependent.

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 being scarcely expressed in roots, consistent with that of the host gene ( Figure 2A). However, the accumulation level of the 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 being mildly downregulated during initial treatment, the Vv-circPTCD1 was significantly downregulated on the fifth day after treatment ( Figure 2B,C). During drought treatment, VvSnRK2I (VIT_207s0197g00080), a marker gene homologous with AtSRK2I/AtSnRK2.3 induced by ABA in Arabidopsis [40], was significantly induced when the soil water potential reduced to −0.8 and −1.0 MPa ( Figure 2D). The expression level of PTCD1 was mildly upregulated 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 barely changed during drought treatment ( Figure 2F). We further investigated the expression pattern during heat stress, and the results showed that the heat shock factor genes VvHsfA7, as a marker of heat stress [22], were remarkably upregulated ( Figure 2G), and the expression level of PTCD1 was downregulated with the prolongation of treatment ( Figure 2H). Although the expression level of Vv-circPTCD1 was downregulated after 4 h of treatment, the expression level was also slightly upregulated after 2 h of treatment ( Figure 2I). Those results indicated that the Vv-circPTCD1 expression always fell behind that of PTCD1 and potentially played a role during abiotic stresses in grapevines.

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, a corresponding linear fragment (Lin-ear_PTCD1) identical to 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-E). The expression level of the host gene and flanking genes were further investigated in all OE callus masses. Among Vv-circPTCD1 lines, the expression level of the host gene was slightly downregulated ( Figure 3F), while the upstream gene (VIT_211s0016g03850) and downstream gene (VIT_211s0016g03820) were not significantly affected ( Figure 3G,H). However, among Linear_PTCD1 OE lines, the expression level of the host gene was discrepant among different lines, which is downregulated in line 2 ( Figure 3I). The upstream gene and downstream gene were also slightly downregulated ( Figure 3J,K). In order to uncover whether the function of Vv-circPTCD1 is conserved among plants, ectopic transformation was also performed in Arabidopsis ( Figure S4).

Phenotype of OE Lines under Heat Stress
To investigate whether Vv-circPTCD1 plays a role in Arabidopsis, the four-week-old seedlings were treated for 12 h at 45 • C followed by exposure to 22 • C for 7 days to recover. Results showed that compared with WT, the dead rosette leaves were significantly increased in Vv-circPTCD1 OE lines (CircPTCD1-OE), and the reproductive growth was unaffected ( Figure 4A). We also counted the death rate of rosette leaves in Arabidopsis and found that the damage rate of rosette leaves was significantly increased in transgenic lines ( Figure 4B). It was a wonder that the rosette leaves were also remarkably damaged in linear fragment OE lines (Linear_PTCD1-OE), 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 for 15 d at 40 • C. We found that all calluses became brown; however, phenotypic differences were not significant, no matter which OE callus mass ( Figures 4C and S5). Those results imply that the function of Vv-circPTCD1 is potentially not conserved in plants. The expression pattern of Vv-circPTCD1 during salt, drought, and heat stress. Differences between mean values of groups are compared using a t-test ("*": p < 0.05; "**": p < 0.01; "***": p < 0.001). The data are presented as the mean ± SE.

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, a corresponding linear fragment (Linear_PTCD1) identical to Vv-circPTCD1 was also constructed as a control ( Figure 3A). The expression pattern of Vv-circPTCD1 during salt, drought, and heat stress. Differences between mean values of groups are compared using a t-test ("*": p < 0.05; "**": p < 0.01; "***": p < 0.001). The data are presented as the mean ± SE.  The red bar chart means the Vv-circPTCD1 OE callus, the turquoise bar chart means the Linear_PTCD1 OE callus. Differences between mean values of groups are compared using a t-test ("*": p < 0.05; "**": p < 0.01; "***": p < 0.001). The data are presented as the mean ± SE.

Phenotype of OE Lines under Heat Stress
To investigate whether Vv-circPTCD1 plays a role in Arabidopsis, the four-week-old seedlings were treated for 12 h at 45 °C followed by exposure to 22 °C for 7 days to recover. Results showed that compared with WT, the dead rosette leaves were significantly increased in Vv-circPTCD1 OE lines (CircPTCD1-OE), and the reproductive growth was unaffected ( Figure 4A). We also counted the death rate of rosette leaves in Arabidopsis and found that the damage rate of rosette leaves was significantly increased in transgenic lines ( Figure 4B). It was a wonder that the rosette leaves were also remarkably damaged in linear fragment OE lines (Linear_PTCD1-OE), 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 for 15 d at 40 °C. We found that all calluses became brown; however, phenotypic differences were not significant, no matter which OE callus mass ( Figures 4C and S5). Those results imply that the function of Vv-circPTCD1 is potentially not conserved in plants. The red bar chart means the Vv-circPTCD1 OE callus, the turquoise bar chart means the Linear_PTCD1 OE callus. Differences between mean values of groups are compared using a t-test ("*": p < 0.05; "**": p < 0.01; "***": p < 0.001). The data are presented as the mean ± SE.

Phenotype of OE Lines under Salt Stress
To further survey the potential function of Vv-circPTCD1 in plants, four-day-old seedlings of Arabidopsis were transplanted into normal and NaCl-containing media to observe the variation. Results suggested that different OE lines and WT seedlings were normally and identically grown in NaCl-free medium ( Figure 5A), while the survival rate

Phenotype of OE Lines under Salt Stress
To further survey the potential function of Vv-circPTCD1 in plants, four-day-old seedlings of Arabidopsis were transplanted into normal and NaCl-containing media to observe the variation. Results suggested that different OE lines and WT seedlings were normally and identically grown in NaCl-free medium ( Figure 5A), while the survival rate of transgenic lines (CircPTCD1-OE) was lower on the NaCl-containing medium after five days compared to WT, although most Arabidopsis seedlings died after salt stress ( Figure 5B,C). The OE lines of the linear fragment (Linear_PTCD1-OE) were also severely influenced and had a lower survival rate on NaCl-containing media ( Figure 5B,C). In grapevine callus, the OE callus mass transformed with linear fragment adversely displayed an albino phenotype, while the OE callus mass transformed with Vv-circPTCD1 also did but much slighter compared to the WT callus mass ( Figure 5D). Those results also showed the Vv-circPTCD1 potentially served as a negative factor for salt stress, and the linear fragment identical with the Vv-circPTCD1 also plays the same roles in Arabidopsis and grapevine callus. The phenotype of OE seedlings and WT on 1/2 MS medium; (B,C) The phenotype and survival rate of OE seedlings and WT on 1/2 MS medium containing 300 mM NaCl; (D) The phenotype of OE and WT callus mass incubated on medium containing 300 mM NaCl. Differences between mean values of groups are compared using a t-test ("*": p < 0.05; "***": p <0.001).

Phenotype of OE lines under Drought Stress
To comprehensively explore the function of Vv-circPTCD1 in Arabidopsis, the fourweek-old seedlings were stopped from watering. Results showed that the transgenic lines of Vv-circPTCD1 (CircPTCD1-OE) were much more susceptible and withered, and the leaves turned purple due to the accumulation of anthocyanins during drought stress ( Figure 6A). The OE lines of the linear fragment (Linear_PTCD1-OE) also turned purple and withered during drought stress compared to the WT ( Figure 6A). The four-day-old seedlings were further transplanted into normal and PEG-containing medium, and it was The phenotype and survival rate of OE seedlings and WT on 1/2 MS medium containing 300 mM NaCl; (D) The phenotype of OE and WT callus mass incubated on medium containing 300 mM NaCl. Differences between mean values of groups are compared using a t-test ("*": p < 0.05; "***": p < 0.001).

Phenotype of OE Lines under Drought Stress
To comprehensively explore the function of Vv-circPTCD1 in Arabidopsis, the fourweek-old seedlings were stopped from watering. Results showed that the transgenic lines of Vv-circPTCD1 (CircPTCD1-OE) were much more susceptible and withered, and the leaves turned purple due to the accumulation of anthocyanins during drought stress ( Figure 6A). The OE lines of the linear fragment (Linear_PTCD1-OE) also turned purple and withered during drought stress compared to the WT ( Figure 6A). The four-day-old seedlings were further transplanted into normal and PEG-containing medium, and it was found that different transgenic lines (CircPTCD1-OE and Linear_PTCD1-OE) and WT seedlings were normally and identically growing in normal media ( Figure 6B), while the growth of rosette leaves was obviously inhibited in Vv-circPTCD1 and linear fragment OE lines (CircPTCD1-OE and Linear_PTCD1-OE), the growth of roots was barely inhibited (Figure 6C-E). Those results suggested that the Vv-circPTCD1 and corresponding linear fragment also play negative roles during drought stress in Arabidopsis. In grapevine callus, on the contrary, the OE callus mass of the Vv-circPTCD1 and linear fragment transgenic callus mass (CircPTCD1-OE and Linear_PTCD1-OE) significantly displayed resistance to osmotic stress mediated by PEG-8000, while the WT had become brown after 30 days of 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 ( Figures 6G and S5). Those results implied that the function of Vv-circPTCD1 is potentially species-dependent.

Discussion
CircRNAs are widespread in plants. Most circRNAs are generated by back-splicing from the exon or exons of the genic primary mRNA; for example, 94.5% are derived from the coding genes in rice, and the remaining are intergenic circRNAs [41]. In addition, the mitochondrion genome is also proven to generate circRNAs, called mitochondrionencoded circular RNAs (mcircRNAs) [17]. Here, we identified a circRNA spliced from the (G) The growth index of callus mass incubated on medium containing PEG-8000. Differences between mean values of groups are compared using a t-test ("*": p < 0.05; "**": p < 0.01; "***": p < 0.001; n.s.: p > 0.05). The data are presented as the mean ± SE.

Discussion
CircRNAs are widespread in plants. Most circRNAs are generated by back-splicing from the exon or exons of the genic primary mRNA; for example, 94.5% are derived from the coding genes in rice, and the remaining are intergenic circRNAs [41]. In addition, the mitochondrion genome is also proven to generate circRNAs, called mitochondrion-encoded circular RNAs (mcircRNAs) [17]. Here, we identified a circRNA spliced from the second exon of PTCD1, a PPR family gene, in which it was spliced based on a canonical GU/AG splicing signal [38] (Figure 1C). In general, hundreds of these superfamily proteins encoded by the PPR family genes located in the nuclear and organelle genomes are found in plants and are characterized by multiple tandem arrays of the PPR domain [42]. Previous findings proved that the PPR proteins, which serve as RNA binding proteins, affect cytoplasmic male sterility, embryogenesis, seed development, and stress [43][44][45][46]. Interestingly, the physical gene structure of PTCD1 orthologs is similar, and the sequence of the CDS backspliced into Vv-circPTCD1 is highly conserved among plants (Figures S1 and S2). However, the verification failure of the corresponding circRNA in Arabidopsis and N. benthamiana implies that circPTCD1 is not naturally conserved among plants. No matter what mammals and plants are, a factually small proportion of circRNAs are supposed to be conserved, such as the overlap of 26% of circRNAs in both humans and mice based on the syntenic region of the genome [8]. In plants, only 8.7% of circRNAs are genomically conserved in dicotyledonous plants, even that is less than half in the Oryza genus [41]. Taken together, Vv-circPTCD1 is not evolutionarily, although the host genes are conserved. This result partially supports the idea that circRNAs result largely from splicing errors [47,48].
We also successfully constructed the artificial vector containing a reverse complementary sequence pair 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) [38]. 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, short interspersed elements in primates, are preferentially circularized [49]. In Populus tomentosa, miniature inverted-repeat transposable elements (MITEs) are enriched in circRNA flanking regions and serve as a cis-regulatory factor to significantly regulate exon circularization [15]. Furthermore, previous findings suggested that the back-splicing of circRNA is also executed by the canonical spliceosome, a highly conserved precursor mRNA splicing mechanism in eukaryotes [50]. Additionally, the flanking intron sequence is crucial for the precise splicing and circularization of one exon and exons, which has been proven by heterologous investigations [38,51]. Therefore, the inverted repeats and flanking sequence commonly contribute to the fidelity and expression level of Vv-circPTCD1 in Arabidopsis and grapevine callus ( Figures 3B,C and 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 [52].
Additionally, previous investigations uncovered that circRNAs are incorporated into chromatin by an R-loop structure that interacts with the nuclear DNA to regulate chromatin stabilization or alternative splicing [13,53]. The host gene and neighboring genes are potentially influenced by overexpressed or deleted circRNA. In rice, a neighboring gene is significantly upregulated at circRNA mutant lines [14]. In this study, Vv-circPTCD1 has a slightly negative effect on the host gene and barely any effect on the neighboring genes after overexpression of Vv-CircPTCD1 ( Figure 3F-H). In P. tomentosa, overexpression of Circ_0003418 significantly increased the expression abundance of its parental gene [15]. Therefore, circRNA potentially has the ability to function by disturbing the expression of host or neighboring genes in plants.
The expression pattern of circRNAs is also determined by tissue and environmental factors and even genotype [15,54]. We found that the Vv-circPTCD1 focused in this research was preferentially expressed in young and mature leaves, which is consistent with the expression pattern of the corresponding host gene, while the expression abundance of Vv-circPTCD1 was less than 2% that of the 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 largely obeyed the expression pattern of host genes in plants, such as grapevine [38]. 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-F), indicating it plays a role in grapevine responses to drought stress. In addition, the expression level of Vv-circPTCD1 is remarkably decreased during salt stress after 5 days, while a decrease in 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 is potentially controlled by the specific abiotic stress, at least during the processing or degradation pathway [55].
The hypothesis of "splicing error" is indeed uncontradictory with the biochemical activities and biological functions of circRNAs. According to 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 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 [50]. The strategy to evade this problem is to transform the linear fragment identically with circRNA as a control. Generally, the linear fragment scarcely functions in plants [13,15,38]. In the present research, the linear counterpart is also transformed as a control ( Figures 3D,E and S4). Interestingly, the identical stress-induced phenotype is observed between Vv-circPTCD1 and its linear counterpart in OE lines (Figures 4-6). It is putative that (1) the incompletely cyclized transcript plays roles identical to its linear counterpart; (2) the Vv-circPTCD1 functions in a sequence-dependent manner. 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 (Figures 3 and 6). Coincidentally, the cognate sequence-generated Vv-circPTCD1 is greatly conserved among multiple species (Figure S2), while the biogenesis is not ( Figure S3). Those results imply that the biological functions of circRNA are potentially species-dependent in plants.

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 a flowerpot in a greenhouse. Cuttings with four well-developed leaves were used for drought stress. 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 • C, 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 • C) and watered with 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 • C. All treatments were carried out on three biological replicates.
Arabidopsis ecotype Col-0 was cultured in a 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, six-week-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 recover. For drought stress treatment, six-week-old WT and transgenic plants were stopped watering until the substrate was completely dry, and the phenotype was observed. The PEG-8000 treatment (the water potential was −0.7 MPa) was performed according to the Verslues's descriptions [56]. The four-day-old seedlings were transferred into the plate containing PEG-8000 and cultured in the chamber for another 5 days. For salt treatment, plates containing 1/2 MS salts supplemented with 300 mM NaCl were prepared, and four-day-old seedlings were incubated. The survival rates were counted after 5 days. Three biological replicates were performed.

Validation of circRNA and RT-qPCR
The total RNA of Arabidopsis and grapevine callus was extracted by the TRIzol reagent (Sangon Biotech, China), and the total RNA of grapevine leaves was extracted by the CTAB method. In brief, 200 mg of the sample was powdered and transferred into a RNase-free 2 mL tube. Then, added quickly was 900 µL pre-heated (65 • C) CTAB buffer (2% CTAB, 100 mM Tris-HCl, 20 mM EDTA, 2 M NaCl, 2% PVP-40, pH8.0) and 20 µL β-mercaptoethanol. The mixture was mixed robustly and then bathed at 65 • C for 15 min. The lysate was extracted with 900 uL of chloroform: isoamylol (24:1, v/v) by vortexing for 30 s and centrifuged at 4 • C for 10 min at 12,000 rpm. This was repeated two times with chloroform: isoamylol extraction, 20 µL NaAC (pH5.2) was added, and then 60 µL pre-cooled (−20 • C) absolute alcohol, and bathed 10 min on ice. Then an equal volume chloroform: isoamylol was added before vortexing. The sample was centrifuged at 4 • C for 10 min at 12,000 rpm, then 150 µL 10 M LiCl was added to 450 µL of supernatant and mixed. The RNA was precipitated at 4 • C for 6 to 8 h and harvested by centrifugation at 4 • C for 10 min at 12,000 rpm. Then we carefully removed the supernatant and washed it two times with 70% pre-cooled (−20 • C) alcohol. The pellet was dried and dissolved by 30 µL DEPC-treated ddH 2 O.
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, a 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 qTOWER 3 Real-Time PCR Detection System (Analytikjena, Germany). Gene expression levels were calculated by the 2 −∆∆Ct method [57]. For each RT-qPCR assay, three biological duplicates were conducted.

Vector Construction
All expression vectors were constructed based on the pHB binary plasmids. The construction of the circRNA expression vector was conducted based on a previously published strategy [38]. For the linear fragment expression vector, the linear fragment derived from the same sequence with circRNA was cloned into the pHB vector, which was digested by BamH I and Xba I. The constructed plasmids were individually transformed into Agrobacterium tumefaciens GV3101 strains and incubated at 28 • C for 2-3 d on 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 • C 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. An equal volume of 50% glycerol was added to the positive clone and stored at −80 • C.

Arabidopsis Transformation and Verification
Arabidopsis transformation was performed according to the floral dip method [58]. T1 seedlings were selected by spraying Glufosinate ammonium 10% solution (dilute 1000×) (Sangon Biotech, Shanghai, 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 the PCR method and RT-qPCR reaction, and the splicing site was verified by further sequencing.

Callus Transformation of 'Thompson Seedless' and Treatment
The embryogenic callus was induced from floral explants of 'Thompson Seedless' according to a previously published protocol [59]. The callus mass was subcultured monthly in MSTP medium (MS base salts, 20 g·L −1 sucrose, 1 mg·L −1 TDZ, 2.2 mg·L −1 picloram, pH5.8) in dark condition at 26 • C. For callus transformation, the A. tumefaciens-containing vector was inoculated in 20 mL of LB liquid medium and incubated until OD 600 reached 1.0. After centrifugation for 5 min at 6000 rpm at 25 • C, the medium was discarded. The pellet was resuspended, and the OD 600 was adjusted to 0.4 with liquid MS medium (0.1 mM acetosyringone, pH5.8). The bacterial mixture was incubated with the embryogenic callus for 30 min at 40 rpm in a 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 conditions at 26 • C for 3 days. After 3 days, the callus was transferred to solid MSTP medium containing 50 mg·L −1 hygromycin, and 200 mg·L −1 Timentin. Petri dishes were subcultured monthly 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 subcultured into MSTP medium for one week at 26 • C and then further cultured at 40 • C for 15 days. The callus mass was subcultured into MSTP medium supplemented with 300 mM NaCl for salt stress at 26 • C. For PEG-8000 treatment, the plates were prepared according to the above description, of which the base salt was replaced by MSTP medium supplemented with 15 g·L −1 agar on the solid layer, but the liquid layer was the MSTP medium without sucrose and agar. Then the callus mass was subcultured on the PEG-containing plates and cultured at 26 • C for one month. For measuring the growth index of callus mass, the callus was shaped into small pellets (about 3 mm in size), subcultured on the PEG-containing plates, and immediately photographed. After one month of culturing, the plates were also photographed. The area of the pellet was analyzed by Image J software (https://imagej.net/downloads). The growth index of callus mass was the ratio of the area of the pellet after and before culturing. Three biological replicates were performed.

Statistical Analysis and Gene Structure
Data were analyzed statistically using the SPSS 19.0 software. Significant differences were determined using a test; p < 0.05 was considered statistically significant. The gene structure was generated by the GSDS 2.0 web tools in Figure S1 [60].

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.

Supplementary Materials:
The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/plants12122332/s1, Figure S1: The physical gene structure of PTCD1 (VIT 211s0016g03830) orthologs in plants. Figure S2: The multiple sequence alignment of the second exon of PTCD1 orthologs. Figure S3: PCR detection of putative Vv circPTCD1 ortholog candidates in Arabidopsis and N. benthamiana. Figure S4: Verification of Vv circPTCD1 and Linear_PTCD1 OE lines in Arabidopsis (A) The gel electrophoresis of PCR amplification; (B) confirmation of back splicing. Figure S5: The phenotype of callus incubated in normal condition. The overexpressed callus mass and WT were incubated at 26 • C on the dark and the phenotypic difference was undetectable. Funding: This work was supported by the National Natural Science Foundation of China (Grant No. 32122076, 32202446), the Shanghai municipal key task projects of "Prospering Agriculture by Science and Technology Plan" (2020-02-08-00-08-F01458), and the earmarked fund for CARS-29.

Data Availability Statement:
No new datasets were generated or analyzed in this study.

Conflicts of Interest:
The authors declare no conflict of interest.